It is also clear to those who can read this display aloud, "Umgogly vaped um callinger, yardidly zingored pey," that meaning is not required to. produce a spoken sentence. Rather, an understanding of the correspondences between printed words and the articulatory programs of the oral language (signs and syntax) permits the utterance of the sentence with no conceptualizing (meaning) necessary. Hence, Smith's contention that "Before you can utter a sentence, you must know what it means" (p. 77) is clearly not true. A more accurate statement is that in order to make a proper articulation of some of the words in some sentences, it is necessary to have developed some understanding of the underlying conceptualization represented by the sentence or parts of the sentence.

One final point raised in Smith's polemic deserves some comment. In a section which intends to demonstrate that a skilled reader could not first identify words and then comprehend what he is reading, and that comprehension must precede identification, he states that "Although visual information sufficient for the identification of four words may be available in a single fixation, and four fixations a second may be made, the skilled reader cannot identify words (that is, read material aloud) at the rate of 16 words per second (almost 1000 words a minute) but instead at barely a quarter that rate (a limit which is not set by the rate at which words can be articulated)." '(pp. 63-64)

What Smith fails to point out is that, (a) although the eye typically may make four fixations per second, the visual information included in each successive fixation is not new but rather has considerable overlapping, hence word identification may still be going on for the information in the overlapping segments of the visual field; and (b) it turns out (see Chapter V, Hypothesis 3) that the rate of reading continuous prose aloud may, indeed, be limited by the rate at which words can be articulated aloud, to about four or five words per second, or 240 to 300 words per minute. But these rates are comparable to the silent reading rates of better readers (college students), and, since such rates of articulation are possible (even aloud, with muscle inertia and lag time between internal articulatory processing and actual movements), it is then completely possible for visual words to be decoded by reference to internal articulatory programs at rates quite compatible with typical (and Smith, p.64, agrees, even optimal) rates of proficient reading.

We find ourselves, then, in disagreement with many of Smith's assertions, and in agreement with Carroll's (1971) position that the printed display is converted to an internal representation of a spoken message, which is then used in conceptualization as discussed in Chapter III. In the early stages of learning to read, it seems likely that the child will attempt to decode such units of print into units of the spoken language as are involved in the teaching process. Thus, a teacher who uses a whole-word approach will stimulate children to decode entire printed words into spoken words (ie., the unit of decoding will be the word). A teacher who uses a strong phonics approach will stimulate the use of letters and letter clusters to decode printed word segments into spoken word segments.