DELE B2 Oral Exam: What Examiners Are Actually Writing Down While You Talk

Nobody tells you what the examiner's pen is doing

You finish your two-minute monologue. There's a polite nod. The examiner writes something. You have no idea if that something is 'impressive range of connectors' or 'struggled to maintain register.' That silence is where most DELE B2 candidates lose marks — not in the room, but in the months before it, when they prepared for the wrong things.

The DELE B2 oral exam (Prueba 4: Expresión e Interacción Orales) is scored on a published rubric. It has five criteria. Each criterion has descriptors. Most prep guides mention this in passing, then give you a list of 'useful phrases' and call it a day. This post is about the rubric itself — what each criterion actually means in practice, what a strong response looks like versus a weak one, and why some very fluent speakers still fail.

The structure of the exam in 90 seconds

Before the scoring, let's be precise about what you're being scored on. The DELE B2 oral exam has three tasks, totalling around 20 minutes (including 6–7 minutes of preparation time beforehand).

• Tarea 1 — Valoración de una propuesta (approx. 6–8 minutes): You're given a written proposal — a hypothetical project, initiative, or plan — and asked to discuss it with the examiner. You evaluate it, defend a position, negotiate.
• Tarea 2 — Descripción e interpretación de datos (approx. 5–6 minutes): You receive a graph, chart, or image and must describe it, interpret trends, and speculate about causes and implications.
• Tarea 3 — Dar tu opinión sobre datos de una encuesta (approx. 4–5 minutes): You react to survey results related to the Tarea 1 topic, compare them to your own experience or context, and discuss with the examiner.

The five scoring criteria — and what they actually measure

The Instituto Cervantes scores Tarea 1 and Tarea 2 separately. Each is marked on five criteria, each worth a maximum of 4 points, giving a total of 40 points across both tasks. Tarea 3 is scored as part of the overall interaction. Here's what the five criteria really mean — translated out of bureaucratic language.

• 1. Coherencia y cohesión del discurso — Can you build an argument that flows? This isn't about using connectors like 'sin embargo' performatively. It's whether your ideas actually connect logically. A response that jumps from point to point with no thread scores low here even if the individual sentences are grammatically perfect.
• 2. Adecuación sociolingüística — Are you speaking to an examiner, not a friend, not a textbook? Formal register where it's expected, appropriate hedging when speculating, neutral tone when presenting data. This is where very informal speakers and overly rigid textbook-speakers both lose marks.
• 3. Amplitud y control de los recursos léxicos — Vocabulary range and precision. Not just 'big words.' Using 'incrementarse notablemente' instead of 'subir mucho' shows control. Using 'se podría deber a' instead of 'es porque' shows range.
• 4. Amplitud y control de los recursos gramaticales — Grammar range and accuracy. B2 means you're expected to use the subjunctive naturally, conditionals correctly, relative clauses without anxiety. Errors that don't impede communication are tolerated. Systematic errors in basic structures are not.
• 5. Fluidez — Not speed. Absence of disruptive pauses, filler overload ('ehhh... pues... como... no sé'), and the ability to self-correct without completely losing the thread.

The criterion most people ignore: coherencia

Coherencia is the criterion that separates candidates who've genuinely reached B2 from those who've memorised enough phrases to sound like B2 in short bursts. The examiner isn't just listening to whether your grammar is right. They're tracking: does this person actually have a position? Are they developing it or just cycling through rehearsed sentences?

A common failure pattern looks like this: candidate gives an opinion, then gives the opposite opinion, then circles back with 'bueno, depende de la situación.' That's not balanced argumentation — it's intellectual evasion, and examiners recognise it immediately. Strong coherencia means you stake a claim, support it with at least one concrete reason, acknowledge a counterargument, and then either defend your original position or revise it explicitly.

Sample response: Tarea 2 (graph description) — weak vs. strong

The topic: a bar chart showing the percentage of young people (18–30) who use social media as their primary news source, across five countries, comparing 2015 and 2023.

WEAK RESPONSE: 'En este gráfico podemos ver información sobre las redes sociales y los jóvenes. En 2015 el número era más bajo. En 2023 el número es más alto en todos los países. Creo que las redes sociales son muy populares ahora. Es interesante.'

Why it fails: It describes without interpreting. It uses vague language ('más alto,' 'más bajo,' 'número'). It offers an observation disguised as an opinion. Fluidez might be fine, but léxico and coherencia both score poorly.

STRONG RESPONSE: 'Lo que más llama la atención en este gráfico es el incremento generalizado entre 2015 y 2023, aunque con diferencias notables entre países. En el caso de España, el porcentaje prácticamente se ha duplicado, lo cual sugiere un cambio estructural en los hábitos informativos de los jóvenes, no solo una tendencia pasajera. Cabe preguntarse si esto refleja una pérdida de confianza en los medios tradicionales o simplemente una cuestión de accesibilidad. En mi opinión, probablemente sea una combinación de ambos factores, aunque el segundo me parece más determinante en contextos con buena cobertura mediática.'

Why it works: It identifies the most significant trend. It uses precise vocabulary ('prácticamente se ha duplicado,' 'cambio estructural,' 'tendencia pasajera'). It speculates with appropriate hedging ('cabe preguntarse,' 'probablemente sea'). It takes a position and briefly defends it. All five criteria are being fed simultaneously.

Sample response: Tarea 1 (negotiation) — the move most candidates miss

Tarea 1 is a dialogue, not a monologue. The examiner will push back on your position. This is intentional. They want to see interacción — not whether you can defend your original answer forever, but whether you can engage with a counterargument without either immediately caving or just repeating yourself louder.

The scenario: A proposal to ban single-use plastics in school canteens. You've been asked to evaluate it.

EXAMINER (playing a sceptic role): 'Pero, ¿no cree usted que esto supondría un coste adicional demasiado elevado para los centros educativos?'

WEAK RESPONSE: 'Sí, tiene razón, es verdad que es caro. Pero es importante para el medioambiente.' (Capitulation + repeated assertion. No development.)

STRONG RESPONSE: 'Es un punto completamente válido, y no sería honesto ignorarlo. Sin embargo, creo que habría que distinguir entre el coste inicial y el coste a largo plazo. Muchos centros que han implementado medidas similares han conseguido reducir gastos al año siguiente gracias a contratos con proveedores locales. Además, si lo enmarcamos como una medida educativa — no solo logística — podría acceder a financiación específica para proyectos medioambientales. Así que más que descartarlo por el coste, propondría explorar modelos de financiación antes de tomar una decisión.'

Why it works: It acknowledges the counterargument genuinely. It introduces a distinction (inicial vs. largo plazo). It offers a concrete alternative framing. It moves the conversation forward rather than defending a trench. This is B2-level interacción.

Register: the silent killer of high scores

Adecuación sociolingüística is the criterion that catches people who have spent most of their Spanish life talking to friends, watching Netflix, or practising with conversation partners in informal contexts. The DELE B2 oral is semi-formal. That doesn't mean stiff. It means consistent.

Watch out for these register mismatches:

• Starting sentences with 'O sea...' or 'La verdad es que...' every few turns — fine occasionally, problematic as a crutch.
• Using 'tío/tía,' 'mogollón,' 'guay' (for Spain-variant speakers) — these are real Spanish, but wrong register here.
• Overcorrecting into stiff formal language that sounds rehearsed: 'En lo que respecta a la cuestión que nos atañe...' before every single point. Examiners notice the performance.
• Hedging too much when speculating — 'No sé, quizás, a lo mejor podría ser que...' — this hurts both register and fluidez scores.
• The target is the register of a thoughtful adult having a serious conversation. Confident, clear, occasionally direct, with appropriate politeness markers.

What fluidez actually looks like in the rubric

The fluidez descriptor at B2 doesn't demand zero pauses. It asks that pauses don't impede communication and that you can maintain discourse at length without frequent searching. The key phrase in the rubric is 'sin excesivas interrupciones o dudas que dificulten la comprensión.'

The practical implication: a one-second pause to find the right word is fine. A five-second pause followed by 'no sé cómo se dice' and a shrug is a fluency event. What examiners mark down is the pattern — candidates who pause disruptively every three or four words are signalling that they're operating at the ceiling of their real level, whatever phrases they've memorised.

This is also why practising your monologue until it's smooth is less useful than practising recovery — what you do when you lose the thread mid-sentence. Strong B2 candidates say things like 'es decir,' 'o dicho de otro modo,' 'lo que quiero decir es que' to buy half a second and relaunch, rather than trailing off or switching to English internally and then stalling visibly.

Grammar: what B2 actually requires you to control

The grammar criterion at B2 isn't about being error-free. It's about range and accuracy together. An examiner seeing only present tense and basic past tense — even if perfect — will score lower than a candidate who uses conditionals, relative clauses, and the subjunctive with occasional minor errors.

The structures that most clearly signal B2 grammar competence in the oral exam:

• Subjunctive in opinion and doubt clauses: 'No creo que eso sea suficiente para...' — examiners notice this.
• Conditional structures for hypothesising: 'Si se implementara esta medida, es probable que los resultados fueran distintos.'
• Relative clauses with prepositions: 'El sector al que más afecta esta política es...'
• Passive and impersonal constructions for formal register: 'Se ha demostrado que,' 'Cabe destacar que.'
• Nominalisation to add density: 'El incremento del uso' instead of 'que la gente usa más.'

The preparation mistake that costs the most marks

It's memorising responses. Seriously. The DELE B2 oral exam gives you 6–7 minutes of preparation time before you speak. The temptation — especially if you've been studying set phrases and model answers — is to use that time to script exactly what you'll say and then deliver it from memory.

Examiners are trained to detect this. A scripted delivery has a rhythm that's different from real-time thinking. When the examiner then pushes back — and they will — the candidate who scripted their opening is suddenly in unscripted territory with no practice being there. That's where the wheels come off.

Use preparation time to sketch a structure, not a script: your main position, two supporting points, one counterargument you can address. Then talk. The fluency that comes from real-time thinking, even with minor errors, scores higher than polished delivery that collapses under a single follow-up question.

Know your real level before exam day

One underrated reason candidates underperform in the oral is that they sit the exam at a level that doesn't actually match where they are. If you're genuinely at B1+, the B2 oral will expose that — not because you can't use a subjunctive, but because the cognitive load of producing extended discourse on abstract topics at pace is different at every level.

If you're not sure whether you're really at B2 or still consolidating, Nivelo's free 5-minute CEFR test gives you an honest read. It's not a streak counter or a confidence metric — it's calibrated to the actual CEFR scale, the same one DELE is built on. Worth doing before you commit to an exam date.

Two weeks out: what to actually practise

With two weeks to go, targeted oral practice beats general conversation practice. Here's how to make that practice count:

• Record yourself doing a full Tarea 2 (graph description + interpretation) using a real DELE B2 sample task. Listen back specifically for: did I interpret or just describe? Did I speculate with appropriate hedging? Count your disruptive pauses.
• Practice the pushback moment: get a study partner or tutor to challenge one of your positions with a counterargument. Your goal is to acknowledge it genuinely, then develop your response — not repeat yourself.
• Read one editorial in Spanish per day (El País, El Mundo, La Vanguardia). Note how writers express opinions, concede points, and hedge. This is the register you're aiming for.
• Drill self-correction phrases: 'Perdona, quería decir...', 'En realidad lo que pretendía señalar es...', 'Déjame reformular eso.' These aren't weaknesses. They're fluency tools.
• Time yourself. A Tarea 2 monologue should be 2–3 minutes of substantive content. Too short means you're under-developing. Too long means you're padding. Both hurt your coherencia score.

The examiner is not your enemy

This sounds obvious. But a lot of candidates enter the room in a defensive posture — trying not to make mistakes, answering as quickly as possible to minimise exposure, treating the examiner's questions as traps. That posture produces short answers, minimal elaboration, and low scores across almost every criterion.

The examiner's job is to give you opportunities to demonstrate B2. Their questions are prompts, not interrogations. When they push back, they're not trying to catch you — they're opening space for you to show interacción. When they ask 'and why do you think that is?', they're inviting you to score points on coherencia and léxico simultaneously.

Treat it like a conversation with a thoughtful, senior colleague about a topic you've had time to think about. That's the room the rubric was written for.