Week 5 - Eye movements and reading

Foveal Vision

Area of retina with highest photoreceptor density; provides sharpest vision

Peripheral Vision

Region outside the fovea with lower visual acuity

Retinal Angle

Unit (in degrees) for measuring visual space on the retina

Cortical Magnification

Foveal input is disproportionately processed in V1 and beyond

Physical eye structure and acuity

1. Vision clearest where photoreceptor density is highest in the fovea 2. Periphery is away from the fovea 3. Measurement is in degrees of retinal angle 4. Cortical processing emphasises fovea, from V1 and up

Fixation Duration

Eye remains still for 250–300ms to process visual input

Saccade

Rapid, ballistic eye movement (up to 700°/second) between fixations

Saccadic Suppression

Temporary inhibition of visual input during saccades

Reading Skipping

Only ~⅔ of words when reading for comprehension

Regression

Backward eye movement to previously read text, regressions (10–15% of fixations)

Four different metrics

can be measured to explore different experimental effects:

1.Fixation Duration Metric

Indicates processing difficulty per word or phrase

2.Regression Rate

Percentage of backward fixations during reading

3.Word Skipping Rate

Frequency of skipped words during reading

4.Saccade Distance/speed

Length/speed of eye movement between fixations

Perceptual Span

Number of characters processed per fixation; depends on language and direction. When you read, your eyes don’t move smoothly — they jump in quick movements (called saccades) and stop briefly (called fixations). During each fixation, you don’t just "see" the word you’re looking at — you also take in some information to the left and right of it. That visible area is your perceptual span

Gaze-Contingent

Display changes stimulus based on real-time eye position (where the reader is fixating)

Hebrew Perceptual Span

(read right-to-left), the perceptual span is the same size as in English, but flipped — readers see more to the left (ahead in that direction).

Japanese Perceptual Span

adjust based on layout: when reading left-to-right, they see about 13 characters to the right; when reading top-to-bottom, the span shifts 6 characters downward

Chinese Perceptual Span

1 character left, 3 right (Inhoff & Liu, 1998)

Text Difficulty and Span

Perceptual span changes with text complexity (Rayner, 1986) wider for easier texts and narrower when text is more challenging

Eye Tracking: Scleral Search Coil Tracking

Uses current shifts from a coil to track eye movement (<0.1° error)

EOG (Electrooculography)

Electrodes detect eye movement via changes in electrical field

Dual Purkinje Image Tracker

Uses infrared reflection to track fine eye movements, combination of lenses and mirrors reflect this light showing any movements made

EyeLink Eye Tracker

Infrared light illuminates the eye, the corneal reflection is monitored using a camera that samples the location of the eye at 1000Hz

Tobii Eye Tracker

Tracks eye movement at 90–120Hz and allows for head movement

Word Frequency Effect

High-frequency words are recognised faster (e.g., 67ms faster in eye-tracking). Up to 300ms difference in lexical decision task and 67ms in eye movement in reading studies

Measures of word frequency

Kucera-Francis Corpus: only 1,014 words from sources that are not the most typical of our language use. SUBTLEX: subtitles of 10,000 films and TV programmes and features 51 million words

Predictability Effect

Contextually expected words (e.g., "birds") are read ~80ms faster

Word Length Effect

Longer words are processed more slowly (~88ms longer)

Phonological Preview Benefit

Phonological information (how a word sounds) influences how we read and recognise words. Previewing a homophone (e.g., pair before pear) speeds recognition by about 20 ms (showing we automatically sound while reading). We use sound-based similarity to support word recognition

Automatic Phonological Access

Phonological code is automatically accessed during reading

Phonological Neighbourhood

Larger neighbourhoods lead to quicker recognition

Phonological Neighbours

Words differing by one phoneme (e.g., “pit” vs “bit”) improve recognition speed… some letters (like first and last) are more important than others in recognising words, we don’t treat all letter positions equally

Syntactic Parsing

mental process of assigning roles (subject, object, verb) to sentence parts; errors cause garden-path effects

Garden Path Sentence

Misleading syntax forces sentence reanalysis (“The horse raced…”), trick the reader because the initial interpretation turns out to be wrong. We make assumptions when parsing syntactic structure

Reanalysis

when we realise our initial parsing was wrong (as in garden path sentences), we reanalyse the sentence to find the correct structure

Incremental Interpretation

Sentence meaning is built word-by-word during reading (don’t wait till the end of the sentence), readers semantically interpret and syntactically parse text on a word-by-word basis

Minimal Attachment Principle

readers try to interpret sentences with the simplest possible syntactic structure

Incremental interpretation, visual world paradigm

eye movements are tracked while participants listen to narrative

Semantic Plausibility

Readers quickly detect semantic anomalies (e.g., “knife to chop carrots”), readers immediately detect semantic anomaly suggesting we incrementally interpret semantic aspects. Implausible (“used a pump to inflate the carrots”)… however can overridden by context (thanks pragmatics!)

Contextual Override

Implausible phrases can be accepted if context supports them (“mouse lit dynamite”)

Semantic Memory in Reading

Words are accessed based on associative semantic meaning

Model of Word Recognition

rely on the notion of having a mental lexicon (‘internal dictionary’). A store of all lexical representations, with their semantic meaning, syntactic role and phonology encoded within these representations. Not a single neural structure attributed to language knowledge

Interactive Activation Model

Higher frequency words are said to have ‘lower threshold’ for activation, more predictable words are said to be ‘primed’ so they have already received from activation, hence the threshold is reached quicker. This is also the explanation for semantic and phonological priming

Dual Route Cascade Model

Two pathways:

Route 1: Grapheme-phoneme conversion

Grapheme is the visual unit that corresponds to a phoneme (i in “pig”, igh in “high”). Conversion rules used to convert each grapheme into a phoneme. Rules determined by the most common grapheme-phoneme association in the language. Good for regular words and non-words, bad for irregular words!

Route 2: Lexicon + semantics

Orthographic input lexicon stores the spelling of all the words you know. Activates meaning and/or phonology (hence semantic and phonological priming effects). Good for reading all familiar words, bad for reading unfamiliar words (hence frequency and length effects)

Connectionist Triangle Model

Links orthography to phonology directly or via semantics. 1. Direct pathway from orthography to phonology. 2. Indirect pathway from orthography to phonology via semantics

EEG in Reading Research

Tracks brain electrical activity during reading; excellent temporal, poor spatial resolution

Self-Paced Reading Method

Measures reading time by button press; lacks natural reading behaviours

Lexical Decision Task

Participants judge whether a string is a real word; reveals recognition, not comprehension

Questionnaire Method

Gathers subjective data on reading experience or comprehension

Corpus Analysis

Studies word patterns and frequency using large-scale text databases

Corpus Choice in Frequency Effects

Word frequency effects differ by corpus (e.g., SUBTLEX vs. Kucera-Francis)

Web Reading Challenges

Navigation and scrolling may alter reading patterns

Language Generalisability

Effects found in English may not apply to other scripts like Chinese

Semantics in Models

Many models underemphasise semantics, reducing their explanatory power

Speed Reading Claims

Apps may boost word recognition, but not necessarily comprehension

Environmental Influences

Factors like caffeine may impact eye movement and reading behaviour