Elfrieda H. Hiebert, University of Michigan
In today's polarized and politicized debate about beginning reading instruction, there are striking differences in views of what constitutes effective early reading instruction (Honig, 1996). Some argue the importance of returning to phonetically regular texts to support students' phonemic awareness and understanding of the written code (Adams et al., 1995). While young children do need to learn such skills, engaging literaturerather than the highly controlled contrived textsinvites children to reread, a primary way to learn about the relationship between oral and written language. Further, compelling literature invites children to study the language system (see Sipe, this volume). Children are motivated by the messages in books with intriguing content such as Say's (1993) Grandfather's journey; by what they can learn from informational books such as Yabuuchi's (1981) Whose baby is this? where they can learn the names of baby animals; and by encountering familiar characters and issues when they read Brown's series of books about Arthur and his friends and family. Finally, they can be entertained by the rhythmic nature and humor in Hattie and the fox (Fox, 1988).
However, not all texts with literary merit designed for young readers are equally helpful in early literacy instruction. How do teachers select books? How should publishers and professional groups make recommendations? Which of the thousands and thousands of tradebooks will be the material for study? In this section, I lay the foundations for a framework that supports selections of texts for teaching beginning readers and writers.
This framework for text selection is based on a model of beginning reading processes. Text selection has to happen in relationship to the strategies and skills that young readers are acquiring in the moment, and those that they are moving toward learning. Thus, decisions about texts are based in an understanding of what children are learning, how they learn it, and the text features that can provide support during this process. To capture a comprehensive view of reading acquisition, I believe that educators must turn to the contributions of several paradigmsparadigms that are frequently juxtaposed (Stanovich, 1990).
The contributions of both emergent literacy and social constructivist perspectives (see also Hiebert & Raphael, 1997) are critical because they emphasize linguistic and social processes that had been ignored previously. Emergent literacy perspectives recognize the importance of linguistic understandings that children bring to the task of reading (Brown, 1973). They recognize the role of writing in reading acquisition, as well as the critical role of authentic functions over a focus emphasizing form (Dyson, 1993). Social constructivist perspectives emphasize teacher scaffolding through language and activity, arguing for the importance of dialogue in children's construction of reading and writing tasks and the meanings of the texts (Vygotsky, 1962). Both emergent literacy and social constructivist perspectives underscore the centrality of meaningful and engaging text in early literacy learning. Unfortunately, neither provides guidance for determining which texts can be beneficial for students of varying literacy competence and experience.
Perspectives from cognitive psychology are also critical in this view because of their emphasis on cognitive processing with respect to beginning reading (see Adams, 1990; Juel, 1991). Thus, I draw on the contributions of cognitive psychology for helping determine appropriate texts for teaching youngsters to read, with the caveat that the texts be used in authentic and meaningful activities involving dialogue between teacher and students and among students, as suggested by those from social constructivist and emergent literacy perspectives. I discuss issues related to what research tells us about:
Amount of text relates to the importance of automaticitypractice leading to fluent reading without having to use up conscious attention to processing strategies. Structures of whole text relates to successful readers' competence in drawing on multiple sources of text information. Words within the text relates to proficient readers' underlying knowledge of important distinctions within our language system. I believe that these three constructsamong many within cognitive psychological perspectivesare critical to creating early reading experiences for young learners.
Expert readers are automatic in applying their skills. Research suggests that many American students' problems in learning to read are due to a lack of automaticity, not a lack of fundamental skills (Campbell & Ashworth, 1995). To apply knowledge automatically requires experience with numerous texts, measured in both the amount of time that children spent reading in classrooms (Fisher & Berliner, 1985), and in the amount of text that is provided within reading programs.
The Issue. In reading instruction within the U.S., we have paid little attention to the amount of reading that we require of beginning readers. It is not uncommon to find a single page of text in a reading program surrounded with a full day or two of related activities. In fact, incredible variation exists in the amount of text with which children have interacted before coming to school (Adams, 1990). Some children have extensive home text experiences while others' text experiences have been confined to kindergarten and preschool sessions. For children whose text experiences are first concentrated in the classroom, opportunities to see many different texts are essential to understanding the purposes and nature of literacy. How successful children are when literature is used as the primary source for beginning reading instruction may well be attributed to the amount of text provided, not simply the texts' characteristics.
What We Have Learned from Research. In the research I have surveyed, remarkably little is known about the amount of text needed to support early literacy instruction, especially for those students who do not have extensive reading and writing experiences prior to school. Further, in my own research, I found that I ended up raising more questions than I answered. For example:
In section three, I will suggest guidelines on the relative amounts and related features of books to include in a literature-based reading program, based on my observations of successful classrooms where initially struggling readers learn to read (Hiebert, Colt, Catto, & Gury, 1992).
In addition to scaffolding beginning readers through the amount of reading in their beginning program, a second important feature of support comes from the way in which the texts are structured. Since the late 1970s, scholars from a cognitive science perspective (e.g., Rumelhart, 1977) have demonstrated how proficient readers orchestrate information from multiple sources. They draw on their prior knowledge as well as information from various units of text (letters, words, sentences and types of text). These researchers emphasized that the structures of text influence and potentially scaffold the reading act.
The Issues. Text structures trigger expectations about both the text's topic and the particular words or language used. When young children use particular expectations regarding stories (Stein & Glenn, 1979), based on their oral language interactions and early experiences hearing stories, they are supported as they try to read on their own. For example, beginning readers expect that there will be characters in a story facing a particular problem, that characters will react to the problem, possibly in different ways, and they expect the language to sound like the language they hear and use. When asked to read a selection such as Tinywith its artificial wording, lack of elaboration, and simplistic plotstory expectations are not met, effectively eliminating one important source of text scaffolding. Sulzby (1985) drew attention to the fact that beginning readers use their understanding of texts as a whole to "read" the illustrations of books or to "read" a memorized chant from a book. When the information at the word level is not yet available to children, their text expectations draw their attention to individual words and support the development of an ever-expanding reading vocabulary.
What We Have Learned from Research. Research about how texts are structured points to at least three features whole texts that can support beginning readers: (1) predictability, (2) imagability and familiarity of concepts, and (3) word density. If each is viewed in terms of a scaffoldthat temporary, adjustable form of support (Wood et al., 1976)each feature could be highly exaggerated in the earliest materials. Then, over time and reader experience, each would be reduced in emphasis. Using word density as an example, first, there would be relatively few different or unique words in a text when compared to all of the words in early texts. Second, the number of total words in a text would be considerably fewer early on than in later texts. If we consider these two featuresunique words to total wordswe could begin to identify the word density of the text as a whole. For beginning readers, texts would be structured so that the word density is very low. Over time, the number of words in a text and the ratio of unique words to total words would be expected to change, creating "denser" text. In Table 1, steps along the way on three text dimensions: predictability, imagability, and word density are presented
Characteristics of Texts
Characteristic
Components
Evaluation Criteria
Predictability
Size of predictable unit
Small predicatable unit
3–5 words
Sizable unit
6–10 words
Large unit
11–15 or more words
No predictable unit
Proportion of text accounted for by predictable unit
Very high
3/4 or more
High
1/2 to 2/3
Moderate
1/4 to 1/3
Low
Less than 1/4
Contextual
Familiarity of concepts
Very familiar
Familiar
Somewhat familiar
Not common
Usefulness of illustrations in identifying key words
Very useful
Useful
Somewhat useful
Not useful
Word Density
Number of distinct words
Count of distinct or different words
Ratio of distinct to total words
Number of all words is divided by the number of distinct words
Characteristics of Distinct Words
Characteristic
Components
Evaluation Criteria
Decodability
Of distinct words in text, proportion of words with V-C rime is:
Very high
3 of 5
High
2 of 5
Moderate
1 of 5
Low
less than 1 of 5
High-Frequency
Of distinct words in text, proportion of words that are 20 most frequent words in written English is:
Very high
3 of 10
High
2 of 10
Moderate
1 of 10
Low
less than 1 of 10
(1) This table is a modified version of one that appears in: E.H. Hiebert & T.E. Raphael (1997). Early literacy instruction. Fort Worth, TX: Harcourt Brace College Publishers, p. 114. Used with permission of the authors and publisher.
Predictable text structures. Books with predictable text structures provide an important source of scaffolding for beginning readers. This is supported by research by cognitive psychologists who found that how texts are structured influences meaning-making (Beck et al., 1984; Stein & Glenn, 1979) and by research on children's language acquisition (e.g., Brown, 1973). What makes a text predictable? It is the rhythm, rhyme, and repetition seen in the nursery rhymes and songs that are read and told to young children from "Patty-Cake" to This is the House That Jack Built. While predictable texts were used in American textbooks over a century ago (see Stickney, 1885), they were discarded with the promise of a new science of reading in the 1930s (Smith, 1934/1965). Because of findings from the research by cognitive psychologists and psycholinguistics, predictable texts were again integrated into post-1980s American reading texts.
Bill Martin's (1967)) Brown Bear, Brown Bear: What Do You See? became the prototype for the "new" predictable text, as did his notion of "wholebooksuccess"that children could "read" the whole book successfully when the syntactic pattern was predictable, and thus grasp the power and pleasure of reading before they had acquired word recognition skill (Martin & Brogan, 1971). Through the "shared book experience," where an adult or experienced reader pointed to the text in the predictable pattern as he or she read aloud (Holdaway, 1979), beginning readers learn to associate written words with the oral words they recite from memory. Analyses of the predictable book genre showed that authors achieved predictability by using a variety of structures, including refrain, cumulative, compare-contrast, and episodic or enumerative patterns (Bridge, 1986).
The particular forms of the predictable patterns may be less important, however, in children's acquisition of independent reading skills than two other aspects of predictability: (a) the size of the repeated unit, and (b) the proportion of the text that is accounted for by the repeated unit (see Table 1). In terms of the predictable unit's size, while large units may be memorized by young children, the task of memorizing may actually hinder their focusing on the individual words. To illustrate, Sulzby (1985) describes a child named Doug who is pretend-reading a book quite familiar to him, The Carrot Seed. Doug consistently "reads" the predictable unit, "I'm afraid it won't come up," as "Nothing's gonna grow." While he has the gist, he clearly isn't attending to either the number or the actual individual words, which could impede instead of support his word learning. The second aspect of predictable text is the proportion of the text devoted to the predictable unit. If a substantial portion of the text is accounted for by the predictable unit, children will be able to apply their knowledge considerably more than if the predictable unit takes up a less significant part of the text.
For the feature of predictability, texts children initially read would have a small predictable unit that accounts for a high proportion of text. Over the course of children's reading experiences, the scaffold would gradually be withdrawn, increasing the size of the predictable unit and decreasing the proportion of text accounted for by a predictable set of sentences or phases. As children's reading proficiency increases, teachers should steadily pull away the scaffolding of the predictable text, using other texts of literary merit and different text structures.
Imagability and familiarity of topics. When children cannot recognize even a handful of words, the illustrations of books may provide important contextual support for highly meaningful words. In earlier texts such as Tiny, illustrations provided the meaning context, but were of little assistance in figuring out particular words. Since only high-frequency words are used in Tiny, the illustrations depicting a young boy and girl chasing a piglet around a barnyard do not map onto the words actually used in the text. The features of illustrations that support children's word identification in connected text are demonstrated in Martin's Brown Bear, Brown Bear, What Do You See? where the key words on the page that are not part of the predictable pattern pertain to familiar concepts and can easily be figured out from the illustrations (yellow duck, red bird, gold fish, and so forth). The illustrations in Brown Bear, Brown Bear, What do You See? demonstrate two characteristics that can scaffold word identification: (a) the imagability of and (b) the familiarity of the concepts.
Even without accompanying illustrations, words with high imagery valuesthose associated with clear and concrete imagesare learned more easily by beginning readers than those with low imagery values (Paivio, 1968). Words that have high imagery values tend to be nouns (van der Veur, 1977), and to a lesser degrees, actions. In contrast, the imagery value of most high-frequency words (e.g., the ten most frequently used words in written Englishthe, of, and, a, to, in, is you, that, it) is uniformly low. Imagery value is illustrated in Eric Carle's illustrations in Brown Bear, Brown Bear, What Do You See? where he created clear representations of brown bear and red bird. To make sense of the text, however, beginning readers need the predictable sentence, "What do you see?", and they need to hear it modeled initially by a capable reader.
Not all illustrations are equally useful as scaffolds for beginning readers. For illustrations to scaffold specific word identification, they must depict concepts with which young children are highly familiar. In a picturebook such as Zin! Zin! Zin! Violin (Moss, 1995), instruments such as a bassoon or clarinet are clearly depicted in the illustrations and named in the text. Yet, these illustrations will be of little use to young children unfamiliar with the instruments.
In Table 1, I provide gradations in dimensions of both imagability and familiarity of concepts, from very high to low, from very familiar to not common. Initially these two dimensions may follow the same course. However, over time, the scaffold of the clear depiction of imagable words will be steadily withdrawn, while the scaffold of familiar content may be retained to support children's use of semantic cues. Even when students' word recognition increases, illustrations may continue to provide a scaffold, helping convey events in a story episode, rather than individual words.
Word Density. The third feature of the whole text that can support, or impede, beginning readers' ability to acquire and apply reading strategies is the density of words. Density compares the number of unique words to the total number of words in the text. Every time beginning readers encounter a different word in a book, they need to draw on their emerging repertoire of word recognition strategies.
Two texts I described earlier, Tiny and If I Had a Pig, illustrate the contrast in processing demands that differences in the word density of books pose for beginning readers. In Tiny, there are 19 different words among the 132 words in the passage. While If I Had a Pig has fewer total words (95), 57 of these are unique. If the unique or different words are viewed in relation to the total number of words, in Tiny every seventh word could be new. In If I Had a Pig, every other word could be new.
However, these two passagesTiny and If I Had a Pigindicate that word density alone tells only a part of the story. All words are not equal in their difficulty. Remembering words, even when the number of unique words is few, can be difficult when the referents are unclear, the letter-sound patterns inconsistent, and the syntactic and semantic contexts ambiguous as is the case with Tiny. The whimsical pig in If I Had a Pig and the child who is narrating the story are more memorable and engaging than Tiny and its two unnamed owners will ever be. But, no matter how much interest children have in a passage, attention to 57 different words can be overwhelming for beginning readers. Vocabulary density means that children cannot consistently apply their growing strategies related to high-frequency words and common rimes. In all likelihood, the teacher will read aloud the book and children will distinguish few, if any, words from it.
Because word learning is a function of numerous factors, it is highly unlikely that precise formulas dictating the number of unique words which beginning readers can successfully process at particular points in their development will or can ever be identified. Some unique words have easily accessible patterns. Other words hold interest for children and are highly imagable or recognizable from the illustrations. While the issue is complex, understanding the range of whole text features that influence word learning can help make text selection less arbitrary.
Research by cognitive psychologists and linguists highlights the rules our language follows. Albeit not always perfectly. Successful learners know critical features of the systems that make up language, including the sounds represented by the symbols, the order in which words appear in a sentence, and the affixes that convey meaning when added to a word. However, proficient readers do not cite the rules, nor do they explicitly apply a rule when encountering an unfamiliar word. Rather, they use their knowledge through analogies (Cunniungham, 1992).
The Issues. Whatever the words children are learningsat, bat, cat, hat or you, to, the, from, have, or play, plays, playing, played, playfulearly theories of learning suggested having students repeat
words they were shown until they could produce the word accurately (e.g., using flash cards). Today, we continue to recognize that repetitive methods are quite helpful when learning words that have less than perfect sound-symbol connections (e.g., have, the). However, for many other words, proficient readers draw on rules as they play out in both words and in morphemic units such as inflected endings, prefixes, and suffixes.
What We Have Learned from Research. How might the materials at the earliest levels support children's early acquisition of word recognition strategies? As I described in the opening section of the chapter, this research has examined: (a) phonetically regular words and (b) high-frequency words.
Phonetically regular words. Reviews of children's reading acquisition report a similar pattern: Children benefit from information about the consistent patterns within written words (Adams, 1990; Anderson et al., 1985; Bond & Dykstra, 1967). Once youngsters grasp the relationship between sounds and written symbols, they can read words that, when decoded, they understand because the words are familiar ; they are part of their oral vocabulary. Once children are facile with common and consistent letter-sound patterns, they can be introduced to the morphemic patterns of English. For example, a group of words with a common morphemeplay, playing, played, playful, playerwill be read first on the basis of its common letter-sound pattern and the shared meaning of the different forms of the word understood subsequently.
Early reading can be supported using texts where common and consistent letter-sound patterns of English are made visible. However, when all the words are selected for consistency, texts run into problems, as illustrated in the earlier example of The Bad Fan. Surprisingly, there isn't much research informing teachers about the best balance between words with the target letter-sound patterns and other words in the text. In my preliminary framework, I focus on the presence of particular patterns within words, based on the following lines of research.
First, Juel and Roper-Schneider's (1985) work indicates that children apply their phonics strategies more consistently when the texts that they are reading have many examples of these patterns. Their line of work suggests that if children are learning about simple vowel patternsa typical point of departure in American reading instructiontexts should contain at least a modicum of words with these patterns. But, which ones should be used?
Wylie and Durrell's (1970) research helps determine which patterns might be used. They found that 37 different rimes (i.e., patterns) account for 500 words that occur in primary-level text, though their frequencies vary markedly (Solso & Juel, 1980). In related research (Juel & Solso, 1981), there is evidence that young readers benefit from repeated exposures to different words using the same pattern (e.g., bat, cat, hat, mat, flat), than seeing the same word repeatedly (e.g., hat). Thus, instead of the perspective reflected in The Bad Fan, readers would be better off if a text contained various iterations of a rime rather than six iterations of a single word like rat. Moreover, it would make sense to emphasize the most common rimes.
Young children begin to develop a "metacognitive stance" toward their readingengaging strategically in trying to decode wordswhen a high portion of words in their stories are decodable (e.g, contain a familiar V-C rime). Strategic approaches to decoding words can be enhanced when young children are taught to generalize from the simplest patterns they are learning. Later in their program of instruction, rimes are introduced with more complex vowel representations such as the V-C-e (ride) and V-V-C (meat). Words with these patterns and from the earlier taught patterns (V-C rimes) should be considered in selecting text.
High-frequency words. For learning high-frequency words, early educational psychologists were on the right track, expecting beginning readers to learn specific focus words at particular points in their reading development. Martin & Hiebert (1997) demonstrated that initially struggling readers who became successful readers during first grade knew few words by mid-year but, once they had acquired a core group of high-frequency words, they progressed rapidly in their word recognition skills. The key is that they learned a small group of such words first, then built from that experience to learn other high-frequency and phonetically regular words much more quickly than occurred in their first smaller set.
Ironically, while the initial researchers in this area were right in expecting beginning readers to memorize a subset of high-frequency words from a common list, such as the Dolch 220 Basic Sight Words, they were wrong to simply divide up the words into subset lists of 40 to 50 words each that were to be learned by children sequentially. The magnitude of such lists is overwhelming to the beginning reader, and the neat division into relatively equal lists assumes an incremental learning process. In Table 1, notice that the difficulty of a text for beginning readers is based on the presence of words among the 20 most frequent words in written English (Carroll, Davies, & Richman, 1971). Acquiring these words will be a challenge for many beginning readers. But once this proficiency is gained, children will move more quickly in learning other words.