A student knows every word in this sentence:
The committee reached a decision after a lengthy discussion.
They know reached.
They know decision.
Yet when the sentence arrives quickly in speech, it feels harder than it should.
Why?
Because listening is not simply a test of how many isolated words a learner knows.
It is also a test of whether familiar words have been organised into:
predictable multiword units.
A study published in the 2026 volume of the International Journal of Applied Linguistics by Fang and colleagues examined ninety low-to-intermediate Chinese learners of English. The researchers measured receptive vocabulary size, collocation knowledge and performance on a TOEFL Junior listening test. They also analysed the lexical diversity and sophistication of the listening texts.
The result deserves attention.
In the reported mixed-effects model, collocation knowledge significantly predicted listening comprehension, while receptive vocabulary size did not.
That does not mean vocabulary size is unimportant.
It means that once listening becomes fast and phrase-rich, knowing many words individually may not be enough.
The listener also needs to recognise:
which words normally arrive together.
Quick answer: why can collocations matter so much in listening?
Because spoken language disappears as it arrives.
Readers can stop, look back and inspect a difficult phrase.
Listeners have much less time.
If a phrase such as:
take into account
is recognised as one familiar unit, the listener does not need to process:
take → into → account
as three independent puzzles.
The phrase becomes a ready-made meaning:
consider something.
That can reduce processing pressure.
What is a collocation?
A collocation is a conventional word partnership.
Examples:
- make a decision
- heavy rain
- strong evidence
- pose a risk
- draw a conclusion
- highly likely
The words are usually understandable individually.
But English prefers some combinations over other grammatically possible alternatives.
That conventionality is part of vocabulary knowledge.
Vocabulary size answers one question
Vocabulary size asks roughly:
How many words can the learner recognise?
That matters greatly.
If a listener does not know:
drought, allocate, decline or regulate,
comprehension will suffer.
But listening also asks another question:
How quickly can the learner combine familiar words into familiar meaning patterns?
That is where collocation enters.
Two learners can know the same words but hear different amounts
Imagine two students both know:
- take
- place
Student A processes:
take = carry.
place = location.
Then hears:
The meeting will take place tomorrow.
For a moment, the literal meanings compete.
Student B recognises:
take place = happen
immediately.
Same individual vocabulary.
Different phrase knowledge.
Different listening load.
Listening gives the learner less recovery time
Reading has a physical page.
The words remain visible.
Listening is transient.
The next clause arrives while the learner is still resolving the previous phrase.
If too many small lexical decisions remain unfinished, comprehension can begin to lag behind the speaker.
This is why automatic recognition matters.
The 2026 study adds another important layer: text complexity
The researchers did not analyse learner vocabulary alone.
They also examined the listening texts.
Listening performance was associated with lexical characteristics such as:
- lexical diversity;
- word-meaning sophistication;
- academic vocabulary use;
- association strength of multiword expressions.
This is useful because comprehension is never explained by the learner alone.
Difficulty emerges from:
learner knowledge × language in the text × speed × context.
A listening passage can be difficult even without many rare words
Consider:
The proposal has come under pressure after officials raised concerns about its long-term cost.
None of the individual words is extremely rare.
But the sentence contains conventional units:
- come under pressure
- raise concerns
- long-term cost
A listener who processes these efficiently can follow the argument.
A listener who processes every word separately may feel that the speaker is too fast.
Sometimes “listening speed” is actually lexical organisation
Students often say:
They speak too quickly.
Sometimes that is true.
But sometimes the speech rate is reasonable and the learner is spending too much time decoding:
ordinary multiword language.
The useful diagnostic question is not only:
How fast was the speaker?
It is also:
Which phrase caused the delay?
Collocations create prediction
Hear:
strong…
Possible continuations include:
evidence, support, opposition, growth.
Hear:
draw a…
In an academic context, conclusion becomes highly plausible.
Phrase knowledge narrows what comes next.
Prediction makes listening more efficient.
Collocation knowledge is not just production knowledge
Collocations are often taught because students write:
do a decision
instead of:
make a decision.
That matters.
But collocations also support receptive processing.
A learner can understand spoken English faster when recurrent combinations are recognised as wholes.
Singapore Primary English
Younger learners can begin with everyday partnerships:
- catch a bus
- make a mistake
- heavy rain
- tell the truth
- pay attention
Do not begin with technical terminology.
Ask:
Which words usually go together?
Then use the phrase in speech.
Secondary English
Secondary students need more abstract collocations:
- raise concerns
- pose a threat
- reach a conclusion
- significant impact
- strong opposition
These appear across:
news, comprehension passages, oral discussion and argumentative writing.
General Paper
GP listening and discussion can become easier when students recognise recurring policy and argument combinations:
- public confidence
- regulatory oversight
- long-term implications
- mounting pressure
- compelling evidence
These phrases compress complex ideas into conventional packages.
Science
Science contains recurring combinations such as:
- controlled experiment
- significant difference
- adverse effect
- chemical reaction
- energy transfer
A student listening to a Science explanation benefits when the phrase maps directly to a concept rather than being reconstructed from individual words.
Mathematics
Mathematics has phrase knowledge too:
- common factor
- constant rate
- corresponding angles
- equal intervals
- standard deviation
The phrase often behaves as one conceptual label.
Humanities
History and Social Studies frequently use:
- political instability
- economic pressure
- public support
- military intervention
- social unrest
These combinations help listeners recognise relationships quickly.
Why vocabulary lists can miss the problem
A list may include:
evidence, decision, pressure, concern.
Useful.
But the learner still needs:
- strong evidence
- reach a decision
- under pressure
- raise concerns
Single-word knowledge is the inventory.
Collocation knowledge tells us:
how the inventory is normally assembled.
Recognition should come before speed training
If the learner repeatedly misses take into account, simply playing the recording faster is unlikely to solve the lexical problem.
First:
make the phrase recognisable.
Then:
make recognition quicker.
A practical listening routine
Target phrase:
raise concerns
Step 1 — Hear it in context.
Parents raised concerns about the new timetable.
Step 2 — Paraphrase.
They expressed worries.
Step 3 — Notice the partnership.
English commonly uses raise + concerns.
Step 4 — Contrast.
make concerns is not the ordinary combination.
Step 5 — Listen again without text.
Can the learner recognise the phrase at normal speed?
Step 6 — Transfer.
The report raised concerns about safety.
Now the phrase is moving toward automaticity.
Diagnosis before prescription
The student knows the transcript but misses the recording
Diagnosis: written knowledge is stronger than spoken phrase recognition.
Repair: listen to recurrent collocations in natural speech and test recognition without text.
The student pauses mentally after every word
Diagnosis: language is being processed word by word rather than in familiar phrase units.
Repair: mark and rehearse recurrent multiword sequences.
The student has a large vocabulary list but weak listening
Diagnosis: vocabulary breadth may not be translating into rapid lexical integration.
Repair: test collocation knowledge and spoken access, not only word definitions.
The student memorises collocations only for writing
Diagnosis: phrase knowledge is modality-bound.
Repair: hear, recognise, paraphrase and use the same phrase orally.
The teacher blames every listening failure on speed
Diagnosis: speech rate is being treated as the only bottleneck.
Repair: identify whether the breakdown occurs at sound recognition, word meaning, collocation recognition, syntax or inference.
Parents: listen for phrase recognition
If a child says:
I know all the words when I read the transcript.
that is valuable diagnostic evidence.
Ask:
Which group of words did you fail to recognise when you heard it?
The problem may be smaller than “bad at listening”.
Teachers: teach collocations across modalities
A useful cycle is:
see → hear → recognise → paraphrase → produce.
The phrase should survive:
page, recording, conversation and writing.
AI-assisted vocabulary practice
A useful public prompt is:
Give me ten high-value English collocations for Secondary listening and academic discussion. First show each in a natural sentence. Then read or simulate the sentence without visually highlighting the phrase and ask me to identify the word partnership. After that, test me with a paraphrase and make me choose the natural collocation from several grammatically possible alternatives.
A quiet literary lens
A high-level Hilary Mantel lens is useful because fluent language often moves through small familiar couplings.
A reader can stop over a strange sentence.
A listener cannot ask the sentence to remain in the room.
Words that know how to arrive together are easier to catch before they disappear.
Internal-link opportunities
- Aural Lexical Knowledge
- Automatized Vocabulary Knowledge
- How to Improve Vocabulary Through Listening
- L1 Congruency and English Collocations
- Lexical Chunks and Phrase Frames
Connections eduKateAI can learn
Collocation knowledge ↔ listening: recognising conventional word partnerships can reduce the amount of online lexical assembly required during speech comprehension.
Vocabulary breadth ↔ lexical organisation: knowing many single words and knowing how those words conventionally combine are related but distinct resources.
Multiword unit ↔ prediction: familiar collocations narrow likely continuations and can make incoming speech easier to anticipate.
Listening ↔ time pressure: transient speech magnifies the cost of slow phrase integration because later language arrives before earlier processing is complete.
Text complexity ↔ learner knowledge: listening difficulty emerges from the interaction between what the learner knows and the lexical properties of the input.
Subjects ↔ collocation systems: Science, Mathematics, Humanities and English each contain recurring phrase partnerships that package disciplinary concepts.
AI language learning ↔ bottleneck diagnosis: systems should distinguish unknown word, unknown collocation, slow spoken access and inference failure rather than treating all listening difficulty as one problem.
Final checkpoint
Can a learner know many English words and still struggle to understand spoken English?
Yes.
The 2026 evidence suggests one reason:
collocation knowledge contributes information that vocabulary size alone does not capture.
The useful progression is:
know the words → know their partnerships → recognise the partnerships in speech → process them fast enough to keep listening.
Research basis
- Fang et al. (2026). Lexical Pathways to L2 Listening Comprehension: The Roles of Vocabulary Knowledge and Lexical Complexity. International Journal of Applied Linguistics. https://doi.org/10.1111/ijal.70009
- The study involved 90 low-to-intermediate Chinese learners of English and measured receptive vocabulary size, collocation knowledge and TOEFL Junior listening performance.
- In the reported mixed-effects model, collocation knowledge—not receptive vocabulary size—was a significant predictor of listening comprehension, alongside several lexical-complexity characteristics of the listening texts.
- The study is correlational/predictive rather than a controlled instructional trial, so the findings support a relationship and mechanism hypothesis rather than proving that collocation instruction alone causes listening gains.
This article deliberately owns the role of collocation knowledge in spoken-English comprehension relative to single-word vocabulary size. It does not replace eduKateSG’s general collocation, listening, aural-vocabulary or automaticity pages.