The foregoing is important because it bears on a fundamental problem in reading, the problem of teaching children to read. In a recent book, deliberately characterized by the author as polemical (p. vi), it is argued that print is not speech written down, and that children do not, indeed cannot, learn to read by decoding print to speech (Smith, 1973). The controversial nature of this assertion is indicated by the statement of a leading researcher in reading that the decoding aspect of reading can be considered as learning to "read out in units of the spoken language what is directed by the graphic units" (E. Gibson, 1969, p. 434). Also, Carroll (1971) states that "Essentially, the reading process is one of using printed or written symbols and sequences of symbols as cues to construct some kind of representation of a spoken message" (p. 133).

Smith's position seems to be based largely on 'a very broad and loose definition of reading as practiced by skilled readers. For instance, he permits "fluent" readers to make many "errors" at the "word" level so long as "meaning" is retained. He states, "If you are reading so that listeners can comprehend you, it is far more important that you comprehend yourself what you are reading than that you identify every word correctly" (p. 79). It is, of course, a truism that we frequently make such "errors", but simply because !!... plausible meaning is maintained does not mean that the meaning intended by the writer was maintained, Clearly, this approach makes reading one-sided, it neglects the fact that reading is a communication activity, and that, from the writer's point of view, a mis-reading an "error", the exact meaning is not preserved, although what change occurs may not be too important to either the writer or the reader.

Smith also asserts that "Individual words do not carry any information about how they should be articulated" (p. 77). But then he points out that even such words as permit (PERmit or perMIT) or read (red or reed) can be read in "list intonation." What he means, of course, is that we need context to know how to pronounce these words in sentences. But clearly they give some information about how they should be articulated even in a list. Furthermore" as we saw in Durkin's study, children who learn to read on their own first read words, not sentences-so do children who are taught to read.