Duolingo English Test Speaking — Speak About the Topic for 130

It tests whether you can produce an organised, extended spoken description of an image under time pressure. After 20 seconds of planning you speak for up to 90 seconds. The Duolingo English Test scores it on task relevance, fluency and pace, grammatical range and accuracy, lexical range, and pronunciation. A 130 needs an answer that covers the whole image systematically, uses colour and location detail, runs close to the full time, and never describes something that is not in the picture.

Use a repeatable arc. In the first ten to fifteen seconds give a one-sentence overview of the whole scene. Then spend the bulk of the time on the foreground, naming people and objects with their colour and position and describing actions with present-continuous verbs. Move to the background and setting. Close with one inference about the situation or mood, framed as likelihood. The arc keeps you on-task and stops you running out of ideas at forty seconds.

The DET is scored from 10 to 160 in five-point bands. A 130 sits at the C1 threshold on the CEFR scale and is the practical competitive bar for top universities and many graduate programs. For Indian study-abroad applicants the test is attractive because it costs roughly 59 to 70 US dollars, is taken at home in about an hour, and returns results within 48 hours, far faster and cheaper than IELTS or TOEFL.

Stopping after thirty or forty seconds instead of using the full ninety, reciting a memorised intro that does not match the actual image, listing isolated nouns with no full sentences, rambling with no main point, leaving pauses longer than two seconds, speaking so fast that pronunciation blurs, and describing people or objects that are not in the picture. Each of these caps the speaking subscore and pulls the overall below 130.

The real DET uses automated AI scoring with no human in the loop. This rehearsal puts a former speaking rater in front of you who reacts the way the scoring model behaves: rewarding full-time coverage, structure, range and clarity, and penalising short, off-image, or memorised answers. You get a spoken debrief and a transcript-backed scorecard instead of a single number, so you can hear exactly where an answer would have lost band.

Since July 2024 the DET reports eight subscores. Your overall is the average of the four individual subscores, including Speaking, so a weak speaking performance caps a 130 even if reading and writing are strong. Speaking also feeds two integrated subscores: Conversation, which is Speaking plus Listening, and Production, which is Writing plus Speaking. Many universities read these integrated subscores for program fit, so one short answer can drag three numbers down at once.

Do not write a script. Scan the image and pick three describable zones: the main subject, the foreground detail, and the background. Note one colour and one action you will mention in each zone, and one closing inference. That gives you a route through the full ninety seconds so you never freeze mid-answer. The goal of planning is a path, not a paragraph.

Add detail, not filler. Every object gets a colour and a location, every person gets an action in present continuous, and the setting gets weather, time of day, and atmosphere. Then close with a speculation about who the people are or why the scene matters. Task relevance is scored, so describing more of the real image is safe while inventing extra content or repeating yourself is not.

Open with a clear position or main point in one sentence, give two or three specific reasons or examples with concrete detail rather than abstractions, link them with connectors so the answer progresses rather than jumps, and close by restating your point. Use the full time, vary sentence structure between simple, compound and complex, and keep a steady pace so pronunciation stays clear.

Do not stop and do not start over. Paraphrase around the missing word and keep moving, because a pause longer than two seconds signals the scoring model that you are stuck and lowers fluency. If you say a wrong word, self-correct in one short phrase and continue. Raters reward a candidate who repairs smoothly far more than one who freezes in search of the perfect word.

The scoring model checks that your response shows real language ability tied to the actual prompt. A generic opening like in this photo we can see various things, or a script that ignores what is really in the image, triggers the off-topic and memorised-sounding penalty. A flexible structure you adapt to each specific image is safe; a fixed paragraph you repeat regardless of the picture is not.