Document View

The purpose of this collaborative schools-university study was to investigate how the explicit instruction of literary devices during designated literacy sessions could improve the quality of children's narrative writing. A guiding question for the study was: Can children's writing can be enhanced by teachers drawing attention to the literary devices used by professional writers or "mentor authors"? The study was conducted with 18 teachers, working as research partners in nine elementary schools over one school year. The research group explored ways of developing children as reflective authors, able to draft and redraft writing in response to peer and teacher feedback. Daily literacy sessions were complemented by weekly writing workshops where students engaged in authorial activity and experienced writers' perspectives and readers' demands (Harwayne, 1992; May, 2004). Methods for data collection included video recording of peer-peer and teacher-led group discussions and audio recording of teacher-child conferences. Samples of children's narrative writing were collected and a comparison was made between the quality of their independent writing at the beginning and end of the research period. The research group documented the importance of peer-peer and teacher-student discourse in the development of children's metalanguage and awareness of audience. The study suggests that reading, discussing, and evaluating mentor texts can have a positive impact on the quality of children's independent writing. [PUBLICATION ABSTRACT]

Current government test results in the United Kingdom show a continuing gap between achievement in reading and writing. Despite some marginal development, writing continues to be an issue, with the Office for Standards in Education (OFSTED, 2005) reporting that too many children are still leaving elementary school (age 11) below the expected level. The call for teachers to move beyond initial stimulus for writing to more explicit teaching and skillful intervention during composition has been a consistent message from research findings and inspection reports for some years. In its evaluation of the second year of the National Literacy Strategy, OFSTED (2000) confirmed that "improving standards in writing had proved to be challenging" and that "too much time is spent on children's practicing writing rather than being taught how to improve it" (p. 9). Myhill (2001) reiterated the need for more explicit teaching, saying that teachers rarely intervene, during composition, to help children understand how to improve their writing. Thus, despite the implementation of a National Literacy Strategy, a systematic approach to teaching writing, including opportunities for sustained independent writing, is not common practice in most British elementary schools.

The purpose of the study was to investigate whether, through explicit teaching and discussions of mentor texts in literacy sessions, children could develop their knowledge of how texts are crafted by accomplished authors. We then wanted to see whether children would use this knowledge during writing workshops to improve the quality of their own independent writing. Underpinning our classroom research was Vygotsky's (1978) contention that learning occurs within a social context, and through adult guidance or in collaboration with more capable peers. We therefore adopted a social constructivist position; our teaching approach was influenced by Bruner's contention that learning takes place most effectively by providing appropriate social interactional frameworks and by scaffolding through structured interplay between teachers and children. The call in the UK for more structured intervention to scaffold children's learning echoes those in the United States (De La Paz, 1997; Leavell & Loannides, 1993). Research in the United States has shown how the quality of children's writing can be improved through the explicit teaching of self-regulatory strategies and processes used by skilled writers (Anderson, 1995; Graham, Harris, & Troia, 1998; Tierney & Shanahan, 1991). This finding is supported by studies in the UK that indicate that although children's writing is often characterized by poor style and weak structure, it can be improved by systematically deploying various graphic aids to thinking and structuring ideas (Frater, 2001; Lewis, 1999).

The research group's approach to teaching writing was shaped not only by its social constructivist theoretical perspective, but also by substantial research that illustrates how children's writing can be enhanced by teachers modeling, demonstrating, and using mentor texts to highlight craft and artistry in writing (Cullinan, 1987; Harwayne, 1992; Kress, 1986; Short & Pierce, 1990). As research in the United States has shown, children's literature can provide a model of quality writing that, along with explicit teacher instruction and peer discussion, allows personal response and encourages children to transfer ideas to their own writing. This transfer from reading to writing was exemplified by Lancia (1997), who showed how children "spontaneously borrow ideas temporarily as they develop their own writing craft" (p. 471). Other researchers argue that through reading and listening to stories, children learn about the language features and organizational aspects of written texts (Perera, 1984; Rosemary & Roskos, 2002).

An aim of the study was to see whether children could move beyond merely copying ideas to developing a conscious awareness of what structural or stylistic choices they were making and why. We wanted to see if children could transfer knowledge of literary devices, gained from reading and discussing mentor texts, to their independent writing. Various studies have confirmed the need for children to develop this metacognitive understanding of their own writing processes (Dyson, 1986; Flower, 1994; Flower & Hayes, 1981; Rosenblatt, 1995; Wray, 1994). Bereiter and Scardamalia (1987) differentiate between knowledge telling and knowledge transformation. In knowledge telling, any information is retrieved from memory and written down, with each new phrase or sentence stimulating the generation of the next idea. Little attention is directed at the needs of the audience or the language and organizational constraints of the genre. The approach is mostly forward-moving, with little recursive interplay between composing processes, and the role of planning, revising, and other self-regulation processes is minimized. Knowledge transformation refers to a process of writing, whereby drafting and revision take place and the writer is constantly reflecting on the content, coherence, form, and style. This transformation manifests itself in children's ability to reflect on their structural and stylistic choices in a piece of writing. Bereiter and Scardamalia (1987) claim that the process of revising text and in rethinking rhetorical choices "provides a strong indication of increasing compositional maturity" (p. 266).

Writing Workshops and Our Concept of Process Writing

A major organizational component of the study was the weekly writing workshop, in which children could draft and redraft work over a sustained period of time. However, a problem facing the research group was that national tests encourage the teaching of test techniques and undervalue the very reflective process of writing we wanted to develop. After comparing teaching practices in several countries, Purves (1992) found that a product approach to teaching writing was particularly prevalent in the UK and the United States. Similarly, Harris and Graham (1996) found that although some classrooms in the United States were making changes, few activities required sustained writing. This trend is echoed in many British schools; some educators believe this problem has been exacerbated by the introduction of a product-focused National Literacy Strategy, which discourages recursive writing (Hilton, 2001). It was, therefore, important for us to clarify our understanding of what we meant by process writing and what procedures we wanted children to go through during composition. In the UK, the notion of process writing has largely been based on the work of Calkins (1986) and Graves (1983), with its emphasis on peer interaction. The research group was not entirely comfortable with this conception, because it seemed to marginalize teachers, rather than viewing children and teachers as partners in the joint construction of knowledge. Our recursive model of writing-involving discussion, planning, organizing, translating, reviewing, and redrafting-was, therefore, influenced by the composition research of Bereiter and Scardamalia (1987), Flower and Hayes (1981), and Hayes (1996). Our belief in the importance of scaffolding through teacher-student dialogue was reinforced by other work conducted in the UK and the United States (Maybin, Mercer, & Stierer, 1992; Nystrand, Greene, & Wiemelt, 1993; Smith & Elley, 1999). Moreover, the value of collaborative peer reading, writing, and discussion has also been well documented (Raphael & McMahon, 1994; Wells, 1989; Whittaker & Spencer, 1991), while several researchers, including Graves and Hansen (1983), have demonstrated the impact of collaborative learning experiences upon children's sense of audience.

Developing a Sense of Audience

There is a corpus of research dealing with children's awareness of audience when writing and whether they actively engage in constructing representations to their readers. A difficulty, as Kroll (1985) points out, is in knowing how sensitivity to audience manifests itself and whether it is apt to manifest itself in different ways for different genres. In an attempt to measure audience awareness, some studies have looked at syntactic complexity, rhetorical structuring, and other stylistic features. Kroll explored the audience-adapted writing skills of children in grades 5, 7, 9, and 11. Analyses of revisions to their texts suggested that younger children tended to focus on word-level vocabulary changes, whereas older children focused on more major structural and stylistic changes. A major question for the research group was whether an awareness of audience could be enhanced in elementary school children who did not appear to move easily between reader-writer positions.

The teachers believed that one way to help children develop a sense of audience, to read like a writer and write like a reader, was through collaborative interchange. Therefore, peer-peer and teacher-led discussions were a major feature of our work in schools. As Rosenblatt (1989) argues, "Such transactions can help students develop metalinguistic insights in a highly personal and hence instructive way" (p. 173). Teachers in the research group aimed to help children develop their sense of audience and transfer ideas, gained from reading and discussing mentor texts, to their own writing. Consequently, the teachers' approach combined explicit instruction and discussion of mentor texts during literacy sessions, along with opportunities for children to work independently on compositions in writing workshops, where they could gain critical but constructive feedback during the drafting and redrafting process.

Methodology

The study involved 18 teachers who worked as research partners in nine elementary schools and collectively covered the age range of 7-11 years. The schools were chosen to represent a variety of socio-economic and cultural contexts and ranged from inner-city to semi-rural locations. The teachers were all experienced practitioners and well-established within their respective schools. The teachers attended research preparatory and developmental meetings and met on a regular basis as a group to discuss progress and evaluate data. Teachers' professional development in the UK and the United States has increasingly been developed through some form of partnership (Furlong, Barton, Miles, Whiting, & Witty, 2000; Wagner, 1997). However, educational researchers often face the dilemma of adopting a distant research approach with an emphasis on objectivity and reliability, or a grassroots position with a concern for validity and practical application. An initial methodological concern was whether to occupy what Cordingley (1999) describes as the high ground, or be immersed in what she terms "the swamp." However, our underlying premise was that through collectively analyzing individual experiences and interpretations, we might gain useful insights, which could be shared with colleagues. In searching for an appropriate partnership research model, we were influenced by the collaborative inquiry process of Buchanan and Schultz (1993) and by the work of Saez and Carretero (1996), who synthesized a number of case studies and developed descriptive narrative events into a collective analysis. Other researchers also have undertaken collaborative research projects, which involved groups of school practitioners working alongside external researchers (Aspland, Macpherson, Proudfoot, & Whitmore, 1996; Cochran-Smith & Lytle, 1999). We, therefore, adopted the cooperative problem-solving approach defined by King and Lonnquist (1996) as collaborative action research. We recognized that "fuzzy" (rather than firm) generalizations would be drawn from our inductive inquiry. However, we felt that our research findings would have pragmatic validity (Miles & Huberman, 1994) and we hoped that patterns and principles could be identified, from which tentative hypotheses might be formulated to stimulate further deductive research.

Literacy Sessions and Writing Workshops

Literacy Sessions

The UK National Literacy Strategy (Department for Education and Employment [DfEE], 1998) requires schools to designate one hour a day to literacy sessions. During these designated sessions, the research teachers used mentor texts (children's literature, written by professional adult authors) as models of particular narrative forms and writing styles. As the focus of the study was on narrative writing, teachers chose mentor texts that were appropriate for the age phase they were teaching and as models of specific narrative types (e.g., folktale, fantasy, mystery, time-slip). Teachers began the daily literacy sessions by reading a mentor text aloud to the whole class and drawing children's attention to specific structural or stylistic features. Teachers used an overhead projector or electronic whiteboard for shared writing. This procedure involved teachers and children working collaboratively to compose sentences or paragraphs using literary techniques identified in the mentor text. After the whole-class period, children worked in small groups to investigate mentor texts further. During this period, teachers would work with one focus or teacher-led group. The class was divided into five groups, so that over one week teachers would spend time with each group. Findings from group investigations were then shared in a whole-class plenary. Literacy sessions were concluded by a period of independent work, when children transferred ideas and useful literary devices into their author notebooks. The aim of these lessons was for children to take both aesthetic and efferent stances (Rosentblatt, 1995), discussing how authors had crafted texts to evoke emotional responses from the reader. For example, a teacher might use the mentor text Fair's Fair by Leon Garfield (2001) to illustrate a story that is written in the oral tradition and also contains structural and linguistic features of a fairy tale. During one lesson the teacher might focus children's attention on aspects of plot such as opening, setting, complications, climax, and resolution. The focus for another session might be the literary devices used by Garfield, such as word choice, sentence variety, simile, metaphor, or personification.

Writing Workshops

In addition to using designated literacy sessions, the teachers organized a weekly one-hour writing workshop where children worked on sustained pieces of writing, in the context of what Cairney and Langbien (1989) and Schultz (1997) call a writing community. Frater (2001) stresses how successful schools "ensured that extended time for extended writing was frequently and regularly available" (p. 13). The research group believed that writing workshops were essential because they allowed children to experience writers' perspectives and readers' demands (Harwayne 1992; May, 2004). During writing workshops, children worked on independent compositions and could gain continual feedback from teachers and response partners (their critical peers).

Data Collection

Collection of data occurred through audio recording of teacher-child conferences and video recording of peer-peer and teacherled group discussions. Each teacher maintained a research portfolio that contained the following sections.

Record of Peer-Peer and Teacher-led Group Discussions

Group discussions of mentor texts were video-recorded and transcribed. They were then examined to see if children were able to recognize, discuss, and evaluate authors' use of literary devices. Discussions were also examined for evidence of children's developing metalanguage.

Record of Teacher-Pupil Conferences

Children's ability to discuss their work and justify the use of structural and stylistic choices was a crucial factor, since this distinguished between Bereiter and Scardamalia's (1987) notion of knowledge telling and knowledge transformation. Conferences were audio-recorded and transcribed. This allowed teachers to gain evidence of when children showed an awareness of audience and had made conscious and deliberate choices during composition.

Sample of Children's Written Work

Criterion-based sampling was used (Merriam, 1988), and each of the 18 teachers selected six case study children from their classes. Samples of independent writing were collected from two low, two average, and two high attaining children in each class at the beginning and end of the research period. Teachers selected case study children on the basis of their attainments on Qualifications and Curriculum Authority (QCA) standard assessment tasks and according to UK national criteria relating to expectations of attainment for each respective age phase. Low scorers were those whose level of attainment fell below the national expectation for their age phase, average scorers were those who met the national target, and high scorers were those who exceeded the national target. Because of logistical difficulties, such as illness and children leaving, some teachers were unable to collect samples from all their case study children. However, a total of 192 writing samples from 96 case study children (45 girls and 51 boys) were assessed.

Data Analysis

Evaluation of Children's Written Work

To evaluate children's work, we developed a framework for analyzing narrative writing (Figure 1). In devising this we used QCA (2001) categories of structure, style, and process but were also influenced by the work of Applebee (1978) and that of Paramour and Wilkinson (1985), who identified developmental criteria for narrative writing.

At the end of the research period, data from all schools were collated and collectively analyzed by the principal researcher. To help negate marker bias in favor of later work, the samples were mixed before scoring. The structural and stylistic features of each sample were equated to the framework by using a colored marker pen. This provided a visual profile for each child and, in accordance with UK assessment procedures (DfEE, 1999), a "best-fit" descriptor level was selected. As children in the study were given relative freedom to write stories of their own choice, we found this best-fit model preferable to the point scoring system used in national examination tests, in which children are given a very specific task and story title. Using the best-fit evaluations, a comparison was made between children's independent writing at the beginning and end of the research period (see samples 1-4 and commentaries).

Results

Development in the Quality of Children's Writing

Children's first writing samples showed features of levels 1 and 2 on the framework for analysis. Children made significant progress in the assessment categories of structure and style, with first samples showing features of levels 1 and 2 and final samples showing features of levels 2 and 3. This is illustrated in Figures 2, 3, and 4.

The national expectation in the UK is that primary children will progress at a rate of one level every two years. Of the 96 case study children, 77 advanced one level and 19 children advanced two levels over a one-year period. This rate of development significantly exceeds the national expectation for normal progress. Writing samples 1 and 2 (below) illustrate the progress made by one child over the research period. The samples have been selected from the data to represent children's work that showed one level of development on our framework for analysis. It is presented to illustrate typical rather than exemplary progress over the research period. As the focus of the research was on structural and stylistic features, any inaccurate spellings have been corrected. The work is evaluated in relation to our conceptual framework, and the commentaries that follow each sample identify those linguistic features that characterize a particular level of development.

Sample 1. Matt: Age 8. Written in September: start of semester one (assessed as level 1).

The naughty witch was a school sweep. In the night the witch went bad She did a spell. She saw a rabbit She cooked the rabbit She ate the rabbit. Yummy she said the witch. She went and she made a ring She had her ring She went back to school No one knew she was a witch.

Matt had not read and discussed specific mentor texts or experienced explicit instruction of literary devices before writing his story. However, he had read some of Jill Murphy's Worst Witch stories, which may account for his story content. There is no recognizable narrative opening or setting to Matt's story. It follows a linear pattern and is a simple chain of loosely connected events. There is no character development and no clear motivation for actions. Simple vocabulary is used and there are few adjectives or adverbs. There is no variation in the SVO sentence pattern (e.g., She saw a rabbit. She cooked the rabbit. She ate the rabbit.). Knowledge transformation (Bereiter & Scardamalia, 1987) is evident when children consider the needs of their readers and respond, review, and revise. Matt does not move between reader-writer positions, posing questions, and providing answers to obvious questions, such as: Why did the witch become bad in the night? What kind of spell did she cast? Why did she make a ring? What was special about the ring? When Matt's teacher asked him if he thought other children would find his story exciting, he simply shrugged.

Sample 2. Matt: Age 8. Written in July: end of semester two (assessed as level 2).

Sam stared at the misty sky. He stared at the gray carpet. He could hear footsteps creaking softly in the bedroom. Grandma entered. She could see Sam was miserable but she told him he would enjoy his school trip. Sam did love it. He loved climbing trees, making dens, pond dipping and doing lots of things in the Forest. Once, deep in the forest, he heard a snuffle so he turned and there was a fox. She had lovely shiny fur and enormous eyes. He froze because he didn't want to scare her away.

By July, Matt's writing was being influenced by the explicit instruction of literary devices. Although his story was not completed in the allocated time, it is well structured. His opening captures the reader's interest and the setting is carefully crafted, using the literary device of pathetic fallacy to create a somber mood (e.g., misty sky, gray carpet). This is a technique Matt has drawn from class and group discussions of various mentor texts, including Storm (2001), in which Kevin Crossley-Holland establishes a narrative mood by describing a marsh that was silent and empty and where the wind whined and seabirds screamed. Matt has also examined the work of Michael Murpurgo and discussed the way he often states, unambiguously, how his characters are feeling (e.g., in Sam's Duck, Murpurgo describes how the main character couldn't bear to watch and couldn't wait to see Grandad). In Matt's story, the characters' feelings are also described explicitly (e.g., She could see Sam was miserable, Sam loved to climb trees). Over the year, Matt's class has explored mentor texts to see how authors choose verbs carefully for effect (e.g., in The Gargoyle by Garry Kilworth (1997), the author describes something stirring and a strange shape that shuddered and clawed itself over a parapet). Matt makes similar judicious choices in his text (e.g., snuffle, creaked, entered, froze). There is an embedded phrase (Once, deep in the forest, he heard a snuffle). Matt's class has studied this technique during literacy sessions and examined how, in South and North, East and West by Michael Rosen (1995), the tales often open with such sentences (e.g., Once upon a time, many years ago, Tau the lion met Mmutla the hare). Punctuation is used to extend sentences (e.g., he loved climbing trees, making dens, pond dipping). Again, this is something the class has explored using a variety of mentor texts, such as Fair's Fair in which Leon Garfield (2001) describes candles in silver candlesticks, pictures in gold frames, china plates on a shining table, roast beef on a sideboard. Finally, there is a variety of connectives, such as and, but, because, and so, demonstrating a progression in cohesion from level one to level two on our conceptual framework.

Samples 3 and 4 have been chosen from the data to represent children's work that showed two levels of development on our framework for analysis.

Sample 3. Joel: Age 9. Written in September: start of semester one (assessed as level 1).

There was once a forest. There was lots of trees in and some animals. A squirrel was looking for nuts he saw some humans with bulldozers then he went to tell the other animals. Owl told the birds to go and find a new home. At last they found one. They all set off to move. Then they came to a road they didn't know how to cross it. Mole and his friends dug a hole then they all went through it. At last they arrived. Mole made a home under ground. Owl perched in a tree and squirrel found a cozy tree.

Joel had not examined specific mentor texts or experienced explicit instruction of literary devices before writing his story. However, he had read several tales from Arnold Lobel's Frog and Toad series, which use a number of simple SVO sentences (e.g., Frog ran up the path, Frog walked into the house, Toad was lying in bed). In his story, Joel appears to be "echoing" this syntactic pattern. His story has a traditional narrative opening, which is followed by a linear chronicle of events in temporal sequence (e.g., There was once, then, at last). There is a logical sequence of events and some motivation for the action, but minimal development of characters. A problem arises and then is resolved, but no suspense is created through complications that lead to a climatic point and final resolution (e.g., Owl told the birds to go and find a new home. At last they found one). Joel uses only two adjectives (new and cozy) and no adverbs. Most of the sentences are simple, although he does use two compound sentences. Basic noun phrases are used (e.g., a squirrel, some animals, some humans, the birds, a road, a hole, a tree).

Sample 4. Joel: Age 9. Written in July: end of semester two (assessed as level 3).

Late one night, the rain was pouring, the wooden gate opened furiously and the black cat purred. A dog was disturbed. Its pure white fur gave a sharp contrast against the sheet of blackness night wears. A mansion stood shadowed against the night sky. The dog moved slowly, silently, slyly towards the inviting warmth of the window. The animal saw flames flickering in the fireplace. The gloomy lead eyes glared thoughtfully at the inviting puppy portrayed on the wall. Some children wearing short jeans embraced the fluffy white puppy in the painting. Suddenly a loud bark came from nowhere, the dog darted across the countryside as fast as a bullet through the trees, branches crunched under its feet. Then, suddenly, the storm stopped. Birds could be heard singing in the trees, the sun broke through and the air became still. The dog ran through the wood and into the church, it jumped onto the grave. Then it disappeared with its master his friend in life ... and his friend in DEATH!

By July, the explicit instruction of literary devices was evident in Joel's work. Over the year, the class has examined a variety of mentor texts and investigated different kinds of narrative structures. In contrast to the linear pattern of sample one, Joel's second writing sample has a steady build-up of tension, leading to a climax and resolution. One of his favorite mentor texts was One Stormy Night by Ruth Brown (1992) and he draws on some of her techniques when composing his own story (e.g., Brown writes, Just before dawn, the wind fell silent, a bright star shone and the sky was clear). Joel also hooks and sustains his reader's attention through the use of embedded phrases and clauses, which give rhythm to the text and also act as hooks to delay the action (Late one windy night, the rain was pouring, the wooden gate opened furiously and the black cat purred). Throughout the year, the class examined mentor texts to see how authors use figurative language to create imagery. As an example, the concept of personification was studied using Beware the Killer Coat by Susan Gates (1998), who describes a coat as scowling with its rows of metal teeth. In his story, Joel uses language effectively to create imagery (Its pure white fur gave a sharp contrast against the sheet of blackness night wears. The dog moved, slowly, silently, slyly. The animal saw flames flickering in the fireplace. The sun broke through. The air became still.). Adjectives and adverbs are used to create interest (The wooden gate opened furiously. Gloomy lead eyes glared thoughtfully at the inviting puppy portrayed....). Sentences are varied through the use of adverbial openings (e.g., Late one windy night.). Punctuation is used skillfully to construct complex sentences by replacing connectives with commas (e.g., Suddenly a loud bark came from nowhere, [and] the dog darted across the country side as fast as a bullet through the trees, [while] branches crunched under its feet). Joel has examined this technique during class and group discussions. For example, in the mentor text The Butterfly Lion (1996), Michael Murpurgo writes, They were waiting on the veranda, his mother in her nightgown, his father in his hat, his horse saddled.

The literary devices that Matt and Joel have used to enrich and enliven their work in the second samples have been taught during literacy sessions when, over the year, the children have examined a variety of mentor texts and discussed how authors have manipulated language for effect. Matt and Joel have integrated these devices into their own linguistic repertoires, which they are then able to draw from effectively when composing their own stories. In the first samples, Matt and Joel copy familiar text structures and echo author styles, but there is no real sense of transformation or personal ownership. Evidence that transformation, rather than simple echoing, was occurring became clear during teacher-student conferences, when the children justified linguistic choices they had made. For example, when discussing the opening of his story, Matt explains why he has used terms like misty sky and gray carpet by saying:

I want to make people feel sad like Sam is feeling . . . like in Storm it makes you feel all cold . . . a bit scary . . . on your own.

When asked about the sentence The dog moved slowly, silently, slyly, Joel says:

I've used verbs together here . . . a verb list with commas instead of saying "and" "and" "and," 'cos then it makes it sound creepier when you read it.

When discussing the sentence Suddenly a loud bark came from nowhere, the dog darted across the countryside as fast as a bullet through the trees, branches crunched under its feet, Joel says:

When they read I want them to go [gestures with hands and looks fearful] so . . . suddenly . . . then a loud bark came and I want them to think where did it come from and then . . . the dog darted. I didn't want to just put "ran" so I put "dashed" then I put "darted" instead and then I put "like a bullet" . . . "as fast as a bullet." I've used a simile so they can tell the dog's like . . . screaming dead fast across the grass . . . hey! I could've put "screamed" instead of "darted."

Development in the Writing Process

From their classroom observations, teachers noted that at the beginning of the research period (beginning of semester one), the children appeared to have a poor awareness of audience. They seemed insensitive to the needs of the reader and unable to move between reader and writer positions as they composed. However, from their observations of children working on the final sample (end of semester two), the teachers detected a significant development in the way that children were able to critically evaluate and redraft their work (Figure 5).

For example, Joel's final piece (Sample 4) was produced after a process of planning, drafting, and redrafting, in which he responded constructively to feedback from his response partner. Joel's first draft is shown below.

Sample 5. Joel: Age 9. First draft.

One night the rain was pouring down. The cat and a dog with white fur were sleeping. The dog thought it heard something and it woke. It saw a light in a window of the house and it went to investigate. It looked in through the window and saw a painting. In the painting there was some children stroking a puppy and it was white as well. Suddenly there was a loud barking noise and the dog ran off as fast as he could across the fields. Then all of a sudden the storm stopped. The sun came out and the birds in the trees started singing but the dog ran into the churchyard and when it got there it jumped onto its master's grave and disappeared! IT WAS A GHOST DOG!

Joel's discussion with his response partner:

Gus: I like this . . . how you've started the story with the rain and the cat and dog asleep. It gives a good image . . . good pathetic fallacy.

Joel: Yeah, thanks, I wanted to do that. How does the opening sound though? Read it aloud so's I can hear.

Gus: (reads Joel's text aloud) It's a bit like and then and then and then.

Joel: Yeah, that's OK sometimes, but I don't want my opening to be like that; it sounds boring.

Gus: Well, why don't you join up some of your sentences then?

Joel: How do you mean . . . you mean like with some connectives?

Gus: Mmm yeah, you could, but . . . I know . . . I know . . . like in One Stormy Night . . . like one night de-dah, dedah, de-dah.

Joel: Oh yeah so . . . one night.

Gus: Was it in the middle of the night . . . why don't you start in the middle of the night.

Joel: In the middle of the night . . . no . . . late one night . . . late one night.

Gus: Late one night when the rain was pouring and the cat and dog were asleep.

Joel: Erm, hang on (Joel now consults his author notebook). Got it . . . in One Stormy Night, it goes one stormy night, the wind was howling, the iron gate creaked and the black cat hissed.

Gus: Yeah, that's good; that is, you could start . . . late one night.

Joel: The rain was pouring.

Gus: And the black cat hissed . . . or snored . . . do cats snore?

Joel: No, not snored . . . something wakes it up . . . a gate opened a gate opened with a bang . . . no no . . . personification . . . personification. I want to personificate it . . . the gate . . . the old wooden gate opened madly.

Gus: Violently.

Joel: It was mad because of the wind blowing it was furious.

Gus: Yeah, furious . . . the old wooden gate opened furiously and the black cat . . . they don't snore they purr, don't they . . . cats purr?

In this short exchange, Gus is helping Joel to modify his text by providing a reader's voice, urging him to reflect on linguistic choices and to consider alternatives. The breakthrough comes when Gus remembers a literacy session when the class had discussed the mentor text One Stormy Night by Ruth Brown. During the literacy session, Joel had made a note of the syntactic structure Brown uses to open her story, and he now adopts it to improve his own text and give it the kind of rhythm he wants his reader to hear.

Peer Group Discussion

The discussion between Joel and Gus illustrates their metalinguistic awareness. Our data confirms findings from research in the United States and the UK that illustrate how children also develop their metalanguage through small-group discussion of mentor texts (Bershon, 1992; Cooper-Hansen, 2004; Corden, 2000; Thompson, 1988). The following transcript, taken from a video recording of a peer group discussion, illustrates this process:

Year 4: Children ages 8-9 discussing the opening chapter of The Iron Man by Ted Hughes (1968).

Joe: (reading from the text) From crash to crash.

Laura: He's repeated the words . . . rock to rock snag to snag.

Joe: Why has he done that?

Alex: It's because he goes from rock to rock and he bashes on that rock, then on that rock, then he bashes on the other rock.

George: He's trying to get the rhythm of him rolling.

Gemma: Trying to show him tumbling down.

Alex: He's trying to get the rhythm of him falling down the rocks, isn't he?

George: Yes, he's trying to get the rhythm of the rocks . . . the rhythm of him tumbling down.

All: Yeah . . . crash! crash! crash!

Laura: He's put silence as a minor sentence.

Joe: It's like ... erm ... slows you down and then.

Gemma: Slows the reader down.

Emma: Yeah.

George: It's going really fast and then it really slows it down.

Joe: And as it crashed and crashed and crashed.

Laura: It repeats itself.

Joe: What's it called ... what's it called ...rep ... ?

All: Repetition.

Joe: (reads from the text) His iron leg fell off. His iron arms broke off and the hands broke off the arms. His great iron ears fell off and his eyes fell out. His great iron head fell off.

Joe: It's all separate.

Emma: His, his, his, his, his, his.

Joe: Oh yeah.

Alex: 'Cos it's like all his body falling off.

Joe: So why ... ?

Emma: 'Cos it's all his own . . . like it doesn't belong to anyone else, it all belongs to him.

Alex: So when he crashes and crashes and crashes he all falls apart.

Joe: CRASH! CRASH! CRASH! He's using graphology ... and an exclamation mark.

The UK National Literacy Strategy (DfEE, 1998) states that as children progress through the elementary phase, they should "become increasingly conscious of the writer's intentions" (p. 7). The children in this discussion are demonstrating their ability to evaluate an author's use of language. Their knowledge of linguistic terms (e.g., minor sentence, repetition, graphology) and an awareness of what effects the author is trying to achieve is evident.

Teacher-led Group Discussion

Research has shown the value of teachers working with children to scaffold their learning through effective discourse that focuses attention on specific features of texts (Webb & Palinscar, 1996). This process is illustrated in the following transcript, in which the teacher now joins the group to discuss its findings:

Teacher: Did you spot lots of techniques?

Alex: Lots.

Teacher: Lots, Alex . . . OK . . . and when you spotted the techniques, were you able to say why Ted Hughes had actually used lots of those techniques?

Alex: Some of them.

Teacher: Some of them ... OK... so which ones in there did you spot ... techniques he was using but you weren't sure why he was using it ... Joe?

Joe: Mmm...on the top of page 12 on the second line ... lifted up his foot... dash.

Teacher: Right... so ... his right foot-his enormous right iron foot lifted and then it's got that dash. OK, you're not sure why he's put the dash in.

Gemma: Is it because he's going up slowly?

Teacher: Yeah, well, what does a dash make you do when you're reading?

Gemma: It makes you slow down.

Teacher: Yes, it does; it makes you slow down, doesn't it ... so his right foot his enormous right foot lifted up out into space ... so he's not going straight down to the bottom, is he?

Emma: He's going up.

Teacher: He's going up, but very slowly, so I think you're right ... the dash slows it down. OK, and again that's a technique you can use in your own writing if you want to do that sort of thing. Are there any other techniques you weren't sure why he was using?

Laura: Snag to snag.

Teacher: Right... so ... from rock to rock snag to snag tumbling slowly.

George: Well, he's trying to use a rhythm so it's like going rolling down the hill.

Emma: Erm ... it's like when he tumbles down, he hits every rock when he tumbles down.

Teacher: Yeah, that's a good way of putting it ... that he hits every rock as he's going down ... so he doesn't just sort of fall off the cliff and go bump right down at the bottom of the cliff does he? Well let's read it and let's actually feel what he's doing.

(Children and teacher now read the text aloud)

Through dialogue with children, the teacher helps to consolidate their learning. His effective questioning technique enables them to clarify or expand on ideas formed during the peer discussion.

Teacher-student Conferences

Wells (1990) suggests that by discussing their own written texts with others, children can extend and refine their knowledge in a "conscious and deliberate manner" (p. 13). Children's conscious awareness became apparent during teacher-student conferences, when teachers were able to make direct references to specific stylistic features of mentor texts and the literary devices used by professional authors. In the following extract, a Year 6 (10-year-old) child discusses her draft narrative with the teacher:

Haylie's Text:

A grand house stood in the middle of a beautiful garden. Pretty flowerbeds trailed along the edge of a path that glistened in the summer sun. A girl stepped up to a wooden door that towered above her. She crept into the huge house. A silver staircase twisted and turned into the shadows of the enormous walls. A rosy red carpet welcomed her into the house.

Conference:

Teacher: I love that ... trailed along the edge of a path that glistened in the summer sun. You've said the girl stepped up to a wooden tower. Was there a tower by the house in the garden?

Haylie: No, that's the house ... the tower ... that's the house, it's like a tower. I was going to put the house was like a big tower towering over her, but then I thought it'd be better if I made the house seem like a tower to her.

Teacher: Oh ... right ... the tower is a metaphor.

Haylie: Yeah, it isn't a tower ... do you understand? It only seems like a tower in her head, 'cos she's just a little girl and it's a massive house.

Haylie's Text:

I woke with a start. What had the dream meant? I strained my memory but all I could remember was a huge wooden door and an elegant silver staircase.

Conference:

Teacher: These sentences are different, aren't they? (teacher reads aloud) I woke with a start. What had the dream meant? Did you want them to stand out?

Haylie: Yeah, because it has to be read like this (Haylie demonstrates in voice and gesture) so the sentences are sudden and sharp ... like a shock.

Haylie's Text:

The desolate house looked as if it was being strangled by the tangled ivy, with its long green arms and fingers slowly creeping round its neck.

Conference:

Teacher: I like how you describe the house. Earlier in the dream you described it as a tower but now you've added to that.

Haylie: Yeah, I wanted to make the house and the garden like characters, not just things, so I've used personification ... made the things more like people.

Haylie's Text:

The trees that surrounded the deserted house were overgrown and unruly like lots of excited little children left to play on their own in the playground. We stepped anxiously up the slimy steps, the door was my height. It looked strangely familiar. We pushed the door. It creaked open. We tiptoed inside. What a creepy house. BANG! The door suddenly slammed shut. Inside the house there were three identical doors with bright red paint peeling off them revealing the rotten wood beneath. An elegant silver staircase tainted with age twisted into the darkness like a slithering snake. I knew I had seen it somewhere, I just couldn't remember where. I looked around. There were no pictures, no wallpaper, no lights. Nothing.

Conference:

Teacher: As a reader, I'm really taken along by this passage. You're building such a vivid image in my mind and building suspense. You've got me wondering whatever they're going to find in the house. You must have given a lot of thought to this part.

Haylie: Mmm ... I did. I wanted to hook people so's they couldn't put it down ... like ... had to keep reading.

Teacher: So tell me how you did that ... tell me what kind of things were going through your mind as you were writing ... and I hope ... reading.

Haylie: Yeah, I was doing the two hats thing and ... well this bit here ... I just wrote the house was surrounded by overgrown trees but I thought I could make that better 'cos I umm ... I wanted it to be like ... not just big trees standing still but as if they were going mad ... like some little kids do in the playground ... act wild and that running around everywhere.

Teacher: So ... uncontrolled ... no one controlling them?

Haylie: Yeah.

Teacher: That's a really good idea; and then I see you've used another literary device as your characters walk up the steps.

Haylie: I've used some alliteration there; is that what you mean? Yeah, I wanted to show the steps were all green and covered in moss and I thought the "ssss" sounded good . . . the "ssss" sibilance.

This extract illustrates how the teacher is helping Haylie to refine her knowledge and consolidate her understanding. Through placing himself in the role of a genuine reader, the teacher is not only showing his appreciation of Haylie's text but also prompting her to explain and justify the linguistic choices she has made (e.g., I love that, Oh right . . . the tower is a metaphor, I like how you describe the house, That's a good idea, So tell me how you did that). By explicitly placing himself in the role of a reader, he is developing Haylie's awareness of audience (e.g., As a reader, I'm really taken along by this passage).

Discussion

Throughout the study, developments in children's linguistic awareness and in their ability to be critical readers and reflective writers became evident during peer and teacher-led discussions. Research has suggested that children in the elementary phase of schooling are able to borrow ideas and plots from literature, but are less likely to transfer more subtle stylistic elements into their own writing (Cairney, 1990). However, Flower (1994) suggests that an effective means of developing composition is through the use of collaborative writing in authentic contexts, whereby strategic skills are modeled and whereby children participate in what Dyson (1997) calls a "pedagogy of responsibility" (p. 180). Evidence from the study indicates that a critical evaluation of literature and an examination of literary devices can help children become more reflective writers. We found that with support from teachers (providing models, demonstrating and drawing attention to the features of mentor texts) and through focused group discussion, children began to develop their awareness of how texts are constructed. One of the most striking features to emerge from our work was the way children gradually developed a metalanguage and were able to use it effectively when discussing their own texts. The use of specific literary terms helped children to clarify their thoughts, identify issues, and engage in lucid, informed discussion. They were able to integrate the stylistic and organizational features of mentor texts into their personal repertoires and use them esuccessfully in their own writing.

However, the marked gain in attainment of the children in our study raises questions about general application and sustainability. Because pragmatic validity was an important methodological concern, the study took place in natural, whole-class settings and involved a variety of teachers working in different school contexts. Although rates of attainment varied, the progress made by most children exceeded the national expectation. Substantial improvements occurred in some schools and the possible reasons for this are something the project will focus on during its next phase (2006-07). Some schools were located in economic and socially depressed areas with high rates of crime, and it may be that class management was a particular challenge. However, the structured support approach (involving explicit teaching during literacy sessions), along with the careful scaffolding of children's learning during writing workshops, did seem to have beneficial effects on most of the children. The children appeared to respond positively to clear, attainable learning goals and continual support throughout the writing process. A common comment, found in children's journals, was that they valued teachers' guidance and feedback during the drafting process. Nevertheless, it seems unlikely that the rate of progress in attainment demonstrated over the research period is sustainable. One conclusion for the significant increase may be that without the structured support of teachers and the opportunity to work on extended pieces of work during writing workshops, most children were dramatically underachieving. Our post-study experience indicates that unless the approach is used consistently throughout a school, the significant gains are eroded. Students' progress was tracked over one semester after they had changed classes and begun a new academic year with different teachers. Steady progress was maintained when teachers continued to use mentor texts for the explicit instruction of literary devices. However, in those classes where teachers did not use the approach, the quality of students' writing deteriorated. Although a longitudinal study would be required to confirm our findings, it does seem that a whole school policy and consistent teaching approach is crucial in maintaining and developing gains in attainment. This illustrates the importance of incisive management and continual professional development in schools.

Some concern has been raised over the explicit teaching of literary technique, with Frater (2004) arguing that it may result in formulaic writing, so that individuality and creativity are stifled and children lack any sense of ownership. Graham (1998) shares this concern and describes teachers using "counter-productive, mechanical tasks" (p. 117) while Gibbons (2001) advises against "grafting technique on to none-too-willing children" (p. 16). Hilton (2001) is critical of what she sees as the "mechanical objectives-led approach" of the UK National Literacy Strategy and argues that its rigidity inhibits sustained independent writing (p. 8). We found little evidence that children's creativity was being stifled. Our study confirmed the view that freedom to experiment is essential for creativity, but so too are skills, knowledge, and understanding. The combination of explicit teaching, opportunities for sustained independent writing, and teacher support during composition enabled children to find and express their aesthetic voices (Bakhtin, 1986). Most children began to write with authority, drawing effectively on their strategic repertoires to orchestrate their writing. However, supporting young writers is a skillful business that requires knowledgeable, sensitive teachers. Scaffolding must be gradually withdrawn so children can become independent creative authors. Teachers need to assume a variety of teaching roles beyond that of expert and respond contingently to the needs of children as they discuss, plan, draft, and re-draft written work. We found that children benefit from experiencing what it is like to be an author-wrestling with problems, drawing on knowledge and experiences, seeking advice, and responding to critical comments. However, we also concluded that teachers need to be artists, themselves, modeling the creative process so that children can "see how we struggle, structure, think, re-read, revise and edit as we write" (Grainger, Gooch, & Lambirth, 2005, p. 167). Our findings suggested that, despite what many practitioners in the UK perceive as a nationally imposed prescriptive curriculum, imaginative and resourceful teachers can create a learning environment where they can work alongside children to develop writing as a generative, reflective process.