Through the eyes of my reader: a strategy for improving audience perspective in children's descriptive writing.

Abstract. The findings in the study suggest "reading-as-the-reader" can improve 5th- and 9th-grade writers" ability to compose descriptive writing consistent with their readers" informational needs. Participants included: 154 writers (78 fifth-graders and 76 ninth-graders) and 52 ninth-grade readers. The study adapted the referential communication design from Traxler and Gernsbacher (1992, 1993) to investigate whether young writers can benefit from a specific perspective-taking condition as they compose and revise their descriptions of tangrams over three separate writing sessions. Three conditions were contrasted: a feedback-only condition, a "rating other" condition, and a "reading-as-the-reader" condition. Readers' correct description-to-tangram matches made for each of three sessions served as the dependent measure. Repeated measures analysis revealed that both the 5th and the 9th-graders showed consistent significant improvement under the "read-as-the-reader" condition when revising their essays and when drafting anew (F(4,296) = 2.96, p=.019). The results indicate that when young writers engage in a process that mirrors their readers" experiences, they can more accurately revise their descriptive writing to meet their readers' informational needs. A qualitative analysis of the writers" reflection comments completed at the end of the experiment found that reading-as-the-reader was a positive task that enabled students to consider descriptive strategies and the interpretive effect that their descriptions had on their readers. An analysis of the readers' reflections on their experiences revealed that readers were assisted most if the writer included in his or her descriptions a familiar "global" analogy, with additional specific spatial language that describes the internal organization and spatial orientation of the tangram figure.

**********

Writing serves numerous purposes. Reading satisfies myriad needs. Lacking a conversational partner, writers are challenged to create a shared perspective with their readers. Writing that serves descriptive purposes presents unique demands for writers. For descriptive writing to be successful, readers need critical information that articulates a clear picture through the writer's words; a writer must paint a picture that the reader can "see." For example, in a "Study of Two Pears," Wallace Stevens (1954) composes a picture of pears resting in a bowl: "They are yellow forms, composed of curves, bulging toward the base. They are touched red.... There are bits of blue. A hard dry leaf hangs from the stem ... the shadows of the pears are blobs on the green cloth" (p. 196). Stevens colors an image with sensuous textures and crystal-like clarity. Without actually seeing a fruit bowl, many readers can delight in this simple but vivid still-life description.

Experienced writers can create pictures with words. Less experienced writers do not always "see" their words as their readers do. Many primary-school and middle-school students find writing personally meaningful, and their readers can marvel at the colorful insight they can create through their writing. At times, however, a student's writing can lack certain information crucial for the reader's full participation in the text. Taking the perspective of the reader represents a developmental accomplishment for young writers.

The following study focused on how 5th- and 9th-grade writers described tangrams to other student readers. The writers were randomly assigned to one of three perspective-taking conditions that varied the amount of reader insight a writer received. Each writer was asked to describe a set of tangrams. The readers used the descriptions to select matching tangrams from other similar-looking tangrams. The guiding question for this study was: Can "reading-as-the-reader" help 5th- and 9th-graders in composing descriptive writing consistent with their readers' informational needs?

Theoretical Overview

Writing is simultaneously an individual struggle and a social undertaking (Dyson & Freedman, 1991; Fitzgerald, 1992; Florio, 1979; Flower, 1994). Writers face the individual cognitive task of selecting what information to communicate and how they will communicate it. Inseparably, writers consider who their readers will be and the context of their reading. Writing scholars (Fitzgerald, 1992; Traxler & Gernsbacher, 1992, 1993) theorize that to meet the informational needs of readers, a writer must coordinate at least three interacting mental representations: a representation of personal communicative intent (what do I want to say?), a representation of the text produced (what have I written?), and a representation of the reader's perspective (how will the reader interpret my writing?).

Establishing reciprocity among the reader, writer, and text is the hallmark of experienced writing (Olson, 1994; Witte, 1992). Considerable research (e.g., Beal, 1996; Bereiter & Scardamalia, 1987) has demonstrated that young writers are particularly challenged in learning this writer-reader-text reciprocity. Specific instructional conditions that foster "comprehension monitoring" (Beal, 1996; Fitzgerald & Markham, 1987) and "knowledge-transforming" (Bereiter & Scardamalia, 1987) can help young writers discriminate their intended message from the actual text they have composed, thus influencing the textual quality of their writing. Fewer studies, however, have outlined conditions that may help improve younger writers' awareness of their readers' possible interpretations. For example, Frank (1992) found that subtle manipulation of "audience specification" in writing prompts led 5th-grade writers to compose their newspaper advertisements differently for two separate audiences. The research literature (e.g., Bonk, 1990) remains unclear about instructional conditions that can help young writers view their text from the perspective of their readers, and thus improve the communicative quality of their writing.

To investigate how older students might become more sensitive to their readers' needs, Traxler and Gernsbacher (1992, 1993) asked college students to compose and revise descriptions of tangrams for anonymous readers. The readers' task was to read each description and then select the matching tangram from a group of similar-looking tangrams. The writers who went through a revision process identical to that of their readers consistently wrote the most effective texts. Traxler and Gernsbacher (1993) concluded that the reciprocity among readers' needs, text, and the writer could be successfully accomplished when writers conduct the read-as-their-readers process; that is, when writers learn to take the informational perspective of their readers. If consideration of the reader's needs is critical to "good thinking during composition" (Fitzgerald, 1992, p. 345), then reading-as-the-reader may enable young writers to consider the descriptive needs of their readers. Reading-as-the-reader may be one strategy whereby young writers can coordinate "what do I want to say?" and "what have I written?" with "how will the reader interpret my writing?"

Literature Review

The following review considers research from three disciplinary perspectives: social cognition in referential communication, composing with "audience awareness," and representation and revision in written communication. Although associated with different research traditions, each perspective helps us understand the underlying complexity that affects the writer-reader-text reciprocity.

The Social Cognition of Perspective Taking in Referential Communication

Social cognition is an ability to think about perspective. This includes thinking about another person's thoughts, emotions, and perceptions, combined with the ability to use metacognitive processes to reflect on one's own thoughts, emotions, and perceptions. This "wheels within wheels" thinking is a developmental accomplishment (e.g., Miller, Kessel, & Flavell, 1970; Selman & Byrne, 1974).

Psychologists have used the terms "egocentrism" and "decentering" to define opposite ends on a social cognitive spectrum. Flavell (1992) characterizes egocentrism as children's limited ability to understand that "their own perspective may differ from those of others" (p. 107). Chandler (1977) suggests that decentering is a social cognitive achievement that enables children to "transport themselves into the role or vantage of someone else" (p. 110).

Some developmental psychologists have created scales whereby the emergence of cognitive, emotional, and moral skills are measured on this egocentric/decentric spectrum (e.g., Cox, 1978; Kohlberg, 1984; Selman & Byrne, 1974). Other scholars argue, however, that "it is not possible to pinpoint a single age of onset of social decentration because different types of social cognitive processes become operative at different ages" (Rubin, 1984, p. 218). The development of literacy skills, for example, requires a decentering ability to distinguish "the said" from "the intended" (Olson, 1994). Jokes, lies, and tall tales are great fun for children as they learn that what is said is not always what is meant. "Children understand that someone may misunderstand as long as the judgment is made from their own perspective ... they have yet to come to understand the possibility of another's beliefs about understanding and misunderstanding" (Olson, 1994, p. 129). Children realize that they can misunderstand others, but they do not yet fully understand that others can misunderstand them. Between the ages of 4 and 5, children acquire the rudimentary social cognitive ability to distinguish discrepancies between a speaker's communicative intent, the listener's possible interpretations, and their "actual words" in oral communication (Demorest, Silberstein, Gardner, & Winner, 1983).

The study of social cognitive skills (also labeled "perspective-taking skills") in oral communication has a detailed history (Chandler, 1977; Flavell, 1992). The "referential communicative task" is one of the most fruitful research designs in studying the development of perspective-taking skills (e.g., Bonitatibus & Flavell, 1985; Chandler, 1977; Clark & Wilkes-Gibbs, 1986; Glucksberg & Krauss, 1967; Schober & Clark, 1989). This experimental design requires participants to communicate clearly the critical distinguishing properties of some object or "complex stimulus array" so that another person can recognize that object, based on what the other participant has attempted to communicate (see Chandler, 1977, p. 114). Glucksberg and Krauss (1967), for example, used a referential communicative task that positioned two people of the same age so that they sat on opposing sides of an opaque screen. Third-, 5th-, 7th-, and 9th-graders participated. One person described irregularly shaped figures ("low-encodable graphic designs") to the other person on the other side of the screen. The listener then attempted to select the appropriate design figure from a collection of nearly identical designs. Both participants were able to freely discuss the figures, yet they could not look at the other's designs. The results suggest "that younger children, in addition to displaying limited response repertoires, failed to edit, i.e., they did not modify their message in socially appropriate ways" (Glucksberg & Krauss, 1967, p. 313). Older children were more informationally accurate in orally describing these abstract figures than were younger children. The study concluded that perspective-taking in oral referential communication is a developmental achievement, moving from egocentric language use to a more decentered language use that includes listener-useful descriptions.

The referential communication paradigm has been primarily used for studies, on spoken communication. In several non-developmental studies, the spatial perspectives that speakers take when they describe different environments represented by maps has been thoroughly investigated (e.g., Taylor & Tversky, 1996). Written communication, however, has received limited attention in the literature on perspective-taking development. Kroll and Lempers (1981) investigated the differences between oral and written modes of discourse by having 4th-graders describe the rules of a game to an experimenter. The 4th-graders were more successful at explaining and describing the rules of the game in speaking than they were in writing. Kroll and Lempers concluded that decentration in writing tended to lag behind in speaking for 4thgraders. Social cognition in successful oral communication requires a speaker and listener to coordinate their interpretive perspectives by using facial gestures, intonation, pauses and stress on the words spoken, and on an assumed contextual familiarity. Learning to write from the perspective of the reader, however, requires different social cognitive processes. Writers have no immediate feedback. Decentering in written communication requires a writer to develop a representation of the reader's possible interpretation, a social cognitive ability problematic for young writers (Kroll, 1978; Olson, 1994).

The literature in the development of perspective-taking in written communication remains unclear in terms of what conditions can enhance perspective-taking skills for young writers and at what developmental juncture perspective-taking can help young writers communicate clearly. For example, Cameron, Hunt, and Linton (1996) reviewed a study wherein 2nd-, 4th-, and 6th graders in a highly supportive writing environment were able to effectively revise the content of their advertisements; they had to rewrite the ad copy, which concerned buying puppies, to reflect the change in their audience from farm-dwellers to city-dwellers. Bracewell, Scardamalia, and Bereiter (1978) concluded from two studies on referential communication and audience prompting that only 12th-graders were successful at making revisions that improved the communicative quality of their descriptions. The two studies showed that when 4th- and 8th-graders revised, the communicative quality of their writing actually decreased. Studies that vary specific audience prompting (i.e., for age, fictional or actual world, teacher, other students) have mixed results on the measured effectiveness in taking the reader's perspective (e.g., Brossell, 1985; Charney, 1989; Matsuhashi & Gordon, 1985; Oliver, 1995).

Although underused in research on writing development, a referential communication task is one research approach that may offer insight into the development of children's social cognitive ability to take the perspective of their readers as they write.

Perspectives on "Audience"

The study of the "reader's perspective" (also known variously as "audience awareness" and "sense of audience") has deep historical roots (e.g., Berkenkotter, 1981; Ede, 1984; Kirsch & Roen, 1990; Park, 1986; Rafoth 1989). In an attempt to characterize the various approaches to the reader's perspective, Kroll (1984) summarized three broad, overlapping approaches: the rhetorical, the informational, and the social. The rhetorical perspective is based on Platonic and Aristotelian notions suggesting that if speakers are to communicate successfully, they must match the details of their speaking with the temperament of the listening audience. This perspective views all discourse as persuasive and assumes that the rhetor can know the audience in some analyzable way. In composition studies, this approach to the reader teaches students "methods of analyzing the knowledge, traits, and beliefs of their readers, so that the students will be able to identify their audience with realistic detail" (Kroll, 1984, p. 175). In such scenarios, the writer needs the perspective of readers' beliefs and values to be communicatively successful.

The informational perspective on audience is grounded in the cognitive model of human information processing. In short, "the writer aims to get information into the reader's head" (Kroll, 1984, p. 176). The assumption underlying this perspective on the reader is that new information is provided from the writer, so that this information then is added to the old information contained within the reader's working and long-term memory. This perspective views discourse as influencing all readers' interpretations in similar ways. The writer needs a perspective about the available and unavailable information the reader has in order to communicate successfully.

The social perspective on audience views writing for readers as a "social activity, entailing processes of inferring the thoughts and feelings of the other persons involved in the act of communication" (Kroll, 1984, p. 179). The writer uses language not to persuade or inform, but rather to induce dialogue, reflection, and imagination. In this perspective, the reader is not as knowable to the writer as are the readers in the rhetorical and informational perspectives. Here, the writer may not rely on the reader's perspective as much as on other communicative purposes, because different readers will be affected differently and unpredictably by the author's text.

Rhetorical, informational, and social perspectives of audience thus assume different relationships among the writer, reader, and text, therefore requiring the writer to make different inferences. Additionally, other scholars suggest that the reader's perspective is influenced by an interaction of the function of the discourse with the reactions that a reader has to a particular text (Britton, 1982; O'Keefe & Delia, 1986).

Britton, Burgess, Martin, McLeod, and Rosen (1975) characterized three main functions in writing that are interdependent with reader perspectives. At one end of the functional spectrum is transactional writing, to the middle is expressive writing, and at the other end of the spectrum is poetic writing. Transactional writing "is language to get things done" (p. 88). Transactional writing functions to inform, advise, persuade, and/or instruct the reader. It requires "accurate and specific reference to what is known about reality" (p. 88). Expressive writing functions "close to the self" (p. 90), with little concern for the reader other than the self. The meaning encoded in expressive language is not made explicit, because the writer assumes that the content can be immediately known or is familiar to the readers. Expressive language "submits itself to the free flow of ideas and feelings [;] it is relatively unstructured" (p. 90). The poetic function of language is "an art medium" (p. 90), and results in "patterned verbalizations" of emotions and concepts that may (or may not) evoke emotional responses from the reader.

Audience is clearly a complex phenomenon that includes the intertwining influences of discourse function and reader perspective. Given this broad theoretical background, the present study focuses on what might be termed transactional informative writing. This writing function requires that the writer provide relevant, detailed, and descriptive information so the reader can share the writer's perceptual and visual fields. Some real-world examples of transactional writing are plant identification, road maps and accompanying directions, instructions for making Lego houses and toys, cooking recipes, and procedural writing in experimental studies.

Representation and Revision as Social Cognition in Written Communication

The reader's perspective plays as crucial a role in models of composition and revision. Flower and Hayes (1984) suggest that "as writers compose they create multiple internal and external representations of meaning" (p. 122). Not only must writers coordinate the cognitive representations of the external text they have created (what have I written?) with their internal communicative intent for that text (what do I want to say?), they also must coordinate an internal representation of the perspective of the potential reader (how will the reader interpret my writing?). Successful written communication depends on "the writer's success at attending to the informational needs of the readers" (Roen & Willey, 1988, p. 82). Attending to the informational needs of the reader, however, is problematic in writing, because the writer does not receive immediate feedback from the reader. To develop a representation of a reader's informational needs, a writer must be able "to think from the reader's perspective, to perceive potential trouble sources, and to think as the reader would think" (Fitzgerald, 1992, p. 342). Revision is one path that can lead writers to think about and represent, to re-think and re-represent, their readers' informational needs.

Revision is a constructive, reiterative process that presents special challenges for young writers. Fitzgerald (1987) suggests that revision "means making any changes at any point in the writing process. Revision involves identifying discrepancies between intended and instantiated text, deciding what could or should be changed in the text and how to make desired changes" (p. 484). Flower, Hayes, Carey, Schriver, and Stratman (1986) theorize two underlying variables are necessary for revision: knowledge and intention (pp. 19-20). Experienced and less experienced writers represent knowledge and intention differently, resulting in distinct composition and revision patterns. Expert writers employ a "diagnose/revise" composition strategy, whereas novice writers employ a "detect/ rewrite" composition strategy.

Theoretically, these two strategies exist on opposite ends of a continuum. "Diagnosis is the act of recognizing and categorizing the problems one detects in the text" (Flower et al., 1986, p. 47). Diagnosis includes the categorizing of textual, intentional, and conceptual problems. Once a problem has been detected, diagnosis brings to attention different solutions to reshape, or revise, the text anew: "That's not exactly what I want to say nor what my reader needs, so I'll reshape this and try it again" is one way to characterize this strategy. With detection, however, "the writer, typically working with individual sentences or even clauses, attempts to capture the same propositional content (or a loose approximation of it) in a different semantic or syntactic form" (Flower et al., 1986, p. 43). There is no attempt to restructure the text at a meaning level; "don't look back, just keep writing" is one way to characterize this strategy.

Both representational strategies are useful, depending upon the functional purpose of the writing (e.g., Elbow, 1987). Younger and less experienced writers, however, have a limited experience with certain linguistic knowledge, topic knowledge, reading their own texts, and an understanding of their readers' perspectives, resulting in a predominant use of the detect/rewrite strategy. On the other hand, more experienced writers can employ both strategies; they have a developed ability to coordinate and modify a representation of their communicative intentions, a representation of the text, and a representation of their readers' perspectives. Revision for experienced writers has the potential to generate new meaning and understanding, and can lead to "knowledge-transforming." For less experienced writers, however, revision can be repetitive and cosmetic, and can lead to "knowledge-telling" (Bereiter & Scardamalia, 1987).

Revising for the readers' informational needs requires that a writer be a skilled reader. Olson (1994) theorizes that learning to differentiate one's own perspective in reading from another's perspective is a benchmark in becoming literate: "Readers frequently fail to consider how texts could be understood or misunderstood by readers other than themselves" (p. 135). Learning to recognize that a text one has created does not correspond to personal communicative intent (What do I want to say?) and that the same created text can be taken differently by another reader (No! That's not what I mean!) is crucial to the revision process. A major developmental accomplishment for a young writer is recognizing that "text is a representation for meaning, not meaning itself' (Beal, 1996, p. 221). The ability to "decenter" from one's own text and reread from the perspective of a reader is interdependent with revision.

Revision entails considerable rereading. Flower et al. (1986) argue that the ability to evaluate and revise a text is a basic reading comprehension process. Rereading can lead to rebuilding, replanning, discovering new ideas, and re-representing a "working image of the text." Research from several differing disciplines demonstrates that rereading for the informational needs of the audience during revision can be an effective strategy that leads to successful written communication (e.g., Frank, 1992; Holliway & McCutchen, 2004; Kroll, 1985; Rafoth, 1985, 1989).

Experimental research has found that, in some revision circumstances, adults can modify their internal representations of text and readers' perspectives, and, ultimately, revise their writing in closer accordance with their readers' informational needs. In a series of five referential communicative experiments (Traxler & Gernsbacher, 1992, 1993), first- and second-year college students wrote and revised descriptions of geometric figures (tangrams). The dependent measure for these studies was the percentage of readers who successfully matched tangram descriptions with the appropriate target-tangrams. The research offers an insight into the informational needs of the reader (the reader's informational needs require enough detailed descriptive information to distinguish one tangram from similar-looking tangrams), into the communicative adequacy of the writing, and thus, an insight into the nature of readers' perspectives in descriptive writing.

Traxler and Gernsbacher (1992) compared two groups of writers: one group received minimal feedback (a figure indicating the number of readers who successfully matched their descriptions with appropriate figures); the other group received no feedback. The writers who received minimal level of feedback revised their descriptions in a way that led readers to make more accurate judgments in a second scoring session than did those writers who received no feedback. Traxler and Gernsbacher (1992) concluded that even a minimal feedback can help writers to "envision how readers interpret their texts" (p. 1) by forming and modifying a representation of their readers' informational needs.

In a second study, Traxler and Gernsbacher (1993) investigated the differences between two reading conditions and their influence on revision over the course of three writing sessions. In the first session, the writers were asked to describe eight geometric figures in enough detail so that an anonymous reader could choose the described figures from other similar-looking figures. In the second writing session, writers were randomly assigned to either the "readers' task" or the "rating task." One group of writers rated texts written by another writer in terms of informational adequacy and clarity (e.g., "how clear was the description?"; "how much information did the description contain?" p. 326), while the other group of writers read descriptions written by other writers and then matched them with figures (just as their readers did).

Those writers who read descriptions written by other writers, then chose target figures they believed to fit with the descriptions, revised and wrote higher quality descriptions than those who rated texts on their informational adequacy. The process of reading-as-the-reader, Traxler and Gernsbacher (1993) reasoned, led these writers to create a more accurate representation of the readers' needs than those writers who simply rated descriptions. Traxler and Gernsbacher concluded that "gaining the reader's perspective helps writers communicate more clearly because perspectivetaking helps the writers form a mental representation of how readers interpret their texts" (1993, p. 311).

By taking the role of the reader on a similarly produced text, writers gain the experience needed to revise their own texts in a way that facilitates composing a text consistent with the informational needs of their readers. By reading-as-the-reader, writers can diagnose and revise their writing in closer approximation to the perspective of their readers. For the writers in the Traxler and Gernsbacher study, reading-as-the-reader provided the rereading and revision experience needed to create and modify representations of their readers' informational needs. It is an indirect, noninstructional condition; yet by stepping "in their readers' shoes" (Traxler & Gernsbacher, 1993, p. 311), these writers were able to communicate more accurately (as measured by an increase in the number of readers who correctly matched the descriptions with their target pictures over a series of three reading sessions) than those who did not read-as-the-reader.

Although writing teachers have used similar "environmental activities," such as describing potatoes or seashells that were then matched by other classmates with the appropriate target objects (Daniles, 1990), there has not been a systematic procedure that reveals what younger writers can do, specifically, to revise their transactional descriptive writing to meet the informational needs of their readers.

Purpose of the Study

The specific conditions that might lead children to construct a representation of the reader's perspective is an area not sufficiently studied in the writing acquisition literature (see Beal, 1996; Cameron, Hunt & Linton, 1996; Frank, 1992). The purpose of this study is to inquire whether reading-as-the-reader can help elementary and middle school students write in accordance with their readers' informational needs.

Traxler and Gerusbacher (1992, 1993) offered compelling, but limited, evidence that has interesting implications for understanding writing development, if similar evidence can be collected from younger writers. This study modified the referential communication design used by Traxler and Gernsbacher to investigate what perspective-taking conditions can help younger writers improve the communicative quality of their descriptive writing. The "feedback only," "rating other," and "reading-as-the-reader" conditions were manipulated to investigate how 5th- and 9th-grade writers compose for the informational needs of their readers.

This study investigated these questions:

1. Can the perspective-taking condition "reading as the reader" assist 5th- and 9th-grade writers in accurately meeting the informational needs of their readers?

2. What do the writers' reflections about the experimental revision conditions reveal about meeting their readers' needs?

3. What information do readers need from writers when they are challenged to connect a text (a description) to an abstract geometric figure (a tangram)?

Method

Participants

The 5th-grade and 9th-grade participants for this study came from four school districts in a large metropolitan area in the northwestern region of the United States. There were two groups of participants: the writers and the readers. There were 154 writers: 78 fifth-graders and 76 ninth-graders. The readers were a separate group of 52 ninth-grade readers in advanced placement English classes. There were a total of 206 participants.

Design

A written referential communicative paradigm adapted from Traxler and Gernsbacher (1992, 1993) was used. There were three writing sessions and three reading sessions. Writing and reading sessions were separated by one-week intervals (see Table 1).

Writers. There were three 30- to 35-minute writing sessions. In the first writing session, all writers were given a notebook that included written instructions and three tangram figures to be described. Each tangram and writing space was provided on separate pages (see Figure 1 for a sample page). In the second session, each writer received a typed version of the descriptions that he or she had composed in the first session. During this session, writers were randomly assigned to one of three revision conditions.

[FIGURE 1 OMITTED]

Condition One (feedback only condition). Writers received a sentence for each description indicating whether their reader had successfully matched the given description with the associated target tangrams (e.g., "Your reader was successful in matching your description with the correct tangram" or "Your reader was not successful in matching your description with the correct tangram.") Writers were then asked to make revisions to their original descriptions.

Condition Two (feedback + rating condition). As in condition one, all writers received a feedback sentence indicating whether the reader had successfully matched the given description with the associated target tangram. Then, all writers in this condition received the same three descriptions written by another student. They were asked to rate the descriptions by considering the informational adequacy of each description. They were then asked to make revisions to their own original descriptions.

Condition Three (feedback + read-as-the-reader condition). As in conditions one and two, all writers received a feedback sentence indicating whether their reader had successfully matched the given description with the associated target tangram. Then, all of the writers in this condition were asked to read the same three descriptions written by another student and asked to match descriptions with tangrams (this is exactly what their readers did). They were then asked to revise their original descriptions.

In the third writing session, writers remained in the experimental condition that they were randomly assigned to in session two. During this session, they again received feedback and, for conditions 2 and 3, respectively, they either rated written descriptions, or read-as-the-reader and matched descriptions with tangrams. They were then asked to describe a new set of three tangrams; writers composed new descriptions for tangrams they had not seen. In addition, all writers in the third session were asked to write a brief reflection comment on their tangram-describing experience. The reflections were used to gather the students' insights about their experiences under the three conditions.

Readers. For the entirety of the experiment, the same reader scored each writer. For each of three reading sessions, each reader read and scored 9 tangram descriptions; readers read one set of descriptions from each of the three perspective conditions. For each description, readers saw four similar-looking tangrams (see Figure 2). At the bottom of each reader's page, the readers were asked to follow this direction: "Circle the tangram that is best described by description X." Readers decided which one of the four tangrams was described. In addition, after each reading session, the readers commented in writing about the overall quality of the descriptions that they read. The reading process was repeated on three different reading sessions; once for the original draft descriptions from Session 1, once for revised/rewritten revisions from Session 2, and once for the new descriptions written in Session 3.

[FIGURE 2 OMITTED]

Materials

The stimulus materials consisted of 72 tangram figures (similar to those used by Clark & Wilkes-Gibbs, 1986; Traxler & Gernsbacher, 1992, 1993) that were divided into 18 sets of four similar-looking tangrams.

Each writer received a notebook with a cover page of written instructions and three separate pages with one target-tangram that they described. If additional writing space was needed, the writers were invited to continue on the back side of the page.

The readers received a notebook and a scorebook. The notebook contained nine student descriptions that were typed by the researcher. The descriptions in the notebook were numbered 1-9 and sequentially arranged by writer (writer 1: descriptions 1, 2, and 3; writer 2: descriptions 4, 5, and 6; and writer 3: descriptions 7, 8, and 9). In the score book, the readers made their description-to-target-tangram matches. The scorebook included a page of four similar-looking tangrams for each description. For each description, the readers read the descriptions in the notebook, then, in the scorebook, circled the tangram they thought was best described by the description.

Results

Quantitative Analysis 

The dependent measure for the 2 (grades) x3 (tasks) x3 (sessions) repeated measures analysis was the number of correct description-to-target-tangram matches that each reader made for each description he or she read (for mean differences, see Table 2).

The ANOVA revealed a main effect of grade, F(1,148)=11.01, p=.001. On average, 9th-grade texts yielded more matches (M=2.25, SD=.48) than did 5th-grade texts (M=1.98, SD=.58). There was no main effect of condition, nor of a grade by condition interaction (F<1 in both cases). The main effect of session, F(2,296)=8.76, p<.001, was significant, with session 1 (M=1.88, SD=.86) yielding fewer matches than session 2 (M=2.22, SD=.87) and session 3 (M=2.24, SD=.83). However, the session main effect was compromised by a significant interaction between session and condition, F(4,296)=2.96, p=.019. Post-hoc analyses (Tukey's Honestly Significant Difference (HSD) approach) established that differences between session 1 and sessions 2 and 3 were significant only for the read-as-the-reader group (critical value=.375, p=.05). No other interactions reached significance (F<1).

Qualitative Analysis

An analysis of the writers' reflections about their "reading-as-the-reader" experiences revealed that the task was very useful for these writers. Students portrayed their writing experiences on a variety of levels, usually characterizing the task in some way as fun or boring, insightful or uninspiring. The student reflections were used to generate a general coding scheme that categorized their experiences as positive or negative, useful or not useful (see Table 3 for some examples of positive student responses in the read-as-the-reader condition).

Based on the coding scheme, the percentage of students in each condition who characterized their writing experience positively was calculated. In the feedback condition, 46.5 percent of the 20 students who responded found their experience positive. Within the rate-other condition, 67.9 percent of the 38 who provided a written reflection found their experience positive, and 70.9 percent of the 39 who responded from the read-as-the-reader condition commented that they found the experience to be positive. Not all of the students provided a written reflection, yet it is clear that students in the feedback condition were significantly less likely to characterize their writing experiences as positive, compared with students in the other two groups (Fisher's Exact=6.787, p=.005).

To investigate the "readers' needs," a profile was compiled, based on the readers' open-ended comments made at the end of each reading session. An analysis of their comments about what they needed from their writers revealed that a global conceptual image, created by an analogy with a balance of local shape and spatial elaborations, helped them discriminate and chose the target-tangram from the group of similar-looking tangrams. For example, one reader commented: "The descriptions that were the best were very detailed in the shapes and what the figure looks like it's doing." The readers' profile revealed that the readers' informational needs were met more efficiently by writers who elaborated on the analogical referent with a balance of shape names (e.g., triangle, parallelogram, square), geometrical qualifications (e.g., zigzaggy, diagonal, pointy), and location descriptors (e.g., to the right, on its left, the left one).

A text analysis of the descriptions revealed that many writers, regardless of condition and grade, began their descriptions with analogies (e.g., "It looks like a running fox," "This tangram looks like a ghost flying.") These "spontaneous analogies" (English, 1997, p. 15) may be one way that writers attempt to establish a common perceptual ground with their readers. Writers varied, however, in the way they elaborated on the spatial and geometric qualities of the tangrams they described. Many writers used an object-centered strategy that focused on the intrinsic details of each tangram. This writer's description represents a common strategy: "It looks like a goose. It has a long zigzagging neck. It has a small head and a pointed beak. Its body is kinda [sic] long and it has two feet on top of each other." The text analyses of the descriptive essays generated in this study reveal few structural differences that can be associated directly with condition and/ or session.

Discussion

Implications for Writing Theory and Future Research

This research contributes to a body of literature (e.g., Beal, 1996; Cameron, Hunt, & Linton, 1996; Frank, 1992; Oliver, 1995) that clarifies some of the instructional conditions that can help young writers envision how their readers interpret the texts they have written. The study contributes to our understanding of how younger writers can learn of the reciprocity among writing, reading, and text (Witte, 1992).

Although all three groups received some form of feedback and improved the effectiveness of their texts as they moved from drafting to revising (the session 1/session 2 revision comparison) and from drafting to drafting anew (the session 1/session 3 transfer comparison), the read-as-the-reader group (i.e., the group that used the texts in the same way their readers did) showed more improvement than the other two groups in the transfer comparison. All three groups received numerical feedback about the accuracy of their reader's choice. The rate-other group also read and evaluated descriptive tangram texts written by other students. However, only the read-as-the-reader group was asked to take their readers' perspectives in the actual task of matching texts to tangrams. The repeated measures results of reader matches are supported by the writers' own comments, which suggest that reading-as-the-reader qualitatively changed the experience of these young writers.

The potency of the read-as-the-reader condition emerged only in the transfer comparison, that is, in session 3. There are at least two explanations for why the read-as-the-reader group did not show larger improvement in the revision comparison (session 2). First, it may be the case that an increased frequency of exposure to the strategy is necessary; perhaps more than one experience with reading-as-the-reader is necessary for younger writers to show the benefits. Second, it also may be the case that the nature of the writing task in session 3, compared to session 2, offers a different experience with written text. In the second session (revision session), writers may have been under the influence of the text they had already created; the actual physical text that they composed in the first session may have constrained the creation of a new text, fresh with detail. As Bereiter and Scardamalia (1987) suggest, "The original version of text, because it is perceptually present [emphasis added], has a direct claim on conscious attention. Unless the writer can deliberately bring alternatives to mind, the original text will win for lack of competition" (p. 87). Reading-as-the-reader had the greatest impact when students were given a chance to apply and recontextualize what they had learned from composing one set of texts to the composition of a similar, but new, text. That is, when writers drafted anew in the third session, unconstrained by an existing less-effective text, they were able to demonstrate what they had learned from reading-as-the-reader.

Both of the above explanations may be true. Young writers may benefit more from multiple experiences reading-as-their-readers, and those experiences may be most observable when students are not influenced by the physical presence of a text they wrote previously. A future study might include intervention duration and writing task as independent variables. For example, a group of writers could be asked to draft descriptions in a first writing session. In a second session, writers would be randomly assigned to one of four conditions: reading-as-the-reader + revision; reading-as-the-reader + drafting; rate-other + revision; and rate-other + drafting. In a third session, writers would remain in their condition, some revising previous texts, others drafting anew. In this way, comparisons could be made between those writers who participated in the reading-as-the-reader condition over two sessions and those under the revise-only or draft-only conditions.

Students' positive responses suggest that reading-as-the-reader gave these writers a perspective on the effects of their writing that they otherwise might not have had. For example, one writer reflected, "I like to read other kids' descriptions because sometimes if I read other kids [sic] descriptions, I can get more ideas ... because when I look back into the pictures I can't see the pictures they see." The analysis of the writers' comments from the read-as-the-reader group suggests that actually doing the task their readers did revealed to them the necessary information they needed to include and the unnecessary information they needed to exclude in their descriptions. Further research might address whether younger students in the 2nd, 3rd, and 4th grades also receive the social cognitive boost that the 5th-graders did. Further research in the development of the readers' perspectives might contribute to our understanding of specific pedagogical situations that would assist help younger writers consider the informational needs of their readers.

Educational Applications

Both 5th- and 9th-grade students' reflective comments suggest that reading another student's descriptions and then matching them with target-tangrams was a motivational push and an awareness-boosting process for the students in this study. An adaptation of the writing/reading exercise "reading-as-the-reader" used in this study might be added to a teacher's repertoire of "optimal environmental activities" (Daniles, 1990). Daniles suggests, "Lessons about effective descriptive writing emerge from experiencing strategies in use" (p. 119). Reading-as-the-reader is a perspective-taking strategy experienced when the writer attempts to create a spatial perceptual description their readers can "see." In a similar fashion, reading as the reader is one perspective-taking condition that can be added to a "writer's tool box" (Harper, 1997). Harper describes five revision tools that she suggests have worked for her as a practicing middle school writing teacher. One such tool is the snapshot. Students compose written snapshots, similar to a detailed photographic snapshot. Snapshots are writing activities that compel students to focus on close, physical detail without relying on subjective thoughts and/or feelings. Reading-as-the-reader could be a tool that writing teachers incorporate into their repertoire of classroom activities to help students become more efficient descriptive writers.

Reading-as-the-reader also may help students to make details explicit and assist students in recognizing other text-creating approaches that could be used with other functions of writing. Transactional descriptive writing is "information heavy." The readers' needs are limited to a specified perceptual visual context; literary interpretations and rich imagistic responses from expressive and poetic writing may not be as susceptible to an objective dependent measure. Descriptive writing, however, is one path whereby students can demonstrate their understanding of particular works of art (Stavropoulos, 1997). Written responses to artwork can be assessed on several dimensions: formal (elements of design, media, and technique); interpretive (meaning, emotion, and expression); historical (names, dates, information regarding the artist); and descriptive (objects and subject matter). As students learn the dimensions of art interpretation and are asked to write about their understanding of artwork, a process similar to reading-as-the-reader may be one writing lesson that can contribute to students describing in more detail various works of art.

Clearly, reading-as-the-reader is a strategy not unlike the common classroom practice of peer editing, in which students are asked to evaluate another student's work for spelling errors, grammatical mistakes, and clarity in meaning. Yet, reading-as-the-reader requires the student writer to go one step further by actually envisioning how a reader might more accurately choose the appropriate figure based on the quality of the writer's description. With reading-as-the-reader, writers learn to "see" their descriptions as if they, too, were readers of their own writing.

Conclusion

In Bruner's cultural psychological approach to education, "the perspectival tenet" states that %o understand well what something 'means' requires some awareness of the alternative meanings that can be attached to the matter under scrutiny" (1996, p. 13). Olson (1994) theorized that the development of competency in literacy is associated with understanding different interpretive perspectives. This would include learning to write from the informational and perceptual perspective of readers. Learning to distinguish the "said" (that which is explicit) from the "intended" (that which is implicit) requires a perspective on the reciprocity among reader, writer, and written text that is challenging even for experienced writers and readers. Reading-as-the-reader may help students to make details explicit and assist students in recognizing other text-creating approaches that could be used with other functions of writing. If reading-as-the-reader is a strategy that worked for younger writers in helping them develop a reader's perspective in transactional writing, it also might be a strategy transferable to other writing purposes. Reading-as-the-reader might be one small step in assisting writers not only in inquiring, "What do I want to write?" and 'What have I written?" but also in asking, "How will the reader interpret my writing?" Reading-as-the-reader may help young writers take a small step towards the literacy of perspectivability.

Table 1
Sequence of Weekly Experimental Sessions for Writers and Readers

Weekly Sessions
                        Writers                       Readers
Week 1           Writers are given a            Readers read
                 notebook of three              descriptions (three
                 target-grams to be             descriptions from three
                 described.                     different writers)
                                                and make description to
                                                target-tangram matches.

Week 2           Writers are randomly assigned  Readers read and make
                 to either feedback,            new description-to-
                 rating-other, or               target-tangram matches.
                 read-as-the-reader condition.
                 After task is performed,
                 all writers revise original
                 descriptions.

Week 3           Students remain in previous    Readers read new
                 perspective-taking condition,  descriptions and make
                 perform task, and then         target-tangram matches.
                 describe a new group of three
                 target-tangrams.

Table 2
Means and Standard Deviations by Grade, Session, and Condition

                Session 1
              N       M       SD
Condition
Feedback
9th-Grade    18      2.16    .923
5th-Grade    25      1.80    .763

Rate-Other
9th-Grade    26      2.23    .764
5th-Grade    30      1.86    1.04

Read-as-
Reader
9th-Grade    32      1.75    .879
5th-Grade    23      1.57    .589

                Session 2

              N       M        SD
Condition
Feedback
9th-Grade     18     2.38     .849
5th-Grade     25     2.20     .763

Rate-Other
9th-Grade     26     2.42     .757
5th-Grade     30     2.00     .946

Read-as-
Reader
9th-Grade     32     2.28     .851
5th-Grade     23     2.13     1.01

                Session 3

              N      M        SD
Condition
Feedback
9th-Grade     18     2.27    .958
5th-Grade     25     1.68    .945

Rate-Other
9th-Grade     26     2.42    .702
5th-Grade     30     2.27    .868

Read-as-
Reader
9th-Grade     32     2.50    .672
5th-Grade     23     2.26    .688

Table 3
Examples of Students' Positive Reflections on Reading-as-the-Reader

5th-graders

This exercise was challenging and fun. But sometimes it was
frustrating. I can now successfully describe tangrams.

This writing activity was a good experience for me because I got to
see how other people than us writes [sic] and think. It helped me
when I was writing my second set of tangrams because what they wrote
gave me ideas. The feedback was also helpful because I knew when I
had to change stuff to make things clearer.

The part I liked most was reading other peoples [sic] descriptions.
It made me feel better about my writing when I read and thought "what
the beck are they talking about?" I felt I wasn't the worst tangram
describer afterall [sic].

My favorite was choosing which one other people were describing. It
was also fun getting feedback and writing our won [sic]. It was more
fun than boring. I learned that it is easier to write what it looks
like instead of what it has on it, and it also works better.

9th-graders

Though tangrams do not contribute to my knowledge, it is a good way
to find out if the directions we use daily are clear or not.

I think that even though it was a nag to rewrite the same tangram
over again, it helped me to condense my descriptions and the reader
still guessed it.

I liked it because it makes us think not only about what we are seeing
but about what your readers will see or might not see.

I think that describing the tangrams got easier as you went along. At
first you didn't know how much each tangram look liked the others. When
I matched descriptions with the pictures it made it easier to describe
the tangrams because I knew what was helpful to the reader mad
what was not.

I think that this exercise was interesting but kinda [sic] hard. It was
hard to describe a figure in your own words and your own perspective to
try to get someone else to understand what you are trying to say and
describe. I learned that descripfulness [sic] helps.

References

Beal, C. (1996). The role of comprehension monitoring in children's revision. Educational Psychology Review, 8(3), 219-238.

Bereiter, C., & Scardamalia, M. (1987). The psychology of written composition. Hillsdale, NJ: Lawrence Erlbaum Associates.

Berkenkotter, C. (1981). Understanding a writer's awareness of audience. College Composition and Communication, 32, 388-399.

Bonitatibus, G. J., & Flavell, J. (1985). Effect of presenting a message in written form on young children's ability to evaluate its communicative adequacy. Developmental Psychology, 21(3), 455-461.

Bonk, C. (1990). A synthesis of social cognition and writing research. Written Communication, 7(1), 136-163.

Bracewell, R., Scardamalia, M., & `, C. (1978). The development of audience awareness in writing. (ERIC Document Reproduction Service No. ED 154 433)

Britton, J. (1982). Spectator role and the beginning of writing. In M. Nystrand (Ed.), What writers know: The language, process, and structure of written discourse (pp. 149-169). New York: Academic Press

Britton, J., Burgess, T., Martin, N., McLeod, A., & Rosen, H. (1975). The development of writing abilities. London: MacMillan Education.

Brossell, G. (1985). Rhetorical specification in essay examination topics. College English, 45, 165-173.

Bruner, J. (1996). The culture of education. Cambridge, MA: Harvard University Press.

Cameron, C., Hunt, K. A., & Linton, M. (1996). Written expression as recontextualization: Children write in social time. Educational Psychology Review, 8(2), 125-150.

Chandler, M. (1977). Social cognition: A select review in current research. In W. Overton & J. McCarthy Gallagher (Eds.), Knowledge and development (pp. 93-148). New York: Plenum Press.

Charney, D. (1989). Fictional scenarios and rhetorical specification in writing tasks: A cautionary note. Journal of Advanced Composition, 9(1-2), 151-161.

Clark, H. H., & Wilkes-Gibbs, D. (1986). Referring as a collaborative process. Cognition, 22, 1-39.

Cox, M. V. (1978). Order of the acquisition of perspective-taking skills. Developmental Psychology, 14(4), 421-1422.

Daniles, H. (1990). Young writers and readers read out: Developing a sense of audience. In Timothy Shanahan (Ed.), Reading and writing together: New perspectives for the classroom. Norwood, MA: Christopher-Gordon Publishers.

Demorest, A., Silberstein, L., Gardner, H., Winner, E. (1983). Telling it as it isn't: Children's understanding of figurative language. British Journal of Developmental Psychology, 1, 121-134.

Dyson, A., & Freedman, S. (1991). Writing. In J. Flood, J. M. Jensen, D. Lapp, and J. R. Squire (Eds.), Handbook on teaching the English language arts (pp. 754-774). New York: Macmillan.

Ede, L. (1984). Audience: An introduction to research. College Composition and Communication, 35(2), 140-154.

Elbow, P. (1987). Closing my eyes as I speak: An argument for ignoring audience. College English, 49(1), 51-69.

English, L. (Ed.). (1997). Mathematical reasoning: Analogies, metaphors, and images. Mahwah, NJ: Lawrence Erlbaum.

Fitzgerald, J., & Markham, L. (1987). Teaching children about revision in writing. Cognition and Instruction, 4(1), 3-24.

Fitzgerald, J. (1987). Research on revision in writing. Review of Educational Research, 57(4), 481506.

Fitzgerald, J. (1992). Variant views about good thinking during composing: Focus on revision. In M. Pressley, K. Harris, & J. Guthrie (Eds.), Promoting academic competence and literacy in school (pp. 337-360). San Diego, CA: Academic Press.

Flavell, J. (1992). Perspectives on perspective taking. In H. Belin & P. B. Pufall (Eds.), Piaget's theory: Prospects and possibilities (pp. 107-139). Hillsdale, NJ: Lawrence Erlbaum Associates.

Florio, S. (1979). The problem of dead letters: Perspective on the teaching of writing. The Elementary School Journal 80(1), 1-7.

Flower, L. (1994). The construction of negotiated meaning: A social cognitive theory of writing. Carbondale, IL: Southern Illinois University Press.

Flower, L,. & Hayes, J. R. (1984). Images, plans, and prose: The representation of meaning in writing. Written Communication, 1(1), 120-160.

Flower, L., Hayes, J. R., Carey, L., Schriver, K.A., & Stratman, J. (1986). Detection, diagnosis, and the strategies of revision. College Composition and Communication, 37(1), 16-55.

Frank, L. (1992). Writing to be read: Young writers' ability to demonstrate audience awareness when evaluated by their readers. Research in the Teaching of English, 26, 277-298.

Glucksberg, S., & Krauss, R. (1967). What do people say after they have learned how to talk? Studies of the development of referential communication. Merill-Palmer Quarterly, 13, 309-316.

Harper, L. (1997). The writer's toolbox: Five tools for active revision instruction. Language Arts, 74, 193-200.

Holliway, D., & McCutchen, D. (2004). Audience perspective in children's descriptive writing: Reading as the reader. In L. Allal, L. Chanquoy, & P. Largy (Eds.), Revision of written language: Cognitive and instructional processes. (pp. 87-101). Amsterdam: Kluwer Academic Press.

Kirsch, G., & Roen, D. (Eds.). (1990). A sense of audience in written communication. Newbury, CA: Sage Publications.

Kohlberg, L. (1984). Essays on moral development. San Francisco: Harper and Row.

Kroll, B. (1978). Cognitive egocentrism and the problem of audience awareness in written discourse. Research in the Teaching of English, 12(3), 269-281.

Kroll, B. (1984). Writing for readers. Three perspectives on audience. College Composition and Communication, 35(2), 172-185.

Kroll, B. (1985). Rewriting a complex story for a young reader: The development of audience-adapted writing skills. Research in the Teaching of English, 19(2), 120-139.

Kroll, B., & Lempers, J. D. (1981). Effects of mode of communication on the informational adequacy of children's explanations. The Journal of Genetic Psychology, 138, 27-35.

Matsuhashi, A., & Gordon, E. (1985). Revision, addition, and the power of the unseen text. In S. Freedman (Ed.), The acquisition of written language: Response and revision (pp. 226-249). Norwood, NJ: Ablex Publishing.

Miller, P., Kessel, F., & Flavell, J. (1970). Thinking about people thinking about people thinking about ... : A study of social cognitive development. Child Development, 41, 613-623.

O'Keefe, B., & Delia, J. (1986). Communicative tasks and communicative practices: The development of audience centered message production. In B. Rafoth & D. Rubin (Eds.), The social construction of written communication (pp. 70-98). Norwood, N J: Ablex Publishing.

Oliver, E. (1995). The writing quality of seventh, ninth, and eleventh graders, and college freshman: Does rhetorical specification in writing prompts make a difference? Research in the Teaching of English, 29(4), 422-450.

Olson, D. (1994). The world on paper. Cambridge, MA: Cambridge University Press.

Park, D. (1986). Analyzing audiences. College Composition and Communications, 37(4), 478-488.

Rafoth, B. (1985).Audience adaptation in the essays of proficient and nonproficient writers. Research in the Teaching of English, 19(3), 237-253.

Rafoth, B. (1989). Audience and information. Research in the Teaching of English, 23(3), 273-290.

Roen, D., & Willey, R.J. (1988). The effects of audience awareness on drafting and revising. Research in the Teaching of English, 22(1), 75-88.

Rubin, D. (1984). Social cognition and written communication. Written Communication, 1(2), 211-245.

Schober, M. F., & Clark, H. H. (1989). Understanding by addressees and overhears. Cognitive Psychology, 21, 211-232.

Selman, R., & Byrne, D. (1974). A structural developmental analysis of levels of role-taking in middle school children. Child Development, 45, 803-806.

Stavropoulos, C. S. (1997). Alternative methodology for diagnostic assessment of written and verbal responses to works of art. In J. Flood, S. H. Heath, & D. Lapp (Eds.), Handbook of research on teaching literacy through the communicative and visual arts (pp. 239-263). New York: Simon and Schuster Macmillan.

Stevens, W. (1954). Collected poems. New York: Knopf.

Taylor, H. A., & Tversky, B. (1996). Perspective in spatial descriptions. The Journal of Memory and Language, 35, 371-391.

Traxler, M., & Gernsbacher, M. (1992). Improving written communication through minimal feedback. Language and Cognitive Process, 7, 1-22.

Traxler, M., & Gerrnsbacher, M. (1993). Improving written communication through perspective-taking. Language and Cognitive Process, 8(3), 311-334.

Witte, S. (1992). Context, text, intertext: Toward a constructionist semiotic of writing. Written Communication, 9(2), 237-308.

David R. Holliway

Marshall University