OET pronunciation for Mandarin speakers

Mandarin is the second-largest L1 in the OET candidate pool worldwide. Nurses and doctors trained in mainland China, Taiwan, Singapore, Malaysia, and the wider Mandarin-speaking diaspora are now a significant cohort migrating to Australia, the UK, New Zealand, and Ireland. AHPRA's overseas-trained-nurse statistics consistently rank China-trained candidates in the top five source countries each year, behind only the Philippines and India. Yet most of the publicly-available OET preparation content is written either for English-as-a-first-language learners or with a generic "international student" lens that ignores the specific L1.

This post fixes that. It is a Mandarin-speaker-specific guide to the intelligibility criterion of OET Speaking — the single criterion that costs Mandarin-L1 candidates the most marks, and the one that is easiest to fix without compromising how you sound.

Before the practical drills, three things have to be clear. The first is what OET intelligibility actually rewards. The second is which specific Mandarin-to-English sound patterns the rubric notices. The third is which patterns are not pronunciation problems at all, even though they feel like they are. Then the drills.

If you have not yet worked through the 9-criteria OET Speaking rubric — read that first. This post assumes you know what intelligibility means inside the rubric and why it is one criterion of nine, not the whole exam.

What OET intelligibility actually rewards

The OET rubric is explicit on one point that most accent-coaching content is fuzzy about: any accent is acceptable as long as a non-native English listener can follow you without effort. That is the literal wording in the public scoring guide. The criterion is called intelligibility, not "accent quality".

For a Mandarin speaker, that means:

• You will never be penalised for sounding Chinese. Examiners are trained to expect candidate accents from across the world, and Mandarin-accented English is one of the most common accents they encounter.
• You will be penalised for specific sound substitutions that change meaning — when "thin" sounds like "sin", or "leave" sounds like "live".
• You will be penalised for word stress that makes multi-syllable medical words unrecognisable — when "examination" comes out with equal weight on every syllable.
• You will be penalised for running consonants together or dropping word endings so badly that the listener has to mentally re-parse what you said.

That is it. Four narrow conditions. Everything else about how you sound — vowel colour, rhythm, the music of your sentences, even residual tone-like pitch glides on individual words — does not lose marks on intelligibility. Mandarin candidates often worry about sounding "too Chinese". The examiners I have spoken with rarely flag that as the issue. What they flag is precisely the five sound-level patterns below.

The five Mandarin → English sound patterns the OET rubric notices

These are the five patterns linguists have studied most extensively in Mandarin L1 / English L2 speakers, and the five examiners consistently flag on real OET role-plays. None of them apply to every Mandarin speaker — your own pattern depends on where you grew up, whether your home dialect was Mandarin, Cantonese, Shanghainese, Hokkien, or something else, and how much exposure you have had to native English in your clinical training so far.

Pattern 1: /θ/ and /ð/ (the "th" sounds)

Mandarin has no /θ/ (the voiceless "th" in "think") and no /ð/ (the voiced "th" in "this"). Both phonemes are foreign to the Mandarin sound inventory. Most adult Mandarin speakers substitute the closest sounds available:

• /θ/ → /s/: "think" sounds like "sink", "three" like "sree" or "tree", "thirty" like "sirty"
• /ð/ → /d/ or /z/: "this" sounds like "dis" or "zis", "the" like "duh" or "zuh"

The OET cost: clinical terms that depend on /θ/ become ambiguous, and connector words that depend on /ð/ break the rhythm of the sentence.

• "thigh" ↔ "sigh" — when you describe a leg injury, this matters
• "throat" → "sroat" or "froat"
• "thumb" → "sum" or "fum"
• "breath" → "bress"
• "smooth" → "smoos" or "smood"

The reverse — overcorrecting /s/ to /θ/ — also happens once you start drilling, and is just as costly. "Sick" → "thick" is an embarrassing slip in any clinical context.

Pattern 2: Final consonant deletion and cluster simplification

Mandarin syllables end in vowels, nasals (-n, -ng), or — in some dialects — a glottal stop. English allows complex word-final clusters: "asked" (-skt), "world" (-rld), "next" (-kst), "tests" (-sts). Under speech pressure, Mandarin speakers tend to drop the final consonants entirely, or to insert an epenthetic vowel between them.

The OET-relevant cases:

• "next" → "neck" or "nex"
• "first" → "firs" or "first-uh"
• "asked" → "ask"
• "tests" → "tes" or "test-uh"
• "diagnosed" → "diagnose" or "diagnose-duh"
• "hospital" → "hospitaw" or "hospi"
• "results" → "result" or "resul"

Most of these are still intelligible individually. The problem is that when the dropped consonant is the past tense marker (asked → ask, diagnosed → diagnose), it changes the meaning. "I diagnosed her with hypertension yesterday" and "I diagnose her with hypertension yesterday" tell the examiner two different stories about your clinical timeline.

The epenthetic-vowel version ("test-uh" instead of "tests") sounds less abrupt to your own ear, but the examiner hears the extra vowel as a hesitation. Two or three of those in a sentence and your fluency score drops too.

Pattern 3: Vowel length conflation

English contrasts long and short vowels phonemically — they change word meanings. "Sheep" (long /iː/) and "ship" (short /ɪ/) are different words. "Leave" and "live", "feel" and "fill", "beat" and "bit", "fool" and "full" — all minimal pairs distinguished only by vowel length plus a small quality difference.

Mandarin vowels do not have phonemic length. All vowels are produced at roughly the same duration, and the small length differences that do exist are not used to distinguish meaning. As a result, Mandarin speakers often produce English long and short vowels as the same vowel.

The OET-relevant cases — and some are clinically embarrassing:

• "sheet" ↔ "shit" — happens more often than candidates realise; the patient role-player will not interrupt, but the examiner notes it
• "leave" ↔ "live" — "she will leave the hospital tomorrow" vs "she will live the hospital tomorrow"
• "feel" ↔ "fill" — "do you feel any pain" vs "do you fill any pain"
• "beat" ↔ "bit" — "her heart rate is beating fast" only works if "beat" is clearly long
• "reach" ↔ "rich" — different verbs with different patient implications
• "pool" ↔ "pull" — relevant in physiotherapy contexts

The single most useful intervention is over-articulating the length contrast: making long vowels deliberately longer than feels natural. Native English speakers do not pronounce "sheep" as a particularly long vowel — but they pronounce it noticeably longer than "ship". Mandarin speakers tend to land in the middle, which is where the ambiguity lives.

Pattern 4: /v/ → /w/

Mandarin has no /v/. Most adult speakers substitute /w/ — the closest articulation that uses the same general region of the mouth.

The OET cost:

• "very" → "wery"
• "vital signs" → "wital signs"
• "vein" → "wayn"
• "vomit" → "womit"
• "ventilator" → "wentilator"
• "examination" → often fine; the /v/ in "valves" or "values" is where it slips

This is one of the easier patterns to fix once you notice it. /v/ is articulated by lightly pressing your top teeth against your lower lip and letting air vibrate through — a labio-dental fricative. /w/ is a rounded-lip glide with no teeth contact. The physical difference is large; you can feel the contrast clearly once you focus on it. The challenge is consistency under speech pressure, not articulation.

A subset of Mandarin speakers — particularly from northern China — produce a labio-dental /v/-like sound natively in some contexts (for the /w/ in pinyin "wei", "wang", some speakers produce it as [v]). For those candidates, the issue is reversed: the /v/ comes out fine in English, but they need to be careful not to over-apply it.

Pattern 5: Word stress and syllable timing

This is the highest-leverage pattern of the five. English is stress-timed: stressed syllables fall at roughly even intervals, and unstressed syllables compress to fit. Mandarin is syllable-timed (more accurately, tone-timed, but the effect on English is the same): every syllable gets roughly equal weight and duration.

When Mandarin speakers carry that habit into English, multi-syllable words come out with all syllables equally stressed. Worse, residual Mandarin tonal patterns sometimes overlay onto English syllables — assigning a rising or falling pitch to each one — which makes the word sound chant-like rather than spoken.

The OET-relevant cases:

• "examination" — should be ex-a-mi-NA-tion (stress on the fourth syllable). Mandarin-default tends toward EX-A-MI-NA-TION (equal) or ex-a-MI-na-tion (penultimate guess)
• "hypertension" — hy-per-TEN-sion (stress on TEN). Mandarin-default often HY-per-ten-SION or hy-PER-ten-sion
• "diabetes" — di-a-BE-tes. Mandarin-default DI-a-BE-tes (two equal stresses) is intelligible but sounds halting
• "comfortable" — COM-fort-a-ble (stress on COM, with the middle syllables compressed). Mandarin-default com-FORT-a-ble is a band-loser
• "antibiotic" — an-ti-bi-O-tic. Mandarin-default an-TI-bi-O-tic
• "asymptomatic" — a-SYMP-to-MAT-ic (two stresses). Mandarin-default any of several equal-weight versions

Word stress under pressure separates Band B from Band C+ on intelligibility for almost every Mandarin candidate I have seen audited. The fix is memorisation of a clinical-vocabulary stress list — drill 3 below.

Patterns that are NOT pronunciation problems

Before the drills, it is worth flagging three things that feel like pronunciation issues but are scored under different criteria. If you spend pronunciation drill time on these, you waste hours.

Direct translations from Mandarin clinical idiom ("Chinglish")

Many Mandarin-trained clinicians have a stock of idioms translated word-for-word from Chinese that sound odd in English but are perfectly pronounceable. "Please pay attention to your blood pressure" (注意血压) — the Chinese is natural, the English is grammatically correct but registers as oddly formal. "Eat the medicine" (吃药) instead of "take the medicine". "Cut the line" (剪线) instead of "remove the sutures".

None of these are intelligibility problems. The patient understands you. They are scored under appropriateness of language or sometimes vocabulary. The fix is exposure to native clinical English — listen to UK or Australian nurses speak to patients, on YouTube or in clinical podcasts, and notice the verb choices. Pronunciation drills will not help here.

Honorifics and over-formality

Mandarin clinical culture is more hierarchical than Australian or UK clinical culture. A patient is "the patient" (病人) or, more respectfully, "the elderly gentleman/lady" (这位先生/女士). When this register transfers into English, it can come out as "the gentleman is feeling unwell today" instead of the more typical "Mr Smith is feeling unwell today". Over-formal register is scored under appropriateness of language, not intelligibility.

The fix is awareness, not pronunciation. In Australian and UK clinical practice, the default is first-name basis after the first introduction. ("My name's Sarah, I'll be your nurse today — what would you like me to call you?") Practise this opening; the rest follows.

Hesitation markers and filled pauses

Mandarin uses 那个 ("nèige" — literally "that one") and 这个 ("zhège" — literally "this one") as hesitation fillers. Under English-language pressure, many Mandarin speakers retain these — saying "那个" softly mid-sentence, or substituting "this... this..." or "that... that..." as a stalling tactic. The English "um" and "uh" are perfectly acceptable on OET; "那个" is intelligible but flagged by examiners as language code-switching.

The fix is not pronunciation training. It is awareness during practice recordings. Count the number of Mandarin hesitation markers in your own speech each week. The number should drop.

Drill 1: /θ/, /ð/, and /v/ minimal pairs

The standard pronunciation drill for these patterns is minimal pairs: word pairs that differ in exactly one sound, like "thin / sin" or "vest / west". Saying them in succession trains your mouth to feel the difference.

Twenty minutes a day for two weeks moves most candidates from "swaps under pressure" to "consistent under pressure". Beyond two weeks the curve flattens — you are at the level where remaining slips come from speaking too fast, not from not knowing the sounds.

Clinical minimal pairs

These are the ones you will actually use in OET role-plays. Drill these, not generic phonetics-textbook pairs.

/θ/ ~ /s/ (think / sink):

• think / sink
• thigh / sigh
• thin / sin
• thank / sank
• three / see (close enough to drill)
• "her sroat is sore" / "her throat is sore"
• "a sin patient" / "a thin patient"
• "I sink she's improving" / "I think she's improving"

/ð/ ~ /d/ (this / dis):

• this / dis
• they / day
• though / dough
• breathe / breed
• "dat is the issue" / "that is the issue"
• "do you fink so" / "do you think so"

/v/ ~ /w/ (vest / west):

• vest / west
• vine / wine
• vary / wary
• veil / wail
• "wital signs" / "vital signs"
• "I'll check your weins" / "I'll check your veins"
• "she's womiting" / "she's vomiting"
• "the wentilator is set to" / "the ventilator is set to"

How to drill

1. Record yourself saying each pair five times: slowly, slowly, normal, normal, fast.
2. Listen back. The /θ/ in "think" should feel like the tip of your tongue lightly between your top and bottom teeth, with a brief puff of air. The /s/ in "sink" is tongue-behind-the-teeth, sustained airflow. The physical difference is visible in a mirror — if your tongue is hidden, you are not making /θ/.
3. For /v/, watch yourself in the mirror. The lower lip should touch the upper teeth. For /w/, the lips round and there is no teeth contact.
4. Drill in clinical sentences, not just isolated words. "I think she has a thin chance of full recovery" is the use case; "think" in isolation is the drill.
5. Vary the speed. Slow drills install the sound; fast drills install the reflex. The exam will demand fast.

After two weeks of this, twenty minutes a day, your /θ/ and /v/ accuracy on a fresh recording should be 90%+ even at conversational speed.

Drill 2: vowel length contrast (sheep / ship)

This is the drill Mandarin candidates most often skip — because the difference between "sheep" and "ship" sounds tiny when you first hear it, and you assume it cannot be scoring you down. It is. Vowel-length confusion is the single most common intelligibility flag for Mandarin candidates after word stress.

The minimal pairs to drill

Long /iː/ vs short /ɪ/:

• sheep / ship
• feel / fill
• leave / live
• reach / rich
• sleep / slip
• beat / bit
• heel / hill
• seat / sit

Long /uː/ vs short /ʊ/:

• pool / pull
• fool / full
• shoe / should
• who / hood

Long /ɑː/ vs short /ʌ/:

• heart / hut
• cart / cut
• park / puck
• last / lust

How to drill

1. Pick one column at a time. Say each pair five times, exaggerating the length of the long vowel. "Sheeeep / ship". "Leeeave / live". Hold the long vowel for what feels like too long.
2. Then say each at normal speed but consciously keeping the length difference. If a non-Mandarin listener cannot tell which one you said, you are not differentiating enough.
3. Embed in clinical sentences:
   • "She wants to leave the hospital tomorrow." (long /iː/)
   • "She wants to live at home." (short /ɪ/)
   • "Do you feel any pain?" (long /iː/)
   • "Please fill out this form." (short /ɪ/)
4. Record. Listen back. The single biggest tell of an unfixed vowel-length issue is hearing a sentence and not being able to tell, from your own audio, which word you said.

Two weeks of this, ten minutes a day, gets most candidates to a reliable contrast. After that, it is maintenance — periodic check-ins on your recordings.

Drill 3: stress on the top 50 clinical words

This is the highest-leverage drill in this post. Word stress, more than any other phonological factor, decides whether a multi-syllable clinical word is intelligible to a non-native-English-trained examiner.

The list

Memorise the stress pattern on each of these words. The stressed syllable is in CAPS.

| Word | Stress | |---|---| | hypertension | hyper-TEN-sion | | hypotension | hypo-TEN-sion | | diabetes | dia-BE-tes | | asthma | ASTH-ma | | anaemia (UK) / anemia (US) | a-NAE-mia | | diagnosis | diag-NO-sis | | symptom | SYMP-tom | | asymptomatic | a-SYMP-to-MA-tic | | comfortable | COM-fort-a-ble | | medication | medi-CA-tion | | examination | exami-NA-tion | | laboratory (UK) | la-BOR-a-tory | | laboratory (US/AUS) | LAB-ra-tory | | antibiotic | anti-bi-O-tic | | paracetamol | para-CE-ta-mol | | ibuprofen | ibu-PRO-fen | | ultrasound | UL-tra-sound | | oncology | on-CO-lo-gy | | cardiology | car-DI-o-lo-gy | | neurology | neu-RO-lo-gy | | pathology | pa-THO-lo-gy | | radiology | ra-DI-o-lo-gy | | pharmacology | pharma-CO-lo-gy | | psychiatry | psy-CHI-a-try | | haematology | hae-ma-TO-lo-gy | | paediatric | pae-di-A-tric | | anaesthetic | anaes-THE-tic | | recovery | re-CO-ve-ry | | consultation | consul-TA-tion | | referral | re-FER-ral | | prescription | pre-SCRIP-tion | | discharge (verb) | dis-CHARGE | | discharge (noun) | DIS-charge | | treatment | TREAT-ment | | condition | con-DI-tion | | reaction | re-AC-tion | | infection | in-FEC-tion | | operation | ope-RA-tion | | procedure | pro-CE-dure | | emergency | e-MER-gen-cy | | pregnancy | PREG-nan-cy | | delivery | de-LI-ve-ry | | paramedic | para-ME-dic | | physiotherapy | physio-THE-ra-py | | occupational | occu-PA-tio-nal | | respiratory (UK) | res-PI-ra-tory | | respiratory (US/AUS) | RES-pi-ra-tory | | immunisation | immuni-SA-tion | | inflammation | inflam-MA-tion | | ventilation | venti-LA-tion |

How to drill

1. Print the list. Tape it somewhere you will see five times a day — the bathroom mirror, the fridge, your car's dashboard.
2. Every time you look at it, say five words at random with deliberate stress on the marked syllable. Five seconds per word; this is a 30-second total drill done many times a day.
3. After three days you will know the list by heart and the look-up is automatic.
4. Then start using the words in sentences during your role-play practice. "Her diaBETes is poorly controlled." "We'll book an examiNAtion." Catch yourself if the stress slips.
5. The trickiest piece for Mandarin speakers is suppressing the equal-syllable-weight default. Force yourself to make the unstressed syllables noticeably shorter and quieter than the stressed one. It will feel unnatural at first; that means it is working.

This drill is the most boring of the three. It is also the one that returns the biggest band-jump in a two-week window.

Self-assessment recording protocol

Drills are useless without a feedback loop. Here is the minimal protocol:

1. Once a week, do one full 5-minute OET-style role-play. Record audio only. Use any case bank, or the OET Live nursing cases.
2. Listen back twice:
   • Pass 1: count slips. Tally /θ/ as /s/, /v/ as /w/, dropped final consonants, vowel-length collapses, wrong-stressed words. Write the tallies down.
   • Pass 2: pick the three most jarring slips — the ones that you found hardest to listen to. Re-record only those words in isolation, then in the sentence they appeared in. Listen to your re-recording against the original.
3. Track the tally over time. Week 1 might be 15 slips. Week 4 should be 6. Week 8 should be 2-3. If the number is not coming down, the issue is not the sounds — it is your practice volume. Add 5 minutes per day.

The number itself matters less than the trend. The single most useful piece of feedback is whether your Mandarin-L1 patterns are getting better over time. They almost always do if the drills are happening; they almost never do if they are not.

For a structured version of this protocol, OET Live AI scoring gives a per-criterion score on every role-play with timestamps for specific moments. We use it ourselves to track week-over-week intelligibility improvement. You can do the same thing manually with a phone recorder and ten minutes of patience.

Hard cases: sounds that keep slipping under exam pressure

If two weeks of drilling have not moved a specific pattern, there are three things to try.

Try 1: Slow your tempo by 10%

Most slips are speed-related, not phonological. Your mouth knows the sound; under pressure your nervous system short-circuits the production. The single highest-leverage intervention is consciously slowing your tempo for the first 30 seconds of the role-play. This anchors you for the rest.

For Mandarin speakers, this also serves a second purpose: slowing down disrupts the equal-syllable-timing reflex. The stressed-timed rhythm of English needs space between content words, and pace gives you that space.

Try 2: Anchor phrase

Pick a phrase that contains the difficult sound and rehearse it to absolute automaticity — to the point you could say it in your sleep. Use it in the first 30 seconds of the role-play.

• For /θ/: "I think the best thing is for you to..."
• For /v/: "Let me check your vital signs."
• For final clusters: "I'd like to ask you a few questions first."
• For vowel length: "Please feel free to ask me anything."
• For word stress: "I'll do an examination now."

These phrases serve two functions. First, they get the sound out cleanly early so your mouth has the muscle memory primed. Second, they buy you 5–10 seconds of structured speech while you settle your nerves.

Try 3: Reframe the criterion you are stuck on

Sometimes the issue is not phonological at all — it is that the criterion you are failing is not intelligibility, and you have been treating it as such because the symptoms (the patient looking confused, the examiner not following) feel like pronunciation problems. The actual issue might be:

• Fluency: you are speaking too fast or with too many fillers, so the listener cannot parse you even though your sounds are fine
• Organisation of talk: you are jumping topics without signposting, so the listener cannot follow even though each sentence is intelligible
• Vocabulary: you are using clinical terms without plain-English follow-up, so the patient (and the examiner role-playing the patient) does not understand
• Grammar under pressure: many Mandarin-L1 speakers drop articles ("a", "the") and tense markers when speaking quickly, and this surfaces as a comprehension issue that feels like pronunciation

If you are stuck on intelligibility after eight weeks of solid drills, get a recording reviewed by someone trained to score the OET rubric (or use a scoring tool that breaks the score down by criterion). The result is often "you are losing marks on fluency or grammar, not intelligibility" — and then you stop drilling pronunciation and switch to the actual issue.

This is the diagnostic problem the 8-week OET Speaking plan for working nurses is designed to solve — the week-1 diagnostic anchors which criteria you actually need to work on, instead of guessing.

A note on Cantonese, Shanghainese, and other home dialects

Many "Mandarin" candidates grew up speaking another Chinese language at home — Cantonese in Hong Kong and Guangdong, Hokkien in Taiwan and parts of Southeast Asia, Shanghainese in the Yangtze delta — and learned Mandarin as a second language. For those candidates, the L1 influence on English is a mix.

Two patterns worth noting:

• Cantonese speakers tend to have a stronger /θ/-as-/f/ substitution ("three" → "free") on top of the Mandarin-typical /θ/-as-/s/. Cantonese also retains stop codas (-p, -t, -k) which means final consonants are sometimes preserved better than in Mandarin speakers — but the stops are unreleased, which can sound like a dropped consonant to an English ear.
• Hokkien and Taiwanese-Mandarin speakers often have a stronger nasalisation pattern and tend to merge /n/ and /l/ in some positions ("nice" / "lice"). Vowel length contrast is usually slightly better than mainland Mandarin speakers because Hokkien has more vowel length distinction natively.

The drills in this post still apply — they are based on the patterns the OET rubric flags, which are largely shared. But if you grew up speaking Cantonese, also drill /θ/ vs /f/ in addition to /θ/ vs /s/. If you grew up speaking Hokkien, also drill /n/ vs /l/ in word-initial position.

The headline summary

If you read nothing else in this post:

1. OET intelligibility scores accent-neutral. Sounding Chinese is fine. Sounding unclear is not.
2. Five Mandarin-L1 sound patterns account for almost every mark lost: /θ/ and /ð/, final consonant deletion, vowel length, /v/ as /w/, wrong word stress.
3. Three drills, twenty minutes a day, for two-to-three weeks move most Mandarin candidates from C+ intelligibility to B intelligibility: th/v minimal pairs, vowel length contrast, multi-syllable clinical word stress.
4. The single highest-leverage drill is multi-syllable word stress on the top 50 clinical words. Boring, repetitive, devastatingly effective — and the one Mandarin candidates most often skip because the equal-syllable-weight default feels natural.
5. Three patterns that look like pronunciation but are not (Chinglish word choice, over-formal honorifics, Mandarin hesitation markers) live under appropriateness and fluency, not intelligibility. Do not waste pronunciation drill time on them.
6. If you are still stuck after eight weeks, the issue is probably not phonology — it is fluency, grammar under pressure, or organisation, dressed up in symptoms that look like pronunciation.

For the broader weekly structure that integrates these drills into a full OET prep plan, see the 8-week roadmap. For what Band A pronunciation actually sounds like in a real role-play, the annotated transcripts post walks through two cases turn-by-turn. For Filipino-L1 speakers, the equivalent post is OET pronunciation for Filipino nurses. For the full 9-criteria rubric these intelligibility marks sit inside, see how OET Speaking is scored.

You do not need to lose your voice to score Band A on OET Speaking. You need to fix five sound patterns. Do that, and the rest of who you are can come through unchanged.