1. Trying to assess the impact of pre-school training in oral language skills on learning to read in the first or second grades before the reading decoding skills are adequately developed. There is no reason to expect oracy training to affect learning the decoding skills of reading (see page 91). Rather, the impact of such training on comprehension should be sought after the bulk of the decoding skills have been acquired-in the third or fourth grades.
  2. Trying to produce speeds of "reading" well over 300 words per minute without recognizing that the speed of reading is limited by the speed of languaging and conceptualizing. The "fact" of high-speed reading is not only peddled by enterprising individuals and concerns, it is also frequently pointed to when distinctions between auding and reading are being drawn in scientific books concerned with the nature of reading (see p. 54; also Neisser, 1967; Mattingly, 1972). But the developmental model (Hypothesis 3 of Chapter V) makes clear that such "reading" cannot occur. It is more likely that speed readers utilize less and less reading (languaging via print) and more and more conceptualizing-creating a story out of fragments of text. Thus, although page after page of text may be scanned and flipped through, it is a gross error to report reading speed in terms of the words on the pages divided by the time needed to flip the pages. Such a process is not the reading process used by the national sample described in Chapter V, nor is it primarily reading at all. It is scanning and constructing a story. At any rate (no pun!) the burden of proof for speed "reading", and hence for the use of this concept as a factor distinguishing languaging by eye and by ear, rests on those making the assertions. We have found no evidence for speed "reading".
  3. Trying to produce tests of "listening" that produce scores not correlated with tests of reading, that is, trying to test a separate "listening" factor. Thus, concern has been expressed that "listening" ability tests and reading ability tests frequently correlate highly, and, therefore, the "listening" tests do not measure some independent "listening" ability and hence are not useful (cf., Russell, in Duker, 1968; Spearritt, 1961; Lundsteen, 1969; Brown, in Duker, 1968). Yet, following the developmental model of reading, we see that auding comprehension and reading comprehension tests should be. highly correlated-as long as auding and reading factors, but not listening and looking or decoding factors, are primarily influencing test performance. In fact, however, most auding tests are called "listening" tests and fail to exclude undue reliance on short-term memory by permitting re-auding; they may include following a sequence of sounds; sometimes they require the recognition of non-speech sounds and language in the same test battery (Witkin, 1971). In short, they may represent a variety of listening and conceptualizing tasks that mayor may not involve language. The surprising factor is that "listening" and reading tests do so frequently correlate above .60. The developmental model suggests that we should strive to construct auding and reading test batteries that are even more highly correlated and that can, therefore, index the closing of the gap between auding and reading as languaging processes during the early period of acquisition of decoding skills, and the equivalency of the two processes for languaging after the decoding skills are acquired.
  4. Trying to test reading comprehension using knowledge derived from "typical" school curricula rather than the language and conceptual bases of the students at hand-to the disadvantage of those students who have not participated in the curricula from which test items are drawn. This problem has come to the foreground today, with the concern for accountability and the emergence of the concept of criterion-referenced tests that a local school district can use to test what is being learned of what is being taught. Careful attention to the interrelationships among conceptual base, language ability, and reading-as in the developmental model-points to the desirability of reading tests that do not rely upon specific content area knowledge for their performance, or that separate content knowledge from assessment of the ability to access language and conceptual bases via the graphic signaling system.

    Admittedly this is a difficult thing to do; testing reading comprehension requires the use of some content. Yet many standardized tests contain "standardized curricula" content, and many students can score high on such tests without even reading the passages accompanying the test items (Tuinman,1972-73). This may penalize the student who lacks the content knowledge and must spend precious testing time trying to extract that knowledge. Perhaps one way to overcome the content problem is to devise procedures for teachers to obtain materials spoken by students which can be typed and presented as reading materials. This would at least ensure that the content of the reading materials is familiar to the student. Carver (1971b) has expressed similar concern about the need for tests that measure how well students can read, rather than what they know about some content area.