Opening Dialogue

Opening Dialogue

UNDERSTANDING THE DYNAMICS OF LANGUAGE AND LEARNING IN THE ENGLISH CLASSROOM

Martin Nystrand WITH

Adam Gamoran, Robert Kachur, & Catherine Prendergast

FOREWORD BY ROBERT GUNDLACH

J E 4 C H E R S © Teachers College Press 1997-2012 ^*-Skfv^ Martin Nystrand 2012-

Teachers College, Columbia University New York and London

The will to learn is an intrinsic motive, one that finds both its source and its reward in its own exercise. The will to learn becomes a "problem" only under specialized circumstances like those of a school, where a curriculum is set, students confined, and a path fixed. The problem exists not so much in learning itself, but in the fact that what the school imposes often fails to enlist the natural energies that sustain spontaneous learning— curiosity, a desire for competence, aspiration to emulate a model, and a deep-sensed commitment to the web of social reciprocity.

—Jerome Bru ner. Toward a Theory of Instruction

CHAPTER 2

The Big Picture: Language and Learning in Hundreds of English Lessons

Martin Nystrand and Adam Gamoran

F E W G O O D M O D E L S currently exist for understanding h o w social processes affect student learning i n classroom settings. There is widespread consensus that research must focus o n teacher-student and peer interaction as i t affects learning i n order to be sensitive to the social context o f learning. Yet most conceptions o f instruct ion view learning as the result o f what teachers plan and provide for students, that is, what teachers do to students. Adherents o f this approach see instruct ion as a one-way transmission o f knowledge f r o m teacher and texts t o students, and they typically assess students' k n o w l edge for its congruence w i t h curricular aims and objectives.

I n our research i n secondary school English classes, however, we had i n m i n d no t what teachers "do to students" but rather what teachers and their students do together, that is, wha t Michaels (1987) calls "the day-to-day practice o f a ' cu r r i cu lum'" (p. 323) . I n this sense, teacher and students negotiate the actual cur r icu lum — as opposed t o the ideal or intended cur r i cu lum (for example, as w r i t t e n up i n a cur r icu lum guide). Superficially, this negot iat ion is visible i n the give-and-take o f classroom talk (Flanders, 1970). However , such ostensible interaction clearly is pedagogically less significant than the cognitive interaction that occurs —or does no t occur, as the case may be — between teacher and students. W h e n minds meet i n this way, the result is a sequence o f shared understandings o f subject matter among members o f the class, and the course o f instruct ion, whether considered o n any given day or examined over an entire school year, may be analyzed i n terms o f h o w classr o o m talk and activities modi fy and expand these understandings. I n this sense, dialogic instruct ion is a negot iat ion o f meaning by and between teacher and students.

We observed hundreds o f eighth- and ninth-grade lessons over 2 years. I n all our observations, we never found the "perfectly dialogic classroom"; indeed, such a perfect classroom probably does no t exist i n the real w o r l d . Rather, we d i d a large empirical study to examine the general effects o f dialogic practices o n achievement and learning. The scope o f our study, under-

30

LANGUAGE AND LEARNING IN HUNDREDS OF ENGLISH LESSONS 31

taken w i t h a large sample o f students i n a large and diverse sample o f classes, schools, and communit ies , enabled us to systematically test hypotheses about such practices.

This chapter reports details and findings f r o m this study, sketching an overall portrai t o f classroom discourse i n middle and h igh school English classes, and helping us address several impor tan t questions about classroom talk and student learning.

• W h a t is instructional discourse generally Hke i n e ighth- and ninth-grade Engl ish and language arts classes?

• H o w m u c h instruct ion is recitation .> H o w m u c h consists o f discussion and small-group work?

• H o w much instruct ion is organized dialogically? Monological ly? • H o w does classroom discourse vary f r o m midd le to h igh school? By

abil i ty group? By subject? A m o n g urban, suburban, and rural schools? • H o w do these practices affect student learning about literature? W h i c h

interactions are appropriate and productive?

Our study, w h i c h shows the role that large-scale data analysis can play i n the investigation o f classroom interaction and its effects o n learning, found that generally students learn more i n classrooms organized more dialogically than monologically. For an overview o f the entire study, see Figure 2 . 1 .

STUDYING CLASSROOM DISCOURSE: DESIGN AND METHODS

I n order t o develop as comprehensive an understanding o f classroom discourse as possible, our study coordinated three separate but related investigations:

1. Surpeys and interviews. We sought first to determine practices and attitudes t o w a r d classroom discourse by surveying bo th students and teachers o n various classroom practices; i n addi t ion, we in terviewed the teachers to learn about their instructional methods and the context o f instruct ion.

2. Class observations. T h r o u g h direct observation, we investigated teachers' allocation o f class t ime to various types o f classroom discourse, inc luding recitation, discussion, and small-group work . We collected data a l lowing us t o gauge the quali ty o f teacher-student interaction by focusing on the characteristics o f questions asked by bo th teachers and students.

3. Hypothesis testing. Us ing statistical techniques, we examined the general effects o f classroom practices and the organization o f instruct ion o n student achievement.

32 OPENING DIALOGUE

Figure 2.1A. Synopsis of Study—Design

The purpose of our research was to investigate the effects of instructional organization on student learning, contrasting the epistemologies of recitation and discussion. This work was conducted in a 2-year study (1987-1989) in 16 middle and junior high schools and 9 high schools in eight midwestern urban, suburban, and rural communities. Participating were 58 eighth-grade and 54 ninth-grade language arts and English classes, involving more than 1,100 students each year. Each class was observed four times, twice in the fall and twice in the spring, providing observational data for more than 200 lessons each year.

T R A C K I N G The study encompassed both middle and high school, fol lowing a subset of students as they moved into high school, in order to understand the mechanism and effects of placement and tracking in high school. The study was designed to provide systematic contrasts of instruction and learning in high- and low-track classes.

INSTRUCTIONAL DISCOURSE Instructional discourse was studied in two ways. First, observers timed instructional activities in order to determine the allocation of class time to various activities, for example, question-answer, discussion, small-group work, seat-work , and other activities. In addition, observers recorded and coded both teacher and student questions for dimensions of dialogic instruction, including (1) authenticity (whether or not questions had "prespecified" answers), (2) uptake (incorporation of previous answers into subsequent questions), and (3) level of evaluation (extent to which the teacher allowed a student response to modify the topic of discourse). More than 23,000 questions were coded. The observational data were supplemented by teacher and student survey data and end-of-year teacher interviews.

L E A R N I N G Learning about literature was tested wi th a written examination in the spring based on several works of literature studied during the year. The test involved a set o f increasingly more probing questions, ranging f rom naming and/or describing as many characters from each story as the student could remember, and explaining the ending of each story to briefly explaining the themes and conflicts o f each story and relating theme, conflict, and ending. A l l students answered the same general questions, although the details of the tests varied depending on the titles studied and selected. For the ninth-grade test, students wrote a brief essay on some character from their readings whom they admired, and explained their admiration.


Figure 2.1B. Synopsis of Study—Results

O V E R A L L Classroom discourse was overwhelmingly monologic. When teachers were not lecturing, students mainly were either answering questions or completing seat-work. The teacher asked nearly all the questions, few questions were authentic, and few teachers followed up student responses. On average, discussion lasted less than 50 seconds per class in eighth grade and less than 15 seconds per class in ninth grade. Small-group work in eighth grade took only about half a minute each day, and only a lit t le more than 2 minutes a day in grade 9.

P R O C E D U R A L V A R I A B L E S The study found a modest effect for time spent on homework, no effect for asking questions in class, and a negative effect for level o f activity during recitation.

D I A L O G I C INSTRUCTION Results provided support for dialogic instruction, indicating that time devoted to discussion, authentic questions, uptake, and high-level teacher evaluation had a strong positive effect on achievement. Discussion in grade 8 had a particularly large effect.

IMPORTANCE OF CONTENT Results of the grade 9 study were consistent wi th grade 8 results but only when controlled for content. Discussion and authentic questions unrelated to literature had a negative effect on learning.

GROUPWORK Groupwork was successful to the extent that teachers clearly defined goals and tasks at the same time that they encouraged students to generate conclusions, solve open-ended problems, and address authentic questions rather than simply manipulate information and answer study questions. Most small-group work in study classes was in fact "collaborative seatwork," however, which had a negative effect on learning.

T R A C K I N G Instruction was more fragmented, contrived, and monologic in low-track than in high-track classes. In grade 8, teachers lectured 40% more in low-track classes than in high-track ones, and low-track discussion time was only half that o f high-track groups. In grade 9, seatwork was nearly four times more frequent in low-track than in high-track classes.

34 OPENING DIALOGUE

Table 2.1. Characteristics of School Sample Number of Schools

School District Type Total Middle Schools High Schools Parochial 8 6 2 Public 17 10 7

Small town/rural 6 3 3 Suburban 3 2 1 Urban 8 5 3

Research Sites and Participants

O u r study lasted 2 years: Eighth-grade classes were observed d u r i n g 1987-88 , ninth-grade classes du r ing 1988-89 . We collected data i n eight m i d -western communit ies , inc lud ing rural , urban, and suburban sites, i n b o t h public and parochial schools. Six o f these communit ies were public school districts; the other t w o were Catholic h igh schools w i t h students f r o m a n u m ber o f urban and suburban K - 8 feeder schools. U n l i k e ninth-grade classes, w h i c h were all called English, eighth-grade classes were variously called language arts, English, reading, communications, literature, and so o n ; we selected the eighth-grade classes that focused most o n reading. Table 2 .1 provides a breakdown o f the communi ty and school types that participated i n our study.

I n each school we observed four English classes. I n the smaller schools, we observed all the English classes; i n the larger schools, we selected a representative sample o f different abil i ty groups as defined by the school (honors or accelerated, regular or average, basic or remedial). I n all we made 4 5 1 observations i n 58 eighth-grade classes i n 16 middle and j u n i o r h igh schools, and 54 ninth-grade classes i n nine h igh schools (which were fed by the j un io r h igh and middle schools i n our eighth-grade study). Between 1,100 and 1,200 students participated each year; o f all eligible students, about 1 0 % were lost t h r o u g h absence or refusal. A b o u t one- third o f all students participated i n b o t h years o f the study. Table 2.2 summarizes these data.

Observational Procedures

Each class was visited four times by a trained observer, twice du r ing fall semester and twice du r ing spring semester.^ W i t h i n these parameters, observations were scheduled at the mutua l convenience o f teachers and observers. O n these occasions, the observer noted the t ime spent i n different instruct ional activities, and recorded and coded all teacher and student questions for a set o f variables contrasting monologic and dialogic ins t ruct ion (coding is explained o n pp. 3 7 f ) .


Table 2.2. Scope of Study

Characteristics Grade 8 Grade 9 Number of students 1,041 1,100 Number of classes 58 54 Number of times each class observed 4 4 Number of observations 227 224 Number of coded questions 12,033 11,043

Discourse Episodes and Segments

Data f r o m each class session were organized according to episodes and segments. A n episode was defined as a coherent classroom activity centering around a particular objective or purpose. A new episode was marked w h e n the teacher addressed a new objective. L ike the start o f a new paragraph, each such shift usually was evident i n the teacher's in i t i a t ion o f a new topic. Usually episodes consisted o f t w o or more activities. For example, i n addressing a particular objective, a teacher m i g h t initiate a question-and-answer session that then w o u l d be interrupted by br ie f periodic lectures and culminate i n a h o m e w o r k assignment. W h e n something like this happened, we d iv ided the episode in to segments, defined as any coherent part o f an episode that differed f r o m other activities o f the episode. Instruct ional activities were classified w i t h durations i n minutes and seconds for the fo l l owing :

1. Classroom management activities Classroom procedures D i r e a i o n s Discipl ine

2. Direct instruction Lecture, f i lm Question-answer Discussion Student presentations Students reading aloud

3. Seatwork Supervised w i t h teacher helping Supervised w i t h teacher m o n i t o r i n g Unsupervised Small-group w o r k

4. Tests and quizzes

W h e n the teacher d i d one t h i n g (e.g., lecture) and some students were allowed to do another (e.g., w h e n the teacher lectured to part o f the class bu t

36 OPENING DIALOGUE

individualized inst ruct ion for others by assigning them seatwork), or the teacher d i d no t object to some do ing w o r k unrelated to the lecture, we classified the activity that most students d i d . Observers wro t e a b r ie f description o f each episode and recorded and coded each question asked by either the teacher or students.

We defined discussion as the free exchange o f in format ion among students and/or between at least three students and the teacher that lasted at least a half minute . Typically discussions came about du r ing question-answer exchanges w h e n a student w o u l d volunteer an observation (rather than ask a question) that the teacher al lowed to substitute for normal evaluation. These discussions, w h i c h interrupted or violated the normal initiation-response-evaluation ( I R E ) sequence o f recitation, included few questions, and those that were asked typically clarified ideas and in format ion ("By that do y o u mean . . .

Questions

Bakhtin's conception o f discourse encompasses far more than just questions and the utterances immediately preceding and fo l l owing them: As we saw i n Chapter 1, Balditin's chains o f utterances also encompass frames o f m i n d and core beliefs, inc luding those related to students' experience ou t o f school. Dialogic analysis is most direcdy accomplished t h r o u g h close analysis o f transcripts o f indiv idual lessons and other forms o f qualitative analysis, especially o f the sort Dyson undertakes i n her Social Worlds of Children Learning to Write (1993) . A l t h o u g h our study under took some such analysis (see Chapter 3) , the main focus o f our research was a comprehensive analysis o f classroom discourse, generally w i t h a special focus o n teacher and student questions. We d i d this for several reasons. First, we wanted to capture the general dimensions o f instruct ion, requir ing us to examine hundreds o f i n structional episodes and lessons. We focused o n question-answer exchanges because they provide an effective me thod for such a general analysis, and because they are so central t o instruct ion, occupying 3 0 % o f class t ime i n the eighth-grade classes we studied and 4 2 % i n the ninth-grade classes. Quest ion-answer exchanges between teachers and students clearly dominate i n struction for most students. They play a key role i n b o t h accommodating and excluding student voices i n the public , authoritative discourse o f the classr o o m , and they are the central instructional mechanism i n American classrooms for assigning epistemic roles to students. As such, they significantly regulate the extent to w h i c h teacher-student interaction can be dialogic.

T w o features o f questions were o f particular interest t o us: authenticity (whether or no t teacher questions had "prespecified" answers) and uptake ( incorporat ion o f previous answers in to subsequent questions). Each is a critical variable affecting the salience o f student voices i n classroom discourse, and


reflecting h o w far the horizons o f classroom discourse extend beyond the question-answer sequences themselves to draw o n student experience central t o student engagement. W h i l e these data cannot detail the dynamics o f particular teacher-student interactions, they nonetheless provide a powerful index o f the extent t o w h i c h teachers open their classes to student voices w h e n they ask the questions they do.

Questions are n o t everything, o f course, and authentic questions, we found , do no t invariably produce learning. Nonetheless, one must no t underestimate the role teachers' questions play i n shaping the character o f classroom discourse as i t affects learning. Questions presume answers. As negotiations o f sorts, question-answer sequences reveal impor tan t features o f teacher-student interaction and hence the character o f instruct ion. M u c h can be learned about teacher-student interaction and talk i n a classroom by det e rmin ing the source o f questions, the extent o f authenticity and uptake, the level o f cognitive activity that questions elicit, and so on . Even the pace o f a teacher's quest ioning can be revealing: Carlsen (1991) cites studies f ind ing that a s low pace o f teacher quest ioning and extended wai t times correlate w i t h greater numbers o f student responses (Honea, 1982) , as we l l as more sustained student responses o f greater complexity and higher-order t h i n k i n g (Fagan, Hassler, & Szabl, 1981).

Early i n our study, we learned that the vast p r o p o r t i o n o f teacher questions (a) are test questions, (b) get a response, (c) do no t involve uptake, and (d) elicit a report o f what is already k n o w n . Indeed, this is the very profile o f monologic classroom discourse, and we soon began to describe such questions, unfortunately, as normal teacher questions. The fo l l owing questions, all f r o m M r . Schmidt's lesson described i n Chapter 1, are examples:

• "According to the poet, what is the subject of The Iliad?'' • "Where does the action o f the first part o f Book I take place when we

enter the story?" • "Wha t is the result o f the quarrel between Agamemnon and Achilles?"

We used data about questions to b u i l d profiles o f instruct ion and classr o o m discourse, coding more than 23,000 questions, and examining each question i n the context o f the whole lesson at the t ime i t was asked.^ N i n t h -grade classes were tape recorded. Whenever observers were uncertain about h o w to interpret classroom activities and code questions, they consulted w i t h the teacher after class. Questions were coded for

• Source: Teacher or student • Response: Yes or no • Authenticity: Whether or no t an answer was prespecified • Uptake: Incorpora t ion o f a previous answer in to a subsequent question

38 OPENING DIALOGUE

• Cognitive level: The type o f cognitive demand made by the question • Level of evaluation: Whether the teacher valorized and elaborated the

students' responses

C o d i n g reliabilities were based o n paired readings o f a sample o f questions.^ D u r i n g the ninth-grade study, observers collected data using a specially

w r i t t e n computer program, C L A S S 2.0,'* d u r i n g class. This program helped w i t h question coding, as wel l as the allocation and t i m i n g o f various instruct ional activities, w h i c h the observer recorded by selecting f r o m a menu and then briefly described. Every 5 minutes du r ing question-answer exchanges and every 2 minutes du r ing seatwork and lecture, the program p rompted the observer to record the number o f students obviously o f f task,^ as we l l as the number actively participating, and to make adjustments i n the number o f students i n case any recendy had entered or left the classroom.

D u r i n g question-answer exchanges, the observer typed i n and coded the questions that teachers and students asked du r ing instruct ion. As observers entered each question in to computer memory, the program p rompted them for codings. W h e n data collection was completed, C L A S S - E D I T 2.0, a companion p rogram to C L A S S 2.0, al lowed proofreading, edi t ing, and revising o f each file for inappropriate codings, and then computed basic statistics for each episode. Generally, we were satisfied that these procedures were min ima l ly i n t r u sive to ins t ruct ion d u r i n g our observations.

Authenticity. Authent ic questions are questions for wh ich the asker has no t prespecified an answer and include requests for in format ion as we l l as open-ended questions w i t h indeterminate answers. Dialogically, authentic teacher questions signal t o students the teacher's interest i n what they th ink and k n o w and no t just whether they can report what someone else thinks or has said. Authent ic questions invite students t o contr ibute something new to the discussion that can change or modi fy i t i n some way.

By contrast, a test question allows students no cont ro l over the flow o f the discussion. Because authentic questions al low an indeterminate number o f acceptable answers and open the floor to students' ideas, they w o r k dialogically. By contrast, a test question allows only one possible r igh t answer, and is hence monologic ( i n Lotman's terms, univocal; see Chapter 1).

Before we started our observations, we wor r i ed that determining the authent ici ty o f questions m i g h t be complicated and unreliable since such determina t ion requires assessing teachers' intent ions; authenticity cannot be determined f r o m words alone. For example, " W h o w o n the W o r l d Series i n 1928?" can be either a test question or an authentic question depending on (a) whether the asker knows the answer and wants t o see i f the person asked also loiows, i n w h i c h case i t is a test question, or (b) whether the asker doesn't


k n o w and wants to f ind ou t by asking someone w h o does know, i n wh ich case i t is authentic.

The nature o f the activity, that is, the genre o f classroom discourse, we discovered, is the most reliable indicator o f authenticity. Hence, when teachers began a lesson by saying, "Okay, class, let's check the answers to your study questions," we quickly learned that the questions were invariably test questions (al though fo l low-up discussions o f students' answers were sometimes authentic). By contrast, when teachers asked about students' personal experiences as lead-ins, for example, t o open-ended discussions o f a poem or short story, we found that these questions were authentic. I n Chapter 1, M s . Turner's in i t ia l question, "Can y o u recall things f rom Huck Finn that, u m , seemed racist t o you?" is an example. Questions asked d u r i n g discussions, for example, the question T o m asks John i n Ms . Lindsay's class, "Is i t M r . H o l -lings's store? Is that i t?" are also authentic since their purpose is no t testing someone's knowledge bu t rather exchanging only that in fo rmat ion the person asking the question actually needed to know.

Whenever the authenticity o f a question was unclear o r ambiguous to us, we consulted the teacher. I n practice (and somewhat to our ini t ia l surprise), coding authenticity proved to be generally quite straightforward. We quickly learned that most classroom discourse is no t subtie: The vast p r o p o r t i o n o f questions teachers ask are test questions, whereas student questions are v i r t u ally always authentic (except when students role play teachers; then they ask test questions!).

Uptake. Uptake occurs when one conversant, for example, a teacher, asks someone else, for example, a student, about something the other person said previously (Coll ins, 1982). Here is an example o f uptake f rom a n in th -grade lesson on The Iliad: The teacher asks, "Wha t do they have to do to Polyphemus?" A student replies, " B l i n d h im . " The teacher then follows up, asking, " H o w come the plan is for Winding Cyclops?" Uptake occurs here when the teacher picks up o n the student's response, asldng about " b l i n d i n g " h i m . Uptake often is marked by the use o f pronouns, for example, " H o w d i d it work?" ; "Wha t caused if?"; "Wha t city grew ou t o f this?'' I n each o f these questions, the italicized p ronoun refers to a previous answer.^ Uptake also may be characterized by ellipsis. I n M s . Turner's class, for example, when L inda says that racism i n The Adventures of Huckleberry Finn makes her "ashamed," M s . Turner's reply, " I n what way?" exhibits uptake since her quest i o n makes Linda's answer the momentary topic o f discourse.^ Teachers use uptake whenever they fo l low up on student responses. As an essential dialogic resource facilitating the negotiat ion o f understandings, uptake plays a p r o m i nent role i n discussion as conversants listen and respond appropriately to each other.

40 OPENING DIALOGUE

Cognitive Level. O u r project also sought t o assess whether the cognitive level o f quest ioning affects student learning. I n this way we examined whether instruct ion stressing higher-order t h ink ing is necessarily dialogic. We therefore coded the level o f cognitive funct ioning that each question sought t o elicit, j u d g in g i t h igh to the extent that the question could no t be answered " th rough the rout ine application o f previously learned knowledge" (New-mann, 1990, p. 44 ; see also Polanyi's (1958) dis t inct ion between rout ine performances and heuristic acts). L ike autiienticity, the cognitive level o f questions cannot be judged altogether f r o m words alone. For example, i f the teacher expected students to answer questions by recit ing in format ion found i n textbooks, we coded questions as reports regardless o f their l inguist ic structure. Hence, a l though a why-quest ion normal ly w i l l elicit an analysis, i t w i l l elicit a report i f the teacher's focus is the recitation o f a textbook's analysis rather than the class' reflection; then "Why?" really means, "According to your text, w h y d i d i t happen this way? D o y o u remember?" Factors affecting the cognitive level o f any question include the fo l l owing :

1. Source of the question. The same question that elicits an analysis f r o m a person w h o has to figure things ou t may wel l elicit a report f r o m another, more knowledgeable ind iv idua l w h o already knows and simply needs to explain. For example, " W h y d i d Odysseus and his men plan deliberately to b l i n d Polyphemus?" may elicit an analysis f r o m students (assuming, o f course, that they have to figure ou t the answer and no t merely recite their textbook o n the p o i n t ) , bu t most l ikely w i l l elicit a report i f a student asks a teacher w h o already knows the answer. W h e n we were unclear, we asked about i t after class.

2. Experience, ability, and prior knowledge o f the person answering the quest i o n , whether student or teacher. I f student answers seemed to require rout ine cognitive operation, we coded questions as el ic i t ing reports. We defined pr ior knowledge as "p r io r t o the previous night's homework."^

3. Nature of the instructional activity. W h e n an episode was devoted to review, our normal expectation for responses was a report , even i f questions had the linguistic f o r m o f higher-level questions (e.g., "What's the difference between a symbol and an image?" when asked as a study question).

4. Source of information required by the question. I n fo rma t ion sources include p r io r experience, textbooks, and previous teacher lectures.

Level o f cogn i t ion elicited by questions was measured o n a five-point linear scale calibrated for level o f abstraction and derived f r o m Applebee (1981) , B r i t t o n , Burgess, M a r t i n , M c L e o d , and Rosen (1975) , and Mof fe t t (1968) . Levels were as fol lows:

1 Record o f an ongo ing event: What's happening? 2 Recitation and report o f o l d in format ion : W h a t happened?


3 Generalization: W h a t happens? 4 Analysis: W h y does i t happen? 5 Speculation: Wha t m i g h t happen?

We coded questions as records i f they elicited descriptions o f what students were observing, feeling, or d i i n k i n g at the time o f the question. Examples include: "Any questions o n that?" and "Wha t [o r w h y ] are y o u th inking about that?" I f the question required students to th ink and no t just report something already I m o w n or previously though t by someone else, then we scored cognitive level higher than 2. De te rmin ing the level involved j u d g i n g whether the student answering the question was bu i ld ing up a generalization, i n w h i c h case we scored i t a 3, or breaking d o w n an argument, i n w h i c h case we coded i t as an analysis and rated its cognitive level as 4. Generalizations display inductive reasoning, bu i l d ing up ideas rather than breaking them d o w n . They address questions such as: "Wha t happens?" and "Wha t do I make o f what happens?" They tie things together; they are no t restatements o f in format ion . Analyses display deductive reasoning, brealdng concepts, ideas, and arguments d o w n rather d ian bu i ld ing up ideas. To be scored as analyses, questions had to require more than restatements o f k n o w n information. Questions were judged to be lower order (i.e., ehcit ing records or reports) i f they elicited o l d in format ion , or higher order (i.e., e l ici t ing generalizations, analyses, or speculations) i f they elicited new informat ion and could no t be answered t h rough the rout ine application o f p r io r knowledge.'^

Unless texts explicidy stated the answers to teachers' questions, we judged most questions about literary texts t o be either generalizations or analyses. Hence, f r o m a ninth-grade class session on To Kill a Mockingbird, the question, " H o w does T o m die?" ehcited a report since the answer is stated i n the novel. By contrast, "What's the overall reaction o f most o f Maycomb's citizens?" ehcited an analysis since the answer required taking in to account and relating disparate in format ion f rom the text. We found that questions like, "Wha t is die meaning ofx.?" or "Wha t is x?" ehcited either (a) a report if , i n recitation, the teacher's in tent was to see whether students knew i t ; (b) a

generalization i f x was a w o r d and the teacher's in tent was t o elicit an or iginal def ini t ion; or (c) an analysis i f x was a phrase i n a book, a line i n a play, a symbol , and so fo r th , and the teacher asked students to situate or relate a part o f the text (phrase, l ine, symbol) to the whole , for example, "So h o w do we explain x?" I n these cases we coded recitations o f meaning as reports, and explications o f meaning as generalizations and analyses.

THE PREVALENCE OF MONOLOGIC INSTRUCTION

The results o f our research found most classroom discourse to be overwhe lming ly monologic . I n this regard, our study replicates numerous previ-

42 OPENING DIALOGUE

ous studies document ing the historical and widespread prevalence o f recitat i o n i n American schools. Indeed, as early as 1860, M o r r i s o n complained that "young teachers are very apt to confound rapid quest ioning and answers w i t h sure and effective teaching" (cited i n Hoetker & Ahlbrand , 1969, p. 153). I n 1912, Stevens complained that the widespread practice o f recitation made "the classroom the place for displaying knowledge instead o f a laboratory for get t ing and using i t " (p. 16). I n 1919, C o l v i n estimated that on ly "about five percent [ o f the teacher questions he studied] could be considered i n any way genuine though t questions" (p. 269) . W r i t i n g about the same t ime. M i l ler (1922) complained that teachers were unable to "endure the silence that must prevail whi le the p u p i l is t h i n k i n g and organizing his material" (quoted i n Hoe tker &: Ah lb rand , 1969, p. 154). Thayer (1928) claimed tiiat recitat i o n was a progressive reform enabling teachers to gauge the mastery o f large groups o f chi ldren by checking the knowledgeabil i ty o f relatively few. Corey (1940) , BeUack, Kl iebard, H y m a n , and S m i d i (1966) , and Hoe tker (1967) all found that teachers talked about two- thi rds o f all instructional t ime and that more than 8 0 % o f all teacher questions sought t o elicit recall i n a recitat i o n format. Recent studies continue to f ind similar results; see Duffy (1981) , D u r k i n ( 1 9 7 8 - 7 9 ) , Hoetker and Ah lb rand (1969) , Goodlad (1984) , Sarason (1983) , and Tharp and Gal l imore (1988) .

Class Time

O u r study generally replicates these depressingly endur ing findings. Recita t ion and lecture were c o m m o n . W h e n teachers were no t lecturing, students mainly were either answering questions or engaged i n seatwork. Indeed, o n average, 8 5 % o f each class day i n bo th eighth- and ninth-grade classes was devoted to a combina t ion o f lecture, question-and-answer recitation, and seatwork. Discussion and small-group w o r k were rare. O n average, discussion t ook 50 seconds per class i n e ighth grade and less than 15 seconds i n grade 9;i° small-group w o r k , w h i c h occupied about ha l f a minute a day i n e ighth grade, took a b i t more than 2 minutes a day i n grade 9.^^

I n grade 8, more than two- th i rds o f all classes had at least 10 minutes o f seatwork daily, and 3 1 % had 20 minutes or more daily; only one class had no seatwork at all. H a l f o f all classes had at least 10 minutes a day o f question-and-answer activity, and 17 .2% had 20 minutes or more. A l l bu t t w o classes rout inely involved lecture; teachers i n four o f the 58 classes lectured 2 1 m i n utes o r more each day. Overall means for all eighth-grade class activities are summarized i n Table 2.3.

I n n i n t h grade, t ime spent o n lecture increased f r o m 5 to 8 minutes per day ( 2 6 % o f all class t ime) , and question-answer recitation increased f r o m 12 to almost 18 minutes per day. Lecture was so ubiqui tous that less than 2 % o f

C3 M a: u

u

-1

u

0 0 O 0 0 0 0

vq en o o c<S ly-i O CN

CN v O O N C O so OCJ U-) VO p r t

oo' —<• o CN

0 0 0 0 ^ O l > t~~; 0 0 m (N CN i r i r^' m'

CN O 0 0 0 0

p GO iri O O CN

C3N r - O N i n m ^ O N so 0 0

O N i n ^ —«' i r i

so O N O O m

^ >o 0 0 CN| so ON r~- so so m (N C N C O

— CN O (N cn —; CN r~ lO (N (N —• O (N

0 0 O O o ^ O N C N O t>

(N SO ^ SO 's^ O m oq o 0 0 o 0 0 c<S C O O O* r n

r<-) so O N so ON CNj r-- r- i > r-̂ so —i ^

m o so so so n-) (N 0 0 m c~-i n c4 o o CN

M 3 s ° .2 -3

c .S

C C

c

< c _o

p. «

5 o -3 -o

O N 0 0 0 \0 O - H O N o

p y-i o o so 0 0

o >o — o o ocj o —<• o

^ ^ — I CN <n O O N O N

r~-, K o <N

O >o - H rn i > ~ c<-i O CN —̂" O —<' —

o S~ «n 0 0 ^ so r - ^ CN Z ^ " <N ( N

p —< ON o i n ON " i n — o o ' O ^ ' O

i n ^ ^

" 1 ^ P ON O r- ' r<i

C O r— O N so o i n o i —• —

CN o o o CO —. in ( N o o o 0 \ CO O

C O r - r - ^ o CN so " 0 0 O

—' d o d

so i n i n o r~; ^ m C O i n

CN co' CN CN

so 0 0 —. r-H U-l O O '—^ so ' — I so ^ ir^ d d

o

3 3 OJ C Q 55 ^ on

(U N ^

I <D -a p

H C/2 ^ g g

H 0 0

0 0

II

>

(U o

c 1^

44 OPENING DIALOGUE

all classes had none at all, and ha l f the classes heard 8 minutes or more o f lecture daily (one class actually had an average o f more than 2 7 minutes a day). A l l classes involved question-and-answer recitation; 5 0 % had at least 16 minutes daily ( t w o classes averaged more than 30 minutes o f recitation each day). O n l y t w o classes had no seatwork at all ; at least 10 minutes daily was c o m m o n for nearly a t h i r d o f all classes. Overall , results on use o f class t ime document the clearly monologic character o f classroom discourse i n the classes we observed. Overall means for use o f class t ime i n grade 9 are summarized i n Table 2.4.

Questions

The character o f instructional questions i n the classes we observed was consistent w i t h the monologic organization o f class t ime. M o r e commonly than not , students were treated as "empty vessels" to be " f i l l ed" by teachers. I n vir tual ly all classes, the teacher asked nearly all the questions; few about literature were authentic, and equally few fo l lowed up o n student responses. I n the eighth-grade classes (see Table 2 .5) , about 35 questions o n average were asked dur ing each class, 9 2 % o f them by the teacher. O n l y 12% o f the teacher questions were authentic, and only 1 1 % exhibited uptake.'^ I n the ninth-grade classes (see Table 2 .6) , teachers asked more than 52 questions each class per iod o n average, or about 5 0 % more than i n the eighth-grade classes: 5 4 % o f all questions involved recitation. These results help explain Nat ional Assessment o f Educat ion Progress ( N A E P ) results, w h i c h perennially show that American students are far more proficient at literal comprehension than at analysis and critical t h i n k i n g (see, e.g., Applebee, Langer, M u l l i s , Latham, & Genti le , 1994).

Given the infrequency o f authentic questions i n eighth grade, we were surprised to find that the p r o p o r t i o n i n ninth-grade classes was twice as h igh . Indeed, authentic questions were asked i n all the ninth-grade classes we observed; half the classes rout inely had 2 5 % or more. I n subsequent analysis, we discovered that much o f the increase was due to the use o f authentic questions to inquire about nonacademic topics. Uptake also was more c o m m o n i n ninth-grade classes, exhibited by 2 6 % o f all questions. L ike eighth-grade instruct ion, however, most ninth-grade classroom interaction tended to be monologic , pe rmi t t ing l i t t ie oppor tun i ty for substantive exchange.

Track Differences

The near-universal preference for "recitable in fo rma t ion" afflicted low-track classes even more than regular- and high-track classes since low-track classes typically go t a refracted, watered d o w n , fragmented rendi t ion o f the

• a

> o

H

o p '•S

< (U

•rs 2 o

f2

U

X

op 0 0 c s m O N T t O N

CN O N , O d c n

oo r~- 0 0 O N i n m i n CN oo" d d CO

0 0 2 ^ c^ •n o 0 0 CO

""̂ S S S C-N O C O O C N C N <N p O N r~ r~- >n d C N d

, m N O ( N i n C N N O i n C O o N O i n d C N c<S

C O — O N O C O

i n — - C N m C O

o6 od d —<' C N

N O i n i n CO N O

t--; CO CO

oo' N O d co" —.'

u

.•̂ o

^ i n o 0 0 i n m i n N O d

(N oo i n CN

0 0 r~: d

—< 0 0 vo O N O ^ N O CO CO

C O i n d d •—<

r ~ i n i n 0 0 NO CO —< oo

CO r-' d d d

c < c c .2 • is w

3 O

C .S

c c T 3 T 3

o ^ " «n •'t CN

CN N O d —<' so"

0 \ N O <N N O q ^

N O d CN r^"

«n O N ^ ^ ^

o CO — — m < '̂ i n d csi r t

^ O > s "E

S 3 .S2 3 - „

J O Q ^ ^

3 0 0

O

3 O

II O N O i n O N N O o ( N

i n d o i

i n i n (o —< O N CNJ CO CN d

CN o II O N i n q

1—1 O d (N d r4

ON NO CO o ON oq — 0 0 o

d co' so d d C N ON '—' It

(N CO i n o II

CO oo ON O i n q d (N ON d d

ON i n i n so NO q i n CO 0 \

CN CO CO i n d •Tt' • n ^ ' It 0 0 CO 0 0 i n

II CO O N C N CNl C N (N —• so' d (N

CO

s

1)

-a >^

CO

•S T 3

^ 3 H O

c o

0 0 (u

E 0 0

Table 2.5. Properties of Classroom Questions in Grade 8 Literature Classes

Average Proportion of

Class Type

Number of Questions per Class Session

Asked by Teacher Uptake

Authentic Teacher

Questions No

Response

High-Level

Evaluation

High-Level

Cognition

Mean Cognirive

Level

Proportion of Students

Off Task

Average Class Size

All classes 34.62 (17.37) 0.92 (0.16) 0.11 (0.07) 0.10 (0.11) 0.02 (0.03) 0.03 (0.04) 0.36 (0.15) 2.66 (0.29) 0.05 (0.06) 20.66(7.31)

Low track 36.91 (18.71) 0.97 (0.05) 0.11 (0.08) 0.12 (0.12) 0.04 (0.04) 0.02 (0.04) 0.37 (0.12) 2.69 (0.29) 0.08 (0.10) 18.80 (7.72)

High track 41.97 (20.91) 0.93 (0.09) 0.16 (0.07) 0.12 (0.10) 0.02 (0.03) 0.03 (0.04) 0.40 (0.14) 2.76 (0.27) 0.02 (0.02) 21.08 (7.89)

Urban 29.22 (18.25) 0.89 (0.20) 0.10 (0.06) 0.13 (0.10) 0.03 (0.03) 0.04 (0.05) 0.36 (0.15) 2.68 (0.30) 0.06 (0.08) 21.91 (7.63)

Suburban 42.52 (14.99) 0.96 (0.04) 0.14 (0.07) 0.12 (0.14) 0.01 (0.02) 0.02 (0.03) 0.40 (0.14) 2.70 (0.26) 0.02 (0.03) 20.00 (6.86)

Rural 41.11 (9.22) 0.98 (0.02) 0.11 (0.08) 0.06 (0.05) 0.04 (0.04) 0.02 (0.03) 0.25 (0.12) 2.46 (0.20) 0.04 (0.04) 19.00 (6.58)

Notes: N = 5S classes, including 15 low-track and 13 high-track classes, and 39 urban, 8 suburban, and 11 rural classes. Numbers in parentheses are standard deviations.

Table 2.6. Properties of Classroom Questions in Grade 9 Literature Classes

Proportion of

Average

Class Type

Number of Questions per Class Session

Asked by Teacher Uptake

Authentic Teacher

Questions No

Response

High-Level

Evaluation

High-Level Mean

Cognition Cognitive Level

Proportion of Students

Off Task Average

Class Size All classes 52.77 (20.25) 0.91 (0.10) 0.26 (0.12) 0.27 (0.19) 0.03 (0.03) 0.01 (0.02) 0.46 (0.19) 2.91 (0.38) 0.04 (0.04) 22.74 (6.36)

Low track 48.62 (22.35) 0.94 (0.05) 0.27 (0.13) 0.25 (0.19) 0.03 (0.04) 0.01 (0.02) 0.47 (0.14) 2.95 (0.35) 0.05 (0.04) 15.00 (4.27)

High track 60.05(11.61) 0.89 (0.12) 0.26 (0.11) 0.28 (0.18) 0.03 (0.02) 0.01 (0.01) 0.44 (0.21) 2.85 (0.37) 0.02 (0.03) 25.38 (5.75)

Urban 53.78 (20.80) 0.92 (0.09) 0.27 (0.13) 0.33 (0.19) 0.03 (0.04) 0.02 (0.03) 0.52 (0.23) 2.99 (0.41) 0.04 (0.05) 24.63 (6.63)

Suburban 45.50 (15.28) 0.82 (0.16) 0.24 (0.12) 0.22 (0.16) 0.02 (0.01) 0.00 (0.00) 0.26 (0.14) 2.51 (0.28) 0.05 (0.04) 21.88 (3.23)

Rural 54.83 (22.12) 0.96 (0.04) 0.22 (0.10) 0.09 (0.04) 0.02 (0.02) 0.00 (0.01) 0.45 (0.20) 2.89 (0.40) 0.03 (0.02) 18.80 (5.19)

Notes: N = 54 classes, including 9 low-track and 13 high-track classes, and 35 urban, 8 suburban, and 11 rural classes. Numbers in parentheses are standard deviations.

48 OPENING DIALOGUE

regular cur r i cu lum; i t was as i f low-track students were t o understand a book by dealing only w i t h the index (Page, 1991). O u r data also show that low-track students, i n contrast to high-track students, engaged i n far more clerical as opposed to composi t ional tasks; indeed many o f their so-called " w r i t i n g " tasks, such as filling-in-the-blanks, were no t discourse at all . Thei r w r i t i n g was more formulaic, and the level o f response to their w r i t i n g was low. I n these low-track classes, the terms o f reciprocity were l i m i t e d mainly to procedures.

To examine differences between high- and low-track classes, we compared the t w o groups o n allocation o f instructional t ime and all discourse variables. I n eighth grade, bo th groups spent 4 0 % or more class time d o i n g seatwork. The biggest differences were i n t ime devoted to lecture — the teacher lectured to low-track students 4 0 % more than to high-track students — and to discussion, w h i c h occupied nearly twice as m u c h t ime i n high-track classes than i n low, hence g iv ing a more dialogic tone to the high-track classes (even t hough the p r o p o r t i o n o f authentic teacher questions and uptake d i d no t differ much) . I n n i n t h grade, seatwork occupied almost 13 minutes a day ( 2 9 % o f instructional t ime) i n the low-track classes bu t less than 4 minutes ( 8 % ) i n the high-track classes. B o t h groups spent close t o 9 minutes i n lecture and less than 1 minu te i n discussion, but the high-track classes spent more time d o i n g small-group w o r k and answering questions. (Tables 2.3 and 2.5 show some o f these results for grade 8; Tables 2.4 and 2.6 provide a breakdown for grade 9.)

These results are consistent w i t h our earlier research (Nystrand &: Gamoran, 1988) , w h i c h found that students i n low-achieving classes are far more l ikely than their higher-achieving counterparts t o be involved i n fragmented, contr ived learning. U s i n g survey data, we found that students i n low-achieving eighth- and ninth-grade English classes

• D i d grammar exercises 2.6 times as frequentiy as d i d their h igh-achieving counterparts

• D i d reports 2.4 times as frequently • Fil led i n blanks 5 times as often • Answered true-false questions 4 times as frequendy • Completed multiple-choice questions 4 .1 times as often

I n their responses to the papers o f students i n low-achieving classes, teachers commented

• 2.3 times as much about spelling ( i n marginal and terminal comments) • 1.8 times as m u c h about punctuat ion • 2 times as much about grammar


I n their responses to high-achieving students' papers, however, teachers commented

• 1.7 and 1.9 times as much about content ( in marginal and terminal comments, respectively) compared w i t h teacher comments on low-track papers

Teachers held w r i t i n g conferences w i t h low-abi l i ty students about as infre-quentiy (about once a m o n t h o n average) as w i t h high-achieving students. However , i n these conferences they discussed spelling 2.6 times as much w i t h students i n low-achieving classes, and they discussed content 1.9 times as frequent iy w i t h high-achieving students.

W h y these differences.> To answer this question, we looked closely at i n terviews o f the teachers i n our study and then compared what they said w i t h what we observed i n their classes. First, despite considerable hp service to "discussion," we observed l i t t le discussion i n any classes i n the sense o f i n -depth exchanges o f ideas in the absence o f teacher evaluation. M o s t teachers w h o spoke readily o f their value for "discussion" — and indeed there were a great many—real ly enacted some version o f recitation. Discussion almost always turned ou t t o be what one teacher described as "question-and-answer discussion" invo lv ing a prescripted, teacher-set exchange. Such discussion was rarely collaborative, thoroughgoing , pushed-to-the-limit sharing and exploration o f student ideas unfo ld ing i n class —what B r i t t o n (1970) described as a "struggle to organize . . . thoughts and feelings, to come up w i t h words that . . . shape an understanding, [a] struggle to rise above the l imita t ions o f [ the] language" (p. 12). Wha t mainly varied i n the lessons we observed was the length o f students' responses as they answered teachers' questions.

Some other teachers expressed a conception o f discussion best described as forensic. One teacher, w h o taught academically talented ninth-grade English i n a large urban h igh school, believed that schooling t oo much favors docile, cooperative students; i n contrast, he l iked aggressively expressive, openly assertive students w h o could readily state and defend their points o f view and were w i l l i n g to argue i n class, even w i t h h i m . R igh t answers weren't enough i n his classes, he said; students had to be able to support them and prevail. Needless to say, this conception o f discussion as debate favored the most confident, verbally articulate, and competit ive o f students. This was a view most commonly heard i n suburban schools and was expressed almost exclusively by male teachers (most o f the suburban teachers i n our study were male; most o f the urban teachers were female).

Given these conceptions o f discussion — one recitation, the other debate—it is perhaps no t surprising that students tracked in to low-abi l i ty classes proved hesitant or "reticent" i n the classroom, as their teachers often de-

50 OPENING DIALOGUE

scribed them. I n observations, these students often seemed unsure o f what teachers were look ing for, and responded to teachers' questions w i t h shy, cryptic guesses (marked by r is ing in tona t ion patterns) more often than "answers" (as M r . Schmidt's students d i d ; see Chapter I ) . I n the homogeneously grouped classes, low-abi l i ty students readily stepped ou t o f the way o f the more confident students. M a n y o f them, especially those w i t h reading problems, d i d no t do homework , and some became discipline problems i n class.

W h e n students d i d no t complete reading assignments for homework , many teachers stopped assigning much and devoted substantial propor t ions o f class t ime to reading aloud. They w o u l d pick very short (2 -3 page) stories that could be b o t h read aloud and fo l lowed up w i t h either seatwork or "discussion," all i n the confines o f a 50-minute class period. One teacher t o l d us she deliberately avoided open-ended questions w i t h these students because she felt obUgated to prepare them for standardized reading tests i n the spring. She therefore concentrated o n w h o , what , and w h e n —though no t so m u c h o n w h y — a n d very few o f her questions focused o n the special demands o f literature (as opposed to nonfic t ion informative prose). Newspaper stories migh t have served just as we l l , and i n fact d i d , especially i n many urban classes. Needless to say, the discourse i n these classes was heavily monologic , and students' recollections o f these very short stories, t o say n o t h i n g o f their understanding, were ephemeral, as we learned w h e n we gave our o w n test i n the spring.

One eighth-grade English teacher at a j un io r h igh school i n a small homogeneous midwestern t o w n was typical o f these teachers. She l iked tracking because she felt i t allowed her t o give each o f the groups the attention they needed. She previously had taught mixed-abil i ty classes, and she said the slower students invariably "got lost" because they were "unable to compete w i t h their college-bound peers"; as they became frustrated, moreover, they got in to t rouble. I n her school, she said, the advanced students covered a l o t more. This teacher described her idea o f an ideal lesson as the engaged, authentic discussion o f students expressing and defending their opinions and relating their readings to their personal experience, and she found this far easier t o do w i t h high-track than w i t h low-track students. The high-track students were better readers, so she could reasonably assign more reading for homework ; class t ime, then, was for discussion. The high-track class read four novels; the low-track class read t w o . The low-abi l i ty students needed a l o t o f help w i t h reading, and they often d i d no t complete assigned readings as homework . Hence, she devoted a lo t o f class time to "decoding" and reading aloud; this left l i t t le time for discussion. Moreover, the low-track students were far "more reticent" about expressing their opinions and "less adept" at relating their readings to their o w n experience than were their comparatively confident and "garrulous" high-abi l i ty classmates, so this made i t far easier.


she said, t o engage the high-track students i n discussion. She d i d no t seem to realize that these differences i n pedagogy and interaction afforded the t w o groups o f students significantiy different learning opportuni t ies , especially favoring the high-abi l i ty students at the expense o f the l o w (Gamoran, Nystrand, Berends, & LePore, 1995).

Far more teachers supported abil i ty g roup ing than opposed i t . Those opposing i t were far more l ikely to be i n suburban and rural schools than i n urban schools, where oppos i t ion was very rare. Rural teachers, w h o seemed to view school as part o f extended families i n their small towns, were more concerned than the other teachers about possible negative effects o n the learni n g and self-image o f their slower students, w h o i n tracked groups w o u l d no t "have anyone to model o r peers w i t h g o o d behavior and good sldlls," as one teacher pu t i t . Several suburban teachers particularly were disaffected w i t h high-abi l i ty students, w h o m one described as "pseudo-intellectuals and pseudo-sophisiticates" w h o tended to be "abrasive." Another teacher said, "Sometimes the higher-level kids get an attitude that T ' m s u p e r i o r - I don ' t have to w o r k hard. '" Some suburban teachers felt that m k e d g roup ing could benefit average students. They d iough t i t could improve smdent w r i t i n g by expanding the stock o f shared experience i n the class: " W r i t i n g is based o n experience," said one. Others though t i t could stimulate discussion and i m prove literature instruct ion. Yet these teachers were i n the minor i ty . We interviewed on ly one teacher w h o claimed that whi le mixed-abi l i ty groups m i g h t make i t more difficult t o teach, they also made i t "more fun." "They're more real," she claimed.

M o s t teachers, however, were content to teach tracked classes. Some cand id ly said they were better able to engage the high-abil i ty students than the l o w simply because they had more i n c o m m o n w i t h the high-abi l i ty students; they had more to talk about together. One such teacher taught eighth-grade reading to "basic" students and literature to "academically talented" students i n a midwestern urban jun io r h igh . A t her school, I Q scores played a major role i n assigning students to different abilit)^ groups, and she agreed that her academically talented students were "innately brighter," especially able to manipulate mul t ip le points o f view, and more l ikely than her basic students to go beyond the literal meaning o f their readings. She believed that m i x i n g students o f different abilities shortchanged everyone: I f teachers pitched things toward the higher-abili ty students, she explained, they lost the slower students; i f they made adjustments for the slower students, their classes were no t as s t imulat ing for the faster students. Hence, she felt such students needed to be separated f r o m the basic students "for their o w n good." She preferred teaching the academically talented because she could do more things w i t h t hem —"more fun things and creative things." She spent " too m u c h time" o n discipline in the basic class and found that the students "hated"

52 OPENING DIALOGUE

reading- they had htde interest i n what she was teaching. She said she was more able to be herself w i t h the high-abi l i ty students. I n die final analysis, di is teacher's ideas o n tracking had more to do w i t h her o w n comfor t level than w i t h the learning needs and opportuni t ies o f her students.

Demographic Differences: Urban, Suburban, and Rural Classes

I n our study, the urban classes tended to be more lifeless than the suburban and rural classes. I n interviews, the suburban teachers tended t o sound more progressive than the practices we actually observed i n their classes. The rural teachers had often taught the siblings and sometimes even the parents o f their students, and they tended to view school as an extension o f their students' families.

Urban Classes. Eighth-grade urban students i n our study were involved i n far fewer interactions w i t h their teachers than either suburban or rural students; for the most part, the urban teachers seemed to "manage" students more than teach them i n any conventional way. This was probably no t due to large class sizes i n urban schools since the average number o f students (22) i n these classes was on ly sl ighdy greater than i n either suburban (20) or rural (19) schools. Rather, instruct ion i n urban schools seems to have suffered because the teachers i n them were tired and burned ou t by the difficulties o f dealing w i t h challenging administrative and disciplinary problems, and felt as abandoned by their school systems as the children they taught.

I n interviews, urban teachers frequendy cited problems o f truancy and attendance, discipline, short a t tent ion spans, general apathy and disengagement , and poor reading skills. A c o m m o n attitude was that students "d idn ' t particularly care" and "just d i d wha t they needed to do to get by—these kids don ' t b r ing pencil and paper half the time." These complaints even extended to some academically talented students, w h o , according to one teacher, were the "roughest kids i n the school" and had "the at tent ion span o f a pea . . . they giggle at things even m y lower-abi l i ty kids w o u l d n ' t react to." Some saw links between reading and discipline problems: "That is w h y y o u have so many struggling, so many fail ing, and so many discipline problems."

These teachers frequendy felt orphaned along w i t h their students by lack o f support f r o m central administrat ion. Teachers said their schools spent l i tde money, and even rat ioned paper. They felt stymied by large classes. One teacher said.

The numbers w i l l k i l l you . The numbers! You're no t go ing to teach 26 . That's a b ig class.


Another teacher said her school system had been

very remiss i n coming up w i t h a plan for these basic kids. I n fact, there isn't one. So all they have done is t h r o w these materials at us, and no one has ever sat d o w n and said, " L o o k , this is what y o u should cover. This is what these kids should know," or whatever. They pretty much leave i t t o the teacher. A n d i f you went t o other schools i n the district, they w o u l d be d o ing something total ly different. It's a very poor setup for these kids. A n d they are the ones that need the most strucmre, and they are no t get t ing i t .

Discouraged, these teachers often set remarkably l o w expectations for dieir students. They said they needed books w i d i shorter chapters, lower-level vocabulary, and higher interest levels. One teacher said that m o r n i n g was the best time to teach because her students' concentration was highest and she had fewer behavioral problems then. Students were more l ikely t o be alert and fresh; there were "no clowns to disturb anyone." Even though her first-per iod class d i d no t do much w o r k , she nonetheless judged i t t o be a g o o d one because the students were obedient and " d i d what they were to ld , " and many o f the troublemakers were "no t alert enough to be disruptive." Absenteeism was h igh , and one- thi rd o f her students were usually late. Fridays were always the hardest day o f the week. Discussion was hard. M o s t readings i n the cur r icu lum were irrelevant to students' lives. M o r e than anything, this teacher said, she wanted her students to k n o w she cared about them. After that, discipline — coming to class, do ing the w o r k , and f o l l o w i n g directions — was her most impor tan t goal.

Yet no t all urban teachers had given up. One teacher, specially trained to teach Tide I students, was upbeat. She l iked her students and found the remediation techniques she had learned h ighly effective. She reported no significant discipline problems and found parents cooperative. She especially favored small-group w o r k because there "you can really discuss." She l iked using newspapers because "you could discuss recent developments," and she claimed success when she asked students to wr i te questions about what they were learning. Another teacher emphasized that i t was up to her as the teacher to "discern gifts, b r ing ou t students, make them grow," and that she was able to do this. She looked at students as individuals — "people w o r l d n g w i t h me" — and her students often came back just to "talk about life." She al lowed regular time for " i m p r o m p t u sessions," and said students became engaged when she t ook the t ime to ask, "Wha t do you t h in k o f this.>" Ano the r teacher stressed the importance o f respect: "Teaching has got to be respectful. A n d y o u have to establish that rapport early because the anger and host i l i t ) ' are very difficult t o wear o f f once they start."

54 OPENING DIALOGUE

Nonetheless, most of the teachers in the urban schools said discussion and small-group work were impossible. They generally had resigned themselves to worksheets and newspapers as the best they could offer. Most just tried to keep order in their classrooms, especially through seatwork, which was the dominant mode of instruction, averaging more than 21 minutes a day. On top of this, eighth-grade urban teachers lectured about 6 minutes a day. Even question-answer exchanges involved too much interaction for these teachers, and it averaged only about 9 minutes a day versus 14 minutes for suburban students and nearly 19 minutes for rural students. Unsurprisingly, discussion was rare in these classes, averaging only 45 seconds a day.

Eighth-grade urban teachers asked about as many authentic questions (13%) as suburban teachers did (12%) but twice as many as rural teachers (6%). In ninth-grade classes, these differences were even more pronounced: 33% of teacher questions asked in urban classes were authentic compared with 22% in suburban classes and 9% in rural classes, although in a subsequent analysis (see p. 58), we discovered that many authentic questions did not concern literature or anything academic. Tables 2.5 and 2.6 summarize these and other data.

In grade 9, seatwork in urban classes diminished to about 7 minutes a day, whereas lecture increased to almost 9 minutes a day and question-answer increased to 18. This was the case even though, unlike the eighth-grade classes, the average number of students in ninth-grade urban classes (about 25) was notably higher than either suburban (22) or rural (19) classes. Tables 2.4 and 2.6 summarize these and additional data.

Suburban Classes. Most suburban teachers in our sample said they prized student autonomy: the ability and willingness to articulate and defend ideas. The suburban teachers were clearly more up-to-date on leading-edge pedagogy and professional buzzwords than were the urban teachers. The fundamental province of their instruction, as they saw it, was the life of the mind. Discussion commonly was cited as a goal for all classes; one teacher represented the expressed views of the others when he said, "It is important for students to participate, to feel free to voice their opinions, to present divergent opinions without feeling intimidated, to do the work, be motivated, and be free and willing to discuss literature." His goals for students were excitement and depth of understanding. Another teacher wanted all his students to "discuss, contribute, and offer original ideas."

In fact, we observed more teacher-student interaction in the suburban than in the urban schools, especially in eighth-grade classes. The suburban teachers did less lecturing than the urban (4 minutes a day versus 6), and their students did less seatwork (15 minutes a day versus more than 21). Yet


"discussion" in these schools was rarely open-ended and generally took the form of question-and-answer recitation: Eighth-grade suburban classes t)^pi-cally had more than 14 minutes a day of question-answer activities versus less than 10 minutes in urban classes, and actual discussion was about 50 seconds a day on average, not much more than in urban schools. In ninth-grade suburban classes, we encountered not even a second of discussion.

If suburban teachers prized student autonomy and independent thought, they also said they prized sharing and cooperation. Many teachers used small groups to promote such values, and in both eighth- and ninth-grade classes, suburban students spent twice as much time as urban students in small groups.

Reading skills of suburban students were stronger than those of urban students, and this difference was evident in the fact that suburban students spent less time reading aloud in class. The eighth-grade classes spent about 2V2 minutes a day reading aloud compared with more than 3 minutes a day in the urban classes; in the ninth-grade classes, this difference became even more pronounced: The suburban classes spent only 43 seconds reading aloud, whereas urban classes spent more than 2 minutes. While urban teachers often sought simplified reading materials, suburban teachers sometimes did just the opposite. As one told us, "Last year I used a basic reader. . . . There were good stories, but they were so watered down that I said I can't use this book. I realize that the stuff I use now, the kids are frustrated with the vocabulary, but that's life." The suburban teachers said they spent more time on interpretation than on reading skills. They told us they encouraged students to give opinions, explain why, and go "beyond remembering" to "using" texts. Yet the cognitive level of questions asked in eighth-grade suburban classes was no higher than in urban classes, and in ninth grade it was actually lower. Nor did suburban teachers ask more authentic questions: In eighth grade, about 12-13% of all teacher questions were authentic in both suburban and urban classes; in ninth grade, urban teachers asked about 50% more. Tables 2.3-2.6 summarize these and other data.

Rural Classes. Schools in small, rural towns were a lot like the towns themselves — places where everybody knew everybody and where secrets could sometimes be hard to keep. Teachers often knew the siblings of their students and kept in close touch with the parents: One teacher regularly recognized his students as "chips off the ol' block." Because the same students often spent the day together through all their classes, they often knew each other's grades, and this was sometimes a problem. Also, because the schools were so small, there was often no ability grouping; everyone was just mixed together. Virtually all the teachers approved of this. They felt that "the slower

56 OPENING DIALOGUE

kids don' t learn as much i f they're grouped," and believed that mixed groups provided these students w i d i role models. They also said i t was impor tan t for the smarter kids to "learn the other side o f life."

Teachers t o l d us they prized "question-and-answer discussion," w h i c h averaged approximately 18 minutes a day i n bo th eighth and n i n t h grades. We found more open-ended discussion i n eighth-grade rural classes than i n anv o f the other schools, o n average a l i tde less than 2 minutes a day Also, seatwork i n grade 8 occupied m u c h less time i n rural than i n urban and suburban schools. Tables 2 .3-2 .6 summarize these and other data.

ANALYZING THE DATA: THE EFFECTS OF DISCOURSE ON LEARNING

H o w concerned should we be w i t h these bleak figures? O u r data enabled us to test specific hypotheses concerning the overall effects o f dialogic elements on learning. There is clearly a trade-off between research, such as our study, diat comprehensively depicts the b ig picture and examines general effects, o n the one hand, and case studies that examine the dynamics o f i nd i vidual cases and episodes, o n the other. Ideally these different perspectives complement each other. O u r large study enabled us t o test empirically many widely debated hypotheses about the effectiveness o f different instructional practices and discourse environments (e.g., recitation, discussion, small-group w o r k ) for learning.

I n a series o f analyses, we examined the effects o f many o f these practices on literature achievement. To assess student learning, we administered a literature test i n the spring to each class. The test required students t o answer a series o f questions about five works o f literature (stories, novels, dramas, and short plays) they had read du r ing the year.^* The questions ranged f r o m simple recall (e.g., " W h o were the main characters i n Roll of Thunder, Hear My Cry?') t o ones requir ing in-depth understanding ("Relate the conflict o f Roll of Thunder, Hear My Cry to the ending and to the theme"). The same types o f questions were asked o f each class, bu t the stories varied, depending on what students actually read du r ing the year. For the ninth-grade test students also wro te a br ie f essay o n a character f r o m their reading w h o m they admired. A n example o f an eighth-grade test is found i n Appendix A .

The literature tests were scored for the fo l l owin g :

1. Extent o f recall 2. D e p t h o f understanding 3. N u m b e r o f endings remembered 4 . Relation o f ending to denouement 5. Relation o f conflict/and or ending to theme


6. Understanding o f the internal motivat ions o f characters 7. Interpretive treatment o f the major selection 8. Level o f discourse used to discuss theme and conflict

Readers read the entire test and then determined a single score for each o f the above variables. Each student's literature score was the sum o f the ind i vidual scores. Each test was scored by t w o readers and the scores were averaged. The overall reliabili ty o f the assessment i n grade 8, computed as a correlation o f the t w o readings, was .90; i n the grade 9 assessment, the reliabihty was .82. The rubric used for scoring the tests is shown i n Appendix B.

We then examined the effects o f instruct ion and classroom discourse o n learning t h rough a statistical technique called regression analysis. Regression analysis makes i t possible to examine the effect o f one cond i t ion (e.g., the amount o f time spent i n discussion) whi le statistically ho ld ing constant other impor tant condit ions (e.g., pr ior abilities m w r i t i n g and readmg, socioeconomic status, and characteristics o f students i n diflferent classes). For example, i f authentic questions are found to be related to higher achievement, regression analysis can reveal whether this is because teachers ask previously higher-achieving students more authentic questions, or whether authentic questions actually p romote higher achievement. Regression analyses systematically estimate the effect o f each variable whi le statistically ho ld ing constant each o f the other variables. O u r analyses control led for the effects o f bo th background variables (sex, race, ethnicity, family socioeconomic status) and pr ior achievement (as measured by fall tests o f reading and w r i t i n g skills).

Overall Results

I n none o f our analyses d i d we ever find that a higher cognitive level o f instructional activities actually enhanced learning. Instead, we could explain the relative effectiveness o f different instructional practices on ly when we examined the ways teachers and students interacted as evidenced by authentic questions, uptake, and especially discussion.

Eighth-Grade Classes. The results o f our analyses o f the eighth-grade classes, control led for w r i t i n g and reading ability, socioeconomic status ( S E S ) , race, and ethnicity, showed unsurprisingly that disengagement, inc luding off-task behavior and failure to complete homework , adversely affected achievement. The results revealed a modest effect for time spent o n homework , no effect for h o w many questions were asked i n class, and a negative effect for level o f activity du r ing recitation. Results indicated that dialogically organized instruct ion, indicated by time devoted to discussion, authentic questions, uptake, and high-level teacher evaluation, had a strong, positive effect o n

58 OPENING DIALOGUE

achievement. Discussion i n particular had a large effect, w h i c h is especially s t r iking when i t is recalled that the average class engaged i n less than a minute o f discussion each day Table 2.7 summarizes these results.

I n our eighth-grade study, we also found that effective teachers o f literature regularly assigned extended pieces o f exposit ion (Nystrand, 1991c); this practice enhanced students' recall and understanding o f the literary works they read. The frequent assignment o f short-answer exercises, however, actually degraded students' overall recall and depth o f understanding. This result is consistent w i t h Applebee's (1984) content ion that, because w r i t i n g tends to promote recall o f what i t focuses on , such "narrow-banded" activities as short-answer exercises are l ikely to hinder total recall — i n other words , helping students t o remember trees at the expense o f understanding the overall shape o f the forest. I n addi t ion , because they elicit cryptic, fragmented discourse, short-answer exercises p romote superficial involvement w i t h literature; i n so do ing , they trivialize students' experiences w i t h literature. A l l i n all , students learn literature best i n classes that encourage substantive and personal student response to literature i n bo th classroom interaction and w r i t i n g . Table 2.8 summarizes the results.

Ninth-Grade Classes. I n analyzing data f r o m the n i n t h grade (Gamoran & Nystrand, 1992), we sought t o replicate our findings f r o m the eighth-grade study, bu t we were in i t ia l ly frustrated to find that discussion had no effect o n learning i n n i n t h grade and that authentic questions appeared to have no effect or even a negative effect. We were further perplexed to discover that authentic questions had positive effects i n high-track classes bu t negative effects i n low-track classes. Table 2.9 summarizes these results.

L o o k i n g more closely at our data, we discovered that the t w o different tracks used authentic questions very differendy. I n the high-track classes, ful ly 6 8 % o f authentic questions concerned literature, whereas on ly 2 5 % o f authentic questions i n low-track classes d i d . I n low-track classes, teachers' authentic questions often concerned such issues as, " H o w do most o f y o u feel about tests?"; "Wha t w o u l d your parents say i f y o u go t an A o n next week's test?"; "Wha t things w o u l d y o u associate w i t h l y ing i n the sun?"; " D o y o u ever have to take notes?" Discussion broke d o w n i n a similar way so that discussion i n the high-track classes tended to be about literature far more than d i d that i n the low-track classes.

Tracking, Instructional Discourse, and Learning

Again and again, researchers have found that tracking and abil i ty grouping p romote inequality o f achievement, as the gap between students i n h igh-track and low-track classes widens over t ime (Gamoran 8c Berends, 1987).

Table 2.7. Effects of Disengagement, Procedural Engagement, and Substantive Engagement on Spring Literature Achievement in Grade 8 (metric regression coefficients)

Model

Independent Variable Background

Sex (1 = female)

Race (1 = Black)

Ethnicity (1 = Hispanic)

SES

Grade (1 = eighth)

Fall reading score

Fall writing score

Disengagement Off task in class

Reading not completed

Writing not completed

Nonresponse to questions

Procedural engagement Active in class

Asking questions

Time on homework

Substantive engagement Authentic questions

Authentic reading

High evaluation of writing

Uptake

Coherence of reading

Discussion time

Small- group time

Procedural Substantive Background Engagement Engagement Full Model Variables Only Variables Variables (All Variables')

0.44 (0.38)

-2.67" (0.66) -1.51" (0.64) 1.62"'

(0.24) 2.09"'

(0.57) 0.39"*

(0.04) 0.93"*

(0.15)

0.47 (0.36) -1.55" (0.64) -0.15 (0.62) 1.10"

(0.24) 1.10*

(0.56) 0.30"*

(0,04) 0.70*"

(0.14)

-0.23"* (0.03) -0.02 (0.01) -0.03*" (0.01) -0.20"*' (0.06)

-0.03" (0.02)

-0.02 (0.02) 0.45*"

(0.17)

0.62* (0.35) -1.75*" (0.62) -1.47" (0.60) 1.45*"

(0.23) i . i r *

(0.57) 0.36"'

(0.04) 0.73**"

(0.14)

.322 .399

0.05*"' (0.02) -0.01 (0.03) -0.39 (0.40) 0.14**"

(0.03) 0.16**"

(0.03) 0.34*"

(0.11) -0.19" (0.07)

.428

0.59* (0.34) -1.10* (0.61) -0.58 (0.59) 1.05""

(0.23) 0.16

(0.56) 0.30'**

(0.04) 0.58"*

(0.14)

-0.16"* ( 0.03) -0.02* (0.01) 0.02"

(0.01) -0.22*"* (0.06)

-0.06"" (0.02) -0.03 (0.02) 0.38"

(0.16)

0.04" (0.02) 0.02

(0.03) -0.03 (0.40) 0.10'"

(0.03) 0.11""

(0.03) 0.29*"

(0.11) -0.23"" (0.07)

.475 Source: Nystrand & Gamoran, 1991a. Notes: Dependent variable; Spring Literature Achievement Test N = 924 students Missina values deleted listw.se. Numbers in parentheses are standard errors ^

P < .10 (margmal trend) ** p < .05 • p < .01 **** p < .001

Table 2.8. Effects of Selected Instructional Variables on Difficulty of Recall and Difficulty of Understanding Literature in Depth in Grade 8 (metric regression coefficients)

Difficulty Difficulty of of Recall Understanding in Depth

Background variables Grade (1 = eighth) 0.054 -0.183

(0.090) (0.118) Race (1 = Black) 0.246 0.068

(0.091) (0.120) Ethnicity (1 = Hispanic) 0.122 0.165

(0.089) (0.117) SES -0.107**** -0.138***

(0.036) (0.048) Sex (1 = female) -0.120** -0.174*

(0.052) (0.069) Fall writing score -0.073**** -0.105****

(0.021) (0.027) Fall reading score -0.034**** -0.036****

(0.005) (0.007) Procedural variables

Time on homework -0.065*** -0.058* (0.024) (0.031)

Reading not completed 0.0004 0.0004 (0.002) (0.003)

Writing not completed 0.005** 0.002** (0.002) (0.003)

Participation in class -0.001 -0.001 (0.002) (0.002)

No response to teacher questions 0.008 0.026* (0.010) (0.013)

Instructional variables Amount of writing -0.256** -0.289**

(0.104) (0.137) Discussion time -0.022 -0.038*

(0.016) (0.022) Authenticity of teacher questions -0.0005 -0.006**

(0.002) (0.003) Authenticity of readings -0.003 -0.018***

(0.005) (0.006) Uptake -0.015**** -0.020****

(0.005) (0.006) Relating discussions to other discus 0.005 0.001

sions and student compositions (0.004) (0.005)

Relating readings to other readings -0.046** -0.018 (0.019) (0.025)

.353 .339 Source: Nystrand, 1991c.

Notes: N = 762 students. Missing values deleted listwise. Numbers in parentheses are standard errors.

* P < .10 (marginal trend) ** p < X)5 *** p < .01 **** p < .001

Table 2.9. Effects of Instruction and Engagement on Ninth-Grade Literature Achievement

Regression Variable Mean SD Coefficient SE Background variables

Sex (1 = female) 0.51 0 .50 1.47** 0 .37 R a c e (1 = b lack) 0 .07 0 .26 - 0 . 4 6 0 .72 Ethnic i ty (1 - Hi spanic ) 0 ,09 0 .28 - 1 . 5 6 * 0 .65 S E S - 0 . 0 2 0 .80 0 .44 0 .25 F a l l reading score 31 .88 5 .34 0 .40** 0 .04 F a l l writ ing score 5.71 1.28 0 .90** 0 .15

Abi l i ty groups

Honors /acce lerated 0 .24 0 .43 0 .25 0 .96 Bas i c / remedia l 0 .10 0 .30 - 1 . 0 9 1.13 Other" 0 .09 0 .29 0 .57 1.16

Psycho log ica l engagement

Engagement scale*" 2 .42 1.40 0 .02 0 .14 Behav iora l engagement

Writ ing completed 87 .88% 19.68 0 . 0 3 * * 0.01 Reading completed 83 .04% 2 4 . 6 2 0 . 0 3 * * 0.01 H ome wor k time ( h r s / w k ) 1.27 1.27 0 .19 0 .15 O f f task in class 3 .22% 3.27 - 0 . 1 2 * 0 .06

Instructional discourse

Authentic questions In honors classes 2 4 . 3 0 % 11.41 0 .10** 0 .03 In regular classes 2 8 . 1 3 % 18.81 - 0 . 0 2 0.01 In remedial classes 27 .40% 18.86 - 0 . 0 9 * * 0 .04 In other classes 36 .90% 2 6 . 0 3 - 0 . 2 0 * * 0 .03

Uptake 2 5 . 9 0 % 11.26 0 .09** 0 .02 Di scuss ion (min/day) 0 .24 0 .48 - 0 . 1 8 0 .40 Coherence*^ 13.01 7 .07 0 . 1 2 * * 0 .03

Source: Gamoran & Nystrand, 1992. Notes: N = 971 students. Dependent variable: Spring Literature Achievement Test (mean =

21.82, SD = 7.66). = .52. " Other classes include two classes in a school-within-a-school program and two classes in a

heterogeneously grouped school. Engagement scale based on student responses to the following questions, coded on a weekly scale: In Engl ish class, how often do you: (a) Try as hard as you can? (b) Think what you are learning is interesting and worthwhile? (c) Find yourself concentrating so hard that time passes quickly? Coherence measure based on teacher responses to the following questions, coded on a weekly scale: (a) About how often do students in your class write about (or in response to) things they have read? (b) About how often do you discuss writing topics with your students before askmg them to write? (c) About how often do you and your class discuss the readings you assign? (d) When you ask students about their reading assignments in class, how frequendy do you attempt to do each of the following: A s k them to relate what they have read to their other readings? (e) About how often does your class relate its discussion to previous discussions you have had? (f) About how often do you and your class discuss things students have written about?

* p < .05 **p < .01

62 OPENING DIALOGUE

A major purpose o f our investigation was to determine wiiether inequali ty resulted f r o m differences i n the quali ty o f discourse i n high-track and low-track classes. This question is no t easy to answer, for t w o reasons. First, p r ior to our study, quantitative measures o f discourse quali ty had no t been available. Second, i t is hard to measure the effects o f tracking because they are confounded w i t h in i t ia l differences among students. Students assigned to different tracks have different achievement f r o m the start, and i t is impor tan t t o take this i n to account w h e n examining inequality.

We addressed the first p rob lem by developing the scheme for measuring classroom discourse described i n this chapter. We addressed the second problem w i t h statistical controls for fall reading and w r i t i n g skills, as we l l as controls for sex, race/ethnicity, and S E S (see note 15). I n addi t ion, for this analysis, we used standardized reading and math test scores as indicators o f an under lying "ab i l i ty" construct, t o add a stricter cont ro l for students' p r ior academic background (Gamoran et al., 1995). Even w i t h these controls, we found that average achievement was almost t w o points higher i n honors classes than i n remedial classes, a statistically significant difference. One standard deviation o n the test was 6.8 points, so the t w o - p o i n t gap was almost 3 0 % o f a standard deviation, a substantial amount (Gamoran et al., 1995).

D o differences i n instructional discourse account for this achievement inequality.^ Yes, to a large extent (Gamoran et al., 1995). Part o f the difference occurred because students i n honors classes were more responsive — they completed more o f their reading and w r i t i n g assignments — w h i c h helped them learn more (see Table 2.10) . Uptake and discourse coherence also helped students learn, bu t these d i d no t account for track differences i n learning , because they occurred at similar rates i n honors and remedial classes (see Tables 2.5 and 2.6) . Authent ic questions also were comparably dis t r ibuted across class types, but they were beneficial only i n honors classes because there they pertained to literature (see Table 2.10). Hence, similar levels o f authent ic i ty led to inequality o f learning, because the content o f the authentic questions differed.

We further observed that students were more often o f f task i n remedial classes, and this was detr imental t o their achievement (see Table 2.10) . I n honors classes, where off-task behavior was less prevalent, its effects were less harmful . Conversely, discussion occurred more often i n honors classes, where i t b rought benefits for literature achievement, and had no benefits i n other classes, presumably because i t focused o n topics other than literature.

These results contradicted the usual rationale for abil i ty g rouping . A b i l i t y g roup ing is supposed to al low teachers and students to engage i n whatever instruct ion is most beneficial i n each context. I n actuality, we found that authentic questions occurred at similar rates i n all classes, but contr ibuted to achievement on ly i n honors classes; off-task behavior was most c o m m o n


Table 2.10. Maximum Likelihood Estimates of Background and Instructional Effects on Literature Achievement in Eighth- and Ninth-Grade Ability-Grouped English Classes Independent Variable Effect SE Background

Sex (1 = female) I.ISS'' 0.252 Minority (1 = Black or Hispanic) -0.652 0.339 SES 0.155 0.174 Fall reading score 0.202" 0.024 Fall writing score 0.512" 0.103 Ability 0.12r 0.018

Instruction Completion of reading 0.022*' 0.006 Completion of wridng 0.025'' 0.007 Off task in class

Honors classes 0.149 0.092 Regular classes -0.193" 0.044 Remedial classes -0.124" 0.028

Authentic teacher questions Honors classes 0.056'-' 0.022 Regular classes 0.000 0.010 Remedial classes' -0.050'-' 0.017

Uptake 0.063" 0.013 Discussion

Honors classes 0.277'= 0.129 Regular classes -1.510'= 0.591 Remedial classes 0.045 0.169

Discourse coherence 0.158" 0.022 Intercepts

Honors classes -8.502" 1.385 Regular classes -7.081" 1.207 Remedial classes -7.144" 1.061

X^(61,iV= 1,564) = 86.33

Source: Gamoran et al., 1995 Coefficient is four times its standard error. Coefficient is three times its standard error. Coefficient is twice its standard error.

where i t was most harmful , i n regular and remedial classes. Discussion fit the pattern o f occurr ing more where its payoff was greater ( in honors classes); this, too , cont r ibuted t o inequality.

I n l igh t o f these results, we concluded that t w o changes must be considered (Gamoran et al., 1995). Ei ther abi l i ty g roup ing i n eighth- and n in th -grade English should be eliminated, o r i t should be implemented very dif-

64 OPENING DIALOGUE

ferently than typically occurs at present. To mit igate the inequali ty that results f r o m g r o u p i n g and tracking, teachers and students i n regular and low-level classes need higher expectations and more engaging discourse that focuses o n academic subject matter.

Small Groups

I n the eighth-grade study, we in i t ia l ly expected small groups to be dependable sites o f dialogic peer interaction, but this was naive. I n fact, t ime spent i n small groups, rare as i t was — 39 seconds a day o n average — turned out to have a p rominen t negative effect o n achievement. We had predicted that small-group w o r k and discussion w o u l d enhance achievement by engagi n g students substantively, especially compared w i t h lecture, d r i l l w o r k , and recitation, w h i c h , like short-answer study questions, typically involve abbreviated responses f r o m students. I n fact, increased t ime spent i n small-group w o r k seemed to result i n lower achievement i n literature. We decided t o examine this f inding i n more detail i n a fo l low-up study o f ninth-grade literature classes. We were particularly interested i n dis t inguishing various kinds o f small-group w o r k and t o see whether some were more effective than others.

I n the ninth-grade study, unlike the eighth-grade study, we audiotaped the classes we observed, and al though we had no t placed microphones i n small groups, the tapes provided a record o f h o w each class proceeded. For classes invo lv ing small-group w o r k , we were able to determine: (a) what preceded and what fo l lowed this w o r k , (b) the kinds o f tasks undertaken i n groups, (c) the instructions the teachers gave to the groups, and (d) the roles teachers played i n setting up and runn ing the groups.

We learned that small-group w o r k involved a great range o f activities. Some small-group w o r k was so highly structured by the teacher — involv ing , for example, students comple t ing worksheets together — that i t m i g h t best be called "collaborative seatwork." Other groups, w h i c h we called problem-solving groups, required students t o come t o consensus concerning some issue or question that the teacher defined. Yet other small-group w o r k , w h i c h we described as "autonomous," was even more open-ended, w i t h the groups themselves defining as wel l as resolving the problems and issues they discussed. Figure 2.2 shows the con t inuum described by these categories.

O u r study looked exclusively at the features o f small-group w o r k that p romote th in ldng. Given our findings that ownership, coherence o f discourse, and student p roduc t ion o f knowledge are impor tan t features o f effective instruct ion, we reasoned that t o promote response t o and th ink ing about literature, small groups also should manifest these traits. M o r e specifically, we predicted that collaborative seatwork, wh ich is essentially w r i t t e n recitation done by students w o r k i n g together, w o u l d be less effective than problem-

LANGUAGE AND LEARNING IN HUNDREDS OF ENGLISH LESSONS

Figure 2.2. Continuum of Small-Group Work

65

Collaborative Structured Autonomous seatwork problem problem

solving solving

solving and autonomous groups, w h i c h al low ownership and thereby maximize the possibili ty o f coherent discussion. I t may be that collaborative seat-w o r k is effective for teaching facts and grammar, bu t ou r research d i d no t examine this possibility.

As we examined the range o f activities that we had coded as small-group w o r k , we confirmed our in i t ia l impression that "small-group t ime" was a misleading category because o f impor tan t differences among the activities that occurred i n small groups. We can illustrate these contrasting activities by focusing o n t w o dimensions o f w o r k i n small groups: student autonomy and student p roduc t ion o f knowledge. Small-group activities occurred i n on ly 29 o f the 216 class sessions we observed. The small-group sessions averaged about 15 minutes w h e n they occurred, but because they were so infrequent, small-group t ime accounted for less than 2 minutes o f the average 50-minute period among all the classes. The g r o u p w o r k we observed was usually closer to collaborative seatwork than to student problem-solving and autonomous g roupwork .

Student Autonomy. Teachers shape g r o u p w o r k by assigning tasks and establishing parameters o f interaction. I n highly "prescripted" g roupwork , the parameters are defined entirely by the teacher, and the task could just as easily be done w i t h o u t interaction among students. The g roup setting is gratuitous, used perhaps more for the teacher's convenience. Here is an example f r o m a ninth-grade English class.

Today whi le y o u are w o r k i n g i n groups y o u w i l l keep the same groups that y o u had yesterday. The same rules w i l l apply, and those are: You must, as a group, f o r m a t igh t circle; y o u must b r ing all o f your materials w i t h y o u to that g roup and y o u may N O T get up f r o m your seat; your voice must stay at a whisper — i f I can hear y o u above anyone else's, that means y o u are t o o l o u d and your name w i l l go up o n the

66 OPENING DIALOGUE

board. You may do one o f t w o things i n your group. You may continue to w o r k o n your paper —there are five, probably six people I have to see yet i n conference. I f y o u are as far as y o u can go i n your groups w i t h your papers — that is, edi t ing, proofreading, all o f those — then as a group, I w o u l d hke y o u t o see i f y o u can fill i n the blanks o n this handou t o n five basic sentence patterns: h o w t o find them, what questions to ask. A n d we w i l l go over this. Remember one section o f your binders should be sentence patterns.

This g r oupwor k , so completely structured by the teacher, promotes neither ownership nor coherent discussion.

I n more autonomous groups, the teacher gives students some lati tude i n their interactions w i t h each other, and a l though students may remain o n a "short leash," the g r o u p w o r k nonetheless displays spontaneous student interactions concerning the substance o f the lesson. I n the most autonomous groups, the teacher clearly defines group tasks but w i t h o u t prescripting the g roupwork . Typically the teacher (a) defines the goal o f the group , for example, a r r iv ing at a consensus concerning some controversial issue; (b) oudines the tasks to be accomplished, for example, the g roup composi t ion o f a letter ou t l i n ing the group's views to a public official; and/or (c) assigns roles t o g roup members, for example, t w o group members argue one side o f the issue whi le t w o others argue the other side and one student acts as the recorder. I n the f o l l o w i n g transcript, for example, the teacher initiates g r o u p w o r k after students have w r i t t e n some or iginal verse. Referring to the students' poems, the teacher says:

I f yours is the best i t can be — instead o f coun t ing off, because we ' l l run ou t o f t ime, w i l l y o u just g roup together w i t h the three or four —three or four m a x i m u m — people around you . Read them over, and choose the one [poem] t h a t . . . looks the most specific t o y o u — see that person that the)^re ta lk ing about.

A l i t t le later, this teacher reminds students that she wants

to hear what you've picked ou t in your groups so that you're hearing some o f the real s trong images o f these people. . . . I ' m asking y o u to brag; I want t o hear the three or four R E A L g o o d ones, so people w h o are having difficulty get a g o o d sample before they go home.

This skillfully organized g r o u p w o r k starts w i t h expressive student w r i t i n g , thus p r o m o t i n g ownership. N o t only because students must w o r k together to find g o o d examples bu t also because the teacher encourages them to articu-


late wha t " g o o d " means i n this context, the ensuing g roup talk is more likely to be coherent conversation than the first session, where students must " f i l l i n the blanks o n this handout o n five basic sentence patterns." Such discussions were rare i n our study: O n l y 1 1 . 1 % o f all small-group w o r k was judged either to be w h o l l y autonomous or t o display significant student interaction i n p roduc ing the outcome o f g roupwork , whereas 7 0 . 4 % o f all small-group w o r k was prestructured by the teacher.

The degree o f student autonomy i n the small groups we observed was coded f r o m audiotapes according to the fo l l owing scale:

1 Teacher-structured groupwork: Task parameters entirely defined, or "prescripted," by teacher. Task can be done w i t h o u t student interact i o n (e.g., worksheets); g roup setting is gratuitous.

2 Prescripted task w i t h obl igatory student interaction. 3 Limited student interaction: G r o u p w o r k involves spontaneous student

interaction concerning substance bu t students are o n "short leash." For example, the teacher m i g h t define some general principle that students i n groups must then apply.

4 Significant student interaction defining shape o f task and outcome, alt h o u g h teacher m i g h t have been able to predict results before class.

5 Autonomous groupwork: Teacher sets up g r o u p w o r k w i t h o u t prescript ing activities; significant student interaction define shape o f task and outcome. Results o f g r o u p w o r k cannot be predicted before class.

Student Production of Knowledge. I n addi t ion to p r o m o t i n g coherent conversation, the second example encourages students t o generate insights and understandings far more than the first session. I n activities such as complet ing worksheets and answering study questions, students are required mainly to manipulate and master in format ion provided by the teacher or a textbook. This w o r k calls only for lower-order cognitive activity, characterized by questions w i t h prespecified r igh t and w r o n g answers, and i t tends no t t o be very coherent because i t does n o t b u i l d o n student responses. I t consequently fails to result i n student p roduc t ion o f knowledge. Here is an example.

Everyone pu t their name i n the upper r ight . Put your name and your per iod and the g roup number i n the upper r ight-hand corner. N o w I've gone over t w o or three times your g roup number so y o u should remember i t —when y o u get in to your groups, i f y o u have forgot ten i t , maybe somebody i n there w i l l remember i t . N o w y o u are to combine eight sentences to f o r m the k i n d o f sentence shown i n parentheses, okay? A n d I mean that . . . when I say a simple sentence, that's wha t I mean: a

68 OPENING DIALOGUE

simple sentence that has one subject and one predicate. W h e n I say comp o u n d sentence, I mean t w o sentences o f equal value and equal importance, pu t together w i t h a conjunct ion or a semicolon or a transitive adverb, okay? Complex sentences: y o u w i l l be asked to either wr i te one w i t h an adjective clause o r an adverb clause. A n d because there were n o n o u n clauses, I have given y o u a task d o w n below here t o wr i t e t w o sentences w i t h n o u n clauses. N o w , all o f the sentences up above — the first eight — are all about Niagara Falls and the river and stuff like that, and though i t isn't absolutely impor tan t that y o u focus o n the Niagara River, i t w o u l d be k i n d o f nice i f y o u d i d i t , all right?

I n another example, f r o m another class, the teacher puts students t o w o r k i n groups o n a worksheet concerning Greek gods.

I k n o w y o u won ' t finish the entire chart, bu t we w i l l be starting . . . sharing, because what we wan t to be sure is that everyone has the same items o n the chart. So I w i l l give y o u the next 15 minutes . . . t o just w o r k o n what y o u have. Be sure i f y o u haven't — some o f y o u maybe start f rom the back; we w o n ' t get them all checked today so that y o u have the chart. This is the chart that y o u w i l l have to memorize, y o u w i l l be tested o n i t . . . I want t o make sure y o u all have the same i n format ion.

I n b o t h these examples, the g r o u p w o r k involves l i t t le o r no p roduc t ion o f knowledge. I n other groups, by contrast, students must sustain a coherent discussion i n order t o w o r k ou t problems that generate new understandings. Typically, this g r o u p w o r k proceeds i n response t o open-ended questions w i t h more than one acceptable answer and involves higher-order cognitive activity. I n the fo l l owing transcript, for example, the teacher asks students to predict the ending o f an Agatha Christie novel . To get them th ink ing , she first asks them to wr i te a br ief p lo t summary.

I n a paragraph, wr i t e o u t wha t happened, what y o u th ink happened — and talk i t ou t first, because there are a l o t o f lights go ing o n i n your eyes, and some o f y o u are st i l l saying, " I have no idea." Get together i n your g roup , and o n a sheet o f loose-leaf tell me what happened.

As students w o r k , the teachers says:

As I came around, a l o t o f y o u picked up a lo t o f the clues as t o h o w she m i g h t have been ki l led and, so then, w h o w o u l d have done i t . W h o ever you're accusing, t h i nk about a mot ive —so some o f y o u are saying.


" O h , the inspector d i d i t , " o r "The colonel d i d i t , " o r "Pollet d i d i t , " o r "Ted d i d i t , " and that's all you're w r i t i n g d o w n . W h a t I came around and asked y o u for was a mot ive , right? Miss Marple's go t that d o w n . So i n your paragraph: w h o d i d i t and a mot ive .

I n another lesson, f r o m another class, the teacher asks her students t o analyze the characterization o f M r . M o r r i s o n i n Roll of Thunder, Hear My Cry.

Here's wha t y o u need to do. . . . First o f all , y o u want t o name three outs tanding character traits. N o w , remember traits refer t o personality, no t physical characteristics. . . . Give suppor t ing quotations for your ideas, one quote for each trai t w i l l be fine, and then give a warrant— that is, explain h o w the quote sets up that trait or h o w i t establishes that t rai t , and then te l l me wha t technique [the author] is using. Does she use a character's language, a character's actions, o r do y o u see the character t h rough the eyes o f another character, or the reaction o f another character?

I n each o f these latter t w o examples, students must no t on ly identify some under lying principle —a motive for murder, a character t r a i t — b u t also find support ing evidence for the interpretation. Once again, we discovered that such g r o u p w o r k is infrequent; only about a quarter o f the g r o u p w o r k we observed involved discussion o f open-ended questions w i t h students actively constructing interpretations (rated a 5 o n the scale i n the list that fol lows). By contrast, two- th i rds was a version o f collaborative seatwork.

I n our analysis o f the extent to w h i c h students were required to actively construct new understandings du r ing small-group work , we coded each session, using audiotapes, according to the f o l l o w i n g scale:

1 Collaborative seatwork. Students receive only prespecified informat i o n ; student activities i n small groups are t igh t ly control led. Work sheets are an example.

2 Students pr imar i ly receive prespecified in fo rmat ion ; occasionally their tasks involve open-ended questions. Student activities are h ighly orchestrated, w i t h topic coverage defined by the teacher.

3 Teacher identifies issue or p rob lem; students find examples and explore applications. Teacher provides t ime and encouragement for exp lo r ing new meanings and implications.

4 Students receive some in format ion ; much o f the g r o u p w o r k involves students using in format ion w i t h open-ended questions or w i t h teacher-provided questions.

5 Students identify b o t h problems and applications. Discussing open-

70 OPENING DIALOGUE

ended questions, students aaively construct interpretations. Teacher sets the parameters of the groupwork, but it is mainly the students who work out ways to address issues and answer questions.

Effective and Ineffective Croupwork. D id small-group work affect achievement, holding other conditions constant .> In our regression analysis, we held constant students' sex, race/ethnicity, SES , and fall reading and writing skills. We also took into account teachers' uses of authentic and follow-up questions, the amount of discussion time outside of small groups, as well as rates of student off-task behavior and homework completion.

We first discovered that, overall, classes spending more time in small groups produced lower achievement, a finding that replicated the surprising conclusion of our eighth-grade study. However, we were now prepared to go beyond simply asking how much time was spent in small groups: We could ask whether the effectiveness of small-group time depended on what was going on in the small groups. To do this, we used the two measures of the quality of group time discussed earlier: student autonomy and student production of knowledge. Although these measures differ in theory—students could be given autonomy but not take the opportunity to produce knowledge — in practice the two tended to occur together: the more student autonomy, the more production of knowledge (the correlation was .78).^^

Analysis of student autonomy showed that the higher the degree of autonomy, the more likely group time was to contribute positively to achievement. For example, a class that averaged 5 minutes per day of highly "pre-scripted," rigidly structured group time, with little autonomy for students, would actually score about 1 point lower on the test than a class with no group time at all. By contrast, a class that averaged 5 minutes of highly autonomous groupwork, in which students worked together to define the task, would gain almost 2 points on the test over a similar class wi th no groupwork. This is a significant effect: I t could move a student from the fiftieth to the fifty-eighth percentile on the 32-point test. We found similar results for student production of knowledge: Collaborative seatwork actually reduces achievement, but groupwork in which students actively construct interpretations promotes achievement.

These results are consistent with Hillocks's (1986) study, which examined dozens of empirical studies conducted over a quarter century involving nearly 12,000 students and found that the most effective writing instruction involved peer-response groups with what Hillocks called an "inquiry" focus: assigned topics involving analysis of readings or odier data and attention to rhetorical strategies.

By replicating our eighth-grade study and moving beyond it , we accounted for the unexpected results. Small-group work appeared ineffective


because the groups were being used ineffectively; many of the assignments could just as well have been done individually. The follow-up study shows that when small-group time allows students to interact over the substance of their problem, defining tasks as well as solutions and construaing interpretations, students benefit from the opportunity to work in small groups.

These findings reaffirm the view that effective small-group work requires coherent activities that result in the sustained production of student knowledge. To promote such activities, the teacher must not overly specify group tasks. I n other words, effective teachers clearly define the general parameters of the tasks, but not the precise character of the activities themselves. Teachers who promote thinking about literature may present clear objectives to students for groupwork—for example, identifying the best poems written by students in each group, articulating character traits, and finding supporting quotations — but avoid telling groups exacdy how to proceed; they do not, for example, specify a list of questions and topics students must answer in a particular order.

When teachers put groups of students together to work on some problem, they send students an important message that developing their own thoughts is important. The teacher above who told her students, "There are a lot o f lights . . . in your eyes," as she prepared them for groupwork sent exactly this message.

The benefits o f direct instruction presumably result from higher on-task behavior when the teacher works directly with students. One could argue that this occurs when the instructional task is the same for both whole-class and small-group settings. For example, students may be more on task when they answer recitation questions as a whole class than when the recitation questions are assigned to small groups for written responses. Learning, in this example, may not differ or may be greater in the whole-class format. However, this format does not take advantage of the opportunities for intellectual collaboration that are made possible by the small-group setting. I f the tasks are the same, one should expect little difference in achievement; the point of small-group instruction, however, should not be to assign the same tasks, but to design work that draws on the potential for cooperation and collaboration in the small group.

I f successful small-group work depends on the teacher setting up open-ended rather than prescripted tasks and on the students having coherent conversations generating insights, then teachers must carefully match small-group work to suitable tasks. For example, i f the objective for a given day requires presenting a lot of new information, a lecture is probably better than small-group work. I f teachers want students to practice some particular skill, recitation and seatwork may be better than small-group work. On the other hand, i f teachers want students to compare ideas, develop a train of thought.

72 OPENING DIALOGUE

air differences, or arrive at a consensus o n some controversial issue, then the forum o f small groups may be just the r ight setting for most students t o carry on intensive conversation and discussion, especially students w h o are t oo shy t o say much i n the larger setting o f the whole class. Teachers must always remember, however, that they cannot just p u t students i n groups and expect them to "go t o i t " w i t h positive results. For g r o u p w o r k to succeed, teachers must carefully design collaborative tasks that are interesting to students (and n o t just t o the teacher). They also must be prepared t o help students develop effective g roup skills; an excellent guide for teachers is Cohen's Designing Groupwork (1986) .

THE BOTTOM LINE; LEARNING TO THINK REQUIRES EFFECTIVE INTERACTION

M o s t o f the t ime when teachers ask students questions, they are no t aski n g to be in formed since they already Icnow the answers themselves. W h e n students are asked to recite for teachers w h o have no need to be in formed, they produce "pseudo-discourse." Authent ic discourse occurs on ly when some informat ion or interpretive stance is really at issue. O n l y authentic discourse can engage students.

Yet the results o f our study suggest that authentic questions, discussion, small-group w o r k , and interaction, t hough impor tant , do n o t categorically produce learning; indeed we observed many classes where this was no t the case. We also found that recitation is no t categorically ineffective; rather, its effectiveness varies depending o n whether and h o w teachers expand I R E sequences. The under lying epistemology o f classroom interaction defines the b o t t o m line for learning: W h a t ul t imately counts is the extent to which instruction requires students to think, not just report someone else^s thinking. As Leont 'ev (1981) migh t pu t i t , the pedagogical usefulness o f each interaction needs to be gauged i n terms o f the particular activity or project involved and, more than this, the system o f social relationships that support i t . Authent ic questions, discussion, and small-group w o r k have impor tan t instructional potent ial , bu t unless they are used i n relation to serious instructional goals and, more impor tant , unless they assign significant and serious epistemic roles t o students that the students themselves can value, they may be l i tde more than pleasant diversions.

I n one English class i n our study, for example, students engaged i n several ostensibly open-ended, imaginative w r i t i n g tasks requir ing them t o wr i te f rom the po in t o f view o f a pencil eraser or a bullet. U p o n close inspection o f these exercises and especially the teacher's responses to the papers, we eventually came to understand that, f r o m the teacher's po in t o f view, the content


o f student responses t o these prompts — imaginative or not—was irrelevant; nearly the on ly t h i n g the teacher responded to i n his marginal comments was whether o r n o t all words had been spelled correcdy. As i t happened, the students i n this class understood the operational rubric for this exercise and played their roles more or less appropriately. However, the ostensible purpose (imaginative discourse) and the actual purpose (correct spelling) o f the w r i t i ng tasks significantly differed; the writers were no t really speaking to a reader w h o was l istening to what they were t ry ing to say. These w r i t i n g tasks were what Bloome and Argumedo (1983) call procedural displays. Students can be involved ful ly and substantively i n reciprocal instruct ion on ly when the ostensible purpose o f the discourse is the same as the actual purpose.

Generally, we may say that reciprocity i n instruct ion occurs most often when students, as wel l as the teacher, have some input in to and control over instructional discourse, and w h e n their previous learning significantly affects the course o f subsequent learning. This concept has been implemented i n some elementary reading instruct ion by Palincsar and B r o w n (1984) i n what they call reciprocal teaching. I n reciprocal teaching, students take turns being the teacher. I n other classrooms where students do n o t play the role o f teacher, the teachers nonetheless honor the terms o f reciprocity when they avoid prespecifying answers to their questions so that student answers potentially can affect subsequent questions and discussion. W h e n teachers ask genuine questions o f this sort, they treat students as full-fledged conversants.

This is no t to suggest that the dialogicality o f instruct ion can be judged i n terms o f the how o f instruct ion — question-answer sequences evidenced i n face-to-face interaction — alone. The study found that the what o f instruct i o n — t h e content and subject matter — is critical t o learning as wel l . Authentic questions must challenge students t o t h i n k and reflect o n the consequences o f their ideas, no t just remember their past experiences. Teachers must prize vigorous discussion i n dialogic classrooms, encouraging what Bakht in (1981) calls the struggle o f "contradictory opinions, points o f view and value judgments" (p. 281) . Bakht in teaches us that meaning and therefore l e a r n i n g -understood as the expansion o f a personally coherent interpretation o f informa t ion and events —are actively constructed and negotiated t h r o u g h language use. Learning is a dialogic event, part the instructor's and part the learner's. I f events and informat ion are to acquire meaning and students are to learn, then teachers must t h in k o f cur r icu lum no t just i n terms o f points to be made, in fo rmat ion to be conveyed, and abstract skills to be mastered. Rather, they must engage students i n activities and projects b r idg ing the purposes o f their students and the goals o f instruct ion. Yet instruct ion creating such bridges, we found, was either rare or altogether missing i n the classes we observed, w h i c h overwhelmingly were given over t o lecture, recitation, short-answer questions, and seatwork.

74 OPENING DIALOGUE

The Bakhtinian explanation for the relative ineffectiveness o f monologic instruct ion i n p r o m o t i n g learning and conceptual change, compared w^ith discussion and instructional conversation (even for basic objectives such as recall and literal comprehension), is that meaning "is realized on ly i n the process o f active, responsive understanding" (Volosinov, 1973, p. 102). Yet i t is just such active, responsive understanding that teachers fail to practice — especially i n low-track classes —when they determine p r io r to a given class the entire sequence o f questions they w i l l ask and what answers they w i l l accept, and when they respond to correct student answers w i t h a mere n o d before m o v i n g o n to the next question, often changing the topic o f discourse. I n do ing so, these individuals make no attempt at active, responsive understanding; they "want, i n effect, t o t u r n o n a l igh t bu lb after having switched o f f the current. O n l y the current o f verbal intercourse endows a w o r d w i t h the l i gh t o f meani n g " (Volosinov, 1973, p. 103).

r

APPENDIX A

Sample Literature Familiarity Test

1. For each o f the f o l l o w i n g stories that y o u have read this year, indicate, by ! checking the correct box, whether each had a happy ending or a sad

ending.

a. "The Adventure o f the Speckled B i r d " b. "The Fifty-first D r a g o n " c. "The M o s t Dangerous Game" d. Of Mice and Men

2. Name as many characters as y o u can remember for each story. I f y o u can't remember their names, briefly describe as many as y o u can.

3. For as many stories as y o u can remember, briefly explain h o w each story ended. Wr i t e no more than t w o sentences for each story.

4. For each story that had a conflict y o u can remember, briefly explain what this conflict was.

Briefly relate the conflict i n Of Mice and Men t o the ending o f the story.

Briefly relate this conflict i n Of Mice and Men t o the theme o f the story.

5. For each stor}' that had a theme, or main idea, that y o u can remember, briefly explain what this theme was.

Briefly give three reasons w h y y o u th ink the theme o f Of Mice and Men is what y o u say.

I l l

APPENDIX B

Scoring Information for Literature Familiarity Test

Categories A - H describe dimensions o f interpretive ability. In fo rma t ion f r o m the complete test should be used to derive a score for each.

INFORMATION

A . Extent o f recall (absolute scale)

0 N o n e — no detail recognizable f r o m any stories 1 Vague — very few details recognizable for on ly one or t w o stories 2 Some details regarding some or most o f the stories 3 M a n y details regarding more than half the stories; r ichly detailed for

less than half the stories 4 Richly detailed —many details regarding more than half o f all the

stories

B. D e p t h o f understanding (absolute scale)

0 Never goes beyond literal understanding; empty generalization 1 Goes beyond literal understanding for only one story 2 Some evaluation and interpretation o f more than one story 3 Some interrelation o f story elements i n more than one story 4 H i g h interrelation o f story elements i n most or all stories

C. N u m b e r o f endings f r o m stones studied in class remembered (absolute scale)

0 N o n e 1 One story 2 T w o stories 3 Three stories 4 Four stories 5 Five stories

113

114 APPENDIX B

LITERARY UNDERSTANDING

D . Relates ending to denouement

FOR STUDENTS ANSWERING FOR ALL STORIES

1 For either no stories or one (barely) 2 For either one really we l l or t w o 3 For three stories 4 For four or five stories

FOR STUDENTS ANSWERING FOR HALF OR LESS OF THE STORIES

1 A lm o s t never or no t at all 2 M o s t o f the t ime 3 Consistendy

E. Relates conflict and/or ending to theme


1 For either no stories or one (barely) 2 For either one really we l l or t w o 3 For three stories 4 For four or five stories


1 For either no stories or one (barely) 2 M o s t o f the t ime 3 Consistendy

F. In tu i t s internal motivat ions o f characters


0 For no stories 1 For only one story 2 For less than half the stories 3 For more than half the stories 4 Four ou t o f five or all

APPENDIX B 115


0 For no stories 1 For less than half the stories 2 For more than half the stories

3 Consistendy

G. Interpretive treatment o f major selection

0 N o response

1 Literal/superficial 2 Literal/substantial 3 Some interpretat ion 4 Sophisticated and original/substantial

H . Level o f discourse regarding theme and conflict

0 N o response 1 Cliche or t ru i sm 2 Li tera l 3 General 4 Universal

Note: Score level o f discourse the h igh po in t reached; that is, i f one story is treated literally and t w o are treated i n terms o f cliche, record a score o f 2 (for li teral). Level o f discourse should be scored as general i f the student goes "beyond the in format ion g iven" to treat characters, plots, conflicts, etc., i n general terms, i.e., oifers more than a literal rendi t ion o f events, bu t nonetheless l imi ts these generalizations to the story itself (e.g., statements about what characters are generally like i n the story). By contrast, level o f discourse should be scored as universal i f the student seeks to apply such generalizations beyond the story to experience i n general and does so i n we l l developed terms, making an explicit case for the universality o f the generalization.

118 NOTES

struggle and tension between self and other —"the differential relation between a center and all that is not that center" (Holquist, 1990, p. 18)-which in their view is why conflict is essential to the meaning we give our experience and the understandings we have of it. "Consciousness takes shape and being," Volosinov (1973) claims, "in the material of signs created by an organized group in the process of its social interaction" (p. 13).

8. The interactive character of discourse is evident even when the conversants fail to interaa: in the conversational management of misunderstandings. When conversants misunderstand one another, they must backtrack to some point in the discourse where they were in synch with each other and properly repair the trouble source; for example, the listener asks, "What.>" The ease with which speakers usually explain the "what" — readily, without first asking what was unclear ("What do you mean by 'what;""') — means that speakers consistendy monitor their utterances in terms of what their listeners already know.

9. Clark and Holquist (1984) persuasively argue that he also meant the Communist Party.

10. As Klancher (1989) explains, "Bakhtin's crucial starting point—the diversity of practical languages rather than a unitary abstract structure — leads him to argue that every effort to impose unir^ on these languages is 'monologic' The institutions of the school, the state, and the church enforce monologic languages as the voice of culture, die voice of authority, the voice of God ventriloquized through the literary critic, the politician, or the priest. His terms dialogic and monologic thus describe uses of language rather than inherent properties of language itself" (p. 85).

11. lames Britton (1969, 1970) argued that conversants are most Hkely to discover insights when their talk is expressive.

12. "Stratification, diversity, and randomness [i.e., heteroglossia] is {sic) not only a static invariant in the life of language, but also what insures its dynamics" (Bakhtin 1981, p. 272).

13. This formulation follows Halliday's (e.g., 1978) conception of register as a unique configuration of tenor (relationship between conversants), field (what discourse is about), and mode (channel [i.e., written or spoken] and genre).

14. "The understanding of any sign [i.e., learning], whether inner or outer, occurs inextricably tied in with the situation in which the sign is implemented. . . .Ix.'is^X-wz.ys z social situation" (Volosinov, 1973, p. 37; emphasis in original).

15. In an analysis of 36 hours of instruaion in six different classes, Corey (1940) found that the average student utterance was 11 words long (cited in Hoetker & Ahlbrand, 1969, p. 159).

16. Popper (1972) called this Worid-3 knowledge, distinguishing it from Worid-1 knowledge of the physical worid and Worid-2 knowledge of experience and thought.

CHAPTER 2

The section "Small Groups" is reprinted with editorial changes from "Using Small Groups for Response to and Thinking about Literature," by Martin Nystrand,

NOTES 119

Adam Gamoran, and Mary Jo Heck, English Journal, January 1993. © 1993 by the National Council of Teachers of EngUsh. Reprinted with permission.

1. For various reasons, 5 eighth-grade classes were observed only three times. 2. If teachers asked a question that ehcited no answer, we coded this as an

aborted question, distinguishing it from a repaired question, which is a question that the teacher asks and, without giving students a chance to answer, revises.

We did not code questions unrelated to instruction, rhetorical questions, procedural questions hke, "Does that answer your question.^" and "Do you have any ques-tions.>" or discourse-management questions, which either (a) control discourse traffic (e.g., "What?" or "Did we talk about that?" or "Where are we [in the text]?") or (b) initiate discourse topics (e.g., "Do you remember our discussion from yesterday?").

3. All codings were double checked and challenged by at least one other person besides the coder; we consulted tapes whenever we had questions, and a project assistant verified all transcribed questions and challenged codings by listening to these tapes. A small sample of 12 observations involving over 600 questions was observed by two observers to determine coding reliability: Reliability was computed both at the question level (percent agreement for all questions pooled) and at the observation level (average correlation between raters for the 12 observations).

In our subsample of 12 observations, raters agreed perfectiy on authenticity for 78% of 619 questions; the Pearson correlation at die observational level was .938. Raters agreed on uptake perfectly for 81.7% of the 619-question subsample used to check for reliability; at the observational level, the interrater correlation was .973. Raters agreed on cognitive level perfectiy for 79.0% of the 619-question sub-sample used to check for reliabihty; at the observational level, interrater correlation was .965.

4. CLASS 2.0 and CLASS-EDIT 2.0 and the accompanying documentation are available from the author to anyone wishing to use them. The documentation also contains all coding rules. Write Martin Nystrand, University of Wisconsin-Madison, Wisconsin Center for Education Research, 1025 West Johnson Street, Madison, WI 53706; email [email protected].

5. Students were counted off task in terms of the teacher's expectations. In some classes, teachers expected every student to pay attention during a filnistrip whereas in other classes students were permitted to either watch the fihnstrip or do homework; students doing homework were counted as off task in the first class but not in the second.

6. Linguists call such references deictic references. 7. To qualify as uptake, a question must incorporate a previous answer, not a

previous question; hence, we did not code as uptake teachers' references to questions or remarks they previously had made or to filmstrips, videos, or texts that previously had been discussed. Nor did we code repeated questions as uptake.

8. If a teacher asked students about die previous night's reading, we coded the source of information as text, whereas if die teacher asked about sometiimg learned prior to that, even from a text, we coded the source of information as prior knowledge. We made no distinction between prior knowledge and personal knowledge.

9. Superficially a question such as, "Do you diink that's important? might

120 NOTES

seem to elicit a record (i.e., referring to what the student is thinking at the time of the question), but the question more typically elicits a higher-cognitive operation such as an analysis of what is important. Hence, for such preformulated questions (c£ French & Maclure, 1981), we distinguished the preformulators ("Do you think . . .") from their nuclear utterances (the remainder of the question: "Is that important?"), coding only the latter.

10. More than half of the eighth-grade classes (58.6%) spent no time at all in discussion; 20.7% spent 1 minute or more on average; only two classes of the 58 regularly spent 7 minutes or more. In ninth grade, 61.1% of all classes had no discussion at all, and only 5.6% had more than a minute daily; only one class of the 54 averaged more than 2 minutes.

11. In grade 8, 91.4% of all classes spent no time in small-group work; only 5 out of 58 classes involved small-group work, and this ranged from an average of 1.75-13.50 minutes. In grade 9, 63% of all classes had no small-group work at all; only 11.1% spent more than 5 minutes daily; and only four classes spent more than 10 minutes daily in small groups.

12. In 36.2% of the eighth-grade classes, teachers asked 5% or fewer authentic questions. Only 6.9% of classes involved more than 30% authentic teacher questions; one class of the 58 involved more than 50%. In 25.9% of the classes, uptake was present 5% or less. Only 8.6% of classes involved more than 20%; two classes involved 28%.

13. In grade 9, 13% of all classes had 50% or more; and in one class, the teacher on average asked 83% authentic questions. Uptake was noted in all ninth-grade classes, although 5.6% of classes involved less than 10%. Half the classes had at least 25%, and 13% of classes involved 40% or more uptake.

14. Four of these tides were chosen as a stratified sample to represent the kinds of literature each class had studied; if half of the tides studied were short stories, then two of the four were short stories, and so on. The fifth selected tide was the one work the class had spent the most time on; typically it was either a novel, such as To Kill a Mockingbird or: A Tale of Two Cities, or a drama, such as Romeo and Juliet.

15. Information on smdent characteristics came from smdent questionnaires. Race and ethnicity were coded when students identified themselves as Black or Hispanic. Family socioeconomic status was indicated by an unweighted additive composite of father's and mother's education, the higher of either parent's occupational stams, and a list of home resources, as reported on student questionnaires. Fall reading skills were measured by a NAEP multiple-choice test of reading comprehension, and fall writing skills were measured by a holistically scored writing sample. Fall writing skills were indicated by a writing sample that was scored by two readers, whose marks were averaged, for (a) level of abstraction, based on Britton and colleagues' (1975) categories of transactional-informative prose; and (b) coherence and elaborateness of argumentation, based on the 1979/1984 NAEP criteria for informative writing (in Applebee, Langer, & Mullis, 1985). Each student's writing score was the sum of these two measures. Interrater reliabihty of scoring this test was .68. (For further details on the background variables and achievement controls, see Gamoran & Nystrand, 1991).

16. The relationship between student autonomy and knowledge production has

NOTES 121

been found in many studies, including studies examining small groups; see King (1992) and Palincsar, David, Winn, Stevens, and Brown (1990).

17. Some small-group work rated lowest on autonomy and student production of knowledge concerned grammar and sentence errors, not literature. All of the highest-rated small-group work concerned literature. We are unable to rule out, for the lowest-rated, small-group sessions, the possibility that their focus on grammar and sentence correction rather than literamre, accounted for the negative impact of collaborative seatwork on literature achievement.

CHAPTER 3

We wish to acknowledge Sarah Bing-Prineas's invaluable help in the early stages of this research.

1. These predictions were the result of a residual analysis based on instructional and background variables outlined in Chapter 2.

2. Students in classes whose format most closely approximated conversation tended to score higher on the literature achievement test, regardless of the actual number of student or teacher questions.

CHAPTER 4

Parts of the sections "Dialogism and Students at Risk" (pp. 104-105) and "Literature Instruction" (pp. 105-108) are reprinted with editorial changes from "High School English Students in Low-Ability Classes: What Helps?" by Martin Nystrand, The Newsletter of the National Center on Effective Secondary Schools, January 1990, pp. 7-8.

1. "The meaning of a word is determined entirely by its context. In fact, there are as many meanings of a word as there are contexts of its usage" (Volosinov, 1973 p. 79). Compare also Volosinov's (1973) observation that ''^the forms of signs are conditioned above all by the social organization ofthe participants involved and also by the immediate conditions of their interaction. When these forms change, so does sign" (p. 21; emphasis in translation).

2. Volosinov (1973) argues that thought, or inner speech, "resemble[s] the alternating lines of a diabgue . . . joined with one another and alternat[ing] with one another not according to the laws of grammar or logic but according to the laws of evaluative (emotive) correspondence, dialogical deployment, etc., in close dependence on the historical conditions of the social situation and the whole pragmatic run of hfe (p. 38; emphasis in translation).

3. When writers fail to elaborate potential trouble sources or do so made-quately, the result is "misconstraint," that is, a mismatch between what the writer has to say and what the reader needs to find out. If the topic is inadequately elaborated, the reader will find the text abstruse (What's this about?). If die writer madequately elaborates what is said about the topic, the reader will find the text ambiguous (So what's the point?). For more, see Nystrand (1986, Ch. 3).

Documents

Opening Dialogue