Language
Pronoun Resolution Task
Also called: pronoun interpretation task, anaphor resolution task, implicit causality task, pronoun assignment task
Measure how readers link he or she to the person it means: participants choose the referent in short English sentences, timed, with gender, meaning and the verb's implied cause as cues.
4-6 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
- 91
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures how readers link a pronoun to the person it means. Participants read 48 short English sentences about two people, such as "Zoe wrote to Carl because she was homesick", and press F or J to say who "he" or "she" refers to. The pronoun's gender picks the person in half; in the rest, only the meaning does. The main score is how much slower answers are when the pronoun goes against the verb's expected cause.
Constructs
- Pronoun resolution
- Implicit causality
- Discourse comprehension
- Sentence comprehension
Research uses
- Psycholinguistics and discourse research
- Second-language reading research
- Within-study comparison of groups and conditions
How the paradigm works
The Pronoun Resolution Task measures how readers link a pronoun to the person it refers to. Each sentence names two people and goes on with "because he" or "because she", and the task is to say, as quickly as possible, whether the pronoun means the person named first or the person named second, as in Caramazza, Grober, Garvey and Yates (1977). Two cues decide: the pronoun's gender, when the two people differ in it, and the meaning of the sentence, when they share it. The verb adds a third. Verbs of implicit causality lead readers to expect the cause in one of the two people: the first after "apologized to", the second after "thanked" (Garvey, Caramazza and Yates 1974; Ferstl, Garnham and Manouilidou 2011). In Caramazza and colleagues' study, people named the person faster when the pronoun went with the verb's expected cause than against it, an effect reliable for eight of the fourteen verbs, and faster when gender told the two people apart than when only the meaning did; the two effects did not interact.
More about the paradigm
Each session uses 24 verbs, 12 that favour the person named first and 12 that favour the person named second, chosen from Ferstl, Garnham and Manouilidou's English norms: each verb's bias is at least 65 points of 100, and the people's genders changed it by less than 15 points. Each verb appears twice, once in a sentence whose ending makes the pronoun refer to the person the verb favours (consistent) and once in a sentence whose ending makes it refer to the other (inconsistent), and once with two people of the same gender and once with a man and a woman. Two lists drawn from the session's seed decide which pairing, so across sessions every ending appears with both kinds of pair. A session has 12 sentences of each of the four kinds, half with "he" and half with "she" and half answered with each key, with no more than 3 answers on one key in a row and a verb's two sentences at least 12 apart. The people have common first names whose gender is plain, drawn from the seed. A cross shows for 500 ms, then the sentence with its pronoun in bold and both names under it; F means the person named first and J the person named second, within 10 seconds, timed from the sentence's onset as in Caramazza and colleagues' study. Six practice sentences with feedback come first. Every sentence was written for this library, none from a published set, and no ending names a job, a social role or anything people tie to one gender, so no stereotype points to an answer. The sentences are English in every session; the instructions follow the session's language.
The main score is the consistency effect: the median time of right answers to inconsistent sentences minus consistent ones, over answers at or after 100 ms. The task also reports the gender effect, the median time of right answers when the two people share a gender minus when they differ, the four medians, and accuracy over all sentences and over each kind. Caramazza and colleagues kept sentences that at least 91% of their judges read the intended way; these sentences were written to be clear but have not been rated, so one that some readers take the other way lowers accuracy on same-gender sentences. Scores describe how a reader used gender and verb meaning in this task, a research measure; they do not diagnose.
Scores
- Consistency Effectms
- Median time of right answers to sentences whose pronoun goes against the verb's expected cause minus sentences whose pronoun goes with it, from the sentence's onset.
- Positive values mean pronouns against the verb took longer
- Gender Effectms
- Median time of right answers when the two people share a gender minus when they differ in it.
- Positive values mean answers took longer when only the meaning told the people apart
- Consistent Sentence Timems
- Median time of right answers to sentences whose pronoun goes with the verb's expected cause.
- Lower is faster
- Inconsistent Sentence Timems
- Median time of right answers to sentences whose pronoun goes against the verb's expected cause.
- Lower is faster
- Same Gender Timems
- Median time of right answers to sentences about two people of the same gender.
- Lower is faster
- Different Gender Timems
- Median time of right answers to sentences about a man and a woman.
- Lower is faster
- Accuracy%
- Right answers as a share of answers at or after 100 ms, over all 48 sentences.
- Below 0.7 is invalid
- Consistent Accuracy%
- Accuracy on sentences whose pronoun goes with the verb's expected cause.
- Higher is more right answers
- Inconsistent Accuracy%
- Accuracy on sentences whose pronoun goes against the verb's expected cause.
- Higher is more right answers
- Same Gender Accuracy%
- Accuracy on sentences about two people of the same gender, where only the meaning tells them apart.
- Higher is more right answers
- Different Gender Accuracy%
- Accuracy on sentences about a man and a woman, where the pronoun's gender tells them apart.
- Higher is more right answers
- Missed Sentencescount
- Test sentences not answered within 10 seconds.
- Expected to be 0
- Anticipationscount
- Answers under 100 ms after the sentence appeared, counted in no score.
- Expected to be 0
- Session In Another Language0/1
- 1 when the session ran in a language other than English; the sentences are English in every session.
- Flagged for review
Use and evidence
91 papers from 2021 to 2025 name this task in their title or abstract, down 12% compared with 2016 to 2020. That ranks it 227 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 18 |
| 2024 | 17 |
| 2023 | 20 |
| 2022 | 16 |
| 2021 | 20 |
| 2020 | 26 |
| 2019 | 22 |
| 2018 | 22 |
| 2017 | 10 |
| 2016 | 24 |
| 2015 | 22 |
| 2014 | 11 |
| 2013 | 17 |
| 2012 | 10 |
| 2011 | 10 |
| 2010 | 4 |
| 2009 | 14 |
| 2008 | 8 |
| 2007 | 6 |
| 2006 | 13 |
| 2005 | 4 |
| 2004 | 1 |
| 2003 | 7 |
| 2002 | 3 |
| 2001 | 3 |
| 2000 | 6 |
Use grade D
30 to 99 papers from 2021 to 2025.
Evidence grade D
Little published measurement evidence found.
- Papers on test-retest reliability
- 0
- Papers on reliability or validity
- 3
- Registered clinical studies using it as an outcome
- 0
Note on the counts: includes theoretical linguistics papers.
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 0.22.0, counted 3 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Sentences
- Sentences written for this test, each about two people: a verb, then "because he" or "because she" and an ending (Zoe wrote to Carl because she was homesick). They are English in every session.
- Verbs
- 24 verbs from Ferstl, Garnham and Manouilidou's (2011) English norms: 12 that lead readers to expect the cause in the person named first (apologized to) and 12 in the person named second (thanked), each with a bias of at least 65 points of 100 and a gender effect under 15 points.
- Kinds
- Each verb appears twice: once with an ending that makes the pronoun refer to the person the verb favours and once with one that makes it refer to the other, once with two people of the same gender and once with a man and a woman. Two lists drawn from the session's seed decide the pairing, so a session has 12 sentences of each of the four kinds.
- Trial
- A cross for 500 ms, then the sentence with its pronoun in bold and both names under it; F for the person named first and J for the person named second, within 10,000 ms of the sentence's onset.
- Session
- 6 practice sentences with feedback, then 48 with none, with no more than 3 answers on one key in a row and a verb's two sentences at least 12 apart.
- Names
- Common first names whose gender is plain, 12 men's and 12 women's, drawn from the seed.
- Scoring
- The consistency effect is the median time of right answers to sentences whose pronoun goes against the verb minus those whose pronoun goes with it. Under 70% right is invalid.
The sentences were written for this test, not taken from Caramazza and colleagues' (1977) set, and no ending names a job, a role or anything people tie to one gender. Answers are keys, where Caramazza and colleagues' participants said the name aloud. Results describe how pronouns were linked in this task, a research measure.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: pronoun_resolution_test from the scored run, F for the person named first or J for the person named second, within 10 seconds of the sentence. Every sentence has a right answer (chance 50%, half on each key). Below chance invalidates and not above chance is flagged; one key on 90% of answers invalidates when accuracy is not above chance. Under 70% right invalidates. A session in a language other than English is flagged for review. Missed responses and responses under 100 ms follow the timed-test rule.
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). 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. Caramazza, Grober, Garvey and Yates (1977) found the pronoun assigned against the sentence's meaning on 3.3% of trials, so under 70% right means the sentences were not being read; that limit is a ConductScience choice. 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.
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 Pronoun Resolution Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Caramazza, Grober, Garvey & Yates, 1977; Ferstl, Garnham & Manouilidou, 2011). The main score was consistency effect (median time of right answers to sentences whose pronoun goes against the verb's expected cause minus sentences whose pronoun goes with it, from the sentence's onset).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Caramazza, A., Grober, E., Garvey, C., & Yates, J. (1977). Comprehension of anaphoric pronouns. Journal of Verbal Learning and Verbal Behavior, 16(5), 601-609. https://doi.org/10.1016/S0022-5371(77)80022-4
- Ferstl, E. C., Garnham, A., & Manouilidou, C. (2011). Implicit causality bias in English: A corpus of 300 verbs. Behavior Research Methods, 43(1), 124-135. https://doi.org/10.3758/s13428-010-0023-2The verbs and their biases, from completions of fragments such as "John liked Mary because..."
- Garvey, C., Caramazza, A., & Yates, J. (1974). Factors influencing assignment of pronoun antecedents. Cognition, 3(3), 227-243. https://doi.org/10.1016/0010-0277(74)90010-9The implicit causality of verbs, from completions of "because he" fragments.
- Hartshorne, J. K., & Snedeker, J. (2013). Verb argument structure predicts implicit causality: The advantages of finer-grained semantics. Language and Cognitive Processes, 28(10), 1474-1508. https://doi.org/10.1080/01690965.2012.689305Implicit causality measured on the web by choosing who a pronoun refers to.
Source papers
- Caramazza A, Grober E, Garvey C, Yates J (1977). Comprehension of anaphoric pronouns. Journal of Verbal Learning and Verbal Behavior, 16(5):601-609.Paradigm source, Adults
- Ferstl EC, Garnham A, Manouilidou C (2011). Implicit causality bias in English: a corpus of 300 verbs. Behavior Research Methods, 43(1):124-135.Paradigm source, Adults
- Garvey C, Caramazza A, Yates J (1974). Factors influencing assignment of pronoun antecedents. Cognition, 3(3):227-243.Paradigm source, Adults and children aged 10 to 13
Questions
How long does the Pronoun Resolution Task take?
About 4 to 6 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?
Consistency Effect, Gender Effect, Consistent Sentence Time, Inconsistent Sentence Time, Same Gender Time, Different Gender Time, Accuracy, Consistent Accuracy, Inconsistent Accuracy, Same Gender Accuracy, Different Gender Accuracy, Missed Sentences, Anticipations and Session In Another Language. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Pronoun Resolution Task used in research?
91 papers from 2021 to 2025 name it in their title or abstract, down 12% compared with 2016 to 2020. That ranks it 227 of 495 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.22.0, counted 3 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Pronoun Resolution Task have?
In the published research, 0 papers report its test-retest reliability and 0 registered clinical studies use it as an outcome (evidence grade D 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.