One of the most popular and widely referenced models for word recognition is the Three Cueing Systems model. This model suggests that there are three cues that good readers make use of to identify individual words in text. The model further places these cues in a hierarchical order of importance.
The first and most important cue is semantics (sometimes called context). The range of possible words is restricted by the context, so when children come to a word they do not know, they can “guess” based upon the context.
The second cue is syntax. English places restrictions on the order that words can be placed in a meaningful sentence, so when semantics and syntax are both considered by a young reader, the model suggests that they can make an even more educated “guess” about individual words in the passage.
The third, and least important cue, according to this model, is the letter-sound information (orthographic information).
So this model suggests that when good readers are reading, they are depending upon semantic and syntactic cues to guess each word, and then they occationally “sample” from the text to confirm their predictions.
At BalancedReading.com, we always defer to the research, and in this case, the research suggests that the Three Cueing Systsms model is inaccurate. Simple logic reveals some of the problems with the model. First of all, if we depend so heavily on context to identify words, how do we identify the first few words in a sentence? Or how do we identify words when they are out of context? Secondly, English syntax is not all that constraining — in the sentence “Mary wrote her answers on ____,” the blank could be an article (“the” or “a”) or it could be a noun (“paper”) or it could be a proper noun (“Billy’s”) or it could be a pronoun (“her” or “his”) or it could be an adjective (“white” “flat” “smooth” “lined” “green” “perfect” etc, etc, etc.).
When you think about it, there is simply no way to accurately “guess” words in sentences, and research has shown this to be true. People are only correct about 1 out of 10 times when they try to “guess” content words in running text. To decode text, the first and really only cue that good readers use is the orthographic information — the visual information contained in the letters in the words.
For further reading, here is a brief paper (in PDF format) that describes the model and some of its more apparent flaws. For further reading on the Three Cueing Systems, see the recent article published by Marilyn Jager Adams.