📖 Reading time: approximately 8 minutes
What You'll Learn
✅ Key changes in the 2026 TOEFL reform ✅ The new scoring system (1-6 scale) explained ✅ What MST adaptive testing means for you ✅ New question types with real examples ✅ How reliable is the AI scoring system?
Introduction: Insights from Official ETS Documentation
In October 2025, ETS released the TOEFL iBT Technical Manual (RR-25-12), the official technical specification document for the reformed TOEFL. This 68-page document details:
- Test structure and time allocation
- Scoring criteria and score interpretation
- Question types and measured abilities
- Reliability and validity data
This article presents the most test-prep-relevant information extracted from this official document.
2026 TOEFL Test Structure
The new TOEFL maintains the four traditional sections: Reading, Listening, Writing, and Speaking.
The biggest change is the introduction of MST (Multistage Adaptive Testing) in the Reading and Listening sections.
What is MST Adaptive Testing?
MST divides Reading and Listening into two stages:
| Stage | Description |
|---|---|
| Stage 1 (Router Module) | All test-takers answer the same medium-difficulty questions |
| Stage 2 (Upper/Lower Module) | Based on Stage 1 performance, the system assigns harder or easier question sets |
Key point: Scoring considers both the number of correct answers AND the difficulty level of questions answered. Even if you're routed to easier questions, your score is statistically calibrated for fairness. "Getting easy questions means you can't score high" is a myth.
Questions Per Section
| Section | Scored Questions | Score Range | Test Type |
|---|---|---|---|
| Reading | 35 questions | 1-6 points | MST adaptive |
| Listening | 35 questions | 1-6 points | MST adaptive |
| Writing | 12 questions | 1-6 points | Fixed format |
| Speaking | 11 questions | 1-6 points | Fixed format |
Note: Reading (up to 15 questions) and Listening (up to 12 questions) may include unscored tryout questions.
New Scoring System: The 1-6 Scale
Score Range
- Section scores: 1-6 points (in 0.5 increments, e.g., 3.5, 4.5)
- Total score: Average of four sections (also 1-6 points)
TOEFL Score to CEFR Mapping
The new TOEFL scores directly correspond to CEFR (Common European Framework of Reference) levels:
| CEFR Level | TOEFL Score | Proficiency Description |
|---|---|---|
| C2 | 6 | Mastery |
| C1 | 5 - 5.5 | Effective Operational Proficiency |
| B2 | 4 - 4.5 | Vantage |
| B1 | 3 - 3.5 | Threshold |
| A2 | 2 - 2.5 | Waystage |
| A1 | 1 - 1.5 | Breakthrough |
Application benchmarks:
- Most US undergraduate programs: B2 or above (4+)
- Graduate programs: C1 (5+)
- Elite institutions: 5.5-6 points
New Question Types: A Complete Breakdown
Reading Section
| Question Type | Skill Measured | Description |
|---|---|---|
| Complete the Words | Vocabulary + Grammar + Spelling | Fill in the missing second half of words (C-test format) |
| Read in Daily Life | Practical comprehension | Emails, notices, social media posts (15-150 words) |
| Read an Academic Passage | Academic comprehension | ~200-word academic text + 5 questions |
🎯 What Do "Complete the Words" Questions Look Like?
This isn't multiple choice. You must write the missing portion of words yourself.
Examples:
- The experiment required precise mea________ of temperature changes.
- The new software greatly improved the eff________ of data processing.
- It is important to dist________ between fact and opinion.
- He worked tire________ to meet the project deadline.
Answers: measurement, efficiency, distinguish, tirelessly
The challenge: Recognizing a word and spelling it correctly are two different skills. Pay special attention to double consonants (committee, accommodation) and vowel combinations (receive, achieve).
🎯 What's Tested in "Read in Daily Life"?
Questions about emails, notices, and social media posts test your ability to read between the lines (understand implied meaning).
Example: A professor's email
"Thanks for your proposal. I think it has potential, but perhaps we could refine the methodology section before moving forward."
- Surface meaning: Your proposal looks promising
- Actual meaning: The methodology has issues and needs revision
Common "diplomatic expressions" in academic settings:
| Expression | Literal Meaning | Actual Meaning |
|---|---|---|
| "I'll take that into consideration." | I'll think about it | I probably won't adopt it |
| "That's an interesting perspective." | That's an interesting viewpoint | I don't quite agree |
| "We might want to revisit this later." | Let's discuss this later | This proposal is rejected |
Listening Section
| Question Type | Description |
|---|---|
| Listen and Choose a Response | Hear a statement and choose the most appropriate response |
| Listen to a Conversation | Comprehension of everyday dialogues |
| Listen to an Announcement | Understanding campus announcements |
| Listen to an Academic Talk | Academic content, 175-250 words |
Important: Listening materials include four accent varieties: North American, British, Australian, and New Zealand.
🎯 "Listen and Choose a Response" Examples
This question type tests your ability to give socially appropriate responses. You need to understand not just the literal meaning, but also the speaker's tone and emotions.
Example 1:
🎧 A: "I was thinking of grabbing coffee after class. Want to join?"
Options:
- A. "Sure, I'd love to." ✅
- B. "I don't really like coffee." ❌
- C. "When did you start drinking coffee?" ❌
- D. "Coffee is very popular." ❌
Correct answer: A
While B is grammatically correct, it sounds abrupt in a social context. A more natural way to decline would be: "I'd love to, but I have to finish an assignment."
Example 2:
🎧 A: "I just submitted the group report by myself. No one else showed up."
Options:
- A. "You should tell the professor right away." ✅
- B. "That's nice of them to help you out." ❌
- C. "Maybe they'll show up later." ❌
- D. "I guess you didn't finish it, then." ❌
Correct answer: A
The speaker is frustrated/annoyed ("No one else showed up"). The correct response should show empathy or offer advice. B, C, and D ignore the emotion and are illogical.
Example 3:
🎧 A: "I thought the exam was next week, but apparently it's tomorrow morning!"
Options:
- A. "That explains why you look so relaxed." ❌
- B. "Well, good luck pulling an all-nighter." ✅
- C. "Then you must have studied hard already." ❌
- D. "Why didn't you tell me that earlier?" ❌
Correct answer: B
The speaker is clearly panicking. A is sarcasm (illogical in context), C contradicts reality (they're not prepared), D blames the other person. B shows empathy with light humor—natural in everyday English conversation.
Writing Section
| Question Type | Count | Description |
|---|---|---|
| Build a Sentence | 10 | Reorder scrambled words/phrases |
| Write an Email | 1 | Write an email based on a given scenario |
| Write for an Academic Discussion | 1 | Class discussion format, minimum 100 words |
🎯 How to Write an "Appropriate" Email
Scoring criteria emphasize social conventions: politeness, register, and information organization.
❌ Poor example:
"Dear Professor Smith, I have a family emergency. I hope you understand. Thank you."
Issues: Too brief, no situation explanation, no specific request, somewhat impolite tone.
✅ Good example:
Dear Professor Smith,
I'm writing to request permission to miss class next Tuesday, October 15th.
I have a family emergency that requires me to travel home that day. I understand we'll be covering the midterm review, and there's also a group presentation scheduled.
Would it be possible to:
- Get the review materials after class?
- Reschedule my group's presentation to the following week?
I've already coordinated with my group members, and they're flexible with the date.
I apologize for any inconvenience and really appreciate your understanding.
Best regards, [Your name]
Key points: State purpose upfront → Explain the specific situation → Make clear requests → Close politely
Speaking Section
| Question Type | Count | Description |
|---|---|---|
| Listen and Repeat | 7 | Repeat sentences immediately after hearing them (progressive difficulty) |
| Take an Interview | 4 | Answer questions about experiences and opinions |
🎯 Why is "Listen and Repeat" Challenging?
According to the official document, this question type measures language processing ability and pronunciation clarity. Sentences get progressively longer, potentially exceeding 10 words in later items.
Example:
"The professor postponed the exam due to unforeseen circumstances."
Common mistakes:
"The professor... uh... postpone exam... because... unexpected things."
Issues: Missing articles (the), tense errors (postpone → postponed), imprecise vocabulary (unforeseen circumstances → unexpected things)
Tip: Don't memorize word by word—remember meaning chunks
- The professor / postponed the exam / due to unforeseen circumstances
- Who / did what / why
AI Scoring: How Reliable Is It?
The new TOEFL uses a hybrid AI + human scoring system for Writing and Speaking.
AI Scoring Accuracy Data
| Section | Human-AI Correlation | Human-Human Correlation |
|---|---|---|
| Writing | 0.86 | 0.85 |
| Speaking | 0.89 | 0.96 |
Interpretation: The agreement between AI and human scoring is nearly equal to—or higher than—the agreement between human raters themselves.
Quality Assurance
- If AI confidence is low, human raters automatically take over
- Regular sampling reviewed by certified raters
- Continuous monitoring of scoring consistency
Reliability Data
| Section | Reliability Coefficient | Standard Error of Measurement (SEM) |
|---|---|---|
| Reading | 0.86 | 0.37 |
| Listening | 0.88 | 0.35 |
| Writing | 0.87 | 0.36 |
| Speaking | 0.94 | 0.22 |
| Total | 0.90 | 0.32 |
Speaking has the highest reliability (0.94), and the total score reliability of 0.90 meets industry standards for high-stakes testing.
Preparation Tips
The following tips are based on analysis of the RR-25-12 document and do not represent official ETS positions.
For MST Adaptive Testing
Your Stage 1 performance determines Stage 2 difficulty. While scoring is calibrated, a stable start helps you enter a question set that matches your level, which is psychologically advantageous.
Tip: Stay calm at the start of the exam. The Router module is medium difficulty, so answer carefully without rushing.
For Complete the Words (C-test)
This tests not just vocabulary, but spelling accuracy and contextual inference ability.
Tips:
- When studying vocabulary, always practice writing the words
- Use dictation exercises to strengthen spelling
- Pay attention to prefix and suffix spelling patterns
For Listen and Repeat
Tips:
- Practice shadowing (repeating almost simultaneously with audio) daily
- Start with short sentences and gradually increase length
- Pay attention to linking and reduction phenomena
For Email Writing
Tips:
- Learn standard English email expressions
- Understand the difference between formal and informal registers
- Practice different scenarios: requests, suggestions, apologies
For Accent Variety
Tips:
- BBC (British English)
- ABC Australia (Australian English)
- TED Talks (various accents)
FAQ
Q: Is the new TOEFL still scored out of 120?
No. The new TOEFL uses a 1-6 scale. Both section scores and total scores range from 1-6 (in 0.5 increments).
Q: Is MST adaptive testing unfair?
No. ETS uses IRT (Item Response Theory) to statistically calibrate scores for test-takers who followed different difficulty paths, ensuring fairness.
Q: Can I still use my old TOEFL scores?
Check with your target institution. Since the scoring system is completely different, universities may need time to adjust their requirements.
Q: Is AI scoring reliable?
According to official data, agreement between AI and human scoring is high (Writing 0.86, Speaking 0.89), and there's a human review system in place.
Summary
Key changes in the 2026 TOEFL reform:
- Scoring system: 0-120 → 1-6 scale (aligned with CEFR)
- Testing method: MST adaptive testing for Reading and Listening
- Question types: New formats including Complete the Words and Listen and Repeat
- Scoring technology: Hybrid AI + human scoring
The RR-25-12 document is currently the most authoritative technical document on the new TOEFL. For more details, consult the original text on the ETS Research website.
Questions? Leave them in the comments.
— WizPrep Academic Team
Tags: TOEFL TOEFL reform TOEFL 2026 new TOEFL TOEFL preparation TOEFL test TOEFL score TOEFL question types MST adaptive testing CEFR English test study abroad study in USA TOEFL iBT AI scoring test prep
Related Articles:
- 2026 TOEFL Test Dates
- TOEFL vs IELTS Score Conversion
- TOEFL Requirements for Top 50 US Universities