The Visual Encoding Stage

The Visual Encoding Stage Involved in the Reading Process

1. The Icon

The term icon refers to the visual information kept in a temporary store for a very brief period. The visual sensory processing of the written word, including the iconic storage that eventually leads to its recognition, constitutes the first step in the reading process.

In cognitive psychology Opens in new window, the process of capturing the meaning or the pronunciation of the word is referred to as lexical access. The term lexicon refers to a putative mental dictionary which contains all the information the reader has about the words s/he knows. The mental lexicon can, therefore, be viewed as part of the long-term memory store.

Consequently, in reading, lexical access refers to the process of relating the printed word to its representation in the mental lexicon. During the reading process, movement of the eyes is not smooth and continuous but consists of a series of fixations interspersed with short, jerky movements called saccades.

The average duration of fixation is thought to be about 250 milliseconds even though factors such as word length, familiarity, and word frequency can affect the duration. The number of words fixated also can vary but, in general, one word is processed during a single fixation; function words such as the, of, and to are not always fixated.

Studies by Just & Carpenter (1980) show that the reader tries to recognize each word as it is being viewed and that the eye remains fixated on a word as long as it is being processed. This indicates that eye movement during reading is not an automatized act but is governed by higher-level cognitive decisions. The visual sensation created during fixation stays in a temporary store, the icon, for a short duration; the eyes move again, fixate and process the next visual unit.

Experiments by Sperling (1960) show that visual information remains in iconic store for only about 250 milliseconds after which it decays. In order to be preserved, information in the icon must be processed further before input from the subsequent eye fixation comes in and wipes out the first icon. Experimenters have taken advantage of this feature of the visual process by superimposing a second visual stimulus on the first icon in order to experimentally obliterate the first input. This experimental process is referred to as masking. A processing mechanism that is sluggish in relaying visual input to the next stage rapidly could be an impediment and, therefore become a potential source of reading difficulty.

Even though the iconic store has a very brief life, it has a relatively large storage capacity of about twelve letters. In a classic study, Sperling (1960) investigated the nature of the icon by presenting, for a fraction of a second, twelve letters arranged in a matrix of four columns and three rows and asking subjects to report the letters in one of the three rows by sounding a high, medium, or low tone that corresponded to one of the three rows. He found that subjects could report almost all the letters in any of three rows provided the tone followed the visual presentation immediately.

This advantage, however, was lost if the tone came 300 milliseconds after the visual presentation. This suggested that for an extremely short duration, the entire matrix of twelve letters was available for report. This relatively large storage capacity of the icon is, therefore, sufficient to hold almost any printed word. Even though some details of Sperling’s interpretations are questioned, the general finding that the iconic memory has a duration of about 200 to 300 milliseconds is confirmed by other studies (e.g., Eriksen & Collins, 1967, 1968).

Letters in a word also appear to be processed simultaneously and in parallel. In one study, Sperling (1970) presented a string of five letters for a very brief period and asked readers to report them. Within 50 milliseconds after the onset of the stimulus, however, the computer changed the fifth and final letter into another letter. During testing, the subjects reported the original letter in the string and not the substituted letter. Since 50 milliseconds is too short for processing letters serially, from left to right, Sperling concluded that letters in a string are processed simultaneously, in a parallel fashion.

Other studies have shown that letters at the two end positions of a word are processed simultaneously, before the ones in the middle (Merikle et al., 1971). It is also known that when a word is presented, not all the letters of the word are relayed as a single unit but the morpheme (root word) and its suffix are segmented and processed as separate units.

Taft (1985) used a lexicon decision task to investigate this phenomenon. In a lexicon decision task a string of letters which may or may not form a word (e.g., rain or rane) is visually presented and the subject is asked to indicate whether the string is a word or a nonword. In order to avoid confounding factors associated with pronunciation, a manual response such as pushing a “yes” button or a “no” button is required. Reaction time is the dependent variable.

Taft found that words such as eating and bigger took longer to correctly identify than words without a suffix such as string and trigger. He concluded that the additional time was required to strip the suffixes off the root morpheme. Words such as string and trigger do not have suffixes and, therefore, require less time than words such as eating and bigger. It must be noted that during oral reading and writing, this process has to be reversed in order to produce the word with its suffix.

Taft administered another lexical decision task by using nonwords with and without suffixes (e.g., molks, widodled, vs. porld, vodinten) in order to locate the stage in which this parsing occurs. Again, suffixed nonwords took longer time to respond than nonwords without suffix.

Since nonwords such as the ones used in this experiment are not present in the mental lexicon, parsing of the word into its root and suffix should have taken place at the prelexical stage, before it was recognized.

It appears, therefore, that suffixes are treated as distinct entities and that their separation from root morphemes occurs without the help of the mental lexicon. Because, in the iconic memory, information is registered the way it looks in print, the suffix stripping probably occurs somewhere between visual encoding and word recognition. The suffix omission errors frequently seen in the reading and writing of dyslexic readers cannot, therefore, be attributed to poor visual perception or inattention.

2. Short-Term Visual Memory

The question of whether short-term memory Opens in new window has a status independent of long-term memory Opens in new window is frequently raised. Whether it has an independent status or is an integral component of a single memory system, short-term memory, as a hypothetical construct and model, has served well in explaining certain aspects of reading behavior. For this reason, in this entry, short-term visual memory will be treated as though it is distinct from long-term memory.

Several experimental studies have shown that visual information can be retained in memory for a few additional seconds even after it has disappeared from the icon . Even though short-term visual memory differs from iconic memory in important ways, its representation is visual in nature and hence the descriptive name.

Unlike the icon, short-term visual memory has a duration of about five or six seconds and it cannot be obliterated by another visual input that closely follows it (i.e., it is resistant to masking). Its capacity, however, is smaller than that of the icon.

It is generally believed that all the information in the icon is transferred to the short-term visual memory which acts as a holding place. Beyond this, the fate of the visual representation of the word can be described in terms of three possibilities:

1. it may be relayed to working-memory where its phonological representation is realize;
2. it may be directly relayed to the semantic lexicon where its meaning is realized;
3. or both these operations may take place concurrently.