Chapter 43

I was going to type this over, but it is clear that won't happen-sorry.
A Theory of Reading: From Eye Fixations to Comprehension
By Marcel Adam Just and Patricia A. Carpenter

In this article, the authors investigate how a reader controls the rate of input through eye fixation when presented with text. The common misconception that readers do not fixate on every word in a text but rather on a small portion of the text is quickly squelched. Data indicates that during ordinary reading, readers do fixate on each and every word that is present in text, reading an average of 1.2 words per fixation. The small exception to this is when readers come across the small words the, of and a. Readers actually vary in gaze duration depending on the word. Longer fixations are attributed to longer processing caused by the word’s infrequency and its thematic importance. Readers also pause longer at the ends of sentences.

There are two other assumptions that people generally have. The first is called immediacy assumption. This is the assumption that a reader tries to interpret every content word of the text as they come upon it when reading. When interpreting a word, the process involves several levels. First, the reader must encode the word. Next, they must assign an object and choose the meaning of the word. Finally a reader must decide how the word relates within the sentence. The second assumption is the eye-mind assumption. This assumption is that the eye fixates on a word as long the word is being processed. Comprehension of words often involves using previously presented textual information. There is no significant lag between what is being fixated and what is being processed.

In this article, Marcel Adam Just and Patricia A. Carpenter examine four major sections. First, they describe a theoretical framework for the processes and structures involved in reading. Next, they describe the reading task and eye fixation results accounted for by the research model they have conducted. Next, they investigate the model itself, including subsections that describe each element of the model. Finally, they discuss some implications of the theory for language comprehension and relate the theory of reading to other approaches.

Theoretical Framework
Reading involves a number of coordinated processing executions. These stages include word encoding, lexical access, assigning semantic roles, and relating the information in a given sentence to previous sentences and prior knowledge.

Look at: A Schematic Diagram of the Major Processes and Structures in Reading Comprehension

Structural and procedural knowledge is stored in the form of condition-action rules, such that a given stimulus condition produces a given action. The brain functions fast and automatically when reading familiar text. When reading less familiar text, the brain slows down to process the variables as well as the contents.


Productions operate on the symbols in a limited-capacity working memory- the symbols of the words activate concepts that are the inputs and out puts of production. Items are put into working memory when they are encoded. Retrieval from long-term memory happens when a production fires and activates a concept, causing it to be inserted into working memory. Long-term memory is a collection of productions that are repositories of both procedural and declarative knowledge. This knowledge includes orthography, phonology, syntax, and semantics of the language, as well as the schemas for particular topics and discourse types. Recent evidence suggests that working memory’s capacity is strongly correlated to individual differences in reading comprehension performance.

Production systems have a mechanism for adaptive sequencing of processes-The items in working memory are given time to enable a given production to fire and insert a new item. The various stages of production can supercede each other.

Research
During this research, scientific text was used. There were 15 passages from Newsweek and Time, averaging 132 words each. The text was identified as “fairly difficult” and “difficult” n the Flesch’s readability scale, with 17 words per sentence and 1.6 syllables per word.

Method and Data Analysis
The readers were 14 undergraduates who read 2 practice texts prior to reading 15 scientific texts in random order. The texts were presented on a television monitor using upper and lower case letters and conventional paragraph layout. To start the passage, the reader had to fixate on a determined section of the passage and press a “ready” button. The passage appeared instantaneously and remained there until the reader signaled that they had finished reading by pushing a response button. The reader’s pupils and corneal reflections were monitored by a camera as they read. This monitoring system assisted with collecting the data needed for the research.

Results
The mean gaze durations gathered by the participants indicated that their reading rates were typical for the text they were presented.

The Reading Model
The reading model included five stages. The five stages include 1. get next input; 2. word encoding and lexical access; 3. case role assignment.; 4. interclause integrations; and 5. sentence wrap-up.

Get Next Input
Involves the cycle that finds information, encodes it, and processes it.
When the goal conditions for processing a word are met, the result is to get the next input.
The duration of the next input is short, consisting of the time for a neural signal to be transmitted to the eye muscle. The process begins again.

Word Encoding and Lexical Access
This stage involves encoding a word into an internal semantic format. It is believed that visual features of common words have been already established in working memory. Possible meanings and information about the context of the word occurs during this stage. Orthographically based subword units such as syllables happen during this stage as well. Other time alternative codes, such as phonological and whole-word codes may also be used. Perceptual encoding, parallel productions and activation by the serial productions that do the major computations in all stages of processing also occur during this time.

Assigning case roles
For comprehension to occur, a reader needs to determine the relations between words, the relation among clauses and the relations among whole units of the text presented. Determining the relation among the words in a clause can be categorized into semantic classes, like agent, recipient, location, time, manner, instrument, action, or stat. The reader not only has to determine the relation, but also has to look for the meaning using the prior semantic and syntactic context, as well as language based inferences. Verbs have been proposed as providing a pivotal source of information to establish the necessary and possible case roles, the noun phrase then is fit into slots to help establish meaning. When a reader immediately assigns case roles, errors sometime occur. Revisions then have to be made for the assignment process to be completed. (Mary loves Jonathan…. Mary loves Jonathan apples).
The strategy of assigning case roles within clauses is to assign a word to the clause processed and make adjustments as needed (soil, when farmers are plowing the soil… When farmers are plowing the soil is most fertile). There is no mapping between particular case roles and the duration of the assignment process. The assignment may depend on the context of the text, prior knowledge of properties of the word or other unidentified areas. Overall, verbs did not take long to process (33msec), connectives (9 msec), adverbs (29 msec), place or time (23 msec), and possessive (16 msec).

Interclause Integration
Clauses and sentences must relate to each other for the reader to make meaning of what is read. For each new clause, the reader must weave previously presented clauses or sentences presented. The reader must integrate the previous information acquired from the text or with the prior knowledge retrieved from long-term memory. A representation between the old and the new information must occur. It takes longer to read a sentence that refers to information presented several sentences earlier than one that refers to recently introduced information. Integration can lead to forgetting in working memory. As a new chunk is formed, it may displace some previous information that was being stored in working memory.
The gaze duration on a sector depends on its text-grammatical role and on the number of concepts it contains.

Sentence Wrap-up
When readers reach the end of a sentence, it is called sentence wrap-up. There are two important things that happen during this stage. Many difficulties within a sentence are clarified by the end of a sentence. The second important thing that happens at the end of a sentence is it signals the end of one thought and the beginning of a new one.
Previous eye fixation studies indicate that if a lexically based inference needs to be made by the reader to relate to a new sentence to some previous part of the text, there is a stronger tendency to pause at the lexical item in question and then at the end of the sentences.
Reading-time studies also indicated that there is extra processing time given at the end of sentences.
Sometimes the wrap-up stage happens at the end of text units or during smaller or larger sentences. Sometimes the wrap-up stage happens at the end of clauses.
The final word in a paragraph may also be a wrap-up point.
The return sweep of the eyes from the right-hand side of one line to the text on the left-hand side of the next line was also looked at. Due to the sweep, the first word on a line may also get increased gaze duration.

Fit of the Model
To test how the model accounts for the data, comparisons of how closely the estimated gaze durations match the observed gaze durations can be looked at. The estimates took into account the processes of encoding, lexical access, case role assignment, sentence wrap-up and the beginning of the next line. Overall, the match was satisfactory in this study as indicated by the table on page 1207 and 1208.

Discussion
Three areas were looked at in the final section of this article. The are the immediacy assumption for language processing in general, how variation in reading modes can be handled by the theory and the relation of the current theory to other theories of reading.

The immediacy assumption implies that a reader’s initial interpretation of a sentences and its meaning are often accurate. Readers will fixate on a word as long as they need to in order to make sense of the symbols presented. By doing so, readers do not overload their memory and can begin to process new information. Reader’s are able to delineate between ambiguous information and quickly can re-establish new meaning if sentences do not fall together quickly. When reading headlines in newspapers, or in many jokes of puns, the reader must often rely on the contrast between two interpretations of an ambiguous word or phrase to make sense of what is read.

Variation in reading
Reading varies. The function of what is being read depends on who is reading, what they are reading and why they are reading. For instance, when reading these article reviews, my eye fixation was much slower as I had to process and make sense of the difficulty of the text. A reader’s goals are probably the most important indicator of how one approaches the reading process. Reading may also vary when the same text is being read. When I read these articles the first time, I read them at a different speed and comprehension than when I read the article a second and even third time. Each time I read the article, a bit more information may have been gleaned. Words that I initially needed to fixate on for a longer period of time, may have needed less eye fixation the next time. My working memory was certainly overloaded the first time I read this article. Unfortunately, my working memory still does not have enough room to fully grasp the true concepts of this article. However, operational capacity may depend on the automat city of basic reading processes such as encoding and lexical access. Poor readers have to devote more time and attention to these processes; hence the inability to maintain previously presented information for comprehension.

Theories of Reading
Current theory attempts to account for reading time on individual words, clauses and sentences. This approach is more adept at understanding the importance of recall, question answering, and summarizing. The current focus on processing time indicates a theory that accounts for the moment-by-moment, real-time characteristics of reading. It also tries to span the various stages of reading by describing mechanisms for the word-encoding and lexical access stages, as well as the parsing and text integration stages. Current theory pays less attention to retrieval and reconstruction of the text read. Current model falls between top-down and bottom-up.

Gough-focus on the word-encoding process, bottom-up model
LaBerge and Samuels bottom-up and top-down approach to reading, bidirectional.
Kintsch and van Dijk -focused on integration.
Schank & Aelson- focused on schema-based theories.
Goodman & Niles- top-down model; analysis-by-synthesis theory; eyes fixate merely to confirm the hypothesis.

Future directions
1. Make computer simulations that are driven by reading performance data. The computer can be used to look at the complex interactions among knowledge sources.
2. Real-time characteristics of reading. Eye movement and reading-time methodologies can measure reading time on successive units of text. Presenting sentences one sentences at a time; another is to present text in phrases, clauses, sentences or entire passages.

In conclusion, in this article, Marcel Adam Just and Patricia A. Carpenter examined four major sections. They began by describing a theoretical framework for the processes and structures involved in reading. Next, they described the reading task and eye fixation results accounted for by the research model they conducted. Next, they investigated the model itself, including subsections that describe each element of the model. Finally, they discussed some implications of the theory for language comprehension and related the theory of reading to other approaches.