Article 1€¦ · Article 1 THE READING WARS 1 P.D AVID PEARSON Thisarticle’sfundamentalargumentisthatreadinginstructionandreadingresearchhavebeen ...

Article 1

THE READING WARS1

P. DAVID PEARSON

This article’s fundamental argument is that reading instruction and reading research have beenshaped by political forces desiring to privilege particular approaches to instruction or particularcombinations of methodological and epistemological perspectives on research. The swings inboth dominant pedagogies and dominant research paradigms are analyzed in terms of thesedetermining forces. The article concludes by championing balance and compatibility across bothinstructional approaches and research methods in hopes of arresting the pendulum swings thathave characterized the field for too many decades.

Keywords: reading; reading research; reading policy; history of reading; reading curriculum

3

Writing about the politics of readingsome 15 years ago (Pearson, 1989), Iwondered whether the whole-

language movement, which was the centerpieceof the reading field’s foray into constructivistpedagogy, was capable of maintaining the mantleof “conventional wisdom,” a status that at thattime, it was on the brink of achieving. I ques-tioned its enduring leadership capacity becauseof the curricular, philosophical, and politicalground on which it stood. Curricularly, I

expected that its guiding principles of authentic-ity (in texts, tasks, and tests) and curricular inte-gration—both within the language arts (acrossreading, writing, speaking, and listening) andbetween the language arts and other curricularareas—would run afoul of the powerful publish-ing lobbies in the United States. Philosophically,it is built on epistemologies of interpretationrather than realism, rejecting the idea of an exter-nal reality that we will eventually find if we justlook hard enough; as such, leaders of the whole

SOURCE: Pearson, P. D. (2004). The Reading Wars. Educational Policy, 18(1), 216–252. Reprinted by permis-sion of Sage Publications, Inc.

language movement would have been thrilled tofind provisional and situation-specific answersto burning policy questions such as, What is thebest way to teach beginning reading? Thoseviews would not sit well, I thought, in congres-sional or state legislative milieus or school boardchambers, places where truth and simple answersto policy questions are serious goals. Politically,I predicted that its commitment to grassrootsdecision making—a commitment requiring thateverything must be done to preserve as muchpower and prerogative for individual teachers(who must, in turn, offer genuine choices to indi-vidual students)—would doom it as a policy ini-tiative. In an atmosphere in which accountabilitysystems driven by externally mandated high-stakes tests lay just over the horizon, I wonderedwhether policy makers, or parents for that matter,would be willing to cede that level or prerogativeto a profession that in terms of its capacity todeliver achievement, seemed to be asleep at thewheel. My overarching question was whetherwhole language could withstand the pressure ofcurricular leadership, with implicit responsibilityfor whatever trends in achievement ensued. Mysuspicion was that it was better situated as aguerilla-like movement that made occasionalsorties into the policy world to snipe at those incurricular power.

In reflecting on those wonderments some 15years later, it is clear that whole language, alongwith its close constructivist cousins—literature-based reading, process writing, and integratedlanguage-arts instruction—did not experience along tenure in the seat of curricular power, atleast in the form in which it and its relationsexisted in the late 1980s and early 1990s.Whether the seeds of its demise were internalshortcomings, as I wondered in 1989, or externalpolitical forces of the sort that dominate the pol-icy conversation today, or some combination ofthe two, is a question that I return to at the end ofthis article after reviewing the important devel-opments in policy and practice that have shapedevents and interpretations in the interim.

THE GOLDEN YEARS OF

WHOLE-LANGUAGE INFLUENCE

Whole language did not suddenly emerge on thereading scene in the 1980s. Its roots(Y. Goodman, 1989) are in Deweyian-inspired,child-centered pedagogy and the integrated cur-riculum movements popular in England,Australia, and New Zealand (e.g., Holdaway,1984). It also owes part of its heritage to earlierAmerican movements, such as individualizedreading (Veatch, 1959) and language experience(Stauffer, 1980). But it was the incredible shift inthe scholarly paradigms that undergirded ourviews of reading acquisition that in my view,really laid the groundwork for its ascendancy(see Pearson & Stephens, 1993, for an account ofthese developments). The psycholinguisticallyoriented work of Roger Brown (1970), FrankSmith (1971), and Kenneth Goodman (1965,1969) sent the message that reading was more alanguage than it was a perceptual process. Thework in reading comprehension inspired by thecognitive revolution in psychology (seeAnderson & Pearson, 1984) established meaningas the core, not the residual outcome, of reading.Advances in sociolinguistic theory in the 1980s(Bloome & Green, 1984; Heath, 1983) and criti-cal literacy in the 1990s (Gee, 1989; Luke, 1995)established the understanding that all languageand, hence, all literacy learning is grounded inthe material motives of human interaction, withall of its social, political, and economic faces(however endearing or ugly they might be) intact.

When whole language emerged as a move-ment in the 1980s, it challenged the conventionalwisdom of basals and questioned the unqualifiedsupport for early code emphases that had grownbetween 1967 and the early 1980s. One of thegreat ironies of whole language is that its ascen-dancy into curricular prominence is best docu-mented by its influence on the one curricular toolit has most consistently and most vehementlyopposed, the basal reader. Basals changed

4 Section I • Planning and Organizing Literacy Instruction

dramatically in the early 1990s, largely, I amconfident, in response to the groundswell of sup-port within the teaching profession for wholelanguage and its close curricular allies, litera-ture-based reading and process writing.

Vocabulary control, already weakened duringthe 1970s in response to Chall’s (1967) admoni-tions, was virtually abandoned in the early 1990sin deference to attempts to incorporate more lit-erature, this time in unexpurgated form (i.e.,without the practices of adaptation and excerpt-ing that had characterized the basals of the 1970sand 1980s) into the Grade 1 program (Hoffmanet al., 1995). Phonics, along with other skills,was backgrounded, and literature moved to cen-ter stage.

Basal programs appropriated or, as somewhole-language advocates have argued, “basal-ized” the activities and tools of whole language.Thus, in the basals of the early 1990s, each unitmight have a writing process component inwhich the rhetoric if not the reality of some ver-sion of process writing was presented to teachersand students. In the 1980s, comprehension ques-tions, probably following a story line, might havesufficed for the guided reading section of themanual (the part that advises teachers on how toread and discuss the story), but in the 1990s,questions and tasks that supported deep probesinto students’ responses to literature becamemore prevalent. Another concession to literature-based reading was the creation and marketing ofclassroom libraries—boxed sets of books, usu-ally thematically related to each unit, thatteachers could use to extend their lessons andunits “horizontally” and enrich children’s literaryopportunities.

Basals also repositioned their “integrated lan-guage arts” and “integrated curriculum” strands.Dating back even to the 1920s and 1930s, basalshad provided at least a “token” section in whichteachers were encouraged to extend the themes orskills of the basal story into related writing (e.g.,rewriting stories), oral language (e.g., transform-ing a story into a play and dramatizing it),

or cross-curricular activities (e.g., conductingcommunity surveys, tallying the results, andreporting them), but these forays were regardedas peripheral rather than core. In the basals ofthe early 1990s, as skills moved into the back-ground, these integrated language-arts activitieswere featured more prominently as core lessoncomponents.2

These changes can, I believe, be traced to theprominent position of whole language as a cur-ricular force during this period (Pearson, 1992).Publishers of basals accomplished this feat ofappropriation not by ridding their programs ofthe skills of previous eras but by subtle reposi-tioning—foregrounding one component whilebackgrounding another, creating optional com-ponents or modules (e.g., an intensive phonicskit or a set of literature books) that could beadded to give the program one or another spin.Unsurprisingly, this created bulkier teachers’manuals and more complex programs.

Acceptance of whole language was not uni-versal. To the contrary, there was considerableresistance to whole language and literature-basedreading throughout the country.3 In many places,whole language never really gained a foothold.In others, what was implemented in the name ofwhole language was not consistent with thephilosophical and curricular principles of themovement; California, whole-language advo-cates would argue, is a case in point. Whole lan-guage got conflated with whole-class instructionand was interpreted to mean that all kids shouldget the same literature, even if teachers had toread it to them.4

Nor was there a single voice within thewhole-language movement. Whole-languagescholars and practitioners differed, and still dif-fer, on a host of issues such as the role of skills,conventions, and strategies within a language-arts program. Some said, if we can just bepatient, skills will emerge from meaningfulcommunication activities; others spur things onby taking advantage of spontaneous opportuni-ties for minilessons; still others were willing to

Article 1 • The Reading Wars 5

spur spontaneity a bit with minilessons andother transparently instructional routines.

Even so, it is fair to conclude that by the early1990s, whole language had become the conven-tional wisdom, the standard against which allelse was referenced. The rhetoric of professionalarticles belies this change. As late as the mid-1980s, articles were written with the presump-tion of a different conventional wisdom—aworld filled with skills, contrived readers, andworkbooks. By 1991–1992, they were writtenwith the presumption that whole-languagereforms, although not fully ensconced inAmerica’s schools, were well on their way toimplementation. The arguments in the 1990swere less about first principles of whole lan-guage and more about fine-tuning teaching reper-toires. The meetings of the Whole LanguageUmbrella grew to be larger than most large stateconventions and regional conferences of theInternational Reading Association. By 1995,whole language was no longer a collection ofguerrilla sorties into the land of skills and basalsthat characterized it through the mid-1980s. Ithad become the conventional wisdom, in rhetoricif not in reality.

THE DEMISE OF

WHOLE LANGUAGE

Toward century’s end, just when it appeared as ifwhole language, supported by its intellectualcousins (process writing, literature-based read-ing, and integrated curriculum), was about toassume the position of conventional wisdom forthe field, the movement was challenged seri-ously, and the pendulum of the pedagogicaldebate began to swing back toward the skills endof the curriculum and instruction continuum.Several factors converged to make the challengecredible, among them (a) unintended curricularcasualties of whole language; (b) questionableapplications of whole language; (c) the growth ofbalanced literacy as a mediating force in thedebate; (d) a paradigm shift in the ideology of

reading research; (e) increasing politicizationof the reading research and policy agenda;(f) increasing pressure for educators of allstripes, especially reading educators, to producemeasurable results; and (g) loss of the moral highground. All of these forces, but especially thosedelineated above as d through f, came together inone place—the reading first component of theNo Child Left Behind Act of 2001 (NCLB,2002). By the time this article was submitted forpublication, in mid-2003, reading first hadassumed the role of conventional wisdom inreading instruction, albeit by mandate rather thangroundswell, and only a few traces of whole lan-guage, which seemed so dominant only 7 yearsearlier, could be found in our schools and curric-ula. How did this remarkable political transfor-mation occur? That is the subject of this article.

Unintended Curricular Consequences

In its ascendancy, whole language changedthe face of reading instruction and in the process,left behind some curricular casualties, few ofwhich were intended by those who supportedwhole language. Those, including many curricu-lar moderates, who supported practices that werediscarded in the rise of whole language had dif-ficulty supporting the whole-language move-ment even though they might have beenphilosophically and curricularly sympathetic tomany of its principles and practices (see Pearson,1996). This lack of enthusiasm from curricularmoderates meant that whole language failed tobuild a base of support that was broad enough tosurvive even modest curricular opposition, letalone the political onslaught that it would expe-rience at century’s turn.

There were four casualties: skills instruction,strategy instruction, an emphasis on text struc-ture, and reading in the content areas. Earlier, Isuggested that one of the consequences of wholelanguage was the relegation of skills to the“appendices” of instructional programs. Inaccepting whole language, we tacitly accepted thepremise that skills are better caught in the act ofreading and writing genuine texts for authentic


purposes than taught directly and explicitly byteachers. The argument is the same for phonics,grammar, text conventions, and structural ele-ments. These entities may be worthy of learning,but they are unworthy of teaching. This positionpresents us with a serious conundrum as a pro-fession. Admit, for the sake of argument, that theskills instruction of the 1970s and earlier, withdecontextualized lessons and practice on “tex-toids” in workbook pages, deserved the criticismaccorded to it by whole-language advocates (andscholars from other traditions). But a retreatfrom most skills instruction into a world of“authentic opportunity” did not provide a satis-factory answer for teachers and scholars whounderstood the positive impact that instructioncan have. Many young readers do not “catch” thealphabetic principle by sheer immersion in printor by listening to others read aloud. For some itseems to require careful planning and hard workby dedicated teachers who are willing to balancesystematic skills instruction with authentic textsand activities (see Hiebert & Taylor, 1994, for adescription of many of the interventionsdesigned to accomplish just this balanced goal).

Strategy instruction (intentional attempts toequip students with meta-cognitive routines forunderstanding text, monitoring comprehension,and fixing things up when they go awry) wasanother casualty. This loss was particularly diffi-cult for scholars who spent the better part of the1980s convincing basal publishers and textbookauthors that the thoughtful teaching of flexiblestrategies for making and monitoring meaningwas a viable alternative to ubiquitous skillinstruction, where skills were taught as thoughthey were only ever to be applied to workbookpages and end-of-unit tests. But the strategylessons that filled basals in the mid- to late 1980swere virtually nonexistent in the basals of theearly to mid-1990s. Although there is no inherentbias in whole language or literature-based read-ing against the learning and use of a whole rangeof cognitive strategies, there is, as with phonicsand grammar, a serious question about whetherdirect, explicit instruction in how to use themwill help. The advice is to let them emerge from

attempts to solve real reading problems and puz-zles, the kind students meet in genuine encoun-ters with authentic texts.

Structural emphasis was also suspect withinwhole language. This suspicion extended to for-mal grammars, story grammars, rhetorical struc-tures, and genre features of texts. As with skillsand strategies, whole-language reformers did notclaim that students should not learn and developcontrol over these structural tools; they simplyclaimed that like skills, they are best inferredfrom reading and writing authentic texts in theprocess of making meaning. So, the advocatesare comfortable in adopting Smith’s (1983)admonition to encourage kids to read like awriter (meaning to read the text with a kind ofcritical eye toward understanding the tools andtricks of the trade that the author uses to makepoints and achieve his or her effects on readers),but they would likely reject a systematic set oflessons designed to teach and assess children’scontrol of story grammar elements (such as plot,characterization, style, mood, or theme) or somesystem for dealing with basic patterns of the var-ious genres of expository text. As with skills andstrategies, many in the field sought a compro-mise alternative to both the formulaic approachof the early 1980s and the “discovery” approachof the new reforms—dealing with these struc-tural elements as they emanate from stories thata group is currently reading can provide someguidance and useful tools for students andteachers.

Content area reading also suffered during theascendancy of whole language and literature-based reading. Content area texts—expositorytexts in general, but especially textbook-likeentries—were not privileged in aworld of literature-based reading. There is a certain irony in thisdevelopment, for it is competence with exposi-tory reading, not narrative reading, that most con-cerns educators and future employers. The costhere has been very dear. Concerned that studentseither cannot or will not read textbook assign-ments, most high school teachers have choseneither to read the text to students or even morelikely, to tell students what they would have


encountered had they or could they have read it.Although understandable, this approach is ulti-mately counterproductive. There comes a time inthe lives of students—either when they go tocollege or enter the world of work—when othersexpect them to read and understand informationaltexts on their own and in printed form rather thanthrough oral or video transformation.5

Because whole language did not go out of itsway to accommodate these structural- and con-tent-focused curricular practices, those who weresympathetic with whole language but also cham-pions of one or another approach were not avail-able to help whole language respond to thecriticism leveled at it in the late 1990s. Buildingallies across boundaries of curricular politicaldivides was not, as it turned out, a strength of themovement.

Questionable Applicationsof Whole Language

One of the dilemmas faced by any curricularinitiative is sustaining the integrity of the move-ment without imposing the very sorts of controlsit is trying to eliminate. Whole language did notfind a way to manage this dilemma, and it suf-fered as a consequence. Many schools, teachers,and institutions appropriated the whole-languagelabel without honoring its fundamental princi-ples of authenticity, integration, and empower-ment. Basal-reader publishers made the mostobvious and widespread appropriation, someeven positioning their basal series as “whole-language” programs. The most egregious misap-plication was the conflation of whole languagewith whole-class instruction. Nowhere was thisconflation more extreme than in the implementa-tion of the California literature framework. Thelogic that prevailed in many classrooms was thatit was better to keep the entire class together, allexperiencing the same texts, even if it meant thatthe teacher had to read the text to those childrenwho lacked the skills to read it on their own.Implicit in this practice are two interestingassumptions: (a) that getting the content of thestories is the most important goal for reading

instruction, and (b) that the skills and processesneeded to read independently will emerge some-how from this environment in which manystudents are pulled through texts that far exceedtheir grasp, given the sophistication of their cur-rent skills repertoire. Needless to say, whole lan-guage had enough on its hands dealing with itsown assumptions and practices; these philosoph-ical and curricular misapplications exposed themovement to a whole set of criticisms thatderived from practices not of its own making.

A plausible explanation for the misapplicationof whole language was its lack of an explicit planfor professional development. Given its grass-roots political assumptions, it is not surprisingthat whole language gave teachers a wide berthfor making curricular and instructional deci-sions. It assumed that teachers who are empow-ered, sincere, and serious about their work wouldbe able to tailor programs and activities to theneeds and interests of individual children. Suchan approach makes sense only when teacherknowledge is widely and richly distributed in ourprofession. To offer these prerogatives in the faceof narrow and shallow knowledge is to guaranteethat misguided practices, even perversions of thevery intent of the movement, will be widespread.The puzzle, of course, is where to begin thereform—by ensuring that the knowledge pre-cedes the prerogative, or by ceding the preroga-tive to teachers as a way of leveraging theirmotivation for greater knowledge. Similar argu-ments have been made for the reform movementsin mathematics (i.e., that the reforms got outahead of the professional knowledge base); inter-estingly the reform movement in mathematicshas experienced a fate similar to that of thewhole-language movement (see Good & Braden,2000; Schoenfeld, 2004.)

Balanced Literacy

Although it has reached its peak in the past 5years, concern about extreme positions, be theyextremely child centered (such as the more radi-cal of whole-language approaches) or extremelycurriculum centered (such as highly structured,


unswerving phonics programs), is not new.Voices from the middle, extolling balancedapproaches or rationalizing the eclectic practicesof teachers, began to be heard even in the earliestdays of whole language’s ascendancy.6 Scholarsand teachers raised a number of concerns aboutthe assumptions and practices of the whole-language movement. Most important, theyexpressed concern about the consequences ofwhole language outlined earlier in this article.They questioned the assumption that skills arebest “caught” during the pursuit of authenticreading activity rather than “taught” directly andexplicitly. They also questioned the insistence onauthentic texts and the corollary ban on “instruc-tional” texts written to permit the application ofskills within the curriculum. They questioned thezeal and commitment of the movement quamovement, with its strong sense of insularity andexclusivity. Finally, they worried that the presstoward the use of authentic literature and litera-ture-based reading would eradicate, albeit unin-tentionally, what little progress had been madetoward the use of informational texts and teach-ing reading in the content areas (Pearson, 1996).

Ironically, in the past few years, these voicesfrom the middle have found themselves respond-ing not to those who hold a radical whole-language position but to those who hold stead-fastly to the phonics first position. Even so, thefact that those with centrist positions were notinclined to defend whole language when thepolitical campaign against it began in the middle1990s undoubtedly hastened the demise of wholelanguage as the pretender to the title of conven-tional wisdom.

Changing Research Paradigms

Prior to the 1980s, qualitative research in anyform had little visibility within the readingresearch community. Among the array of qualita-tive efforts, only miscue analysis7 and some earlyforays into sociolinguistic and anthropologicalaccounts of literacy had achieved much in theway of archival status.8However, all that changedin the 1980s and early 1990s. Qualitative

research more generally, along with more spe-cific lines of inquiry taking a critical perspectiveon literacy as a social and pedagogical phenom-enon, became more widely accepted as part ofthe mainstream archival literature.9 Treatisespointing out the shortcomings of traditionalforms of quantitative inquiry, especially experi-mental research, appeared frequently in educa-tional research journals.10 Much of the researchthat undergirds whole language comes from thismore qualitative, more interpretive, more criticaltradition. Thus, the credibility of this type ofresearch increased in concert with the influenceof whole language as a curricular movement.

Somewhere in the mid-1990s, the discourseof literacy research began to take a new turn.Stimulated by research supported by theNational Institute for Child Health and HumanDevelopment, a “new” brand of experimentalwork began to appear, beginning in the mid-1980s and gathering momentum steadily sincethat time (Lyon, 1995; Lyon & Chhaba, 1996).This is experimentalism reborn from the 1950sand 1960s, with great emphasis placed on “reli-able, replicable research,” large samples, randomassignment of treatments to teachers and/orschools, and tried and true outcome measures.11

It finds its aegis in the experimental rhetoric ofscience and medicine and in the laboratoryresearch that has examined reading as a percep-tual process.12 Although it was not broadlyaccepted by the reading education communitywhen it first appeared, this work found a verysympathetic ear in the public policy arena.13

The political positioning of this research isimportant, but so is its substance. Two themesfrom this work have been particularly importantin shaping a new set of instructional practices—phonemic awareness and phonics instruction.

The absolutely critical role played by phone-mic awareness (the ability to segment the speechstream of a spoken word, e.g., /cat/ into compo-nent phonemes /cuh + ah + tuh/ and/or to blendseparately heard sounds, e.g., /cuh + ah + tuh/into a normally spoken word /cat/) in the devel-opment of the ability to decode and to read formeaning has been well documented in the past


decade and a half (Adams, 1990; Juel, 1988;Snow, Burns, & Griffin, 1998). Irrespective ofmode of instruction, the overwhelming evidencesuggests that phonemic awareness is a necessarybut not a sufficient condition for the develop-ment of decoding and reading. First, childrenwho possess high degrees of phonemic aware-ness in kindergarten or early in first grade arevery likely to be good readers throughout theirelementary school careers (Juel, 1988). Second,almost no children who are successful readers atthe end of Grade 1 exhibit a low level of masteryof phonemic awareness. On the other hand, asubstantial proportion of unsuccessful end-of-Grade-1 readers possess better than averagephonemic awareness; this evidence is the criticalpiece in establishing that phonemic awareness isa necessary but not a sufficient condition forreading success. Although we can be confidentof its critical role in learning to read, we are lesssure about the optimal way to enhance its devel-opment. Many scholars have documented theefficacy of teaching it directly, but they alsoadmit that it is highly likely to develop as a con-sequence of learning phonics, learning to read, orespecially learning to write, especially whenteachers encourage students to use inventedspellings (see Adams, 1990; Juel, 1991).Research in whole-language classrooms (Clarke,1988; Winsor & Pearson, 1992) suggests thatwriting is the medium through which bothphonemic awareness and phonics knowledgedevelop—the former because students have tosegment the speech stream of spoken words tofocus on a phoneme and the latter because thereis substantial transfer value from the focus onsound-symbol information in spelling to symbol-sound knowledge in reading.

The second consistent thread in the newexperimentalism of the 1990s was the emphasison the code in the early stages of learning toread. Reminiscent of Chall’s (1967) earlier con-clusions, scholars in this tradition advocatedphonics first, fast, and simple.14 Less well docu-mented, and surely less well agreed on, is theoptimal course of instruction to facilitate phonicsdevelopment. Even Gough (Gough & Hillinger,

1980), a classic bottom-up theorist, while argu-ing that what distinguishes the good reader fromthe poor reader is swift and accurate word iden-tification, suggested that an early insistence onreading for meaning may be the best way todevelop such decoding proficiency. Both Juel(1991) and Gough are convinced that studentscan learn how to read when they have cryptoan-alytic intent (a disposition to decipher thespecific letter-to-sound codes), phonemic aware-ness, an appreciation of the alphabetic principle(i.e., regardless of the numerous exceptions, let-ters do stand for sounds), and “data” (some textsto read and someone to assist when the goinggets tough).

After reviewing available instructional evi-dence, two of the most respected scholars in thistradition, Marilyn Adams and Connie Juel, inde-pendently concluded that children can andshould learn the “cipher” through a combinationof explicit instruction in phonemic awarenessand letter-sound correspondences, a steady insis-tence on invented spellings as the route to con-ventional spellings in writing activities, and lotsof opportunity to read connected text (especiallywhen the texts contain enough decodable wordsto allow students to apply the phonics informa-tion they are learning through explicit instruc-tion). Both of these reviewers, known for theirsympathies toward instruction in the code, arequick to add that rich experiences with language,environmental print, patterned stories, and “bigbooks” should also be a staple of effective earlyreading instruction (Adams, 1990; Juel, 1991).15

This new research paradigm became offi-cially codified by the appearance, in rapid suc-cession, of two research syntheses—thepublication of the report of the NationalAcademy of Science’s Committee on PreventingReading Difficulties (Snow et al., 1998) and thereport of the National Reading Panel (NRP)(2000). These are very different documents, andthey have exerted very different influences on thereading field, particularly on reading policy. ThePreventing Reading Difficulties report was con-ducted in the tradition of “best evidence” synthe-ses: well-established scholars meet, decide on


the issues, the domain of relevant research, andsome subdivision of labor, do the work, write upthe results, and turn the manuscript over to a setof editors to bring some synthetic clarity to theentire effort. As such, it considered a range ofstudies conducted within very different researchtraditions using very different research methods.The result was an apology for a balanced view ofreading instruction, but with a special nod tophonemic awareness and phonics first and fast.A solid piece of scholarship many of us thought,but not much news (Pearson, 1999).

Authorized by congressional mandate, theNRP report used the most “scientific” reviewapproaches (i.e., meta-analysis, at least whereverthey could) available to them to distill from exist-ing research what we knew about the efficacy ofteaching phonemic awareness, phonics, fluency(instantiated as either guided reading instructionor independent reading), comprehension, andvocabulary; in addition, they investigated the sta-tus of the research base on teacher education andprofessional development and attempted toreview research on technology and literacy. It isinteresting to note that according to CatherineSnow (2001), one of the lead authors of thePreventing Reading Difficulties report, officialssuch as G. Reid Lyon and DuaneAlexander fromthe National Institute for Child Health andHuman Development, one of the sponsoringagencies of the NRP, were concerned about thePreventing Reading Difficulties report because itwas vague and did not discriminate betweentrustworthy and untrustworthy research. TheNRP report is noteworthy on a number ofgrounds. First, the actual conclusions in the mainreport are consistent with earlier attempts tosummarize the knowledge base on these keyissues, such as Becoming a Nation of Readers(Anderson, Hiebert, Scott, & Wilkinson, 1984)and Preventing Reading Difficulties (Snow et al.,1998), and point to a balanced approach to teach-ing reading. Second, although the vote of confi-dence in teaching phonics and phonemicawareness was strong and direct, it was moder-ated by important caveats that limit the applica-bility of these important instructional tools. For

example, phonics was found to be a usefulinstructional approach, but only in a particulartime frame (Grades K–1); it was not effective forolder students. Moreover, although the analysisprivileged systematic phonics, nothing in theanalysis implicated a particular approach (e.g.,synthetic or letter-by-letter phonics vs. analyticphonics), nor was there any explicit support fordecodable text. Also, the authors of the NRPreport were careful, in their conclusions, to sug-gest that phonics by itself was not the total read-ing program: “Finally, it is important toemphasize that systematic phonics instructionshould be integrated with other reading instruc-tion to create a balanced reading program.Phonics instruction is never a total readingprogram ” (NRP, 2000, p. 2-135).

Third, the authors of the NRP report werevery clear about which topics and studies wouldbe included. It would review only those topics forwhich there existed a sufficiently large pool of“potentially viable” experimental studies. Henceissues of grouping, the relationship of reading towriting, the role of texts in reading acquisition—just to name a few of the more obvious issuesthat schools and teachers must address in craft-ing local reading programs—are not addressed atall. Regarding specific studies, they wouldinclude only those that met minimal criteria:employ an experimental or quasi-experimentaldesign with an identifiable comparison group,measure reading as an outcome, describe partic-ipants, interventions, study methods, and out-come measures in sufficient detail to “contributeto the validity of any conclusions drawn.”Natural experiments of the sort found in large-scale evaluation efforts or epidemiological inves-tigations of relationships between methods andoutcomes were excluded.

Vis-à-vis whole language, the point isstraightforward: The changes in the dominantparadigm meant that the research base on whichwhole language was grounded (all of those closeethnographies of individual classrooms andteacher action stories) was no longer privilegedin official conversations about “research”-basedpractice. Numbers, not compelling stories, were


the order of a new day; and it was not clearwhether there was a place for constructivist ped-agogy in general or whole language in particular,in these new conversations.

Politicization of the ReadingResearch and Policy Agenda

From its beginnings, one of the great hopes ofeducational research (and those who conduct it)has been that policy makers will take researchseriously when they establish policy at a local,state, or national level. After all, the improve-ment of educational practice is the ultimate goalof educational research, and policy is oursociety’s most transparent tool for educationalimprovement. Historically, however, research hasbeen regarded as one among many informationsources consulted in policy formation—includingexpert testimony from practitioners, informationabout school organization and finance, and eval-uations of compelling cases. In the past halfdecade, research, at least selective bits ofresearch, has never been taken more seriously.Several laws in California make direct referencesto research. For example, in 1998, CaliforniaAssembly Bill 1086 prohibited the use of Goals2000: Educate America Act of 1994 money forprofessional developers who advocated the useof context clues over phonics or supported theuse of “inventive [sic] spellings” in children’swriting. The federally sponsored ReadingExcellence Act of 1998, which allocatedU.S.$240,000,000 for staff development in read-ing, required that both state and local applica-tions for funding base their programs on researchthat meets scientifically rigorous standards. Thescientifically rigorous phrase was a late entry; inall but the penultimate version of the bill, thephrase was reliable, replicable research, whichhad been interpreted as a code word for experi-mental research. As of early 1999, “phonicsbills” (bills mandating either the use of phonicsmaterials or some sort of teacher training toacquaint teachers with knowledge of the Englishsound-symbol system and its use in teaching)had been passed or were pending in 36 states.16

The NCLB made this goal of “evidence-basedpractice” even more explicit, with the phrase sci-entifically based reading research appearingmore than 110 times in the Reading First portionof this act reauthorizing Title I.

Policy makers like to shroud mandates andinitiatives in the rhetoric of science, and some-times that practice results in strained, if not inde-fensible, extrapolations from research. This hashappened consistently in the reading policyarena in the past decade. Three examples makethe point vividly. First, California Assembly Bill1086, with its prohibition on context clues andinvented spelling, represents an ironic applica-tion of research to policy. The irony stems fromthe fact that many of the advocates of a return tocode emphasis, such as Marilyn Adams, read theresearch as supporting the use of inventedspellings in the development of phonemic aware-ness and phonics (Adams, 1990). Second, themandate in several states calling for the use ofdecodable text (usually defined as text consistingof words that could be sounded out using a com-bination of the phonics rules taught up to thatpoint in the program plus some instant recogni-tion of a few highly frequent “sight” words) isbased on the thinnest of research bases. The ideais that children will learn to use their phonicsbetter, faster, and more efficiently if the textsthey read permit facile application of the princi-ples they are learning. Although it all soundsvery logical, there is precious little research evi-dence to support the systematic and exclusiveuse of decodable text.17 This lack of evidence,however, does not seem to have deterred advo-cates who, on the phonics issues, championedscientific evidence as the gold standard for pol-icy implementation.

The third example comes from the state ofCalifornia’s application for Reading First funds.The Reading First provision of NCLB requiresthat all elements of a program’s application—instructional materials, assessments, and profes-sional development—be supported by scientificallybased reading research. Scientifically based profes-sional development was defined in the Californiaapplication as the professional development


required to help teachers implement the two state-adopted commercial reading programs; the pro-posal was accepted by federal officials withoutobjection to this definition. This developmentwas convenient in a financially troubled statethat could ill afford to pay for the professionaldevelopment for its new adoptions entirely on itsown hook. The irony here, of course, is that thetwo commercially adopted programs inCalifornia, although they might be able to trace15% to 20% of their practices to scientificresearch, are no more research based, let alonescientifically based (i.e., they have not regularlyused randomized trials to test their efficacy) thanthe average run-of-the-mill commercial program.They are now, of course, officially blessed asscientific.

When research moves into the policy arena,one of two outcomes are most likely. If theresearch is widely accepted by members of theprofession from which it comes, widespreadacceptance and implementation usually follows.This often occurs in medical, pharmaceutical, oragricultural research. If widespread consensuson what the research says about practice is notreached, then research-based policy initiativesare likely to sharpen and deepen the schisms thatalready exist and the whole enterprise is likely tobe regarded as a “war” among balkanized fac-tions within the field. The latter scenario appearsto characterize the reading field. The entry ofscience into the reading research community, andits accompanying blessing of particularapproaches to teaching reading, has met withconsiderable resistance, some overt and somequiet, within the reading research community.The most vocal and prominent voices in theresistance have been Elaine Garan, DennyTaylor, and Richard Allington. Soon after thepublication of the report of the Committee onPreventing Reading Difficulties of the NationalAcademy of Science in 1998, D. Taylor (1998)published her treatise unveiling the “spin doctorsof science.” Essentially, D. Taylor attempted toshow how the conservatives involved in promot-ing the “new-phonics” agenda had used publicrelations techniques rather than science to

accomplish two goals: (a) to convince policymakers and the general public that the answer toteaching reading was more phonics earlier, and(b) to discredit public education more generally.Garan’s (2001, 2002) critique focused on thereport of the NRP, and essentially, she offers twotypes of critique: internal and external. The inter-nal critique holds the methodology of meta-analysis to its own standards, and she tried toshow that the NRP effort was a fundamentallyflawed approach to meta-analysis. For example, aprinciple of meta-analysis (Salkind, 2000) is thatalthough the outcome measures need not beidentical from one study to another, they shouldrepresent the same underlying construct; theNRP phonics analysis, Garan argued, fails thisstandard. She pointed out the many internal con-tradictions in method: for one group, eight stud-ies are too few to move ahead with themeta-analysis whereas for another, nine isenough. However, perhaps most important,Garan pointed out that the statements included inthe executive summary of the report are ofteninconsistent with comparable statements in themore elaborated reports of the various subgroups(on phonics, comprehension, and the like). Icould not agree more with this last critique; as Iwill point out later, these discrepancies with theelaborated report only worsen when we examinethe more “popular” version of the report writtenfor general consumption and the headlines dis-tilled by reporters for headlines and newspaperarticles. Allington (2002) took a third approach.He enlisted the help of several colleagues in hisedited volume to make the case that for the past30 years, a conservative lobby has been trying tomanipulate several policy levers (standards,assessment, professional development, and evi-dence-based practice) to shape a national readingpolicy that privileges basic skills for students andlimits teacher education to training rather thaneducative practices. The case he made could becharacterized as a sort of “skill the kids and de-skill the teachers” approach (my words, not his).

Interestingly, the debate, accompanied by itswarlike metaphors, appears to have more life inthe public and professional press than it does in


our schools. Reporters and scholars revel inkeeping the debate alive and well, portrayingclearly divided sides and detailing a host of dif-ferences of a philosophical, political, and peda-gogical nature (see Manzo, 1997, 1998a, 1998b).Teachers, by contrast, often talk about, and moreimportant enact, more balanced approaches. Forexample, several scholars, in documenting thepractices of highly effective, highly regardedteachers, found that these exemplary teachersemployed a wide array of practices, some ofwhich appear decidedly whole language in char-acter (e.g., process writing, literature groups, andcontextualized skills practice) and some ofwhich appear remarkably skills oriented (explicitphonics lessons, sight word practice, and com-prehension strategy instruction). Exemplaryteachers (e.g., Pressley et al., 2001; B. M. Taylor,Pearson, Clark, & Walpole, 2000; Wharton-MacDonald, Pressley, & Hampton, 1998) appearto find an easier path to balance than eitherscholars or policy pundits.

Producing Measurable Results

Evaluation has always posed a conundrum forwhole-language supporters. First, some advo-cates oppose the use of any sort of externallymandated or administered assessments as amatter of principle, holding that assessment isultimately the responsibility of a teacher in col-laboration with a student and his or her parents.Second, even those supporters who are open toexternal forms of accountability, or at leastreporting outside the boundaries of the class-room or school, often claim that standardizedtests, state assessments, and other external mea-sures of student accomplishment do not providesensitive indicators of the goals of curriculabased on whole-language principles. Mostappealing would be assessments that are class-room based and individualized in nature, withthe option of aggregating these sorts of data atthe classroom and school levels when account-ability comes knocking. During the 1990s, manyfelt that the increased emphasis on performanceassessment and portfolios would fill this need.18

In an age of high expectations, explicit stan-dards, and school- and classroom-level account-ability, none of these options is a good fit withthe views and desires of policy makers and thepublic. Both of these constituents seem quiteuneasy about the quality of our schools and oureducational system, so uneasy that leavingassessment in the hands of our teachers seems anunlikely outcome. It is not at all clear to me thatthe proponents of at least strong versions ofwhole language can, or will be willing to, holdthemselves accountable to the sorts of measuresthat the public and policy makers find credible.

Loss of the Moral High Ground

One other factor, although difficult to docu-ment, seems to be operating in the rhetoric of thefield in the first years of the 21st century. Wholelanguage, and constructivist approaches gener-ally, has always privileged the role of the teacheras the primary curriculum decision maker.Teachers, the argument goes, are in the best posi-tion to serve this important role because of theirvast knowledge of language and literacy devel-opment, their skills as diagnosticians (they areexpert “kid watchers”), and the materials andteaching strategies they have at their disposal.And in the arguments against more structuredapproaches, this is exactly the approach whole-language advocates have taken: “Don’t makethese decisions at the state, district, or even theschool level. Arm teachers with the professionalprerogative (and corollary levels of professionalknowledge) they need in order to craft uniquedecisions for individual children.” Although thismay seem a reasonable, even admirable position,it has recently been turned into an apology for aself-serving teacher ideology.19The counter argu-ment suggests that the broad base of privilegeaccorded to teachers may come at the expense ofstudents and their parents. Thus, those who advo-cate a strong phonics-first position often take themoral high ground: “We are doing this forAmerica’s children (and for YOUR child!)—sothat they have the right to read for themselves.”Even if one opposes this rhetorical move, it is


hard not to appreciate the clever repositioning onthe part of those who want to return to morephonics and skills.

The Net Effect

Taken together, these factors created a policyenvironment in which whole language, or anyother constructivist movement for that matter,was unlikely to flourish as the mainstreamapproach to teaching reading and writing. In thefinal analysis, however, I believe that the reluc-tance to own up to the “measurable results” stan-dards was the Achilles’ heel of whole language.If whole-language advocates had been willing toplay by the rules of external accountability, toassert that students who experience good instruc-tion based on solid principles of progressive ped-agogy will perform well on standardized testsand other standards of performance, they wouldhave stood a better chance of gaining a sympa-thetic ear with the public and with policy makers.And as long as the criteria for what counts as evi-dence for growth and accomplishment are vagueor left to individual teachers, the public couldquestion the movement and wonder whose inter-ests were being served by an unwillingness tocommit to common standards.

LOOKING AHEAD

So where has this journey left us? And wherewill it take us next? I want to divide my analysisof the future of reading policy into two strands,research and curriculum, because these two facesof reading policy, although often joined at thehip, occasionally privilege different themes andissues. I will close by bringing them backtogether.

Research Policy

Complementarity as a Scientific Value

In the current research context, literacy schol-ars find themselves between a rock and a hard

place. The official views of research promul-gated by the federal government in its researchprograms administered within the Department ofEducation are weighted toward quantitative andexperimental work.At the same time, the work ofmany, perhaps even most, literacy researchersand doctoral students in research training pro-grams is decidedly qualitative, narrative, and/orethnographic in character. An impending crisis?A confrontation of the immovable object and theirresistible force? Or just the exclusion of a widearray of literacy scholars from federally fundedresearch efforts? I would bet on the exclusion,but I hope and argue for a rapprochement amongmethods and even epistemologies.

Regarding science, my fundamental claim isthat reading research can never be truly rigorous,indeed truly scientific, until and unless it privi-leges all of the empirical and theoretical method-ologies that characterize the scientific disciplines.Included among those methodologies wouldsurely be experimentation and of course random-ized field trials of the sort that are being proposedfor several federally sponsored programs, but therange of scientific methods would extend to:

• careful descriptions of phenomena in their nat-ural settings (just like Darwin did and just liketoday’s environmental scientists);

• examinations of natural correlations amongvariables in an environment, just to see whatgoes with what;

• natural experiments in which we take advantageof the differences that serendipity and the nor-mal course of events have created between twoor more settings that are otherwise remarkablysimilar—the most common form of this effortin education being outlier studies and the evenmore common approach in public health’s epi-demiological studies;

• data gathered in the name of theory buildingand evaluation—just to see if we can explainthe nature of things;

• design experiments in which we adopt a plan-ful, incremental approach to knowledge refine-ment, with each successive step buildingcarefully on what was learned in the last; and


• the use of qualitative tools such as ethnographyand discourse analysis in concert with random-ized experiments to describe what is reallygoing on inside those randomly assigned treat-ments, so that we can explain why a treatmentworked or did not work, or whether the range ofvariation in treatments is so great across sitesthat it is doubtful that it can really be called thesame intervention across sites, or what the con-sequences, especially the unintended conse-quences, of an intervention might be.

As good as randomized experiments are fordetermining the overall efficacy of interventions,they are very short on details about the interven-tions, such as why, how, for whom, and underwhat conditions interventions work. For that weneed complementary methods, and this is wherequalitative methods come into play. DonaldCampbell (1984), one of the foremost designmethodologists of the 20th century and the coau-thor of the infamous book on quasi experiments(Campbell & Stanley, 1963), the classic treat-ment of threats to internal and external validity,recognized this need for complementarity:

To rule out plausible rival hypotheses we needsituation-specific wisdom. The lack of this knowl-edge (whether it be called ethnography, programhistory, or gossip) makes us incompetent estima-tors of program impacts, turning out conclusionsthat are not only wrong, but are often wrong insocially destructive ways. . . .

There is the mistaken belief that quantitativemeasures replace qualitative knowledge. Instead,qualitative knowing is absolutely essential as a pre-requisite for quantification in any science. Withoutcompetence at the qualitative level, one’s computerprintout is misleading or meaningless. (pp. 141–142)

We hear a lot of talk about randomized fieldtrials in medical and pharmaceutical research,and we are advised to follow their lead. I agree.But if we follow medicine and pharmacology,then we should follow them all the way down theroad of science. Let us remember that beforeresearchers in those fields get to the last 10% ofthe journey, which is when they invoke random-ized field trials in anticipation of advocacy and

policy recommendations, they have already useda much wider range of methodologies, includingmuch observation, description, examinations ofrelationships, and just plain messing around (thatis a technical term used by scientists to describewhat they spend most of their time doing) to travelthe first 90% of that journey. So let us talk aboutcomplementarities and convergence amongmethods rather than competition and displace-ment of one worldview with another. This is themessage of the recent report on educationalresearch by a committee empanelled by theNational Academy of Science (Shavelson &Towne, 2002), a message I heartily endorse.

If we rush too soon to the last 10% of the jour-ney and enamor ourselves of randomized field tri-als for their own sake, we are likely to end upconducting expensive experiments on interven-tions that were not worth evaluating in the firstplace. A drug company would never think of con-ducting a randomized field trial on a new drugthat had not gone through a thorough basicresearch phase in which biochemical theories, try-outs on nonhuman organisms, correlationalresearch on chemical components of the drug inthe natural environment, and probably someserendipitous case studies of individual subjectswho volunteered to use the drug out of despera-tion all played a key role.We should ask no less ofeducational interventions and programs. An inter-vention that is based on bad theory or no theory isnot likely to yield a significant contribution topractice in the long run. To know that somethingworked without a clue about how and why itworked does not advance either our scientific orprofessional understanding of an educationalissue.We cannot afford blind experimentation andhorse races with interventions of unknown theo-retical characteristics. As our candidates for ran-domized field trials, we want treatments andinterventions that have gone through these variousstages of scientific development.

I fear that as a profession we have fallen into amethodological trap.We have become so attachedto our methodologies and to their epistemological(some would say ideological) underbellies thatwe, as individuals, are likely to begin our work by


looking for a question that fits our methodologi-cal preferences rather than the other way around.This does not serve our profession well, for itallows us to address questions that may or maynot be of great relevance to policy and practice.We must return to the ethic of insisting that just asform follows function in language, so methodsmust follow questions in research. And if we donot, as individuals, possess the range of method-ological expertise to address different sorts ofquestions, then we ought to align ourselves withscholarly communities in which such expertise isdistributed among its members.

As a curious and ironic footnote, I wouldpoint out that complementarity across methods isconsistent with the definition of scientificallybased reading research in the Reading First por-tion of NCLB; the definition includes thesestandards:

• employs systematic, empirical methods thatdraw on observation and experiment;

• involves rigorous data analyses that are ade-quate to test the stated hypotheses and justifythe general conclusions;

• relies on measures that provide valid dataacross observers and occasions; and

• is published in peer-reviewed journals (orreviewed by a duly constituted panel).

The Complexity of Researchin Education

Complexity in the policy arena is always adouble-edged sword. To assert that educationalresearch is complex is to imply that there issomething categorically different about educa-tional research in comparison to research in agri-culture, physics, chemistry, medicine, or evenpsychology. Usually the complexity is attributedto the human factor and the variation introducedby human activity:

• that individuals differ from one another;

• that they live and work in groups;

• that the members of the group influence whatothers do, how they act, and what they believe;and

• that when humans are involved, things changein unpredictable ways.

David Berliner (2002), in a persuasive articlein a recent issue of Educational Researcher, putsforward just such a view. And there is much truthin the argument. I know this all too well from myown experience in trying to do large-scaleresearch on best practices (B. M. Taylor, Pearson,Peterson, & Rodriguez, 2003). We tried to do anoutlier analysis of schools that beat the odds pre-dicted by their demographics. However, wefound that high poverty/high performance statusis not a static characteristic. Some schools thatentered the study with a record of high achieve-ment foundered; some with reputations as failingthe mark changed their ways. Had we not col-lected a wide array of student outcomes (whichallowed us to build post hoc indicators of whowas and was not beating the odds) and an evenwider array of indicators of school reform effortsand teacher practices, we would not have beenable to unearth program and instructional char-acteristics that explain variation in achievementgrowth. Moreover, we found that a combinationof quantitative and qualitative approaches wereabsolutely essential in teasing out important rela-tionships between programs and outcomes.

We continue to find, in our more recent work(B. M. Taylor et al., 2003) with low-income, low-performing, aspiring schools, that things are notalways what they appear to be—that there isincredible variability among our interventionschools in the degree to which the intervention isactually implemented, both across schools andacross classrooms within schools. We also findthat the degree of fidelity to the intervention, notto a set of specific instructional practices but toa set of broad principles outlining the process tobe followed and the issues to be addressed, is agood predictor of achievement growth, againboth across schools and across classrooms withinschools. My point is simple—no matter how wellplanned an intervention might be, things happen


and variation will occur. In many studies, thevariation within treatments is often equal to thevariation between treatments. Thus, it is criticalin research involving programmatic and instruc-tional reform, to document carefully the natureof the actual practices across schools and class-rooms. And there is no better tool to do this thanethnographic descriptions of classroom instruc-tion and professional development meetings. Inshort, we need all the tools we can muster toaddress the inherent complexity of researchinvolving human beings who live and work ingroups.

The final question about randomized field tri-als is whether we will be willing to pay the pricetag. It is one thing to randomly assign collegefreshmen who happen to have the misfortune tobe enrolled in Psychology 1-A to different treat-ments. It is quite another to randomly assignteachers, classrooms, and even schools to a par-ticular treatment. In the psychology class, I test30 subjects and I get 30 data points for my analy-sis. In the classroom, I test 30 students and getone data point—the classroom mean. That is onecost factor. But there are others: For example, ifwe want to know if the treatment generalizesacross types of students and types of schools,then we will either have to draw very large sam-ples or very carefully shaped selective samples.

The Treacherous RoadFrom Research to Policy

The road from research to policy is fraughtwith many dangers—potholes, blind corners,road hogs, and detours that can frustrate even themost thoughtful traveler. Both researchers andpolicy makers must be aware of these threats asthey do their best to draw valid inferences fromresearch for practice and policy. Let me unpacksome of the dangers and some guidelines for min-imizing risk to students, parents, and teachers.

When research travels to the land of policy,often only the headlines make the journey, leavingthe details and the nuance behind. The conse-quences of this fact of policy life are depicted withreal examples of the discrepancy in Table 1.1.

I am not sure the journalists are to blame; thereporting of educational research probably doesnot differ much from the reporting of medical,pharmaceutical, agricultural, or public healthresearch. Lest we think that education is differ-ent, compare these two headlines, one com-posed by a staff writer of the New York Timesand the other by a staff writer of theWashington Post in reporting the findings froman article released by the New England Journalof Medicine about the relative effectiveness ofsurgery versus benign neglect in treatingprostate cancer in men:20

From the New York Times on September 12, 2002: “ProstateCancer Surgery Found to Cut Death Risk” (Kolata,2002)

From theWashington Post on September 12, 2002: “ProstateCancer Therapies About Equal” (D. Brown, 2002)

One glass is half full; the other half empty. Aperson contemplating surgery would muchrather be reading the Times! The point is that asa society we must find a way to cope with thepersistent problem of interpretation that tendstoward oversimplification, whether it occurs inthe press, the Congress, or our statehouses.Perhaps we should require that policy makers(or members of their staffs) be required to readbeyond the headlines of educational reportsbefore setting policy in concrete. Nuance maynot make things simple, but nuance is a fact oflife in most policy contexts, including publichealth and medicine.

Research is often used in a selective, uneven, andopportunistic manner by policy makers. Anunfortunate corollary of this surface-levelapproach to summarizing research for policypurposes is the uneven use of the research card insetting policy. The danger is that those who setpolicy will choose to play that card when the evi-dence swings in their favor; and when it does not,they will appeal to common sense, the conven-tional wisdom of practice, or authoritative opin-ion. So, for example, when the NRP reportblessed the systematic teaching of phonics,


basic-skills advocates were quick to point to thescientific evidence underlying their policy initia-tives. But the use of decodable text in conjunc-tion with those programs, which could not bejustified by the available evidence, was rational-ized as a commonsense adjunct to a systematicapproach to teaching phonics. In a similar vein,the Los Angeles Unified School District has

mandated that all high schools use a remedialprogram with the ironic title of Language for allof its low-performing secondary students.Language is a decodable text (“Dan can fanNan”) throwback to the linguistic readers of themid-1960s. The noteworthy aspect of its adop-tion is it was adopted not by appealing to the evi-dence (the NRP could not document the use of


Headline Source Digging Deeper Into the Actual Report

Systematic, Explicit,Synthetic PhonicsImproves ReadingAchievement

Foorman, Francis, Fletcher,Schatschneider, and Mehta (1998)

When a program includes systematic, syntheticphonics among many other elements (lots ofwriting, lots of reading of a whole range oftexts, and lots of supplementary activities), asmall but robust effect for a subset of thepopulation is found on a measure that requireskids to read lists of pseudowords.

Phonemic AwarenessImproves Later ReadingAchievement

National Reading Panel (2000) Phonemic awareness helps . . .• If taught early (K–1);• Mostly on measures of word

identification;• If taught with letter-sound instruction;• If limited in scope (from 18 to 20 hours).

Phonics Wins National Reading Panel (2000) Phonics helps . . .• If it is taught early (not great beyond

Grade 1);• More on word recognition than

comprehension;• If it is systematic and explicit (no

evidence for one approach over another);• If it is embedded in a rich curriculum;• If caveats are recognized, for example,

that there is no evidence for decodabletext.

Independent ReadingDoes Not Help—If YouWant to Do It, Assign Itas Homework

National Reading Panel (2000) The National Reading Panel did not studyindependent reading but rather the impact onfluency of instructional interventions designedto increase the amount of independent readingdone in classrooms. From the paltry array ofstudies they were able to assemble, theyconcluded that the research on the efficacy ofsuch interventions was inconclusive.

Table 1.1 Headlines Versus Details in the Reporting of Reading Research

phonics for readers in that age range) but byappealing to common sense (these kids clearlymissed out on phonics the first time around solet’s go back to square one and do it right).Allington andWoodside-Jiron (1998) have docu-mented similar enactments of selective attentionto research in noting the widespread adoption ofdecodable texts in state textbook standards.

Some science is more important than otherscience. Another corollary of this uneven use ofresearch is a kind of first among equals conspir-acy of good intentions. And it applies to the useof the NRP report in setting policy. The chapterson comprehension and vocabulary in the NRPare laudatory in their praise for the work in theseareas (although they eschewed meta-analysis infavor of a best evidence synthesis on the groundsof too few studies) and enthusiastic in recom-mendations for renewed attention to strategyinstruction and ambitious vocabulary teaching.Moreover, the even more recent Rand report(Snow, 2002) advocates a renaissance in researchin comprehension instruction and assessment.But I have not witnessed a groundswell in advo-cacy for comprehension and vocabulary instruc-tion as the fundamental solution to America’sliteracy problems. There seems to be a kind of“first things first” ethic, suggesting that “ofcourse we’ll get to comprehension and vocabu-lary . . . , but first let’s make sure we have thebasics in place.” The same could be said for thesection on teacher education and professionaldevelopment in the comprehension chapter ofNRP; we get glowing recommendations for theefficacy of professional-education models toincrease teacher capacity to teach comprehen-sion but little action on the policy and “scientifi-cally based” professional development fronts.One of the other ironies of professional develop-ment in NCLB is that what it means to conductevidence-based professional development is thatthe content of the professional development ses-sions must be based on scientifically based read-ing research about how young children learn toread, but need not attend at all to the substantialbody of research documenting the optimal ways

to promote teacher learning. We know from ahost of studies (see Richardson & Placier, 2002;Wilson & Berne, 1999) that professional learn-ing is at its best when teachers have a voice in itsdesign, when it is long term and school based,when it is focused on analyses of teaching andstudent learning, and most important, when thefocus is on establishing learning-sustained com-munities. Yet NCLB, as committed as it is toscience, is moot on the point of how professionaldevelopment (itself to be based on scientificallybased reading research) ought to be organized ordelivered.

When we do not have definitive research toanswer a question about policy or practice, wecan easily slip over the line and privilege ideol-ogy and belief over evidence. The gold standardin research leading to policy implementation issurely the randomized field trial (see Mosteller& Baruch, 2002, for a series of articles extollingthe virtues and assessing the limitations of therandomized field trial). Other things being equal,it is better to have experimental evidence to sup-port a claim or validate a practice. But what areeducators, especially school- and district-basededucators, to do when they must establish curric-ular practices or forge new programs without thebenefit of randomized experiments? Do we gostraight from randomized trials to personalbeliefs? Or do we establish a principle thatrequires us to use the best evidence available tous in any situation? If there are no randomizedfield trials, can we rely on the evidence fromquasi experiments? Natural experiments?Correlational research? Best-practice research?They are all quantitative, but they cannot easilyrule out rival hypotheses. How about case stud-ies? Ethnographies? They are both empirical-and data-driven but they have a different set ofevidentiary rules and a different notion of gener-alizability (Firestone, 1987). And when do weresort to professional consensus, the wisdom ofexperience, and personal belief? As a profession,we have not established any sort of hierarchy ofevidentiary sources (Peirce, 1885, as cited inHartshorne, Weiss, & Burks, 1931–1958).


Perhaps it is time we did. It is my personal con-viction that it is our moral and ethical obligationto use the best evidence we can muster for mak-ing policy decisions of consequence. Further,when we have no evidence, we must fess up tothat fact and make it clear that we are basing pol-icy on values, beliefs, and hunches. If we fol-lowed some sort of evidentiary guidelines, wewould not have so many intensive phonics pro-grams for older students, so much decodable textin our commercial programs, or so little time forindependent reading.

The independent reading issue is particularlytroubling because it took a double hit. First, theNRP chose to examine it even though there wereprecious few studies that could pass through theeye of the needle imposed by their standards forinclusion. When they did, they stated their con-clusion in a way that allowed readers to movefrom what they did say, “There is no evidence tosupport the efficacy of school-based programsthat promote independent reading,” to what somewanted to hear, “Independent reading is a wasteof time.”

Second, the NRP did not look at anything butexperiments, thus eliminating some powerful evi-dence documenting the importance of everydayreading. Had the NRP examined a wider array ofresearch, including several experiments con-ducted in other countries for second-languagelearners (e.g., Elley, 1998), a few smaller scaleexperiments (e.g., B. Taylor, Frye, & Maruyama,1990), some impressive naturalistic studies con-ducted in an epidemiological tradition (e.g.,Anderson, Wilson, & Fielding, 1988), and a widearray of best-practice research (e.g., B. M. Taylor,Pearson, et al., 2000; Wharton-MacDonald et al.,1998), they would have reached a very differentconclusion about the efficacy of just plain read-ing. And in a situation in which the experimentalresearch is moot, that would have been the highroad to take on such a key professional issue.

Curricular Policy

Many recent developments suggest that we areretreating to a more familiar, more comfortable

paradigm of basic skills in which phonics, skills,and controlled text dominate our practices. Otherdevelopments suggest that we are on the verge ofa new paradigm, a hybrid that weds some of theprinciples of whole language (integrated instruc-tion and authentic texts and tasks) with some ofthe traditions of earlier eras (explicit attention toskills and strategies, some vocabulary control ofearly readers, and lots of early emphasis on thecode) in an “ecologically balanced” approach toreading instruction.21 The most cynical among usmight even argue that we are just riding the nat-ural swing of a pendulum that will, if we have thepatience, take us back to whole language, orwhatever its pedagogically constructivist, child-centered descendant turns out to be, in a decadeor so. Before making a prediction about thedirection the field will take, let me play out thefirst two scenarios: phonics first and balancedreading instruction.

Two Different Worlds

If those who have advocated most strongly fora return to phonics and a heavy skills orientationhave their way—if they are able to influence fed-eral, state, and local policy as well as the educa-tional publishing industry—we will experienceeven bigger shifts in the very earliest stages oflearning to read—preschool, kindergarten, andGrade 1. They suggest explicit instruction onphonemic awareness and phonics, with a strongpreference for decodable texts in the earlygrades. When it comes to writing, literature,response, and comprehension, the phonics-firstadvocates seem quite content to cede curricularauthority to the practices that emerged during the1980s and early 1990s, those associated withwhole language, literature-based reading, andprocess writing (see Adams & Bruck, 1995;Fletcher & Lyon, 1998). Thus, looking broadly atthe entire elementary reading curriculum (therange of materials and the range of pedagogicalpractices), things might on the surface look sim-ilar to the early 1990s, with some retreat to the1980s, especially in terms of skill and strategyinstruction.


But beneath that curricular surface, majorchanges would have occurred. For example, therole of the teacher and the learner would havereverted to what they were before the ascendancyof constructivist teaching reforms, such as wholelanguage. The role of the teacher would be totransmit the received knowledge of the field, asreflected in research-based curricular mandates,to students. Students would eventually beregarded as active meaning makers, but onlyafter they had received the tools of decodingfrom their teachers. The greatest changes of allwould have taken place in the underlying modelof reading and reading acquisition. The simpleview of reading (that reading comprehension isthe simple product of decoding prowess and lis-tening comprehension) would have returned infull force, and the job of young readers would beto acquire the decoding knowledge they lackwhen they begin to learn to read.

If those who are pushing for ecological bal-ance carry the day, the field will experience lessdramatic shifts. A balanced approach will privi-lege authentic texts and tasks, a heavy emphasison writing, literature, response, and comprehen-sion, but it will also call for an ambitious programof explicit instruction for phonics, word identifi-cation, comprehension, spelling, and writing. Abalanced approach is likely to look like someinstantiations of whole language from the early1990s, but recalibrated to redress the unintendedcurricular consequences outlined earlier in thischapter. Major differences between a balancedapproach and the new phonics are likely to mani-fest themselves most vividly in kindergarten andGrade 1, where a rich set of language and literacyexperiences would provide the context fromwhich teachers would carve out scaffoldedinstructional activities to spotlight necessaryskills and strategies—phonemic awareness, letter-sound knowledge, concepts of print, and concep-tual development. Thus instruction, althoughfocused and explicit, would retain the highly con-textualized patina of whole language.

Beneath the curricular surface, balancedapproaches seem to share slightly more in com-mon, at least on a philosophical plane, with

whole-language than with new-phonics approaches.The teacher is both facilitator and instructor. Theteacher facilitates learning by establishing authenticactivities, intervening where necessary to providethe scaffolding and explicit instruction requiredto help students take the next step toward inde-pendence. The student is, as in whole language,an active meaning maker from day one ofpreschool. Reading is a process of constructingmeaning in response to texts encountered in aspecific context, and the emergent literacymetaphor, not the readiness metaphor, character-izes the acquisition process.

An Ecologically Balanced Approach

Just in case my personal bias has not emerged,let me declare it unequivocally. I favor the con-ceptual map of the ecologically balancedapproach, both for research and curricular policy.

I hope my reasons for supporting ecologicalbalance—or as Howe and Eisenhart (1990) havecharacterized it, compatabalism—in researchmethods are transparent. The problems we faceare too vexing to limit ourselves to a singlemethodology or epistemology. Multiplicity in thetradition of Spiro’s (Spiro & Jehng, 1990) cogni-tive flexibility theory is what is needed now. Wesurely need to know what works, but we alsoneed to know why it works, for whom, and underwhat conditions; interestingly, this sort ofapproach is appealing to many research policyleaders, including those who have led the chargetoward more experimental approaches (see Lyon,2003; Whitehurst, 2001). For example, in testify-ing to Congress, Lyon (1999) expressed just thesort of complementarity I have argued for:

In order to develop the most effective instructionalapproaches and interventions, we must clearly definewhat works, the conditions under which it works, andwhat may not be helpful. This requires a thoughtfulintegration of experimental, quasi-experimental, andqualitative/descriptive methodologies.

To paraphrase Spiro and his colleagues, neithersimplemindedness nor muddleheadedness willserve our interests, or those of the nation, well.


There are several reasons for favoring an eco-logically balanced, or comprehensive, stancetoward curriculum. First, my reading of the read-ing research points to the balanced-curricularposition, not to the new-phonics position or thewhole-language position, and it does so on both atheoretical and a pedagogical plane. I do not seemuch support for the simple view of reading thatunderlies the new phonics; readers do constructmeaning, they do not just find it lying there in thetext. Regarding pedagogical research, my readingrequires me to side with Chall’s (1967) view thatalthough some sort of early, focused, and system-atic emphasis on the code is called for, no partic-ular approach can be singled out. Even the recentreport of the NRP (2000) took exactly that posi-tion.And although I readily accept the findings ofthe phonemic awareness research, I do not readthem as supporting drill and practice approachesto this important linguistic understanding; to thecontrary, highly embedded approaches, such asinvented spelling, are equally as strongly impli-cated in the research (see Clarke, 1988; NRP,2000; Winsor & Pearson, 1992).

Second, an ecologically balanced approach ismore respectful of the entire range of research in

our field. It does not have to exclude majorresearch paradigms or methodological approachesto sustain its integrity.

Third, an ecologically balanced approachalso respects the wisdom of practice. It is noaccident that studies of exemplary teachers,those who are respected by their peers and nur-ture high student achievement, consistentlyfind that they exhibit a balanced repertoire ofinstructional strategies. Teachers who are facedwith the variations in achievement, experience,and aptitude found in today’s classrooms need,and deserve, a full toolbox of pedagogicalpractices.

Finally, an ecologically balanced approachrespects our professional history. It retains thepractices that have proved useful from each erabut transforms and extends them, rendering themmore effective, more useful, and more supportiveof teachers and students. And it may representour only alternative to the pendulum-swing viewof our pedagogical history that seems to haveplagued the field of reading for most of the 20thcentury. A transformative rather than a cyclicalview of progress would be a nice start for a newcentury.


P. David Pearson is professor and dean within the Graduate School of Education at the University of California,Berkeley, where he pursues a program of scholarship in reading pedagogy, assessment, and policy analysis.

Article 1€¦ · Article 1 THE READING WARS 1 P.D AVID PEARSON Thisarticle’sfundamentalargumentisthatreadinginstructionandreadingresearchhavebeen ...

Documents