Refixation probability is the proportion of eligible word encounters in which the reader makes another eye fixation on the same word after the initial fixation, under a clearly defined rule. Classic research found that immediate refixations in English reading are least frequent when the initial landing is near the word’s centre and become more frequent toward its edges. It is an eye-movement statistic, not a personal measure of intelligence or comprehension.
Did you know that a longer word can sometimes receive two fixations because the first landing did not provide enough clear visual information? The eye may make a short corrective movement before continuing. That does not mean the reader has forgotten the word. Vocabulary recognition depends on how visual information is sampled and connected to familiar spelling and meaning. A second glance can be efficient when it improves accuracy.
This guide explains the 1989 McConkie–Kerr–Reddix–Zola–Jacobs findings, the relation between landing location and a U-shaped refixation curve, and how refixation probability differs from gaze duration, word skipping and backward sentence regressions. It includes a fictional count calculation, references to oculomotor models, and practical lessons for Primary and Secondary English, Mathematics, Science, History, GP and translation.
Editorial note: All arithmetic examples are invented and no eye-tracking data were collected for this article. The published research is cited with its defined experimental setting.
Contents: why another fixation can be part of proficient reading
Open 51 chapters, mathematics and teaching checks
- 1. A refixation is another fixation on a word
- 2. Why vocabulary recognition may need another glance
- 3. A probability needs a denominator
- 4. The first fixation sets the start of the comparison
- 5. What an immediate refixation means
- 6. McConkie and colleagues’ 1989 result
- 7. The curve is often described as U-shaped
- 8. A parabola is a model of the relationship
- 9. Why edge landings can demand more information
- 10. The optimal viewing position is related
- 11. Preferred viewing location is another construct
- 12. A long word can invite more than one fixation
- 13. Word frequency can also influence gaze
- 14. Launch site influences the landing
- 15. Saccadic targeting is not perfectly accurate
- 16. Mislocated fixations provide another account
- 17. A corrective movement can be fast
- 18. Refixation probability is not gaze duration
- 19. A refixation adds to first-pass time
- 20. First fixation duration is a narrower measure
- 21. Single fixation duration uses selected trials
- 22. Regression and refixation are not synonymous
- 23. Later rereading differs from immediate refixation
- 24. Regression path duration can include other words
- 25. Word skipping has its own denominator
- 26. A fictional thirty-two-trial calculation
- 27. The effect of adding excluded trials
- 28. The same probability can conceal different times
- 29. A group curve is not an individual forecast
- 30. A word can be familiar but visually difficult
- 31. Word boundaries depend on script
- 32. The reading task changes fixation behaviour
- 33. Proofreading can encourage reinspection
- 34. Inverted optimal-viewing effects need care
- 35. The two central curves answer different questions
- 36. Data quality and landing accuracy matter
- 37. Automatic classification can make errors
- 38. Why multiple comparisons require restraint
- 39. Primary English: the goal is not zero refixations
- 40. Primary English: check the whole word
- 41. Secondary English: abstract words may deserve inspection
- 42. Secondary English: sentence reanalysis is another issue
- 43. Mathematics: numbers require exact positions
- 44. Science: technical suffixes may be decisive
- 45. History: names and dates justify a second check
- 46. General Paper: reread the exact claim
- 47. Translation: a second visual check protects meaning
- 48. Workplace identifiers need full verification
- 49. A classroom diagram can show the two stops
- 50. A six-field research record
- 51. The final distinction: one more look can be useful
Quick test: refixation is a probability, not a duration
| Measure | Definition | Usual unit |
|---|---|---|
| Refixation probability | Eligible initially fixated target occurrences with a second first-pass fixation divided by eligible occurrences | Proportion or percentage |
| First fixation duration | Time of first target fixation | Milliseconds |
| Gaze duration | Sum of first-pass fixations on the target before first exit | Milliseconds |
| Word-skipping probability | Eligible target occurrences passed without first-pass fixation divided by eligible appearances | Proportion or percentage |
| Regression path duration | Fixation time until first forward exit, including some backward detours under its rule | Milliseconds |
A 25% refixation probability and a 250-ms gaze duration answer different questions. They must not be compared as if they represented the same kind of measurement.
1. A refixation is another fixation on a word
A refixation is a further eye fixation on a word after an initial fixation, often before the reader first moves away from that word. Refixation probability describes how often such an additional fixation occurs across eligible trials under the study’s rules.
The statistic counts events, not how much the reader comprehended. A refixation may help gather more visual information or reflect a landing location that left useful letters difficult to identify. It is therefore an informative property of reading behaviour, not a personal grade.
2. Why vocabulary recognition may need another glance
Printed words vary in length, spelling familiarity and the number of other lexical entries they resemble. A first fixation may supply enough information for one word yet leave ambiguity about another, making an additional fixation useful.
The eyes are part of the vocabulary-recognition system. They sample the printed form while the reader connects letters to sound and meaning. But an extra fixation is not automatically a sign of a vocabulary deficit; a skilled reader can refixate when the visual or linguistic conditions require it.
3. A probability needs a denominator
To calculate a refixation probability, the researcher needs a clearly specified set of eligible initially fixated word occurrences and a rule for what counts as an immediate within-word refixation.
If 8 of 32 fictional eligible cases receive a second fixation before first leaving the word, the observed proportion is 8/32 = 0.25, or 25%. That arithmetic is invented for teaching, not an empirical norm. It demonstrates why the counting rule matters.
4. The first fixation sets the start of the comparison
A target must usually receive an initial fixation to be eligible for an immediate within-word refixation analysis. The location of that first fixation can help predict whether the eyes land on the same word again.
This differs from word skipping, in which no direct fixation occurs on the first pass. Treating skipped cases as ordinary non-refixated trials can alter the denominator and the meaning of the statistic. A research report should state which trials were eligible.
5. What an immediate refixation means
An immediate refixation commonly refers to moving from the first fixation on a word to another fixation on that same word without first looking away to another word. The movement may be relatively small compared with an interword saccade.
Later revisits after the reader has already left the region form a different class of event. They may reflect sentence reanalysis rather than initial visual sampling. The reader should identify the exact temporal relationship before equating all second visits.
6. McConkie and colleagues’ 1989 result
George McConkie, Paul Kerr, Michael Reddix, David Zola and Arthur Jacobs analysed more than 40,000 fixations in research published in 1989. They examined how immediate word refixation frequency depended on where the initial fixation landed.
The proportion was lowest near the word’s centre and increased toward the edges in the studied adult reading data. The relation was well described by a parabolic curve. That is evidence of a structured eye-movement pattern, not an instruction for every pupil to fixate the centre intentionally.
7. The curve is often described as U-shaped
A conceptual plot places initial landing position along the horizontal axis and refixation probability on the vertical axis. A central minimum and higher probabilities at word edges make a U-shaped pattern under the reported conditions.
The exact measured curve depends on word length, participant sample and task. A teacher can draw the qualitative shape without inserting invented percentages from the original experiment. The important observation is that an edge landing may encourage another fixation.
8. A parabola is a model of the relationship
The 1989 authors reported that a parabolic function fit their refixation data well. In mathematical terms, a quadratic curve can represent probability rising as the initial position moves away from a central optimum.
The fitted curve is not a universal formula for every English word. It is an empirical approximation in a defined dataset. A student can learn the shape without being told that all real reading obeys an exact textbook parabola.
9. Why edge landings can demand more information
A fixation near the beginning of a word leaves later letters farther from central vision; one near the end leaves early letters farther away. A second fixation may improve access to the portion insufficiently sampled initially.
This is a useful perceptual explanation of some refixations. It does not follow that every letter has equal informational value, or that the central fixation always identifies a word. Word-specific spelling patterns and lexical alternatives can influence what needs to be inspected.
10. The optimal viewing position is related
The optimal viewing position is a place within a word that supports relatively efficient identification in specified experimental tasks. It is often near the centre for many isolated-word settings, though information distribution can alter the optimum.
Refixation probability can help evaluate whether a landing location was sufficient for recognition, but an OVP and an observed probability curve are not identical constructs. One concerns efficient viewing; the other counts additional eye movements.
11. Preferred viewing location is another construct
Preferred viewing location describes where readers’ first fixations naturally tend to land, often around or slightly to the left of the centre of words in English reading.
The typical landing location need not equal the point of minimum refixation for every word. Launch-site geometry and motor variability can shift actual saccade endpoints. Distinguishing preferred from optimal avoids a familiar error in eye-tracking explanations.
12. A long word can invite more than one fixation
As a word becomes longer, one fixation may not provide sufficient high-resolution access to all its letters. Additional within-word fixations can therefore be more common in long targets under certain reading conditions.
Word length is not a direct measure of conceptual difficulty. A student might know a long frequently encountered school term better than a short unusual verb. The eye’s sampling needs and the learner’s semantic knowledge should be assessed separately.
13. Word frequency can also influence gaze
Frequently encountered words may be identified efficiently in some reading tasks, while low-frequency items can require additional processing. Refixation probability can vary with lexical factors as well as viewing location.
A researcher who compares common short words with rare long words cannot attribute a difference to frequency alone. Length, launch site, familiarity and contextual predictability must be considered. A careful analysis distinguishes these variables.
14. Launch site influences the landing
The launch site is the position of the previous fixation from which the eye began its movement toward the target. Eye-movement range error means the eventual landing depends in part on that starting distance.
If the first landing is less favourable, refixation probability may rise. The 1988 McConkie and colleagues study found systematic relations between launch site and landing distributions. Refixation is therefore partly an outcome of oculomotor geometry, not merely a conscious desire to reread.
15. Saccadic targeting is not perfectly accurate
An eye movement aimed toward one location can land short or long because of systematic range error and ordinary variability. The reader may need a corrective fixation inside the word before continuing.
These events do not necessarily imply that the brain failed to understand the word’s definition. A visual motor correction and a semantic repair are distinct possibilities. Research must evaluate the observed path and task conditions before assigning a mechanism.
16. Mislocated fixations provide another account
Researchers studying mislocated fixations have proposed that a fixation may accidentally land on an unintended word, triggering a rapid corrective movement. Such events can contribute to observed refixation and skipping distributions.
Nuthmann, Engbert and Kliegl’s work on self-consistent estimation of mislocated fixations examined how targeting error can help explain reading eye movements. This complicates explanations that assume each fixation was placed deliberately to resolve a vocabulary problem.
17. A corrective movement can be fast
A short fixation followed by another movement may occur because the reader quickly recognises a mislocated landing or because an eye-movement program was already in preparation.
A small fixation duration should therefore not automatically be interpreted as effortless word understanding. Motor corrections can produce fast events that need separate analysis. The measure is powerful only when its underlying movement sequence is considered.
18. Refixation probability is not gaze duration
Refixation probability is the fraction of eligible target encounters receiving another fixation. Gaze duration is the total time spent fixating that word before the first exit.
The two are related but have different units. More second fixations can raise average gaze duration even if individual fixations are relatively short. A probability of 0.25 is not equivalent to 250 milliseconds or any other duration.
19. A refixation adds to first-pass time
Suppose a fictional word receives a 160-ms fixation and a 140-ms refixation before the eyes leave. The word’s gaze duration is 300 ms, and that trial counts as a refixation under the stated immediate rule.
Another fictional word receiving only one 220-ms fixation has gaze duration 220 ms but no immediate refixation. The examples demonstrate how counts and times are separate. Neither fictional reader should be ranked by intelligence from those values.
20. First fixation duration is a narrower measure
First fixation duration is the time of the first fixation on a word. Whether a refixation follows does not change that first fixation’s observed duration.
Gaze duration can be larger because it includes additional first-pass within-word fixations. A research report should avoid treating first fixation and gaze duration as independent evidence when they share the same initial event. Statistical planning matters.
21. Single fixation duration uses selected trials
Single fixation duration is commonly computed only on cases where the word received exactly one fixation on its first pass. Trials with immediate refixations are excluded from that particular measure.
This creates a selection condition. A comparison of single fixation durations may concern a different subset of words than a comparison of gaze durations. A student reading tables should ask which trials are represented in each number.
22. Regression and refixation are not synonymous
A regression often means an eye movement backward to earlier text. A refixation can be another stop within the same word without moving to a previous word.
A reader can refixate a long word before moving on, or regress to an earlier sentence after encountering a confusing pronoun. The two paths may arise for different reasons. Accurate use of terms helps researchers choose appropriate measures.
23. Later rereading differs from immediate refixation
A word can be fixated, left behind, and revisited after the reader processes another phrase. That later return is part of second-pass or total viewing behaviour under many definitions.
The original immediate refixation probability concerns a different moment. Combining all later returns with within-word corrections can blur recognition and sentence reanalysis. A study must describe its event classification.
24. Regression path duration can include other words
Go-past or regression path duration may include fixation time spent on earlier words before gaze first progresses beyond the target region. That calculation can be much larger than the target’s first-pass gaze duration.
It is a duration rather than a probability, and it captures a different route through the text. A reader who can distinguish all three measures—refixation probability, gaze duration and go-past time—is ready to interpret many eye-tracking tables.
25. Word skipping has its own denominator
A skipped target has no first-pass fixation. Its occurrence contributes to a word-skipping analysis but may be excluded from conditional refixation probability.
A study could report one proportion among all eligible target appearances and another among only initially fixated words. Both can be valid if labelled clearly. Swapping denominators without explanation changes the scientific claim.
26. A fictional thirty-two-trial calculation
Imagine 32 fabricated valid trials in which the target was initially fixated. Eight of them included a second within-word fixation before the eyes left the target. Under the stated rule, refixation probability equals 8 divided by 32.
The result is 0.25 or 25%. If there had also been eight skipped target presentations, those would need a separate policy; they would not automatically enter the denominator of the initially fixated set. The example demonstrates operational precision.
27. The effect of adding excluded trials
Suppose the fictional study recorded 40 presentations but 8 were skipped or invalid for the chosen refixation analysis. Then the eligible denominator is 32, not 40. Reporting 8/40 would answer a different question.
This illustrates why scientific tables need inclusion rules. The raw count of refixations is not sufficient. A tutor can use the arithmetic to teach fractions, conditional probability and honest statistical reporting.
28. The same probability can conceal different times
Two groups might each have a 25% refixation rate while differing in how long the first and second fixations lasted. Another comparison could have different rates but similar average gaze durations.
Probability and time therefore complement one another. The researcher should inspect both the event counts and fixation distributions if the hypothesis concerns reading effort. One summary statistic never describes the entire eye path.
29. A group curve is not an individual forecast
The 1989 central-minimum curve was estimated from many college readers’ fixations. It does not guarantee that an individual pupil will refixate a word every time their gaze lands near its edge.
Group patterns include variability across words, people and conditions. A child should not be given a personal learning diagnosis on the basis of one supposedly suboptimal landing. Appropriate education uses multiple meaningful assessments.
30. A word can be familiar but visually difficult
Even a well-known word may be displayed in a small font, unusual typeface or low contrast. Such conditions can reduce the clarity of its visual form and affect refixation patterns.
A teacher should check accessibility and text presentation before attributing extra fixations to poor vocabulary. The reading environment is part of the task. Better legibility may help without changing the learner’s underlying lexical knowledge.
31. Word boundaries depend on script
English usually uses spaces to separate words, allowing researchers to define many target regions readily. Other writing systems may not use equivalent spacing, or may arrange text in different directions.
A refixation must be defined relative to units meaningful for that script and research question. Comparing English word-based probabilities directly with a character-based analysis in another language can be misleading. The unit belongs in the methods.
32. The reading task changes fixation behaviour
A participant searching for a particular name may make different eye movements from someone reading a story for comprehension. The need to inspect detailed letters can vary with the goal.
Thus refixation probability is a conditional behavioural outcome, not a universal measure of word quality. An experiment must state whether participants were reading, searching, proofreading or performing another task. The interpretation follows the goal.
33. Proofreading can encourage reinspection
When proofreading, readers may deliberately check exact spellings, punctuation or technical terms. Additional fixations can be appropriate if they help locate an error.
A low refixation rate would not automatically be desirable under this task. The aim is accurate detection and correction. The same eye movement can have different practical value depending on whether the person is reading a novel or verifying a legal name.
34. Inverted optimal-viewing effects need care
Some studies of first fixation duration find a counterintuitive relation in which fixations near a word’s centre can be longer than those near its boundaries. This is discussed as an inverted optimal viewing position effect.
One explanation involves rapidly corrected mislocated fixations near edges. A shorter edge fixation can therefore reflect a correction rather than superior recognition. The distinction between refixation probability and fixation duration prevents a misleading interpretation of the curve.
35. The two central curves answer different questions
The probability of making another fixation may be lowest near the centre, even while the duration of the first fixation can follow another pattern. One curve concerns whether a subsequent event occurs; the other concerns the length of the initial event.
There is no contradiction when the variables differ. A child learning research vocabulary should name the vertical axis and denominator before announcing which landing position is ‘best’. The definition of the outcome is the foundation of interpretation.
36. Data quality and landing accuracy matter
Eye trackers estimate fixation coordinates with measurement error. A gaze point near the space between two words may be assigned to one target or the other depending on calibration and region rules.
That affects both refixation classification and calculated landing position. Good research reports valid trials, filtering and appropriate calibration procedures. A precise-looking decimal does not make a poorly measured eye trace reliable.
37. Automatic classification can make errors
Analysis software may group very short fixations, remove tracking loss or define regions automatically. These operations can change whether an apparent second stop counts as a genuine refixation.
Researchers should inspect the tool’s operational definitions and validate important classifications. A school article cannot infer a child’s refixation probability from ordinary webcam footage. Controlled eye tracking is required for such a claim.
38. Why multiple comparisons require restraint
Eye-tracking papers often test first fixation duration, gaze duration, refixation probability, skipping and total time. The outcomes share observations and may be statistically dependent.
Research on false positives warns against interpreting every nominally significant result as independent confirmation. A strong analysis identifies its hypotheses, reports relevant measures and controls statistical risk. Methodological quality matters more than a long list of impressive numbers.
39. Primary English: the goal is not zero refixations
A young reader may need to inspect a longer unfamiliar word in two places before recognising it. That additional eye movement can support accurate decoding.
Teach spelling patterns, sound–letter correspondences and meaning. Do not demand that the child stare at one fixed point or race through new words. A successful pupil can identify the complete form and use it correctly, regardless of an unmeasured refixation pattern.
40. Primary English: check the whole word
A student might infer a familiar word from its beginning but miss a contrasting ending. Rechecking the full spelling can prevent a false interpretation.
Ask for a paraphrase of the sentence and a new example. The educational task is to use visual evidence to read correctly. A refixation is useful when it improves that result; the fixation count is not itself a grade.
41. Secondary English: abstract words may deserve inspection
A Secondary learner may meet inference and interference in similar subject contexts. Their overlapping letters can create confusion when read quickly.
Teach the difference in spelling, grammatical role and meaning. A second look can be sensible. The final assessment should ask which concept the author meant rather than whether the learner’s eye moved twice.
42. Secondary English: sentence reanalysis is another issue
A student can recognise a long word perfectly yet need to reread an earlier clause because a later pronoun is unclear. That backward movement is not a simple immediate within-word refixation.
Distinguish visual recognition from discourse integration. A teacher should offer clause and reference analysis when the latter is the problem. Using accurate eye-movement terminology helps avoid prescribing the wrong lesson.
43. Mathematics: numbers require exact positions
A number with several digits can be misread if one place value is overlooked. Additional inspection may be appropriate to verify the complete value.
Ask students to check the digits, units and operation directly. A mathematics solution cannot be validated by a reading-speed measure. Numerical precision is the learning objective, not an artificial minimum number of eye stops.
44. Science: technical suffixes may be decisive
Scientific words can share roots while differing in suffixes that name distinct processes or fields. An extra fixation on the relevant part of a term can help the reader avoid a false substitution.
The tutor then teaches the exact concept and mechanism. An accurate scientific explanation requires more than visual inspection, but visual accuracy supports the correct lexical choice. Do not replace the word with an unverified near-match.
45. History: names and dates justify a second check
Historical sources may contain similar surnames, places or dates. A careful student can legitimately look twice to distinguish them.
A teacher should reward faithful transcription and source verification. A higher refixation count would not show poor History knowledge without other evidence. The record’s accuracy is the meaningful outcome.
46. General Paper: reread the exact claim
An argument may use a familiar phrase with a subtle exception or qualification. Looking again at that wording can prevent an exaggerated paraphrase.
Students should identify the claim, scope and supporting evidence. Whether the eyes refixated a word is secondary to whether the resulting interpretation is defensible. Vocabulary supports critical reasoning through precision.
47. Translation: a second visual check protects meaning
A source term can resemble a different valid term, particularly across related languages or within technical documents. A translator who rereads the spelling may avoid a false friend or a changed referent.
The target sentence must preserve the original meaning, agency and conditions. Eye-tracking research can describe behaviour, but it cannot certify translation equivalence. The correct form and concept remain the acceptance test.
48. Workplace identifiers need full verification
An equipment code or customer name may differ from another in one character. A worker may need a second inspection to prevent a costly mismatch.
Use the authorised record and the current procedure. No research finding about refixation minima justifies skipping that verification. Vocabulary and visual recognition are technologies for reliability when stakes are high.
49. A classroom diagram can show the two stops
Draw a word with one dot near its beginning and another inside the same printed region. Explain that a genuine refixation measure would require the dots to represent separately recorded fixations on a valid trial.
The picture is only a schematic. It does not demonstrate which eye movement a real reader made or why. Students can still learn the counting logic and ask what additional evidence would support a psychological explanation.
50. A six-field research record
Record the target, whether it received an initial fixation, the landing location, whether an immediate second fixation occurred, the word length and task condition. Add participant group, tracking quality and count denominators.
This creates a defensible probability rather than an opaque number. A researcher can then examine whether landing position, frequency or task predicts refixation while controlling other variables. Definitions and data must precede interpretation.
51. The final distinction: one more look can be useful
Refixation probability is a measure of how often readers make another fixation on the same word under a specified rule. Classic studies reveal a strong relationship with where the initial fixation landed, along with other visual and lexical influences.
The educational objective is not to force the eyes to stop only once. It is to recognise the correct word and understand its meaning. A second look that prevents a mistake can be part of proficient reading.
Full worked case: eight extra fixations among thirty-two trials
| Fictional count | Number | What it means |
|---|---|---|
| Initially fixated, valid target occurrences | 32 | Denominator for the stated conditional measure |
| Targets immediately refixated | 8 | Numerator |
| Targets not immediately refixated | 24 | Other eligible trials |
| Refixation probability | 8/32 = 0.25 | 25% in this fictional example |
This example is invented solely to teach how the statistic is constructed. If another eight presentations were skipped, they would need their own defined treatment rather than automatically entering the denominator. The original McConkie study analysed real adult eye movements with a much larger sample.
Eight-stage research and vocabulary audit
- Define the target word region and whether the analysis concerns immediate first-pass refixation.
- Identify the valid trials in which the target was initially fixated.
- Record the location of the initial fixation within the word.
- Count whether a second fixation occurred on the same word before leaving it.
- Calculate the fraction with a stated denominator and distinguish it from skipped trials.
- Check word length, frequency, launch distance and display legibility as possible influences.
- Review whether the claimed curve is an actual measured group relation or just a teaching sketch.
- Evaluate vocabulary comprehension separately by asking for a definition, context and new sentence.
Frequently asked questions
1. What is refixation probability?
The fraction of eligible target encounters with another fixation on the same word under a specified rule.
2. Is a refixation the same as a backward regression?
No. A refixation can remain within the same word; a regression usually moves to earlier text.
3. Why are edge landings often refixated?
They can leave some letters less clearly sampled, encouraging another fixation.
4. What did the 1989 study find?
A low refixation rate near the word centre with rising rates toward both ends in the studied data.
5. Is the curve exactly the same for every word?
No. The original parabolic fit describes a group relationship under specific conditions.
6. Does one refixation imply weak vocabulary?
No. Visual sampling and eye-movement variability can explain extra stops.
7. Does a longer word always need a second fixation?
No. Length affects probability, not a guaranteed response.
8. Why exclude skipped words from some denominators?
They did not receive an initial fixation, which a conditional refixation analysis requires.
9. Is refixation probability the same as gaze duration?
No. One is a proportion of events; the other is a sum of fixation time.
10. What should pupils learn?
Recognise words accurately and check meaning, regardless of unmeasured eye-movement counts.
Original research and connected vocabulary guides
- McConkie, Kerr, Reddix, Zola and Jacobs (1989): Eye Movement Control During Reading II — Frequency of Refixating a Word — study of more than 40,000 fixations.
- McConkie, Kerr, Reddix and Zola (1988): Eye Movement Control During Reading I — Initial Fixation Locations — launch and landing relation.
- Nuthmann, Engbert and Kliegl: Self-Consistent Estimation of Mislocated Fixations During Reading — targeting and corrective movements.
- A Beginner’s Guide to Eye Tracking for Psycholinguistic Studies of Reading — definitions of fixation measures and refixations.
- The Fundamentals of Eye Tracking, Part 2: Operationalization — measurement choices and denominators.
- Reichle and colleagues (1998): Toward a Model of Eye Movement Control in Reading — word identification and eye-movement models.
- Vocabulary | What Is Gaze Duration? — understand the time sum when another fixation occurs.
- Vocabulary | What Is the Optimal Viewing Position? — why landing location matters.
- Vocabulary | What Is Word Skipping? — separate skipped targets from refixations.
- Vocabulary | What Is Regression Path Duration? — separate within-word correction from backward rereading.
- Vocabulary Learning Hub — more science of words and comprehension.
Refixation probability captures a common and often useful feature of visual word recognition: the eyes sometimes need another look. The strongest vocabulary practice does not demand fewer fixations. It helps learners use all available evidence to identify the correct word and communicate the right meaning.