Prepared by Melissa Mitchell (additions and comments by Frawley)
The cognitive science issue in reading is as follows: how could we build a device that goes from printed input to meaning recognition and makes the same kinds of errors as humans do?
Reading in English or other alphabetic languages is done in a jumpy fashion (saccadic scan) of approximately 250 msec each and each encompassing about 1.1 words. During each fixation, the printed entity is recognized, information is integrated, and motor plans are made for the next jump.
Reading is unique in that numbers and letters are the first entities a child encounters in which orientation and order are essential to identity. This results in backwards letters, etc.
In fact, if children do not reverse letters and numbers, this indicates a problem. Every other object they need to recognize has to be held constant under different viewpoints, so it is not surprising that children call "u" "n," e.g. They analyze "u" as a constant "n" under inversion.To gain insight into how we read, lexical decision tasks are used, in which the subject is asked, as quickly as possible, to say whether a stimulus is a word or not. These tasks measure such things as time of response as a function of the phonemic and orthographic regularity of input. While these tasks are obviously not the whole story in reading, they do tap one essential process: recognizing and processing words.
Note that these tasks and results are designed to shed light on one of the fundamental issues in reading: is there any -- or only -- phonemic mediation? Does the processor have to go through phonology to assign meaning to printed input? Some claim that all reading is phonemically mediated, especially in languages with close phoneme-grapheme relationships: e.g., Turkish and Finnish. Others claim that there are two routes, phonemic and visual.
One standard result from these tasks illustrates that phonology can play some role in reading: the pseudohomophone effect -- words that sound like real words but are spelled like nonwords take much longer to reject. But the effect is eliminated when word is obviously spelled wrong, indicating that there can be no phonological check before checking spelling. This suggests that the primary route for reading is visual; when this route is slow, phonology, semantics, and context are used to determine meaning. When a person becomes proficient at reading, visual recognition becomes automatic, if not modularized, resulting in increased speed and inattention to detail.
That is, phonology is called only when needed. This is an instance of a rather robust effect throughout the literature on linguistic processing. In morphology, for example, it looks like the human processor prefers unanalyzed morphological wholes, but will analyze into morphological components when necessary. So the preferred visual route in reading suggests that the processor prefers whole word visual recognition, but will analyze into grapheme-phoneme correspondences when necessary. Note that reading English, in many ways, benefits from this strategy: the word-level cue for plural, e.g., is constant visually "s," but varies when analyzed phonetically [s], [z]...
------------------------
Reading: Green Chapter 6
The lexicon is the stored mental representation of words; the input lexicon is visual input and the output lexicon is phonological. Perhaps in reading, print is changed to an inner voice, and then converted to meaning. Or meaning may be derived directly from print: note the Russian Novel Phenomenon -- names of characters are never pronounced but identity is understood.
There are three stages in learning to read:
1. logographic -- earliest mental representation of printed alphabet. Words are recognized like pictures independent of one another; words are interesting to child; the first letter only is usually salient
2. alphabetic -- child isolates phonemes in mind's phonological reps. and maps these onto written letters using phoneme-grapheme rules; this must be explicitly taught.
3. orthographic -- consequence of interactive reading and linguistic knowledge.