Duolingo English Test Speaking — Interactive Speaking for 120

It tests whether you can answer a spoken question spontaneously in about 35 seconds and then adapt when a follow-up changes the angle. You hear each question once with no replay and no preparation time. Across 6 to 8 turns the model judges fluency, coherence, task completion, vocabulary range, grammar, and clear pronunciation. This mock mirrors that: a clear opinion in the first two seconds, one concrete example, full use of the window, and a real adjustment when the follow-up pivots rather than repeating your first answer.

Open with a direct answer or clear opinion within the first two seconds, before any thinking-out-loud filler. Give one specific personal example rather than several thin generalities. Add a short reason that connects the example to your opinion, then close a complete thought before the timer ends. The familiar opinion, reason, example, outcome scaffold helps you reach the 25 to 35 second target without rambling. The goal is one developed idea delivered cleanly, not a list of shallow points.

Finishing in roughly ten seconds and leaving the window unused, drifting off the asked question halfway through, pausing longer than two seconds so fluency reads low, reciting a memorised template that does not fit the prompt, and repeating your first answer when the follow-up clearly pivoted. Filler like um and like instead of content, and long dead air at the end, also lower the score. Each of these is a behaviour the AI flags as not showing real spontaneous language ability.

The real DET scorer is automated and silent and gives you a number later. This examiner is a coach in character who debriefs you out loud after each turn, names the specific behaviour the model would read, and gives you one redo on the same prompt. It probes every answer at least once and never praises you mid-session. It will not act like a machine or break character, but it knows exactly how the DET speaking model rewards relevance, time use, and spontaneity.

Interactive Speaking feeds your Speaking individual subscore, and that subscore rolls into the Conversation and Production integrated subscores. The overall is the average of the four individual subscores rounded to the nearest 5. So a weak speaking performance pulls your overall below 120 even if your reading is strong, and admissions teams often read the Speaking and Conversation subscores alongside the overall when a program has a spoken-English bar.

Check your microphone is clear and you are in a quiet room, because pronunciation is scored from the recording. Decide your default opening move: a one-sentence direct answer, then one personal example. Remind yourself the question is heard once, so listen for the exact ask. Plan to start speaking within two seconds of the prompt ending and to keep going until the timer stops rather than quitting at ten seconds.

Treat every follow-up as a new question that builds on the last, not a cue to repeat your first answer. Listen for the pivot word, the part of the new question that is different from the first, and answer that specifically. If the first turn asked your opinion and the follow-up asks for a counterexample or a recommendation, give a fresh example aimed at the new angle. Adapting to the pivot is what separates a 120-level turn from a 100-level one.

It opens with a clear position in the first sentence, supports it with one specific personal example with concrete detail, uses varied connectors like because, for instance, and as a result, keeps a steady pace with no pause over two seconds, and closes a complete thought just as the time runs out. On the follow-up it visibly shifts to the new angle rather than recycling the first answer. It sounds spontaneous and on-topic, not rehearsed.

A 120 maps to roughly IELTS 7.0 and CEFR C1 and sits around the 75th percentile. Most universities treat 115 to 120 as competitive, which is why Indian first-attempt applicants commonly target 120. Selective programs ask 125 to 135, so confirm your specific program's requirement. Because each attempt costs about 59 US dollars, a clean first attempt at 120 with a balanced profile and no weak subscore is the practical goal.

The model scores coherence and task completion, so an answer that drifts off the asked question loses points even when the grammar is correct. A polished general speech that fits no specific prompt is read as memorised and not real language ability. Answering the exact question, developing one idea, and staying on topic until time ends is what protects the speaking subscore and keeps your overall at 120.

Aim to use the full 25 to 35 second window. Quitting at ten seconds caps the score even with correct grammar, and long dead air at the end hurts you too. If you run dry, extend with a second concrete example, a consequence of your opinion, or a brief contrast, rather than stopping. Keep one developed idea moving rather than stacking shallow points, and finish a complete sentence as the timer ends.