26
Despite the growing evidence of the language and literacy benefits of collaborative discussions for English language learners, the factors contributing to productive discussions that promote ELLs’ positive language outcomes are less understood. This study examined the influence of teacher talk, students’ initial language and literacy skills, and home language backgrounds on the discussion proficiency of four groups participating in eight peer-led literature discussions, called collabora- tive reasoning (CR), in two 5th-grade classrooms serving mainly Spanish-speaking ELLs. Levels of discussion proficiency were determined using a holistic rating approach and utterance-by- utterance coding of discourse features. Teachers’ scaffolding moves were coded. Students’ pre- and post-intervention language and literacy skills and home language backgrounds were assessed. Results showed greater group variation in discussion proficiency in the mainstream class than in the bilingual class. The two teachers differed in their ways of facilitating CR discussions. Group discussion proficiency was associated with oral English skills (sentence grammar) and reading comprehension, as well as student English language use at home and parental assistance with homework. The talk volume and indicators of high-level comprehension such as articulating and responding to alternative perspectives, elaborations, extratextual connections, and uses of textual evidence were associated with post-intervention language and literacy outcomes. These findings contribute to the understanding of sources of variations in discussion proficiency among groups composed predominantly of ELLs and provide implications for teacher scaffolding strategies to facilitate ELLs’ learning and participation in classroom discussions. Quality classroom talk is associated with improved student problem solving, understanding, and learning (e.g., Applebee, Langer, Nystrand, & Gamoran, 2003; Mercer, 1995, 2002; Nystrand, 1997; Webb, 1991). Instructional approaches emphasizing student collaboration and meaningful and extended communica- tion in small-group discussions provide opportunities for language development for culturally and linguistically diverse learners (e.g., Goatley, Brock, & Raphael, Jie Zhang University of Houston Chunling Niu University of Kentucky Research in the Teaching of English Volume 51, Number 2, November 2016 183 What Makes a More Proficient Discussion Group in English Language Learners’ Classrooms? Influence of Teacher Talk and Student Backgrounds Shahbaz Munawar Western Kentucky University Richard C. Anderson University of Illinois at Urbana-Champaign

What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 183

Despite the growing evidence of the language and literacy benefits of collaborative discussions for

English language learners, the factors contributing to productive discussions that promote ELLs’

positive language outcomes are less understood. This study examined the influence of teacher talk,

students’ initial language and literacy skills, and home language backgrounds on the discussion

proficiency of four groups participating in eight peer-led literature discussions, called collabora-

tive reasoning (CR), in two 5th-grade classrooms serving mainly Spanish-speaking ELLs. Levels

of discussion proficiency were determined using a holistic rating approach and utterance-by-

utterance coding of discourse features. Teachers’ scaffolding moves were coded. Students’ pre- and

post-intervention language and literacy skills and home language backgrounds were assessed.

Results showed greater group variation in discussion proficiency in the mainstream class than in

the bilingual class. The two teachers differed in their ways of facilitating CR discussions. Group

discussion proficiency was associated with oral English skills (sentence grammar) and reading

comprehension, as well as student English language use at home and parental assistance with

homework. The talk volume and indicators of high-level comprehension such as articulating and

responding to alternative perspectives, elaborations, extratextual connections, and uses of textual

evidence were associated with post-intervention language and literacy outcomes. These findings

contribute to the understanding of sources of variations in discussion proficiency among groups

composed predominantly of ELLs and provide implications for teacher scaffolding strategies to

facilitate ELLs’ learning and participation in classroom discussions.

Quality classroom talk is associated with improved student problem solving, understanding, and learning (e.g., Applebee, Langer, Nystrand, & Gamoran, 2003; Mercer, 1995, 2002; Nystrand, 1997; Webb, 1991). Instructional approaches emphasizing student collaboration and meaningful and extended communica-tion in small-group discussions provide opportunities for language development for culturally and linguistically diverse learners (e.g., Goatley, Brock, & Raphael,

Jie ZhangUniversity of Houston

Chunling Niu University of Kentucky

Research in the Teaching of English Volume 51, Number 2, November 2016 183

What Makes a More Proficient Discussion Group in English Language Learners’ Classrooms? Influence of Teacher Talk and Student Backgrounds

Shahbaz MunawarWestern Kentucky University

Richard C. Anderson University of Illinois at Urbana-Champaign

g183-208-Nov16-RTE.indd 183 11/22/16 3:51 PM

chartman
Text Box
Copyright © 2016 by the National Council of Teachers of English. All rights reserved.
Page 2: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

184 Research in the Teaching of English Volume 51 November 2016

1995; Martinez-Roldan & Lopez-Robertson, 1999; Saunders & Goldenberg, 1999; Zhang, Anderson, & Nguyen-Jahiel, 2013). In small-group literature discussions, research has reported bilingual students “not only learning language and learning about language, but also learning through language” (Martinez-Roldan & Lopez-Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions in todays’ elementary classrooms, research into teachers’ and students’ use of text discussion is urgently needed.

The achievement gaps in text-level skills between ELLs and native English speakers are large and persistent in upper elementary grades. Literacy instruction practices for ELLs too often feature individual seatwork and teacher-directed whole-class instruction, depriving ELLs of extended and meaningful experiences with oral English and thus contributing to their low academic achievement and engagement (August & Shanahan, 2006). Despite the growing evidence of the language and literacy benefits of collaborative discussions for ELLs, the discourse processes of peer-led small-group discussions that promote positive language outcomes are less understood. Specifically, several questions remain underexplored, including: What constitutes a proficient discussion? To what extent does teacher talk explain the variation in the quality of small-group discussions in ELLs’ classrooms? Little is known about the influence of student academic and home language backgrounds on discussion quality and how teacher talk supports or constrains the develop-ment of student discussion skills as a result of peer-led small-group discussions.

The purpose of this study was threefold. First, it investigated the variation in discussion proficiency of four peer-led discussion groups in two 5th-grade classrooms serving mainly Spanish-speaking ELLs, using both a holistic rating approach and utterance-level coding of discourse processes. Second, the study examined the influence of teacher talk, student academic background, and home language background on the discussion proficiency of these groups. Third, the study examined the association between discourse features in peer-led small-group discussions and students’ language and literacy outcomes. Understanding the factors that contribute to successful or less successful peer-led group discussions and the mechanisms by which such discussions foster gains in diverse learners’ language and literacy skills will enable teachers and researchers to design effective literacy instruction for these students.

Programs featuring authentic and rich language usage benefit English language learners (e.g., Kucer & Silva, 1999). To date, several small-group discussion ap-proaches have been successfully implemented in classrooms composed predomi-nantly of ELLs: instructional conversations (Saunders & Goldenberg, 1999), book clubs (Kong & Fitch, 2002; Kong & Pearson, 2003), literature circles (Peralta-Nash & Dutch, 2000), and collaborative reasoning (Zhang et al., 2013). Instructional ap-proaches like instructional conversations and book clubs offer ELLs opportunities to engage in extended language use and develop language, social, and cognitive skills (Kong & Pearson, 2003; Williams, 2001).

The current study used collaborative reasoning (CR), a peer-led small-group discussion approach (Clark et al., 2003), aiming to promote Spanish-speaking

g183-208-Nov16-RTE.indd 184 11/22/16 3:51 PM

Page 3: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 185

ELLs’ intellectual engagement, reflective thinking, and language development. In CR, students read a text that raises a dilemma or unresolved issue with multiple and competing points of view. Students then gather in groups of five to eight to deliberate on the Big Question (e.g., Are zoos good places for animals?) raised by the text. Students are expected to present their positions on the Big Question, support the positions with reasons and evidence, carefully listen, evaluate, respond to one another’s arguments, and challenge each other when they disagree. CR features open participation, in which students speak freely without being nominated by the teacher. Students are encouraged to manage all aspects of discussion as indepen-dently as possible. The teacher facilitates the discussion from the side and provides scaffolding to foster student participation and argumentation skills as needed. A comprehensive list of teachers’ scaffolding moves can be found in Clark et al. (2003).

Zhang et al. (2013) provided empirical evidence of the positive impact of collaborative reasoning on English language learners’ oral and written language development. After participating in eight CR discussions over one month, fifth-grade Spanish-speaking ELLs who participated in CR outperformed their control counterparts in English listening, speaking, reading, and writing. The differences were most significant in speaking and writing. Prompted by a wordless picture book, CR students produced much longer and more coherent stories than those in the control group. Reflective essays written by the CR students in response to a new text were longer; had more diverse vocabulary; and contained a greater number of satisfactory reasons, counterarguments, and uses of text evidence.

Despite the positive intervention effects, Zhang et al. (2013) did not provide in-depth discourse analyses to uncover the mechanism by which collaborative reasoning promotes ELLs’ language and literacy development. In-depth analyses of discourse processes in ELLs’ collaborative discussions are scarce in existing literature. In particular, few studies on classroom discussion have investigated what constitutes a proficient discussion group in ELLs’ classrooms or how ELLs develop their discussion skills over time. Almasi, O’Flahavan, and Arya (2001) found that proficient literature discussion groups were more able to sustain topics of conversation by revisiting old topics, making links between topics, and embedding topics within one another, while the conversation of less proficient groups included substantially fewer linkages and embedded topics because the teachers and students initiated large amounts of metatalk (talk about one’s talk or how the group should function), which caused disjuncture in peer discussion and impaired the groups’ ability to maintain topics. Almasi et al.’s (2001) intensive microanalysis of discourse captured how more proficient and less proficient peer discussion groups created conversational coherence and how they managed their group process. However, all participants in the study were European-American students, and the researchers did not consider the cultural variations and conver-sational conventions of non–English speaking groups.

Going beyond previous research, the present study examined the variation in discussion proficiency of small groups in ELLs’ classrooms and the influence of teacher talk and student backgrounds on the groups’ discussion proficiency. Levels

g183-208-Nov16-RTE.indd 185 11/22/16 3:51 PM

Page 4: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

186 Research in the Teaching of English Volume 51 November 2016

of discussion proficiency were determined using two approaches: Almasi et al.’s (2001) discussion proficiency rating scale and utterance-by-utterance coding of discourse features, which will be described in detail in the Method section.

Over the past decades, researchers have attempted to identify distinctive dis-course features of quality classroom talk. In a meta-analysis of nine small-group discussion approaches (including CR), Soter et al. (2008) coded representative transcripts from each approach for various discourse features and identified several elements of productive talk or indicators of higher-level thinking and compre-hension: authentic questions and uptake (Nystrand, 1997); higher-level thinking questions that elicit generalization, analysis, and speculation (Nystrand, 1997); questions that elicit extratextual connections; the density of reasoning words (We-gerif & Mercer, 1997); elaborated explanations (Webb, 1991); and exploratory talk (Mercer, 1996). Chinn, Anderson, and Waggoner (2001) found more extensive use of cognitive processes—elaborations, predictions, providing textual evidence, and articulating alternative perspectives—in student CR discussions than in recitations. These cognitive processes served as the indicators of proficient discussions in the present study because these processes are aligned with the indicators of higher-level thinking and comprehension in small-group text discussions (Soter et al., 2008).

There is emergent research on the influence of teacher talk on student discourse features in elementary classroom discussions. In fourth-grade CR discussions, Jadallah et al. (2011) observed that teacher instructional moves (prompting and praising the use of evidence, asking for clarification and challenging) initiated a profound influence on students’ ways of thinking and talking. Some students appropriated the teacher’s instructional moves and began to use these moves spontaneously in later discussions (Jadallah et al., 2011). Similarly, Lin et al. (2015) found that CR teachers’ prompts for relational thinking had an immediate effect on student relational thinking, triggering further relational thinking from stu-dents over several speaking turns. Boyd and Rubin (2002) analyzed fourth- and fifth-grade ELL students’ elaborated talk in a six-week, literature-rich science unit, and found that the teacher played a crucial role in students’ extended dialogue by facilitating interpretation through strategic questioning. Taken together, these findings underscore the importance of teacher influence in peer discussions despite the low teacher talk rate. The current study aimed to understand how teachers’ language use or scaffolding moves affect the discussion proficiency of various groups composed mainly of ELLs.

Home language and literacy backgrounds play an important role in ELLs’ literacy skills (e.g., Duursma, Romero-Contreras, Szuber, Proctor, & Snow, 2007). To our knowledge, no prior studies have examined how ELLs’ home language back-grounds might affect their participation in small-group discussions. Specifically, do ELLs who use English more and are exposed to English more at home participate in CR discussions more actively? Are home literacy resources, parent-child reading frequencies, and parental literacy support associated with children’s participation in CR discussions? These questions were addressed in the present study.

Another purpose of the current study was to examine the linkages between discourse features in CR discussions and student language and literacy outcomes.

g183-208-Nov16-RTE.indd 186 11/22/16 3:51 PM

Page 5: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 187

Much previous research on classroom discourse has examined the ways in which the frequency of individual utterances or cognitive processes (e.g., giving explana-tions or elaborations) is related to later individual learning outcomes (e.g., Chinn et al., 2001; Webb, 1991; also, see a review in Murphy, Wilkinson, Soter, Hennessey, & Alexander, 2009). Less attention has been paid to the issue of joint meaning construction, collective reasoning, and co-construction of arguments, all critical discourse features of collaborative reasoning.

Berkowitz and Gibbs (1983) described collective reasoning as “reasoning that operates on the reasoning of another” (p. 402), or transactive discussion. Kruger and Tomasello (1986) further distinguished transactive utterances from nontransactive utterances, another type of communication that does not contain reasoning, such as a statement, explanation, prediction, request, agreement, or disagreement. Mer-cer (1996) proposed three ways of talking and thinking in collaborative learning: disputational talk (similar to nontransactive talk), cumulative talk, and exploratory talk. Definitions and examples of these types of talk are summarized in Table 1.

The present study investigated cognitive processes and the degree of collec-tive reasoning in CR discussions and how these discourse features may promote the language and literacy skills of ELLs. Student talk was analyzed in terms of the cognitive processes and three ways of talking and thinking (nontransactive talk, cumulative talk, and exploratory talk). The current study asked three specific questions:

1. To what extent do small groups vary in discussion proficiency, as measured by a holistic rating and an utterance-level coding procedure? The answer to this questions may contribute to the development of a useful tool for teachers to gauge the quality of classroom discussions, and has implica-tions for teacher professional development.

2. To what extent do student background and teacher talk affect discussion proficiency across groups? Students’ backgrounds included their ethnic-ity, pretest achievement level, home language, and literacy background. Features of teacher talk included teacher questioning types and scaffold-ing moves.

3. What peer discourse features in CR talk are associated with language and literacy outcomes (listening, speaking, reading, and writing) for ELLs? Answering this question will advance understanding of peer discourse features that contribute to language and literacy gains, and will inform instruction practices for ELLs.

MethodParticipants The participants were 27 fifth graders in a mainstream class (n = 14) and a bilingual class (n = 13), who received collaborative reasoning intervention for a period of 4 weeks. The larger study reported in Zhang et al. (2013) involved four 5th-grade classrooms (N = 90), two CR and two control classrooms, in an elementary school

g183-208-Nov16-RTE.indd 187 11/22/16 3:51 PM

Page 6: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

188 Research in the Teaching of English Volume 51 November 2016

in Illinois. The school serves low-income families. Of the 90 students in all four classrooms, 75 (83.3%) were Latino and 12 (13.3%) were African American.

The study sample of four groups was selected from the two CR classes. Students in the mainstream classrooms, primarily Latino and African American, received instruction only in English. The bilingual classrooms served Latino students who, due to their limited English proficiency, sometimes required Spanish support. Discussion groups were color-coded and remained the same throughout the study period. Students were grouped heterogeneously based on gender, ethnicity, talkativeness, and reading level. In the mainstream class, there were two African American girls and five Latinos (two girls, three boys) in the blue group, and three African American (two girls, one boy) and four Latino (one girl, three boys) stu-dents in the green group. In the bilingual class, there were two boys and five girls in the red group, and two boys and four girls in the yellow group.

The teacher in the mainstream class, Mrs. Lehmann,1 had over 10 years of teaching experience and participated in a CR project prior to the study. The teacher in the bilingual class, Mr. Herrera, spoke Spanish and had no prior CR experience.

ProcedurePre-interventionPrior to the intervention, the participating teachers attended a full-day workshop on CR. All students were assessed on vocabulary, sentence grammar, and reading comprehension as detailed below. A home language background survey was also sent home for parents or guardians to complete.

Vocabulary checklist. We administered a wide-range, general vocabulary test in a checklist format (Anderson & Freebody, 1983). Students read through a list of 180 words, 50 nonwords (e.g., pensile, jerbal), and 30 pseudo-derivatives (e.g., ac-ceptment, inhappy) and indicated whether they knew the meaning of each item. Scores were corrected for guessing using the high-threshold formula described by Anderson and Freebody (1983).

sentence grammaticality judgment test. We used a modified version of the sen-tence grammaticality judgment test developed by Johnson and Newport (1989). Students listened to 40 sentences and decided whether each sentence was expressed the right way in English. Examples of ungrammatical sentences are: Bill and Joe is good friends. The little boy is speak to a policeman.

gates-macginitie reading comprehension test. Students took the reading com-prehension subtest from the Gates-MacGinitie Reading Tests (MacGinitie, Mac-Ginitie, Maria, & Dreyer, 2000), which required them to read short passages and then answer multiple-choice questions.

home language surVey. Parents or guardians of the participants completed a three-page survey about student home language use, home language exposure, home literacy resources, child reading frequencies at home, reading starting age, parent education levels, parent help with literacy activities, and demographic information.

g183-208-Nov16-RTE.indd 188 11/22/16 3:51 PM

Page 7: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 189

InterventionAfter the pretests, students in the CR classes participated in eight CR discussion sessions over a period of 4 weeks with two sessions per week, each approximately 20 minutes long. A total of 16 videotaped discussions (Sessions 1, 4, 5, and 8 recorded for each of the four groups) were analyzed in the present study. The eight stories selected for the CR discussions were interesting and relevant to the children’s life experience, and appropriate for their reading level. The stories described dilemmas faced by the characters, creating contexts for consideration of topics such as friend-ship, fairness, honesty and integrity, winning and losing, and public policy. The stories and the order of discussions were identical between the two CR classrooms.

Post-interventionAfter the intervention, students were assessed on English listening, speaking, read-ing, and writing. Below is a brief summary of these measures. For more details of the measures and scoring methods used, see Zhang et al. (2013).

the sentence Verification technique for listening and reading comprehension. The sentence verification technique (SVT) was developed by Royer and colleagues (e.g., Marchant, Royer, & Greene, 1988; Royer & Carlo, 1991). Students listened to or read a passage and then judged a series of sentences and decided whether a given sentence had the same meaning as a sentence in the story they had just heard or read. Three passages were included in each modality (listening and reading).

storytelling. English speaking ability was assessed with a storytelling task using a wordless picture book, Frog, Where Are You? (Mayer, 1969). Oral narratives were audiotaped and transcribed following the Systematic Analysis of Language Tran-scripts (SALT) conventions (Miller & Chapman, 2003). Each language sample was segmented into T-units, a T-unit being “a single main clause plus whatever other subordinate clauses or nonclauses are attached to, or embedded within, that one main clause” (Hunt, 1977, p. 92). The final transcripts were analyzed using SALT software, and all of the standard measures were computed using transcript length (total words, T-units), vocabulary diversity (type/token ratio, or TTR), syntactic complexity (mean length of utterance, or MLU), and verbal fluency (words per minute). Another hand-coded score, Narrative Scoring Scheme (NSS), an index of a child’s ability to produce a coherent narrative, was also computed as a measure of quality of storytelling.

reflectiVe essay writing. The prompt for the essay writing posttest was a three-page story, “The Pinewood Derby” (McNurlen, 1998). The story had not previously been read or discussed by any of the students. It tells about a boy named Thomas who wins a contest building and racing model cars, but breaks the rules by not making his car by himself. He confides to his classmate, Jack, that he has received help from his older brother in making his car. Students were asked to write an essay reflect-ing on whether Jack should tell on Thomas, and were given 40 minutes to write.

Following Reznitskaya et al.’s (2001) coding scheme, written language samples were coded in terms of argument quality using NUD*IST 6 (Qualitative Solutions

g183-208-Nov16-RTE.indd 189 11/22/16 3:51 PM

Page 8: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

190 Research in the Teaching of English Volume 51 November 2016

and Research, 2002) by the first author. Argument quality indicators included rea-sons, counterarguments, rebuttals, and use of textual evidence. A counterargument was defined as an idea unit in support of the opposite of the chosen position. A rebuttal was defined as an idea unit that responds to the counterargument and fur-ther supports the chosen position after recognizing a possible opposing argument.

Transcription and CodingA total of 16 CR discussions were transcribed and coded, including 4 discussions from the beginning (Session 1), middle (Sessions 4 and 5), and end of the inter-vention (Session 8) for each of the 4 selected groups. Videotapes of discussions were transcribed using Transtool (Kumar & Miller, 2003) and coded using NVivo 9 (Qualitative Solutions and Research, 2002). The analyzed utterances of the 16 discussions included 32,442 words spoken over a total running time of 5.4 hours.

The coding was conducted in four phases. In the first phase, full speaking turns were identified and fragmented turns were removed from all transcripts following Chinn et al.’s (2001) procedure. In the second phase, all student full turns were coded for discourse features as shown in Table 1. In the third phase, the discourse feature codes were further categorized into three types of talk: nontransactive talk (explanations and predictions), cumulative talk (mainly confirmations and explanations), and exploratory talk (counterarguments and high-level questions). In the fourth phase, teacher talk was coded in terms of scaffolding moves and questioning types described in Table 2.

Code Discourse Feature Definition Example

FQ Factual questions Questions that require a single correct answer

What’s the girl’s name?

High-level questions

Open-ended questions that allow for analysis, generalization, and/or speculation

Uta Uptakea A question in which there is follow-up about something someone said before

What do you think about what Jennifer just said?

RQCe Requests for clarifications and explanations

Asking for clarifications or elaborations of ideas

You think Willy should win? Are you saying . . . ?

RQPR Requests for positions and reasons

Asking for opinions or support for a claim

What do you think? Did you change your mind? Why . . . ?

RQeVa Requests for evidence a

Asking for supporting evidence You can use facts?!

Cl Clarifications Correcting other students’ misinformation or misinterpretation

No. I mean they are good.

continued on next page

table 1. Definitions and Examples of Student Discourse Features

g183-208-Nov16-RTE.indd 190 11/22/16 3:51 PM

Page 9: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 191

nt Nontransactive talk Short exchanges consisting of assertions and counterasser-tions; not connected to previous utterance

Student 1: That’s right! Student 2: I don’t agree.

Ct Cumulative talk Students taking over, integrating, and applying the perspectives of their peers, but without real disagreement or construc-tive conflict; characterized by confirmations, expansions, and elaborations

Student 1: I think he should practice more so he could get better.Student 2: Yeah, because in the story it says he held his bats like in the wrong way and he run in the wrong way. So I think he need to practice a little bit more.

et Exploratory talk Students challenging or coun-terchallenging while justifying challenges and offering alterna-tive perspectives; characterized by constructive and critical engagements

Student 3: But he should get some chance [interrupted]Student 2: Yeah he should get a chance but first he should practice a lot more.Student 3: But he has a spirit I know since he has a lot of confidence he should be better.

eX Explanations Telling why events occur and why predicted events could occur

. . . because computers could help you to search.

COnF Confirmations Agreeing with, adding to, or expanding on previous contributions

Yeah, exactly, I agree . . .

eXCn Extratextual connections

Affective responses, shared knowledge responses, and intertextual connections (e.g., using other texts or works of art, media, TV, newspapers, etc.)

If I were Ronald Morgan, I would be sad. Like Ruben said . . . . It happened to me before . . .

teVD Textual evidence Using specific textual evidence to support a claim

On page 30, it says . . .It says in the story . . .

PReD Predictions Discussing what will happen next or what might happen in different scenarios

Ronald Morgan has spirits and maybe he will win.

Ct Counterarguments/alternative perspectives

Ideas intended to challenge and express disagreements with another student’s idea; often appearing in the form of counterarguments and rebuttals

What if . . . ? But . . . . I disagree . . .

elab Elaboration Explaining thoughts in some detail through extension, building of an idea step-by-step, giving justified reasons for a statement, or expanding on a statement

No because the textbook you could take them home and computer you can’t take because its the property from the school and when you need to look for information the textbook is there to help you and that computer is not.

aRequests for evidence and uptake appeared very few times in student talk and were dropped from further analysis.

table 1. Continued

g183-208-Nov16-RTE.indd 191 11/22/16 3:51 PM

Page 10: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

192 Research in the Teaching of English Volume 51 November 2016

Code Discourse Feature Definition Example

BQ Big Question Central question Should the coach let Ronald Morgan play with the team?

FQ Factual questions Questions that require a single correct answer

You know that Ronald Morgan, was he a very good baseball player?

TTM Turn-taking management

Nominating students to talk Anybody else? What do you think about that, Allison? [looking at Allison]

CR Comments and repetitions

Confirmation and simple evaluation

So he should let him play? OK. [looking at student]

PC Personal connection Asking for affective and shared-knowledge responses

Say if you were Stone Fox, if your home has been taken away by someone else, how would you feel?

AC Asking for clarification

Asking students to be clear in their argumentation

Can you explain about that (Ronald Morgan) closing his eyes?

CL Challenging Considering alternative points of view

Do you rather have everybody play and lose or have a chance to win and not everybody play?What about the zoos that have animals in cages?

EC Encouraging Acknowledging and praising students for progress in reasoning and participation skills

These are really good reasons.I like the way you challenge each other.

FI Fostering independence

Moves that support the gradual transfer of student responsibility; asking students to respond to one another’s ideas (uptake)

What do you think about John’s argument?You don’t have to agree, just go with what you believe.

GS Goal setting Explaining ground norms for discussion and setting up clear expectations

Make sure when you are talking to your group that you are looking at your group members.When you give your thoughts or opinions, you are not talking over one another.

PT Prompting Asking or telling students to state their positions, reasons, or evidence

What is your position/opinion? Why do you think computers are good?

SRM Stepping in and reminding

Stepping in and reminding students of the ground rules and norms for the discussion

Remember we don’t need to raise our hands. Just go ahead.One person talks at a time.

SRF Summing up and refocusing

Summing up what students have said and helping the group refocus on the Big Question

OK, let’s get back to the Big Question. You guys all gave good reasons to support your opinion, and many of you think, how many of you still think zoos are good places for animals?

TDB Teacher debriefing Metadiscussion on the quality of the reasoning and participation dynamics in the group

You did a really good job. You did great, um, yeah, we mentioned that you used evidence: “it says in the story, on which page, blah blah blah, . . . You are not only answering the question, but you are talking, responding to each other’s opinion.

continued on next page

table 2. Definitions and Examples of Teacher Discourse Features

g183-208-Nov16-RTE.indd 192 11/22/16 3:51 PM

Page 11: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 193

The coding scheme was adapted from Chinn et al. (2001), Clark et al. (2003), and Soter et al. (2008). The second and third authors were trained to follow the coding scheme with two rounds of practice. The first and second authors coded half of the transcripts independently. As a reliability checker, the third author indepen-dently coded randomly selected transcripts representing 20% of the total sample (20% from each coder’s batch). The coding agreements ranged from 80% to 99%.

To determine the proficiency levels of the four discussion groups, one outside viewer who had expertise in classroom discussions but was blind to the research design rated each of the 16 discussions on a 4-point scale (from 0, indicating no evidence, to 3, indicating adequate evidence) based on Almasi et al.’s (2001) nine indicators of proficient discussions: (a) students add on to interaction and interpretation, (b) students refer to text, (c) students respond to one another, (d) students relate to personal experience, (e) students ask questions, (f) students monitor group process, (g) students extend comments by adding on or asking questions, (h) students critically evaluate the text and argument, and (i) teacher scaffolds interaction (does not dominate the discussion).

Data Analysis Procedures To answer the first research question, on the extent to which small groups vary in discussion proficiency, independent samples t tests were conducted on discussion proficiency ratings and utterance-level coding results between the more and less proficient discussion groups in the mainstream and bilingual classes. The second research question concerned the influence of student backgrounds and teacher talk on discussion proficiency across groups. Independent samples t tests were conducted to compare the teacher talk differences between the mainstream class and bilingual class and the differences in teacher talk, initial language and literacy performance, and home language backgrounds between the more and less proficient groups in each class. To answer the third research question, on what peer discourse features in CR talk are associated with ELLs’ language and literacy outcomes, Pearson correlation coefficients were calculated between student talk features and post-intervention language and literacy outcomes. The level of significance for all statistics was set at α = .05.

SDB Encouraging student debriefing

Asking students to evaluate their group performance

What are some of the things that you think your group did really well? What are things your group needs to work on?

TOL Thinking out loud Describing what is going on inside one’s mind as one works through the thinking process

Well, if I am on this baseball team, there is no way that I want R. M. to play on my team. He is afraid of the ball. He closes his eyes when he hits. . . . So if I am on this team, my coach better not let him play otherwise I am gonna have problems with it. So I don’t think he should be allowed to play.

table 2. Continued

g183-208-Nov16-RTE.indd 193 11/22/16 3:51 PM

Page 12: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

194 Research in the Teaching of English Volume 51 November 2016

ResultsDiscussion Proficiency Variation across GroupsTable 3 presents the outside expert’s holistic ratings of discussion proficiency across four sessions. The four groups varied in total discussion proficiency ratings: the mainstream blue group (22), the mainstream green group (8.75), the bilingual red group (17.5), and the bilingual yellow group (9). In the mainstream class, the blue and green groups were classified as the more and less proficient discussion groups, respectively. In the bilingual class, the red and yellow groups were classi-fied as the more and less proficient groups, respectively. Overall, the difference in total average discussion proficiency ratings between the more and less proficient discussion groups was significant for the mainstream class, t (6) = 5.38, p = .002, but not for the bilingual class, t (6) = 1.32, p = .23. We further compared the rat-ings on each discussion proficiency indicator between the more and less proficient groups in each class. In the mainstream class, independent samples t test results showed that compared with the less proficient group, the more proficient group responded more to one another, t (6) = 2.32, p = .05; related more to personal experience, t (6) = 4.70, p = .003; monitored group process more, t (6) = 7.00, p < .001; and extended comments by adding on or asking questions more, t (6) = 4.24, p = .005. Also, in the mainstream class there was a nonsignificant trend toward students in the more proficient group critically evaluating the text and argument more than those in the less proficient group, t (6) = 1.98, p = .09. No significant group difference was found on any of the nine indicators in the bilingual class.

table 3. Average Holistic Ratings of Discussion Proficiency across Four Discussions by Group

Mainstream Class Bilingual Class

Indicator MORe PROFiCient (n = 7)blUe

less PROFiCient (n = 7)gReen

MORe PROFiCient (n = 7)ReD

less PROFiCient (n = 6)YellOw

stUDents aDD On tO inteRaCtiOn anD inteRPRetatiOn

3.00 1.00 1.75 1.25

stUDents ReFeR tO teXt 2.00 1.00 0.75 0.75

stUDents ResPOnD tO One anOtheR* 2.50 1.00 1.50 1.50

stUDents Relate tO PeRsOnal eXPeRienCe**

2.25 0.00 0.00 0.00

stUDents ask QUestiOns 2.50 1.25 1.50 1.75

stUDents MOnitOR gROUP PROCess** 2.75 1.00 0.75 0.75

stUDents eXtenD COMMents bY aDDing On OR asking QUestiOns**

2.75 1.25 1.50 1.75

stUDents CRitiCallY eValUate the teXt anD aRgUMent

2.25 1.00 0.50 0.00

teaCheR sCaFFOlDs inteRaCtiOn (DOes nOt DOMinate the DisCUssiOn)

2.00 1.25 1.50 1.25

tOtal** 22 8.75 17.50 9.0

Note. Ratings use a 0–3 point scale. * p < .05. ** p < .01.

g183-208-Nov16-RTE.indd 194 11/22/16 3:51 PM

Page 13: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 195

Utterance-level coding of student and teacher talk was analyzed to further investigate the variability of discussion proficiency across groups. Table 4 presents the average talk volume and percentages of each indicator of student talk quality by group across four discussions. Overall, students talked a lot longer with more turns and greater fluency in the mainstream class (especially the more proficient group) than in the bilingual class. The teacher in the mainstream class talked much less than the teacher in the bilingual class. The ratio of teacher to student talk in the bilingual classes (red group 0.42, yellow group 0.42) was almost three times that of the mainstream classes (blue group 0.08, green group 0.14).

Independent samples t tests were conducted on student talk quality indicators within each class. In the mainstream class, the more proficient group had signifi-cantly more student talk turns than the less proficient group, t (6) = 2.84, p = .02. Compared with the more proficient group, the less proficient group used more

Note. More proficient groups: mainstream blue and bilingual red; less proficient groups: mainstream green and bilingual yellow.* p < .05.

table 4. Means and Percentage Use of Each Utterance Type by Group across Four Discussions

Mainstream Class Bilingual Class

Variable MORe PROFiCient (n = 7)

less PROFiCient (n = 7)

MORe PROFiCient (n = 7)

less PROFiCient (n = 6)

talk QUantitY (Means)

Length (min.) 24 20 17 15

Student turns* 276 84 68 89

Student turns per min. 11 4 4 5

Student words* 2,964 1,920 1,350 1,111

Student words per min. 123 97 74 68

Teacher turns 18 11 22 25

Teacher-to-student turn ratio 0.08 0.14 0.42 0.42

stUDent talk QUalitY (% Use OF eaCh UtteRanCe tYPe bY gROUP aCROss FOUR DisCUssiOns)

Request for clarification and explanation 3.62 4.84 3.45 4.11

Request for position and reason 4.45 5.32 7.92 24.26

Counterargument/alternative perspective

38.24 28.95 15.54 6.06

Clarification and explanation 20 37.58 40.54 49.56

Confirmation* 14.89 31.81 14.40 11.08

Extratextual connection 10.84 8.07 2.84 1.99

Textual evidence 5.80 17.64 3.75 5.28

Prediction 4.08 8.54 15.46 18.28

Nontransactive talk* 9.75 31.75 31.5 45

Cumulative talk* 16.25 32 18 12.5

Exploratory talk 38.25 30 19 7

g183-208-Nov16-RTE.indd 195 11/22/16 3:51 PM

Page 14: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

196 Research in the Teaching of English Volume 51 November 2016

confirmations, t (6) = -2.50, p = .04; more nontransactive talk, t (6) = -4.10, p = .006; and more cumulative talk, t (6) = -2.40, p = .05. Despite a general trend of greater talk quality for the more proficient group, no significant group difference was found on any utterance-level code in the bilingual class. The smaller group variability in utterance-level codes in the bilingual class was consistent with the pattern of holistic ratings presented in Table 3. An interesting pattern was that nontransactive talk accounted for more than one third of the student talk in the bilingual class, while the mainstream class had a large proportion of exploratory talk, as shown in Table 4.

Discussions of More and Less Proficient Groups To illustrate the variation in discussion proficiency across groups, two excerpts, taken from the fourth discussions of the two groups in the bilingual class, are presented below. Excerpt 1 is chosen from a discussion by the less proficient group (yellow), made up of six Latino ELLs (two boys and four girls), about the story Crystal’s Vote (Nguyen-Jahiel, 1996). In the story, Crystal is a student member of a school committee that has to decide whether they should replace the worn-out fifth-grade math textbooks with new ones or buy a computer program that teaches math. The Big Question for this discussion was: Should Crystal vote for the textbooks or the computer program? In the beginning of the discussion, most students favored computers and discussion tended to be single-sided until the teacher stepped in and challenged student thinking. The excerpt starts after the teacher prompted for alternative perspectives in favor of buying textbooks.

OnOreliO: What do you think? SalvadOr: Because texts are better, you could study about them.OnOreliO: What do you think?Jennifer: Because she could . . . she could learn . . . she could learn more search-

ing, searching on the computer. Jennifer: You? Sandy?Sandy: I think she could learn more because if she needs help on the searching,

she could search that the thing she needs.Jennifer: You? liSSette: I think yeah, because it might help her . . . to study more.OnOreliO: Yeah, because the computers are better and they can use them again.Jennifer: You? Maria: I think yeah, computers because . . . you can get help in there. Jennifer: What made you change your decision? [looks at Salvador]SalvadOr: Because you could study them and computers you cannot take them

home. That’s why I changed my decision.

By our standards, this was a full-fledged CR discussion. Students self-managed free-flowing student-student interactions for an extended period of time. Jennifer acted as a teacher, managing turn taking. However, Excerpt 1 features nontrans-

g183-208-Nov16-RTE.indd 196 11/22/16 3:51 PM

Page 15: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 197

active talk, as students gave their own opinions briefly, without connecting to previous student talk. There was no evidence of joint, collaborative engagement in the talk. Student responses were simple and short, and did not present arguments with supporting reasons and evidence, nor did they consider others’ arguments or respond to one another.

Excerpt 2, below, is taken from the same discussion in the more proficient group (red) in the bilingual class. The group consisted of seven Latino ELLs, two boys and five girls. Differing from the interaction patterns found in Excerpt 1, in this excerpt Sigfrido, rather than waiting for the teacher to step in and prompt for alternative points of view, took the lead and successfully challenged those who voted to buy computers.

SigfridO: How much do the supplies cost then?Student: Forty.SigfridO: I think with the computers, you are not gonna be able to buy the large

supplies. You could buy few supplies. And with the textbooks you could buy supplies and then you could buy PE stuff and music.

alita: Yeah, but there is also a lot of things too. And there is also a lot things from gym and music that you need. Um . . . like . . . um . . . gym you need a lot of stock because many school have things that sometimes . . . um . . . teachers say that we will have in the school. And that’s why they should buy computers because if they . . . if they get that money from the computers, because they don’t pay, they are gonna have like $5,000. And with that they can buy supplies.

SigfridO: OK listen. Let me ask you. I think they should buy the textbooks now because instead of buying only one thing like only the computers, you can buy the textbooks, supplies, music equipment, and PE equipment instead of one thing like computer.

aiert: But we are gonna have, um . . . we gonna have, we are gonna still have um . . . money left over to buy the computers.

SigfridO: We are able to . . . I think to buy the stock of music and PE and textbooks.

nina: I agree with Sigfrido because um that we will need more supplies than just computers and it will be boring to have just computers.

Excerpt 2 illustrates exploratory talk and co-construction of arguments. As the teacher minimized his talk, students produced longer utterances, which in turn generated more complex reasoning and thinking. Students incorporated conflicting perspectives (buying computers vs. textbooks) and engaged critically but constructively with each other’s ideas. Students counterargued, but counter-arguments and rebuttals were justified and alternative perspectives were offered. Collective reasoning and joint decision-making were evident in Excerpt 2.

g183-208-Nov16-RTE.indd 197 11/22/16 3:51 PM

Page 16: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

198 Research in the Teaching of English Volume 51 November 2016

Influence of Teacher Talk on Discussion ProficiencyTable 5 presents percentages of teacher talk by group and by discourse feature. To investigate the teacher talk differences, independent samples t tests were conducted to compare the average percentages of teacher talk features between the mainstream class and bilingual class. Compared with the teacher in the mainstream class, Mrs. Lehmann, the teacher in the bilingual class, Mr. Herrera, asked more factual ques-tions, t (14) = 2.06, p = .05; and had more turn-taking management moves, t (14) = 1.99, p = .06. However, Mrs. Lehmann used more effective scaffolding moves, such as stepping in and reminding, t (14) = -3.33, p < .001. We further compared the teacher talk differences between the more and less proficient groups in each class. The only significant difference was that, in the mainstream class, the teacher asked for more clarifications in the more proficient group’s discussions than in the discussions of the less proficient group, t (6) = 5.22, p = .002. In the bilingual class, there was no evidence that teacher talk differed between the more and less proficient groups.

table 5. Percentages of Teacher Talk by Group across Four Discussions

Mainstream Class Bilingual Class

Variable MORe PROFiCient (n = 7)

less PROFiCient (n = 7)

aVeRage MORe PROFiCient (n = 7)

less PROFiCient (n = 6)

aVeRage

big QUestiOn 6.23 10.07 8.15 5.87 5.73 5.80

FaCtUal QUestiOns* 0.00 0.00 0.00 11.70 7.40 9.55

tURn-taking ManageMent* 1.67 1.04 1.35 13.52 8.09 10.81

COMMents anD RePetitiOns 3.33 1.04 2.19 16.12 5.44 10.78

PeRsOnal COnneCtiOn 6.48 0.00 3.24 2.05 2.4 2.22

asking FOR ClaRiFiCatiOn* 9.75 0.00 4.87 6.07 5.53 5.80

Challenging 11.53 7.99 9.76 17.6 15.27 16.44

enCOURaging 5.77 10.42 8.09 3.75 2.50 3.12

FOsteRing inDePenDenCe 2.06 5.21 3.64 0.66 2.21 1.43

gOal setting 1.67 0.00 0.83 0.00 0.00 0.00

PROMPting FOR OPiniOn OR eViDenCe 11.09 20.83 15.96 25.71 28.33 27.02

stePPing in anD ReMinDing** 7.87 12.15 10.01 0.66 0.00 0.33

sUMMing UP anD ReFOCUsing 10.19 8.33 9.26 2.43 4.17 3.30

teaCheR DebRieFing 1.67 10.42 6.04 1.04 0.93 0.98

enCOURaging stUDent DebRieFing

thinking OUt lOUD

4.63

2.80

2.78

3.82

3.70

3.31

3.82

0.00

0.00

5.00

1.91

2.50

* p < .05. ** p < .01.

g183-208-Nov16-RTE.indd 198 11/22/16 3:51 PM

Page 17: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 199

Illustrations of Teacher Talk in the Mainstream and Bilingual ClassesTo illustrate teacher talk differences in the mainstream and bilingual classes, two excerpts, taken from the first discussion, are presented to complement the quan-titative analyses already described. Excerpt 3 is chosen from the beginning of the first discussion in the red (more proficient) group of the bilingual class. The group was made up of seven Latino ELLs, two boys and five girls. In this excerpt, students discussed the story Ronald Morgan Goes to Bat (Giff, 1990), in which Ronald is completely inept at playing baseball: he cannot hit the ball, lets the ball roll between his legs, and even runs to the wrong base. He does, however, have great team spirit and is good at cheering on the team. The Big Question is: Should the coach let Ronald play on the team?

Mr. Herrera: Should Mr. Spano let Ronald Morgan play with the team?StudentS: Yes.Mr. Herrera: Okay? Why do you think? Give me your opinion why do you

think Mr. Spano should let him play in the team? [Students raise hands.]Mr. Herrera: Remember, we don’t need to raise our hands. Just go ahead.SigfridO: I think Mr. Spano should let, should let . . . Ronald Morgan play

because he has . . . spirit.Mr. Herrera: Okay. All right.JOuHann: I think . . . Ronald . . . Mr. Spano should let R. M. play because ev-

erybody has a chance. [looks at the teacher]Mr. Herrera: OK.azli: I think Mr. Spano should let R. M. play so he could learn uh, what to do.Mr. Herrera: Okay. What to do with what?azli: With . . . um . . . with . . . the . . . sport. [looks at the teacher]Mr. Herrera: OK.

As shown in the excerpt, Mr. Herrera seemed uncomfortable about letting go of students and retained authority in turn taking and topic management in the first discussion. Therefore, the less proficient discussion in the bilingual class cannot be attributed solely to lower English proficiency, but also to the teacher’s unwillingness to relinquish control over topic and turn taking.

Excerpt 4, below, is chosen from the same first discussion in the green (less proficient) group of the mainstream class. Mrs. Lehmann set up clear goals and expectations from the very first discussion and allowed students to manage their own free-flowing discussions. She used critical scaffolding moves to foster stu-dent independence and challenge students’ thinking when their arguments were single-sided.

g183-208-Nov16-RTE.indd 199 11/22/16 3:51 PM

Page 18: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

200 Research in the Teaching of English Volume 51 November 2016

dOMinica: Wait, I am still talking [waves hand to Kaelyn and students laugh]. He cannot throw the ball. He can’t even . . . he holds the bat upside down . . . and he still don’t know how to play baseball . . . and baseball takes a lot of skills to play.

MariO: But the coach said that he had team spirit.dOMinica: That’s team spirit. He can be a cheerleader. But he cannot play.Kaelyn: But . . . but, umm . . . it, it’s supposed to be yes because he could have

chances and in the third chance he could make it.MariO: Yeah, because look at these things. Can he just tries until he get the

hang of it and be good it?dOMinica: But I still think no.Kaelyn: At least say yes.dOMinica: OK, I will go with yes then. [Students laugh.]MrS. leHMann: You don’t have to change your answer just because. You can feel

your own way. You don’t have to say yes because they say yes.

Influence of Student Backgrounds on Discussion ProficiencyTable 6 presents the descriptive statistics of pretest performance and home lan-guage backgrounds for each group. Independent samples t tests showed that, in the bilingual class, there was a nonsignificant trend toward the more proficient group performing better than the less proficient group on sentence grammar, t (11) = 1.98, p = .07. In the mainstream class, there was a nonsignificant trend toward the more proficient group having better reading comprehension than the less proficient group, t (12) = 1.98, p = .07. No significant group difference was found for vocabulary in either class. These results suggest that group discussion quality is associated with oral English skills (sentence grammar) for limited Eng-lish proficient students and that reading comprehension is critical for discussion proficiency among fluent English language learners.

To investigate which aspects of students’ home language and literacy envi-ronment can distinguish more proficient discussion groups from less proficient groups, independent samples t tests were performed. Results showed that in the bilingual class, the more proficient group spoke more English at home than the less proficient group, t (11) = 1. 82, p = .09; and parents of the less proficient group reported greater frequencies of helping with their children’s homework, t (10) = -1.57, p = .04. In the mainstream class, parents in the less proficient group reported greater frequencies of helping their children with homework, t (8) = -3.05, p = .01. No significant group difference was found in other home language background factors in either class. These results suggest a connection between home literacy support and students’ discussion proficiency.

g183-208-Nov16-RTE.indd 200 11/22/16 3:51 PM

Page 19: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 201

Student Talk Features and Literacy Outcomes Table 7 displays the Pearson correlations between student talk features and post-intervention student language and literacy performance. Students who talked more in CR discussions tended to have better listening comprehension skills, to produce oral narratives with greater vocabulary diversity (TTR) and greater fluency (words per minute), and to write reflective essays with a greater number of satisfactory rea-sons, p < .05. The frequencies of alternative perspectives, elaborations, extratextual connections, and uses of textual evidence had moderate correlations with language outcome measures (e.g., listening, reading, vocabulary diversity and fluency in narratives, and vocabulary diversity in writing)(r = .18–.59). The frequencies of questions (prompting for clarifications and explanations, requesting positions and reasons), clarifications and explanations, confirmations, and predictions had relatively low correlations with language outcomes (r = .03–.47). Unexpectedly, the discourse features had negative or no correlations with the length of narratives and reflective essays, or narrative quality. Discourse features had negative or low correlations with the use of counterarguments, rebuttals, and textual evidence in reflective essays, which may have been due to low frequencies of these argument elements in student essays.

table 6. Means and Standard Deviations of Pretest Scores and Home Language Backgrounds by Discussion Group

Mainstream Class Bilingual Class

MORe PROFiCient (n = 7) M (sD)

less PROFiCient (n = 7) M (sD)

MORe PROFiCient (n = 7) M (sD)

less PROFiCient (n = 6) M (sD)

Pretests

VOCabUlaRY 0.66 (0.12) 0.63 (0.13) 0.48 (0.07) 0.39 (0.15)

sentenCe gRaMMaR 0.88 (0.01) 0.88 (0.02) 0.87 (0.02) 0.78 (0.07)

gates-MaCginitie ReaDing COMPRehensiOn

0.76 (0.12) 0.61 (0.15) 0.56 (0.14) 0.49 (0.15)

Home language backgrounds

ChilD langUage Use* 3.58 (1.36) 3.50 (1.04) 2.98 (0.40) 2.14 (0.61)

ChilD langUage eXPOsURe 3.48 (1.45) 3.20 (1.28) 2.52 (0.41) 2.19 (0.20)

ReaDing FReQUenCies 4.92 (0.26) 4.70 (0.48) 4.30 (0.85) 3.28 (0.99)

ChilD bOOks ResOURCes 1.07 (0.26) 2.60 (2.06) 1.76 (0.83) 1.00 (0.00)

ReaDing staRt age 3.85 (1.40) 3.20 (1.03) 3.30 (1.93) 1.64 (1.69)

MOtheR eDUCatiOn 4.00 (1.92) 3.40 (0.51) 2.53 (0.77) 3.70 (1.04)

FatheR eDUCatiOn 3.64 (1.33) 3.00 (0.00) 3.15 (0.98) 3.94 (1.02)

PaRent assistanCe with hOMewORk* 1.83 (0.75) 3.75 (1.25) 3.00 (1.00) 4.50 (1.00)

Note: Background scores were calculated as follows: child language use/exposure: only Spanish (1), mostly Spanish (2), equal amounts of Spanish and English (3), mostly English (4), only English (5); reading frequencies: never (1), once a month (2), once a week (3), 2–3 times a week (4), every day (5); child book resources: 1–20 (1), 21–40 (2), 41–60 (3), 61–80 (4), more (5); reading start age: 4 years old or later (1), 3–4 years old (2), 2–3 years old (3), 1–2 years old (4), 0–1 year old (5); parent education: elementary school (1), middle school (2), high school (3), two-year college (4), four-year college (5), graduate school and above (6); parent assistance with homework: never (1), once a month (2), once a week (3), 2–3 times a week (4), every day (5). * p < .05.

Bilingual Class

g183-208-Nov16-RTE.indd 201 11/22/16 3:51 PM

Page 20: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

202 Research in the Teaching of English Volume 51 November 2016

DiscussionThis study sought to investigate the variation in the proficiency of small-group peer-led literature discussions and the influence of teacher talk and student back-grounds on group discussion proficiency in two 5th-grade classrooms serving mainly Spanish-speaking English language learners. Greater group variability in discussion proficiency was found in the mainstream class than in the bilingual class, a consistent finding from both the holistic ratings and utterance-level coding. In the mainstream class, the more proficient group talked significantly more and used less nontransactive talk and cumulative talk than the less proficient group, while the more proficient groups in both classes tended to use more exploratory talk. Nontransactive talk was predominant in the bilingual classroom. As shown

table 7. Pearson Correlations between Discourse Features of Collaborative Reasoning Discussions and Language Performance (N = 27)

Note. S_Turn: student total turns; RQ_CE: request for clarification and explanation; RQ_PR: request for position and reason; ALT: alternative perspective; ELAB: elaboration; CLEX: clarification and explanation; CONF: confir-mation; EXCN: extratextual connection; TEVD: textual evidence; PRED: prediction; Gates-MacGinitie reading comprehension; TTR: type-token ratio; MLU: mean length of utterance; NSS: narrative scoring scheme; CTA: counterargument; RBT: rebuttal.* p < .05

S_Turn

RQ_CE RQ_PR ALT ELAB CLEX CONF EXCN TEVD PRED

Posttests

listening .39* .20 -.09 .47* .45* .27 .28 .34 .49* .29

ReaDing .16 .14 .31 .39* .18 .00 .34 .41* .35 -.04

Narrative

t-Unit -.36 -.26 -.07 -.48* -.46* -.29 -.41* -.43* -.48* -.21

ttR .42* .30 .00 .59* .59* .29 .42* .48* .56* .09

MlU .03 .05 -.23 .14 .11 -.00 .24 .23 .37 -.14

wORDs/Min. .52* .04 -.14 .58* .52* .45* .47* .51* .46* .45*

nss -.02 -.12 -.47* .12 -.01 -.06 .13 .15 -.03 -.14

Writing

t-Unit .06 .18 .28 -.14 -.23 -.11 -.10 -.14 -.00 -.02

ttR .26 .27 .08 .40* .45* .20 .39* .34 .32 .04

MlU .29 -.02 -.11 .36 .08 .15 .03 .29 .05 .19

ReasOn .47* .34 .28 .39* .03 .05 .23 .29 .28 .04

Cta/Rbt .11 .16 -.01 .18 -.05 -.13 -.09 .04 .02 -.10

eViDenCe -.18 -.06 .11 -.38* -.33 -.12 -.27 -.29 -.31 .20

g183-208-Nov16-RTE.indd 202 11/22/16 3:51 PM

Page 21: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 203

in the discussion excerpts, students in the bilingual class (especially the less pro-ficient group) seemed more reluctant, or less able, to elaborate ideas in extended speech. These findings reflect the fact that co-construction of argument, the most advanced form of argumentation, takes time to develop, especially for limited English proficiency students. As students become more skillful in managing their own discussions and developing a sound argument schema, they gradually appropriate exploratory talk by engaging in critical consideration of multiple perspectives, weighing arguments, and collectively searching for the best solution to the central issue.

The second finding is that the two teachers differed greatly in their ways of fa-cilitating CR discussions and that teachers’ practices influenced the level of student participation in discussions. The teacher in the mainstream class, Mrs. Lehmann, was skillful in setting up clear goals and expectations, employing a wide range of effective scaffolding moves to facilitate student participation and argumentation, modeling debriefing, and encouraging student reflections. She relinquished control over turn taking from the very beginning. ELLs in her class were active participants and expressed ideas freely. Friendly peer relations between ELLs and non-ELLs were evidenced in her class, as ELLs were constantly encouraged and complimented by their native English speaking peers. However, the teacher in the bilingual class, Mr. Herrera, had difficulty adjusting to the new discourse mode in the beginning. He retained authority over turn taking and topics, asked more factual questions, and constantly evaluated student responses in the first discussion. He was able to gradually release control and encourage more student interactions over time, but used limited scaffolding moves (e.g., prompting for opinion and challenging).

The current study provides quantitative discourse data and representative snapshots characterizing the student talk and teacher talk differences between the discussion groups designated as more or less proficient. To our knowledge, this study is among the first attempts to investigate variation in discussion proficiency and the influence of teacher talk and student backgrounds on discussion profi-ciency in classrooms serving predominantly Spanish-speaking ELLs. Given that these students belong to demographic groups that are consistently underserved in today’s schools, analysis of teacher practices and snapshots of the character-istics of discussion groups that differ in their discussion proficiency levels make a unique contribution to the field of literacy instruction for ELLs in at least two ways. First, the current findings add to the growing literature about the critical role of teacher modeling and scaffolding in productive text-based discussions, especially in classrooms serving mainly Spanish-speaking ELLs. Our findings support previous research in mainly English-speaking children that suggests pro-ductive discussions occur when students hold the floor for an extended period of time and have interpretive authority and control of turn taking (see Murphy et al., 2009, for a meta-analysis). Second, the findings have implications for teacher professional development intended to promote quality talk in ELLs’ classrooms. Previous research suggests that conceptualizing talk moves as tools helps teachers learn to facilitate academically productive talk (Michaels & O’ Connor, 2015).

g183-208-Nov16-RTE.indd 203 11/22/16 3:51 PM

Page 22: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

204 Research in the Teaching of English Volume 51 November 2016

Teachers need to understand how and why a particular talk move might support or hinder students’ thinking about text in order to make sustained shifts in their discourse practices.

Other important sources of variation in students’ discussion proficiency are their prior language and literacy backgrounds and home language environment. Students in the more proficient discussion groups demonstrated better sentence grammar skills and/or reading comprehension skills than students in the less pro-ficient groups. In the bilingual class, two aspects of home language environment differentiated the more and less proficient discussion groups: English language use at home and parent assistance with homework. These findings expand previous evidence about the role of home literacy and language environment on Spanish-English bilinguals’ vocabulary and oral language development (Duursma et al., 2007).

The current results highlight the importance of talk quantity and extended dialogue (alternative perspectives and elaborations) as well as inter- or extratextual connections for English learners’ language and reasoning development. Consistent with Soter et al.’s (2008) meta-analysis, in this study, student talk quantity and discourse indicators of high-level comprehension were significantly correlated with student language and reasoning outcomes. These findings have extended the previous intervention studies of classroom discussion that compared the discourse features and learning outcomes of the experimental group and control group (e.g., Reznitskaya et al., 2012; Rojas-Drummod, Pérez, Vélez, Gómez, & Mendoza, 2003).

The current study has several limitations that suggest directions for future research. First of all, given the small sample size and the fact that only a small fraction of the school and nonschool (home language background) influences on student learning were examined during the period of observation, it is possible that the observed discourse changes over time were due to factors other than the CR intervention. Additional research is needed to replicate the results with a larger sample size, a longer duration of intervention, and a full account of school and home influencing factors.

Second, the direct comparison of the two teachers in the current study may simplify the complexities of teacher influence on student talk, because the teacher in the mainstream class had more experience in CR prior to the study and the teacher in the bilingual class was new to CR. Although not explicitly examined in the cur-rent study, the teachers’ pedagogical approaches and student peer relationships may have contributed to the different levels of student participation in the mainstream and bilingual classes. The two teachers may differ in their understanding of ELLs’ social and cultural needs in a mainstream classroom or in a sheltered bilingual classroom (Yoon, 2008), which could contribute to different teaching practices. Based on our observations, ELLs and native English speakers in the proficient discussion group in the mainstream class were friendly and supportive, which may have contributed to the better discussion quality in those groups. Future research should also examine the native English speakers’ view of ELLs in the mainstream class and discourse differences between ELLs who participate more or less.

g183-208-Nov16-RTE.indd 204 11/22/16 3:51 PM

Page 23: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 205

The results of this study have both theoretical and practical implications. The findings contribute to our understanding of the sources of variation in discus-sion proficiency among groups consisting predominantly of ELLs, as well as to our understanding of small-group discussion as a tool for promoting language and literacy development of ELLs. The study provides important insights into the complex interactional classroom dynamics that influence the teaching and learn-ing of ELLs. The results indicate that teachers’ interactions with the ELLs varied and that the teachers’ different approaches were connected to the ELLs’ differing levels of participation in group discussions. These findings will help teachers bet-ter understand the social dynamics of the ELL classroom and increase teachers’ awareness of how their language use and teaching practices support or constrain ELLs’ learning and participatory behaviors in the classroom.

ACkNOWLEDGMENT

The research reported in this paper was partially supported by the Institute of Education Sci-

ences, U.S. Department of Education, through Grant R305A080347 to the University of Illinois

at Urbana-Champaign (principal investigator: Richard C. Anderson). The opinions expressed are

those of the authors and do not represent views of the Institute of Education Sciences or the U.S.

Department of Education. The research is also supported by the startup funds provided to the

first author at Western Kentucky University.

NOTE

1. All teacher and student names have been replaced with pseudonyms.

REFERENCES

AlmAsi, J. F., O’FlAhAvAn, J. F., & AryA, P. (2001). A comparative analysis of student and teacher development in more and less proficient discussions of literature. Reading Research Quarterly, 36, 96–120.

AndersOn, r. C., & FreebOdy, P. (1983). Read-ing comprehension and the assessment and acquisition of word knowledge. In B. Hutson (Ed.), Advances in reading/language research (Vol. 2, pp. 231–256). Greenwich, CT: JAI Press.

Applebee, A., lAnger, J., nystrAnd, m., & gAmOrAn, A. (2003). Discussion-based approaches to develop understanding: Class-room instruction and student performance

in middle and high school English. American Educational Research Journal, 40, 685–730.

August, d., & shAnAhAn, t. (2006). Devel-oping literacy in second-language learners: Report of the National Literacy Panel on language-minority children and youth. Mah-wah, NJ: Erlbaum.

berkOwitz, m., & gibbs, J. (1983). Measuring the developmental features of moral discus-sion. Merrill-Palmer Quarterly, 29, 399–410.

bOyd, m. p., & rubin, d. (2002). Elaborated student talk in an elementary ESoL class-room. Research in the Teaching of English, 36, 495–530.

g183-208-Nov16-RTE.indd 205 11/22/16 3:51 PM

Page 24: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

206 Research in the Teaching of English Volume 51 November 2016

Chinn, C., AndersOn, r. C., & wAggOner, M. (2001). Patterns of discourse during two kinds of literature discussion. Reading Research Quarterly, 36, 378–411.

ClArk, A. m., AndersOn, r. C., ArChOdidOu, A., nguyen-JAhiel, k., kuO, l.-J., & kim, I. (2003). Collaborative reasoning: Expanding ways for children to talk and think in the classroom. Educational Psychology Review, 15, 181–198.

duursmA, e., rOmerO-COntrerAs, s., szuber, A., prOCtOr, p., & snOw, C. E. (2007). The role of home literacy and language environ-ment on bilinguals’ English and Spanish vocabulary development. Applied Psycholin-guistics, 28, 171–190.

giFF, P. R. (1990). Ronald Morgan goes to bat. New York: Puffin Books.

gOAtley, v. J., brOCk, C. h., & rAphAel, t. e. (1995). Diverse learners participating in regular education book clubs. Reading Research Quarterly, 30, 352–380.

hunt, K. W. (1977). Early blooming and late blooming syntactic structures. In C. R. Cooper & L. Odell (Eds.), Evaluating writ-ing: Describing, measuring, and judging (pp. 91–104). Urbana, IL: National Council of Teachers of English.

JAdAllAh, m., AndersOn, r., nguyen-JAhiel, k., miller, b. w., kim, i.-h., kuO, l.-J., dOng, t., & wu, X. (2011). Influence of a teacher’s scaffolding moves during child-led small-group discussions. American Educa-tional Research Journal, 48, 194–230.

JOhnsOn, J. s., & newpOrt, e. l. (1989). Critical period effects in second language learning: The influence of maturational stage on the acquisition of English as a second language. Cognitive Psychology, 21(1), 60–99.

kOng, A., & FitCh, e. (2002). Using book club to engage culturally and linguistically diverse learners in reading, writing, and talking about books. The Reading Teacher, 56, 352–362.

kOng, A., & peArsOn, P. D. (2003). The road to participation: The construction of a

literacy practice in a learning community of linguistically diverse learners. Research in the Teaching of English, 38, 85–124.

kruger, A. C., & tOmAsellO, m. (1986). Trans-active discussions with peers and adults. De-velopmental Psychology, 22, 681–685.

kuCer, s. b., & silvA, C. (1999). The English literacy development of bilingual students within a transitional whole language cur-riculum. Bilingual Research Journal, 23, 347–371.

kumAr, s., & miller, k. F. (2003). Let SMIL be your umbrella: Software tools for tran-scribing, coding, and presenting digital video in behavioral research. Behavior Research Methods, 37, 359–367.

lin, t.-J., JAdAllAh, m., AndersOn, r. C., bAker, A. r., nguyen-JAhiel, k., kim, i.-h., . . . wu, X. (2015). Less is more: Teachers’ influence during peer collaboration. Journal of Educational Psychology, 107, 609–629.

MACginitie, w. h., mACginitie, r. k., mAriA, k., & dreyer, l. g. (2000). Gates-MacGinitie Reading Tests: Level 4, Form S (4th ed.). Itasca, IL: Riverside.

MArChAnt, h. g., rOyer, J. m., & greene, B. A. (1988). Superior reliability and validity for a new form of the sentence verification technique for measuring comprehension. Educational and Psychological Measurement, 48, 827–834.

mArtinez-rOldAn, C. m., & lOpez-rObert-sOn, J. M. (1999). Initiating literature circles in a first-grade bilingual classroom. The Reading Teacher, 53, 270–281.

mAyer, M. (1969). Frog, where are you? New York: Dial Books.

mCnurlen, B. (1998). Pine wood derby. Cham-paign, IL: Center for the Study of Reading.

merCer, N. (1995). The guided construction of knowledge: Talk amongst teachers and learn-ers. Philadelphia: Multilingual Matters.

merCer, N. (1996). The quality of talk in children’s collaborative activity in the class-room. Learning and Instruction, 6, 359–378.

g183-208-Nov16-RTE.indd 206 11/22/16 3:51 PM

Page 25: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

Zhang et al. What Makes a More Proficient Discussion Group? 207

merCer, n. (2002). Diversity and common-ality in the analysis of talk. Journal of the Learning Sciences, 11, 369–371.

miChAels, s., & O’COnnOr, C. (2015). Con-ceptualizing talk moves as tools: Profession-al development approaches for academically productive discussions. In L. B. Resnic, C. Asterhan, & S. N. Clarke (Eds.), Social-izing intelligence through academic talk and dialogue (pp. 347–362). Washington, DC: American Educational Research Associa-tion.

miller, J., & ChApmAn, R. (2003). Systematic analysis of language transcripts [Computer software]. Madison: Language Analysis Laboratory, Waisman Center, University of Wisconsin.

murphy, p. k., wilkinsOn, i. A. g., sOter, A. O., hennessey, m. n., & AleXAnder, J. F. (2009). Examining the effects of classroom discussion on students’ high-level compre-hension of text: A meta-analysis. Journal of Educational Psychology, 101, 740–764.

nguyen-JAhiel, k. (1996). Crystal’s vote. Champaign, IL: Center for the Study of Reading.

nystrAnd, M. (1997). Opening dialogue: Understanding the dynamics of language and learning in the English classroom. New York: Teachers College Press.

perAltA-nAsh, C., & dutCh, J. A. (2000). Lit-erature circles: Creating an environment for choice. Primary Voice K–6, 8(4), 29–37.

QuAlitAtive sOlutiOns And reseArCh. (2002). QSR NVivo [Computer software]. Victoria, Australia: Author.

reznitskAyA, A., AndersOn, r. C., mCnurlen, b., nguyen-JAhiel, k., ArChOdidOu, A., & kim, S. (2001). Influence of oral discussion on written argument. Discourse Processes, 32, 155–175.

reznitskAyA, A., glinA, m., CArOlAn, b., mi-ChAud, O., rOgers, J., & seQueriA, l. (2012). Examining transfer effects from dialogic dis-cussions to new tasks and contexts. Contem-porary Educational Psychology, 37, 288–306.

rOJAs-drummOd, s., pérez, v., vélez, m., gómez, l., & mendOzA, A. (2003). Talking for reasoning among Mexican primary school children. Learning and Instruction, 13, 653-670.

rOyer, J. m., & CArlO, m. s. (1991). Assessing the language acquisition progress of limited English proficient students: Problems and a new alternative. Applied Measurement in Education, 4(2), 85–113.

sAunders, w. m., & gOldenberg, C. (1999). Effects of instructional conversations and lit-erature logs on limited- and fluent-English-proficient students’ story comprehension and thematic understanding. The Elementary School Journal, 99, 277–301.

sOter, A. O., wilkinsOn, i. A. g., murphy, p. k., rudge, l., reninger, k., & edwArds, m. (2008). What the discourse tells us: Talk and indicators of high-level comprehension. International Journal of Educational Research, 47, 372–391.

webb, N. (1991). Task-related verbal interac-tion and mathematical learning in small groups. Journal of Research in Mathematics Education, 22, 366–389.

wegeriF, r., & merCer, n. (1997). A dialogi-cal framework for investigating talk. In R. Wegerif & P. Scrimshaw (Eds.), Comput-ers and talk in the primary classroom (pp. 49–65). Clevedon, United Kingdom: Multi-lingual Matters.

williAms, J. (2001). Classroom conversations: Opportunities to learn for ESL students in mainstream classrooms. The Reading Teacher, 54, 750–757.

yOOn, B. (2008). Uninvited guests: The influ-ence of teachers’ roles and pedagogies on the positioning of English language learners in the regular classroom. American Educational Research Journal, 45, 495–522.

ZhAng, J., AndersOn, r. C., & nguyen-JAhiel, k. (2013). Language-rich discussions for English language learners. International Jour-nal of Educational Research, 58, 44–60.

g183-208-Nov16-RTE.indd 207 11/22/16 3:51 PM

Page 26: What Makes a More Proficient Discussion Group in English ... · Robertson, 1999). Given the benefits of text discussion for English language learners and the underutilization of discussions

208 Research in the Teaching of English Volume 51 November 2016

Jie Zhang is an associate professor in the Department of Curriculum and Instruction at the University of Houston. Chunling Niu is a research coordinator in the College of Social Work, University of Kentucky.Shahbaz Munawar was a research assistant at Western Kentucky University.Richard C. Anderson is a professor emeritus and university scholar at the University of Illinois at Urbana-Champaign.

Initial submission: September 11, 2014Final revision submitted: March 31, 2016

Accepted: May 25, 2016

Cultivating New Voices among Scholars of Color Program 2016–2018

The NCTE Research Foundation’s Cultivating New Voices among Scholars of Color (CNV) program is designed to provide two years of support, mentoring, and network-ing opportunities for early-career scholars of color. The program aims to work with doctoral candidates and early-career postsecondary faculty of color to cultivate the ability to draw from their own cultural and linguistic perspectives as they conceptual-ize, plan, conduct, write, and disseminate findings from their research. The program provides socialization into the research community and interaction with established scholars whose own work can be enriched by their engagement with new ideas and perspectives. The 2016–2018 CNV program participants are listed here:

Blanca Caldas, University of Minnesota-Twin CitiesJustin Coles, Michigan State University, East LansingMarcus Croom, University of Illinois at ChicagoMichael Domínguez, University of North Carolina at Chapel HillTracey Flores, Arizona State University, TempeBrooke Harris Garad, The Ohio State University, ColumbusLaura Gonzales, University of Texas at El PasoMónica González, University of Colorado, BoulderFahima Ife, Louisiana State University, Baton RougeGrace MyHyun Kim, University of California, BerkeleyJamila Lyiscott, Teachers College, Columbia University, New York, NYEsther O. Ohito, Mills College, Oakland, CAGrace Player, University of Pennsylvania, PhiladelphiaDywanna E. Smith, University of South Carolina, Columbia

g183-208-Nov16-RTE.indd 208 11/22/16 3:51 PM