Language
Word Partners Task
Also called: collocation task, collocation test, collocational knowledge task, collocation judgment task, phrase acceptability task
Measure collocation knowledge. Participants choose the word that usually goes with another, then judge frequent, rare and unused English word pairs against the clock.
7-9 minTimed: desktop or laptop with a mouse or trackpad
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Try it here first, with no account.
- Papers, 2021 to 2025
- 215
- Evidence grade
- C
- Registered clinical studies
- 0
What it measures
Measures collocation knowledge, knowing which words usually go together. Participants pick the word that usually goes with another, such as keen interest, then press F or J to judge whether word pairs are usual in English, with frequent pairs, rare pairs and pairs that are not used. Scores accuracy, and how much faster frequent pairs are recognized than rare ones. The words are English in every session.
Constructs
- Collocation knowledge
- Vocabulary depth
- Lexical processing
Research uses
- Second-language research
- Vocabulary research
- Psycholinguistics
How the paradigm works
Knowing which words usually go together, such as heavy rain rather than strong rain, marks fluent use of a language and is a frequent difficulty for second-language learners. This collocation task has two parts, with English items in every session. In part 1, a word with a gap appears with three words, and participants choose the one that usually goes in the gap, with no deadline: ten adjectives for nouns and ten verbs for nouns, the other two options near-synonyms or the verbs learners use instead, the kinds of error Nesselhauf found in learners' writing. In part 2, after a cross for 250 ms and a blank for 50 ms, an adjective and a noun appear together, and participants press F or J within 4 seconds to say whether the pair is a usual one in English, as in the acceptability judgments of Wolter and Gyllstad and of Öksüz, Brezina and Rebuschat. Twenty adjectives each make one frequent pair and one rare pair, and each of their forty nouns appears again with another adjective in a pair that is not used, so no single word gives the answer away; the frequencies come from Google Books. Which key means usual is balanced across sessions. Two practice items and four practice pairs come first, with feedback.
Scores
- Usual Partners Chosen%
- Percentage of part 1 items answered with the usual partner.
- Higher means more correct
- Adjective Partners Chosen%
- The same for the ten adjective + noun items.
- Higher means more correct
- Verb Partners Chosen%
- The same for the ten verb + noun items.
- Higher means more correct
- Frequency Effectms
- Median time to accept rare pairs minus median time to accept frequent pairs, over right answers.
- Positive means frequent pairs were recognized faster
- Pairs Judged Correctly%
- Right answers in part 2, as a percentage of right and wrong answers.
- Higher means more correct
- Frequent Pairs Accepted%
- Frequent pairs answered usual.
- Higher means more correct
- Rare Pairs Accepted%
- Rare pairs answered usual.
- Higher means more correct
- Wrong Pairs Rejected%
- Pairs that are not used, answered not usual.
- Higher means more correct
- Median Time, Frequent Pairsms
- Median time of right answers to frequent pairs.
- Lower means faster
- Median Time, Rare Pairsms
- Median time of right answers to rare pairs.
- Lower means faster
- Median Time, Wrong Pairsms
- Median time of right answers to pairs that are not used.
- Lower means faster
- Missed Pairscount
- Part 2 pairs left unanswered for 4 seconds.
- 0 expected
- Answers Under 100 mscount
- Answers given within 100 ms in either part.
- 0 expected
- Part 1 Items Answeredcount
- Part 1 items answered in the most complete run.
- 20 standard
- Part 1 Items Plannedcount
- Part 1 items in the run.
- 20 standard
- Part 2 Pairs Showncount
- Part 2 pairs recorded in the most complete run, answered or missed.
- 80 standard
- Part 2 Pairs Plannedcount
- Part 2 pairs in the run.
- 80 standard
- Other Instruction Languageflag
- 1 when the English items were shown with instructions in another language.
- Review English comprehension
Use and evidence
215 papers from 2021 to 2025 name this task in their title or abstract, up 5% compared with 2016 to 2020. That ranks it 160 of 711 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 55 |
| 2024 | 50 |
| 2023 | 43 |
| 2022 | 33 |
| 2021 | 34 |
| 2020 | 40 |
| 2019 | 40 |
| 2018 | 35 |
| 2017 | 43 |
| 2016 | 46 |
| 2015 | 36 |
| 2014 | 40 |
| 2013 | 29 |
| 2012 | 23 |
| 2011 | 16 |
| 2010 | 19 |
| 2009 | 16 |
| 2008 | 7 |
| 2007 | 8 |
| 2006 | 5 |
| 2005 | 5 |
| 2004 | 2 |
| 2003 | 3 |
| 2002 | 2 |
| 2001 | 0 |
| 2000 | 2 |
Use grade C
100 to 299 papers from 2021 to 2025.
Evidence grade C
At least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies.
- Papers on test-retest reliability
- 2
- Papers on reliability or validity
- 34
- Registered clinical studies using it as an outcome
- 0
Measured on ConductCognition
Pooled across the studies that feed our platform norms, by device group. A figure needs 50 or more people.
No pooled figure yet.
Note on the counts: mostly second-language learning studies.
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.58.0, counted 13 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Part 1
- 20 English items with no deadline: a word with a gap and 3 options, 10 adjectives for nouns and 10 verbs for nouns. The usual partner is at least 20 times as frequent with the word as the other 2.
- Part 2
- 80 adjective + noun pairs after a cross for 250 ms and a blank for 50 ms, each judged usual or not with F or J within 4 seconds.
- Pairs
- 20 adjectives each make 1 frequent and 1 rare pair; each of their 40 nouns appears again in a pair that is not used, with 1 of 20 other adjectives, so no single word gives the answer away.
- Frequencies
- From Google Books (English, 2010 to 2019): frequent pairs at least 0.4 per million, rare pairs 0.02 to 0.15, unused pairs at most 0.003. Nouns of frequent and rare pairs are matched on word frequency and length.
- Practice
- 2 items and 4 pairs with feedback. Which key means usual is balanced across sessions.
- Scores
- Percentage of usual partners chosen in part 1, and the frequency effect: the median time to accept rare pairs minus frequent pairs. Acceptance, rejection and median times by kind of pair, missed pairs and answers under 100 ms.
The items are our own, chosen with open frequency lists, and they are English in every session, as in second-language collocation studies. Scores support research and do not diagnose. No percentile output.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: word_partners_choice and word_partners_judgment from the most complete run, checked against the canonical items: one of three words in part 1, F or J for usual or not within 4 seconds in part 2 (half the pairs are usual). Below chance invalidates and not above chance is flagged, and one option on 90% of answers invalidates when accuracy is not above chance. Missed responses and responses under 100 ms follow the timed-test rule. Fewer than half the items of a part invalidates and any shortfall is flagged. A session in a language other than English is flagged for review. A hidden tab withholds the percentile.
Why these rules
Accuracy significantly below chance (one-sided exact test, p < .05) cannot come from guessing; it means the response mapping was reversed or the task was not followed (Slick, Sherman and Iverson 1999 treat below-chance forced-choice performance as definite response bias). Chance-level accuracy can be a real result in impaired participants, so it is only flagged. One option on at least 90% of answered trials, when the correct options are balanced and accuracy is not above chance, means the answers did not depend on the items (the task counterpart of long-string responding in surveys; Curran 2016). The usual partner is in each of the three positions equally often, and half the pairs are usual. Missed responses and responses under 100 ms keep the platform's timed-test rule: more than max(3, 25%) missed responses or more than max(2, 20%) responses under 100 ms invalidate, and any is flagged. A response under 100 ms cannot be a reaction to the stimulus. The two parts measure different things, so an incomplete part changes the scores' meaning. The word lists are English. A session run in another language shows English words to a participant who may not read English fluently, so it is flagged for review. Another tab can be used to look a phrase up, so a percentile is withheld after a hidden tab.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
This test runs with fixed settings.
Need a different version of this test?
Devices and languages
- Timed test: participants need a desktop or laptop with a mouse or trackpad. Participant devices
- Instructions and screens: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese. Words and sentences: English.
- Available on: Every plan, including the free plan.
- No percentile. Scores and quality flags only. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Word Partners Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Nesselhauf N, 2003; Wolter B, Gyllstad H, 2011, 2013; Öksüz D, Brezina V, Rebuschat P, 2021). The main score was usual partners chosen (percentage of part 1 items answered with the usual partner).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Nesselhauf, N. (2003). The use of collocations by advanced learners of English and some implications for teaching. Applied Linguistics, 24(2), 223–242. https://doi.org/10.1093/applin/24.2.223Verb + noun collocations in advanced German-speaking learners' free writing: the first language shaped their errors far more than earlier studies predicted.
- Wolter, B., & Gyllstad, H. (2011). Collocational links in the L2 mental lexicon and the influence of L1 intralexical knowledge. Applied Linguistics, 32(4), 430–449. https://doi.org/10.1093/applin/amr011A primed lexical decision task and a receptive collocation test: learners responded faster to collocations that have a translation in their first language.
- Wolter, B., & Gyllstad, H. (2013). Frequency of input and L2 collocational processing: A comparison of congruent and incongruent collocations. Studies in Second Language Acquisition, 35(3), 451–482. https://doi.org/10.1017/S0272263113000107An acceptability judgment task with collocations and matched unrelated pairs: native and advanced learners were highly sensitive to collocation frequency.
- Öksüz, D., Brezina, V., & Rebuschat, P. (2021). Collocational processing in L1 and L2: The effects of word frequency, collocational frequency, and association. Language Learning, 71(1), 55–98. https://doi.org/10.1111/lang.12427Judgments of whether adjective + noun pairs are commonly used, with high- and low-frequency collocations and baseline pairs that reuse their nouns: both groups were faster for frequent collocations.
Source papers
- Wolter and Gyllstad (2013); Öksüz, Brezina and Rebuschat (2021): collocation acceptability judgments.Paradigm source, Adults who read English
Questions
How long does the Word Partners Task take?
About 7 to 9 minutes, as listed in the test library.
What device does a participant need?
Timed test: participants need a desktop or laptop with a mouse or trackpad. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Usual Partners Chosen, Adjective Partners Chosen, Verb Partners Chosen, Frequency Effect, Pairs Judged Correctly, Frequent Pairs Accepted, Rare Pairs Accepted, Wrong Pairs Rejected, Median Time, Frequent Pairs, Median Time, Rare Pairs, Median Time, Wrong Pairs, Missed Pairs, Answers Under 100 ms, Part 1 Items Answered, Part 1 Items Planned, Part 2 Pairs Shown, Part 2 Pairs Planned and Other Instruction Language. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Word Partners Task used in research?
215 papers from 2021 to 2025 name it in their title or abstract, up 5% compared with 2016 to 2020. That ranks it 160 of 711 test families in the ConductCognition test census (use grade C). Source: ConductCognition test census 0.58.0, counted 13 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Word Partners Task have?
In the published research, 2 papers report its test-retest reliability and 0 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which languages can participants use?
The instructions and screens are in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese. The words and sentences stay in English, so sessions in other languages are flagged for review. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.