Paper presented at the Annual Meeting of the American

An argument is made that early reading ought to be studied from three perspectives: the function of print, the form of print and the conventions of print. In so doing it may be possible to avoid some of the hazards that have plagued the field, namely, unsubstantiated assumptions about beginning reading and how it ought to be taught, erroneous beliefs that maturation plays an overriding role in learning to read, and shortsighted approaches to assessment of young children's knowledge of and progress in early reading. Acquisition of Knowledge About Reading in the Preschool Period: An Update and Extension Think back to your childhood. Do you have a memory about learning to read? Many of us do. When I ask this question to those who have a distinct memory about when or how they learned, I find that it is often tied to a particular book. For myself, it was Beatrix Potter's book, Peter Rabbit. Of course, I have no idea now whether it is an accurate memory and whether it helped me to read in school. Did I actually learn to read the book or was I reciting it from memory? What did I learn by memorizing the story and did it help me read other stories? These are questions none of us who have such memories can answer. Furthermore, since young children might read in ways that are unlike adults or older children, the process probably cannot be extrapolated from models of skilled reading. Nevertheless, while it is difficult to gather reliable retrospective data, it is possible to construct processing models from analyses of children's early attempts to read, recite and interpret printed information, and in so doing to chart the development of their approaches to reading. What a typical child knows about reading before going to school would seem to be a reasonable question. Yet it is one that is fraught with hazards, influenced not as much by research as by the implicit models we have of reading and by the hidden Acquisition of Knowledge 3 Acquisition of Knowledge 4 assumptions we make about how children learn. I will describe three hazards so it will be more apparent why the question has been difficult to answer. Following this I will propose a model of early reading and then will describe data we collected that support some aspects of the model. Three Hazards to the Study of Early Reading Instructional assumptions. One hazard to the question, what do preschool children know about reading, is that our views of how reading takes place, and extrapolating from that, how it should be taught, interferes or biases the way we ask the question. This is partly because the field is not in agreement about how reading occurs and, as a result, about how to teach children to read. Look, for example, at the number of alternative programs purporting to show effective ways to teach beginning reading (Aukerman, 1971). To reduce complexity we typically classify them in terms of one or another assumed reading processing model, skill-based or holistic. Skill-based programs rely on a model of reading in which the beginning reading process is assumed to have a linear quality. The more strictly organized of these is called a code-emphasis program (Beck, McKeown, & McCaslin, 1979) or a linguistic program (Chall, 1967). Proponents of this model, as evidenced from the quotes below, emphasize that the process is initiated with letters, words, or their sounds and then proceeds to larger units of text. It is a "bottom-up" model. Once a child begins his progression from spoken language to written language, there are, I think, three phases to be considered. They represent three different kinds of learning tasks, and they are roughly sequential, though there must be considerable overlapping. These three phases are: learning to differentiate graphic symbols; learning to decode letters to sounds; and using progressively high order units of structure. (E. Gibson, 1976, p. 254) In the information-processing approach that we have proposed, reading involves the successive recognition of larger and more abstract meaning . . . from the recognition of word meaning to the recognition of the meaning of phrases, sentences and stories. (Venezky, Massaro, & Weber, 1976, p. 695) . .. the transformation of written stimuli into meanings involves a sequence of stages of information processing. (LaBerge & Samuels, 1976, p. 551) Holistic programs assume that the reading process, as well as its instruction, is not linear but interactive and tightly bound to meaning. Some basal reading programs from the 1940's and 1 9 50's (those which featured a whole word approach to beginning reading) and, more recently, language experience programs follow many characteristics of this model. In the next quotes, notice the assumption that reading instruction must be Acquisition of Knowledge 5 Acquisition of Knowledge 7 formed around understanding and interpreting text. They are "top-down" approaches. Reading is a psycholinguistic guessing game. It involves an interaction between thought and language. (Goodman, 1976, p. 498) . . . a child learns to read by reading. (Smith, 1980, p. 421) If learning to read and write is to constitute an act of knowing, the learners must assume from the beginning the role of creative subjects. It is not a matter of memorizing and repeating given syllables, words, and phrases, but rather of reflecting critically on the process of reading and writing itself, and on the profound significance of language. (Friere, 1980, p. 369) The viewpoint described by the first set of quotes is usually interpreted to indicate that reading has a hierarchical nature. The second emphasizes the interaction between meaning or language and print. A problem with the first viewpoint is that, while the research does indicate that our eyes read and process very small bits of text at a time (see, for example, McConkie, 1982), it can neither be assumed that the young child reads in the same way as an adult nor that the most effective instruction is to recognize first letters, then words, then larger units of text. One argument against that ordering for instructional purposes is that letters having no intrinsic meaning are not easier to learn than words. Further, words, if placed out of context, often carry very little of their intended meaning (Anderson & Ortony, 1975; Bollinger, 1965). As we showed in a study with children (Mason, Kinseley, & Kendall, 1979), being able to identify printed words (e.g., polysemous words such as pitch, jam, switch) does not guarantee that appropriate contextderived meanings are recognized. A problem with the second viewpoint is that it lacks a clearly formulated instructional approach. The "look-say" or sight word approach was rejected as a result of Chall's 1967 survey of instructional effects. Other meaning oriented approaches either have not been rigorously evaluated (language experience) or are still being studied (Tharp, 1982). As a result, beginning reading instruction is more influenced by a hierarchical model of reading than by one that focuses on meaning. While the instructional issue has not been resolved, it can be hedged by taking great care that teachers encourage text understanding and interpretation. More specific changes await evidence from long-term investigations of young children's developing knowledge of reading. By tracking children's knowledge from or before kindergarten when they more often can choose what and how to spend their time and learn, and then Acquisition of Knowledge 6 Acquisition of Knowledge 8 follow them into school, tracking their reading instruction, it might be possible to separate school instructional effects from early home learning effects. For example, in a 1980 study, I observed and tested children throughout their year in a university nursery school. I found that they began learning to read by recognizing their own name, food labels, and traffic signs. Their early awareness of print was centered on highly meaningful words in context and was followed by active attempts to spell words and to analyze words in terms of their letter sounds. Informal follow up indicate that they continued to excel in reading. Bissex (1980) who observed her young son from age four, found that he began learning to read by merging reading and writing with its meaning. Yet teachers are often urged to begin reading instruction with meaningless, out-of-context, letter-sound and word recognition drill. Is this really the most effective initiation into reading? We don't know, but clearly, this is an issue that must be studied. Learning assumptions. A second hazard to answering questions about what a chilld knows about reading before going to school is found in assumptions about how children learn. Despite research evidence to the contrary (for example, Brown, 1975; Chi, 1976), many educators appear to believe that what children do and are able to learn is profoundly limited by their age or maturity. The field of reading particularly has been influenced by statements that focus on effects of the chronological or mental age of the child. For example, a long-standing statement is that "the age of six is the crucial age" for learning to read (Morphett & Washburne, 1931; Heffernan, 1960; Hildreth, 1950). Further, research from the 1920's and 1930's often emphasized how intellectual endowment affects the age a child can learn to read (for example, Cox, 1926; Davison, 1931). What they and others failed to study in the same depth are relationships between age (or intellectual endowment) and home background experience in learning to read. Hence, conclusions that only age and IQ form important ties to reading have misled educators into believing that early instruction is unimportant. At one point, an even stronger argument was made that early reading instruction could harm children. Here, for example, is the way Gesell stated the issue: The attempt to force reading [by the age of six] frequently leads to temporary or permanent maladjustment and more or less