7.4 The Speaking Sample: Graded Long-Form Speaking That Institutions Also Watch

Key Takeaways

  • The Speaking Sample has been graded since 1 May 2023: its official page lists the subscores Speaking, Conversation, and Production, and Duolingo groups it with the open-response types evaluated by the grading engine.
  • The directions appear for 30 seconds so you can read and consider the written prompt, and the official duration of the response is 3:00.
  • The recording is also sent to every institution that receives your certificate, so it is simultaneously a scored item and a piece of your application.
  • Duolingo removed the minimum response time for this task on 1 July 2025, so pacing is entirely self-imposed.
  • It appears once per test, and is one of only four question types that feed the Speaking subscore at all.
Last updated: September 2026

7.4 The Speaking Sample: Graded Long-Form Speaking That Institutions Also Watch

Quick Summary: The Speaking Sample asks you to "speak about the topic below." You read a written prompt while the directions are displayed for 30 seconds, then record a response with an official duration of 3:00. Two facts define it: it is graded — its official subscores are Speaking, Conversation, and Productionand the recording is sent to every institution that receives your results. It is scored by the engine and watched by people, which is why it deserves more preparation than any other single item on the test.

[!CAUTION] The Speaking Sample is not ungraded. This is the single most persistent piece of misinformation in DET preparation, and it appears in guides, forum posts, and video walkthroughs that were accurate years ago and have never been updated. Duolingo began grading the Speaking Sample on 1 May 2023. Its current "How the Items Are Scored" article lists the Speaking Sample among the open response question types whose answers are "evaluated by a grading engine," and its official task page lists the subscores plainly: Speaking | Conversation | Production. A candidate who treats this task as a throwaway is discarding one of only four items on the entire test that feed the Speaking subscore.


The Dual Function: Scored Item and Application Material

The Speaking Sample does two jobs at once, and preparing for only one of them is how candidates get caught out.

+-------------------------------------------------------------------------+
|                  THE SPEAKING SAMPLE: OFFICIAL SPEC                     |
+-----------------------+-------------------------------------------------+
| Official Directions   | "Speak about the topic below."                  |
| Prompt Format         | A written prompt, displayed on screen            |
| Directions Window     | 30 seconds to read and consider the prompt      |
| Official Duration     | 3:00                                            |
| Official Frequency    | 1 item per test                                 |
| Official Subscores    | Speaking | Conversation | Production            |
| Graded?               | YES - graded since 1 May 2023                   |
| Minimum Response Time | NONE - removed 1 July 2025                      |
| Also Shared?          | Yes - the recording goes to score recipients    |
+-----------------------+-------------------------------------------------+

Job One: It Moves Three of Your Eight Numbers

Only four question types feed the Speaking individual subscore — Speak About the Photo, Read Then Speak, Interactive Speaking, and this one. Because Conversation averages Listening and Speaking, and Production averages Writing and Speaking, anything that moves Speaking moves three reported scores at once. The Speaking Sample is also the longest speaking opportunity on the test at 3:00, which means it is the single item where you have the most room to demonstrate developed content, organisation, and grammatical range.

Job Two: Institutions Watch It

The recording is transmitted to every institution that receives your certificate. Duolingo's parallel advice on the Writing Sample is instructive here — it tells test takers that "the topic will be shared with the recipients of your results, so make sure your response is relevant to the topic" and that recipients "may use the Writing Sample as additional information about your writing skills." The Speaking Sample occupies the same position for spoken English.

What that means practically:

  1. Relevance is doubly important. An off-topic response is both a content problem for the engine and a strange artefact for a human reviewer who can see the prompt next to your answer.
  2. Presentation matters in a way it does not elsewhere. Nobody watches your Speak About the Photo recording. Somebody may well watch this one.
  3. It corroborates your written application. A fluent, organised spoken response supports the writing in your application; a very short or disengaged one raises questions it did not need to raise.

What the Two Audiences Look For

The grading engine and a human reviewer are looking at overlapping but not identical things. The engine assesses content, coherence, vocabulary and grammar, and fluency and pronunciation. A human reviewer additionally forms an impression of:

  1. Consistency with the written application: Does this candidate's spontaneous spoken English sit plausibly alongside the vocabulary and structure of their personal statement?
  2. Classroom Seminar Readiness: Can this student engage spontaneously in dynamic undergraduate seminars, articulate independent viewpoints, and defend intellectual positions without a scripted teleprompter?
  3. Maturity, Poise, and Resilience: How does the student carry themselves under pressure? Do they present themselves with warmth, intellectual curiosity, and self-possession, or do they appear panicked, disengaged, or robotic?
  4. Identity and integrity: The recording is one more point of correspondence between the person who took the test and the person applying.

Using the 30-Second Directions Window

Duolingo's description of this window is specific: "The directions appear for 30 seconds before recording begins. During this time you will read and consider the prompt, and prepare to respond." You are reading the prompt — the official page describes a single written prompt, not a menu of topics. Spend the window planning, not choosing.

+-------------------------------------------------------------------------+
|                THE 30-SECOND PLANNING PROTOCOL                          |
+-------------------------------------------------------------------------+
|                                                                         |
|   Seconds 00-08: READ AND PARSE                                         |
|   - Read the prompt fully. Identify every sub-question it contains.     |
|   - Decide your position in one sentence.                               |
|                                                                         |
|   Seconds 08-20: CHOOSE ONE CONCRETE EXAMPLE                            |
|   - Pick a single specific experience, event, or case you can describe  |
|     in detail. Specificity is what carries a 3-minute response.         |
|                                                                         |
|   Seconds 20-30: FIX THE OPENING AND THE ENDING                         |
|   - Rehearse your first sentence silently so it comes out cleanly.      |
|   - Decide the note you will finish on, so the ending is not improvised.|
|                                                                         |
+-------------------------------------------------------------------------+

[!WARNING] Do not spend this window deliberating about whether you like the topic. Duolingo scores relevance and development, not topic preference. The most expensive thing you can do with these 30 seconds is reach the record screen without a first sentence and without an example, because those are exactly the two things you cannot invent while talking.


Physical Demands & Video Presence: The Digital Interview Studio

Because the recording is sent to institutions, your delivery and demeanour carry weight beyond the score itself. Treat the webcam as an interviewer sitting across a table — while still observing every proctoring rule, which the guidance below is designed to be compatible with.

+-------------------------------------------------------------------------+
|               EXECUTIVE VIDEO PRESENCE: BEST PRACTICES                  |
+-----------------------+-------------------------------------------------+
| Camera Eye Contact    | Look directly into the webcam lens, NOT down at |
|                       | your own image or around the physical room.     |
+-----------------------+-------------------------------------------------+
| Framing & Posture     | Sit upright with shoulders squared; position    |
|                       | yourself 18-24 inches away so head and upper    |
|                       | chest are fully centered in frame.              |
+-----------------------+-------------------------------------------------+
| Facial Expressiveness | Smile naturally during opening and closing; use |
|                       | active brow engagement to convey enthusiasm.    |
+-----------------------+-------------------------------------------------+
| Vocal Projection      | Speak with diaphragmatic support at a confident |
|                       | presentation volume; articulate final consonants|
+-----------------------+-------------------------------------------------+
| Hand Gestures         | Keep gestures natural and controlled within the |
|                       | frame; avoid fidgeting or touching your face.   |
+-----------------------+-------------------------------------------------+

The "Monitor Gaze" Pitfall

Most candidates watch their own face on screen while speaking. To a viewer, that reads as looking slightly downward and away. Address the camera lens instead.

[!NOTE] This does not conflict with the proctoring rules, provided you do it the right way. The published rule is "don't look away from the screen, except when typing," and on a laptop the lens sits directly above the display — so looking at the lens keeps your face square to the screen. What you must not do is turn your head, look at the ceiling to think, or glance around the room. Eyes forward, head still, gaze at the lens rather than at your own thumbnail.


The 4-Stage Admissions Monologue Architecture

To construct an impactful 2-to-3 minute oral presentation, execute the 4-Stage Admissions Monologue Architecture:

[Stage 1: 00-30s] The Narrative Hook & Intellectual Stance
       |
       v
[Stage 2: 30-90s] Concrete Episodic Exposition (The Challenge/Story)
       |
       v
[Stage 3: 90-135s] Philosophical Reflection & Personal Transformation
       |
       v
[Stage 4: 135-165s] Forward-Looking Academic Horizon (University Fit)

Stage 1: The Narrative Hook & Intellectual Stance (Seconds 00–30)

Open with an authentic, engaging thesis that establishes both your perspective and a personal hook. Avoid formulaic openings like "Hello admissions committee, today I am going to talk about prompt number one." Instead, dive directly into the core theme:

  • "When contemplating the balance between preserving historical tradition and embracing modernization, I am constantly drawn back to my hometown's debate over preserving our centuries-old municipal canal system."

Stage 2: Concrete Episodic Exposition (Seconds 30–90)

Anchor your reflection in a detailed, lived experience. Generalizations ("I worked hard on many projects and learned a lot") are forgettable; specific narratives ("During our high school robotics championship, our primary pneumatic intake failed thirty minutes before our qualifying match") are unforgettable. Describe:

  • The specific dilemma or tension encountered.
  • The proactive intellectual or collaborative steps you undertook.
  • The immediate outcome of those actions.

Stage 3: Philosophical Reflection & Personal Transformation (Seconds 90–135)

Elevate the narrative from a simple story into evidence of intellectual maturity. Answer the crucial reflective questions:

  • How did this experience challenge your prior assumptions?
  • What did it teach you about intellectual humility, teamwork, or resilience?
  • How did your methodology for solving complex problems evolve?

Stage 4: Forward-Looking Academic Horizon (Seconds 135–165)

Conclude by connecting your insights to your future undergraduate journey. Demonstrate how the habits of mind developed through this experience will contribute to the campus community:

  • "As I prepare to embark on my undergraduate studies in biomedical engineering, I plan to carry this iterative approach into laboratory research—welcoming unexpected setbacks not as failures, but as the essential catalyst for scientific discovery."

Duration: You May Now Stop Early, and You Should Not

On 1 July 2025 Duolingo removed the minimum response time from five question types, the Speaking Sample among them. Nothing now locks the submit control. That makes this the one task where an anxious candidate can quietly discard the most valuable minute of speaking on the whole test.

Think about it in terms of what the engine is asked to judge. It assesses whether your response is relevant and well developed, whether it is organised and easy to follow, and whether you use a wide range of appropriate language. Those are the criteria of an extended answer.

Response lengthWhat it can demonstrate
~60 secondsA position and one thin reason. Almost no development; little organisational structure to follow.
~120 secondsA position, a developed example, and a conclusion. Adequate, but leaves range unshown.
~150 secondsPosition, developed example, reflection, and a considered close — with room for subordination, precise vocabulary, and natural pacing.
Right up to 3:00Risks being cut mid-clause, which undercuts coherence and grammar at exactly the wrong moment.

Aim for roughly 2:15 to 2:45, finish your final thought, and submit deliberately with a few seconds in hand. The goal is not to fill the clock — remember that fluency scoring penalises repetition and filler — but to use enough of it that there is something substantial to evaluate.


Holistic University Admissions Rubric

Admissions deans evaluate your video across four primary qualitative dimensions:

DimensionExemplary (Ivy / Top-Tier Standard)Competent (Standard Admissible)Deficient (Admissions Concern)
1. Spoken Authenticity & SpontaneityNatural, fluid monologue; authentic thought progression; zero signs of canned memorizationGenerally natural delivery; occasional formulaic phrasing but responds genuinelyStiff, robotic recitation of pre-memorized scripts; disconnect from prompt
2. Narrative Depth & MaturityDeep self-reflection; acknowledges nuances and mistakes; articulates clear personal growthClear story with standard moral lesson; limited analytical complexitySuperficial listing of accomplishments; boastful or childish narrative
3. Lexical & Syntactic CommandRich academic vocabulary deployed effortlessly; complex sentence boundaries; precise idiomsSolid conversational vocabulary; mostly correct grammar with minor slipsRepetitive elementary phrasing; frequent grammatical breakdowns
4. Executive Presence & DemeanorDirect camera eye contact; open, confident posture; engaging facial energy and voice projectionAcceptable eye contact; slight nervous fidgeting; clear audio volumeLooking downward at screen; slumped posture; flat monotone; mumbling

Full Exemplar Spoken Transcript & Admissions Walkthrough

Consider this authentic admissions-grade response delivered to a top-tier university:

The Prompt Chosen

"Describe a significant intellectual challenge or personal project that did not go according to plan. What unexpected obstacles arose, how did you respond, and what did this experience reveal about your character?"

Model Monologue Transcript (Duration: ~2 Minutes 35 Seconds | Word Count: 335 Words)

*"When reflecting on intellectual challenges that completely upended my expectations, my mind immediately turns to my junior year science fair project, where I attempted to build an inexpensive water filtration prototype using locally sourced biochar.

Going into the project, I was admittedly overconfident. I had read several academic papers and assumed that translating published chemical formulas into a physical apparatus would be straightforward. However, during our initial trials two weeks before the regional exhibition, the gravitational flow rate was so catastrophically slow that water backed up and ruptured our primary acrylic column, flooding our school's biology prep room.

In that moment of intense panic, my initial instinct was to abandon the custom filter housing and purchase standard commercial tubing. But after stepping back and analyzing the failure with my chemistry teacher, I realized that the core problem wasn't the biochar material itself, but my flawed grain-size distribution, which was causing severe particulate compaction under pressure. Over the next ten days, I spent every afternoon after school sieving biochar into graded particle diameters, redesigning the pressure gradient, and rebuilding the acrylic housing with reinforced silicone gaskets. When we retested the unit forty-eight hours before the fair, the filtration system operated flawlessly, purifying our synthetic runoff samples within established safety thresholds.

That experience was deeply humbling, and it fundamentally reshaped my intellectual approach. It taught me that real engineering is rarely a linear progression from textbook theory to finished prototype; rather, it is a messy, iterative process that demands emotional resilience, empirical patience, and the humility to diagnose your own mistakes without defensiveness.

As I look forward to joining a collaborative university campus, I am eager to bring this exact persistence to undergraduate laboratory research—embracing unexpected experimental failures not as roadblocks, but as the foundational stepping stones toward true scientific innovation. Thank you."*

Admissions Dean Evaluation

  • Authenticity & Vulnerability: The applicant admits overconfidence and describes a genuine failure (flooding the biology prep room) rather than offering a disguised brag.
  • Problem-Solving Rigor: Explains the technical cause (particulate compaction, grain-size distribution) and the methodical engineering solution (sieving into graded diameters, redesigning pressure gradient).
  • Campus Value Proposition: Concludes with an inspiring philosophical synthesis connecting directly to undergraduate laboratory contributions.

High-Frequency Candidate Blunders

Ensure you eliminate these four recurring errors when recording your Speaking Sample:

  1. The Script Recitation Blunder: Attempting to recite a memorized personal statement. Admissions officers review thousands of applications; they instantly recognize when a student is reciting memorized prose. The delivery sounds mechanical, the eyes glaze over, and pauses occur at unnatural rhythmic intervals. Speak organically from mental bullet points.
  2. The Early Departure: Submitting after a minute because the option is now there. Since July 2025 nothing stops you, which is precisely the problem — you are discarding the longest speaking opportunity on the test, on an item that feeds Speaking, Conversation, and Production and is also watched by institutions. Sustain the response past the two-minute mark.
  3. The "Humorous Casual" Trap: Treating the recording like an informal TikTok or social media video. While warmth and authenticity are prized, the register must remain professionally academic. Avoid slang, casual profanity, or inappropriate self-deprecation.
  4. Failing to Wrap Up Naturally: Stalling at second 140 and staring into the lens in awkward silence. When you reach your concluding thought, deliver your final sentence, offer a warm smile, say a polite "Thank you for your time," and deliberately click the submit button.
Test Your Knowledge

Which statement correctly describes how the Speaking Sample is treated on the current Duolingo English Test?

A
B
C
D
Test Your Knowledge

How should a candidate use the 30-second directions window that precedes the Speaking Sample recording?

A
B
C
D
Test Your Knowledge

The Speaking Sample has an official duration of 3:00 and, since July 2025, no minimum response time. What is a sensible target length?

A
B
C
D