Clay was a distinguished scholar at Ohio State University in 1984-85, where she joined Gay Su Pinnell, a recent Ohio State Ph.D, who was doing reseach on first grade reading instruction.

Called Reading Recovery, the program emphasized the teaching of whole words rather than word parts, particularly deemphasizing teaching the sounds of the letters as in direct phonics instruction.

In her last works, Clay summarized the complexity of teaching children to read, and made significant remarks about brain networks at a time when the wiring together of visual and auditory brain areas were just being explained by Stanislas Dehaene...

When a child first looks at a written code and tries to make something of it, this is a new experience for his brain which can already deal with oral language. The new task is about learning to recognise the visible symbols and making some invisible links to how we speak. How do we bring these things together?

Often people choose to explain the visible/invisible relationships by helping the child to link the smallest features of the letters with the phonemic features of the oral language. In my observations of young children trying to make any sense of the code I chose to define reading as a message-getting, problem solving activity, and writing as a message-sending, problem-solving activity. Both activities involve linking invisible patterns of oral language with visible symbols.

When you think about it, deaf people learn to ‘speak’ with their hands, and blind people learn to ‘read’ with their hands. So the human brain is really quite flexible about how it can deal with linguistically coded messages on paper...

How should we think about these scanning rules and the links we make? We can assume that we create networks in the brain linking things we see (print on a page) and things we hear (the language we speak). Messages flow in and out of those networks. In the context of reading and writing this is often called literacy processing. Familiar marks on the page can be linked to familiar language networks in the brain, and they allow us to make sense of novel messages never read before. Processing activities may involve only one network or many networks ‘talking’ to each other!..

After only a brief time at school children have acquired many wa) checking words in the text they are reading. They move from print to message and back at each decision point. Letter by letter, word after word, they select a response and begin to construct the phrase or sentence. From the word groups they get meaning to aid the next bit of solving, and that helps them to solve the meaning further along the text. Probabilities or rules of occurrence are derived from our everyday experiences with texts. As the child reads, his brain links the sounds of speech and the squiggles on a page of print and computes the probability of information. Children construct their personal rules about written language from the print you expose them to.

Most written language occurs as continuous text, so the focal task for the learner is to problem-solve the message(s) of continuous text. That is another one of my assumptions. Teaching a child 100 words, or 26 letters, in isolation before you allow him to read a text does not seem like the appropriate learning context for laying down the foundational neural networks.

Literacy Lssons, Part One, "A Simple View of a Complex Theory (2008), pp.9-10