ABSTRACT
This chapter demonstrates strongly that much experimental research is based on unexamined theory — in this case, the theory that the word is a self-evident unit of language and meaning. In that view, words are recognized (matched with a kind of mental template) as the basic act in reading. Context is at best an aid in word identification. When that theory is tested against real English texts rather than texts contrived to control words, it is shown to be grossly inadequate. Peter Fries, a systemic-functional linguist, uses the tools of corpus linguistics to examine the extent to which word-centered and meaning-centered views of reading provide the best fit for the realities of real English texts. Fries shows through corpus linguistics that words cannot be recognized except in the multiple contexts of language as readers construct meaning. He demonstrates this by showing the data of corpus linguistics on even the most common words such as look.
