A data-accurate guide for test-takers and nerds.
Listen & Repeat looks simple. You hear a short sentence, and you repeat it. The scoring guide says a top score requires an “exact” repetition. Many learners interpret this as: “I must repeat every word exactly as it would be written.”
That interpretation is incorrect.
The scoring system evaluates your response based on the audio you hear, not the written transcript. Contractions like “you’re” and “you are” are pronunciation variants of the same lexical item. They preserve meaning, they preserve the structure of the utterance, and they produce no penalty under the official scoring construct.
This article explains exactly why. The reasoning below matches ETS scoring logic and the engineering principles behind automated spoken-language evaluation.
How Listen & Repeat Is Actually Scored
The task assesses how accurately you repeat the spoken prompt. Scoring is based on meaning preservation and intelligibility, not visual text matching.
The key idea:
A contraction is not a meaning change.
Meaning is the anchor of the scoring system.
Why Contractions Are Always Considered Correct
Four reasons, grounded in real scoring logic:
1. The task is audio-based
“You’re” and “you are” have nearly identical acoustic signals. Scoring engines evaluate the sound, not written form.
2. Contractions do not change meaning
They do not alter tense, aspect, content, or intent.
3. ASR transcriptions flip them routinely
A speaker says “you are,” and the ASR may output “you’re.”
A speaker says “you’re,” and the ASR may output “you are.”
This happens because the acoustic difference is minimal.
4. Penalizing contractions would cause false score drops
If the system scored literal text strings, high-performing speakers would routinely receive lower scores due to ASR output variation.
This is why scoring engines normalize contractions automatically.
The Correct Examples: All Contraction-Based, All Accurate
These examples follow the Listen & Repeat rubric exactly.
Score 5 Example: Contraction swap with identical meaning
Prompt:
“You are guiding students through a maze.”
Response:
“You’re guiding students through a maze.”
Reasoning:
Only the contraction changes. Meaning, structure, and content are identical. Fully intelligible. This earns a 5.
Score 4 Example: Perfect contraction swap, but delivery has minor issues
Prompt:
“They will check your ID before you enter.”
Response:
“They’ll uh check your ID before you enter.”
Reasoning:
The wording is identical except for the contraction.
But the delivery includes a filled pause (uh). Same if it has a clipped syllable, or a pronunciation flaw, the meaning remains intact but the performance is not fully “5-level.”
This fits the 4 range: meaning preserved with minor delivery differences.
This would also be a 4: “They will check your IDs before you enter.”
Score 3 Example: Contraction correct, meaning changed
Prompt:
“They will check your ID before you enter.”
Response:
“They’ll check your ID before you can enter.”
Reasoning:
The contraction is fine. The added modal “can” changes the condition.
Meaning altered → Score 3.
Score 2 Example: Contraction correct, content missing
Prompt:
“You are welcome to visit the front desk for help.”
Response:
“You’re welcome to visit for help.”
Reasoning:
Missing “the front desk.”
Meaning incomplete → Score 2.
Score 1 Example: Contraction correct, almost everything else missing
Prompt:
“You are welcome to visit the front desk for help.”
Response:
“You’re uh welcome.”
Reasoning:
Only a fragment of the prompt is repeated.
Meaning mostly lost → Score 1.
Score 0 Example: Contraction irrelevant
Prompt:
“You are welcome to visit the front desk for help.”
Response:
“You’re… uh… water… hello.”
Reasoning:
Irrelevant and unintelligible.
Why Scoring Systems Must Normalize Contractions
A scoring engine that compares raw ASR text to a reference transcript will mis-score learners unless contractions are normalized.
Normalization ensures:
• “you’re” = “you are”
• “they’ll” = “they will”
• “it’s” = “it is”
• “I’m” = “I am”
Without normalization, ASR variability becomes a scoring error instead of a linguistic feature.
This is why My Speaking Score hard-codes contraction equivalence directly into the scoring pipeline.
Contraction Normalization Table
FAQ
Are contractions officially allowed in Listen & Repeat?
Yes. They do not alter meaning and occur naturally in fluent speech. ETS scoring reflects this.
Is a contraction ever considered an error?
Only if the surrounding words change meaning.
The contraction itself is never the error.
Does ASR inconsistency affect my score?
No. Scoring systems normalize variants so ASR output does not penalize performance.
Should I intentionally use contractions?
Use what feels natural. Both forms are scored the same.
Final Takeaway
Listen & Repeat rewards accurate meaning and intelligibility. Contractions are pronunciation variants, not content changes. When scoring engines treat them as equivalent, scores reflect true performance instead of transcription noise. That alignment is essential for a fair, data-driven system.