If your first language is Turkish or Vietnamese, the marks you lose in Cambridge Speaking are rarely about accent. The B2 First handbook says intelligible speech can come with "a strong or unfamiliar accent". The losses come from a handful of predictable habits. For Turkish speakers: vowels inserted into consonant clusters ("sipor" for sport), v for w, missing articles, "he" and "she" mixed, and a long turn where the main point arrives last. For Vietnamese speakers: final consonants dropped, so plurals and past tenses disappear, clusters reduced, every syllable stressed, and turns that stop before the minute is up. For both: Part 3 turns that answer the partner without adding to the conversation. This page maps each habit to the criterion it costs in B2 First and C1 Advanced, shows how Cambridge's own examiners mark the same problems, and gives a 60-second drill for each.
The short answer
Cambridge Speaking is paired and marked by two examiners. Per the B2 First handbook, the assessor scores Grammar and Vocabulary, Discourse Management, Pronunciation and Interactive Communication on a 0 to 5 scale, and the interlocutor gives one Global Achievement mark; at C1 Advanced, Grammar and Vocabulary is split into Grammatical Resource and Lexical Resource. Candidates "are assessed on their own individual performance and not in relation to each other", so a stronger or weaker partner does not change your mark.
The good news for both groups: the habits below are well understood, and each has a mechanical fix. Turkish speakers often have strong grammar from school and lose marks mainly on sounds and on the order of ideas. Vietnamese speakers often have good vocabulary and lose marks on endings and rhythm, which also hide grammar they actually know. In both cases, the fastest gains are in Discourse Management and Interactive Communication: longer, better-organised turns and more questions to the partner.
- Accent is not marked. Intelligibility, intonation, stress and clear sounds are (B2 First handbook, Band 5 Pronunciation).
- Turkish: inserted vowels, w as v, th, final devoicing, articles, he and she, the main point last, the present continuous for habits.
- Vietnamese: final consonants, clusters, every syllable stressed, s and sh, missing -s and -ed, "because... so...", "although... but...", short turns.
- Both: answering without developing in Part 3. The fix is a question or a link in every second turn.
- These lists are TalkDrill's judgement, not Cambridge data. Record yourself and fix only what you hear.
How the criteria turn these habits into marks
The B2 First Band 5 Pronunciation descriptor reads: "Is intelligible. Intonation is appropriate. Sentence and word stress is accurately placed. Individual sounds are articulated clearly." Band 3 says the same with "generally" in each line. A habit that appears in every sentence, such as a dropped final s, keeps you at "generally". For Interactive Communication, Band 5 asks you to initiate and respond "linking contributions to those of other speakers" and to maintain and develop the interaction and negotiate "towards an outcome". The handbook glossary defines developing the interaction as "saying more than the minimum" or "proactively involving the other candidate with a suggestion or question", with the examples "What about bringing a camera for the holiday?" and "Why's that?".
| Criterion | Turkish habits that cost it (our judgement) | Vietnamese habits that cost it (our judgement) |
|---|---|---|
| Grammar and Vocabulary | Missing articles; he and she mixed; present continuous for habits ("I am going to school every day"); "do a mistake" | Missing -s and -ed; "because... so...", "although... but..."; missing "be" ("she very kind"); "In the photo have..." |
| Discourse Management | The answer arrives last, after the reasons; "yani" and "şey" as fillers | Turns that stop at 30 or 40 seconds; ideas listed, not linked |
| Pronunciation | Inserted vowels; w as v; th; final b, d, g devoiced; stress moved to the last syllable | Final consonants lost; clusters reduced; every syllable stressed; s and sh merged; th as t or d |
| Interactive Communication | Few questions to the partner; long reasons before a position | Waiting to be asked; "yes" that means "I am listening" read as agreement |
What Cambridge's examiner comments show about the same problems
Cambridge publishes Speaking test videos with written examiner comments and marks on its B2 First preparation page. Cambridge does not say what first language these candidates speak, and none of them is presented here as Turkish or Vietnamese. But the comments show exactly how examiners treat the problems on this page.
| Problem | What the examiners wrote | Mark |
|---|---|---|
| A short long turn | "Her response to part 2 is rather short at only 36 seconds rather than a full minute." (B2 First, Natalia) | Discourse Management 3.5 |
| Answering without developing | "With her partner, in parts 3 and 4, her development of the conversation mostly consists of answering his questions, brief agreement, and simple questions." (Natalia) | Interactive Communication 4 |
| Moving on instead of building | In Part 3 he "does not develop his partner's point, but chooses to move on to another point instead", and "does little to negotiate towards an outcome". (B2 First, Enzo) | Interactive Communication 4 |
| th replaced | "She opts to use /d/ and /f/ or /t/ sounds instead of" the th sounds "in words like that, this, or with"; she "could stress the main words in her sentences to avoid losing focus". (B2 First, Tabatha) | Pronunciation 3.5 |
| Staccato rhythm | "His speech can sometimes seem a little staccato, particularly in part 2, however, all sounds are clearly articulated." (C1 Advanced, Antonio) | Pronunciation 4.5 |
Two lessons. First, these are half-band losses, not failures: every candidate above was described as intelligible. Second, the interaction comments are the most useful for both of our groups. A candidate can be fluent and accurate and still sit at 4 for Interactive Communication because the turns answer but do not build. That is the most fixable mark in the test.
The habits at a glance
This table is TalkDrill's judgement from how Turkish and Vietnamese sounds and grammar differ from English. Tick what you hear in your own recording and drill no more than three habits at once.
| # | Habit | Who | Criterion | What the examiner hears | Fix |
|---|---|---|---|---|---|
| 1 | Vowel inserted into a cluster | Turkish | Pronunciation | "sipor", "tirain", "ischool" | Drill 1: build the cluster from the end |
| 2 | w as v | Turkish | Pronunciation | "ve vork vell" | Drill 2: lips round, no teeth |
| 3 | Final consonant devoiced | Turkish | Pronunciation | "bak" for bag, "hat" for had | Lengthen the vowel before a voiced ending |
| 4 | Missing articles; he and she mixed | Turkish | Grammar and Vocabulary | "in first photo", "my sister, he works" | Drill 5: phrases with the article inside |
| 5 | Main point last | Turkish | Discourse Management | Reasons for 40 seconds, the answer at 55 | Drill 6: answer first |
| 6 | Present continuous for habits | Turkish | Grammar and Vocabulary | "I am going to the gym every day" | "I go to the gym every day" |
| 7 | Final consonants dropped | Vietnamese | Pronunciation, Grammar and Vocabulary | "fi" for five, "two photo", "she like" | Drill 3: hold the ending |
| 8 | Clusters reduced or split | Vietnamese | Pronunciation | "stit" or "sơ-treet" for street | Drill 1 |
| 9 | Every syllable stressed | Vietnamese | Pronunciation | Staccato rhythm; no weak "a", "to", "of" | Drill 4: tap the stress, swallow the rest |
| 10 | Double connectors | Vietnamese | Grammar and Vocabulary | "Because it rains, so we stay"; "Although it's cheap, but..." | One connector per sentence |
| 11 | Short turns | Vietnamese, often Turkish | Discourse Management | A long turn of 30 to 40 seconds | Drill 7: the three-part minute |
| 12 | Answering without developing | Both | Interactive Communication | "Yes, I agree." and silence | Drill 8: agree, add, ask |
Turkish speakers
Sounds
- Inserted vowels. Turkish does not start native words with two consonants, which is why "club" became "kulüp" and "tren" (train) is often said "tiren". In English the extra vowel adds a syllable: "sipeak", "tirain", "sipor" for sport. In a long turn about sport and school it can appear in every sentence (drill 1).
- w and v. Turkish has v but no w, so "we", "work", "well" and "why" all start with v. Practise with our v and w guide (drill 2).
- th. Usually t and d ("tink", "dis"), sometimes s and z. Our th guide shows the tongue position.
- Final devoicing. Turkish turns a final b, d or g into p, t or k (compare "kitap" and "kitabı"), so "bag" sounds like "back" and "had" like "hat". English listeners hear the length of the vowel before the ending, so stretch the vowel in "bag" and keep it short in "back".
- Vowels. Short and long vowels merge (ship and sheep), and "bad" and "bed", "man" and "men" can sound the same.
- Stress. Turkish stress usually falls on the last syllable, so English words stressed early can move: chocoLATE, photoGRAPH, interesTING. Learn the English stress with the word.
Grammar and organisation
- Articles. Turkish has none; "bir" means one. So "in first photo", "I am student", "it is good idea", and "one man is sitting" for "a man". Our articles guide has the rules; for the test, learn ten Part 2 phrases with the article inside (drill 5).
- He and she. Turkish "o" covers both, so "my sister... he" slips out under pressure. In Part 1 and Part 4, where you talk about family, it is noticed.
- The present continuous for habits. Spoken Turkish uses the "-iyor" form for routines ("her gün okula gidiyorum"), so "I am going to school every day" sounds right. In English: "I go to school every day".
- Do and make. "Yapmak" covers both: "do a mistake", "make homework". In English: "make a mistake", "do homework".
- The main point last. Turkish puts the verb last and reasons before conclusions, so a Turkish-shaped answer builds up to its point. In Part 2 the question is printed above the photographs, and a minute is short: if the answer to the question arrives at 55 seconds, it may be cut off. Put the answer first, then the reasons (drill 6).
- Fillers. "Yani" (I mean) and "şey" (um, thing) appear under pressure. "I mean" and "well" are English cohesion; "yani" is not.
Vietnamese speakers
Sounds
- Final consonants. Vietnamese allows only a few final consonants and does not release final stops, so English endings vanish: "five" becomes "fi", "nice" becomes "nai", "school" loses its l. Because the endings carry grammar, a dropped s hides a plural and a dropped d hides a past tense, so the habit costs Grammar and Vocabulary as well as Pronunciation (drill 3).
- Clusters. Vietnamese has no consonant clusters, so "street" is either reduced ("stit") or split with a vowel. Final clusters are hardest: "texts", "months", "asked".
- Every syllable stressed. Vietnamese is tonal and most words are one syllable, so each syllable gets full weight and its own pitch. English rhythm depends on stressed syllables standing out and weak ones ("a", "the", "to", "of", "and") shrinking. Without that contrast, speech sounds staccato and the listener cannot find the key word (drill 4). Our connected speech sentences help.
- Pitch. Tones can carry into English as pitch jumps that sound like questions or surprise. English uses pitch for meaning across the sentence: a fall at the end of a statement, a rise on a yes or no question. See our intonation patterns guide.
- s and sh, z and r, n and l. Many northern speakers say s and sh the same ("see" and "she"), and some say an English r close to z; in some northern areas n and l also merge ("night" and "light"). Southern speakers have different patterns. Check which apply to you in your own recording.
- th. Vietnamese "th" is an aspirated t, so "think" becomes "tink" and "the" becomes "de" or "ze".
Grammar and organisation
- Endings. Vietnamese marks time and number with words, not endings, so "Yesterday I go", "two photo", "she like" are grammar habits as much as sound habits. Fix them together: say the ending and mean it.
- Double connectors. "Vì... nên..." becomes "Because it was raining, so we stayed in", and "mặc dù... nhưng..." becomes "Although it's cheap, but it's crowded". English uses one: "Because it was raining, we stayed in" or "It was raining, so we stayed in".
- Missing "be". Vietnamese adjectives work like verbs, so "She very kind" and "The city very crowded" are natural. In English: "She's very kind".
- "Have" for "there is". "Có" means both, so Part 2 openings come out as "In the photo have two people". In English: "In the photo there are two people".
- "Very like". From "rất thích". In English: "I really like it" or "I like it very much".
Both groups: making Part 3 a conversation
Whatever the reason (politeness, shyness, or a classroom habit of waiting to be asked), short answers and few questions in Part 3 read as not initiating. The format pages describe Part 3 as a conversation with the other candidate in which you exchange ideas, agree or disagree, suggest and reach "a decision through negotiation". The B2 First handbook adds that candidates "are not penalised if they fail to reach a negotiated decision": what is marked is how you hold the conversation. Three rules fix most of it.
- Every second turn ends with a question. "What do you think?", "Would you agree?", "Shall we move on to...?" Questions like these are what the handbook glossary means by developing the interaction (its own examples are "What about bringing a camera for the holiday?" and "Why's that?").
- Every reply starts with a link. "That's a good point, and...", "Going back to what you said about...". Enzo's comment shows the cost of moving on without one.
- Disagree once, politely. "I see what you mean, but I'd put... first." In both Turkish and Vietnamese, contradicting someone you have just met can feel rude. In Part 3 it is the task. A polite disagreement followed by a compromise is what "negotiates towards an outcome" sounds like.
One Vietnamese-specific point: "yes" (the "vâng" or "dạ" of acknowledgement) can mean "I am listening". An English listener hears agreement. If you disagree, start with "OK, but..." rather than "Yes, but...". The Part 3 phrases page has the full set of moves.
Also useful if you speak Greek or Polish
Greek and Polish speakers are among the candidates this hub is written for, and they share several habits with the groups above. The short list, also TalkDrill's judgement:
| First language | Watch for | Fix |
|---|---|---|
| Greek | Five vowels, so ship and sheep, full and fool merge; s and sh merge ("see" for she); stress moved on Greek-rooted words (photoGRAPH); "the" before abstract nouns ("The education is important") | Minimal pairs from your own Part 2 lines; check stress in a dictionary; "Education is important". Greek already has both th sounds: use them |
| Polish | Missing articles ("in second picture"); th as f, t or d; final devoicing ("dok" for dog); the letter w read as v; the present for the present perfect ("I know him since five years") | Phrases with the article inside (drill 5); "I've known him for five years"; drill 2 for w |
Eight 60-second drills
All drills are written by TalkDrill. Say the lines once slowly, once at speaking speed, then make one new sentence of your own with the target. Record the last run.
- Clusters from the end (habits 1 and 8). Build backwards so no vowel can slip in: "eet, reet, street"; "ort, port, sport"; "ain, rain, train". Then: "Students in the street are playing sport after school."
- Round lips for w (habit 2). Say "oo", then open into the vowel: "oo-e, we"; "oo-ork, work". Then: "We work well when we want to." Teeth never touch the lip.
- Hold the ending (habit 7). Over-pronounce every final sound for one minute: "five books, two photos, she likes it, he walked, they asked". Then a Part 2 line: "Both photos show people who have travelled a long way."
- Tap the stress (habit 9). Tap the table on the stressed syllable only: "I THINK the FIRST PHOto is MORE reLAXing." Keep "the", "is" and "more" quiet. Then do the same with your own Part 1 answer.
- Article inside the phrase (habit 4). Learn as single units: "the first photo", "the second photo", "a good idea", "the most important thing", "a lot of people", "the other one". Say each three times, then use four in one long turn.
- Answer first (habit 5). For any question: answer in one sentence, then give the reason, then an example. "The violin lesson looks harder. That's because the teacher sees every mistake. For example, she's correcting the girl's hand right now."
- The three-part minute (habit 11). Twenty seconds comparing, twenty answering the printed question, ten summing up, ten spare. Time it. If you finish at 40 seconds, add one more difference.
- Agree, add, ask (habit 12). With the AI as your partner, answer six Part 3 turns in the same shape: "Yes, and... What about...?" or "I see what you mean, but... Do you think...?"
Before and after: two worked examples
Both examples are written by TalkDrill to show the habits and the fixes; neither is a real candidate.
Part 2, a Turkish-shaped long turn: before (written by TalkDrill)
Question above the photos: "What might be difficult about learning in these situations?"
"In first photo there is one girl, she is playing violin and her teacher, he is helping. In second photo there are many students, they are listening the lecture, some of them using phone. The girl, maybe she is nervous because teacher is very near, and in lecture students cannot ask question because there are too many people, so I think both of them difficult but in different ways."
What costs marks: five missing articles, "her teacher, he" for a teacher who could be either, "listening the lecture", and the answer to the printed question arriving only at the end.
After (written by TalkDrill)
"Both photos show learning that could be difficult, but for opposite reasons. In the first photo, a girl is having a violin lesson, and the teacher is standing right next to her, so every mistake is noticed. That might be stressful. In the second photo, it's the opposite: there are hundreds of students, so it would be hard to ask a question, and some of them are already looking at their phones. So I'd say the lesson is difficult because of too much attention, and the lecture because of too little."
Part 3, a Vietnamese-shaped exchange: before (written by TalkDrill)
B: "I think the sports centre is the most important. What do you think?"
A: "Yes. I agree. Because young people need exercise, so it is good."
B: "And the music venue?"
A: "Yes, music venue also good."
What costs marks: "because... so...", missing "is", short turns, and no question back, so B does all the initiating.
After (written by TalkDrill)
B: "I think the sports centre is the most important. What do you think?"
A: "I agree it's useful, because a lot of young people don't do enough exercise. But what about transport? If the buses are bad, nobody can get to the centre anyway."
B: "That's true. And the music venue?"
A: "It's a good idea, although I think fewer people would use it. Shall we choose between the sports centre and transport?"
What changed: one connector per sentence, every ending pronounced, "is" restored, a new idea in each turn, and two questions that move the task towards a decision.
Your practice route
- Record one Part 2 and one Part 3 task from the paired practice bank, with the AI as your partner.
- Tick the habits you hear in the table above. Choose three.
- Ten minutes a day for two weeks: read-aloud practice for the sound habits, role-play practice for the turn-taking ones.
- Re-record the same tasks and score yourself on the four criteria.
- Fit it into a plan. The eight-week B2 First plan has first-language days in weeks 1, 2 and 5, and the Cambridge English hub has the dates, the scale and results timing.
Spanish, Italian or Portuguese speakers should read the Romance-language page instead. For the format of each part, see the official B2 First exam format page.