The current Speaking section has two task types: seven Listen and Repeat items and four Take an Interview responses, in roughly eight minutes. Practice both accurate repetition and clear, developed answers to questions about experiences and opinions.
Test Ninjas Research · Updated September 14, 2026
Speaking — at a glance
Attribute
Value
Items
11
Raw points
55 of 145 (38%)
Delivery
Linear — same tasks for everyone on a form
Band scale
1.0–6.0 in half-band steps
Weight in overall score
One quarter — all four sections are averaged equally
Practice every Speaking task type with original questions and feedback. Start today’s Speaking practice and return tomorrow for a fresh set.
How the section is built
Composition
Part
Detail
Listen and Repeat
7 items — repeat the heard sentence accurately and intelligibly
Take an Interview
4 items — answer questions about experiences and opinions
Estimated time
About 8 minutes
Scoring
AI-scored — natural language processing assesses fluency, coherence and grammar
Format
Linear — everyone on a given form receives the same tasks
Responding to a question without lengthy preparation time
Speaking in academic navigational contexts
Maintaining fluency under a short clock
Organizing a spoken answer without writing it first
Speaking clearly enough for automated assessment
Repeating a sentence without dropping or changing important words
Sustaining coherence across a short response
Recovering from a stumble without losing the thread
How to approach it
Eleven items in eight minutes changes the game
The old section gave you a handful of tasks with structured preparation time. Eleven items in roughly eight minutes means answers are short and the pace is brisk. Rehearsed templates that worked on the old format do not fit, and long preambles cost you a disproportionate share of a short response.
It is AI-scored, which rewards clarity over flourish
Responses are assessed with natural language processing for features including fluency, coherence and grammar. That is not a reason to sound robotic, but it does mean clear structure and steady delivery are worth more than ambitious vocabulary delivered hesitantly.
Speaking carries the most raw points
At 55 raw points from 11 items, Speaking has by far the highest points-per-item weighting of any section. It is a quarter of your overall band like every other section, but each individual response counts for a great deal within it.
Speaking contributes one quarter of your overall band, the same as every other section. See what a change here does to your overall score with the TOEFL score calculator.
Most also require the GRE, which ETS administers alongside the TOEFL. Our GRE course covers Verbal and Quant in full, with adaptive practice tests and instant scoring.
About 8 minutes for 11 items: seven Listen and Repeat items and four Take an Interview responses. The tasks differ from the independent and integrated tasks in the older format.
It is AI-scored. ETS uses natural language processing and machine learning to assess constructed responses on features such as fluency, coherence and grammar. Speaking and the two written responses are the only AI-scored parts of the exam; everything else is machine-scored against predefined logic.
No. Speaking and Writing are linear — every test taker on a given form receives the same set of tasks. Only Reading and Listening use the two-stage adaptive design.
Your responses produce a Speaking band from 1.0 to 6.0 in half-band steps, which is then one of the four bands averaged into your overall score. A band 5 indicates CEFR C1 proficiency for the skill.