HUMBEHV 2AP3 - Educational Psychology
The Science of Reading: A Comprehensive Overview
Introduction to Reading as a Cognitive Process
Fundamental Question: Is reading primarily a bottom-up or top-down cognitive process? This question explores whether visual input dictates understanding (bottom-up) or if higher-level cognitive processes guide our visual attention (top-down).
Illustrative Example: When you encounter the concept "ice cream," do your eyes process the individual letters first and send signals to the visual cortex for interpretation, or does your frontal cortex (responsible for higher cognitive functions) actively direct your eyes to specifically search for and process the word "ice cream"?
How Does Reading Work? Insights from the Patterson & Shewel Model
A Dual-Route Model: The model proposed by Patterson & Shewel (1987), as cited in Coltheart et al. (2001), offers a schematic of processes involved in both speech production and reading. It integrates both bottom-up and top-down mechanisms.
Bottom-Up Processes in the Model:
For Speech:
Auditory Input is subjected to Acoustic Analysis.
This feeds into the Acoustic-to-Phonological Conversion at the sub-word level.
The output then goes to a Phonological Response Buffer.
Finally, it leads to the Speech Output Lexicon and speech production.
For Print (Reading):
Orthographic Input (the visual representation of words) undergoes Orthographic Analysis.
This results in Orthographic-to-Phonological Conversion at the sub-word level.
The converted information is sent to the Phonological Response Buffer.
Ultimately, this routes to the Speech Output Lexicon and speech production.
Top-Down and Interacting Processes:
Lexicons: Dedicated mental dictionaries exist for both auditory and orthographic input, storing known words and their properties.
Cognitive System: A central cognitive system interacts with these processes, overseeing and guiding them.
Phonological-to-Orthographic Conversion: This sub-word level conversion allows for mapping sounds back to letters, crucial for spelling and recognition.
Graphemic Output Buffer: Prepares print output.
Orthographic Output Lexicon: Stores knowledge for written word production.
Key Conversion: A critical aspect of this model is the understanding of how a word sounds (phonology) relates to how it looks (orthography).
Reading Aloud: The model, especially when applied to reading out loud, highlights the importance of the grapheme-phoneme rule system. This system dictates how letters (graphemes) map to sounds (phonemes) and how this interaction with orthographic input leads to phonological output. The semantic system also plays a guiding role, influencing both phonological output and the final verbal response.
Modeling Human Reading Behavior:
Whole Word Processing: Human behavior suggests that for learned words, the number of letters does not significantly impact the time it takes to read a word. However, for made-up words or non-words, word length does affect reading time. This implies that learned words are often processed as "whole" units rather than letter-by-letter. This is illustrated by graphs showing "Naming latency" as a function of "Length of word (letters)" (Coltheart et al., 2001).
Modeling Phonological Dyslexia: The model helps understand conditions like phonological dyslexia by examining how people process different types of nonwords:
Pseudohomophonic words (PSHs): Nonwords that sound like real words when read (e.g., "fantom" sounding like "phantom").
Controls: Nonwords that do not resemble real words.
Close variants: Nonwords differing from a real word by only one letter.
Distant variants: Nonwords differing from a real word by more letters. Accuracy in reading these categories provides insight into the different routes used in reading.
Model's Utility and Limitations:
The Patterson & Shewel model successfully captures many observed human reading behaviors (Coltheart et al., 2001).
However, accounting for behavior does not equate to "proving" how reading works. Using an analogy, penguins can swim, but this doesn't mean they are fish, as there are many ways to swim. Similarly, even if a model explains all known reading behavior, it cannot definitively prove the mechanism because it's impossible to eliminate all other conceivable explanatory models. We can only suggest that a model accurately describes reading, not proves it.
How Do Children Learn to Read?
Importance of Letters and Sounds:
Phonological Awareness: This is the ability to recognize and manipulate the individual sounds (phonemes) within a language. It is identified as the number one predictor of a child's reading ability.
Letter Knowledge: Children's knowledge of letters significantly supports their acquisition of phonemes. Exposure, familiarity, and relevance (e.g., their own name) contribute to this learning.
Developmental Milestones: The majority of 3-year-olds and virtually all 4 and 5-year-olds can identify the first letter of their names.
Learning Progression: Typically, children learn the names of letters faster than the sounds associated with those letters (e.g., knowing "W" vs. its /w/ sound, or "N" vs. its /n/ sound). (Rebecca Treiman, 2013).
Kindergarten Study on Sound Knowledge (Treiman, 2013):
Kindergarteners were tested on their knowledge of letter sounds both before and after formal instruction in school.
For the letter "W": Only 20% were correct before the lesson (with 28% incorrectly saying /de/); this increased to 53% correct after the lesson.
For the letter "Y": Only 13% were correct before the lesson (with 60% incorrectly saying /we/); this only slightly increased to 18% correct after the lesson.
Observation: Letters with names that sound similar to their phonemic representation (e.g., "s" for /s/) are easier for children to learn than letters whose names are dissimilar (e.g., "y" for /y/).
Implication for Curriculum: Rather than a standard approach of introducing one new letter per week, it would be more effective to dedicate more instructional time to letters that are phonologically "difficult" for children to learn.
Letters and Sounds Help Spelling:
Early Spelling Errors: Young children often make characteristic spelling errors that reveal their reliance on phonetic encoding and letter knowledge.
For example, spelling "farm" as "frm" indicates that the child perceives the sound /a/ to be contained within the letter name "r" (/ar/).
Similarly, spelling "help" as "hlp" suggests the child hears the sound /e/ within the letter name "l" (/el/).
Integrated Learning: These errors demonstrate that young children utilize all available information for language processing: their knowledge about letter names, their developing phonological awareness, and their recognition of print patterns. (Rebecca Treiman, 2013).
Are We Actually Reading "Whole" Words? Evidence from Eye Movements
Mechanics of Eye Movements in Reading:
Saccades: These are rapid, ballistic movements of the eyes across the text.
Fixations: These are the brief pauses (typically between 200 and 250 milliseconds) during which the eyes are relatively still, allowing for visual encoding of text. Skilled readers typically process about 7 to 9 letters during a fixation.
Regressions: These are saccades that move the eyes backward in the text, usually to re-read a word or phrase that was not initially understood. In skilled readers, regressions account for about 10-15% of reading time. (Rayner et al., 2006).
Impact of Text Difficulty: When reading more difficult text, readers generally exhibit shorter saccades (smaller jumps), longer fixations (more processing time per segment), and a higher frequency of regressions.
Developmental Changes in Eye Movements (Rayner et al., 2006):
Grade 1 Readers: Tend to have longer fixations (greater than 350 milliseconds), make 2-3 fixations per word, and exhibit about 30% regressions.
Grade 4-5 Readers: Show stable saccade and fixation patterns when reading age-appropriate books.
Causality: Crucially, these changes in eye movement patterns are understood to represent the development of reading comprehension and understanding, rather than being the cause of immaturities in comprehension.
Study on Text Difficulty and Eye Movements (Rayner et al., 2006):
Experiment: 16 English-speaking university participants had their eye movements recorded while silently reading 32 passages of text with an eye-tracker.
Stimuli: The passages were pre-rated for difficulty by a separate group of participants, ranging from easy (2.8/10) to difficult (6.6/10).
Results: As passage difficulty increased, readers consistently showed longer fixations, a greater number of fixations overall, and spent more total time reading the passage.
Inconsistencies Affect Eye Movements (Rayner et al., 2006):
Anaphor-Antecedent Consistency: Reading comprehension relies on linking pronouns (anaphors) to their referents (antecedents). For example, "Alison decided to order some celery sticks… The waiter brought them…" (consistent) vs. "Alison decided to order some celery sticks… The waiter brought her some water and the carrot sticks…" (inconsistent).
Experiment: 18 adult skilled readers read 36 paragraphs containing either consistent or inconsistent anaphor-antecedent relationships, while their eye movements were tracked.
Results:
Readers exhibited significantly longer gaze durations when encountering an inconsistent anaphor compared to a consistent one.
Readers were also more likely to make regressions in the condition where the inconsistency was immediate/near (18.1%) compared to a consistent near condition (5.3%).
Implication: These findings suggest that visual processing (a bottom-up process) is directly influenced by higher-level cognitive processes, such as semantic understanding and whether the sentence "makes sense" (a top-down process). This demonstrates that reading is fundamentally a cognitive process.
Ongoing Eye Movement Control (Bicknell, Levy, & Rayner, 2020):
Initial Observation: Missing the first 3 letters of a word makes it much harder to read than missing the last 3 letters, highlighting the importance of word initial recognition.
Hypothesis: If reading were purely bottom-up, eye gaze would consistently move from left to right across a page, except for general difficulties.
Experiment: Participants silently read 160 single-line sentences, each with 7 words and a target word, while their eye movements were recorded.
Conditions: The target word either did not move, moved slightly to the right, or moved slightly to the left from its expected position.
Findings:
When an eye gaze landed too far behind the target word (missing crucial initial letters), participants spent a longer time looking at that word.
Furthermore, when the eye gaze landed too far behind the word, participants were significantly more likely to make a subsequent re-fixation (another saccade) back to that same word.
Conclusion: Eye-tracking data indicates that cognitive processing is continuously involved in targeting saccades, even within a single word. This suggests that all letters within a word are perceived and contribute to ongoing processing. Saccade targeting is optimized for efficiency, with eyes often scanning forward into words after perceiving their initial letters. This intricate control means that reading requires constant, ongoing cognitive processing, even at the micro-level of eye movements within individual words, challenging the common perception of reading as solely a "visual" skill.
Conclusion: The Iterative Nature of Reading
Reading is a dynamic, constant, and iterative process that involves a continuous comparison and integration of visual input (what we see), prior knowledge (our understanding of language and the world), and phonological output (how words sound). This complex interplay underscores reading as a highly sophisticated cognitive skill.