Language
Sentence Verification Task
Also called: sentence-picture verification, picture-sentence verification, Clark and Chase task, sentence-picture comparison
Measure how quickly people check a sentence against a picture, with negation and above or below varied, in the participant's own language.
8-11 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. Preview any test free after you sign up.
- Papers, 2021 to 2025
- 53
- Evidence grade
- C
- Registered clinical studies
- 3
What it measures
Measures how quickly people check a sentence against a picture. Participants read a short sentence about a star and a plus sign, such as "The star is not above the plus sign", then press F or J to say whether it is true of a picture of one above the other. The sentences vary in negation and in above or below, and are shown in the participant's language.
Constructs
- Sentence comprehension
- Processing of negation
- Comparing language with what is seen
Research uses
- Language comprehension research
- Individual differences in verbal and spatial strategies
- Within-study group comparison
How the paradigm works
Sentence verification measures how quickly a person checks what a sentence says against what a picture shows. In the version of Clark and Chase (1972), a sentence such as "The star is not above the plus sign" is compared with a picture of a star and a plus sign, one above the other. Sentences vary in their subject (star or plus sign), their relation (above or below) and negation, and each is paired with both pictures, so half the pairs are true. In their first experiment, Clark and Chase estimated that a negation added 685 ms, a mismatch between the sentence and the picture 187 ms, and below rather than above 93 ms, and showed that these times add up, as separate mental operations would.
More about the paradigm
This test uses the two response times of MacLeod, Hunt and Mathews (1978). The participant reads the sentence at their own pace and presses the space bar, which gives the reading time; 500 ms later the picture appears, and the participant presses F or J for true or false within 5,000 ms, which gives the verification time. Which key means true is drawn from the session's random seed. 16 practice trials with feedback come first. The test then runs each of the 16 sentence-picture pairs 6 times (96 trials) by default, or 4 or 8 times as a setting, in a seeded order with no more than 3 of one answer in a row and a short break halfway. The sentences are written in each of the eight languages, so the test runs in the participant's language. Scores support research and do not diagnose.
The main score is the median verification time of correct answers. In MacLeod and colleagues' 70 students, mean verification times ran from 773 ms for true affirmatives to 1,195 ms for false negatives, split-half reliability was .97 to .99, and faster verification went with higher verbal and higher spatial ability. The constituent comparison model of Carpenter and Just (1975) fitted most of them, but a smaller group was fitted poorly and appeared to compare pictures rather than sentences; the model fit is reported for each session.
Scores
- Median Verification Timems
- Median time from the picture's onset to a correct true or false answer, leaving out answers under 100 ms.
- Lower means faster checking of sentences against pictures
- Percent Correct%
- Test trials answered correctly; a picture left unanswered counts as wrong.
- Higher means more accurate
- Median Reading Timems
- Median time from a sentence's onset to the space bar, on trials answered correctly.
- Describes reading pace, which differs between languages
- Negation Effectms
- Median verification time after negative sentences minus after affirmative ones, on correct answers.
- Positive means negative sentences took longer to verify, the usual result
- Negation Effect in Readingms
- Median reading time of negative sentences minus affirmative ones, on correct answers.
- Positive means negative sentences took longer to read
- Below Minus Abovems
- Median verification time after sentences with below minus after those with above, on correct answers.
- Positive means below took longer, as in Clark and Chase (1972)
- True Affirmative Timems
- Median verification time on true affirmative trials; needs at least 5 correct answers.
- Lower means faster
- False Affirmative Timems
- Median verification time on false affirmative trials; needs at least 5 correct answers.
- Lower means faster
- True Negative Timems
- Median verification time on true negative trials; needs at least 5 correct answers.
- Lower means faster
- False Negative Timems
- Median verification time on false negative trials; needs at least 5 correct answers.
- Lower means faster
- Negation by Truth Interactionms
- True negative minus true affirmative time, minus false negative minus false affirmative time.
- Positive means negation slowed true answers more than false ones, as in MacLeod and colleagues (1978)
- Constituent Comparison Fitscore
- Correlation of the four trial-type times with the comparisons of Carpenter and Just (1975): 0, 1, 4 and 5 for true affirmative, false affirmative, false negative and true negative.
- Near 1 fits a sentence-based strategy; near 0 or below suggests a picture-based one (MacLeod and colleagues 1978)
- Time per Comparisonms
- Least-squares slope of the four trial-type times on those comparisons.
- Higher means each comparison took longer; read it with the fit
- Answers Under 100 mscount
- Answers within 100 ms of the picture's onset.
- Expected to be 0
- Missed Answerscount
- Pictures left unanswered for 5,000 ms.
- Lower means fewer missed answers
- Test Trials Recordedtrials
- Test trials in the scored run.
- Fewer than 90% of those set is invalid
- Test Trials Settrials
- Test trials the session set out to show.
- 64, 96 or 128
Use and evidence
53 papers from 2021 to 2025 name this task in their title or abstract, down 18% compared with 2016 to 2020. That ranks it 251 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 9 |
| 2024 | 8 |
| 2023 | 13 |
| 2022 | 10 |
| 2021 | 13 |
| 2020 | 12 |
| 2019 | 9 |
| 2018 | 12 |
| 2017 | 17 |
| 2016 | 15 |
| 2015 | 12 |
| 2014 | 19 |
| 2013 | 16 |
| 2012 | 13 |
| 2011 | 12 |
| 2010 | 8 |
| 2009 | 16 |
| 2008 | 3 |
| 2007 | 9 |
| 2006 | 2 |
| 2005 | 6 |
| 2004 | 6 |
| 2003 | 6 |
| 2002 | 9 |
| 2001 | 5 |
| 2000 | 8 |
Use grade D
30 to 99 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
- 1
- Papers on reliability or validity
- 48
- Registered clinical studies using it as an outcome
- 3
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 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Items
- 8 sentences from every combination of subject (star or plus sign), relation (above or below) and negation, each paired with both pictures: 16 pairs, half true, 4 in each trial type.
- Languages
- The sentences are written in each of the eight participant languages, so the whole test runs in the participant's language.
- Trial
- A dot for 500 ms, then the sentence until the space bar (reading time), a blank for 500 ms, then the picture until F or J for true or false, for up to 5,000 ms (verification time).
- Keys
- Which of F and J means true is drawn from the session's seed and saved, so half the sessions answer true with each hand.
- Session
- 16 practice trials with feedback, then each pair 6 times (96 trials) by default, or 4 or 8 times as a setting, with no feedback, a break halfway and no more than 3 of one answer in a row.
- Scoring
- Median verification time of correct answers is the main score. Accuracy, reading time, the negation and below-minus-above effects, the four trial types and the fit to the constituent comparison model are also reported.
The sentences and pictures were made for this test from the design of Clark and Chase (1972). Results from different languages are compared within a language.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: sentence_verification_test from the scored run, F or J for true or false within 5,000 ms of the picture (chance 50%, half the pairs are true), correctness from the key for the pair. Below chance invalidates and not above chance is flagged. One key on 90% of answers invalidates when accuracy is not above chance. Missed responses and responses under 100 ms follow the timed-test rule. A run with fewer than 90% of its planned trials is invalid.
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. A median over a much shorter run cannot be compared with one over the full run. The sentences are shown in the session's language, so a session in any of the eight languages is scored the same way.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Test trials | 96 trials (standard) | 64 trials, 96 trials (standard), 128 trials |
Yearly plans can change settings within these limits. Free and pay-as-you-go studies use the standard settings.
Need a setting this test does not have?
Devices and languages
- Timed test: participants need a desktop or laptop with a mouse or trackpad. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Sentence Verification Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Clark & Chase, 1972; MacLeod, Hunt & Mathews, 1978), with the standard settings (test trials: 96 trials). The main score was median verification time (median time from the picture's onset to a correct true or false answer, leaving out answers under 100 ms).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Clark, H. H., & Chase, W. G. (1972). On the process of comparing sentences against pictures. Cognitive Psychology, 3(3), 472-517. https://doi.org/10.1016/0010-0285(72)90019-9
- Carpenter, P. A., & Just, M. A. (1975). Sentence comprehension: A psycholinguistic processing model of verification. Psychological Review, 82(1), 45-73. https://doi.org/10.1037/h0076248
- MacLeod, C. M., Hunt, E. B., & Mathews, N. N. (1978). Individual differences in the verification of sentence-picture relationships. Journal of Verbal Learning and Verbal Behavior, 17(5), 493-507. https://doi.org/10.1016/S0022-5371(78)90293-1
- Roberts, M. J., Wood, D. J., & Gilmore, D. J. (1994). The sentence-picture verification task: Methodological and theoretical difficulties. British Journal of Psychology, 85(3), 413-432. https://doi.org/10.1111/j.2044-8295.1994.tb02532.x
Source papers
- Clark HH, Chase WG (1972). On the process of comparing sentences against pictures. Cognitive Psychology, 3(3):472-517.Paradigm source, Adults
- MacLeod CM, Hunt EB, Mathews NN (1978). Individual differences in the verification of sentence-picture relationships. Journal of Verbal Learning and Verbal Behavior, 17(5):493-507.Paradigm source, Young adults
Questions
How long does the Sentence Verification Task take?
About 8 to 11 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?
Median Verification Time, Percent Correct, Median Reading Time, Negation Effect, Negation Effect in Reading, Below Minus Above, True Affirmative Time, False Affirmative Time, True Negative Time, False Negative Time, Negation by Truth Interaction, Constituent Comparison Fit, Time per Comparison, Answers Under 100 ms, Missed Answers, Test Trials Recorded and Test Trials Set. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Sentence Verification Task used in research?
53 papers from 2021 to 2025 name it in their title or abstract, down 18% compared with 2016 to 2020. That ranks it 251 of 395 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Sentence Verification Task have?
In the published research, 1 paper reports its test-retest reliability and 3 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Which settings can a researcher change?
Test trials (64 trials, 96 trials (standard), 128 trials). Settings can be changed on yearly plans; free and pay-as-you-go studies use the standard settings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese, with its words in the same language. 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.