You are on page 1of 58
DEVELOPMENTAL REVIEW 14, 245-302 (1994) On the Efficacy of Reading to Preschoolers Hou ts S. SCARBOROUGH AND WANDA Dosricu Brooklyn College, City University of New York We review more than three decades of empirical research pertaining to the hypothesized influence of parent-preschooler reading experiences on the devel: ‘opment of language and literacy skills. The literature provides evidence for this association, although the magnitudes of the observed effects have been quite variable within and between samples and, on average, have been unexpectedly modest. Demographic, attitudinal, and skill differences among preschoolers all apparently made stronger direct contributions to prediction in investigations that permitted such comparisons. Implications of the findings for theory and research on literacy acquisition, for educational practice, and for parental guidance are discussed, © 194 Academic Pres. Reading loud is one of the best-kept secrets of good parenting. It has a tremendous impact on a young mind and gives youngsters a head start on their education Barbara Bush in Reader's Digest (1990) Reading aloud to your children .. . is the best way to help youngsters become readers and also promotes a good attitude toward reading that they carry with them throughout ther lives. Parent Teachers Association in Redbook (1987) The single most important activity for building the knowledge required for eventual success in reading is reading aloud (o children ‘Commission on Reading, National Academy of Education (1985) ‘And of course we've read to her every night since she was one, like you're sup- posed to, Recent comment by a well-educated young mother It is common knowledge that reading to young children is important. This message has been conveyed to today’s parents in numerous articles in popular family magazines (e.g., Bush, 1990; Marzollo & Sulzby, 198% PTA, 1987). Assertions regarding the efficacy of reading to preschoolers can also be found in many more scholarly sources, including the report of the Commission on Reading of the National Academy of Education (1985), numerous books and articles by prominent educational research- ers (e.g., Adams, 1990; Goldfield & Snow, 1984; Teale, 1984, 1986), and We thank B. DeBaryshe, C. Crain-Thoreson, P, Dale, D. Dickinson, D. Dunning, and R. Watson for generously providing the additional information that we requested about their work, We are also grateful to S. Choi and M. Waibel for their assistance during the prep- aration of this review and to the March of Dimes Birth Defects Foundation for financial support. Address reprint requests to Hollis S. Scarborough, Department of Psychology, Brooklyn College of CUNY, Brooklyn, NY 11210-2889, us 0273-2297198 $6.00 Copyright © 1988 by Academe Press, ‘Almas of reprodocton in any foo reserved 246 SCARBOROUGH AND DOBRICH college textbooks. In short, there is widespread agreement that joint par- ent-preschooler reading is a highly beneficial parental practice that pro- motes the acquisition of literacy-related knowledge and, consequently, paves the way for successful achievement, The potential importance of early literacy experiences is underscored by recent longitudinal findings on the development of reading. It is now abundantly clear that children differ widely in their acquired knowledge about books and reading at the time of school entry and that these dif- ferences are predictive of their subsequent academic achievement in the early school years and beyond. Identifying the sources of these differ- ences in preparedness is thus necessary for arriving at a full understand- ing of the process of learning to read from its inception during the pre- school years and is crucial to enable us to provide accurate guidance to parents and preschool educators regarding the conditions that best foster such preparedness. ‘This issue is also of particular interest nowadays because the presump- tion that reading to young children is beneficial to literacy development has increasingly come to influence contemporary educational theory and practice. A basic tenet of literature-based approaches to early reading instruction is that pleasurable and purposeful adult-guided exposure to books is a more natural and effective means of promoting the acquisition of literacy than are more traditional “code-based™ curricula (e.g., Clay, 1979; Goodman, 1987; Taylor, Blum, & Logsdon, 1986). Detailed obser- vations of individual children have documented that parent-preschooler reading and many other literacy experiences in the home can provide a wealth of opportunities for acquiring knowledge about reading and writing (e.g., Bisex, 1980; Ferreiro & Teberosky, 1982; Heath, 1983; Scollon & Scollon, 1981; Snow & Ninio, 1986; Taylor, 1983). Although these eth- nographic and case studies contain many accounts of children’s apparent learning of specific aspects of literacy skill through prior exposure to particular experiences, it is difficult to establish definitive cause-and- effect relations from such descriptive and anecdotal reports. Neverthe- less, on the basis of these and some more quantitative research findings (reviewed below), many educators have argued that the family interac- tional practices prevalent in the histories of good readers are largely re- sponsible for those children’s success in learning. Consequently, litera- ture-based educational programs often seek to provide similar opportu- nities for enjoyable and facilitative interactions within school settings, with the presumption that the acquisition of reading and writing will be enhanced, Furthermore, parental practices—especially joint parent-child reading, but also library excursions, book ownership, and other literacy-related activitieshave been hypothesized to explain not just individual differ- READING TO PRESCHOOLERS 247 ences in literacy development but also group differences. As will be dis cussed below, children from lower socioeconomic and nonmainstream cultural communities often exhibit somewhat poorer school achievement, and their less-extensive knowledge of literacy is evident even by the time they begin school. The amount and nature of parent-child joint literacy experiences during these children’s preschool years may be very different from those typical of mainstream families, and such cultural differences in parental practices may set the stage for future differences in academic. achievement. This line of thinking, however, is contingent upon the pre- sumption that parental practices indeed play the important role that has been ascribed to them. But how important for literacy development is reading aloud to pre- hoolers? It did not occur to us to question the efficacy of shared reading with young children until we began to study its relation to literacy achievement in our own longitudinal research (Scarborough, 1989; Scar- borough, Dobrich, & Hager, 1991). In conjunction with that effort, we looked more closely at the body of prior work on this issue and discov- ered, as did Teale (1984), that the notion that reading to preschoolers makes an important contribution to literacy development has usually been accepted uncritically. We also discovered that the evidence in support of, this assumption was not as strong or consistent as we expected it to be, given the widespread acceptance of this hypothesis. The purpose of this review, therefore, is to acquaint others with that body of work and with some of our concerns about what can be concluded from it. Potential Benefits of Parent-Preschooler Shared Reading: Outcome Measures Many desirable outcomes might conceivably be fostered by reading to preschoolers in the home. For example, shared bookreading might serve to strengthen the emotional ties between parent and child, to acquaint the child with factual information about the world, to provide the child with an appreciation of pictorial representations, and so forth. Nevertheless, it is generally believed that the primary benefit of reading regularly to young children is the fostering of literacy development, and it is on this issue that ‘we will focus. In studies of the efficacy of parent-preschooler reading, several kinds of measures have served as indices of children’s emerging literacy skills: traditional achievement scores during the primary grades; “emergent” literacy skills prior to school entry; and oral language abili- ties during the preschool period. Those outcome measures, and the ra- tionale for using each, will be compared briefly here. Achievement tests. Literacy achievement is most traditionally viewed as a constellation of skills that improve during the school years as a result of formal instruction and practice. Reading, spelling, and writing skills are 248, SCARBOROUGH AND DOBRICH all important components of literacy achievement. Individual differences in these abilities have been shown to be quite stable from year to year, in that children who acquire literacy skills most readily in the earliest grades tend to remain the best readers, writers, and spellers throughout the elementary school years, and children who have difficulty at the start often continue to exhibit relatively weak skills (e.g., Butler, 1988; Juel, 1988; Mason, Stewart, Peterman, & Dunning, 1992; Wells, 1986), Primary grade acquisition of reading, writing, and spelling skills is thus an appro- priate outcome measure for evaluating the efficacy of parent-preschooler reading, and it is undoubtedly the foremost concern of parents who seek to prepare their preschoolers for academic success. Several studies to be reviewed have assessed these traditional literacy achievements through. the use of standardized and nonstandardized tests and, occasionally, classroom teachers’ evaluations. Emergent literacy skills. The year-to-year stability of individual differ- ences in literacy skill also extends downward into the preschool years. That is, literacy-related skill differences among children when (or shortly before) they enter school are moderately to strongly correlated with sub- sequent performance on traditional achievement measures (e.g., Mason et al., 1992; Scarborough, 1989; Share, Jorm, Maclean, & Matthews, 1984; Wells, Barnes, & Wells, 1984). These “emergent literacy” or “preliteracy” abilities include identifying and printing letters, under- standing some rudimentary letter-sound relations, recognizing some printed words, knowledge of the mechanics and purposes of bookreading, familiarity with narrative and expository prose forms, and so forth. When obtaining outcome data at older ages is not feasible, assessments of such early literacy skills may serve as a reasonable estimate of later achieve- ment outcomes. Moreover, preliteracy assessments are not affected by instructional factors after school entry and thus may be more sensitive to the influence of preschool experiences such as shared reading in the home. Several studies to be reviewed have examined the efficacy of par- ent-preschooler reading in relation to this type of outcome measure. Oral language abilities. Another well-established precursor to differ- ences in literacy ability is oral language proficiency. Differences in se- mantic, phonological, and syntactic abilities during the late preschool years are typically among the best predictors of literacy achievement a year or more later (e.g., Bishop & Adams, 1990; Butler, Marsh, Shep- pard, & Sheppard, 1985; Mann & Ditunno, 1990; Share et al., 1984; Wolf, Bally, & Mortis, 1986). Furthermore, there is growing evidence that oral language abilities at even earlier ages (2% to 4 years) can be good pre- dictors of literacy achievements many years later (e.g., Bryant, Bradley, Maclean, & Crossland, 1989; Scarborough, 1990, 1991; Stuart & Colth- cart, 1988). In light of this association between developing oral and wri READING TO PRESCHOOLERS 249, ten language skills, some studies of the efficacy of reading to preschool- ers, especially those in which the samples are very young, have justifiably elected to use language measures as outcome variables. Needless to say, literacy and language are not monolithic abilities, but rather are constellations of related but separable processes and skills. It entirely possible that the effects of parent-preschooler shared reading may be limited to only some facets of language and literacy development. For example, preschool differences in formal language proficiency (pho- nological and syntactic skills) may be more closely linked to mastery of the more formal aspects of literacy acquisition (like decoding and spell- ing), while early differences in language content (lexical-semantic and pragmatic skills) may be forerunners of later reading comprehension. Contextual, in addition to form/content, distinctions may also be impor- tant. For example, children’s grasp of ““decontextualized” language skills (the reliance on language alone, rather than in conjunction with contex- tual cues, to convey messages) may be more relevant to literacy acqui- sition than their speaking and listening abilities under conditions of greater contextual support. As will be seen, however, few studies of the efficacy of parent-preschooler reading have been concerned with these potentially important distinctions. Other Factors Related to Outcome Measures Preschool emergent literacy skills and oral language abilities are not the only predictors of subsequent achievement during the school years. In this section, we will briefly discuss two other factors that are associated with differences in reading acquisition and that should be taken into ac- count in interpreting research on the relation of parent-preschooler read- ing to literacy/language outcomes. Socioeconomic status (SES). Differences in literacy achievement be- tween mainstream children and those from disadvantaged backgrounds have often been compared. Most investigations of SES-related differ- ences in achievement have relied on demographic indices such as parental income, occupation, and education. In such studies, the correlations of demographic differences with individual differences in literacy ability have often been surprisingly weak (accounting for less than 15% of vari- ance) even in large, population-representative samples (e.g., Alwin & Thornton, 1984; Estrada, Arsenio, Hess, & Holloway, 1987; National Assessment of Educational Progress, 1985; Rowe, 1991; Walberg & Tsai, 1985; White, 1982). These SES-related differences in achievement tend to be more pronounced in the higher school grades than at the outset of schooling (e.g., Applebee, Langer, & Mullis, 1988; Hanson & Robinson, 1967), but can nevertheless be observed in studies of preliteracy abilities during the late preschool years (¢.g., Dickinson & Snow, 1987; Estrada et 250 SCARBOROUGH AND DOBRICH al., 1987; McCormick & Mason, 1986; Morrow, 1983; Mortenson, 1968; Telegdy, 1974; Wells, 1986). More erratic findings have been obtained, however, regarding associations between SES and preschool oral lan- guage abilities (e.g., Butler, 1988; Dickinson & Snow, 1987; Howard, Hoops, & McKinnon, 1970; Ninio, 1980; Raz & Bryant, 1990; Tough, 1977; Warren-Leubecker, 1987). Because how well individual children acquire literacy skills is not ter- ribly well predicted by conventional demographic indices of SES, it has been hypothesized that measures of the home environment may be more accurate predictors of achievement, and this notion has been borne out in several large-scale analyses of samples of grade school and high school students (e.g., Cox, 1987; Iverson & Walberg, 1982; Teale, 1984; Wal- berg, 1984; Walberg & Tsai, 1985; White, 1982). For instance, in a sample of 1459 9-year-olds from the National Assessment of Educational Prog- ress data base, Walberg and Tsai (1985) determined that reading achieve- ‘ment was less strongly correlated with demographic indicators (.22) than with the amount of reading materials in the children's homes (.30). Sim- ilarly, ina meta analysis of 101 studies, White (1982) found that the mean correlation with achievement was higher for “home atmosphere” predic- tors (.58) than for demographic indices (.19-.33). As summarized by Wal- berg (1984), the results of these investigations of older children suggest that “what might be called ‘the curriculum of the home’ predicts aca- demic learning twice as well as the socioeconomic status of the family"” (Walberg, 1984, p. 400). At earlier ages also, home environmental differences may be more important than demographic indices of SES in accounting for variation in children’s literacy development, and parent-preschooler reading experi- ences could contribute substantially to the overall literacy environment of a young child's home (e.g., Teale, 1984). As will be seen, in several studies of the efficacy of reading to preschoolers, the relative contribu- tions of early home experiences and demographic SES indicators to out- come differences can be compared. Attitudinal differences. During the school years, motivation and inter- est are related to academic achievement (Juel, 1988; Paris, Olson & Stevenson, 1983; Wigfield & Asher, 1984). Attitudinal influences have largely been neglected, however, in theoretical accounts of the preschool antecedents to literacy acquisition, even though it is reasonable to pre- sume that a child who has the desire to learn to read, who is interested in literacy-related activities, and who voluntarily engages in them will, all else being equal, turn out to be a better reader, writer, and speller than a child with lower interest and motivation. Although there has been less research on attitudinal than on cognitive, linguistic, and socioeconomic READING TO PRESCHOOLERS 281 antecedents to literacy, the available findings suggest that early differ- ences in attitude may also be important. We will describe those results in some detail here because they have not been reviewed elsewhere and may be of particular relevance to the problem of assessing the effects of par- ent-preschooler shared reading on literacy development. Early interest and motivation toward literacy have usually been as- sessed by using parental reports about the child's perceived desires and preferred activities. In a very early study, for instance, Almy (1949) in- terviewed the parents of 106 first graders and evaluated the children’s reading skills. A composite score based on 22 interview responses, 19 of which concerned the child’s behaviors and attitudes toward literacy ac- tivities, was correlated (r = .26) with reading scores. (Because severe floor effects were obtained for the achievement measures, this correlation was probably attenuated by the restricted range in the sample.) Despite Almy's conclusion that success in beginning reading is related to the preschooler’s “responses to opportunities for reading™ (p. 111), relatively few researchers have pursued this question in the 40 years since her well known study was published. One valuable source of information about relationships between pre- school development and later achievement is the Bristol project directed by Gordon Wells (Wells, 1985a, 1985b, 1986; Wells et al., 1984). When the children in that longitudinal study were 5 years old, the parents of a subsample of 32 subjects were interviewed about the children’s literacy- related experiences during the preceding 2 years. Preliteracy knowledge scores at age 5 (which were strongly related to subsequent reading achievement at ages 7 and 10) were correlated with parental reports of: the degree of concentration that the preschooler had displayed when en- gaged in activities associated with literacy, r = .56 (Well, 1985b; Wells et al., 1984); the child’s perceived interest in literacy, r = .45 (Wells, 1985b); and how long the preschooler typically chose to spend on activities as- sociated with literacy, r = .65 (Wells, 1986). When actual observations of children’s responses to literacy experiences were examined, the results generally coincided with those based on parental reports. Moon and Wells (1979) selected from Wells’ Bristol sample a subgroup of 20 children who represented the two extreme socioeconomic classes. Preliteracy skill at age 5 was found to correlate (r = .43) with the observed frequency of the child’s literacy-related actions and utterances during audiorecorded lan- guage samples that were taken in the subjects’ homes over the preceding 2-year period. Somewhat weaker correlations (r = .33 to .40) were ob- tained with reading scores at age 7. (A composite interview-based mea- sure of preschool interest in literacy was not reliably correlated with the observational measure, r = .27, and was only weakly related to outcomes 282 SCARBOROUGH AND DOBRICH in the sample, r = .20 to .38; the latter results are inexplicably at odds with the stronger correlations obtained for the larger, more representative sample in Wells’ subsequent analyses of the same interview responses.) In another observational study, Crain-Thoresen and Dale (1992) con- ducted a prospective longitudinal investigation of a sample of 25 children whose language skills at age 20 months were exceptionally advanced From videorecordings of parent-child joint reading at age 24 months, a measure of children’s “engagement” in the task was derived and ana- lyzed as a predictor of subsequent differences in language and preliteracy abilities. Indications of engagement included the child’s directing his or her gaze at the book, pointing to pages, making book-related comments, holding the book appropriately, turning pages, and so forth. Engagement was predictive of receptive vocabulary and syntax at age 2% years (with receptive vocabulary scores at 24 months statistically controlled in the regression analyses) and with knowledge of print concepts at age 47 (with MLU at 24 months and amount of preschool reading instruction con- trolled in the analyses). Scores on other emergent literacy measures, however, were not related to engagement Morrow (1983) examined motivational factors in an observational study of somewhat older children, From classroom observations of 396 kinder- garteners over an 8-week period, the 75 children with the highest literacy interest and the 97 with the lowest interest were identified according to a scale that primarily reflected how often the children engaged in literacy activities during free activity periods and how long they continued to look at books during recreational literacy periods. A large difference between the groups was seen in their mean percentile scores on a standardized reading readiness test (86th vs 58th). Conversely, only 8 children with poor preliteracy skills had displayed high interest in the classroom, and only 11 with good preliteracy skills had exhibited low interest. Further- more, Morrow’s study provided some evidence for the validity of the results of studies that have relied on parent report data. Questionnaire responses that were also obtained for the sample indicated that 70% of the children who were observed to show high literacy interest in the class- room were reported to look at books daily when at home, but only 2% of the children with low classroom interest reportedly did so. Studies of “precocious” readers (children who learn to read at an es- pecially early age) provide additional evidence for an association between carly attitudes and reading acquisition. These studies, all based on ma- ternal interviews, have indicated that interest in literacy and motivation to learn were important factors that distinguished these unusual children from other, equally bright preschoolers. In an early study, Durkin’s (1966) analyses of interviews with mothers of 30 precocious and 30 nonpreco- cious beginning first graders revealed that more of the former were re- READING TO PRESCHOOLERS. 253 ported to have shown preschool interest in literacy (100% vs 739%) and that the greater amounts of actual instruction provided to the precocious readers by their parents was said to have been prompted by the child’s desires rather than by the parents" goals. Durkin’s conclusion that “'the very special importance of the child’s own interest needs to be highlighted” (p. 135) in seeking to understand precocity receives support from several other sources. For instance, Thomas (1984) compared 28 4-year-old readers and 28 nonreaders matched for age, sex, socioeconomic status, and IQ. These children’s mothers were interviewed about the children’s activity preferences at ages 2, 3, and 4 years. At all ages, the precocious readers were more often described as preferring literacy-related toys (e.g., books, letter sets, workbooks) than were the nonreaders, who were reported to be most interested in gross motor activities (¢.g., ball games, pull toys, swings) at the younger ages and in fantasy toys (e.g... dolls, puppets, cars and trucks) at older ages. Very similar differences were obtained by Manning and Manning (1984) between small samples of precocious and nonprecocious readers from low SES backgrounds. Motivational factors have also been highlighted in some descriptive studies of precocious readers that have not included control groups (e.g., Anbar, 1986; Clark, 1976; Plessas & Oakes, 1964; Price, 1976; Jackson, Donaldson & Cleland, 1988). Finally, given these indications that children who become precocious readers are characterized by especially strong early interest in literacy, we hypothesized that children who become poor readers might show the opposite inclination (Scarborough et al., 1991). Parental interview re- sponses collected during the preschool years were compared for children who, when followed up at the end of Grade 2, had become disabled readers (N = 22) or normal readers (N = 34), despite similar SES and IQ levels of the groups. Maternal reports at ages 36, 42, and 48 months regarding how often the children amused themselves alone with books revealed substantial and consistent differences. On average, children who became better readers reportedly engaged in solitary book activity almost daily, while those who developed reading disabilities reportedly did so only about twice per week. In sum, in addition to oral language abilities and emergent literacy skills, predictors of literacy achievement also include socioeconomic background and early attitudes toward literacy. We now turn our atten- tion to the predictor variable that is the main focus of this review, namely parent-preschooler reading experiences. Relevant Aspects of Parent-Preschooler Reading: Predictor Variables Although ethnographic and diary studies have shown that parents ex- pose their preschool children to literacy in a wide variety of ways during 254 SCARBOROUGH AND DOBRICH household activities (e.g., Anderson & Stokes, 1984; Bisex, 1980; Heath, 1983; Scollon & Scollon, 1981; Snow & Ninio, 1986; Taylor, 1983), the specific situation of parent-preschooler bookreading has been the pri- mary focus of most theory and research.' Before examining the evidence for the efficacy of shared bookreading, we will briefly describe how it has been measured by researchers and what has been described about the frequency and nature of this activity in the preschool years. Parental reports have most often been the basis for estimating the amount of shared bookreading that preschoolers experience. In different studies, these reports have been collected before (prospectively), after (retrospectively), or concurrently with the assessment of outcomes. In- terview or questionnaire responses by parents of middle-class or mixed- SES samples of 2- to 5-year-old British and American children suggest that the typical mainstream preschooler is probably read to on a regular basis: 43 to 75% of subjects in different samples have reportedly been read to daily or more often; sample means of 4.5 to 10.5 times per week have been obtained; and from about 1 to 14 h per week have been said to be spent on shared reading (DeBaryshe, Huntley, Daly, & Rodarmel, 1992a; Dunn, 1981; Mason, 1980; Mason & McCormick, 1981; Mason et al., 1992; Raz & Bryant, 1990; Romatowski & Trepanier, 1977; Scarbor- ough et al., 1991; Thomas, 1984; Walker & Kuerbitz, 1979; Wells, 1985a; Whitehurst et al., 1988). Most, but not all, available research supports the hypothesis that children from less-advantaged backgrounds are read to less regularly by their parents (Anderson & Stokes, 1984; Feitelson & Goldstein, 1986; Harris & Smith, 1987; Mason & McCormick, 1981; Miller, 1969; Raz & Bryant, 1990; Romatowski & Trepanier, 1977; Swin- son, 1985; Wells, 1985b). All together, this body of research suggests that there are sizable differences between demographically defined groups, between samples of equivalent SES, and among individual subjects within those samples. The frequency of shared reading is only one facet of home literacy experiences, however. Consequently, instead of analyzing responses to a single item about parent-preschooler reading, some researchers have in- stead chosen to derive a composite score by summing or factor analyzing responses to many items pertaining to literacy-related activities in the home. To the extent that parents who read frequently to their children are " 1s beyond the scope of this review to evaluate the related bodies of research on the effects of shared home literacy experiences during the school years (e.g., Meyer & Linn 1990; Rowe, 1991; Tizard, Schofield, & Hewison, 1982; Topping & Wolfendale, 1985) and fon the effectiveness of reading aloud to children in school settings (e.g. Karweit, 1989; Mason, Kerr, Sinha, & McCormick, 1990; Morrow, 1992; Valdez-Menchaca & Whitehurst 1992) READING TO PRESCHOOLERS 255 also likely to read more themselves, have more books (and children’s books) in the home, take their young children to the library, and so on, using a composite index of home literacy, may yield more reliable and meaningful estimates of the efficacy of parental practices. As will be seen, however, composite scores have sometimes been based on such a diver- sity of items that their interpretation is clouded. ‘A major drawback to using parental reports is, of course, the possibility that social desirability biases may lead parents unwittingly to inflate their estimates according to what they believe a socially acceptable answer would be. The fact that sizable proportions of parents, even in very recent and middle-class samples, have reported reading to their children less than daily, however, suggests that such biases may not be too severe a problem. To our knowledge, however, no direct investigations have been done to assess the reliability of such reported frequencies. Instead of parent reports, naturalistic observations of parent— preschooler reading have been obtained in some of the studies that follow. As will be seen, these studies have mainly focussed on the qualitative features, rather than frequency, of parent-child bookreading. How par- ents read to their children varies considerably, and in addition to individ- ual differences there are some reported socioeconomic and age differ- ences in the nature of shared bookreading (e.g., DeBaryshe et al., 1991; Heath, 1982, 1983; Hoff-Ginsberg, 1989; Ninio, 1980; Pellegrini, Brody, & Sigel, 1985). A key challenge, of course, is selecting the most pertinent facets of adult behavior to measure. Also, although social desirability biases are more worrisome with regard to parental reports, there is nev- ertheless the possibility that mothers who know that they are being ob- served may alter their usual manner during bookreading, so that it ad- heres more closely to their notions of the most acceptable way of inter- acting with a preschooler. Reliability of observational data can also be difficult to achieve, and, because some studies have relied on only one or two observations of bookreading per dyad or have recorded these ses- sions in the laboratory rather than the home, it cannot necessarily be presumed that representative samples of behavior have been observed. Nevertheless, given the unconfirmed veracity of parental reports, obser- vational studies can be an important source of information about the consequences of reading to young children. Last, some other investigators have used intervention designs, in which the frequency or quality of bookreading experiences has been modified for experimental groups that participate in a research program, and in which outcomes relative to a control group are obtained. Particularly when the adherence of parents to the treatment protocol is monitored, this approach is most likely to provide an unambiguous assessment of, parent-preschooler reading, It is also the only approach that is not simply 256 SCARBOROUGH AND DOBRICH correlational but instead can provide evidence for causality. The chal- lenge remains, of course, to decide which aspects of parental behavior are most relevant and modifiable with regard to fostering language and liter- acy development. Also, when the nature of the intervention is such that few parents ordinarily engage in it, the effects observed may only indi- rectly pertain to the question of whether reading to preschoolers has beneficial effects in normal circumstances RESEARCH ON THE EFFICACY OF READING TO PRESCHOOLERS IN THE HOME: 1960-1993 It should be clear from the foregoing discussion that there is a funda- mental problem to be faced when trying to evaluate the contributions of parent-preschooler bookreading experiences to language and literacy de- velopment. That is, the young child’s language abilities, acquired knowl- edge about books and reading, and preschool interest in literacy are all associated with differences in later achievement. From any observed cor- relation between parent-child reading and one of these other variables, therefore, it is difficult to determine the direction of effect. That is, does reading to preschoolers promote child interest, language, and literacy knowledge? Or, do children with high interest and stronger skills elicit more or better reading by parents? Or, are there ongoing reciprocal in- teractions among all of the predictors during the preschool years? Be- cause most of the studies to be reviewed are correlational in nature, those results do not permit such questions to be answered but can only tell us about whether there is an association between parent-preschooler read- ing and various outcomes. The initial goal to be addressed in this review, therefore, is to determine whether there is indeed a relation between parent-preschooler reading and desired outcomes and to estimate its re- liability and magnitude based on the available evidence To address this issue, the most pertinent findings are necessarily those from studies in which measures were made both of parent-preschooler reading and of language or literacy skill in the preschool years or later. To the best of our knowledge, every published study (article, book, chapter, or technical report) since 1960 that meets these criteria has been included in the review that follows. We will also discuss several new studies that have been presented at national conferences. We have excluded no such work that we know of, but it is possible that other recent unpublished results pertaining to the issues have not come to our attention. Studies with similar variables and methodology will be presented to- gether, beginning with examinations of the associations involving the fre- quency of parent-preschooler reading—first with literacy achievement in the primary grades, then with emergent literacy skills prior to school entry, and then with preschool language abilities. We will then discuss the READING TO PRESCHOOLERS 287 relatively few investigations that have looked for associations between the quality of shared reading and preschool skills and will conclude with a review of intervention research pertaining to the efficacy of parent~ preschooler reading. To facilitate the comparison of results across studies and paradigms, we have taken several steps. First, the research discussed in each of the next five sections of this review has been summarized in Tables | through 5, respectively. For the reader's convenience, the studies in each table are isted in the order that they are introduced in the text. Second, when the investigator has not done so, and when possible from the descriptive information provided, we have conducted significance tests that we deem to be appropriate. Third, because most analyses have been correlational but some have looked at group differences, for the latter we have pro- vided in the tables the correlational equivalents (point biserial, rank bi- serial, or phi coefficients) as estimates of effect sizes, whenever possible. Fourth, like most investigators, we have applied a two-tailed (.05) signif- icance level to all findings; thus, for the sake of uniformity, in the few instances in which published results have been based on one-tailed infer- ential tests, we will instead describe them in terms of the two-tailed prob- abilities. Associations between the Frequency of Parent-Preschooler Reading and Literacy Achievement during the School Years (Table 1) ‘As summarized in Table I, several studies have looked directly at dif- ferences in primary grade literacy achievement in relation to the amount of parent-preschooler bookreading that the children participated in. We will first describe the investigations in which the predictor variable was based on parental reports, including those that have looked specifically at the frequency of shared reading and those in which a broader composite measure of home literacy experiences was derived. This section will con- clude with a description of results involving actual observations of par- ent-preschooler reading frequency One early investigation, which is often cited as providing evidence for the efficacy of reading to preschoolers, compared the preschool experi- ences of precocious and nonprecocious children. Durkin (1966) tested 4465 beginning first graders from 40 New York City public schools. Of the 156 children identified as already able to read, 30 were randomly selected for comparison with 30 1Q-matched controls. Maternal interviews re- vealed that 100% of the precocious readers but only 73% of the controls were read to at home during the preceding year, (x1) = 5.7, p < .02). While demonstrating the hypothesized relation between parent-pre- schooler reading and literacy achievement, it should be taken into ac- count when interpreting these results that the precocious readers also 258 SCARBOROUGH AND DOBRICH TABLE | TESTS OF THE HYPOTHESIZED ASSOCIATION BETWEEN THE FREQUENCY OF PARENT-PRESCHOOLER READING AND LITERACY ACHIEVEMENT OUTCOMES DURING THE Scoot YEARS Outcome Study agearade Desien™ Nowe fects Durkin (966) BeeGi RIK) @ RF A= R Walker Kuerbitz End GI RipreK) MOORE ise = 28) (979) End G3 2S. Gey = —.09) Share etal 0984) End KK PYbeg K) ss RR rae End G1 B) 12 3 se text or ‘alivante esate Scarborough etal. End G22 years)—father 56" RF y= issn) Pave)~moiher Bs = 20 GS years)—mather Bs (re = 00) POV) ~motber nse = 18) PC years)-mother Bs. (re = 9) Miler (1965) EodK oC St Cl Confotnded see text) End G1 Read K) Minois Project, Mason & Dunning End K —Ptbeg.K) tw ORF oy (1386) Beg Gi G Ent GL Mason et al End KPlbeg. KY 127 Ch Motivate a (1982) Bee Gi tie only (se tex) End Gl 109 End G2 * End G3 5 Bristol Project ‘Wells (19886) Tyears——PUUS-A2 months) = «32OF = 40 P3960 months) = «20 OF = 6 Moon & Wells 999) Tycars F960 months) 20 CL = 57, 68, 69, 69 = R retrospective tothe agelgrade listed; C. concurrent measurement of predictors and outcomes: P, prospective from the age/rade listed Independent variable: RF. reported frequency: OF, observed frequency; CI, composite index. "Comparison of precocious and nonprecocious readers 4'Comparion of eutcomes for groupe iring in shared reading experience “Comparison of groups who became good and poor Fears. received more explicit home instruction in literacy skills, had better- educated mothers (33% vs 10% college graduates), and, as noted earlier, were reported to have shown more early interest in literacy (also 100% vs 73%). Consequently, the findings do not permit the relative roles of home instruction, SES, child motivation, and parent-child reading to be disen- tangled, Walker and Kuerbitz (1979) also interviewed the parents of first graders about their children’s joint bookreading experiences prior to school entry. ‘Twenty children who had been read to daily or more often as preschoolers were compared to 16 other children with respect to reading achievement in first and third grade. Slight and nonsignificant differences were found READING TO PRESCHOOLERS 259 between the groups’ mean reading scores (expressed as grade equiva- ents) in Grade 1 (2.7 vs 2.3, respectively: #(34) = 1.72, p = .09) and in Grade 3 (4.8 vs 5.1; 434) = 0.69). These results thus provide little evidence in support of the hypothesis under review, Share et al. (1984) used a battery of 39 measures at the time of entry to kindergarten to predict reading achievement of a large sample of Austra- lian children at the end of kindergarten (N = 525) and first grade (N 479). Of the 19 cognitive, linguistic, and preliteracy skills assessed at the outset, 14 were well correlated with outcomes (r = .30 to .68, median r = 43), Whether parents read to their children was not a strong predictor in its own right at either grade (r = .26 and .21, respectively) and, in multiple regression analyses of Grade 1 outcomes, accounted for no additional variance beyond the 58% accounted for by a set of five variables reflect- ing skill and gender differences. SES, as indexed by the father’s occupa- tional level, was somewhat better correlated with outcomes (r = .37 and 30) but also made no unique contribution in the multivariate analyses. These results suggest that parent-preschooler reading experience makes a weak indirect contribution to developing literacy, Another prospective study was our comparison, mentioned earlier, be- tween 22 children who became disabled readers and 34 who did not (Scar- borough et al., 1991). Distributions of SES and preschool 1Q were similar for these groups. A question about the frequency of shared reading was included in interviews with the children’s fathers when the children were 2 years old, and in maternal questionnaires at ages 214, 3, 3%, and 4 years. Significantly less-frequent shared bookreading was reported for the chil- dren who became poor readers only for father-child reading (36% vs 71% read to more than once per week). Less reliable differences were seen in the proportions read to daily or more often by their mothers: 50% vs 71% al age 21; 55% vs 53% at age 3; 45% vs 61% at age 3¥; and 67% vs 80% at age 4; all p > .10. In contrast, early language abilities, emergent literacy skills, and reported interest in solitary book activities of these groups were more strongly and consistently predictive of outcome reading status, (Scarborough, 1990, 1991; Scarborough, Dobrich, & Hager, 1990).? ‘As noted earlier, a few investigators have used composite scores, rather than just responses to single items, as predictor variables. Miller (1969) interviewed mothers of 19 middle class, 19 upper-lower class, and 17 tower-lower class kindergartners about “*home prereading activities,”” including parent-child reading, various other preschool activities (arts } The two groups who became good readers were collapsed in computing the coeMicients in Table 1. Also, in earlier analyses of a sample that included many of these subjects, nearly equivalent amounts of parent-child reading (summed over the entire preschool period) were ound for children who became good and poor readers (Scarborough, 1989) 260 SCARBOROUGH AND DOBRICH and crafts, family trips, library visits, dramatic play), and, unfortunately, mastery of several literacy-related skills such as rhyming, letter knowl- edge, and visual discrimination of forms and letters. In view of this, it is pethaps not surprising that the composite score derived from interview responses was correlated with concurrent performance on a test of read- ing readiness. When reading achievement was tested 1 year later (in first grade), an association between the composite scores and reading test scores was obtained only for the middie class sample (although it appears to us that if the samples had been combined in the analyses, there would have been a moderate overall correlation). Given the way that the pre- dictor variable was defined, however, the results of this study are entirely uninformative regarding the efficacy of parent-preschooler shared read- ing or of broader differences in parental practices. Both a composite index and responses to single items were analyzed for a large-scale longitudinal study directed by Jana Mason, which we will henceforth refer to as the Illinois project (Mason, 1992; Mason & Dun- ning, 1986; Mason et al., 1992). A 59-item questionnaire was given to parents of a sample of 127 beginning kindergartners, whose initial skill levels and subsequent achievements at the end of kindergarten, beginning and end of Grade 1, and end of Grades 2 and 3 were also measured. Ina preliminary report of results through the end of first grade, simple corre- lations were reported between outcome measures and parental responses to relevant questions (Mason & Dunning, 1986, also described in Mason, 1992). As summarized in Table 1, how regularly the parents read to their children (a 5-point scale) was correlated (r = .25 to .37) with all but one of the five outcome measures taken between the end of kindergarten and the end of Grade 1 In more recent analyses of the data through Grade 3, a composite “reading to children’ score was derived from a factor analysis (which also yielded “child interest,”” “home problems,” and “SES” factors) Items that loaded most highly (.72 to .81) on this “reading to children” factor were whether (and if so, for how long) the child was read to on the previous day and how often the child was read to at home. However, whether the child asked to be read to, whether oral storytelling was a regular household activity, how often the child looked at books, and how often the child tried to read to his/her parents also had loadings of .51 to 63. Thus, while more heavily weighted toward parent-child reading, this predictor variable also reflects some variation in the children’s perceived literacy-related skills and interests. Even so, this factor score accounted for no additional variance in multiple regression analyses, in which the largest proportions of variance were accounted for by differences in chil- dren's preliteracy and language comprehension skills at the start of kin- dergarten or by literacy skill levels in prior grades. Hence, despite the use READING TO PRESCHOOLERS 261 of a broader composite index, the results were similar to those already reviewed, in suggesting that in comparison to existing individual differ- ences in children’s abilities when they begin school, parent-preschooler reading is a relatively weaker, or less direct, predictor of subsequent achievement, Finally, in the Bristol project mentioned earlier, both interview data and actual observations of the frequency of parent-preschooler reading in the home served as the predictor in several analyses. For a representative subsample of 32 children from that project for whom later reading scores were available, Wells (1985b) examined recordings that had been made in the children’s homes between the ages of 15 and 42 months to determine the frequency with which parents read stories to their preschoolers during the day. He found that reading comprehension at age 7 was moderately correlated (r, = 40) with the observed frequencies of parent-preschooler story reading.* Reading scores were predicted as well or better by other preschool differences—in oral language (r = .64), emergent literacy (r = .82), early interest in literacy (r = .63), and SES (r = .52}—in the sample, and in multiple regression analyses no additional variance was accounted for by parent-preschooler reading frequency beyond the 69 to 83% that was accounted for by more powerful predictors (Wells et al., 1984). These findings, based on an observation-derived rather than interview-derived ‘measure of parent-preschooler reading frequency, thus concur with those of other studies in suggesting that the contribution of shared reading experiences to literacy achievement in the school years is relatively mod- est and only indirectly reflected in the contributions of other factors. Wells (1985b) also reanalyzed in a similar manner the smaller (N = 20) subsample that had previously been studied by Moon and Wells (1979). In these analyses, however, home recordings made at older ages (39 to 60 months) were used to derive the observational estimates of story reading frequency, and the association between observed frequency and later reading comprehension was even stronger (r, = .66). Interestingly, this is quite similar in magnitude to the correlations of .57 to .69 (with four different reading scores at age 7) that Moon and Wells (1979) had obtained for the same sample using an interview-based composite reflecting “par- ent interest in, and provision of, literacy” in the home. In light of this, the reason that generally stronger correlations were obtained in analyses of the Bristol project does not seem to have been solely that an observa- tional measure was used. ‘Summary. Seven research samples (median N = 56) have been exam- ined with regard to the association between parent-preschooler reading, ® This Glass rank biserial correlation and those that follow were converted from Mann- Whitney U statistics in the published reports, 262 SCARBOROUGH AND DOBRICH frequency (alone or in conjunction with other indices of the home envi- ronment) and literacy achievement during the early school years. Bivati ate and between-group analyses of those samples have yielded 21 esti- mates of the strength of the hypothesized relation. Of these, 14 (679) were statistically significant at the level adopted. Median coefficients were .24 and .28, respectively, for the two large and socioeconomically diverse samples (Mason & Dunning, 1986; Share et al., 1984), and .32, 08, .17, and .67 for the four smaller samples; the overall (unweighted) median of the 21 coefficients was .28, and the median of the six sample medians was also .28. These results suggest that there is indeed an asso- ciation between literacy outcomes and reading to preschoolers but that it probably accounts outright for no more than about 8% of the overall variance in achievement. Additional analyses of several samples suggest further that the relative contribution of parent-preschooler reading to prediction is somewhat weaker than those of other factors. Without exception, stronger (but not necessarily statistically different) coefficients were obtained for: demo- graphic indices of SES (Share et al., 1984; Wells, 1985a); early interest in literacy (Mason & Dunning, 1986; Scarborough et al., 1991); and pre- school language and literacy abilities (Mason et al., 1992; Scarborough, 1989, 1991; Scarborough et al., 1991; Share et al., 1984). Consequently, when these other differences have been taken into account, differences in how often parents read to children have made only negligible unique contributions to prediction. The results suggest that when the dependent variable is a measure of literacy achievement during the early school years, the effects of shared reading on such outcomes are subsumed within, or mediated by, the effects of predictors with stronger and more direct associations with achievement. The modest average degree of association between parent-preschooler reading and literacy achievement in this group of studies must be inter- preted with caution, however, because the measurements of shared read- ing frequency have been less than ideal. Dichotomous or limited ordinal scaling of reading frequency, for instance, could have failed to capture the full variability in several samples (Durkin, 1966; Mason & Dunning, 1986; Scarborough et al., 1991; Share et al., 1984; Walker & Kuerbitz, 1979), and questions can be raised about the way composite indices of home literacy were derived in others (Mason et al., 1992; Moon & Wells, 1979), Also, observational data from the Bristol project were collected during daytime hours only, so that bedtime story experiences were not reflected in the measure of shared reading. Itis difficult, without additional data, to judge how the use of more sensitive measures might have affected the results of the studies reviewed in this section. Another possible explanation for the modest size of the observed ef- READING TO PRESCHOOLERS 263 fects is that instructional differences once children begin school are strong enough to obscure the influences of home factors such as parent~ preschooler reading experiences. If so, a more appropriate hypothesis to investigate is that the influence of shared reading on literacy development occurs primarily over the shorter term and only indirectly affects later achievement by promoting growth in emergent literacy skills. Results pertaining to this hypothesis are reviewed next. Associations between the Frequency of Parent-Preschooler Reading and Literacy-Related Skills Prior to School Instruction (Table 2) As mentioned earlier, although few children can actually read or write before entering kindergarten, most have begun to develop some “emer- gent literacy” (or “‘preliteracy"’ or “readiness") skills. Differences in such skills among children around the time that they begin school served as the outcome of interest in the research to be reviewed in this section. TABLE? ‘Tests oF THE HYPOTHESIZED ASSOCIATION BETWEEN FREQUENCY OF PARENT-PRESCHOOLER READING AND LITERACY-RELATED SKILLS PRIOR TO Sciool Instaucrion Outcome Study age/grade Devien? Now ters linois Project, Mason & Dunning (1986) Bek C RF r= a8, 21 Mason ct al BeK OC 1 CL Mativanate analyses (19), only (se text) Bristol Project ‘Wells 19886) Syears—— PUSA2 months) «32 OF a PL39-60 months) 20 OF 6 Moon & Wells a9) Syears RSG months) «20 CL r= 77 Wells (984) Syears—RG9-6 months) RF ° Crain-Thoreson & —4¥4 years? years) BORE rw 38s. = 28, Dale (1992) nS. (7 2" 09) see text for mulivar ste results Mason (1980) Bee peK —C MORE asa) Eee reefer aste= dD) ¢ ns.(r = 0} una (1981) Syeus € RF AS trend (e = 26) Briegs & Elkind 97) Bek C CLs as. ‘Thomas (1988) ayes Comother Se ORE ons Cather ne =. retrospective to the age/grade listed: C, concurrent measurement of predictors and outcomes: P prospective from the agsigrade listed "Independent Varabe: RF, reported frequency; OF, abserved frequency: Cl, composite index. « Comparison of precacious and nonprecocious Feades 264 SCARBOROUGH AND DOBRICH We will first discuss the results for the two samples in which both emer- gent and later literacy skills were assessed (the Illinois and Bristol pro- jects) and then introduce studies in which no longer-term outcome mea- sures were taken. In the Illinois project (Mason, 1992; Mason & Dunning, 1986; Mason et al., 1992), rudimentary decoding skills and labelling of environmental print were the two preliteracy skills assessed at the start of kindergarten, and each was related (r = .39 and .21, respectively) to how often the children were read to. In a multiple regression that accounted altogether for 44% of the variance in early reading skill (a combined score), the “reading to children’” factor accounted for some unique variance, al- though it made a weaker contribution (B = .16) than did three other predictors (SES, 8 = .46; child interest, 8 = .39; and home problems, 8 = —.18), In comparison to the findings at later school grades (Table 1), these results are thus consistent with the hypothesis that home influences may have somewhat stronger effects over the shorter term, although those effects are still relatively modest. In the Bristol project, literacy knowledge scores (reflecting letter skills and familiarity with books and print) at age 5 were correlated with: the reported frequency with which parents read stories to their preschool children from age 3% to 5 years (r = .49; Wells et al., 1984); the frequency estimates derived from naturalistic observations from ages 15 to 42 months for the same sample of 32 children (r, = .42, Wells, 1985b); the interview-based index of parental provision of literacy in the smaller sub- sample (r = .77; Moon & Wells, 1979); and the observational measure of story reading from age 39 to 60 months in the reanalysis of that group of 20 children (7, = .61; Wells, 1985b). These values are all quite similar to the corresponding coefficients that were obtained with reading scores at age 7 as the outcome measure (see Table 1), In the more representative subsample of 32 children, emergent literacy skills were also strongly related to demographic SES differences (r = 52), to oral language ability (r = .57), and to preschool interest in literacy activities (r = .56). Nevertheless, a composite “parent literacy influ- ence” variable (reflecting the range and amount of parental writing in addition to the frequency of reading stories aloud) was found to account for some unique variance in outcomes other than the proportion predicted by language skill and literacy interest (Wells et al., 1984). (SES was not included among the variables entered into that analysis.) Hence, these results, like those for the Illinois project, suggest that there is indeed an association between parent-preschooler reading experiences and early literacy development and that it may be somewhat stronger than the effects for longer term outcomes. ‘Analyses of the relations between the reported frequency of parent~ READING TO PRESCHOOLERS 265 child story reading at age 24 months and several emergent literacy indices at age 4% years were recently conducted by Crain-Thoreson and Dale (1992), who studied a longitudinal sample of 25 linguistically precocious preschoolers (i.e., children whose oral language levels at age 2 years were exceptionally advanced). Reported frequencies of shared reading were correlated with letter knowledge (r = .38) but less strongly with under- standing of print concepts (r = .28) and a measure of “‘invented spelling” facility (r = —.09). Similarly, in multiple regressions that controlled for initial differences among the subjects in mean utterance length and for differences in the amount of reading instruction received (at home or in preschool), reading frequency accounted for an additional 14% of the variance in letter knowledge scores and a marginal 9% (p < .10) of addi- tional variance in knowledge of print concepts. Performance on the spell- ing measure was associated only with differences in preschool reading instruction. The results for this unusual sample, therefore, are quite con- sistent with those of the Illinois and Bristol projects. In contrast, several other investigations have not yielded as clear evi- dence for an association between emergent literacy skills and parent— preschooler reading. Mason (1980) collected parental questionnaires at the beginning and end of the school year from parents of 38 middle- to upper-middle-class 4-year-olds enrolled in a preschool program. One item. concerned whether the child was read to at home for less than half an hour, about I h, or more than 2 h per week. Responses to this item were completely unrelated to which of four levels of early reading ability the children had exhibited at the start of the year (r = .07) and at year's end (r = .02 with the first set of parental reports and .11 with the second). Although nonparametric analyses would probably have been more appro- priate given the ordinal scaling of both variables, it is unlikely that a change in method would have yielded substantially stronger results. There remains the possibility, of course, that the questionnaire item's insensitivity to the full variability in parent-preschooler reading frequen- cies is responsible for the null findings in this study. Dunn (1981), however, used a potentially more informative predictor variable in her study of a middle-class sample of 40 5-year-olds who all attended the same half-day nursery program (which did not emphasize academic instruction). Fourteen aspects of the children’s emerging liter- acy skills were rated by their preschool teachers, and the composite scores of prereading ability derived from this checklist ranged from I to 14 (M = 84, SD = 3.2). In addition, the children's mothers completed questionnaires and kept detailed diaries of all parent-child activities that occurred during 5 weekdays. Prereading ability was found to correlate with interview responses regarding mothers’ beliefs about their teaching roles (r = .41), but not as strongly (r = .26, p = .10) with the amount of 266 SCARBOROUGH AND DOBRICH parent-preschooler reading that was recorded in the diaries, which ranged from 0 to 39 min per day and averaged 8 min daily, or about the time it may take to read aloud one or two short children’s books (e.g., Cornell, Senechal, & Broda, 1988; Pellegrini, Brody, & Sigel, 1985). Somewhat similarly, Briggs and Elkind (1977) concluded that parents’ inclinations to push their preschoolers toward literacy, rather than how much they read to their children, was a key difference between the back- grounds of precocious and nonprecocious readers. Interviews were con- ducted with the mothers of two groups of beginning kindergartners: 33 precocious readers and 33 other children of equivalent chronological age and verbal IQ. From the 47-item interviews, five factors were extracted and compared for the two groups. Oddly, the two most pertinent items, concerning the amount of time and the frequency of parent-preschooler reading, loaded on separate factors (that also reflected aspects of cl dren's play preferences, television habits, and interest in learning to read). Neither factor differentiated the precocious readers from controls. Instead, the early readers had higher “socioeconomic status,”” “parent achievement orientation,”” and “family interest in language" factor scores, leading the authors to conclude that SES-linked differences in parental encouragement of literacy learning were the main force behind the precocious acquisition of reading. It is unfortunate that data for indi- vidual interview items were not provided separately, as this would have permitted a clearer picture as to whether differences in parent- preschooler reading were indeed entirely absent ‘Thomas (1984) also found no differences in the reported amounts of parent-preschooler shared reading in her comparison of 28 4-year-old readers with 28 nonreaders matched for age, sex, SES, and 1Q. Mothers and fathers of each group reported that, on average, they spent about I h per day reading to their children. No ranges or standard deviations were provided, so it is possible that there was little variance within each group for the predictor. Nevertheless, given that these groups differed so mark- edly in their early activity and toy preferences (as discussed earlier), the results suggest that becoming a precocious reader may be more strongly related to children's own interests than to parental practices such as shared reading, Summary. The association between parent-preschooler reading fre- quency and literacy-related skills prior to the start of schooling has been examined in eight research samples (median N = 39). Of the 17 bivariate and between-group tests of the hypothesis that were conducted, 7 (41%) met the significance criterion, From the limited information available, we estimate that the median coefficient across these 17 tests was between .21 and .26 and that the median of the eight sample medians was .27. In the Bristol and Illinois projects, which were the two studies in which achieve- READING TO PRESCHOOLERS 267 ment was also assessed during the school years, roughly equivalent ef- fects were obtained for early and later literacy outcomes. Contrary to expectation, therefore, the research reviewed in this section suggests that ‘emergent literacy skills are probably not much more strongly associated with parent-preschooler reading than are later achievement scores. As jowever, such conclusions must be made cautiously in light of the insensitivity of the measures used in some of these inves Although all of these studies looked at early literacy abilities, several different aspects of literacy-related skill were measured. A few assess- ments reflected the more modern emphasis on evaluating children’s fa- miliarity with the concepts, conventions, and purposes of written lan- guage, whereas others adhered more to the traditional approach of mea- suring basic skill components. The results from this small set of studies do not indicate clearly whether the kinds of knowledge tapped by the newer measures might be more strongly fostered by parent-preschooler shared reading, as one might suppose. For example, Mason and Dunning (1986) found that the child's ability to recognize environmental print (e.g., on signs and product labels) was not as well correlated with parent— preschooler reading as decoding ability was at the start of kindergarten, but the opposite pattern was seen at the end of the year (see Table 1). Crain-Thoreson and Dale (1992) also found somewhat larger effects for a traditional measure of letter skills than for more modern measures of print, concept knowledge and invented spelling performance, but weak effects were found in other studies that relied on skill-oriented outcome mea- sures. Finally, both skills and concepts were included in the composite literacy knowledge scores used in the Bristol project, in which the largest correlations were obtained. In additional analyses that were conducted for some samples, emergent literacy outcomes were not more strongly related to shared reading in the home than to demographic SES indices (Mason et al., 1992; Wells, 1985b) and, except in one small sample (Crain-Thoreson & Dale, 1992), to early child interest in literacy (Mason & Dunning, 1986; Thomas, 1984; Wells, 1985b). In short, the research in this section, taken all together, is con- sistent with the work in the preceding section in suggesting that there is an association between reading to preschoolers and emergent literacy skills, but that this relation is not a particularly strong one in comparison to the contributions of other predictors. Associations between the Frequency of Parent-Preschooler Reading and Preschool Language Abilities (Table 3) Another reasonable hypothesis (e.g., Tough, 1983) is that the benefits of parent-preschooler reading pertain primarily to language abilities be- TABLE 3 ‘Tests oF THE HYPOTHESIZED ASSOCIATION BETWEEN THE FREQUENCY OF PARENT-PRESCHOOLER READING AND ORAL LANGUAGE SKILLS Outcome ‘Language Study agefgrade Now measures Brisol Project ‘Wells ta. (1984) 5 years ROS years) ORF pvr ns. tend = 33) Comprehension ina. tend (r= 33), Teacher rating is. tend ( = 33) Wiens (19853) ™ c 8 RF_——_—Production composite roe insis Projet ‘Mason & Dunning (1986) Beg.k c 1 RE Comprehension composite r= 39 Urterance length Bs. ir = 13) Mason etal. (192) Begik c moa Comprehension composite Multivariate analyses only (see tex) Dona (1981) 5 years c RF ope, 1s. trend (7 = 26) Comprehension rating asin at) Production rating as te= 19 CCrain-Thoreson & Dale (1992) 2 years c 2 RF PVT, ns = 5) ‘Uterance length 10) m PO years) PevT 3) Syntax comprehension “% PO years) Pevt =.) Syntax comprehension 03) Phon, awareness a8 (e= Bens. (= 36) See text for milivarat results DeBaryshe (1993) 2 years c 41 RF_——_RDLS receptive P= 30,08, = 08) DLS expressive 03,08." = ~.13) DeBaryshe etal. (1991) Dhyers € = oa pevr, EOWPVT TTPA.VE, Detaryshe eta. (1928) 2Dsyean€ % ot Semantic composite soa Semantic componite *R, retrospective to the age listed; C, concurrent measurement of predictors and outcomes; P, prospective from the age listed » Independent Variable: RF, reported frequency; Cl, composite index. 997 HoRidod anv HONOWoRAVaS READING TO PRESCHOOLERS 269 cause interactions between adults and children during shared story read- ing rarely focus on print or the reading process itself, but rather on the verbal labels for pictured objects and events, and on the semantic content of the stories. This may explain why relatively modest effects have been obtained in most of the studies that have measured literacy-specific knowledge, since those effects would only indirectly reflect influences of shared reading experiences on oral language skills, which in turn contrib- ute to literacy development. Research in which oral language abilities have served as outcome variables for evaluating the efficacy of frequent parent-child reading will be reviewed in this section, ‘The studies that follow differ in several respects, as can be seen in Table 3. First, about half of the samples have included children from 2 to 4 years of age, while the others contained older preschoolers. Second, the outcome variables have been quite diverse. Some investiga tors have looked specifically at measures of language content (word meanings and semantic comprehension) and/or language structure (syn- tactic and phonological skills), while others have relied on broad com- posite indices of language proficiency. In most studies, both receptive and expressive language measures have been included. For the reader's convenience, we will again begin by describing some further results of studies that were introduced in the two previous sections and will then review several investigations in which only language outcomes have been examined. First, contrary to the hypothesis under consideration, parent-reported frequencies of reading to children were not more strongly related to lan- guage outcomes than to developing literacy skills in the Bristol project. For the subsample of 32 children that was studied most intensively (Wells, et al., 1984), marginally significant correlations of .33 were obtained with each of three language indices at age 5: the English Picture Vocabulary Test (EPVT), which requires children to indicate which of several pic- tured stimuli has been named by the examiner; an oral comprehension test; and teacher ratings of the children’s overall language abilities. Sim- ilarly, for the entire sample of 128 children, a correlation of .36 was obtained between the frequency of parent-preschooler reading and a combined scale of syntactic and semantic production that was derived from natural language samples recorded in the children’s homes (Wells, 1985a). These coefficients clearly do not exceed those obtained when reading achievement and emergent literacy outcomes were examined for this sample (Tables | and 2). The hypothesis was also not strongly supported by data from the Illi- nois project (Mason, 1992; Mason & Dunning, 1986; Mason et al., 1992) and from Dunn's (1981) study. Mason and Dunning (1986) found that the regularity of parent-preschooler reading was not correlated with chil- 270 SCARBOROUGH AND DOBRICH dren's mean utterance lengths at the start of kindergarten (r = .13) but was related to a composite semantic comprehension measure (r = .39) at that age. In multiple regression analyses of the latter outcome measure (Mason et al., 1992), the contribution of the ‘reading to preschoolers” composite (which accounted for some unique variance in emergent liter- acy scores, as discussed earlier) approached significance. Similarly, in her sample of 40 5-year-olds, Dunn found that amount of time spent on shared reading (as recorded in the mothers” diaries) was not better cor- related with Peabody Picture Vocabulary Test (PPVT) scores (r = .26, p = .10) or to teachers’ ratings of language comprehension (r = .11) and productive fluency (r = —.15) than with emergent literacy ability (Table 2). Even at younger ages, Crain-Thoreson and Dale (1992) found few strong associations between parent-preschooler reading frequencies (re- ported at age 2) and seven language measures that were obtained for the 25 linguistically precocious children in their sample. At age 24 months, neither utterance length nor PPVT performance was related to the re- ported frequency of shared reading (r = .10 and .09, respectively). At age 30 months, in multiple regression analyses that controlled for initial PPVT differences, reported frequencies of shared reading accounted for almost no additional variance in PPVT performance, but did contribute to the prediction of syntax comprehension scores (R* increase = .19). Finally, at age 4¥4, neither receptive vocabulary nor syntax comprehension was related to the amount of parent-child reading, and correlations of shared reading frequency with phonological awareness scores only approached significance (R increase = .13, p < .10) in an analysis controlling for earlier differences in language abilities and amounts of preschool instruc- tion in reading. The simple correlations for all of these relationshi kindly provided by the authors (personal communication), showed a sim- ilar pattern; these are provided in Table 3. Last, mixed results were obtained in a series of studies by Barbara DeBaryshe. First, for a sample of 41 middle and upper-middle class 2-year-olds, scores on the Receptive and Expressive portions of the stan- dardized Reynell Developmental Language Scales (RDLS) served as the outcome variables (DeBaryshe, 1993). Four measures of home literacy experience were derived from maternal interviews: the frequency of shared bookreading sessions, the number of stories read per week, the age at which parents began reading to their children, and the number of visits to the library. Onset age was correlated with both Receptive (r = —.39) and Expressive (r = ~.33) abilities, the number of stories per week was related only to Receptive scores (r = .30), and frequencies of library visits and shared reading sessions were associated with neither language measure, READING TO PRESCHOOLERS m DeBaryshe’s other two studies focussed on preschool development of lexical-semantic content, as reflected in performance on the PPVT, the Expressive One Word Picture Vocabulary Test of picture-labeling skill (EOWPVT), and the Verbal Expression subtest of the Illinois Test of Psycholinguistic Abilities (ITPA-VE), which taps semantic complexity of descriptive speech. In analyses of a sample of 28 lower- to upper-middle class 2¥-year-olds (DeBaryshe et al., 1991), an interview-derived com- posite index of reading exposure (reflecting the frequency of parent- preschooler reading, the number of stories read per week, and the age at which this activity began) was well correlated with scores on the PPVT (r = .43) and the EOWPVT (r = .41) and less strongly with ITPA-VE performance (r = .26). The same reading exposure index and a composite score based on the same three language scores were analyzed for two additional samples of 2- to 5-year-olds: 56 children in a Head Start pro- gram, and 54 low to middle income children enrolled in daycare programs (DeBaryshe, 1992a). In neither sample was the amount of exposure to reading in the home related to children’s language abilities (r = .03 and 22, respectively). Summary. The association between parent-preschooler reading fre- quency and developing language skills has been investigated in nine re- search samples (median N = 41). Bivariate analyses have yielded 26 estimates of the strength of the hypothesized correlation. Only 6 (23%) of. the 26 tests met the established significance criterion. The overall median correlation was .23, and the median of the nine sample medians was .22. These estimates may be misleading, however, because results involving different aspects of language have been averaged. For the 10 associations involving dependent measures of language content (lexical-semantic abil- ities), the overall median coefficient was .26; for the six associations with measures of language structure (syntactic and phonological abilities), the median effect size was .17; and for the 10 analyses involving broader composite measures of language ability, a median correlation of .21 was obtained, While these differences suggest that shared reading may pro- mote more growth in language content than in language structure, such a conclusion may be premature given the relatively few tests that can be compared, particularly within samples. We tentatively conclude, however, that the results for at Jeast some kinds of language abilities are comparable in strength to those for emer- gent literacy and achievement outcomes that were discussed in the pre- vious two sections of this review. Contrary to the hypothesis posed at the beginning of this section, however, it does not appear that reading to preschoolers is more strongly related to oral language development than to the acquisition of print-specific literacy skills. m SCARBOROUGH AND DOBRICH Associations between the Quality of Parent-Preschooler Reading and Language!Literacy Outcomes (Table 4) Although most investigations of the efficacy of parent-preschooler reading have focussed on the frequency of shared reading, it is possible that what is more important for promoting literacy and language devel- ‘opment is the quality of those interactions. In this section, we will review the studies that have sought evidence for such a relationship. To our knowledge, only five correlational studies have made observations of the qualitative nature of parents’ behaviors while reading to their preschool children and have sought to determine which kinds of differences among mothers are associated with children’s concurrent or subsequent abilities. (Similar hypotheses have also been examined in some intervention re- search to be reviewed in the next section.) We will first describe the results pertaining to the prediction of emer- gent literacy skills and then those that involved developing language skills. Because these investigations have looked at from two to nine as- pects of parental behavior and from one to seven outcome measures, and thus conducted a total of from 9 to 36 statistical tests each, the summary of the 85 results in Table 4 does not provide information about which pairs of variables were correlated but only lists the proportion of effects that were significant in each study. Emergent literacy outcomes. In an early investigation of qualitative aspects of shared bookreading, Flood (1977) studied a socioeconomically heterogeneous sample of 36 mothers and their preschool children (age 31 to 4% years). Audiorecordings were made in the dyads’ homes, and the same book was provided to all mothers, who were asked to read to their children as they usually would. Based on the children’s performance on 10 preliteracy tasks, a factor score of prereading ability was derived. Frequencies of five child behaviors were coded from the tapes, and three of these (words spoken, questions answered, and questions asked) cova- ried with each other and with prereading scores, and the first one alone explained the most variance in a stepwise multiple regression in which were also entered nine observational measures of maternal behavior. Al- though no maternal variables were correlated outright with the reading scores, three did make a contribution to prediction in the multivariate analysis: frequency of warm-up questions prior to beginning the story; frequency of evaluative questions after the story, and frequency of pos- itive reinforcement.* The aspects of maternal behavior that were unre- * One-taled significance tests were used, and only the largest of the 14 zero-order coef ficients is known (r = .27). Neither this nor, hence, any of those involving maternal vari- ables met a two-tailed criterion, TABLE 4. ‘TESTS OF THE HYPOTHESIZED ASSOCIATION BETWEEN THE QUALITY OF PARENT-PRESCHOOLER READING AND (A) EMERGENT LITERACY SKILLS ‘AND (B) LANGUAGE ABILITIES No. of Outcome No. of, study N age (years) Design” Ws Dvs ‘A, Emergent Literacy Outcomes Flood (1977) 6 4 c 9 ' Bus & van Uzendoor (1988) 30 SMhand SC 6 5 Dickinson & Tabors (1991) 9 5 PG years) 2 t Peavey 2 1 ‘Watson & Shapiro (1988) 20 4 Paw) 4 7 B, Oral Language Outcomes Watson & Shapiro (1988); Watson (1989) 20 4 Pew) 4 2 Dickinson & Tabors (1991) » 5 POS years) 2 6 2 6 DeBaryshe et al. (1991) 28 ™ 4 3 *C, concurrent measurement of predictors and outcomes; P, prospective from the age listed. BMfects 9 of 9 n..; see text for multivariate results Confounded (see text) 1 of 2 ns. 2of2ns. 25 of Wns. 4of8ns VW of 12 ns. 12 of 12 ns. Hof 12 ns. SUITOOHISTA OL ONIGVAE ez 24 SCARBOROUGH AND DORRICH lated to differences in emergent literacy included frequencies of words spoken, factual questions asked, interpretive questions asked, elabora- tions of children’s answers, correction of the child, and off-task remarks. These results suggest that what parents do before and after reading a book may be particularly important, but this suggestion has not been pursued in subsequent research. More recently, Bus and Van zendoorn (1988) observed mothers of 3¥-year-olds (N = 15) and 5¥2-year-olds (N = 13). Videorecorded ob- servations of joint reading of a picture-story book and an alphabet book were made in a laboratory playroom, and the production rates were com- puted, separately for each book, for three categories of maternal behavior that were coded from the tapes: narration (pointing out or asking about the meanings of book elements); instruction (providing information about print); and discipline (attempts to maintain or restore the child’s attention on the task). Five emergent literacy abilities were assessed, and maternal use of narration was inversely correlated with several of these (r= —.23 to ~.55). Unfortunately, because the two age groups were combined in the correlational analyses, the results are uninterpretable. That is, since mothers of the younger group relied more heavily on narration, and since the younger children earned lower emergent literacy scores, the inverse correlations were probably due primarily to age differences, which were not controlled in the analyses. Dickinson and Tabors (1991) conducted observations in the homes of 39 children from low-income families at ages 3 and 4¥2 years. At each age, the mother was asked to read a familiar book and a novel one (provided by the researchers) to her preschooler, and the number of “‘nonimmedi- ate” utterances (such as relating events in the book to the child’s expe- rience or asking the child to think about how a character in the book feels) that the mother produced was computed for each book. Such ““decontex- tualized” language, in which meaning is conveyed with little reliance on. a shared physical context, was hypothesized to resemble the kinds of verbal skills required for school success. When the children were 5 years old, several emergent literacy skills were assessed, yielding a composite score. This outcome measure was correlated only with the mother's use of nonimmediate utterances while reading the unfamiliar book at age 3 (7 AD. Watson and Shapiro (1988) conducted a longitudinal study of a sample of 20 preschoolers. When the children were 2% years old, they and their parents (3 fathers, 17 mothers) were videorecorded during two bookread- ing sessions (with three different books) at the preschool that the children attended. It was hypothesized that adult usage of words referring to men- tal states and superordinate categories may foster growth in metalinguis- tic awareness and “‘decontextualized” use of language, which may be READING TO PRESCHOOLERS 275 particularly important for later literacy acquisition. Accordingly, the pro- portions of superordinates (¢.g., animal), basic-level terms (e.g., dog), subordinates (e.g., poodle), cognitive verbs (e.g., think), affective words (e.g., yukky) and verbs of speech (e.g., tell) were computed for each parent. A year later, seven emergent literacy scores (including four subtests and a total score on a print concepts test) were obtained for the children. As expected, parental use of basic level and subordinate terms was unrelated to outcomes. Of the 28 correlations that would be consis- tent with the hypothesis, only three (11%) were significant: between verbs, of speech by parents and children’s letter-word concepts (r = .46), and between affect terms and understanding of print direction (r = 46) and the total Concepts of Print score (r = .47). The other 25 coefficients ranged from ~.24 to .37 (median r = .09; all p > .10, two-tailed). Given the probable nonindependence of the outcome scores, the few large effects probably do not exceed the number that would occur by chance. ‘The findings of the preceding four studies suggest that even when the quality, rather than just the frequency, of parent-preschooler reading is examined, a stronger degree of association with emerging literacy skill does not appear to be found. Because Flood (1977) did not report any zero order correlations, and because coefficients for nonsignificant results were excluded from the correlation matrices presented by Dickinson and Tabors (1991), a median effect size cannot be computed. Given that only 10% of the total of 41 tests met the significance criterion, and given that the median effect size across Watson and Shapiro's (1988) 28 tests was only .10, we can surmise that the average correlation was less than .20 in these samples. Oral language outcomes. We turn next to a consideration of the three correlational studies in which language outcomes have been examined in relation to qualitative differences in parental behavior during shared read- ing. First, in addition to emergent literacy skills, selected language abil- ities were assessed when the children in Watson and Shapiro's sample were 3 years old (Watson, 1989; Watson & Shapiro, 1988). Specifically, measures of the child’s use of superordinate categories both expressively (when defining words) and receptively (in a verification task) were ob- tained when the children were 3 years old. Consistent with the authors” hypotheses, expressive categorization abilities were correlated with pa- ental usage of superordinates (r = .55), cognitive verbs (r = .43), and speech verbs (r = .49) during shared reading a year earlier. Correlations with children's receptive categorization abilities were strong for parental production of superordinates (r = .63), and approached significance for cognitive and speech terms (r = .43 and .39, p < .10). Proportions of affective terms in the parents’ speech, however, were uncorrelated with either measure of child language (r = ~.09 and .35). These results sug- 276 SCARBOROUGH AND DOBRICH gest that there may be rather specific links between the semantic aspects of parental bookreading and the lexical-semantic development of their children. Second, many facets of developing language were also measured at age 5 years for the sample in the aforementioned study by Dickinson and ‘Tabors (1991). These included PPVT scores and five measures derived from nonstandardized language production tasks such as defining words, answering questions about a story, describing a pictured scene, and con- structing a narrative about a sequence of viewed slides. These were an- alyzed in relation to the four observational indices of mothers’ nonimme- diate talk, described above. How often mothers produced these decon- textualized utterances while reading the novel book at age 3 was related to the formal adequacy of narratives children produced (r = .42). Of the total of 24 correlations tested, however, this was the only one to reach significance, so this finding is likely to have occurred by chance. Last, in the previous section of this review, we discussed DeBaryshe et al.'s (1991) findings regarding the relations between language outcome measures and reported amounts of shared reading in the home. For that sample of 28 2/4-year-olds, home observations were also videorecorded, and the proportions of [4 categories of maternal behavior were coded. A. principal components analysis yielded four factors that accounted for 80% of the variance: basic-level prompting of child participation (e.g., asking, easily answered questions, maintaining attention on the task); high-level prompting (e.g., asking more open-ended questions); close feedback (e.g., praising, repeating, or paraphrasing children's utterances); and far feedback (e.g., continuing the conversational topic, correcting the child's speech). Despite the small sample size, which can be problematic for such analyses, the derived factors were highly compatible with the authors’ theoretical expectations concerning parental prompting and feedback strategies. Not surprisingly, differences among parents along these four dimensions were related to differences in the children’s behavior during the bookreading sessions, such as the amount of child participation (R 89) and the quality of the child's contributions (R = .58). Twelve partial correlations, controlling for reported exposure to shared reading, were computed between the four qualitative dimensions of parent reading and the three language test scores (PPVT, EOWPVT, and ITPA-VE). Of these, only the one between basic level prompting and the ITPA-VE was significant ( the other coefficients ranged from —.31 to .16 (median 1 = ~.02). Simple correlations showed a similar pattern of results (me- dian r = .02). In summary, the qualitative aspects of shared reading that have been examined in the few studies to date do not appear to be more strongly related to children’s developing language abilities than to their emergent READING TO PRESCHOOLERS 2 literacy abilities. A median effect size cannot be calculated from the lim- ited data available, and it is difficult to make any estimate given that so much weaker effects were obtained by Dickinson and Tabors (4% of tests significant) and DeBaryshe (median r = .02) than by Watson and Shapiro (median r = .44). What sets the latter apart, perhaps, is the high degree of similarity between the parental and child measures, so that it could be demonstrated that cognitively challenging lexical choices by parents dur- ing shared reading were related to the corresponding semantic abilities of their children. While these findings are promising, it must also be noted that the sample size was unusually small, so converging data from other sources are needed to verify the relationship. Summary. Compared to the frequency of parent-preschooler reading, the qualitative nature of parental behavior during shared reading sessions has not generally been a better predictor of language or literacy outcomes in the five samples in which such relations have been examined. This conclusion, however, is based only on the correlational results presented in this section. Additional tests of the relation between the quality of adult reading and children’s developing abilities have been conducted in several intervention studies, which we will consider next. Intervention Studies: Effects of Modifying Parent-Preschooler Reading Practices (Table 5) ‘The research that we have summarized in the preceding four sections has all been correlational in nature. The associations between parent- preschooler reading and developing language and literacy skills that have been demonstrated in those studies cannot, therefore, necessarily be in- terpreted as indicating that how, or how much, parents read to their preschoolers influences the children’s development. The presumed causal direction of the relationship can only be determined through intervention studies, in which greater gains are demonstrated when the amount or quality of shared reading is increased for participants in a treatment pro- gram. In this section, we will review the seven studies that have com- pared experimental and control groups on language and emergent literacy outcomes following interventions designed to alter parental shared read- ing practices. First, we should mention an early report by Brzeinski (1964) that is sometimes cited as having provided such evidence for the efficacy of parent-preschooler shared reading. Subsequent to reporting the results of an experimental kindergarten program in the Denver public schools, Brzeinski described a “parallel study to determine how effectively par- ents could prepare their preschool children for reading" (p. 19). Three groups were compared, the parents of which, respectively: (1) were told to continue with the activities normally carried on within their house- SCARBOROUGH AND DOBRICH 208 con Buea SV sanuow g 10) ° a sere Apes Ke 3008 (sap aosuins roouss ° ay on sky ‘askin std eg ° 1 ot simak y ° on ot sak ¢ (61) Ayou00q sea OF uy sourvegd oN on we sek (6 wn samnseoye equowy resin pms Hoavasayl NOULNIAWSIN] 40 SLINS3Y, sa1avL Mason & McCormick (1983), MeCormick & Mason (1986) Whitehurst ea (1988) Lonigan (1993) DeBaryshe eta 1926) Books and guidance for # months Same for $ months ‘Gn cohor Bookreading behaviors for 4 weeks Bookreading behaviors for 6 weeks Bookreading behaviors or? weeks Bookreading frequency for 7 weeks Pre 23 years 35 years 24 years 6 » a same read “Lunch book read “Town” book spelling ‘eat. envi, print fread envir, pat fetter naming printing Sword reading Jeter sounds read "Farm book ‘ead “Bed” book speling | word reading letter knowledge teacherrated ach, reading group ead "Ghosts book ead “Apples” Book Spelling sword reading letter knowledge pevr EOWPVT TPAVE ppv, EOWPVT TTPAVE, Pevr EOWPVT TTPAVE PeVT EOWPVT HPAVE pov EOWPVT TPAVE E> Creat tate os en Pst 16 aSug = ap nae Sm ee 2 om shen 2 aa meee = 8 en Le nate? ao nsdn = 2) nse = 16) nse = 08) BS ty = 30 BSC ig = 36) BS. Un 19) ns. = = 09) 3s. = 08) 1. © =18) nae, = 29) ESC ry ESC = © ns. ty = 0) Rs. tad (gp = 34) 1s. tend (be = 3) 2S. yy = 08) n= an Bees sO 26 meee = Mn mee, acu Sas nee = ty meg 2 oan Note. E, experimental: C, control. Point biserial correlation coefficients are estimates based on available group means and standard deviations. SwaTOOHDsI OL ONIaVTY Lz 280 SCARBOROUGH AND DOBRICH holds; (2) were instructed to teach their children “the skills which are basic to beginning reading” using a guidebook and educational television programs; or (3) received guidance of experienced teachers via discus- sion/demonstration groups, as a supplement to the basic intervention pro- gram of Group 2. Achievement gains were reportedly made by those who practiced the beginning reading activities at least 30 min per week. Re- gardless of group, children who were read to (according to parental in- terviews) were said to show an increase in test scores, with the best performance exhibited by children who had both been read to and in- structed in reading by their parents frequently. Neither the sizes nor the ages of the three groups were reported, however, nor was it stated wheth- er children were assigned to groups randomly or whether the groups were equivalent in various important respects (¢.g., socioeconomic status, age, initial level of literacy knowledge) at the outset of the program. The duration of the intervention was not given, and it is not clear when or how the achievement of the children was assessed. Finally, these results were termed “interim,” pending a final evaluation at the end of the ongoing longitudinal project. To our knowledge, no subsequent report of the final results has appeared. In view of the paucity of information provided about the methods and results, this study is of limited usefulness in evaluating the hypothesis under review. In another early study (Irwin, 1960), the mothers of 24 infants from working class families were instructed to read aloud daily from children’s storybooks, and more generally to expose their babies to a lot of speech This regimen began when the children were 13 months old, and continued to age 30 months. At 2-month intervals, home visits were made to these children and to 10 others, of similar background, who received no enrich- ment of language input. A sample of the child's vocalizations produced on 30 breaths was transcribed using phonetic notation, and the number of phonemes (tokens) in each sample was tallied. From age 19 months on- ward, increasingly larger phoneme counts were obtained for the experi- mental group. These striking results are difficult to interpret, however, due to the language measure that was used. It is not known what relation, if any, phoneme frequencies measured in this manner have to the devel- ‘opment of other language skills. Questions can also be raised, of course, about how reliable the transcription method was, and about whether the transcriber was “blind” to the purposes of the study and to the group to which the family was assigned. In the absence of any replication, this evidence is difficult to evaluate. ‘Two British studies have also demonstrated increases in the developing skills of children whose parents were instructed to read to them more frequently at home. Donachy (1976) examined performance on the EPVT and the Receptive and Expressive scales of the RDLS before and after a READING TO PRESCHOOLERS 281 S-month intervention period. Three experimental groups of 16 children each were compared with three corresponding EPVT-matched control groups: 3-years-olds; 4-year-olds; and 4-year-olds who also attended a half-day nursery school. Although the analyses presented were inappro- priate and must be disregarded, it is rather clear from the descriptive information provided that the language scores of the first two experimen- tal groups rose more than those of the control groups in four of six com- parisons. In contrast, the third experimental group (who were read to at home and also attended nursery school) made smaller gains than the other two experimental groups, and did not outperform the controls who at- tended the nursery school. Although these findings are largely consistent with the hypothesis that increasing parent-preschooler reading fosters language growth (at least for children not enrolled in a preschool pro- gram), it should also be noted that the intervention program included a great deal more. Each week, while the children attended a playgroup, their mothers met for tea and discussion groups with teacher volunteers, received books and instructions for conducting other educational activi- ties at home, and were given feedback on the previous week's parent— child interactions. In short, it is not clear to what extent the results of the intervention can be attributed to daily bookreading per se, especially since 85% of the experimental group mothers reported in interviews that they read to their children prior to joining the program, and only 56% said that they changed their bookreading behavior as a result of participation. ‘Swinson (1985) studied 3- and 4-year-old children from working class families in which parent-preschooler reading prior to intervention was a daily activity for only 15%. The 13 experimental and 12 control subjects were nursery school classmates. The randomly assigned groups did not, differ initially with respect to age, gender, EPVT performance, or scores on the Verbal Comprehension and Naming Vocabulary subtests of the British Ability Scales (BAS). Mothers of the experimental group met initially with the child’s nursery teacher and attended a training session on making joint reading an enjoyable and verbally stimulating daily activity. In addition, a lending library for these mothers’ use was maintained in the classroom. After 9 months, the three language tests were readministered, and larger gains were found for the experimental group. When six non- standardized measures of preliteracy skill were later assessed for both groups, however, only on a “word matching” test (which we assume was a perceptual task) did the experimental group outperform the control group. This small study suggests that the language (especially lexical semantic) abilities of preschoolers who are not regularly read to by their parents may be enhanced when mothers are encouraged to read to them, more often, but that gains in subsequent literacy acquisition may be more difficult to bring about through such an intervention program.

You might also like