The word-identification approach has long been the basis for teaching reading in both native and second languages. Recent work has a new approach, particularly that of Frank Smith, who developed a feature analysis model of reading. As we can identify letters in different configurations or before all the strokes are completed, so can we identify words before all the letters are filled in, and comprehend sentences and paragraphs even if some words are left out. The process of reading consists of reducing uncertainty about a text by assigning the various configurations to mental or semantic categories established during the acquisition of language. Feature analysis is consistent with speed reading, as a person uses the redundancy of the text and finds predictive features to extract information from a text quickly. A project at the Macalester English Language Institute in 1975 used feature analysis to train Iranian students to develop English reading skills and speed by reading for specific letters or words, or to find answers to certain questions. Further research is suggested in: the affective domain of language learning, grammar and reading, and testing reading skills. (CHK)