learning to read and write genre-specific text: roles of...

39
Reading Research Quarterly Vol. 42, No. 1 January/February/March 2007 © 2007 International Reading Association (pp. 8–45) doi:10.1598/RRQ.42.1.1 T Learning to read and write genre-specific text: Roles of authentic experience and explicit teaching VICTORIA PURCELL-GATES University of British Columbia, Vancouver, Canada NELL K. DUKE Michigan State University, East Lansing, USA JOSEPH A. MARTINEAU State of Michigan K–12 Assessment Programs, East Lansing, USA his study addresses the long-held debate regarding how language is best learned, particularly language forms that are not acquired as one’s primary discourse (Gee, 1992), such as reading and writing. On one end of a continuum of belief are those who hold firm to the notion that language cannot be taught and can be acquired only through experience by participating in situated use of particular language forms. At the other end of the belief continuum are those who assert that language is best learned through tutelage and explicit instruction in its structures and forms. As is often the case with continua anchored by extremes, many theorists, acade- mics, and teachers hold a moderated middle ground: Language is best learned through a combination (so far, unspecified) of experience and explicit instruction. The question becomes what combination of experience and explicit instruction best facilitates learning of new language forms. This question has been addressed in a great deal of writing and research re- garding decoding of individual words, with explicit teaching being shown to be more effective than immersion alone (e.g., National Institute of Child Health and Human Development [NICHD], 2000), but it has rarely been addressed with 8

Upload: others

Post on 19-Feb-2021

0 views

Category:

Documents


0 download

TRANSCRIPT

  • Reading Research QuarterlyVol. 42, No. 1

    January/February/March 2007© 2007 International Reading Association

    (pp. 8–45)doi:10.1598/RRQ.42.1.1

    T

    Learning to read and writegenre-specific text: Roles of authentic experienceand explicit teachingVICTORIA PURCELL-GATES University of British Columbia, Vancouver, Canada

    NELL K. DUKEMichigan State University, East Lansing, USA

    JOSEPH A. MARTINEAUState of Michigan K–12 Assessment Programs, East Lansing, USA

    his study addresses the long-held debate regarding how language is best learned,particularly language forms that are not acquired as one’s primary discourse (Gee,1992), such as reading and writing. On one end of a continuum of belief are thosewho hold firm to the notion that language cannot be taught and can be acquiredonly through experience by participating in situated use of particular languageforms. At the other end of the belief continuum are those who assert that languageis best learned through tutelage and explicit instruction in its structures and forms.As is often the case with continua anchored by extremes, many theorists, acade-mics, and teachers hold a moderated middle ground: Language is best learnedthrough a combination (so far, unspecified) of experience and explicit instruction.The question becomes what combination of experience and explicit instructionbest facilitates learning of new language forms.

    This question has been addressed in a great deal of writing and research re-garding decoding of individual words, with explicit teaching being shown to bemore effective than immersion alone (e.g., National Institute of Child Health andHuman Development [NICHD], 2000), but it has rarely been addressed with

    8

  • 9

    THIS STUDY explored, with both experimental and correlational designs, the roles of (a) authentic, communica-tively functional reading and writing and (b) the explicit explanation of genre function and features on growth ingenre-specific reading and writing abilities of children in grades two and three. The genres used for this explo-ration were informational and procedural science texts. Sixteen grade 2 classes participated, 10 of which were fol-lowed through grade 3, (N = 420), in one of two conditions: (a) authentic reading/writing of science information-al and procedural texts or (b) authentic reading and writing of these genres with the addition of explicit explanationof language features typical of each. Growth was modeled across six assessment time points using Hierarchical LinearModeling. Results showed no effect of explicit teaching on reading and writing growth for six of seven outcomes.Similarly, correlational analyses showed no relationship between teachers’ degree of explicitness and growth for sixof seven measures. However, correlational analyses showed a strong relationship between degree of authenticity ofreading and writing activities during science instruction and growth for four of seven outcomes, with an interac-tion with degree of explicitness for a fifth. Children from homes with lower levels of parental education grew atthe same rate as those from homes with higher levels, and findings regarding explicitness and authenticity also didnot differ by level of education. These results add to the growing empirical evidence regarding the efficacy of in-volving students in reading and writing for real-life purposes in the classroom. They also contribute to a growingknowledge base regarding the complexities of language learning in school.

    Learning to readand write genre-specific text:Roles ofauthenticexperience andexplicit teaching

    MEDIANTE UN diseño experimental y correlacional, este estudio exploró los roles de: a) la lectura y la escrituraauténticas y comunicativamente funcionales y b) la explicación de la función y los rasgos del género sobre el crec-imiento en las habilidades de lectura y escritura de textos de géneros específicos en niños de segundo y tercer gra-do. Los géneros utilizados fueron el informativo y procedimental en textos científicos. Participaron dieciséis clasesde 2° grado, diez de las cuales fueron seguidas en 3° grado (N = 420), en una de dos condiciones: a) lectura y escrituraauténticas de textos científicos informativos y procedimentales o b) lectura y escritura auténticas de estos géneroscon el agregado de explicaciones explícitas sobre los rasgos lingüísticos típicos de cada uno de ellos. El crecimientose formalizó tomando seis puntos de evaluación en el tiempo mediante el uso de un Modelo Lineal Jerárquico. Nose observaron efectos de la enseñanza explícita sobre el crecimiento en lectura y escritura en seis de siete resultados.De forma similar, los análisis correlacionales no mostraron relación entre el grado de explicitud de los docentes y elcrecimiento en seis de siete medidas. Sin embargo, los análisis correlacionales mostraron una fuerte relación entreel grado de autenticidad de las actividades de lectura y escritura durante la enseñanza de la ciencia y el crecimientoen cuatro de siete resultados, con una interacción con grado de explicitud en un quinto. Los niños de hogares conbajos niveles de educación de los padres crecieron al mismo ritmo que aquellos de hogares con niveles más altos. Loshallazgos referentes a la explicitud y la autenticidad tampoco difirieron según el nivel educativo. Estos resultados sesuman a la creciente evidencia empírica acerca de la eficacia de involucrar a los estudiantes en actividades de lec-tura y escritura que tengan objetivos en la vida real. También constituyen un aporte a la base de conocimientos so-bre las complejidades del aprendizaje del lenguaje en la escuela.

    Aprender a leer yescribir textos degénerosespecíficos: Losroles de laauténticaexperiencia y laenseñanzaexplícita

    DIESE STUDIE untersuchte mit Hilfe von beiden, experimentellen und korrelationalen Designs, die Rollen von(a) authentischem, kommunikativ-funktionalem Lesen und Schreiben und (b) die explizite Darlegung vonGenrefunktion und Eigenschaften übers Wachstum in genrespezifischen Lese-und Schreibfähigkeiten von Kindernin der zweiten und dritten Klasse. Die für diese Untersuchung benutzten Genres waren informatorische undprozedurale Naturkundetexte. Sechzehn zweite Klassen partizipierten, davon wurden zehn bis zur dritten Klassebeobachtet (N = 420); in einer von zwei Konditionen: (a) authentisches Lesen/Schreiben von informativen undprozeduralen Naturkundetexten oder (b) authentisches Lesen und Schreiben dieser Genres unter Hinzufügung einerausführlichen Erklärung von jeweils typischen Spracheigenschaften. Steigerungen wurden quer durch sechsBewertungspunkte unter Nutzung des hierarchischen linearen Modells geformt. Die Ergebnisse zeigten keineWirkung auf Steigerungen im Lesen und Schreiben durch explizites Unterrichten bei sechs von sieben Ergebnissen.Ähnlich zeigten korrelative Analysen keine Beziehung zwischen dem Grad der expliziten Ausführlichkeit der Lehrerund der Steigerung bei sechs von sieben Messungen. Dennoch, korrelative Analysen zeigten eine starke Verbindungzwischen dem Grad der Authentizität von Lese- und Schreibaktivitäten während des Naturkundeunterrichts undSteigerungen in vier von sieben Ergebnissen, bei einer Wechselwirkung mit einem gewissen Grad an Klarheit beieinem fünften Ergebnis. Kinder aus Familien mit geringeren Bildungsstufen der Eltern steigerten sich in gleichemMaße wie jene aus Familien mit höherem Niveau, und die Ergebnisse bezüglich Ausdruck und Authentizität wichenebenfalls nicht vom Grad der Ausbildung ab. Diese Ergebnisse tragen zum wachsenden empirischen Beweisbezüglich der Effizienz der teilnehmenden Schüler im Lesen und Schreiben von wirklichkeitsnahen Aufgaben imKlassenraum bei. Sie tragen auch zu einer wachsenden Erkenntnisbasis über die Komplexität des Erlernens derSprache in der Schule bei.

    Lernen des Lesensund Schreibensvon genre-spezifischem Text:RollenauthentischerErfahrung undexplizitenUnterrichtens

    ABSTRACTS

  • 10

    CETTE ÉTUDE a exploré, à l’aide de plans de recherche expérimentaux et de corrrélations, les rôles a) d’une lectureet d’une écriture authentiques et fonctionnelles en termes de communication, et b) d’une présentation explicite des car-actéristiques et des fonctions du genre sur les progrès dans les compétences des enfants de seconde et troisième année àlire et à écrire dans un genre particulier. Les genres utilisés pour cette étude exploratoire étaient des textes informatifset procéduraux de science. Seize classes ont participé à la recherche, dont 10 ont été également suivies au long de la 3°année (N = 420), dans l’une des deux conditions suivantes : a) lecture/écriture authentiques de textes informatifs etprocéduraux, ou b) lecture/écriture authentiques de ces genres avec en plus présentation explicite de caractéristiques delangage propres à chacun d’eux. Les progrès ont été modélisés au moyen d’une échelle d’évaluation temporelle en sixpoints en utilisant un Modèle hiérarchique linéaire. Les résultats ne montrent aucun effet de l’enseignement explicite surles progrès en lecture et en écriture pour six des sept indicateurs. De la même façon, les analyses de corrélation n’ontpas montré de relation entre le degré d’explicitation des enseignants et les progrès dans six des sept mesures. Toutefois,les anlyses de corrélation ont montré une forte relation entre le degré d’authenticité des activités de lecture et d’écriturependant l’enseignement scientifique et les progrès pour quatre des sept indicateurs, ainsi qu’une interaction avec le de-gré d’explicitation pour le cinquième. Les enfants provenant de familles dont les parents avaient un faible niveau d’é-tudes ont progressé autant que ceux provenant de familles avaient un niveau d’études élevé, et les résultats relatifs à l’-explicitation et à l’authenticité ne diffèrent pas non plus en fonction du niveau d’études. Ces résultats s’ajoutent au corpscroissant de preuves empiriques relatives à l’efficacité d’une pédagogie impliquant les élèves dans des activités de lec-ture et d’écriture ayant des buts réels dans la vie. Ils contribuent aussi au développement d’une base de connaissancescroissante relative à la complexité de l’apprentissage du langage à l’école.

    Apprendre à lireet à écrire destextes d’un genrespécifique : lesrôles del’expérienceauthentique et del’enseignementexplicite

    Что влияет на становление умений читать и создавать тексты определенного жанра умладших школьников? Для экспериментального и корреляционного исследования(a) умений в области аутентичного, выполняющего коммуникативную функциючтения и письма, и (б) роли, которую играет учительское объяснение в становленииэтих умений, были выбраны два типа научного текста: информационный ипроцедурный. В исследовании участвовали 16 вторых классов, 10 из которыхобследовались и на следующий год (N=420). В одних классах исследовалисьсобственно умения детей читать и писать информационные и процедурные тексты. Вдругих классах развитие этих умений подкреплялось работой учителей, объяснявшихтипичную для каждого жанра терминологию, и исследовалась зависимость междуобретенными умениями и качеством объяснений учителя. Динамика отслеживалась вшести временных точках с использованием Иерархического линейногомоделирования. По шести параметрам (из семи учитывавшихся) результаты невыявили влияния учительского вмешательства на становление умений читать и писатьжанровый текст. Корреляционное исследование также не выявило взаимосвязи междукачеством и подробностью учительских объяснений и становлением шести из семипараметров. Однако корреляционное исследование показало, что если действияучащихся аутентичны и преследуют жизненно важные цели, это напрямую влияет нарост четырех из семи параметров, при этом рост пятого параметра также связан скачеством объяснений учителя. Дети из семей с более низким уровнем образованияродителей, продемонстрировали ту же динамику, что и дети из более образованныхсемей; объяснения учителя также никак не коррелировали с уровнем образованияродителей. Полученные результаты подтверждают накопленные эмпирическиесвидетельства об эффективности использования в учебных целях аутентичногочтения и письма, связанного с реальной жизнью. Они также вносят вклад в растущуюбазу наших знаний о сложных процессах освоения языка в школе

    ABSTRACTS

    Роль опыта иобучения причтении инаписаниитекстовопределенногожанра

  • respect to larger units of discourse, which constitutethe bulk of what one actually reads and writes as aliterate person. One notable exception lies in teach-ing text structure; on the whole, research in this areahas indicated that explicit teaching (in this case ex-plicit teaching of particular text structures) can im-prove comprehension and composition of thosestructures (e.g., Dickson, Simmons, & Kame’enui,1998a, 1998b; Gersten, Fuchs, Williams, & Baker,2001; NICHD).

    However, aside from the text-structure research,positions taken on the authentic literacy activity/explicit teaching continuum for learning of larger unitsof discourse are based, for the most part, on theoreticalrather than empirical evidence. We designed this study,therefore, to provide empirical data to inform these is-sues regarding the teaching and learning of written dis-course in school. We were ourselves informed andguided by the recent work in genre studies, selectingthe specific genres of science informational and proce-dural written texts as the contexts for our exploration.We further specified our exploration by focusing onprimary-age children (grades 2 and 3), a developmen-tal level rarely included in studies of genre learning ofcontent area written discourse.

    Theoretical frameWe situate this study within a theory of lan-

    guage use and development that views language as in-extricably woven into social action, or practice(Barton, Hamilton, & Ivanic̆, 2000; Gee, 1992;Street, 1984; Vygotsky, 1934/1962). The essentialcharacteristic of language as social reflects aBakhtinian view of language as dialogic, communica-tive, and socially constructed (Bakhtin, 1981, 1986).

    Texts and genre theoryWithin this larger frame of language as social,

    dialogic, communicative, and constructed, Halliday’stheory of social semiotics helps us see language formsthrough a genre lens (Halliday & Hasan, 1985).Genres, in the instance of written language, are dif-ferentiated and identifiable written text types. Thesocial semiotic view of language considers genres associally constructed language practices, reflectingcommunity norms and expectations. These normsare not static but change to reflect changing socio-cultural needs and contexts (Berkenkotter &Huckin, 1995; Martin, Christie, & Rothery, 1987).Thus, genres are identifiable linguistic forms that aredynamic and fluid, and members of flexible and ac-

    tive communities of practice construct genres toserve social purposes that are situated within socio-cultural contexts. Written language forms serve func-tions that require written, rather than oral, purposesand contexts. The forms, themselves, reflect theseconstraints, purposes, and contexts (Chafe &Danielewicz, 1986; Purcell-Gates, 1988, 1995). Inthis way, written genre function always drives writtengenre form. We develop this further in our discus-sion regarding the variable of authentic literacy activi-ty in a later section of this article.

    Our theoretical lens for language development,or learning, fits within this view of language as essen-tially social. We situate this study within a construc-tivist perspective (Wells, 1994) that is expanded andspecified to place cognition and learning within so-cial worlds (Gee, 1992; Purcell-Gates, Jacobson, &Degener, 2004). Learners within social contexts oflanguage usage construct language to serve the func-tions called for within those contexts.

    Rationale and review of the literature

    The educational discussion (or debates in somecases) regarding the relative merits or roles of situatedexperience and explicit teaching with respect to learn-ing specific forms of academic discourse have yet tobe empirically resolved. Different instructional ap-proaches over the years have forefronted one or theother. Those approaches reflect the relative ends ofthe continuum and operate out of different theoreti-cal lenses, ranging from literacy as social practice toliteracy as cognitive skill (Purcell-Gates et al., 2004).From a social constructivist perspective, the wholelanguage and process movements (Calkins, 1986;Goodman, 1986) emphasized experience and immer-sion over explicit teaching and incrementalism inwritten language learning, with explicit instructioneither downplayed or discouraged. This led to claimsthat students were expected to intuit the functions,features, and forms of academic discourse solely orprimarily through exposure, with little explicit guid-ance from teachers (Chall, 1996; Delpit, 1986, 1988,1992; Stotsky, 1992). Critiques also arose claiming anoverreliance on the personal: personal responses to lit-erature, for example, with much personal writing be-lieved to allow students to find and establish theirindividual voices as writers (Atwell, 1987; Elbow,1973; Murray, 1991). This was accompanied, accord-ing to critics, by a problematic decrease in the class-

    Learning to read and write genre-specific text 11

  • room of more depersonalized expository “academic”writing (Stotsky, 1992; Venezky, 1982).

    Recently, reflecting a cognitive lens, education-al policy and practice have swung back toward ex-plicit teaching with explicit or, at times, directinstruction in language forms (Carnine, Silbert, &Kame’enui, 1990; Moats, 1994; Moats & Foorman,1997; Moats & Lyon, 1996; Slavin, Madden, Dolan,& Wasik, 1996), relying primarily on the research ondecoding and comprehension strategies. Other ap-proaches have struggled to construct different blendsof the two to create what is sometimes referred to asbalanced approaches to learning to read and write inschool (McIntyre & Pressley, 1996; Pressley, 2005).

    This pedagogical swing just referred to reflectsprimarily the U.S. educational landscape. However,similar struggles have, and are, occurring in othercountries such as the United Kingdom, Canada,Australia, and New Zealand. This is reflected in thearea of genre studies that began in Australia, influencedby Hallidayan linguistic theory. Much of this work hasresulted in classroom materials in use throughoutsoutheast Asia that explicitly teach students languagefeatures of a number of different genres. This approachhas been critiqued as overly prescriptive and reflectinga static vision of genre (Freedman & Medway, 1994)as well as ignoring, or glossing over, issues of cultureand power, despite its lens of genre as social and cultur-al phenomena (Luke, 1994).

    The genreists respond that today’s schools areturning out students ill-prepared to do any type ofreading or writing beyond the personal narrative andthat this is especially true for those learners who donot have access to the more academic and expositorylanguage of highly literate homes (Christie, 1987;Martin et al., 1987).

    The genreists joined with other educationaltheorists and psycholinguists in their claim thatwhereas one’s own “home,” language appears to belearned deductively through experience and immer-sion, this is not enough for secondary discourses, in-cluding those of print (Cazden, 1993). Many hotlycontested this position, citing fear that it would leadto a return to the “bad old days” of skill and drill,teaching empty forms and structures that have noappreciable effect on children’s literacy achievement(Dixon, 1987; Sawyer & Watson, 1987). Christieand others associated with the Hallidayan perspec-tive (Martin et al., 1987) denied this intent; on thecontrary, they asserted, genre knowledge is itself theresult of socially situated language practices, reflect-ing community norms and expectations. Thesenorms are not static but change to reflect changingsociocognitive needs and contexts (Berkenkotter &

    Huckin, 1995; Martin et al., 1987). That these so-cially constructed norms require explicit teaching orpointing out, reflects their status as secondary dis-courses, secondary to one’s primary oral language.

    As noted, many responses to the genreists’ po-sition center on issues of power and agency. Some ar-gue that the teaching of the forms and structures ofthe different genres will (a) reify stilted academicforms that should be challenged or (b) deny studentschoice in how they wish to learn and to communi-cate their knowledge and meanings in writing (e.g.,Freedman, 1994; Green, 1987). Genreists reply thatto not teach the forms and structures of the differentacademic genres (which are socially constructed andchanging, not dictated and static) actually deniesstudents choice in that it limits their knowledge oflanguage forms appropriate to different given situa-tions (e.g., Bazerman, 1992; Kress, 1987).

    Though easily obscured by contention over therole of explicit teaching in genre learning, an impor-tant consensus lurks in these debates. Within the lit-eracy field, both proponents and opponents ofexplicit teaching of genre and discourse (languageunits larger than single words—the focus of decodingresearch) agree that authentic reading and writing ac-tivities are essential to genre learning. This is evidentnot only in the classic debate between genreists andopponents in Australia (e.g., see Reid, 1987). In theU.S. context, Lemke (1994) identifies as common toboth genreists and process learning advocates the po-sition that language “is always language-in-use, thatevery text has a context of situation, with a history ina culture, as part of a social activity or practice,” andthat “texts whose only context is an arbitrary curricu-lum task are inferior as learning contexts to texts...that have a wider social context” (p. 11). Similarly,both proponents and opponents of explicit teachingof discourse agree that authentic reading and writingactivity is critical. Delpit (1988), a staunch advocateof explicit teaching of discourse, argues that

    merely adopting direct instruction is not the answer. Actualwriting for real audiences and real purposes is a vital ele-ment in helping students to understand that they have animportant voice in their own learning processes. (p. 289)

    The New London Group (1996), upon reviewingthe positions of a range of scholars, concluded,

    Recent work in cognitive science, social cognition, and so-ciocultural approaches to language and literacy...argues thatif one of our pedagogical goals is a degree of mastery in prac-tice, then immersion in a community of learning engagedin authentic versions of such practice is necessary. (p. 84)

    12 Reading Research Quarterly JANUARY/FEBRUARY/MARCH 2007 42/1

  • Learning to read and write genre-specific text 13

    Thus, a critical mass of scholars from a rangeof theoretical and pedagogical orientations agree thatauthentic literacy activity is essential to genre anddiscourse learning. However, surprisingly little actualresearch has been conducted on authentic literacy ac-tivity, nor is there even a well-established meaning ofthis variable.

    Situating ourselves within this perspective ofauthentic activity, we designed a study that exploredthe impact of explicit instruction, or explicit explana-tion, of language features on genre learning withincontexts of authentic reading and writing activity. Weexpected natural variation among teachers in the ex-tent to which they provided authentic literacy experi-ences, as well as explicit instruction of genre features,and, in accounting for this variation, we were alsoable to examine with correlational analyses the rela-tionship between degree of authenticity of literacy ac-tivities or degree of explicit teaching of genre featuresand children’s literacy learning. Thus we addressedthe following two research questions in this study:

    1. What is the impact of explicit teaching of genre features ofinformational and procedural text in science on children’sability to read and write these texts? Does this impact dif-fer for children from homes of different parental educa-tion levels?

    2. Is the degree of authenticity of literacy activities, or the de-gree of explicit teaching of language features, with thesetexts related to children’s growth in the ability to readand write them? Does this impact differ for children fromhomes of different parental education levels?

    Why genre-specific reading and writing?Before we begin to describe the procedures of

    this research, we wish to briefly address questionsreaders may have regarding our use of the term genre-specific, throughout this report as well as our use ofscience informational texts and science procedural texts.Regarding our inclusion of the term genre to specifythe texts in general, we reveal our theoretical stancethat language, in this case written language, is em-bodied as sociocultural forms of practice. This meansthat when studying reading and writing, one is by ne-cessity studying reading and writing of some languageform, or genre, whether the genre of spelling lists,short stories, plays, mysteries, personal essays, or as-sessment items. As we discussed, genre forms reflectgenre functions; as functions of textual forms shift,even slightly, so will their forms. We, therefore, find itmore useful, clear, and specific to specify texts read orwritten by learners according to their genres ratherthan the more inclusive text or written text. Further,as researchers, we prefer to specify as much as is possi-

    ble and feasible the genres that are used in our re-search and of which we are studying children’s acqui-sition of reading and writing competency. With this,we hope to forestall the possibility that the researchresults will be overgeneralized or simplistically inter-preted. This is our rationale for specifying the genresof science informational text and science procedural textrather than the more general science text or informa-tional text and procedural text.

    Related to this discussion of genre is the issueof the nature of the relationship between genre formand purpose and content. Whereas we absolutely as-cribe to the theoretical tenet that language form andcontent are inextricably linked (Lakoff, 2000;Slobin, 1979, 1994), we forefronted language formover the content of science for this study. We ac-counted for the constraints of context on genrelearning (i.e., the situated nature of genre learning)by limiting our research into genre learning to thecontent of science. Our data collection and assess-ments, however, focus exclusively on genre issues.

    MethodTo address our research questions regarding the

    roles of authentic literacy activity and explicit teach-ing of genre features, we designed a longitudinalexperimental (at the classroom level) or quasi-experi-mental (at the student level) study with correlationalcomponents. To address the first question, we ran-domly assigned classes to receive instruction eitherincorporating or not incorporating explicit teachingof features of target genres—science informationaland science procedural texts—over a one- to two-year period. We chose informational and proceduralgenres because, in our knowledge, experience, andobservation, these are two of the most active genres,not only within science instruction but within thescience field as a whole. We also chose these twogenres because second- and third-grade children areunlikely to have had extensive previous exposure tothese genres to the degree they might with, for exam-ple, fictional narrative text in language arts (Duke,2000; Duke, Bennett-Armistead, & Roberts, 2003).We assigned all classes to incorporate authentic litera-cy activities in two 45-minute periods a week of theirscience instruction, so that the element of authentici-ty of literacy activities would not differ between thetwo conditions. In both groups, we measured stu-dents’ abilities to read and write science informationaland procedural texts at up to six time points over twoyears (beginning, middle, and end of grade 2 and ofgrade 3). Using growth modeling, we determined

  • growth trajectories for reading and writing of eachtarget genre and compared those trajectories for thegroup with explicit teaching and the group withoutexplicit teaching.

    To address the second research question, wecarefully monitored the degree of authenticity of lit-eracy activities in each classroom as well as the de-gree of explicit teaching. We then examinedchildren’s growth in reading and writing of the targetgenres by the degree of authenticity of literacy activi-ties and the degree of explicit instruction. All analy-ses included the parental levels of education at boththe individual and the average class level to ascertainwhether there were differential impacts of, or rela-tionships with, instruction.

    Operational definitions

    Authentic reading and writing activitiesFor purposes of this study, this term relates

    specifically to the nature of the reading and writingacts, or events, in which students engage while par-ticipating in the activity of schooling. Within thiscontext of literacy instruction, we consider the au-thenticity of a literacy activity according to two di-mensions: (a) text type, or genre, and (b) purpose forreading or writing those texts. For this study, authen-tic literacy activity is defined as (a) reading and writ-ing of textual types, or genres, that occur outside of alearning-to-read-and-write context and purpose, and(b) reading and writing those texts for the purposesfor which they are read or written outside of a learn-ing-to-read-and-write context and purpose. Withthis, we are defining authentic literacy activity bydifferentiating it from reading and writing that isdone with texts that are written primarily to helplearners develop as readers and writers and for pur-poses that reflect this literacy learning function. Werefer to this last activity type with the term school-only to reflect the purpose and contextual constraintsof this type of reading and writing activity in school.

    To judge the authenticity of a literacy eventwithin this frame, we looked at two dimensions of theevent—purpose/function of the reading or writing andthe nature of the text being read or written. Authenticpurpose or function, for us, meant that the literacy eventserves a social communicative purpose, such as readingfor information that one wants or needs to know orwriting to provide information for someone whowants or needs it. Authentic textual forms, or genres(written or read), must be like, or very much like, textsthat are used by readers and writers outside of a learn-ing-to-read-and-write context and purpose to serve the

    communicative, real-life purposes or functions that areaccomplished with those textual forms.

    According to this definition, prototypical au-thentic texts would include such written genres asnovels, news articles, fliers, memos, health proce-dures, greeting cards, and so on—all of the text typesread and written by people as part of their social andcultural lives. Authentic purposes for reading thesedifferent texts would include reading a novel for re-laxation, escape, or entertainment; reading news arti-cles for the latest news about an event of interest;reading fliers to see which music group is appearingin town; reading health procedures to manage one’shealth; and so on. Authentic purposes for writingsuch texts would include writing a newspaper articleto inform a reader about the latest news of an event,composing a health brochure to help a reader followappropriate procedures for managing a disease, com-posing a greeting card (or writing on one) to facili-tate social bonding, and so forth.

    Prototypical school-only texts, as we definedthem for purposes of this research, are genres such asworksheets, spelling lists, short passages with com-prehension questions, flash cards, lists of sentences tobe punctuated, and so forth. School-only purposesfor reading these texts are to learn or improve read-ing and writing skills. School-only purposes for writ-ing these texts would be to assist in the teaching andlearning of literacy skills.

    We conceived of these variables of authenticityand school-only along a continuum, so activities canbe somewhat authentic, highly authentic, and so on.Also, the authenticity of text and purpose can differ.Authentic texts can be read or written with school-only purposes and vice versa, rendering the literacyactivity less authentic (i.e., more school-only) becauseour determination of the degree to which literacy ac-tivity within instructional contexts is authentic isbased on both text type and purpose for reading/writing specific texts. For example, students can readnovels in preparation for an exam on one’s compre-hension and interpretive skills, read news articles forpurposes of identifying new vocabulary words, orcompose fliers for purposes of completing a historyunit with an assignment designed to link to art andlanguage arts—in this case the teacher might post thefliers on the bulletin board as decoration and displayof learning. Conversely, teachers or students can ren-der activities that use texts typically thought of asschool-only, such as content textbooks, more authen-tic if readers read them seeking answers to their ownquestions (due to curiosity or to find information fora friend or relative). Our point of view regardingthese judgments of degree of authenticity was always

    14 Reading Research Quarterly JANUARY/FEBRUARY/MARCH 2007 42/1

  • Learning to read and write genre-specific text 15

    from the perspective (i.e., which purposes would thereader or writer cite for the literacy activity?) of thereaders or writers (i.e., the second- or third-grade stu-dents in our sample classrooms).

    It is important to note that a teacher’s instruc-tional purposes can co-occur with students’ authenticpurposes. We do not see these definitions of authen-tic and school-only literacy activities as mutually ex-clusive. By definition, authentic literacy activitiesoccur, for our purposes, within classroom contexts;therefore, they always co-occur with school-onlyones. Teachers make texts available for students andassign, or facilitate, purposes for reading and writingthose texts within a context of teaching and learning,and, thus, always with the goal of furthering literacydevelopment, even while furthering content learning.Their choice of texts and purposes, though, can addthe dimension of authenticity to their within-schoolliteracy activities for their students. (See Duke,Purcell-Gates, Hall, & Tower, 2006, for a descriptionof ways that teachers can facilitate authentic literacywithin science instructional contexts. Also seeJacobson, Degener, and Purcell-Gates, 2003, forsimilar examples from adult education contexts.)

    Science informational textWe defined science informational text as text

    written for the purpose of conveying informationabout the natural world, typically from one pre-sumed to be more knowledgeable on the subject tosomeone presumed to be less so. To qualify as an au-thentic writing purpose, a real audience, or reader(see next sentence), must be involved. The authenticpurpose for reading science informational text is toacquire information about the natural world thatone needs or wants to acquire for other than school-only purposes (e.g., other than passing a test or an-swering comprehension questions).

    Science procedural textWe defined science procedural text as text that

    is written for the purpose of instructing a reader inhow to conduct investigations or experiments relatedto science content, typically written by someone whoknows how to do the procedure to someone whomust rely on the written procedures to conduct theinvestigation appropriately. To qualify as an authen-tic science procedural text writing purpose, a realreader must be involved—one who will actually fol-low the procedures and enact them. The authenticpurpose for reading science procedural text is toproperly conduct the procedure.

    Explicit instruction in genre featuresOur conceptualization of explicit teaching of

    genre features involves naming, describing, and ex-plaining the function of genres and genre features, inthis case science informational and science proceduralgenres. We considered modeling the use of genre fea-tures as explicit teaching if that modeling was accom-panied by verbalization or thinking aloud. Underlyingthis conceptualization of explicit is the occurrence oftalking or saying. The genre features of each of ourgenres will be listed below (see Tables 2 and 3).

    SampleParticipants in the study were 420 students in

    16 classes, selected at random from school districtsthat met the following criteria: (a) the district waswithin a one-hour drive from our university base, and(b) science was taught by regular second- and third-grade classroom teachers. Within those districts, wefurther limited the pool of eligible schools by includ-ing only those whose demographic informationabout parents on official census reports listed them aswithin the 25th to 75th quartiles in college graduates.This last criterion was to allow us to implement theinterventions within classrooms where a mix ofparental education levels existed in order to addressimplications in the literature that children from fami-lies of lower socioeconomic status (SES) need explicitteaching more than middle class children.

    For this study, we took a theoretical approachto the issue of SES. Socioeconomic status is com-posed of several variables, including income, parentaloccupation, and education. Clearly, these dimen-sions of SES overlap, and we feel it is important tospecify the operational paths between these variablesand literacy learning and achievement. Parental edu-cation seems to us to be the most clearly related toacademic genre learning. We can make this argu-ment in several ways. First, education likely has a di-rect relationship with the variety and types of textualgenres available and engaged with in the homes ofyoung children. This dimension is of clear theoreti-cal relation to our study of genre learning in schoolas research has shown that children learn the formsand functions of the written texts that are engagedwithin their homes (Purcell-Gates, 1996; Snow,Burns, & Griffin, 1998). Second, one can imagineseveral scenarios where more highly educated parentslive in, or close to, poverty (such as ministers andteachers) while the influence of parent education onhome genres still operates. With this line of think-ing, we chose parental education as our SES variable

  • and control. This, however, obscures a pragmaticconstraint to our sample that must be noted herewithin the context of our discussion of how we oper-ationalized SES. Our SES-related research questionswere motivated by the implication in the research lit-erature that children of low-income homes need orbenefit more from explicit teaching of languageforms than do middle class children. In such cases,SES is sometimes used or interpreted as code forrace/ethnicity. We fail to apprehend the logic of suchan argument that children of color require explicitinstruction whereas white children do not. Further,we were unable to test this hypothesis due to thefact, given the geographical location of this study, wehave very few children of color in the sample.

    We do, however, have a representation of thedimension that we believe is most likely to be opera-tive—parental education and its implication for chil-dren’s genre experiences outside of school. The meanproportion of residents with a bachelor’s degree orhigher among districts from which the final samplewas derived was 16.96%. As reported in Table 1, ourfinal sample did include large numbers of childrenwhose mother and/or father did not have a collegeeducation as well as large numbers of children whosemother and/or father did have a college education. Amuch smaller number of children in the sample hadparents with less than a high school or more than a

    four-year college education, representative of nation-al educational statistics. Data regarding parental edu-cation come from the consent forms, on whichparents were asked to indicate their highest level ofeducation obtained.

    In the first year of data collection, we had 11second-grade classes and one split second-grade/third-grade class (due to a last-minute decision bythe principal to make the participating classroom asplit class), each from a different school, from eightschool districts. All of these were classes from schoolsin which (a) one second- and one third-grade teacherhad agreed to participate in the study and (b) theschool and parents had agreed to allow children tomove with their second-grade class as a cohort intothe class of the participating third-grade teacher. Wesought parental permission for participation for allstudents in the classes with the exception of the splitsecond-grade/third grade-class, from which wesought permission only for the second graders.

    In the second year of data collection we fol-lowed 10 of the 12 second-grade classes throughthird grade, with their third-grade teachers partici-pating in the same treatment condition. One class,the split second/third-grade class, was not followedbecause the school was closed; the other class wasnot followed because the third-grade teacher whohad agreed to participate changed grades and an al-

    16 Reading Research Quarterly JANUARY/FEBRUARY/MARCH 2007 42/1

    Variable N M SD

    Female (proportion) 419 0.47 0.50Parental education* 420 2.43 0.55Number of absences (days) 420 6.94 5.84Limited English Proficiency (proportion) 420 0.002 0.049Learning disabled (proportion) 420 0.026 0.16In study both years (proportion) 418 0.50 0.50

    *Where parental education ranges from 0 = less than eighth grade to 4 = graduate degree.

    Participant characteristics per classroomVariable N M SD

    Female students (proportion) 16 0.49 0.10Average parental education* 16 2.38 0.26Average number of absences (days) 16 6.96 1.87Class size, grade 2 (students) 16 23.00 3.50Class size, grade 3 (students) 16 25.80 1.93Number of special ed students, grade 2 16 1.44 1.15Number of Limited English Proficient students, grade 2 16 0.19 0.54Number of special ed students, grade 3 16 1.50 1.11Number of Limited English Proficient students, grade 3 16 0.10 0.25

    *Where parental education ranges from 0 = less than eighth grade to 4 = graduate degree. Note. Ten classes were followed through both second and third grade, with the remaining six being followed only through second grade.

    TABLE 1PARTICIPANT CHARACTERISTICS FOR TOTAL SAMPLE

  • Learning to read and write genre-specific text 17

    ternate participating teacher could not be found.Analyses showed that the second-grade classes in thetwo sites that were not followed through third gradewere not significantly different on any predictors oroutcomes of interest than those that were followed.In addition, we added four new second-grade classesin three school districts to our sample in this secondyear of the study, selecting them through the sameprocedures described, for the purpose of increasingour power to detect second-grade effects.

    Thus, across the two years of data collection wehad data for 16 second-grade classes, with 10 ofthese classes studied through their third-grade year aswell, from a total of 11 school districts. Our overallrate of consent was 99%.

    Teachers participating in the study (16 secondgrade and 10 third grade) agreed to attend summerworkshops (see below), teach science (however theydefined science) at least two times per week for 45 to60 minutes each time, implement the type of in-struction to which they were randomly assigned, al-low one or more researchers into their classroomsduring science on a weekly basis to code their imple-mentation of and to coach them on the instructionalintervention, facilitate assessment of students atthree time points per year, and assist researchers incollecting student/parent consent forms.

    We gave each teacher US$750 for the purchaseof science informational and procedural texts fortheir classrooms. The informational texts were forthe most part books that centered on a particular sci-ence topic or content and were identical, or nearlyidentical, to the types of books one could find foradults or older readers in bookstores and in tradebook form. Teachers may recognize these types ofbooks as part of a more general category of “chil-dren’s literature,” in this case literature that is exposi-tory and focused on science topics. The proceduraltexts came in the form of collections of procedures ina book (e.g., “Fun Experiments”) or as single proce-dures collected, for example, from websites that fo-cus on science for children. Research assistantsreviewed all books before teachers purchased them toensure that they met our definition of authentic sci-ence informational or science procedural texts.Research assistants also informed teachers if textsfrom other sources (e.g., the library) that they wereplanning to use met our definitions of informationalor procedural text.

    InterventionsWe asked the teachers to address the same sci-

    ence curriculum that they would have had they not

    been participating in the study. The science curricu-lum varied from classroom to classroom and schoolto school as is typical in this state. So, for example,whereas some second-grade classrooms do a unit onlife cycles, others do not. One school might addresssimple machines in second grade, whereas anothermight do simple machines in third grade. Similarly,whereas some teachers used commercially preparedscience kits, others developed their own science activ-ities, and still others relied on resources from teachercolleagues. We believe this approach of asking teach-ers to embed the intervention into their existing in-struction had greater ecological validity than if wehad mandated some set of science units and sciencematerials to be used by all teachers in the study. Wealso believe it was more advisable for students whowould otherwise miss some units and repeat othersover time as different schools address units in differ-ent grades, and was methodologically appropriategiven that the focus of this study was genre learning,not science content (see previous discussion).

    As would be expected, given that differentteachers were doing different units, teachers selectedtheir own texts to use although, as noted, researchassistants checked these books for fit with our defini-tion of informational and procedural text. In sum,this was not a study of a prescribed curriculum, butof two changes within teachers’ existing curricu-lum—increased use of authentic literacy activitieswith informational and procedural text in scienceand (in one group of classes) explicit teaching ofgenre functions and features.

    All teachers, in both the authentic-only (A-only) and the authentic-plus-explicit (A+E) groups,attended a total of 15 hours of initial professionaldevelopment in their assigned teaching model in theform of workshops, held during the summer prior totheir participation in the study. Both workshops fo-cused on authentic literacy activities with informa-tional and procedural text in science; one group’sworkshop also focused on explicit teaching of genrefeatures of informational and procedural text.

    During the summer workshops we presentedour definition of authentic literacy events as well asour definitions (by function) of informational andprocedural texts to both groups. We engaged teach-ers in activities designed to help them understandthe variable of authentic literacy and to learn tojudge the degree of authenticity of different texts andreading and writing activities.

    We used the instructional models in Figures 1and 2 as a framework for teachers to think about howto embed their science instruction with authentic lit-eracy. According to this instructional model, authentic

  • science literacy purposes are to stem from authenticscience purposes, which in turn stem from science ex-periences and activities. In this way, science purposesand literacy purposes are interrelated. Notice also thatboth models call for science purposes leading to bothtext and nontext activities or what Palincsar andMagnusson (2000) would call second-hand (throughtext) and first-hand (through actual experiences,“hands-on”) investigations (see also Guthrie, Wigfield,& Perencevich, 2004). That these different kinds of

    investigations both stem from the same overarchingscience purpose demonstrates the embedding of litera-cy within science that we designed for this study.

    We presented teachers with examples of lessonsthat followed from the instructional model (for thatcondition/group) and asked each teacher to generateand share such an example as well. We also provideda handout of suggested strategies for establishing au-thentic science and science literacy purposes forreading and writing information or procedural texts.

    18 Reading Research Quarterly JANUARY/FEBRUARY/MARCH 2007 42/1

    FIGURE 1THE TEACHING MODEL LINKING SCIENCE INSTRUCTION AND LEARNING TO AUTHENTICREADING AND WRITING OF SCIENCE PROCEDURAL AND INSTRUCTIONAL TEXTS WITHOUTEXPLICIT TEACHING OF THE TEXTUAL FEATURES OF THE TWO GENRES

    Science experience/activity SCIENCEPURPOSE

    NONTEXTSCIENCE

    PURPOSES

    SCIENCELITERACYPURPOSES

    Informational texts

    Procedural texts

    FIGURE 2 THE TEACHING MODEL USED WITH TEACHERS IN THE EXPERIMENTAL CONDITION INWHICH AUTHENTIC READING AND WRITING OF SCIENCE PROCEDURAL ANDINFORMATIONAL TEXTS WAS SUPPLEMENTED WITH EXPLICIT TEACHING OF THE TEXTUALFEATURES OF THE TWO GENRES

    Science experience/activity SCIENCEPURPOSE

    NONTEXTSCIENCE

    PURPOSES

    SCIENCELITERACYPURPOSES

    Informational texts

    Procedural texts

    • Genrenames

    • Genrefunction

    • Genrefeature

    • Naming genrefeatures

    • Modeling use ofgenre features

    • Describinggenre features

    • Explainingfunction ofgenre features

    read

    aloud

    share

    d readi

    ng guided reading independent reading

    modeled

    writingshared writing guided writ

    ingind

    epe

    nden

    t writ

    ing

  • Learning to read and write genre-specific text 19

    Explicit teaching of informational andprocedural text features in science

    We also gave teachers in the A+E group train-ing in explicit teaching of informational and proce-dural text features. During the summer workshops,we provided teachers in the A+E group with a list offeatures of science informational text and features ofscience procedural text (see Tables 2 and 3). We de-

    rived these features from a discourse analysis of 25science informational and 25 science proceduraltexts available outside a schooling context and at areading level appropriate for children in second andthird grades (Purcell-Gates & Duke, 2001). We alsodeveloped with the teachers the four ways of beingexplicit in the teaching of genre features: (a) naming,(b) modeling, (c) describing, and (d) explaining the

    Almost always• Has description of attributes/components (e.g., “Ants have six legs.”)• Identifies characteristic events (e.g., “Ants eat sugar.”)• Includes at least one definition• Has compare and contrast (talking about ways things are alike and not alike)• Employs some denotative language (e.g., “Most worms are between 1 and 4 inches long.”)• Uses timeless verb constructions (e.g., “Ants eat sugar.” versus “The ant ate sugar.” or “The ant is eating sugar.”)• Uses generic noun constructions (e.g., “Ants have legs.” versus “That ant has legs.” or “Joe the ant has legs.”)• Has some specialized vocabulary• Has realistic illustrations or photographs

    Often• Has an opening statement/general classification (e.g., “Ants are a kind of insect.”)• Has a general statement/closing • Has headings• Has an index• Includes classification (at either the sentence level—“Bees are a kind of insect.”—or larger units—“Some frogs live in ponds. Some frogs live in

    trees….”)• Has labels and/or captions• Has graphical devices (diagrams, tables, charts, boldface/italicized vocabulary)

    Less often• Has a table of contents

    TABLE 2FEATURES OF SCIENCE INFORMATIONAL TEXTS

    Almost always• Has a statement of goal (sometimes in the title) (e.g., “How to Grow a Plant”)• Has a materials section• Includes methods/procedures/steps• Uses letters or numbers to indicate the order of the steps• Has graphics, and the graphics are almost always demonstrative• Has an explicit, clear description of materials• Has explicit information about procedures (how, when, etc.; e.g., “In two weeks look at your plant” versus “Look at your plant later”)• Uses you, if any personal pronoun• Employs imperative verbs (e.g., “Put…,” “Make….” )

    Often• Has an evaluation of the outcome (e.g., “Now look at your two plants. Which one grew better?”)• Has headings/subcategories• Lists materials in order of use• Has a graphic of the end product (e.g., a grown plant)• Uses units of measure (e.g., centimeters, inches)• Provides a scientific explanation for the results• Indicates the expected results• Has an inquiry question

    Less often• Includes temporal terms (e.g., first, then, next, before, after)• Has the procedure represent something (e.g., a “tornado” in a bottle to represent a real tornado; stacking pillows to represent sedimentary rock)

    TABLE 3FEATURES OF SCIENCE PROCEDURAL TEXTS

  • function of the genre and how the genre featuresserve that function (see Instructional Portraits belowfor an example of how this was instantiated in oneclassroom). We modeled the use of each of thesestrategies for the teachers and they, in turn, present-ed their own lessons using these strategies in thesummer workshops.

    Workshop activitiesTo help teachers understand the text features

    and explicit teaching, we led a number of activitiesduring the workshops. These included practicematching implicit statements to explicit statementsand translating implicit statements to explicit state-ments (for the general notion of explicitness), exam-ining texts in terms of their features, participating ina model lesson that included explicit teaching ofgenre features, and developing their own lesson fromtheir own curriculum that included explicit teachingof genre features within authentic reading and writ-ing. Following the workshops, we gave A+E teachersgrids that had two examples—one for reading andone for writing—of explicit teaching for each featureby each strategy for explicit teaching (naming, mod-eling, describing, and explaining function). Examplesare provided in Table 4. Consonant with our social

    semiotic theoretical frame, we stressed both how thegenre features worked to advance the specific purposeof each genre of focus, and how to talk explicitlyabout that with their students. Finally, we gave teach-ers in both groups the content topics involved in theassessments and the specific names of books used forthe informational reading comprehension assessmentand asked them not to teach those topics or use thosebooks as part of their instruction.

    Overall, the workshop goals were to providethe A-only and A+E teachers with foundational un-derstandings of their respective teaching models andto give them some practice in planning and imple-menting science lessons that incorporated authenticor authentic-with-explicit-teaching-of-genre-featuresreading and writing.

    Ongoing coaching/fidelity to treatmentWe gave teachers in both conditions weekly

    feedback and coaching from research assistants or theinvestigators throughout the duration of the study.The purpose of this coaching was to maintain and/orincrease the authenticity (both groups) or explicitness(one group) of the genre teaching to increase as muchas possible the validity of the results. With few excep-

    20 Reading Research Quarterly JANUARY/FEBRUARY/MARCH 2007 42/1

    Informational text Naming Modeling Describing Explaining function

    TABLE 4EXAMPLES OF EXPLICIT STATEMENTS FOR ONE FEATURE OF INFORMATIONAL TEXT ANDONE FEATURE OF PROCEDURAL TEXT

    Has realistic illustrations or photographs

    Representational art orphotographs within atext.

    “See these photographs?Information text oftenhas photographs.”

    “I’ve never actually seen avolcano in real life, so I’mgoing to look very care-fully at this photo.”

    “These illustrations lookvery much like the plantslook in real life. They’renot like those Monetpaintings we have beenlooking at.”

    “Information books give readers in-formation about their topic withpictures as well as with words.”

    Procedural text Naming Modeling Describing Explaining function

    Employs imperativeverbs (e.g., Put…Make…)

    The verbs are in im-perative form (e.g., no-s inflection, no -inginflection). They tellor command some-one to do something.

    “They’re using the com-mand verbs—put, make,etc.”

    “I’m reading to see whatdo to next. It says ‘Thenmix. . .’ I know that’s acommand verb telling mewhat to do.”

    “One way you can recog-nize a command verb isthat they don’t have end-ings on them like -s or -ing.”

    “The purpose of procedural text isto tell people how to do things, sothe verbs in them are verbs that tellpeople to do—like place, make,stop, stir, and so on.”

    “Notice how realistic I’mbeing with my illustra-tions. This kind of bookusually has realistic illus-trations or photographs.”

    “I’m going to work hardto make my picture lookas much like the realflower as possible.”

    “Our class book is goingto have real photos of ourfish; it won’t look likeSwimmy or something.”

    “It will help your reader learn aboutyour topic if you include picturesthat really show what you are talk-ing about.”

    “What do we call thosekinds of verbs? Right, im-perative verbs.”

    “I’m telling the readerwhat to do next so I’mgoing to use a telling orcommand verb—‘Mix….’”

    “Be sure to use your im-perative verb forms—no -s or -ing.”

    “You’re telling people how to dosomething, so it makes sense thatyou would use telling or commandverbs.”

  • Learning to read and write genre-specific text 21

    tions, each research assistant was assigned to twoclassrooms, one in each condition, to hold constantany research-assistant effects across condition. We vis-ited classrooms individually in both conditions ap-proximately equal numbers of times and attemptedto provide approximately equivalent levels of feed-back and coaching across conditions.

    Instructional portraitsTo aid reader comprehension of this complex

    study, we now provide brief portraits of what the in-struction looked like when implemented in the twoconditions. Each portrait is based on data from oneteacher in each group.

    Authentic-only instructionThe class of second graders settled down noisily into their seats.They had recently returned from a field trip to a nearby naturepreserve as part of their science instruction on pond life, and theywere still excited about the samples of live pond life they hadbrought back to the classroom to add to their growing collection.Ms. Jones (all names are pseudonyms), their teacher, called themto attention with an excited, “Guess what came in the mail thismorning?” They all turned to her eagerly. “Mr. Hernandez wroteus a letter. Let me read it to you.” What followed was the resultof a collaboration between Ms. Jones and Mr. Hernandez, thedirector of the nature preserve and the person who had led thechildren through their tour.

    Dear Boys and Girls, I hope you enjoyed your visit to our pond. Ienjoyed answering your many good questions about what lives inponds. After you left, I thought about all of the other children who vis-it us and who also have many of the same questions. I thought itmight be a good idea to have a brochure for them with answers tosome of their questions. I am writing to ask if you would prepare abrochure like this. It could be called something like, “Questions andAnswers About Pond Life.” You could include some of your questionsthat you had before you visited us. If you write this, I will have manycopies printed that we can put on a stand in the main office. Thatway, people can pick one up when they come or as they are leaving. Ihope you can do this for us. Sincerely, Mr. Hernandez

    Following discussion and a vote, the children began workingin groups. They studied similar brochures collected from muse-ums and other sites of natural science. They brainstormed ques-tions for the brochure (“What would children want to knowabout what lives in the pond?”), after which they researched an-swers by reading from a variety of science texts. Finally, they wrotedrafts of their brochure text until they agreed that it would serveas a useful information brochure for the public. Ms. Jones placedthe brochures in a stand in the front office of the nature park.

    Within this vignette, you can identify elementsof authentic reading and writing activity: the real-lifetext of a question/answer brochure with its socialfunction (providing information) driving its form(note that information text can take different formsand is not exclusively tied to the book format). Thewriting of this brochure was directed at an actual au-

    dience, helping to ensure its authenticity as a writingevent. Other authentic informational texts includedthe different brochures and informational books onpond life read by the children as part of this literacyactivity embedded within science instruction.

    Authentic plus explicit teaching of genre featuresMs. Roberts introduced the day’s science activity by remindingher third-grade class that they would once again be creating writ-ten procedures for the second graders on the other side of theroom divider. The topic was force and motion, and the thirdgraders were busy conducting experiments with tilt boards and avariety of objects with the goal of identifying which objects wouldslide faster and further when tilted. Each group had pads of pa-per and pencils and was simultaneously thinking of how to com-pose the steps of the procedures the groups were conducting sothat the second graders could conduct them also. Ms. Roberts re-minded them of the need for the following elements: a list of ma-terials the second graders would need to do the procedure; thesteps they needed to include so that someone could do it right; theneed to put those steps in order; and so on. After a period of timeof experimentation and drafting of procedures, Ms. Roberts calledthe class members together again. Using her pointer, she drewtheir attention to a large model of a procedural handout with sec-tions for title, materials, steps, and conclusions. She went througheach, calling on individual groups to read what they had writtenfor that section, providing feedback such as, “You say ‘Put theeraser on the board and tilt it.’ But remember that proceduresneed to be very exact and clear. So, how could you write it sothat the person reading it will know WHERE on the board toput the eraser and HOW MUCH to tilt the board?” In the midstof all of this, the second-grade teacher pulled back the dividerand stuck her head into the room, asking for reassurance that thechildren were preparing the procedures for her students and re-minding them how much they enjoyed doing the procedures thethird graders had prepared earlier. “It helps the second graderslearn about how science works!” she emphasized.

    This brief glimpse into Ms. Roberts’s room re-veals the writing of an authentic science proceduraltext for a real audience while the teacher points outand names features of this genre and their functionsin the light of the overall function of a proceduraltext. This writing is embedded within the students’own science learning.

    Space prohibits our providing a more completepicture of the two instructional approaches, and ofcourse the degree of implementation of the ap-proaches—how authentic and how explicit teacherswere—varied by classroom. Suffice it to say thatboth reading and writing took place in both—reading and writing of authentic science informa-tional and science procedural texts for authentic pur-poses and to authentic audiences. The A+E teacherssupplemented this with embedded teaching and ex-plaining of the genre features of whichever genretheir students were reading or writing.

  • Data collection procedures

    AssessmentsWe designed assessments to measure children’s

    ability to comprehend and compose informationaland procedural texts in science as no measures suit-able for growth modeling existed for second- andthird-grade students. In developing the assessments,we addressed the following: (a) the need for equiva-lent forms of the measures to allow for multiple as-sessments over time; (b) the need for reliable andvalid measures of the two genres for learners at thisearly stage of development; (c) the need for measuresthat captured critical features of the intervention—explicit explanation of the linguistic features of thetwo genres—while measuring actual reading andwriting competence; (d) the need to situate the mea-sures within science; and (e) the need for elements ofthe measures that reflected the underlying theoreticalframe of the study: language forms are driven by thecommunicative purposes they serve in people’s lives.An important implication of this last point was thatassessment tasks needed to involve reading or writingauthentic texts for authentic purposes as much aspossible (see previous discussion of authenticity liter-acy activities). We also needed to ensure that thetexts (for reading comprehension) and topics (forwriting) were ones for which students would havehigh background knowledge, so that they were morelikely to be assessments of reading and writingachievement and not of prior or science knowledge(not the focus of this study).

    We piloted all of the assessments for contentand wording. In addition, we piloted and checkedeach item using classical test theory (CTT) measuresof item performance, and an internal consistencymeasure of reliability was calculated for each assess-ment as a whole. We piloted multiple forms of eachtype of assessment: informational reading compre-hension, procedural reading comprehension, infor-mational writing, and procedural writing.

    We then selected and refined three forms ofeach informational assessment (given three times ayear each of the two years) and two forms of eachprocedural assessment (given three times across thetwo years). We also evaluated the extent to which theassessments held up to the psychometric demands ofuse over two years of schooling (ceiling and floor ef-fects) by calculating a CTT internal consistency relia-bility index for each form at each administration. Wealso evaluated interrater reliability for each assessmentwithin the complete context of the study (multipleoccasions, tasks, forms, and raters) using generaliz-

    ability theory (g-theory) to assess the impact of differ-ent raters on students’ scores with different tasks,forms, and occasions (Shavelson & Webb, 1991). Weconducted the theory analyses consistent with the de-sign of the study. Where facets of the design werenested, we represented them as nested in the g-theoryanalyses; where they were crossed, we representedthem as crossed. Where facets should be consideredrandom factors, we considered them random; wherethey should be considered the entire universe of inter-est, we treated them accordingly. We assumed thatoccasions would produce significant variance in stu-dent scores given that we expected students to grow,and that tasks and forms would produce variation be-cause they are expected to vary slightly in difficulty.However, we expected raters to produce little varia-tion in student scores. In general, we found thatraters had very small effects either alone or in interac-tion with any other factor in the design. We, there-fore, determined that the use of item response theory(IRT) was appropriate for creating the final outcomemeasures.

    To create outcome measures amenable to statis-tical analysis, we used IRT to produce student scores.We treated each student’s scored response string ateach occasion as a separate response in a large concur-rent two-parameter logistic IRT calibration of asparse response matrix including all forms, tasks, andoccasions. The analyses showed reasonable fit of bothpersons and items to the specified IRT model. Wealso calculated empirical IRT reliabilities for the en-tire set of student ability estimates using the variancein ability estimates and their pooled standard errorsto estimate the ratio of true score variance (variancein ability estimates minus pooled standard error vari-ance) to observed score variance (variance in abilityestimates). We now discuss each assessment in turn.

    Informational reading comprehensionEach child was handed an entire information

    book, provided by an educational publisher of chil-dren’s information books, that mimicked in formand content information books for children to befound in bookstores. The books were on a sciencetopic and written for children at these grade levelsand were accompanied by a list of 13 to 14 (depend-ing on the form) multiple-choice questions to an-swer. There were three books reproduced andquestions written for each. These constituted thethree forms of the informational reading comprehen-sion assessment. The respective topics of the bookswere (a) animal eggs, (b) peanuts, and (c) insecthabitats. Children had 30 minutes to read the book(range of readability levels 2.5–2.9) and answer the

    22 Reading Research Quarterly JANUARY/FEBRUARY/MARCH 2007 42/1

  • Learning to read and write genre-specific text 23

    questions. The children were focused on answeringthe questions according to what they could learn inthe book, reflecting the purpose of science informa-tional text.

    The questions reflected information that couldbe obtained from the book and where or how this in-formation could be obtained from the book (e.g., onwhat pages particular information could be found, re-quiring the use of the index). We constructed thequestions from discourse analyses that identified theinformation and information levels in the text, using aconceptual map as suggested by Lipson and Wixsonfor assessing comprehension of informational text(Lipson & Wixson, 1991; Wixson & Peters, 1989).The questions also called on a hierarchy of thinking:literal, inferential, and critical. There were three formsof this assessment. To control for the influence of de-coding skill for this comprehension assessment, wetold the children that if they came to a word that theycould not read, they were to raise their hand and theteacher or research assistant would give it to them. Weaddressed face and content validity (McDaniel, 1994)by exclusively focusing on information to be learnedfrom the book and strategies (such as using the index)for locating it as well as controlling for decoding abili-ty. We addressed content validity through the use ofscience informational texts identical in genre form andformat to the science informational books used by theteachers as part of the study.

    Psychometric analyses of the informationalreading comprehension assessment indicated that themeasure was acceptably accurate at all time points.However, there was some ceiling effect on the mea-sure such that it was unable to reliably detect differ-ences among students in later third grade. Cronbach’salpha reliabilities for the instrument over measure-ment occasions (timepoints) are .83 to .62 for the A-only group and .84 to .43 for the A+E group. Hereand throughout, the reliability statistics were internalconsistency alpha reliabilities, and they were exam-ined at each measurement occasion.

    Standard error of measurement was .11 to .08over time for both groups. The IRT empirical relia-bility for the assessment (taking all data from all timepoints to inform the single reliability score) was .83.Because these were all multiple-choice questions, wedid not complete an interrater reliability analysis forthis assessment.

    Procedural reading comprehensionTo assess the children’s ability to comprehend writtenprocedures in order to successfully carry them out,we constructed a performance assessment to maxi-mize face and content validity (McDaniel, 1994).

    We gave each child one of two forms of a written sci-ence procedure, written by us in consultation withelementary science educators. We asked the childrento read and follow the directions. We limited thenumber of procedural text assessment time points tothree across the two years (as compared with six forthe informational text assessments) because (a) theywere much more expensive to produce, involving asthey did an array of materials for each child in thestudy; and (b) it was possible that the children wouldremember more the last one they did due to itshands-on nature. By conducting this assessmentonly three times across the two years, we could spaceout the assessment periods longer to ameliorate thiseffect on the children’s memories and thus weakenthe threat to validity that such an effect would carry.

    As for the informational reading task, wehelped the children with words they could not read.Both procedural forms involved carrying out a sci-ence investigation, one involving characteristics ofvarious seeds, the other regarding whether oil and wa-ter mix. Along with the procedural text assigned tothem, children were given materials to complete theprocedure, including distracter materials that were ac-tually not needed to complete the procedure. In orderto ensure that tasks were completed independently, asneeded (typically for every other child at a table), wealso gave the children large trifold poster boards sothat their work could be obscured from that of otherchildren around them. We provided records for thechildren to keep in the course of the procedure, suchas a record of how seeds were sorted (achieved by glu-ing seeds to the paper in sorted groups) and a recordof what the oil and water looked like before, during,and after shaking it (achieved by using colored pen-cils to represent this on empty containers drawn on ablack-line master). We also constructed questions forthe children to answer at the conclusion of each pro-cedure that were intended to tap the science knowl-edge that each child learned from the investigation, afeature typical of science investigation procedural text(e.g., “Do you think that oil and water will staymixed together? Why or why not?”).

    The procedures were constructed and therecords scored in such a way as to identify if, when,and where in the procedure the child miscompre-hended; answers to questions could also be scoredwith respect to whether the procedure was successful.Raters scored the assessment blind to condition. Intotal, there were 12 items on one form of the assess-ment and 13 items on the other.

    Psychometric analyses of this assessment indi-cate that it was an adequately accurate measure. Thestandard error of measurement remained stable

  • over all time points for both groups, at .11. TheCronbach’s alpha reliabilities are .80 to .68 over timefor the A-only group and .81 to .76 over time for theA+E group. Generalizability theory analyses showedthat individual raters had little or no effect on scores,with G = .84–.90. The IRT empirical reliability forthe assessment was .87.

    Assessment of writingThe purpose of the writing assessments was to

    measure and describe growth in the children’s abilityto compose informational and procedural text with-in a science content context. Again, we focused ongenre development, not growth in science contentknowledge (see previous discussion). We devisedwriting assessments to meet the criteria for valid as-sessment of writing as specified by Cooper andO’Dell (1977) and Lloyd-Jones (1977): (a) we speci-fied the universe of discourse, (b) we devised writingtasks that sampled that universe precisely (science in-formational and procedural texts), (c) we ensured co-operation of the writers with enticing activities andattractive books within which to write, (d) we choosetopics within the legitimate knowledge of the writerby choosing from pilot work only those topics aboutwhich the children seemed to know the most, and(e) we created situation-bound writing prompts andexplicitly specified purpose and audience and provid-ed concrete textual space (booklets for informationalwriting and handouts for procedural texts). At thesame time, we needed the following psychometric as-pects for our assessments to allow us to use growthmodeling in our analysis: (a) standardized assessmentconditions, (b) equivalent forms within and acrossassessment time points, and (c) acceptable reliability.We also addressed face and content validity throughour selection of text types similar to those being usedas part of the study, and through our scoring strate-gy, which focused only on the genre features thatwere identified in our discourse analyses.

    INFORMATIONAL WRITING. For the informationalwriting assessments, we gave the children 30 minutesto compose an information book. We gave eachchild a book shell, with a poster board front andback cover and paper pages on the inside—lines onthe right and blank (for illustrations if the child sochose) on the left. The title of the book was on thefront cover as All About Dogs, All About Cats, or AllAbout Bugs, depending on the form of the assess-ment, followed by the word by and a place for thechild’s name. Each form included a written promptthat was read to the children prior to their beginningto write. Prompts varied by form but each involved

    setting out an actual audience who wanted or neededthe information in the book. For example, theprompt for the All About Cats book stated,

    Hi, I own a bookstore in Ann Arbor. A lot of people ask mefor books about animals, especially cats. They do not have acat, and they do not know much about cats. They want tolearn about cats. I already have a lot of books about how totake care of a cat, but I don’t have any books all about cats.Please write a book that will teach people about cats. Makepictures, too, to help them learn. I can give it to people whoask for one. Thank you.

    We chose these prompts in order to provide aconcrete audience rather than true authentic writingcontexts; the resulting texts were therefore not actu-ally sent to a bookstore, and generally studentsseemed to understand this. We allowed the childrento ask the assessor for the spelling of any word theyneeded; however, we encouraged them simply tospell, as much as possible, the best they could.

    PROCEDURAL WRITING. For the procedural writingassessment, we gave the children 30 minutes to com-pose a procedural text handout. We gave them eachan 8.5"�11" sheet of paper with horizontal linesand a generous left margin (for illustrations shouldchildren chose to include them) and a title, eitherHow to Brush Your Teeth or How to Freeze Water IntoIce (these, particularly the former, are admittedly astretch as “science” but we were limited to proce-dures for which children would likely have extensiverelevant background knowledge). As with informa-tional writing, we read a prompt to the children thatindicated a purpose and an audience for thewriting—for a person who wants to conduct theprocedure but who does not know how. Also, aswith informational writing, we allowed children toask the research assistant or the teacher for thespelling of any word they needed; however, weencouraged them to spell the best they could.

    SCORING. Raters scored the results of the informa-tional and procedural writing assessments blind tocondition. Our procedures for scoring the results ofthe writing assessments were guided by our focus onchildren’s growth in the ability to compose the gen-res of science informational and procedural texts. Wedid not score for accuracy of science content,spelling, punctuation, or handwriting. We first tran-scribed all writing assessment texts, with correctedspellings of words, into T-units, defined by Hunt(1977) as a single subject/verb clause plus whateverother subordinate clauses or phrases are attached to,or embedded within, that independent clause. Wethen numbered and listed the T-units separately.

    24 Reading Research Quarterly JANUARY/FEBRUARY/MARCH 2007 42/1

  • Learning to read and write genre-specific text 25

    This transcription procedure was to prevent hand-writing and spelling from influencing the scoring ofgenre considerations and to provide a common unitwith which to examine the length of the texts. Weconducted subsequent scoring using the T-units,with scorers going back to the original books to ex-amine any illustrations or graphical elements.

    We scored each text, informational and proce-dural, two times, first holistically and second with afeature analysis. We based the holistic score on howeffective the entire text was as its genre type. For ex-ample, raters judged how well the text worked as ascience informational text, using a 5-point Likertscale, with a 1 being “Text does not work at all as aninformational text,” and a 5 being “Text works wellas an informational text.” For scoring, we made

    judgments as to how well a text worked, or how ef-fective it was as an informational or procedural text,from the perspective of genre purpose.

    For the second analysis, we scored each text ana-lytically, based on the individual features of science in-formational or procedural texts previously identifiedand given to the teachers in the A+E group for explicitinstruction (see Table 2, page 19, for a list of these fea-tures). We based the scores on how effective the use ofthe individual feature was in providing informationfor the reader (focus on purpose of the textual genre)not the number of times, or if, a student used a partic-ular feature. See Figures 3 and 4 for examples of ab-breviated scoring guides that were used by the raters.

    For the informational texts, factor analysis of thescores did not result in one strong main factor. Rather,

    Entire text0 No response, no written text1 Text does not work at all as an informational text2 Text really does not work as an informational text but has isolated moment(s) of attempting to do so3 Text is vaguely like an informational text; it works a little like an informational text4 Text is close to working as an informational text5 Text works well as an informational text

    Feature � feature assessment(For each feature of informational text below, the scorer used the following rubric, with each score point containing some feature-specific guidance.The one for “attributes” is included in the example below.)

    0 Does not use this feature and would have been inappropriate to do so, given topic.1 Never uses this feature and could have to provide information reader may want.3 Uses this feature but not entirely effectively; the attributes or components provided are not related to key topics or not complete enough; or

    more use of this feature would have been more effective.5 Uses this feature quite effectively to inform the reader; attributes or components are provided and are informative and appropriate; and use of

    the feature is sufficient, given the textual content.

    Effectively uses description(s) of attributes/components (e.g., “Ants have six legs.”) to inform the reader

    Effectively identifies characteristic events (e.g., “Ants eat sugar.”) to inform the reader

    Effectively includes definitions to inform the reader

    Effectively uses compare and contrast (talking about ways things are alike and different) to inform the reader

    Effectively uses denotative language (e.g., “Most worms are between 1 and 4 inches long.”) to inform the reader

    Effectively uses timeless verb constructions (e.g., “Ants eat sugar.” versus “The ant ate sugar.” or “The ant is eating sugar.”) to inform the reader

    Effectively uses generic noun constructions (e.g., “Ants have legs.” versus “That ant has legs.” or “Joe the ant has legs.”) to inform the reader

    Effectively uses specialized vocabulary to inform the reader

    Effectively uses realistic illustrations to inform the reader

    Effectively uses an opening statement/general classification to orient the reader to the information to be conveyed

    Effectively uses a general statement/closing to summarize or comment upon the information conveyed previously in the text

    Effectively uses headings to assist reader in locating information or to orient reader to information to come

    Effectively includes an index to assist reader

    Includes classification (at either the sentence level—“Bees are a kind of insect.”—or larger units—“Some frogs live in ponds. Some frogs live intrees….”)

    Has labels and/or captions

    Has graphical devices (diagrams, tables, charts, boldface/italicized vocabulary

    Has a table of contents

    FIGURE 3ABBREVIATED (FOR PURPOSES OF PUBLICATION) SCORING GUIDE FOR INFORMATIONALTEXT WRITING ASSESSMENTS

  • factor analysis revealed that features clustered intothree groups: (a) a small group of verbal features, (b) alarger group of verbal features—including the smallgroup plus several others, and (c) a group of primarilyvisual features (specifically graphical devices, realisticillustrations, headings, index, and labels and captions).These clusters reflect the findings of a factor analysisof the individual feature scores of children’s writing,not the findings from the initial discourse analysis ofscience informational texts reported earlier. Thus, thecategories are not isomorphic. Because the first clusteridentified in the cluster analysis is completely encom-passed in the second cluster, we used only two of thethree clusters—the larger group of verbal features and

    the group of primarily visual features—as outcomemeasures. We formed each of these two outcome mea-sures by adding and then averaging scores on each fea-ture in the cluster.

    Psychometric analyses indicate that this was anadequately accurate measure. Standard Error ofMeasurement was .15–.18 over time points.Cronbach’s alpha reliabilities were .86 to .78 over timepoints. Generalizability theory analyses showed thatindividual raters (of which there were three, with in-terrater reliability assessed on approximately 1