EIGHT TWEETERS TWEETING: A MULTI - CASE EXPLORATION OF YOUNG CHILDREN WRITING IN AN ONLINE SPACE By Holly Ann Marich A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Educational Psychology and Educational Technology Doctor of Philosophy 2020 ABSTRACT EIGHT TWEETERS TWEETING: A MULTI - CASE EXPLORATION OF YOUNG CHILDREN WRITING IN AN ONLINE SPACE By Holly Ann Marich need to know to be effective communicators online can inform elementary writing instruction and technology integration in writing cl assrooms. This study adds to the nascent research on composition moves of eight second - - form writing online for t heir class Twitter account. With a modified version of the Cognitive Writing Processes Model (Hayes, 2012) as a theoretical lens, I conducted a multi - case study, collecting data from field notes, written artifacts, screen capture, talk aloud transcripts, a nd video - short - form writing processes include a motivation to tweet, goal setting, in - advance, and in - the - moment planning, and specific writing schema knowledg e related to Twitter. Additionally, situationally specific and unique to Twitter in some ways, and shaped by the curriculum. Copyright by HOLLY ANN MARICH 20 20 iv This dissertation is dedicated to my husband, Bob and our children, Hannah, Bobby, and Benjamin . v ACKNOWLEDGEMENTS I wish to show my gratitude to my advisor, Dr. Douglas K. Hartman for his mentorship, patient guidance, and kind efforts in helping me finally reach the finish line to this monumental task. I also thank my committee, Dr. Christ ine Greenhow, Dr. Matthew Koehler, and Dr. Ralph Putnam for their leadership, guidance, and encouragement. The students who participated in this research, the teacher, principal, and families who permitted their children to participate all deserve special acknowledgment. My biggest thanks I give to the four people who have given the most and without whom this never would have been possible. Bob, Hannah, Bobby, and Benjamin my heart swells with love and happiness because of you. vi TABLE OF CONTENTS LIST O F TABLES xi LIST OF FIGURES xiii Chapter 1: Eight Tweeters Tweeting 1 Purpose of the Study 2 Question 3 Nature of the Study 3 Significance of the study 4 Overview 4 Chapter 2: Literature Review 5 Cognitive Writing Process Theories and Models 5 Experiential Theory 6 Cognitive Theories 8 Social Cognitive Theory 15 Sociocultural Theory 16 Cognitive vs. Socio Cognitive vs. Sociocultural Theory 17 Techno Cognitive Theory 18 Technologic Views 20 Children Writing in Online Spaces 21 Children Writing with Twitter 26 Selecting a Writing Process Model for this Study 27 Chapter Summary 30 C hapter 3: Method 32 Research Design 32 Setting and Participants 33 Community 33 School 33 Classroom 34 Teacher 35 Students 35 Consent Procedures 38 Role of the Researcher 39 Data Collection 40 Screen Capture Data 41 Talk - Aloud Data 42 Explaining the Task 44 Modeling a Talk Aloud 44 Student Practice 45 Student Composing a Tweet 46 vii Video Stimulated Recall Interview Data 46 Follow - up Prompts 47 Student - Written Artifact Data 48 Field Notes Data 49 Data Analysis 50 Adapting the Hayes (2012) Model 50 Writing Processes and Comprehensive Moves Analyse s 60 Ethical Considerations 62 Potential Risks and Benefits 62 Privacy 63 Bias 64 Chapter 4: Results 67 Individual Case Descriptions 67 Hal: High Scoring Male 68 General Description 68 Writing Processes 68 Writing Processes: Motivation 68 Writing Processes: Goal Setting 69 Writing Processes: Planning 69 Writing Processes: Writing Schemas 70 Compositio n Moves 71 Composition Moves: Letter, Word and Phrase Level Analysis 71 Composition Moves: Complete Tweet Analysis 76 Emergent Features 82 Reading URLs 82 Magnifying Glass Activation 83 Technology as a Knowing - Other 83 Summary of Hal -- Overall 85 Hope: High Scoring Female 85 General Description 85 Writing Processes 85 Writing Processes: Motivation 85 Writing Processes: Goal Setting 86 Writing Processes: Planning 86 Writing Processes: Writing Schemas 87 Composition Moves 89 Composition Moves: Letter, Wor d and Phrase Level Analysis 89 Composition Moves : Complete Tweet Analysis 94 Emergent Features 99 Blue Highlighted Words 99 Black Box Words 99 Copyright 100 Summary of Hope -- Overall 101 Inez: High Scoring Female 101 Gener al Description 101 viii Writing Processes 102 Writing Processes: Motivation 102 Writing Processes: Goal Setting 102 Writing Processes: Planning 103 Writing Processes: Writing Schemas 103 Composition Moves 105 Composition Moves: Letter, Word and Phrase Level Analysis 105 Composition Moves: Complete Tweet Analysis 111 Emergent Features 116 Word Tools: Red Underline, Blue H ighlight , and Autocomplete 117 Technology as a Knowing - Other 117 Summary of Inez -- Overall 118 Irene: Mid Scoring Female 119 General Description 119 Writing Processes 119 Writing Processes: Motivation 119 Writing Processes: Goal Setting 120 Writing Processes: Planning 120 Writing Processes: Writing Schemas 121 Composition Moves 122 Composition Moves: Letter, Word and Phrase Level Analysis 122 Composition Moves: Complete Tweet Analysis 130 Emergent Features 134 Word Tools: Red Underline, Blue Highlight, Magnifying Glass and Autocomplete 134 Summary of Irene -- Overall 135 Kip: Average - Low Scoring Male: 136 General Description 136 Writing Processes 136 Writing Processes: Motivation 136 Writing Processes: Goal Setting 137 Writing Processes: Planning 137 Writing Processes: Writing Schemas 137 Composition Moves 139 Composition Moves: Letter, Word and Phrase Level Analysis 139 Composition Moves: Complete Tweet Analysis 144 Emergent Features 149 Word Tools: Red Und erline and Autocomplete 149 Summary of Kip -- Overall 150 Kayla: Average - Low Scoring Female: 151 General Description 151 Writing Processes 151 Writing Processes: Motivation 151 Writing Processes: Goal Setting 152 Writing Processes: Planning 152 Writing Processes: Writing Schemas 153 ix Composition Moves 154 Composition Moves: Letter, Word and Phrase Level Analysis 154 Composition Moves: Complete Tweet Analysis 159 Emergent Features 164 Summary of Kayla -- Overall 164 Luke: Low Scoring Male: 165 General Description 165 Writing Processes 165 Writing Process es: Motivation 165 Writing Processes: Goal S etting 166 Writin g Processes: Planning 166 Writing Proc esses: Writing Schemas 167 Composition Moves 168 Composition Moves: Letter, Word and Phrase Level Analysis 168 Composition Moves: Comp lete Tweet Analysis 172 Emergent Features 176 Summary of Luke -- Overall 176 Lori: Low Achieving Female: 177 General Description 177 Writing Processes 1 77 Writing Processes: Motivation 178 Writing Processes: Goal Setting 178 Writing Processes: Planning 178 Writing Processes: Writing Schemas 178 Composition Moves 180 Composition Moves: Letter, Word and Phrase Level Analysis 180 Composition Moves: Complete Tweet Analysis 186 Emergent Features 191 Summary of Lori -- Overall 191 Cross - Case Comparisons 192 Cross - Case Comparisons Writing Processes 192 Cross - Case Comparisons Composition Moves 196 Cross - Case Comparisons of Com position Moves: Letter, Word and Phrase Level Analysis 196 Cross - Case Comparisons of Composition Moves: Complete Tweet Analysis: Overall 206 Cross - Case Comparisons of Composition Moves: Complete Tweet Analysis: in Thirds 210 Cross - Case Comparison s of Emergent Features 212 Autocomplete 216 Red Underline 217 Blue Highlight 218 Technology as Knowing - Other 220 Magnifying Glass Activation 220 Black Box Words 221 Copyright 222 x Reading URL 222 Su mmary of Cross - Case Comparison -- Overall 223 Chapter Summary 223 Individual Case Summaries 224 Cross - Case Comparisons Summary 228 Chapter 5: Discussion, Limitations, and Implic ations 229 230 Writing Processes: Motivation 230 Writing Processes: Goal Setting 231 Writing Processes: Planning 231 Writing Processes: Writing Schemas 232 Composition Moves: Composing 232 Composition Moves: Revision Processes 233 Composition Moves: Task Environment 234 Composition Moves: Resource Level 234 235 Emergent Features: Autocomplete, Red Underline, and Blue Highlight 235 Emergent Features: Magnifying Glass Activation, Black Box Words, Copyright and Reading URL s 236 Emergent Features: Technology as Knowing - Other 236 Twitter Writing is Shaped by the Curriculum 237 Curriculum and Motivation 238 Curriculum and Goal Setting 239 Curriculum and Planning 239 Curriculum and Writing Schemas 240 Limitations 241 Implicat ions 242 Future Research 243 APPENDICES 247 APPE NDIX A: Assent Script 248 APPENDIX B: Talk Aloud and Interview Analysis Example 250 APPENDIX C: Screen Capture an d Think Aloud Data Example 262 APPENDIX D : Eight Tweeters T weeting Code Book 286 REFERENCES 2 99 xi LIST OF TABLES Table 1 Overview of differences between the current study and existing literature 26 Table 2 Study participants 37 Table 3 Data Collection Tools : Description and Purpose 40 online writing processes 54 frequency - of - use in thirds 80 fre quency of use in thirds 97 frequency of use in thirds 115 frequency of use in thirds 132 frequency of use in thirds 147 e in thirds 162 frequency - of - use in t hirds 175 frequency of use in t hirds 189 Table 13 Cross - indicate patterns 193 Table 14 Composing code to tals for Luke and Irene 198 Table 15 Revision Processes code t otals for Inez an d Kayla 200 Table 16 Task Environment code to tals for Luke and Hal 203 Table 17 Resource Level code to tals for Inez and Irene 205 Table 18 Cross - indicate patt erns 207 Table 19 Most Frequent code occurrence divided into three time periods, highlighted to indicate patterns 210 Table 20 Summary of All Cases Composition Moves frequency of use in thirds 211 Table 21 Cross - case mergent features highlighted to xii indicate patterns 21 3 Table 22 Talk Aloud and Interview Analysis Example Writing Processes 250 Table 23 Control Level Aspects of Genre Knowledge and Text Structur e 253 Table 24 Control Level Goal Setting 254 Table 25 Talk Aloud and Interview Analysis Example Composition Moves 255 Table 26 Talk Aloud and Interview Analysis Example Editing for Spelling, Spacing, Capital Letters, and Punct uation 256 Table 27 Other Interesting Notables Keyboard Technology Confusion and Understanding 257 Table 28 Other: Possibly Related to Social Norms 261 Table 29 Screen Capture & Think Aloud Data Example 262 Table 30 Writing Processes Codebook 286 Table 31 Composition Moves Codebook 289 Table 32 Process Level Codes: Composition Moves related to Transcription 291 Table 33 Process Level Codes: Composition Moves related to Revision 292 T able 34 Process Level Codes: Composition Moves related to Task Environment 294 Table 35 Process Level Codes: Composition Moves related to Resource Level 296 Table 36 Emergent Features Codes 297 xiii LIST OF FIGURES Figure 1 The Cognitive Process Model of the Composing Process (Flower and Hayes, 1981) 9 Figure 2 A Recursive Model, Hayes (1996) 10 11 Figure 4 Bereiter and Scardamalia Knowledge telling model (1987) 12 Figure 5 Bereiter and Scardamalia Knowledge transforming model (1987) 12 Figure 6 The Flexible Focus Model (Hayes, 2011) 14 Figure 7 The Fixed Topic Model (Hayes, 2011 ) 14 Figure 8 The Topic Elaboration Model (Hayes, 2011) 14 Figure 9 A Model for Writing Development (Andrews & Smith, 2011) 19 Figure 10 Field Notes Example 49 Figure 11 Flowchart for 2nd - grade tweeters genre k nowledge skill levels 59 - of - composing 72 ency of Use Timeline Analysis 78 81 - of - composing 90 ncy of Use Timeline Analysis 94 97 - of - composing 106 quency of Use Timeline Analysis 11 2 115 xiv tter/word/phrase flow - of - composing 123 - of - composing 124 quency of Use Timeline Analysis 130 omposition moves in thirds 133 - of - composing 140 quency of Use Timeline Analysis 145 thirds 148 - of - composing 155 quency of Use Timeline Analysis 160 163 - of - composing 169 Timelin e Analysis 173 175 ter/word/phrase flow - of - composing 180 quency of Use Timeline Analysis 187 189 Figure 37 Side - by - flow - of - composing figures 198 Figure 38 Side - by - flow - of - composing figures 201 Figure 39 S ide - by - flow - of - composing figure 203 xv Figure 40 Side - by - flow - of - composing figures 205 Figure 41 Teacher - created handout for Tweeting 238 Figure 42 Pedagogical possibilities for supporting children com posing short - form prose online 243 Figure 43 Published Tweet Example 255 1 Chapter 1: Eight Tweeters Tweetin g 21st century writing technologies are altering the uses of writing largely because of the Internet and the ever - evolving digital technologies that it affords (Leu, Kinzer, Coiro, & Byrne, Zawilinski, McVerry & Everett - Cacopardo, 2009). For example, composing emails, blogs, and webpages (which are typical Internet writing spaces) requires knowledge of unique coding scripts, genres, keyboarding tools, layout designs, touchscreen deftne ss, and other cognitive, cultural, technological, and social skills. In short, to communicate via digital writing today is quite different from the earliest analog records on clay tablets, both in terms of the tools, skills, strategies, and dispositions, b ut also in terms of the rapidity with which these tools, skills, strategies, and dispositions evolve. Thus, the evolution of writing, especially in the modern era, has highlighted the need to conceive of writing in broad, technologically deictic and protea n terms -- especially when it comes to understanding the online composing of so - The communicative use of digital technologies by young people has been of scholarly interest for more than a decade (e.g., Brandt, 2014 ; Burnett & Merchant, 2015). While consumptive uses of the Internet have garnered much of this scholarly attention related to literacy (i.e., reading, viewing, listening), the creating and designing affordances of the Internet (i.e., writing, speaking, vis the online writing processes of younger children shows signs of development. Two t ypes of writing is tracked and logged (e.g. Van Waes, Leijten, L indgren, & Wengelin, 2011). These 2 studies, while informative, have limited use for building a more comprehensive view of what happens when children write online. For example, these studies do not take into account what children think as they compose. The absence of scholarship on the writing processes of children writing in the digital medium indicates a need for further study (Harris, Graham, Brindle, & Sandmel, 2009). To be sure, there are a handful of phenomenological studies that examine the processe s children use to compose multimodal digital text, such as email (Burnett & Myers, 2006; Maher, 2010), blogs (McGrail & Davis, 2011) and reports (Mitchell, Thompson, & Anderson, 2017). Generally, the digital writing in these studies occurs in spaces that a re not on the open Internet but occur within highly restricted spaces that simulate the Internet. These moves are taken by educators as a safety measure and by researchers as a study - control design measure. As a result, the findings provide one view into y oung children composing with digital technologies but this lens is insufficient for a comprehensive view (Coiro, Knobel, Lankshear & Leu, 2008). In sum, the scarcity of information on children writing in online spaces is regrettable because this type of wr iting will be an important part of a 21 st century skills set (Marsh, 2014; Rideout, 2017). What young writers know or need to know to be effective communicators online can inform better ways to teach students. Therefore, this study aims to contribute to th e knowledge base by examining the writing processes and composition moves young children use when composing online with Twitter, a microblogging, social - network space on the Internet. Purpose of the Study The purpose of this multiple case study was to exa processes and composition moves (e.g. development of ideas, revision, editing, awareness of audience) while composing tweets for their class Twitter account. Themes were generated from 3 observations, written artifacts, and conc urrent talk aloud and video - stimulated recall interview transcripts from second grade writers. The analysis focused on the composing process from the generation of an initial tweet idea through the transcription onto an iPad and out into the Internet as a published tweet. Question The question guiding the collection and analysis of data for this stud y was: What are the writing processes and composition moves made by second graders when composing tweets for online publication? Nature of the Study A multi ple case study design was used for this study to better understand what happens when young children write in one type of online space. Because writing has been conceived of as a complex social phenomenon by a number of scholars (e.g., Graves, 1973; McKee & Porter, conceived of as a personal experience (e.g., Hyland, 2015; Kellogg, 1999) pos sibly leaving them vulnerable to the criticism of others or themselves (Johnston, 2012), a case study design was selected because I wanted to learn about the personal experiences of writers. For these reasons, a multiple case study approach was best suite d to answer the research question stated above. Five types of data were generated and collected for this case study: screen capture recordings, talk - aloud transcripts, video - stimulated recall interview transcripts, student - written artifacts, and field notes. To identify themes and patterns, the data were analyzed using emic, descriptive codes discovered from the data and a priori codes influenced by t he 2012 writing process model by John Hayes. 4 Significance of the Study This case study is important for several reasons. First, it deepens our understanding of the writing processes and composition moves that young writers employ while composing online. T o date, there is limited understanding. Second, the study extends our understanding by identifying new writing processes and composition moves unique to composing online. To date, there is insufficient understanding. Third, the study expands our understand ing of the limitations and possibilities of composing online. To date, there is underdeveloped understanding. Finally, this understanding. Overview This dissertation follows the traditional five - chapter structure. This first chapter briefly describes the continuously evolving nature of writing as a human technology, followed by a brief explanation for the purpose, nature, and significance of the propo sed study. Chapter 2 reviews a body of writing research literature, acknowledging the complex and vast nature of writing research, narrowing in on writing research specific to writing processes specific to cognitive theories and models of writing and child ren writing in online spaces . Chapter 3 describes the methods for answering the research question by establishing the use of case study methodology as appropriate for addressing the specific research question previously established, including a description of data collection. Chapter 4 provides a detailed analysis of the data for each of the eight individual case studies followed by a cross - case analysis and a final summary section. Finally, Chapter 5 provides the discussion of results, limitations and impl ications of the study. 5 Chapter 2: Literature Review My focus on writing processes does not disregard two other important bodies of literature: the online presence and identity dimensions of writing and the social and cultural elements of writing. part of a larger project that will in time examine the many facets of composing digitally through different identities, presences, cultures, and social arrangements. Therefore, th is review of literature zooms in on writing research specific to some of the well established cognitive writing process theories and models, identifying a best - fit model as a theoretical framework for this study. This review of literature also addresses th processes in situ while composing in online spaces. Cognitive Writing Process Theories and Models Theoretical traditions in writing research have developed over time . Early scholars of the writing process, infl uenced by psychologist Jerome Bruner, thought of writing as a cognitive process (i.e., Emig, 1971; and Moffett, 1968). Wilkinson, Barnsley, Hannah, and Swan (1980) extended that thinking to include the affective, moral, and stylistic aspects of the writing process. include the sociocognitive aspects of writing (Flower, 1996; Bazerman & Prior, 2005). In the sections that follow I provide a diachronic review of the co gnitive - oriented theories and models in writing research. These theories and models focus on what the writer is mentally doing in moments of composition. The account begins with well - established theories and models of writing, both the process approach and the cognitive process approach theories of writing. Then it continues with the earliest social cognition theory of writing. Next, the account traces writing theories into the 21st century, such as the theory of developing writers which proposes a 6 theoreti cal possibility for future writing process research. Finally, the section concludes with a theoretic account of the writing process bases for short - form writing and a nominal review of the literature on children writing in online spaces. Experiential The ory Early writing research focused on product of writing rather than process, which held a dominant place in the field. Examples of this product - (1965) T - y of writing moved to focus on the process of writing when Gordon Rohman (1965) examined what writers were doing while writing in school. From his observations of writers, a three - stage linear model of a writing process emerged, which included prewriting, writing, and rewriting. Subsequent research was not limited to these three stages nor was it a linear process (Emig 1971, Murray 1984, Graves, 1983). To better understand the writing p rocess, Emig (1971) watched her 12th graders write and asked them to think aloud as they wrote. Her findings invited scholars to question the three - stage linear model of the writing process introduced by Rohman (1965). The result was a more contemporary de scription of the writing process while still accounting for the three major processes: prewriting, writing and revising/editing. Emig captured the complex and messy reality of writing often described in current literature: ...the writing process is a recur sive, idiosyncratic, situation - dependent set of activities we engage in to produce a piece of writing. These activities are embedded within broader categories or phases, the hallmark of the writing process: prewriting, writing, and re - writing (Loc. 1390 of 4954 Andrews and Smith 2011). 7 were heralded , in time, her research was criticized for a lack of scientific rigor (North, 1987). The process approach eventually led to considering the classroom environment and pedagogical moves that (e.g., Murray, 1968, 1985; Graves, 1983, 1994; and Calkins, 1986, 1987). Within the structure of a writing workshop, a single process of writing was not dictated to students. Rather, each writer adopted a process that worked best for him/her based on strategic instruction about what writers research observing young children in the act of writing. When ask ed how he came up with his 1971 dissertation topic about children as writers, Graves admitted, When I reviewed the research on writing, no one had ever sat next to kids and watched what they did when they wrote. Janet Emig had sat next to 12th graders, know it at the time believe no one had actually sat down next to kids; so I did it (Routman, 1995 p. 2). Throughout his career, Graves continued observing kids as they wrote and developed a repertoire of li terature for practitioners about children and writing. Graves also built on the work of Don Murray, (a journalist before becoming a teacher of writing) who taught college writing aves found that children benefited from the evidence - based findings that children want to write and can write (if given the time, resources, instruction, purposes, audiences, and independ ence) framed his legacy. P edagogical philosophies supporting children as writers championed by Graves have been q uestioned (see 8 Graham, 2006). Nonetheless , his work set the stage for subsequent studies that examined in - the - moment cognitive processes of wri ters. Cognitive Theories Overlapping the work of Graves and others during the 1980s and 1990s was a strand of research that focused on writing from a cognitive processes perspective (Becker, 2004). Based on the computer metaphor, these information processi ng models depicted the mind working similar to the input/output algorithmic functions of computers. The most prominent cognitive model of writing during these years was developed by Flower and Hayes (1981) accounting for the recursive nature of writing thr ough a hierarchical rather than linear description (Cooper & Holzman, 1983). Their initial information processing model portrayed writing as a problem - solving activity, made up of four internal -- in - the - head -- conditions (planning, translating, reviewing, a nd monitoring) (See Figure 1). The model made a first - of - its - kind contribution to writing research, but in time was critiqued for two primary limitations: (a) the model did not account for context, and (b) the model represented expert writers rather than n ovice writers. A lesser criticism was the focus on planning through goal setting without an emphasis on scripts , which were detailed steps to produce what has been planned (Cooper & Holzman, 1983). 9 Figure 1 The Cognitive Process Model of the Composi ng Process (Flower and Hayes, 1981) In response to these criticisms, Hayes revised the original model in 1996 to account for external conditions (i.e., context) that influenced writing tasks (See Figure 2). This revised model privileged three cognitive writing processes, a) text interpretation, b) reflection, and c) text production. The revised model also added and clarified the cognitive writing processes of long - term and working memory, motivation, and affect. A limitation of this revised model was an absence of the different strategies employed at the task level. According to Deane, Odendahl, Quinlan, Fowlers, Welsh, and Bivens - Tatum (2008), given the complexity of writing, each task calls upon a different set of cognitive strategies. For instance, tex t interpretation calls on reading comprehension skills while text production includes transcription skills such as spelling (Berninger, Cartwright, Yates, Swanson, & Abbott, 1994). 10 Figure 2 A Recursive Model, Hayes (1996) More recently, Hayes (2012) developed an expanded model to address early concerns and to more accurately represent new developments in the field regarding cognitive writing processes (See Figure 3). This more elaborate model was considerably different than previous models. For example, the monitor, intended to represent individual differences in writers, was removed because he thought it misleadingly appeared to be the center of all writer actions. Additionally, this more recent model was divided into three levels: (a) control level, (b) process level, and (c) the resource level. A limitation of this 2012 model, acknowledged by Hayes himself, is the limited scope motivation plays in writing. He agrees the model provides sufficient detail regarding motivati on and goals setting, but motivation related to other aspects of writing -- such as transcription or evaluation -- are not represented (Hayes, 2012). 11 The central concern about the Hayes 2012 model, as it relates to the proposed study is that it is informed by evidence from studies that examined offline writing processes. Furthermore, the 2012 model, like the models that preceded it have been developed based on evidence of adult writers and long - form writing. Figure 3 Haye s ses of writing Based on data from the cognitive writing processes and products of child writers, Bereiter and Scardamalia (1987) developed two models of varying degree of sophistication: (a) the knowledge telling model , a leaner and simpler model which was based on evidence from knowledge transforming model , a richer and more complex model which was based model (see Figure 4) represented a focus on local issue s such as spelling and the automatic retrieval of information. The knowledge transforming model (see Figure 5) represented a focus on more global issues such as thesis development and the strategic retrieval of information (Deane, et. al, 2008; Hayes, 2012 ). 12 Figure 4 Bereiter and Scardamalia Knowledge telling model (1987) Figure 5 Bereiter and Scardamalia Knowledge transforming model (1987) 13 The limitation of both the knowledge telling and knowledge transforming models is the broad - stroke overview they represent of the writing process (Hayes, 2011). In spite of this limitation, Hayes (2011) recognized potential in the knowledge telling model through additional layers of detail he called sub - strategies. Other limitations raised by Gagnon (2014) related t o problem solving: he thought novice writing did not require problem solving, and that young writers can overestimate their abilities, thereby misrepresenting the writing task (Gagnon, 2006). were developed from evidence of adult wr iting, he sought to with sub - strategies for expository writing based on structures identified by Fuller (1995). By analyzing expository essays written by 1st - thr ough 9th - grade students, a subset of the same data set used by Fuller (1995), Hayes identified three sub - (1987) knowledge telling strategy. These sub - strategies were, listed in increasing complexity of writing abili ties: (a) flexible focus, where the writer does not maintain focus on a general topic (See Figure 6); (b) fixed topic, where every sentence connects to one topic, found commonly in grades 1 through 5, (See Figure 7); and (c) topic elaboration where a ge neral topic maintains the focus with subtopics introduced, found most often in grades 6 - 9 (See Figure 8) (Hayes & Berninger, 2014). 14 Figure 6 The Flexible Focus Model (Hayes, 2011) Figure 7 The Fixed Topic Model (Hayes, 2011) Figure 8 The Topic Elaboration Model (Hayes, 2011) 15 Hayes then designed computer programs using the Python language (Hayes, 2012) to check whether his models for his three sub - actually produce the text structures that they [were] designed to produce sub - strategy model and text with 96% of the essays. Hayes concluded that these three sub - strategies better guided the details of instruction for strategy use when based on student cognitive skills th an when based on the overarching strategy of knowledge - telling proposed by Bereiter and Scardamalia (1987). Social Cognitive Theory Extending the mostly cognitive - oriented models and theories of writing presented in the preceding sections, Linda Flower (19 94) proposed a theory of writing that explicitly acknowledged the social elements that work in concert with the cognitive elements of writing. She and others had expressed concerns that the cognitive information processing models (e.g., Flower and Hayes, 1 981) were incomplete (Flower, 1989). At the time, a clear epistemological and methodological division between social theories and cognitive theories of writing were visible. Discussing what she called the social and cognitive continuum, Flower explained, . ..[T]here is no way to isolate a social process from the minds that carry it out . Although we can treat public statements, social conventions, or interpersonal events as independent objects, if we look closer, they are the collaborative creation of individ ual minds over time . They only exist as meaning in the interpretations individual readers and writers give to them (p. 31 of 338 Google Play digital text, 1994). Based on this logic, she argued to integrate the cognitive and social when trying to understa nd the process of writing because one constructs the other. The social context builds 16 cognition, and in turn, cognition mediates the building of social context (Flower, 1989). Flower also called attention to the limitations of a social cognitive theory of writing, which was constructed from the methods of observation and analysis of social or cognitive activity. She argued that which is observed or noticed by the observer and the unique interpretation by that observer provides the data and tells the story w hich theorize a phenomenon. Flower calls the digital text, 1994). Because of their bluntness, the social cognitive theory of writing is considered both prominent a nd more comprehensive for modern day research than two other prominent theories, the cognitive process theory of writing mentioned in the previous section and the sociocultural theory of writing (Leggette et al., 2015). Sociocultural Theory Similar to soci al cognitive theory, sociocultural theory expanded the lens for understanding acts of writing. Unlike social cognitive theory, sociocultural theory did not account for what happens in the brain while writing. Rather, sociocultural theory viewed the cogniti ve dimensions of writing development as embedded in social and cultural interaction (Vygotsky, 1980). As a result, writing as seen through a sociocultural lens situated any act of composing g (Prior, 2006). Furthermore, sociocultural theory viewed writing as an artifact mediated by cultural tools and as a practice embodied by culture and context (Graham & Olinghouse, 2009; Prior, 2006). Indeed, Prior (2006) elaborates on the rich and complex nature of sociocultural theories of writing, Sociocultural theories of writing have found, however, that they cannot live easily within the borders of a folk notion of writing, so studies increasingly explore more 17 erest in writing leads to writing and reading, talk and listening, observation and action, and feeling and thinking in the world (Prior, 2006, Kindle location 1332 of 10594). Based on sociocultural theory, writing is a collaborative social activity that i s embedded with motivation, affect, and cultural influences on cognitive processes (Hodges, 2017). Sociocultural research examines how writing is learned and used in a range of settings and how writing permeates sociocultural practices (Perry, 2012). It al so focuses on learning to write from between independence and inability to acco mplish a learning goal. Within this zone of learning, the support of a more knowledgeable other helps the learner gain independence of the learning goal. Critics of sociocultural theory point to the ambiguity of identifying and measuring an ne of Proximal Development ( Allal & Ducrey, 2000) and the context specific nature of sociocultural theory, which limits study - specific results from being synthesized with other results across multiple contexts (Perry, 2012). While sociocultural theory views cognition as a collaborative process, the appl ication of this theory to writing research has yielded little toward understanding the cognitive processes used by writers that extends beyond what cognitive theories of writing have developed (Leggette, et al., 2015). Cognitive vs. Socio Cognitive vs. Sociocultural Theory Leggette and colleagues (2015) applied theory evaluation criteria (Dudley - Brown, 1997) to the three theories addressed in the previous sections ( the cognitive, social cognitive, and 18 sociocultural theories of writing) to evaluate their abstract page). The seven criteria used in their analysis were: accuracy, which depicts components of the writing process; consistency, which is based on internal consistency and evidence of reliability; fruitf ulness, which means the theory has research potential; simplicity/complexity, which means the concepts identified in the theory are consistently simple or consistently complex; scope, which signifies dependence on the phenomenon and its context; accepta bility, which indicates the level to which the theory has been adopted ; and sociocultural utility, which means the theory accounts for cultural differences (Leggette et al., 2015). The social cognitive theory of writing was evaluated as the most complete using the Hayes and Berninger (2014), for instance, claimed that cognitive theory acco unted for sociocultural influences because both social and cultural elemen ts constitute long - term memory and task - were not represented in long - 9 of draft). Techno Cognitive Th eory The last decade has seen the emergence of a techno cognitive theory for writing (and the production of other sign systems) (Schürer, 2006). The seeds for this theoretical work (which is at the intersection of technology, cognition, and writing) were s students at Carnegie Mellon University (e.g., Ackerman, 1994; Ackerman & Oates, 1996) who 19 studied the workplace writing of architects and other professionals. Building on this work, Andrews and Smith (2011) made the argument some y ears later for a cognitive theory of writing that integrates the role of digital technologies. At the heart of their argument is the claim that current theories of writing emerged from the material conditions of off - line, long - form writing. As such, they a re limited and insufficient for understanding on - line, technologically mediated short - form writing. Andrews and Smith further argued for a new theory of writing, one which described the developing writer rather than writing development . See Figure 9. Taken together, their arguments put the writer, rather than the writing process or product, at the center of a theory that has been broadened to explain writing across media (on - and off - line) and forms (from long - to short - form), thus requiring an architecture 190, Kindle loc 2018 of 4954). Figure 9 A Model for Writing Development (Andrews & Smith, 2011) 20 Technologic Views Complicating the theoretical challenge of our time has been the steeping of technology challenge was to examine trends in the literature over a 30 - year period. She identifies five technologically - mediated writing tools commonly addressed: (a) word processing; (b) e - mail; (c) chat and discussion boards; (d) instant messaging; and (e) social networking software. She also identifies five multimodal composition modes: (a) visual; (b) aural; (c) video; (d) performative; and (e) three - dimensional. Despite the presence of technologically - mediated and multi - modal writing tools in the literature, she notes that there is little research which addresses the composing processes with these technological tools and compositional modes in situ (p. 3). favors a larger focus on cultures of literacy rather than a smaller focus on individuals and their unique writing processes. As a result, she argues for the close examinat writing processes in a network space. For her research, Takayoshi chose to start by examining the visual mode of short form writing in a social networking software application (Takayoshi, - be considered trivial short - form writing. For example, Takayoshi found in her study: a number of tradit ional writing processes are used when writing short - are too. For instance, she identified so - called horizontal and vertical processes (p. 9). The horizontal processes describe the multiple writing spaces, audiences, co ntexts, and genres (email, Twitter, posting to discussion boards, word processing) one might give attention while also composing on Facebook, the primary online space of her study. Vertical writing processes are 21 characterized by the forward and backward re Although the composition process of writing on Facebook has been examined by Takayoshi and several others (e.g., Shepherd, 2015), the short form medium of Twitter has not. However, there exists anecdotal accounts of children writing and Twitter (e.g., Kurtz, 2009; Marich, 2016; & Waller, 2010). is a nominal review of the literature because there are only a half dozen studies to date. The literature is divided into two sections: (1) children writing in online spaces (the research literature generally), and (2) children writing with Twitter (the professional li terature specifically). The first section is further divided into subsections according to the platform: (a) emailing, (b) blogging, and (c) social networking profiles. Children Writing in Online Spaces While there are a number of studies that look at adult writing in online spaces, (e.g., Mills & Chandra, 2011; Riley, 2015;Takayoshi, 2015), there are only a couple of studies that look at children writing in online spaces. One such study, conducted in the United Kingdom, surveyed children age 8 to 16 about their writing practices. The results indicated that children who blogged or had a social network profile were more confident writers and displayed a positive attitude toward writing and computer use comp ared to those who did not blog or participate in a social provides evidence that young children are writing online, little consideration has been given to the wr iting processes employed by young children while writing online. I summarize a selection 22 of literature about children composing emails, blog posts, and social network site profiles, noting how these existing studies overlap but do not directly align with t he proposed study. Writing an email. Burnett and Mayers (2006), Merchant (2005b), and Wollman - Bonilla (2003) investigated the composition moves visible in the emails of children. The Burnett and Myers and Merchant studies (which drew upon the same data), for instance, found that the 5th - graders in their study used a formal writing style with initial emails between neighboring - school writing partners. The writers were conscious of surface level features including spelling, punctuation, and word choice. It w - to - face that the email writing style became informal, showing less concern for mechanics. Furthermore, students were confident and enthusiastic about their writing, recognized multimodal elements as key to meaning making, and engaged in ongoing revision as the composition developed, regularly checking that their writing made sense. To enhance verbal meaning with visual effects students used emoticons at both the overall message and individual word levels. For exampl e, within individual words a smiley - d (Merchant, 2005 b, p. 56 ). Wollman - Bonilla & Carpenter, (2003) observed her six - year - old daughter Rosa as she engaged in ongoing corresponden ce with relatives through both email and traditional paper - pencil mail. Although Rosa used correct punctuation and capital letters when writing paper - pencil mail, the email writing lacked conventional punctuation. When writing paper - pencil letters Rosa wro te with a formal style, indicating an awareness of audience. Conversely, her email writing included an informal conversational style, reflecting an assumption that temporal, 23 Writing a blog. McGrail and Davis (2011) studied the composition moves of students engaged in a 5th grade classroom blogging project. While conventions and mechanics were not emphasized during the project, students paid considerable attention to them because they wanted to present themselves well to their audience and connect with them. Thus, the impact of their mind. This audience awareness did not happen immediately. Rather, it developed over time, as students transitioned away from thinking of the teacher as their primary audience. Blogging also increased student confidence and motivation as writers. For examp le, the students were assertive when blogging about social topics like the importance of recycling and the unacceptable conditions of dirty public restrooms (p. 429). Confidence was also demonstrated through blog comments between fellow students about topi cs such as the social implications of correct spelling when blogging. Over time, as student confidence grew their writing showed evidence of taking ownership of the writing aborate ideas, and using playful language through idioms and metaphors. Writing on a social network. Lindstrom and Niederhauser (2016) and Dowdall (2009) both studied the composition moves visible in the social network writing of children. Lindstrom and Ni ederhauser, for instance, studied literacy - related activity of three 5th grade female students using a closed social network site, Ning. They found that profile curation and writing style of r example, one student, identified as an experienced social network site user, modified her profile page more often than novice social network site users and wrote with a less formal style similar to instant messaging. As a result, this student experienced social success among her peers within the Ning online space. In contrast, another student, identified as a novice social network site user, gave little attention to 24 her profile page, used a formal in - school style of writing, and often posted personal or s ensitive Similarly, Dowdall (2009) studied the literacy - related activity of one 12 year - old female student using the closed social network site, Bebo. Dowdall found competing tensions between the site structure and agency of writers on the site. For example, because of the co - authored st manage the elements added by friends, followers, and commenters, but has limited control. Furthermore, when additional writing - space elements are added, the primary author can control to maintain her chosen online identity. The sociocultural aspects inh overlooked or misrepresented in typical school curriculums for writing. The current study. The current study builds upon, but differs from, the studies reviewed above. First, my study used a different online platform, namely Twitter. While children today continue to write using email, blogs, & social networks, writing on Twitter (a hybrid of the three, - messaging feature, microblogging format, and social media connectivit y) looks poised to be used by teachers and younger students given the prominence and access to the platform. Second, the current study examined younger participants (8 year - olds) rather than the pre - adolescent and adolescent children in five of the six stu dies reviewed. There is some evidence that younger children are doing more writing online than previously thought (Internet Foundation in Sweden ( Davidsson, & Findahl, 2016). Third, the current study uses a writing process model framework, Hayes (2012). The studies reviewed above have largely been agnostic when it came to use of a model, framework, 25 or theory. My intentional use of an a priori writing framework better situates the current study within existing literature addressing contemporary writing - pro cess theories and models. Fourth, the current study focused not only on composition moves (as the studies above did), but extends into an examination of the writing processes of children writing, allowing for a more comprehensive representation of the writ recordings and talk aloud data I can identify the specific number of composing or revising moves a child uses while also providing what the child was thinking about during that time. And fifth, the current study used different data collection methods. Data collection for the current study involved one - on - one audio and video sessions of student talk aloud matched with video screen capture of what the student is doing on the screen. The talk aloud session was immediately followed by audio and video recorded video stimulated recall interviews. Specific to data collection, all six studies reviewed used observations of students writing within the context of the classroom. Both this approach and the out - of - classroo m one - on - one approach used for the current study has its own cost benefit. For example, the in - classroom observations preserves the original writing situation at a cost of a thinner data set of detailed on - screen composition moves. While the one - on - one obs ervations described above allows access to a more comprehensive and thick data set used to describe the writing processes and composition moves, in situ . This is at the cost of the student writing away from the original classroom writing situation. This co st seems reasonable given the current study research question is focused on student writing processes and comprehension moves in situ . Table 1 provides an overview of how each of the six studies reviewed differ from the current study in four of the five ar eas: online platform, student age, writing process model framework, and data collection methods. 26 Table 1 Overview of differences between the current study and existing literature Study Online platform Student average age writing process model framework o r writing elements to frame analysis data collection methods Present Study Open SNS Twitter 2nd grade 7 & 8 yr Hayes (2012) Screen - capture recordings With simultaneous talk - aloud Video - stimulated retrospective interviews Product analysis Burnett & Mayers (2006) email 5th grade 10 & 11 yr in class observations, after project completion interviews, product analysis Merchant (2005b) email 5th grade 10 & 11 yr language use experience of digital communication visual affordances critical awareness in class observations, after project completion interviews, product analysis Wollman - Bonilla and Carpenter (2003) email Kindergarten, 6 yr style, audience awareness, mechanics in home observations, after project completion interviews, product analysis McGrail and Davis (2011) blog 5th grade 10 & 11 yr attitude, content, voice, connections and relationships, thinking, craft in class observations, after project completion interviews, product analysis Dowdall (2009) Closed SNS Bebo 6th/7th grade 12 yr different types of representation in art Impressions Improvisations compositions interviews product analysis Lindstrom & Niederhauser, (2016) Closed SNS Ning 5th grade 10 & 11 yr Not specified interviews product analysis Children Writing with Twitter Practitioner accounts by Waller (2010) and Kurtz (2009) described the Twitter - based 27 primary classroom. While the descriptions of these three practitioners are limited in scholarly scope and rigor, they suggest several themes relevant to the study of Twitter - based writing. Waller, for instance, noted that his young writers began to recogni writers were more effective at revising and editing their writing as they composed tweets in notebooks before publication than these same writing tasks during writing workshop. And Marich noted that the primary teacher also observed most of her students rereading their tweets, fixing grammar, punct uation, and capitalization before publishing. While hard and fast conclusions can not be drawn from these three case studies, they suggest possible reasons why further study of writing processes and composition moves with Twitter is needed. Selecting a Wri ting Process Model for this Study In this chapter I have reviewed several writing process theories and models. Because of the comprehensive nature of the writing process model by Hayes, (2012) I will use this model -- with revisions reflecting how children and adults differ (Hayes & Olinghouse, 2015) -- as a theoretical lens to examine the writing of children in an online space. Like nesting dolls, where all other dolls in a set become a part of the largest doll, I recognize elements of the other writing pro cess theories and models previously mentioned have a part in the Hayes (2012) model. For example, the foundational writing processes identified by Rohman (1965) and better understood in practice by Emig (1971) are nested within the writing processes sectio n of the process level in the Hayes (2012) model. The motivational aspects of an authentic audience and purpose for writing within a learning context that values all learners as writers (e.g. Calkins, 1986; Graves, 1983; Murray, 1968) as well as the multid imensional and recursive nature of the writing process 28 presented by Andrews and Smith (2011) are nested within the task environment section of the process level in the Hayes (2012) model. (This connection is based on adjusting the model to include motivat ion with the process level.) Finally, the knowledge telling and knowledge transforming strategies identified by Bereiter and Scardamalia (1987) and further detailed by Hayes (2011) are nested within the writing schemas section of the control level in the H ayes (2012) model. Moreover, given the data collected during my pilot work, as well as scholarship by other researchers, The Hayes (2012) model provides the best fit to answering my research question for five reasons. First, the model lends itself well to understanding the observable behaviors of screen - capture video. As children wrote, it was apparent how the visible behaviors on video could be writing before addin g the next word or phrase, it signaled a clear connection to the Resource used for problem solving, speaking, and decision making. Second, the model accounts for motivation and genre knowledge, as well as the physical task environment accounting for the technological elements. Students were clearly motivated to compose tweets and displayed specific genre knowledge. For example, while composing tweets students wer e familiar with adding hashtags and emojis to match the intended meaning of their message. Students were also familiar with the technology, using the iPad keyboard, using the automatic word selection, and fixing up their spelling based on red underlined wo rds. Third, the Hayes model accounts for the internal writing processes of composing. For example, the process level of the model identifies work of the transcriber and evaluator. Transcribing includes the act of writing/typing the text (which may cause gr eater difficulty for 29 children than adults) and evaluating includes the act of checking for accuracy during the act of Olinghouse (2015) I did observe some children clearly frustrated while transcribing. For instance, one child became clearly frustrated when she could not figure out, after multiple attempts, the correct spelling of a word underlined in red. Also, counter to Hayes and Olinghouse (2015), I observed chi ldren repeatedly rereading to both edit and revise their text as they composed their tweet. This may have been initiated by the 140 character limit imposed by the platform at the time of my observations, and may not be observed now that the character limit is character limit increased she replied, So I am liking the 280 characters because some of my students are able to explain more in depth. In fact I had one the other day I had a student that went over the 280 character limit and I had to show her how to edit her work. The editing was something that I forgot that I taught all the time with 140 character count. So in that aspect (teaching editing and refinement in wording) the 280 character is a downfall. (Personal email communication, January 23rd, 2018) Fourth, an important reason, though not tied directly to my pilot data, is how the model has shown alignment with the Common Core State Standards, CCSS (Hayes & Olinghouse, 2015). Teachers of writing typically follow a writing process approach characterized by planning, drafting, revising, editing, and publishing (Lacina & Silva, 2010). The Common Core State Standards, CCSS in writing also direct the use of pla nning, revising, editing, and rewriting. Recently, Hayes and Olinghouse (2015) compared the Hayes 2012 cognitive model of writing nted in the process level of the 2012 model (p.491). 30 Finally, other scholars have selected the Hayes (2012) model for similar reasons. Berdanier and Trellinger (2017), for example, developed a method to study screen - capture video of technologically mediat ed real - time writing processes using a modified version of the Hayes (2012) model. They made this determination during an initial open coding session while watching a segment of screen - capture video. Chapter three includes an explanation on adapting the Ha yes (2012) model for the current study. Chapter Summary To recap the text thus far, the literature on process - oriented theories and models of writing traces a tradition of s cholarship where researchers sat side - by - side with novice and expert writers, making sense of the how and why of what writers do. Since the 1990s, this tradition has given way to social and cultural perspectives, leaving much still to be learned about a wr in situ . Additionally, three patterns cut across the review of literature in the previous pages. First, three - phase writing proces s (prewriting, writing, and rewriting) to broader and more robust models that focus on the social and in - the - head processes of writing (e.g., Flower and Hayes, 1981; Bereiter and Scardamalia, 1987; Flower, 1994). Second, the scholarship has also evolved fr om the study of offline to online writing behaviors and processes. Taken together, the evolutions summarized in the previous two sentences represent an effort to develop a more complete and rounded understanding of what writers do when composing, outside o r inside the head, and regardless of medium. And third, newly emergent scholarship on writing is evolving 31 from the study of longer forms of writing (e.g., stories, essays, blogs) toward shorter forms of writing (e.g., summaries, instant messaging, texting) . Building on this previous scholarship, I aim to extend our understanding of what writers focus is on a short form of writing . By building primarily on the work of Takayoshi (2015), the study outlined in the next chapter extends the scholarship on writing by examining the writing processes and composition moves of young writers composing online using an unstudied short form genr e (i.e., Twitter). Finally, because existing models and theories of writing processes do not explicitly account for the "how" (composition moves) and "why" (writing processes) of what young writers do as they write for specific online spaces such as s ocial networking sites, I will use what I have identified as the most comprehensive model (Hayes, 2012) with modifications, as a theoretical lens to examine the writing processes and composition moves of children in an online space. 32 Chapter 3: Method In the following pages I outline the methods used for collecting and analyzing data on editing, awareness of audience). First, I describe the scholarly tradition that informs my research design. Then, I provide community, school, classroom, and participant information. Finally , I conclude with a detailed explanation of the data collection and analysis. Research Design Case study research has a well - articulated traditio n and has been used extensively for research about writing (e.g., Edwards - Groves, 2011; Ranker, 2007). Because the goal of this study was to better understand the writing processes and composition moves of individual children as they wrote in online spaces research question and method. Furthermore, case study design lends itself well to examine the 2). Case study res earch also permits in - depth description of a phenomenon like online writing. Using data collection tools that capture in - the - moment and in - depth writing processes and composition moves, a case study design is the most congruent with the aim of this study. Correspondingly, scholars who study writing processes in online spaces argue that the most appropriate data collection methods are those collected in the moment of composition via case study methods. Takayoshi (2016), for instance, explains, Particularly w ith research located closer to the act of composing, research that combines methods (for example, screen capture or eye tracking with think - aloud or retrospective verbal protocols) can move toward a fuller (yet always impartial) understanding of what write rs are doing and their decision - making processes p. 6. 33 To gather a data set rich enough for the type of understanding that Takayoshi describes, the current study design focused on case - een capture that included synchronous audio recorded talk - aloud protocol narration and video stimulated recall interviews that used proximal video excerpts from the screen capture as context for discussion. Setting and Participants Community The study was conducted in a community located in a small rural town in the Western generally familiar with small town 4th of July parades, September pari - mutuel horse betting races and county fairs with strong youth 4 - H representation. Hunting elk, deer, antelope and sage grouse is something many children talk about in the fall. Learning to swim on summer days in an outdoor warm spring or camping in the lush mountains thick wit h quaking aspen trees is also a familiar family experience in this area. Snowmobiling and sledding in the large open hillsides are common winter experiences. A major community event is the annual cowboy poetry gathering, with poets from around the country visiting. In addition to this strong western influence on the community, local ethnic group experiences from local Basque clubs and nearby Shoshone tribe reservations shape the life of children and teachers. Children in this small town generally travel a f ew hundred miles to the closest larger city. Beyond these influences the children of this small town are relatively sheltered from the larger world. School This school has three to four classrooms of each grade level, K - 5th grade with approximately 550 st udents total. According to 2014 - 2015 state demographic data this school is 34 84.7% white/Caucasian; 9.8 Hispanic/Latino; and 1.8% American Indian. 23% of students qualify for free or reduced lunch. Families generally work in agriculture, open - pit and undergr ound mining, and small business ownership. Levels of parent education range from graduate and professional degrees living within the mid to upper middle class to those without education beyond middle or high school living below poverty. It is not uncommon to have homeless students and very - well - to - do students in one classroom. Classroom Ten male, and eight female. There were no identified English Language Learners (ELL) a nd no students receiving additional academic support through special education. One student was being testing for special services and one student may be tested in the near future . Two students have speech Individual Education Plan (IEPs) with pull out ser vices provided. This classroom w as equipped with a cart of iPads which students used daily during center rotations and for tweeting. Four desktop computers with Microsoft software were available but rarely used. These devices had been replaced in popular ity by the iPads. The teacher reported, district has decided to not support the computers in my classroom anymore, so they are outdated personal communication, Oct. 2017). An interactive whiteboard perched prominently at the front of this classroom was used daily by teacher and students. These digital technologies were an integral part of the established classroom tool - kit, just as a penci l, notebook, or table might be. Mrs. Howe welcomed me as a researcher into her classroom to examine what happens when her students composed tweets. 35 Teacher The classroom teacher, Mrs. Howe, was in her 9th year of teaching. She described her 2003; Graves, 1980) usually involved a 30 to 45 - minute block of time with the first 10 to 15 minutes designated for direct instruction about a concept related to student needs and content standards. This instruction was then followed by students writing on their own as the teacher worked the room working with individuals and small g roups to provide additional instruction. The workshop usually ended with a time for students to share their writing during the last 10 minutes of the workshop. Keenly beyond their small community, Mr s. Howe adopted tweeting as a regular classroom practice to expand student awareness and knowledge of the world beyond their isolated community. This is her fourth year using Twitter in her classroom. Students The primary participants in this study were st - grade class. All students that wanted to participate in the data collection activity were provided that opportunity if parent consent had been granted. Cresswell & Poth (2018) recommend collecting data from a sample size that i s larger than the number of cases that will eventually be analyzed, so that sufficient data is generated to be adequately analyzed during the time frame. Selecting participants. For my sample to be selected purposefully, three criteria were drawn from the literature for the selection of 8 students, which served as in depth case studies: gender, academic level, and technology attitudes and dispositions. These criteria were differentiating features of North American elementary school classrooms where the po pulations of students generally include both male and female students with varying academic levels 36 (McGeown, Goodwin, Henderson, & Wright, 2012) and a range of experiences with technology (National Educational Technology Plan, 2016). Therefore, to obtain d ata on the widest possible range of writing processes and composition moves, the second graders in this study were selected with these criteria in mind. Gender. To obtain data on the widest possible range of writing processes and composition moves, the sam ple the study sample plan was initially intended to include both four male and four female students. These students were determined in an attempt to align gender pairs over degrees of academic level for possible cross - case analysis categories. When identif ying the four male students an error was made selecting one student who I assumed had a typical male name. It was not until after long into the study I noticed something was not right. Going back into the raw data and connecting the pseudonym with the stud realized the student I had listed as male was female. I searched my available case choices for a male student with a similar achievement level to accommodate my error and did not find a match. The remaining male cases were categorized as lower achieving and I needed a high to high - average achieving male. For this reason the final eight cases included three male students and five female students. Academic levels. To obtain data on the widest possible range of writ ing processes and composition moves, the sample included students representing various academic levels determined by the most recent results of the Measured Academic Progress (MAP). MAP is a nationally normed, state - required computer - adaptive test complete d two or three times per year in reading, language usage, math, and science. Specifically, scores from the Language Usage section were used to determine various academic level performance (Northwest Evaluation Association, 2013). Table 2 provides a list of the eight students selected to participate in the 37 current study. The students are listed starting with the student pairs who scored highest on the language usage measure, then progressively presenting the next highest - scoring pair, and so on until the low est - scoring pair is presented. Table 2 Study participants Student Male/Female Language Usage %tile range Hope Female 92 - 94 - 96 Hal Male 54 - 61 - 68 Inez Female 56 - 63 - 70 Irene Female 51 - 59 - 66 Kip Male 48 - 56 - 63 Kayla Female 48 - 56 - 63 Luke Male 27 - 33 - 40 Lori Female 14 - 19 - 24 Technology. Finally, to obtain data on the widest possible range of writing processes and composition moves the sample plan was to include students who indicate positive and negative dispositions with technology. Part one of the & Knezek,1992) was used to determine positive and negative dispositions. This five - part instrument was designed to measure attitudes and dispositions about technology use for 1st to 6th grade school children. Part one focused on computer enjoyment and computer importance, which was most relevant to the current st udy. Part one also fit within the time frame available to meet with students, including eleven 4 - point Likert scale questions about computer enjoyment and 38 importance. Survey results were similar across all students, indicating a positive disposition with technology. For this reason, technology was not used as a factor for sample selection. Consent Procedures Because I was working with children, I realiz ed they may feel some coercion to participate. However, there were two built in safeguards that addressed any possible coercion. First, data collection would come from the well established classroom practices and curriculum for tweeting which the teacher a nd students had been doing all school year. Second, writing of tweets was not graded by the teacher rather the information was used to communicate to a larger audience, including parents and teachers, what the children were learning and why they were learn ing it, as established for their class purpose for tweeting. In addition to parental consent to participate, the children were asked for their assent (verbal script is provided in appendix). Because all children in the 2nd grade class were welcome to parti cipate in the data collection session, the teacher did not have a knowledge of which case would be included in the study, minimizing coercion or undue influence. Students were (a) asked by their classroom teacher to bring home a letter of consent to their parents, or (b) parents would be invited to the classroom for a teacher - directed session explaining the consent form. The letter outlined the purpose of the study, the research activities in which their child would be involved , and the risks and benefits o f participation. Children were also asked to give their assent to participate at: (a) the same time their parents signed the consent and, (b) the time the first interview occurred. Because the participants were children ages 7 and 8, they were given an ass ent form that was a modified version of the consent form their parents received. Modifications (a) adjusted the formal language to more appropriate student - friendly language and (b) removed information of importance to parents but 39 not of relevance to the students. The assent script (See Appendix A) was read aloud to students before they completed any research related task. The classroom teacher read it aloud to students and I was available to answer any questions the classroom teacher could not answer. Onc e students listened to the language of the form read to them they were asked to sign the form, giving their assent to participate in the study. Before meeting with me for a talk - aloud - session students were verbally reminded that participation in the study was entirely voluntary and they could choose to stop participating for any reason , at any time with absolutely no consequence to them. I considered t he teacher m ight also feel some coercion to participate. I d id not know the children in the 2nd grade class but I was associated with the volunteering 2nd grade teacher. The teacher and I have been teacher/mentor colleagues since 2014. When the teacher decided to use Twitter in her classroom, she asked me for guidance . It ha d been almost four years since this original work. For the past few years before the study, discussion between the teacher and I ha d been infrequent, centered on questions the teacher ha d about what she c ould do to improve her teaching. I also consi dered t he teacher m ight feel obligated to let me conduct research in her classroom because of the help provided in past years. To minimize the possibility of the teacher feeling obligated or coerced, I made clear that I h ad other teachers that were willing to provide access to their classrooms for my research and that it wa s of no consequence if she would rather not have her students participate. Knowing there wa s an alternative classroom for my research removed undo feelings of obligation by the teacher. R ole of the Researcher I did not know the children in the 2nd grade class but was and still am professionally associated with Mrs. Howe. We have been teacher/mentor colleagues since 2014. Because I have 40 in a qualitative case study, our continuing relationship through this study may have had some bearing on the data collection and analysis (Creswell, 2009 p.177). Most notably, I, the researcher, spent considerable time both meeting with the students and observing in the results (Cresw ell, 2009). To ensure reliability, I took thick descriptive notes of what was happening and being discussed during each talk - aloud session and any necessary classroom observations as needed. To attend to validity, I employed 009) by asking the students and classroom teacher to review summary statements for accuracy. Data Collection In order to create a comprehensive and thick data set used to describe the writing processes and composition moves, in situ, five types of data wer e collected for the study: screen capture recordings, talk - aloud transcripts, video stimulated recall interview transcripts, student - written artifacts, and field notes. Table 3 provides a short description and purpose for each of the data collection tools . Table 3 Data Collection Tools: Description and Purpose Tool Description Purpose screen capture recordings A video recording of the computer screen showing actions made by the computer user (i.e., moving cursor, typing, changing keyboards, changing screens). These recordings provided a record of student composition moves while writing a tweet. From these data I can identify the temporal, quantity, and composing moves in situ. 41 Tool Description Purpose talk aloud transcripts A written account of all words spoken by both the writer and researcher during a writing session. These transcripts provided a record of student writing processes while writing a tweet. These writing processes were determined by what the student said while talking about what they were doing while writing a tweet. These data were necessary to writing processes in situ. video stimulated recall interview transcripts A written account of all words spoken by both the writer and researcher while the writer watched the screen capture recording of themselves while writing a tweet and the researcher asked the writer questions to better understand T hese transcripts provided a second record of student writing processes while writing a tweet based on watching their writing and talking about what they did and why they did a particular action. These data were necessary to better understand a ting processes in situ. student - written artifacts A screen capture image of the during our writing session together This published writing provided data about composition moves, specifically related to conventions and ele ments of text structure. These data were necessary to better understand a and convention knowledge, processes and composition moves. field notes notes taken during and after research observati ons These notes provided a space for thinking and making sense of the writing processes and composition moves observed in the data. These data were necessary to better understand a composition moves in situ. 42 A detailed explanation of how each type of data was collected follows. Screen Capture Data All screen capture technologies were set up before inviting students one at a time to the data collection session in a nearby classroom. The steps for setting up the technologies were: 1. Enable the do not disturb mode and lock the orientation on the personal iPad and iPhone. 2. Disable the automatic notifications on the laptop. 3. Connect all devices to the same wireless network. 4. Activate the Reflector Director app loaded on b oth my iPad and my laptop to connect devices. 5. iPad screen is now mirrored on the laptop screen. 6. Open the class Twitter account on the iPad. 7. Start the Camtasia 2 screenca pture in the background of the laptop to video and audio record student tweeting. 8. Plug in the external microphone. 9. Turn the iPhone memo recorder on for a backup audio recording. 10. Begin recording using the Camtasia screen capture after the student has gone through the think aloud practice protocol. To mirror the iPad screen onto my laptop screen, I use Reflector Director as the mirroring tool. Using a local network connection between my MacBook P ro laptop and my iPad, Reflector Director allowed mirroring in real time the iPad screen onto my laptop screen. For screen capturing (Takayoshi, 2016) I then used the screen capturing tool, Camtasia, to video record both the mirrored iPad screen and a thum bnail video image of the student as he/she was composing a tweet. Camtasia is a screen recording and video editing software developed by TechSmith ( https://www.techsmith.com/ ), which is recommended for technolo gy enhanced research (Cox, 2007) and used in similar talk - aloud research (e.g.,Coiro, 2006). Talk - Aloud Data The literature on using talk - alouds to understand the writing process is well established (e.g., Brandt, 1992; Hayes & Flower, 1980; 1981; Van Wei jen, 2009; Young, 2005). Using talk 43 aloud protocols with young children is less common (Coiro & Dobler, 2007; Seipel, Carlson, & Clinton, 2017; Pressley & Afflerbach,1995; Young, 2005). While Ericsson & Simon (1984) established relatively strict protocols for talk alouds, more recent uses locate talk aloud protocols located at the end of this continuum where strict control is used by the researcher. Careful applicati on of scripts and limited comments by the researcher allow for less task interference. At the other end of the continuum is a laissez - faire, open interaction between participant and researcher. The passive, monologic participant expected from Ericsson and Simon (1984) lends itself to particular participants, questions, and tasks, whereas the more dialogic, interactive participant recognized approach lends itself to others (Sibly & Watts, 2015). Given the participants, questions, and task used in this study, I designed a talk aloud protocol that was located in the middle of the continuum. The protocol included dialogic elements from speech communication theory to design a complementary approach, controlling some aspects of the protocols, such as maintaining a consistent set of questions for every interview (tight control), but only asking these questions if they were not initiated by the writer and asking them only at points in the session when it felt conversationally natural (lose control) (Boren & Ramey, 2 000). ures used are spelled out below. The procedures used for collecting talk - aloud data during the tweeting session began with explaining the task according to a script of keywords and phrases. This explanation included information about the screen capture se tup and other technologies in the room. What items were 44 called and how they would be used for the research were mentioned to address student initial session. Dur ing each talk - aloud session the student was guided to sit down at a designated table with an iPad tablet open to the class Twitter feed that had been set up with the needed screen mirroring application. We began the recorded session after the practice phas e (e.g., Pressley & Afflerbach, 1995). Once audio and video recordings were turned on we progressed through both the talk aloud script for composing a tweet and video stimulated recall conversation. While the script was visibly available to me, I used it as a guide rather than verbatim - script to create a natural and comfortable conversation - feeling with the students. I used keywords and phrases from this script rather than a word - for - word rendition (Boren & Ramey, 2000). See scripts below. Explaining the Task The first part of the protocol addresses student initial curiosities about the what, how, and why of our work together , intended to establish a safe and comfortable rapport and working environment for the student. thinking when they write in Twitter. To learn about this I need to know what you are thinking as you write. To know what you are thinking you will have to think really good notes about what you do and say. Instead of trying to write everything down I use an app that lets me video and audio record what you picture on a note card. Can you quickly draw a picture that would help you remember to talk about what you are thinking are done with yo ur tweet we will watch the recording and talk some more about what you were thinking 45 Show student the laptop and iPad and how they record and mirror the screen. Modeling a Talk Aloud This part of the protocol is a demon stration of what it might sound like for someone to share their thinking out loud while engaged in a task. talk aloud, or think aloud looks and sounds like. I will write a tweet and talk/ think As I start to compose a tweet I will say variations of the listed phrases: with @ mrshammer 2ndgrade ! #learn # think ] thinking that thinking I mixed Hammers 2ndGrade to me from looking think I will learn all about how second graders] think and learn and write and tweet. #schooliscool] I think that sounds good. I am thinking I am thinking I should reread to see if it makes sen I thinking thinking [Type delete #learn # think ] Student Pract ice This part of the protocol provided the student an opportunity to practice talking aloud with support. This support was gradually released as the student gained confidence and capacity for the task. so me think or talk Gradually release responsibility giving the student full control of the practice session. thinki ng 46 Student Composing a Tweet This part of the protocol reminds the student of his/her job as a thinker, talker and writer - taker. know your thinking as you tweet so be sure to talk about your thinking you to talk about your thinking . Sometimes I will point to the card to remind you to think Begin recording . Video Stimulated Recall Interview Data Video stimulated recall interviews were used as a follow up to talk - alouds to strengthen the validity of analysis done with the talk aloud data (Creswell, 2007; Koro - Ljungberg, Douglas, Therriault, Malcolm, & McNeill, 2013; Kuusela & Pallab 2000). Once the student published his/her tweet I paused and suggested we stand and stretch and walk to the drinking fountain or restroom if needed . Once back in the classroom I explained the next step of our work, the video stimulated recall interview. The laptop ran Camtasia once again, this time to capture the student talking about his/her video. The external microphone was also used . With the t alk aloud video queued to watch from the beginning, and Camtasia ready to go, I began by using phrases from the script below. thinking what you were thinking when you did something that I want to know 47 thinking we I will then began recording and started the video playback. At various moments in the thinking paused the video when the writer had paused from composing for Additionally, I prompted the child to signal when he/she would like to talk about a selection of the video allowing the child agency and ownership in the interv iew process. I ended the video stimulated recall interview Follow - up Prompts Based on conversations with early elementary students in previous years, I had generated prompts I was likely to pose for better understanding what the children were thinking and doing as they composed tweets. These prompts were similar to those used as follow - up prompts during the talk aloud protocol and the video stimulated recall interview s. These prompts are organized into three categories: (a) general statements or questions for example, know (b) related to transcrip tion from card to tweet, awareness of clear message, audience, and purpose for example, 48 were thinking (c) related to specific and unique elements of the social networking site or digital affordances, for example, to me about the red underline, what you know know Student - Written Artifact Data I expected to collect two written artifacts: (a) a student - generated handwritten draft of text they planned to use as their tweet text, and (b) a screenshot of the coinciding completed tweet published during the talk aloud session. Screenshots were collected using Ma cBook pro screenshot capabilities (Shift + CTL + 4). Images were automatically downloaded to the MacBook desktop and then relocated to a secure location in the research files. At the time of the data collection, (the last weeks of the school year) the teac her had not been asking her students to write their tweet text draft on a notecard before tweeting. This had become a practice she abandoned in the fall of the school year. To maintain business as usual with students and their current tweeting practices, I did not collect handwritten drafts of tweet text. 49 Field Notes Data As I collected and analyzed the data I kept field notes. Following recommendations from Saldaña (2016), these notes were generated in the moment of the tweet session (i.e., task environment elements beyond the keyboard such as environmental print the stude nt may have glanced at) as well as after the session while reflecting on screen capture recordings coupled with transcripts of the talk - aloud protocols and transcripts of the video stimulated recall interviews. As explained in the previous section, the rea son this type of data was import for this study had to do with making better sense of the transcript and video data. To illustrate, Figure 10 provides an example of the field notes collected while reading transcripts and watching screen capture video. Figu re 10 Field Notes Example The field note excerpt above, for example, helped me identify a composition move pattern. I also created a document for sorting these notes to identify patterns based on my selected theoretical framework writing processes and co mposition moves categories. , see Appendix B for example. 50 Data Analysis The analysis of data focused on the writing processes and composition moves from the generation of an initial idea through the transcription onto an iPad and onto the Internet as a published tweet. Rather than use a grounded theory approach (because a previous theory or framework is not available for the interpretive work of analysis), this study used a semi - grounded theory approach for data analysis (Corbin & Strauss, 2008; Glas er & Strauss, 1967). As such, the interpretive work of data analysis was initially informed by a theoretical frame, but remained open to new patterns that could extend or refine the existing theoretical framework. Specifically, a modified version of the Ha yes (2012) writing processes model (etic codes) and other noticings that did not seem to fit the framework (emic codes) were used to develop a typology of different writing processes and composition moves. This approach is appropriate given my research que stion is both open - ended (allowing for emic codes) and related to a well established body of research, providing etic codes. Adapting the Hayes (2012) Model Seven criteria were used to adapt the Hayes (2012) model to study the online writing processes an d composition moves of children. The first three criteria were drawn from Berdanier writing processes. The criteria indicate that a model needs to capture the writin unique task requirements, (b) technological aspects, and (c) behaviors that are observable. The 1 on adapting the Hayes criteria indicate that a model needs to capture 1 Hayes and Olinghouse (2015) suggest careful consideration of four areas of the Hayes (2012) model when adapting for children: (a) transcription, or acts of handwriting/typing, spelling, punctuation and capital letter usage require greater effort from chil dren (b) writing schemas, or strategies used by children to construct text are typically knowledge - telling rather than the more complex knowledge transforming (Bereiter and Scardamalia, 1987) (c) planning, and (d) revising, young children are less likely t o employ advanced planning strategies or to revise their text. 51 revising. These seven criteria were then used as a guide to adapt each level of the Hayes (2012) model: the contro l, process , and resource levels. Because each level includes multiple elements associated with writing, adaptations are presented by level. The Control Level . Hayes originally represented this level by teasing out four elements associated with shaping and directing the writing activity: Motivation: key to the writing process Goal Setting: based on what the writer want to achieve The Current Plan: includes sub - goals to do the written work, and in memory otherwise it becomes a written plan found in the process level Writing Schemas: strategies to produce text, and genre knowledge al element of the current plan was deleted because of the observable behavior criteria (c). Also, a detailed description of the writing schemas elements was necessary to more accurately code these elements. This description is provided in the section follo wing Table 4. The Process Level . Hayes originally represented this level by first separating the internal writing processes and external factors that influence them. He delineated four basic internal writing processes: Proposer: generates ideas influenced by the external environment, personal experiences, collaborators, goals, etc. 52 Translator: forms the non - verbal ideas into verbal representation (for adults, this is often the source for detering writing fluency) Transcriber: forms the verbal ideas into wr itten text, applying spelling, handwriting/typing, capital letters, punctuation (for children, this is often the source for detering writing fluency) Evaluator: checks for accuracy at any stage of writing - rethinking ideas, selecting new words, fixing up what has been written down The external factors were partitioned into environment and complex activities. The two environmental activities were: Physical task environment: technology tools, assignment materials, text written so far Social task environment: input at the time of writing - collaborators, verbal comments, distractions in the room that may limit working memory The two complex activities, each with their own writing schema, employed all or some of the basic processes: Planning: in - advance plannin - - the - moments of composing (the evaluator may not engage entirely) Revising: a complex activity with the evaluator front and center dapted to proposer and translator were deleted because of the observable behavior criteria (c). The Resource Level . Finally, Hayes originally represented this l evel by fleshing out four ways that writing can be used for problem solving, speaking, and decision making: 53 Reading: editing and revision use repeated reading of text, purpose for reading based on task goal Attention: aka executive function or executive c ontrol, self - regulation, staying focused among distractions Long - Term Memory: knowledge of facts, events, schemas, information about a topic Working Memory: used to accomplish the current task, short term memory system with two sections a) verbal, and b) visual or spatial - term and working memory were deleted because of the observable behavior criteria (c). 54 Table 4 using the Hayes 2012 Cognitive Writing Processes model. This codebook is a modified version of the codebook by Berdanier and Trellinger (2017). Level Definition of Level Sublevel Code Definitions Writing Processes Monitoring of process overall Motivation Motivation Does the child enjoy this writing? Goal setting Goal(s) What does the writer want to achieve? Planning In - advanced planning What planning does the writer do before tweeting? In - the - moment planning What panning does the writer do while tweeting? Writing schemas Genre knowledge fundamental elements 280 character short - form writing, and teacher - assigned student signature Genre knowledge secondary elements photo(s) and a hashtag. Emojis can also be considered secondary elements but are not necessary Genre knowledge contextual elements followers, which can also be considered the audience, likes, cybersafety, and networking. Text structure text structure classifications include both the teacher - assigned text structure and the three child - created expository text structures identified by Hayes (2011). Control Level 55 d) Level Definition of Level Sublevel Code Definitions Composition Moves Divided into internal and external processes involved in the process of writing text Composing Processes (Transcription component skills) Typing emoji Adds emoji to message Typing single letter Types one letter followed by keyboard search < 2 sec. Typing multiple letters Types many letters without stopping, > 2 sec between letters Typing spacebar Types the spacebar to move forward space Typing backspace Types the delete button to move back a space Typing punctuation Types a form of punctuation or special character Typing hashtag Types a hashtag Planning Planning what to write while writing Revision Processes (Evaluating) Addition of new text Identification of area needing detail Editing Local editing: word choice, grammar, spelling Rewriting Rewrite a sentence from scratch Process Level 56 Revising Alter sentence to add value without rewriting the sentence Reorganizing Moving text around, including copy and paste Deleting Deleting text from document without replacing Task Environment (external environment) Camera Taking or accessing photo(s) Tweet app Opening space to work on iPad and submitting a tweet Keyboard search Pausing to look for letter/item on the keyboard(s) Keyboard open or close Changing, disappearing and reappearing keyboard Keyboard autocomplete Selecting word suggestion at top of the keyboard Cursor reposition Touching screen to position cursor, including magnifying glass Scrolling Swiping the screen to view content below or above the working area Task materials Looking at the iPad screen (student is not searching the keyboard or reading text.) looking at external sources around the room (e.g., wall posters, classroom objects associated with a learning task) Already accessed photo Looking at the image captured before meeting with me 57 Writing prompt Looking at the writing prompt reminder page Internal memories and general purpose processes that processes at the other level can call on Resource Level Reading Reading followed by a continuation of writing or deciding to submit a tweet Attention diverted Attending to something other than the tweeting task Pausing Cannot observe or identify from the interview the reason for pause Resource Level 58 Writing Schemas classification of genre knowledge. Writing schema includes knowledge of both genre elements and text structure. To determine tweet Genre Knowledge I used the sed other genre elements mentioned and or used by students during their tweeting and our discussions. Based on this information, tweet genre knowledge for this study includes three categories of increasing knowledge. First, knowledge of fundamental element s: 280 character short - form writing, and teacher - assigned student signature. Next, knowledge of secondary elements: photo(s) and a hashtag. Emojis can also be considered genre knowledge . The third category includes contextual elements of Twitter: followers , which can also be considered the audience, likes, cybersafety, and networking . These contextual elements are accounted for through conversation with the student and observable behaviors. K nowledge of each contextual element is accounted for when the student communicates the following: followers or audience as a known or unknown other that can read their tweet; Likes as a form of communication among users; cyber safety in the way of followin g safety lessons taught by their teacher (e.g., measures to maintain anonymity); and networking as reading and talking about what others have posted to the class Twitter page. 19 The Step - by - Step Guide to Go from Novice to Expert in Any Skill, based on the work of Dreyfus and Dreyfus (1980) I constructed a flowchart for 2nd - grade tweeters genre knowledge skill levels. See figure 11. 59 Figure 11 Flowchart for 2nd - grade tweeters genre knowledge skill levels *note: error in flowchart (240 characters will be changed to 280 characters) 60 Writing Schemas classification of Text Structure. To determine student text structure classifications I used both the teacher - assigned text structu re and the three child - created expository text structures identified by Hayes (2011). Teacher - assigned structure : The taught text structure includes a statement about what the student is learning and a second statement about why this learning is important and/or, how this learning will be helpful. Another way to classify text structure is by the three child - created expository text structures identified by Hayes. These three structures are listed by increased level of sophistication. First, flexible focus, is a stream of consciousness writing with no coordinating theme and the only writer - evaluation to text is to check for a sufficient amount of writing. Next, is the fixed topic structure. Every sentence is about the topic and the quality of output, (e.g., s pelling, capital letters, word choice) is evaluated by the writer. Finally, in the topic - elaboration (Hayes & Berninger, p. 13, 2014). Writing P rocesses and Composition Moves Analyses Analyses were conducted to identify the writing processes (thoughts made known through talk - alouds and video - stimulated recall), composition moves (actions recorded by the screen - capture video and further understood by talk - alouds), and emergent features (thoughts and actions from all data sources) of eight second - grade tweeters. Both individual case descriptions and cross - case comparisons were conducted. Writing processes analysis was informed by the adapted Contr ol Level elements in - aloud and video - stimulated recall data were analyzed in terms of motivation , goal setting , planning , and writing schemas . 61 Composition moves analysis was info rmed by the adapted Process and Resource Level mergent features were also accounted for based on what came to light during the data analysis, which were unexpected and noteworthy, extending beyond the a priori analyses of writing processes and composition moves . For the composition moves talk - aloud and screen - capture video data were analyzed at two levels: the letter/word/phrase level and the complete tweet level. The letter/word/phrase level analysis was informed by an approach developed by scholars who use miscue analysis to understand reading moves by young readers (e.g., Goodman, 1973; Perl, 1979). As a result, the letter, word and phrase analysis captured every asp ect of the case real - time composition moves, which included all the letters, words and phrases composed, deleted, and revised, whether by typing or the iPad making tools available (like the autocomplete and speech - to - text tools). The complete tweet level was further divided into two sections: complete tweet overall and c omplete tweet in thirds . The complete tweet overall analysis represents second - by - second, real - time composition moves over the course of composing a tweet. These data were displayed as a t imeline informed by the work of Berdanier and Trellinger (2017). Complete tweet overall data were also categorized by levels of use or percentage of occurrences overall to investigate which of the four categories ( composing, revision processes, task envir onment, and resource level ) were most frequently coded. To generate a quick way to categorize this information I counted the total number of occurrences across all 27 coded areas of the four categories From that total, I counted total occurrences for each of the four categories and divided that by the total overall composition moves occurrences to provide a percentage of use. For sorting purposes, categories with 0 to 25% of the total occurrences were ranked low. Categories with 26% to 74% of the total 62 occu rrences were ranked moderate. Categories with 75% to 100% of the total occurrences were ranked high. Finally, the complete tweet analysis in thirds composition moves were divided into three time periods (beginning, middle, and final). Dividing the timeline into three equal time periods provided a clustered temporal view of the four composition moves categories. A summary of the occurrence of use in thirds by time was represented as both a table and histogram. E mergent features analysis was informed by wha t came to light during the data analysis, which were unexpected and noteworthy, extending beyond the a priori analyses of writing processes and composition moves engagement with the autocomplete feature. like different spelling, so, if you're trying to spell a word you could see it up there, if it's up there, Ethical Considerations The research methods used for this case study are designed to protec t the rights and welfare of research participants. The following sections explain the awareness and management plan for participant potential risks and benefits, privacy, and bias. Potential Risks and Benefits There were two possible minor risks associated with participation in this study. The first is fatigue from the 45 - minute interviews. The second is missing class time while participating in the study. Safeguards were employed to protect against these risks. To prevent participation fatigue, data collection sessions were limited to a maximum 45 - minute talk - aloud session with a 63 short break to stretch, walk around, have a snack, and use the restroom after the first 20 or 25 minutes. To prevent loss of class tim e, student participants were not held inside the classroom for data collection purposes during any regularly scheduled recess or lunch break. I worked closely with the cooperating teacher to plan data collection sessions around essential classroom instruct ion to avoid students missing having make - up work due to their absence from class. There were several potential benefits for students participating in this study. For example, students may have learned something about online writing and about the scientif ic process of understanding of online writing of young children. Long term benefits may be that this research will contribute to the development of new literacie s writing curriculum for young children. Privacy classroom were present during the consent process. The cooperating teacher was asked to collect the permission slips a s they were returned to school. These forms were sealed in an envelope and sent home with the letter. The teacher was given a folder in which to collect the permission forms. She kept the forms in a locked filing cabinet in the school office until I transp orted them to a secure location on campus. Additional privacy measures included using pseudonyms and participant numbers rather than names in association with coded data. Only the principal and secondary investigator had access to the list of participant names, pseudonyms, and corresponding numbers. This were password protected. These data were initially password protected and stored on a secure 64 laptop. Fina office. These data will be stored for ten years after publication and then destroyed. Finally, the content of each tweet was published within a secure password - protected c lass Twitter account. Only those Twitter followers the teacher has allowed to follow her class account can see tweets composed during the data - collection talk - aloud session. Only teacher - assigned numbers for each student are used in the tweets sent from th is account. Use of pseudonyms will be used in publications and presentations when tweet content is reported. The classroom teacher initiated anonymity of tweets at the beginning of the school year. Each student was given a random three - digit number in plac e of using his/her name in tweets. Parents have been given the three - digit number associated with their child only. If for some reason a - digit code, I instituted a second authentication factor to ensure anonymity and confidentiality. To do so, I asked the child to include in his/her tweet a new numeric code I assigned. Also, I kept notes on the content of each student's tweet, indicating if the content includes information unique to that student. E xamples of those content notes could be: the book they are reading or a topic they are writing about for a class assignment. These tweets c ould be removed from the pool of case studies if necessary. Based on a recent scan of the existing class Twitter fee d, most tweets are about general above actions became necessary. Bi as In conducting this study, I brought a number of assumptions about writing, technology and children that influenced how the research was carried out. These assumptions have grown 65 out of my work as a (a) teacher of writing for 4th grade students, (b) regi onal coordinator for teacher professional development in the teaching of writing, and (c) scholar who reviews and synthesizes research on writing. Specifically , I was influenced by the following : Writing is a meaning making process (Nystrand, 2006). An author writes to present an intended message for a reader. This message must make sense and convey meaning for the reader to comprehend. New technologies redefine what it means to be a writer (Leu, 2016). Generally, writing has traditionally been defined as words on paper. Using the technologies of today, writing may include multiple modes to convey meaning within one text. For example, in addition to words, a writer may include moving images, audio, and hy perlinks to help convey meaning. Writing is a context specific social act (Shaughnessy, 1977). Every instance of writing is embedded in a particular setting and imbued with distinct technologies. For instance, unique to the micro - blogging within the settin g of Twitter, this writing may involve re - - character constraint. Just as online reading requires new skills, strategies and dispositions (Coiro & Dobler, 2007), the same can be said of a three - dimensional space of Internet links while comprehending a message requires skills not necessary for comprehending a paper text. Similarly , when writing within an online space, generating text appropriate for a social media audience and purpose (e.g. hashtags, usernames, embedded URLs) requires knowledge unique to this online writing experience. 66 The relationship between technological development and literacy development are - e adopt new ways of communicating according to available technologies. For example, the advent of communication. Children do not develop as writers in predictable and linear progression toward sophistication. Rather, complex and beginning skills develop in tandem (Andrews & Smith, 2011). For example, learning how to construct a complete sentence (beginning skill) wh ile also learning about the importance of communicating an intended message for a specific audience (more advanced skill) can develop in tandem. By articulating these assumptions, I aimed to be mindful of their potential influence on my data collection and analysis. Without this awareness, confirmation bias (Wason, 1968), was more likely to shape my interpretations so they align with my existing as sumptions. To be sure, 2017). Thus, I consulted the Qualitative Legitimation Model as a self - checking precaution (Onwuegbuzie & Leech, 2007). 67 Chapter 4: R esults - grounded theory analysis was conducted to identify the writing processes (thoughts made known through talk - alouds and video - stimulated recall), composition moves (actions recorded by the screen - capture video and further understood by talk - alouds), and emergent features (thoughts and actions from all data sources) of eight second - grade tweeters. Thi s chapter presents the results of my analysis in three sections: (a) individual - case descriptions, (b) cross - case comparisons, and (c) overall summary. The results in the first section are presented by case, starting with the student pairs who scored highe st on the language usage measure, then progressively presenting the next highest - scoring pair, and so on until the lowest - scoring pair is presented. The results in the second section are presented as cross - case comparisons which identify the differences an d similarities between the eight cases. The final section presents an overall summary of the data analysis in response to the research question. Individual - Case Descriptions In this first section, I present each of the eight cases in five parts. First, I p resent a general description of the student based on observational impressions recorded while working writing processes analysis which is informed by the Control Level gnitive Writing Processes model. composition moves analysis which is informed by the Process and Resource Level Fourth, I present the emergent features that came to light during the data analysis, which were unexpected and noteworthy, extending beyond the a priori analyses of writing processes and 68 composition moves . Finally, I present a case summary , synthesizing the individual - case analysis so a typology can be designed for use in the cross - case comparisons of section two. Hal: High Scoring Male General Description Hal appeared indifferent about meeting with me, neither excited or reluctant. He was, however, curious about the technology set up for data Twitter feed. When I asked questions, although attentive, Hal sometimes yawned and periodically looked out the windo w into the hallway outside our meeting room. The analysis of his writing processes and composition moves, however, suggest a more inspired engagement with tweeting. Writing Processes - aloud and video - stimulated recall data were analyzed in terms of motivation , goal setting , planning , and writing schemas . Writing Processes: Motivation motivation to tweet appeared positive. He seemed motivated by the joy of tweeting and the efficient completion of it by using technological tools. For example, while tweeting, he often smiled after saying the words of a message toward the iPad microphone, then waiting for his words to appear o n the screen. His delight may have been more about using the speech - to - text tool than about actually tweeting. But when I asked if he used the speech - to - text tool in class, he bashfully one. When asked if he 69 Ironically, the data showed that Hal chose to use the speech - to - text tool more often than his typing skills. Writing Processes: Goal Setti ng Hal expressed two goals for tweeting: developing technical skills and creating meaningful messages. When asked about his first goal , he talked about how becoming a better to do on your computer you can like type it fast. So you get it done and you don't have to do it like after that the purpose of tweeting is to communicate with others . This intent was most visibly expressed in his awareness of the teacher - established tweeting guidelines. He could recite them Writing Processes: Planning planning included both in - advance and in - the - moment planning. In both cases, he Evid in - advance planning occurred when he selected an object and took a photo of it to go with his tweet, using the object and photo as tools for advanced planning . When asked then we take a picture of in - the - moment planning appeared when he added a second sentence to a tweet and struggled to generate an idea. My suggestion to look at his photo helped as he quickly started generating text (using the speech - to - text feature) that giraffes live in rainforests, something he learned from reading a book. 70 Writing Processes: Writing Schemas ree aspects of writing schemas: genre knowledge , contextual elements , and text structure . Writing schemas: genre knowledge. For starters, Hal displayed knowledge about the fundamental and secondary elements of the Twitter genre: 280 character limit, stude nt signature, photo, hashtag, and emoji. When he wanted to include an emoji in a tweet, but could not find one to match the meaning of the text, he decided that emoji - use would not enhance his message. Conversely, when the hashtag and photo Hal included in his tweet matched the meaning of the text, with little explanation necessary, he assumed that the audience would understand their the giraffe was built with Legos, which was something not stated in the text. Interestingly, Hal did not include an explanation of the other items in the photo and how they may or may not be related to the Lego building. Writing schemas: contextual elements. Hal displayed knowledge of f our contextual elements as he tweeted: followers , likes , cybersafety , and networking . For instance, without followers reads his tweets. When asked likes cybersafety protocols, Hal acted in ways consistent with the protocols established by his teacher. And when it came to the contextual element of networking tweets. For example, after publishing his tweet, Hal quickly started scrolling through the class twitter feed to see w hat others had posted. He paused at one post showing the cover of a book 71 and questioned who might have posted the tweet based on how it was written. Writing sche mas: text structure. fixed topic text structure . His topic was the giraffe. Although his second sentence is about giraffes it does not extend his initial statement about building a giraffe. Following the teach er - established guidelines to tweet about what he is learning and why or how this learning would be helpful, Hal did not explicitly say that he was learning about a STEM - related concept. Instead, he described what was in his photo, again assuming his reader would understand how building an object with Legos will help him learn about STEM - related concepts. Composition Moves - aloud and screen - capture video data were analyzed at two levels: the letter/word/phrase level and the complete tweet level. Composition Moves: Letter, Word and Phrase Level Analysis. Based on the screen - capture and talk - phrases was informed by an approach developed by s cholars who use miscue analysis to understand reading moves by young readers (e.g., Goodman, 1973) and the composing style sheets created by Sondra Perl (1979). As a result, the letter, word and phrase analysis (see Figure s flow - of - composing which included all the letters, words and phrases he composed, deleted, and revised, whether by his typing or the iPad making tools available (like the autocomplete and speech - to - text tools). To illustrate, Figure 12 represents his revi sing with blue delete marks. And the points where Hal reread are shown as vertical purple 72 Figure 12 - of - composing ned in red and deleted started to say I would help him. But he inter used. I asked whether he had used it before. He nodded no . I asked if he wanted to try it. He nodded yes . Like an experienced expert with the technology, Hal pressed the microphone button, 73 actions suggested that using the microphone was not new for him. Later he admitted he was not t home. while saying aloud, noticed it in the autocomplete bar. Hal then typed autocomplete bar. He paused a moment, rubbed his eye and without hesitating used the on top of his cupped hand, then wait ed for the word to appear. Once the word popped up (with a putting the cursor back next to the period. He then spoke under his breath, while pressing the up - me for permission to commence typing the standard closing to his tweet. At this point, two minutes and 15 seconds had passed since Hal started drafting his tweet. b ackspaced adding his assigned Twitter number, 531. Again, without adding a space he typed the appeared in his tweet text box he sat back in his chair and announced he was done. The process to this point took Hal two minutes and 44 seconds. 74 The procedures for tweeting in this classroom prompted students to check with their teacher before publishing their tweet. As Hal sat back in his chair, swinging his feet, I asked, oulders and said, as Hal started deleting his hashtag text. He deleted more than intended so he retyped his arted scrolling to find a giraffe. He pressed the icon for animal emojis and scrolled left to view the available animal emojis. He did not see a giraffe so he closed the emoji keyboard, opened the alphabet keyboard and then the character keyboard. He typed the hashtag character and used the if he added his picture. Being reminded, he quickly accessed the photos and selected his previously captured image. I asked i suggesting his forgetting of the picture, in this case, could have been created by the interview context. Returning to what I knew to be the classroom procedures for tweeting, I wanted to learn more abou you have any questions or do you want me to check anything before you send it of 75 sounding the word out. I asked if he wanted me to check any other parts of his work and he replied by asking if he needed to put a period anywhere. It looked like he was thinking through the expectations of checking his work established by his teacher. Confirming his punctuation I shook his head and said he would make another sentence. He started deleting the we do that [the standard closing that includes the assigned Twitter number and a self - selected At this point, I prompted a lot of kiddos say they look at the picture and that helps them get ideas. Do you want to look at and he said that he didn At this point, he had been working on his tweet for 8 minutes and 12 seconds. 76 As I look ed over the real - time text generation letter/word/phrase - tweet I noticed frequent changes in keyboards. He easily switched back and forth from the letter to numeric keyboards. I also noticed he moved to the special character keyboar d to access the hashtag rather than returning to the letter keyboard where the hashtag is more easily accessed. Furthermore, Hal closed his keyboard to exit the emoji keyboard, then reopened the keyboard to access letters. Taken together, these moves sugge st Hal made use of inefficient keyboard routines. composition moves include both efficient and inefficient technology use. Working to maximize efficiency, more than half of accompany Twitter (aspects of the task environment). For instance, Hal used autocomplete for four of the five consecutive words in his first sentence. He then used speech - to - text to generate words he was not sure how to spell and his entire second sentence. Inefficient technology use is found in less prominent patterns with text deletion and keyboard switching. For instance, Hal deleted entire sections of text rather than repositioning his cursor t o preserve his valediction text access needed functions. Composition Moves: Complete Tweet Analysis Based on the screen - lete tweet was informed by a model of composition moves developed by Hayes (2012) and adapted by Berdanier and Trellinger (2017). The model provides a systematic way to code the salient features of real - time composition moves. Based on a careful review of the literature and pilot work, the codebook 77 developed for this analysis made use of four composition move elements: composing , r evision processes , task environment , and resource level . As can be seen in Figure 13, the complete tweet analysis used a time line to represent - by - second, real - time composition moves over the course of composing his tweet. Moving from left to right for each code, a colored bar represents the duration of a composition move element, while the white space along the time line (i.e., which is not occupied by a colored bar) indicates the duration of a talk - aloud moment. The green bars indicate the observed duration of a composing move, which included eight subcategories of composing: typing emoji, typing single letter, typi ng multiple letters, typing spacebar, typing backspace, typing punctuation, typing hashtags, and in - the - moment planning. The yellow bars indicate the observed duration of revision processes moves, which included six subcategories: addition of new text, ed iting, rewriting, revising, reorganizing, and deleting. The red bars indicate the observed duration of task environment moves, which included ten subcategories: camera, tweet app beginning and end, keyboard searching, keyboard open or close, keyboard autocomplete, cursor position/magnifying glass, scrolling, task materials, already accessed photo, and writing prompt. Fina lly, the blue bars indicate the observed duration of resource level moves, which included three subcategories: reading text, attention diverted, and pause. Taken together, this view of the . 78 Figure 13 Complete tweet analysis: Overall. As the composing section of Figure13 indicates (colored green), the most frequently occurring composition moves observed were typing multiple letters typing spacebar after typing a word (n=11). These frequencies indicate there were twelve instances when Hal typed multiple letters without stopping and eleven instances when Hall tapped the spacebar. The appearance of both happened quickly as if they were automatic composition moves for Hal. In addition, his typing backspace occurred five times (to reposition the cursor) and his typing single letter o ccurred four times (three times to activate the included three occurrences of typing punctuation and three of typing hashtags , but after revisions, his final tw eet included only two punctuation marks and one hashtag. Finally, planning 79 in - the - moment was observed when Hal spontaneously added a sentence about giraffes living in the rainforest to the end of his tweet. Although Hal considered it, typing emoji was not evidenced in his composition moves. The revision processes section of Figure 13 (colored yellow) shows only one instance each of the editing , revising , and deleting composition moves in the tweet. The editing move occurred when correcting the spelling of the first word added to his tweet. Revising occurred when Hal noticed he could add an emoji to the tweet. And deleting occurred when the hashtag text used to search for a giraffe emoji yielded nothing, Hal decided to leave an emoji out of his tweet, and th itself. The other three revision processes codes ( add new text , rewriting , reorganizing ) were not evidenced in his composition moves. For the task environment section of Fig ure 13 (colored red), the most frequently occurring composition moves were keeping the keyboard open or close to access the microphone and keyboards (n=10), and using task materials Twitter app, such as the speech - to - text microphone, which Hal first used to spell the unknown (n=14). In addition, he engaged in keyboard searching to find needed letters seven times and used the keyboard autocomplete tool five times. Hal accessed the tweet app beginning and end twice, first when he started tweeting and again when he submitted his tweet. Finally, Hal accessed the camera to add his previously taken photo to the tweet, e ngaged in scrolling for an emoji, looked at his already accessed photo and had a hand in repositioning his cursor one time each. Writing prompt was the only task environment code not evidenced in his composition moves. 80 Finally, the resource level section of Figure 13 (colored blue), shows there were three occurrences of rereading text , twice during the first few minutes of his composing and again before submitting to publish. He had his attention diverted only once, it was around the midpoint of his compos ing time. He noticed another student walking through the hallway and stopped to watch him. Hal quickly returned to his composing work. Five pauses were recorded, once before the tweeting started, twice toward the end of tweeting and twice around the midpoi nt of his composing, all of which appeared to be moments of thinking. Complete tweet analysis: In thirds. time periods (beginning, middle, and final) as shown in Table 5 and Figure 14. Dividing the timeline into three equal time periods provides a clustered temporal view of the four composition moves: composing, revision, task environment, and resource level . Table 5 - of - use in thirds Beginning Third % Middle Thir d % Final Third % Total % Composing 32 2 9 43 Revision 1 2 0 3 Task Environment 22 11 11 44 Resource level 5 2 3 10 Total % 60 17 23 100 81 Figure 14 focused on getting something written down. He concentrated most of his composing within the beginning third of his timeline, quickly typing his message as if following a well - practiced protocol for composing a tweet about what he has learned and how it wil l help him. During the beginning third task environment elements were less frequently accessed. Also, during this time Hal did a lot of keyboard searching and used the keyboard autocomplete . Revision elements and resource level elements were of low use, ed iting once, reading twice and pausing once during the beginning third. down using the task materials for speech - to - text and keyboard autocomplete features. Hal also frequen tly used the keyboard open and closed during this time, aligning with when Hal decided to add an emoji , retyping his ending text which included letters, numbers, and the hashtag, each accessed from its own keyboard. It makes sense that the largest amount o f revision occurred 82 to be manipulated and during his final third the text was generated using speech - to - text. For this reason, he may not have considered the tex t as something to revise. After pausing once, Hal had his attention diverted during this time. multi - letter and single letter typing and punctuation as Hal finished his message. This time also included looking at his already accessed photo when thinking of what to add to his tweet, then generating a final sentence using the task materials speech - to - text feature. Hal ended his composition moves reading once before publishing his tweet and pausing twice. Emergent Features Based on all the data gathered from Hal, three additional elements were identified in - composing technology. These elements were categorized as emergent findings. Dur ing our conversations, Hal seemed confident in explaining different aspects of the technology used to compose his tweet. Even when it was apparent that he did not know specifics of a particular feature he confidently created an explanation. Two examples ar e reading URLs and magnifying glass activation the technology as knowing - other . Reading URLs After publishing his tweet, Hal scrolled through the class Twitter feed stopping to read what his cla ssmates posted. While scrolling he noticed a tweet from a teacher that his class follows. Reading the tweet aloud, he sounded out all the letters in the embedded URL as if think 83 students need to learn related to genre knowledge. He believed the URL represented important words. Magnifying Glass Activation and I coul the red underline indicates a misspelled word I asked him what he does to help himself with these words. He said he keeps on tweeting his ideas then goes back to fix th e words. Then he could do this, [he presses down on the misspelled word, and nothing happens] oh, probably period then pressed down on the word again, and the magnifying glass bubble appeared. The magnifying glass bubble allowed Hal to see his word up close and position his cursor within the letters to make corrections. His misconception that a punctuation mark like a period was needed to activate the magnifying glass bubble illustrated that his understanding of the feature was not fully developed. Technology as a Knowing - Other this language when talking a 84 s his approach to create a concrete explanation to an abstract concept or anthropomorphising technology (Bernstein & Crowley, 2008; Heider & Simmel, 1944). Sum mary of Hal -- Overall The data interpreted for Hal resulted in a profile that indicates he is a resourceful and efficient, just - get - it - writing processes he seemed motivated by the enjoyment of t he task and the efficiency of the technology. His tweeting goals included developing technical skills and writing a coherent message. Hal planned his tweet content before writing and required some support for in - the - moment plans while writing. He used ge nre knowledge to compose short - form writing including a photo and hashtag within a fixed topic text structure and showed a growing understanding of followers, likes, cybersafety, and networking composition moves were dominated by composing (transcri ption component skills) and task environment components associated with the keyboard and task materials. Hal navigated the iPad keyboard quickly and confidently, typing multiple letters in a word quickly. He edited misspelled words but did not edit the cap ital letter miscues. He revised his text to enhance meaning with an emoji and used the autocomplete tool often. that were later retyped rather than repositionin g his cursor to change his text. Finally, Hal has an emerging understanding of the iPad keyboard affordances inventing explanations to how and why things work without fully understanding the technology. 85 Hope: High Scoring Female General Description body language gave the appearance of comfort, confidence, and a down - to - business seriousness while working with me. For instance, when getting started, she positioned herself on the chair next to me, lounging back a bit, her left elbow resting on the back of her chair. Eventually, Hope dropped her elbow off the seatback to rest her left hand in her lap. When asked about her thinking, she smiled as she provided clear and detailed explanations. Her high comfort - level at the beginning of our time together mat ched the level of seriousness she employed as she composed her tweet. She leaned into the tabletop, focused solely on the iPad. As instructed, Hope talked aloud as she worked on her tweet but did not ask questions or initiate conversation. It was not until after publishing her tweet that she again relaxed. Swinging her feet and quickly tapping her left and right index fingers on her chair arms, she attentively responded on moves confirmed her confident and serious engagement with tweeting. Writing Processes - stimulated recall and talk - aloud data were analyzed in terms of motivation , goal setting , planning , and writing schemas . Writing Processes: Motivation motivation to tweet appeared positive. She quickly engaged in her work and enjoyed talking to me about Twitter and her own tweeting. Early on, Hope made statements about being motivated by feedback from her readers. When asked about what she wanted her readers to th ink when they 86 During our conversation, Hope frequently mentioned motivation to tweet seemed linked to the act of sharing and responding with others. Writing Processes: Goal Setting Hope expressed three goals for tweeting: creating meaningful messages, capturing the erest, and teaching her readers. When asked about her first goal, Hope spoke from memory about following the teacher - established tweet expectation: to tweet about something she is learning and how it will help her. When asked about her second goal she show ed an awareness keep on repeating we wouldn't keep getting more likes and if we get new things it will be Writing Processes: Planning planning included both in - advance and in - the - moment planning. Evidence of her in - advance planning occurred when talking with a partner about tw eet ideas and generating tweet topic possibilities over time, in anticipation for what she might writing on different days. [tweeted about] this before and I've bee n wanting to [tweet about the Twitter checklist] but I just pos ted in the tweet box she paused a short moment before beginning to tweet using the photo of in - the - 87 moment planning appeared when she generated specific ideas, words, a nd emojis to create her end of this first sentence, making sure her emoji selection matched the meaning of her message. Writing Processes: Writing Schemas writing schemas: genre knowledge , contextual elements , and text structure . Wri ting schemas: genre knowledge. Hope displayed knowledge about the fundamental and secondary elements of the Twitter genre: 280 character limit, student signature, photo, hashtag, and emoji. The photo Hope included was an image of the class Twitter sheet, w hich hashtags and you don't really feel like emojis on that tweet, even though there's a matching one you don't tion tool that when used, should match the meaning of the text. Writing schemas: contextual elements. Hope displayed knowledge of four contextual elements as she tweeted: followers , likes , cybersafety , and networking . As mentioned in a previous section, followers related to her topic selection and planning. She also talked about the need to choose words based on the possible reading needs of her followers. 88 it would be quicker and I could do it, well, so, some kids that don't really know how to spell the word second umm, they use a 2 so I did that in case the kids look without their parents on here, so they can owers. For instance, when talking about the tweet tweet. When I asked how she knows other people read her tweets she told me about likes know that they read it, and sometimes I am really 100% sure because people have liked it [her cybersafety, Hope explained the possible effects of if I put my name someone could, like get ... do something and find me and do some really bad about networking , Hope talked about classes post problems and they tweet back with possible solutions. She talked about other teachers and students commenting bac k to her class. She also mentioned an example of students and then [added the picture of] this little green guy, a monster [with] books [that] kept falling down h describing a GIF. Writing schemas: text structure. fixed topic text structure . Her topic was the Twitter sheet. Following the tea cher - established guidelines to tweet about what she is learning and why or how this learning will be helpful Hope explained that the Twitter sheet is helpful by showing her the steps to tweet. She ended her tweet stating 89 explicitly communicate that she learned to tweet or that she learned to use the Twitter sheet. Similar to Hal, she names the object in her photo, assuming the reader knows this is what she is lear reader to infer what Hope was learning. Composition Moves - aloud and screen - capture video data were analyzed at two levels: the letter/word/phrase level and the complete tweet level. Composition Moves: Letter, Word and Phrase Level Analysis Based on the screen - capture and talk - - time composition moves, which includes all the text composed, by either typing or using tools made available by the iPad to g enerate words, delete marks and black horizontal lines through text. And, the points in where she and a space are shown as dark pink circles. A detailed acc composition moves follow. 90 Figure 15 - of - composing he typed, without looking away 91 autocomplete tool. Next, Hope looked up at the computer screen to see her text for the first time. At this point, she had typed six words of her sentence. During our interview, I asked Hope to explain wha t she did with the autocomplete. She find the words. It has it spelled up there [pointing to the top of the keyboard] so I can just click on it and then it sh learn it in the autocomplete will provide the complete word. After speaking with me, Hope read the text she had written so far and then quickly added screen, highlighted in blue. The autocom how to spell st rereading the text so far and adding the rest of her sentence. 92 - arrow key to This m using a well - k at the iPad word using autocomplete. Hope noticed the autocomplete also automatically added a space after ding a period. At this point, two minutes and 55 seconds had passed since Hope started drafting her tweet. After completing this sentence Hope reread, deciding she wanted to revise her sentence e location of the word she deleted hted in pink and three suggestions in black boxes appeared above the pink highlight. These boxes read pressing the delete multiple times as if to delete each l etter individually. Once she noticed the deleted the automatic space and typed a period. 93 Hope reread her sentence and decided to add an emoji. She did not place a space between the period and her selected emoji. She paused for a moment. She pressed the spacebar and started The autocomplete added the sele word across tweet, pausing to decide her next move. Without adding a space after her punctuation she typed he letter keyboard to the numeric keyboar d to the character keyboard and tweet, along with my questions and her answers, took Hope six minutes and 52 seconds to complete. the text generation tools that accom pany Twitter. For instance, Hope used autocomplete for five of six consecutive words in her last sentence. Inefficient technology use is found in less prominent patterns with text deletion and keyboard switching. For instance, Hope deleted sections of text keyboard changes to access needed functions. Hope mistyped single letters with immediate corrections and unsuccessfully tried to reposition her cursor, quickly defaulting to text del etion as a solution. 94 Composition Moves: Complete Tweet Analysis As described in this same section for Hal, four composition move elements: composing , r evision processes , task environment , and resource level were used for this analysis. As can be seen in - by - second real - time composition moves over the course of composing a tweet. Taken together, this time. Figure 16 Complete tweet analysis: Overall . As the composing section of Figure 16 indicates (colored green), the most frequently occurring composition moves observed were typing multiple letters typing single letters (n=12). Single letters were often typed when correcting 95 an error. For example, typing errors with letters posit composing included nine occurren ces of typing spacebar and five of typing punctuation , but after revisions, her final tweet included only two punctuation marks. In addition, typing backspace occurred twice, in response to the automatically created space after using the autocomplete featu re. Finally, there was one instance each of the typing emoji , typing hashtag , and planning in - the - moment. The planning in the moment was observed when Hope added the text. The revision processes section of Figure 16 (colored yellow) shows five editing occurrences, t hree for correcting the spelling of words after typing the wrong letter as described Deleti ng occurred three Revising occurred when already complete sentence. The other three revision processes codes ( add new text , rewriting , reorganizing For the task environment section of Figure 16 (colored red), the mo st frequently occurring composition moves were keyboard searching to find needed letters (n=15) keyboard autocomplete usually after multiple letters in a word had already been typed (n=14) keeping the keyboard open or close to navigate between the alphabet , numeric, and special character 96 keyboards (n=9), and using task materials external objects, in this case Hope glanced at the class Twitter sheet (n=8). Hope accessed the tweet app beginning and end twice, first when she started tweeting and again when she submitted his tweet. Accessing the camera to add her previously taken photo, and scrolling for an emoji were employed one time each while composing. Scrolling for a meaningful emoji required more ti me in one moment than other composition move elements recorded. Writing prompt, already accessed photo, and cursor reposition were the only task environment code not evidenced in his composition moves. Only after watching the screen capture video repeatedl y did I recognize cursor reposition this moment was an attempt at a cursor reposition or using a word suggestion tool, therefore, I coded this move as task materials . Finally, the resource level section of Figure 16 (colored blue), shows there were six occurrences of rereading text , once around the beginning of her writing time, after making her first edit, four times spaced out around the middle of her writing time after pausi ng to think or after making corrections, and again before submitting to publish. Two pauses were recorded, once before the tweeting started, and again around the midpoint of her composing. As I stated with Hal, these pauses may have been moments of thinkin g but I did not collect evidence to support this claim. Attention diverted Complete tweet analysis: In thirds. time periods (beginning, middle, and final) as shown in Table 6 and Figure 17. Dividing the timeline into three equal time periods provides a clustered temporal view of the four composition moves: composing, revision, task environment, and resource level . 97 Table 6 frequency of use in thirds Beginning Third Middle Third End Third Total % Composing 19 15 11 45 Revision 2 4 2 8 Task Environment 19 8 13 40 Resource level 2 4 1 7 Total % 42 31 27 100 Figure 17 composing as single - letter and spacebar typing mixed with multi - letter typing . Hope also did some revision work, deleting editing task environment work included accessing the camera to add her photo, keyboard 98 searching, autocomplete, and task materials (glancing at the Twitter checklist paper on the resource level w ork included the pause waiting for the Twitter app to load and rereading her text after making revisions . multi - letter typing with a few single - letter and spacebar move s. This part of the timeline also included backspacing to remove the automatic space provided after an autocomplete suggestion, punctuation , and adding an emoji. revision moves included three edits , 1 revision , and 1 delete . Here edits were to corre ct automatically noticed spelling miscues, (started typing Twitter when she meant to type Tweet), the revision the delete task environme nt time included scrolling for an emoji and task materials (looking at the Twitter checklist paper). She also searched the keyboard, opened or closed the keyboard , and used the autocomplete . resource level moves included reading her text four times and pausing once during a transition from answering my question and getting back into her writing. composing time with multi - letter typing , only two single - letter typin g moves, a long moment adding punctuation and a short moment adding the hashtag revision work included one edit and one delete , for similar reasons as previous edits and deletes. The task environment work included some keyboard searching, keyboard open or close and autocomplete moves but the majority of this time was task materials , looking at her Twitter checklist paper, looking at the resource level work included a final reading of her tweet before publishing. 99 Emergent Features Based on all the data gathered from Hope, three additional elements associated with the tweet - composing technology features and the tweeting context were notable in the analysis of ata. These elements were categorized as emergent findings. During our conversations, Hope talked comfortably and confidently about her use of tweet - composing technology in three notable areas. The first was her explanation of the blue highlighted words. Th e second was her explanation of the black box words what I have identified as copyright . Blue Highlighted Words is wrong or a word isn't spelled that way it highlights it to let us know that it isn't a real word; so, it n it, so it shows that I know that it's there signaled an intention to return and correct the text, thereby making sense of something she did not fully unders tand at first. Black Box Words actually have that spelling pattern, like th e words that spell like that. But, I just pressed the between her words. Her attempt created an unexpected result with the text highlighted in pink and the black box words appearing above her text. She did not tell me about trying to move her 100 questioning moved to ask about the black box words rather than asking about her atte mpt to reposition the cursor. Similar to Hal, Hope admitted that she did not notice the black boxes when she tweets. Copyright When asked what parts of tweeting are frustrating, Hope brought up issues of copying. people actually copy me and then write some other other people 's work but some people actually do and it wastes their time and they are supposed to be doing their work and if you just pick off the answers and they get it wrong then you get it wrong and if you would have worked it out yourself you would have maybe had it right or a still tweet about that same idea and she said really think about it and they don't use their brain and it shuts their brain down and it's really hard for me to experience to the abstract, Internet - related concept of open access, fair - us e, and copyright may be an effective instructional approach. 101 Summary of Hope -- Overall The data interpreted for Hope resulted in a profile that indicates she is a conscientious and confident reader - writing processes she se emed motivated by feedback from her readers and sharing her learning with others. Her tweeting goals included capture the interest and teaching her readers, and following the teacher - established expectation to tweet about the what and how/why of her learn ing. Hoped planned her tweet content before writing and made in - the - moment plans while writing. She used genre knowledge to compose short - form writing including a photo, hashtag, and emoji, within a fixed topic text structure and showed a growing understa nding of followers, likes, cybersafety, and networking composition moves were dominated by composing (transcription component skills) and task environment efficient and e asy but included multiple typing miscues based on side - by - side keyboard location. For this reason, Hope had multiple edits and deletes to correct these miscues. Hope also revised Task environment components included freq uent keyboard searching and autocomplete . She also spent time scrolling for an emoji and looking at the twitter and deleting entire text chunks that were late r retyped rather than repositioning the cursor to change her text. Finally, Hope has an emerging understanding of the composing technology affordances and a growing concern for copyright. Inez: Mid Scoring Female General Description Inez appeared very co mfortable and was very confident while meeting with me. Her confidence is evidenced by how she quickly started her tweet work, accessing the iPad camera 102 and standing up with the iPad to take photos of the objects she planned to write about (which was a boo k about penguins and her sheet for taking notes while reading). Inez noticed her first picture was blurry and tried again. Once she felt satisfied with a photo(s) she settled into a chair and quickly added the two photos to her tweet space. She then paused , taking time to think about the words she would write. Writing Processes - aloud and video - stimulated recall data were analyzed in terms of motivation , goal setting , planning , and writing schemas . Writing Processes: Motivation motivation to tweet appeared positive. When asked if she liked to tweet she shook her head with a big smile, world see what you are working on and then you can see what the people think about the stuff you are twe Writing Processes: Goal Setting Inez expressed two goals for tweeting: to create meaningful messages and to share her thinking. When asked about her first goal, she said that tweets should make sense and that the purpose of tweeting is to communic ate with others. This intent was most visibly expressed in her awareness of the teacher - established tweeting guidelines. She could recite them from memory, knowing she should tweet about something she is learning. When communicating about the 103 second goal s Writing Processes: Planning planning included both in - advance and in - the - moment planning . Evidence of in - advance planning included collecting tw eet topic possibilities over time, in anticipation for what she might writing on different days. For the tweet she composed while working with about the book she was reading and the sheet she used to take notes about her reading] because I i n - the - moment planning appeared when she generated specific ideas, words, and emojis as she created her message. For example, I asked Inez if she knew what she was going to write or if she was going to figure it out as she went along. She replied Writing Processes: W riting Schemas writing schemas: genre knowledge , contextual elements , and text structure . Writing schemas: genre knowledge. For starters, Inez showed an understanding of primary and secondary elements of the Twitter genre: 280 character limit, student signature, photo, hashtag, and emoji. Her tweet represented a complete message using short - form writing, with meaningful multimodal elements, (two photos and a relevant hashtag). The photos Inez 104 - taking sheet. Her hashtag [hashtags] like pull words together and it hel ps people, like if they just read that hashtag then emojis as an elective communication tool that when used, should match the meaning of the text. For this tweet s Writing schemas: contextual elements. Inez displayed knowledge of four contextual elements as she tweeted: followers , likes , cybersafety , and networking . For instance, when asked ab out followers world like all the states and stuff because it goes online. She then talked about correct spelling ds because if they aren't spelled right then people that read your tweets can't read them because they would be like that word isn't When I asked how she knows other people read her tweets With likes] you can see what the people think about the stuff you are cybersafety protocols established by her networking what you are learning and it makes you feel good that peo ple know what you are learning. So, then if anybody sees it, [your tweet] and they think they might know how to help you if you are with her reading because she t hinks about other stuff when she's reading) 105 Writing schemas: text structure. fixed topic text structure . Her topic was the book she was reading and the sheet for notes. Following the teacher - established guidel ines to tweet about what she was learning and why or how this learning will be helpful, Inez explained that she writes down her books to keep track of what she is reading. The photo matches and extends the meaning because it showed the reader what the pape photos did not include additional items and it was more likely the reader could infer what Inez was learning. Composition Moves Based on the composition move model o - aloud and screen - capture video data were analyzed at two levels: the letter/word/phrase level and the complete tweet level. Composition Moves: Letter, Word and Phrase Level Analysis Based on the screen - capture and talk - aloud phrases captured her real - time composition moves, which included all the text composed, by either typing or using tools made available by the iPad to generate words, (i.e., autocomplete and autocorrect). Inez underline of text followed by a bright green circle and the teal circles show when Inez changes keyboards. The points where she rereads are shown as vertical purple marks. A deta iled account 106 Figure 18 - of - composing Inez started her tweet session using the first two minutes and thirty seconds to take a photo of her book, and a second photo of her note - taking sheet. The first photo of her book was blurry so she took another photo then added the photos to her tweet. Ine z started her first bar, then quickly switched from the alphabet ke yboard to the number keyboard to access the period. Inez then returned to the alphabet keyboard to add a space. After adding this space the keyboard automatically applied the shift key, automatically capitalizing the next letter typed. Inez typed this firs t sentence in 32 seconds. After completing this sentence she paused and 107 at her reading note - Inez opened her book, held the pages open to the title page and looked at the words she wanted glancing at the book, then searching the iPad keyboard for the next letter or letters and typing it with her left index finger. Inez ty the number keyboard, typed an exclamation mark, returned to the alphabet keyboard and pressed the s pacebar. Again, Inez navigated this keyboard switching sequence as if a routine composition move. note - c 108 think. Inez quietly restated the phras highlighted blue and she paused to look at the screen as if to check that she added a space. Once I at the iPad screen. Thus, Inez did not notice that when she pressed the spaceb ar the autocomplete this point, she stopped to look at her tex paused, appearing to think, and 109 but this time pressed the spacebar and then the period. Pressing the spacebar again brought her back to the alphabet keyboard. I She read again and paused again. I asked her what she was thinking. She laughed a nd said, deleted the word. Next, Inez reread the text up to the point where she would type and added he tell me why you decided to words or if she was avoiding the high - cognitive demand task of justifying the mismatch. She keyboard, typed a period, switched to the alphabet keyboard and without adding a space type d the hashtag symbol. A list of previously used hashtags appeared. Inez scrolled through this list 110 Inez deleted this space and the list of hashtags appeared again. She ignored the list and typed message. When rereading her text she read the words she planned to write even though it did not at highlighted pink and she deleted the word. She then type Next, I asked Inez if there were any words in her tweet that she wanted to ask about. She wing she worked on this word earlier, trying written was ready to publish and tweeted her message. This tweet session ran fou rteen minutes and seventeen seconds. moves ranged from typing multiple and single letters, pressing the spacebar, searching the keyboard, to looking at her task materials. She looked at the keyboard to type multiple words before looking at the text written so far. This created a problem with the text when the 111 revising occurred as she w orked through sentences, usually after completing a sentence and rereading her text so far. Her composition moves also included typical Second - r she typed, and trying various vowel patterns. Her use of the autocomplete technology was minimal, only three times, with the last time using the word sugge stions thus changing her intended message. Composition Moves: Complete Tweet Analysis As previously stated, four composition move elements were used for this analysis: composing , r evision processes , task environment , and resource level . As can be seen i n Figure - by - second real - time composition moves over the course of composing a tweet. Taken together, this view of the data time. 112 Figure 19 Complete tweet analysis: Overall. As the composing section of Figure19 indicates (colored green), the most frequently occurring composition moves observed were typing multiple letters typing spacebar after a word or to switch keyboards (n=33) and ty ping single letters, often letters at the beginning and medial four occurrences of typing punctuation (three periods and one exclamation mark) and one typing has htags in which the hashtag text was selected from a scrollable list of hashtag suggestions. Finally, planning in - the - moment was observed after Inez read the first part of her tweet and then 113 from her spoken to written version may have been influenced by the autocomplete word planning in the moment outcome was influenced by the task environment. The other two composing codes ( typing emoji and typing backspace The revision processes section of Figure 19 (colored yellow) shows seven deleting Editing occurred four times, editing Revising wanted to see if she made this revision to avoid spelling an unknown word. Inez did not attribute her revision to spelling, rather she changed her text to better match what she thought sounded right. When asked about this revision what I wanted to leaning toward. And, I felt like the second one so revision processes codes ( add new text , rewriting , reorganizing composition moves. 114 For the task environment section of Figure 19 (colored red), the most frequently occurring composition mov es were task materials to glance at the iPad screen or look at both the book and the sheet for note - taking (n=25), keyboard searching to find needed letters (n=22) and keyboard open or close to navigate between the alphabet, numeric, and special character keyboards (n=9). In addition, Inez used the keyboard autocomplete three times, once to add the - quently accessed task environment code was cursor reposition used once to access a misspelled word in the text. Similar to Hope, cursor reposition attempt, trying to co an attempt at a cursor reposition or using a word suggestion tool, therefore, this move was coded as task materials . Inez accessed the tweet app beginning and end twice, first when she sta rted tweeting and again when she submitted his tweet. Accessing the camera to add her previously taken photo, and scrolling for a hashtag were employed one time each while composing. Writing prompt, and already accessed photo were the only task environment code not evidenced in his composition moves. Finally, the resource level section of Figure 19 (colored blue), shows there were eight occurrences of rereading text , once around the beginning of her writing time, multiple times spaced out around the middle of her writing time after pausing to think or after making corrections, and again before submitting to publish. One pause was recorded to account for the time before the tweeting started. Attention diverted Complete tweet analysis: In thirds. time periods (beginning, middle, and final) as shown in Table 7 and Figure 20. Dividing the 115 timeline into three equal time periods provides a clustered temporal view of t he four composition moves: composing, revision, task environment, and resource level . Table 7 Beginning Third Middle Third End Third Total % Composing 11 23 18 52 Revision 1 3 2 6 Task Environment 7 14 15 36 Resource level 1 2 3 6 Total % 20 42 38 100 Figure 20 pausing to get set up with the technology tools followed by using the camera to take photos of the objects Inez planned to tweet about. The beginning third revision included two instances of deleting the letter 116 typing multiple and single letters for most of her first sentence. of total composition moves. During this time both typing multiple letters and typing spacebar were numerou deleting with one revision . The task environment moves were almost entirely keyboard searching and task materials . Finally, during this time Inez read her text written so far three times. The composition moves to the high frequency of typing multiple letters and typing spacebar this time also included the use of typing punctuation , typing hashtags , and in the moment planning . All of the editin g moves occurred during this end third. And, the task environment composition moves included a continuation of keyboard searching , and task materials with the inclusion of scrolling , keyboard autocomplete and keyboard open or close . Finally, the end third included five different readings of the text written so far. Emergent Features Based on all the data gathered from Inez, two additional elements associated with the tweet - composing technology features and the tweeting context were notable in the analysis of figure out how it looks and it looks wrong it makes me mad and gives me a headache because I - composing technology in two notable areas. The first was her explanation of various word tools ( red 117 underline , b lue highlight, and autocomplete when talking about the technology as knowing - other (or a partner) in the composing process. Word tools: Red Underline, Blue Highlight, and Autocomplete The data indicated I nez had a very familiar understanding of the red underline, blue highlight , and autocomplete features. When asked about these features she talked about how each was associated with misspelled words but could not explain how the features differed. During th e video - stimulated recall interview, I pointed out the red underline appeared under a word and then disappeared after she pressed the spacebar. Asking Inez to explain what was t without blue highlighted means I spelt it wrong. If you are spelling it wrong and you keep on going with it and it's wrong then it will highlight it because it's wrong so then not try to make sense of why one feature disappeared (red underline) and the other feature remained (blue highlight). Finally, asking Inez to explain the autocomplete suggestion bar she like if you don't know the word then there might be a word they think you are giving composing platform (Twitter) with the composing tools (iPad keyboard). Technology as a Knowing - Other In addition to associating Twitter with the iPad keyboa 118 indicated something about her appr oach for creating a concrete explanation to an abstract word in the re when I moved on because that's the word that was highlighted, and right, and so it thought I might have wanted that word, so then when I clicked to go on to the next word then it Summary of Inez -- Overall The data interpreted for Inez resulted in a profile that indicates she is a confident tweeter writing processes she seemed motivated by sharing her work with others. Her tweeting goals included sharing her thinking and what she is doing in class and following the teacher - e stablished expectation to tweet about the what and how/why of her learning. Inez planned her tweet content before writing and made in - the - moment plans while writing. She used genre knowledge to compose short - form writing including two photos and a hashtag within a fixed topic text structure and showed a growing understanding of followers, likes, cybersafety, and networking composition moves were predominately composing (transcription component skills) and task environment components associated with typing miscues based on side - by - reason, Inez had multiple edits and deletes to correct these miscues. Ine z also revised her tweet by changing a sentence from her initially planned text. Task environment components included frequent keyboard searching and looking at the book and papers for spelling and writing ideas. 119 She also spent time scrolling for a hashtag and used the autocomplete sparingly. Similar to Hal text chunks but also used the efficient cursor repositioning to change other text. Finally, similar to Hal, Inez has an emerging understanding of the iPad keyboard affordances inventing explanations to how and why things work without fully understanding the technology. Irene: Mid Scoring Female General Description Irene appeared assured when meeting with me. For instance, she immediately spoke several sentences into the external microphone about her preselected tweet topic. She then added her previously taken photo to the tweet space and quickly started typing. When editing her writing she boldly deleted words, added punctuation and checked her work several times. The analysis of her writing processes and composition moves confirmed her confidence as she enthusiastically answered my questions and composed her text, p ausing only once to ask how to spell a word. Writing Processes - aloud and video - stimulated recall data were analyzed in terms of motivation , goal setting , planning , and writing schemas . Writi ng Processes: Motivation motivation to tweet appeared positive. She quickly engaged in her work and enjoyed talking to me about Twitter and her own tweeting. Irene made statements about being motivated by feedback from 120 Thus, like Hope, her motivation to tweet seemed to be linked to the act of sharing with others. Writing Processes: Goal Setting Irene expressed three goals for tweeting: creating meaningful messages, capturing the tweets should make sense and that the purpose of tweeting is to communicate with others. This intent was most visibly expressed in her awareness of the teacher - established tweeting guidelines. Before publishing her tweet she read through the teacher - provided checklist of expectations to check her work and make necessary changes. When asked about the seco nd goal ideas with everyone in the whole world, they see your imagination and it helps them know what Writing Processes: Planning planning included both in - advance and in - the - moment planning. In both cases, d her in - advance planning occurred when she selected an object and took a photo of it to go with her tweet, using the object and photo as tools for advanced planning . Irene also talked through her idea before tweeting. Evidence in - the - moment planning appeared when she started tweeting and the sentences she talked through before tweeting did not in - the - moment planning appeared to be influenced by 121 the autocomplete suggestions. Thirty - one of the thirty - three words in her first draft were associated with the autocomplete or autocorrect features. Writing Processes: Writing Schemas writing sch emas: genre knowledge , contextual elements , and text structure . Writing schemas: genre knowledge. For starters, Irene displayed knowledge about the fundamental and secondary elements of the Twitter genre: 280 character limit, student signature, photo, h ashtag, and emoji. Although her published tweet did not include an emoji, Irene talked atched the meaning of the text, with little explanation necessary, she assumed that the audience would understand that the word of the day was displayed as a coll ection of words posted on classroom wall - cupboard doors, which was something not stated in the text. Interestingly, similar to Hal, Irene did not include an explanation of the other items in the photo and how they may or may not have been related to the wo rd of the day. Writing schemas: contextual elements. Irene displayed knowledge of four contextual elements as she tweeted: followers , likes , cybersafety , and networking . For instance, when explaining followers everyone in our family, umm, more people in our states, and other cities, everyone in the whole likes w the - friendly terms, she cybersafety protocols, 122 Irene acted in ways consistent with the protocols established by here teacher, also elaborating on networking what you are saying to other people. ..it's fun because you get to really like learn more stuff and Writing schemas: text structure. fixed topic text structure . Her topic was either a reading and/or the word of the day. Both were mentioned without elaboration. For instance, her first elaborate on how the word of the day helps her with chunking. Although her second sentence photo, eluded to how the chunking strategy helped with hard words. In addition, including the wo rd of the day as part of her topic suggested these words have been used to practice the chunking strategy. Composition Moves: - aloud and screen - capture video data were analyzed at two leve ls: the letter/word/phrase level and the complete tweet level. Composition Moves: Letter, Word and Phrase Level Analysis Based on the screen - capture and talk - s real - time composition moves, which includes all the text composed, by either typing or using tools made available by the iPad to generate words, 123 delete m arks and the gray circles represent student - activated cursor relocation. A detailed Figure 21 - of - composing Legend 124 Figure 22 - of - composing Irene started her tweet time talking directly into the external microphone a string of words she intended to write: This is word of the day it helps me with hard words It'll help me be a better reader. It's fun I like to use word of the day it will help me chunk hard words and have umm I can chunk the biggest word like fourth, fifth, third, or sixth grade words umm I ca n also be a better reader cuz um if there's words that need to be chunking I can just look at word of the day and that it will help me chunk with the words I need help chunking. hat I need 125 Next, Irene added her photo to the tweet space and began to type, selecting the word oticing the autocomplete xt six words were each added using the followed by ta pping the matching words in the autocomplete bar. Next, she searched the followed tapping this same word in the autocomplete bar. Thus far, I - by - side location on the key board or Irene thinks the word She looked at the autocorrect word bar, did not make a selection, and deleted the letters until she 126 and he omplete for the same word. highlight she added end punctuation which caused the highlighted text to autocorrect to helps me with my w the space and typed an end punctuation mark. Directly after, she engaged the magnifying glass essed the 127 ed line appeared under this newly typed text. Once going to rema o longer a spa ce and typed her end punctuation mark. Next, she used the magnifying glass to position the text. Paused, noticed the red un 128 r ereading, she positioned her cursor after the ending punctuation mark of her tweet and helps me with my words. That I need some chunking. Words that are hard word s I can look up seconds. and deleted both the hashtag mark and the end punctuat ion mark. She retyped the period and She used the magnifying glass t added a space. When she added the space the keyboard automatically switched from the special character keyboard to the alphabetic keyboard. She switched to the numeric keyboard, repositioned her learn ing about? Why are you learning about it? How will you use this information? Read your With a less - than - one - ed to reading the 129 weet she nodded her head and said, eaned over the table to look closely at her iPad screen. She moved her lips, silently reading her writing. Without talking she stopped reading, repositioned the cursor from the end of her tweet to the position after the period ending her first sentence. Sh e deleted the period, moved the cursor back to the end of her tweet and moved the cursor again to that I need some chunking. Words that are hard words I can look up The prominent patterns autocomplete o r autocorrect for thirty - one of the thirty - three words in her first draft. Inefficient technology use is found in less prominent patterns with text deletion and selecting autocomplete after typing the complete word. For instance, Irene deleted sections of text then immediately retyped this same text. Irene also completely typed out nine words followed by tapping these 130 same words in the autocomplete bar. Irene successfully repositioned her cursor sixteen times and reread sections of her tweet nine times. Co mposition Moves: Complete Tweet Analysis As previously stated, four composition move elements were used for this analysis: composing , r evision processes , task environment , and resource level . As can be seen in Figure 23, the complete tweet analysis used - by - second real - time composition moves over the course of composing a tweet. Taken together, this view of the data Figure 23 osition Moves Frequency of Use Timeline Analysis Complete tweet analysis: Overall. As the composing section of Figure 23 indicates (colored green), the most frequently occurring composition moves observed were typing multiple letters like chunks of wo typing spacebar after a word (n=11) and typing single letters, often letters at the beginning and medial ed 131 six occurrences of typing punctuation. The first punctuation mark was typed after the last word of her complete tweet. The additional punctuation marks were included after Irene checked her work and made multiple edits. Additionally, two typing hashtags were observed, but after edits, her final tweet included only two punctuation marks and one hashtag. Finally, due to a lack of evidence, planning in - the - moment was not recorded but the possible influence of the autocomplete suggestions should be noted, su planning in - the - moment outcome was influenced by the task environment . The other two composing codes ( typing emoji and typing backspace The revision processes section of Figure 23 (colored yellow) shows a high frequency of editing to add punctuation, capital letters and fix spelling (n=11). Similarly, eight deleting occurrences were observed, often in conjunction with the editing occurrences. For example, Irene punctuation and newly added capital letter. After further edits and deletes to the second sentence, attempt to address syntax miscues. The other four revision processes codes ( revising , add new text , rewriting , reorganizing ) were not evidenced in her composition moves. For the task environment section of Fig ure 23 (colored red), the most frequently occurring composition moves were keyboard autocomplete to generate the majority of text (n=30), task materials to glance at the iPad screen and when reading the twitter sheet to check her work (n=25), keyboard sea rching to locate letters (n = 18) and cursor reposition/magnifying 132 glass to access and text to be edited (n=16). Irene employed the keyboard open or close seven times, once after adding her photo and six times toward the end of tweeting when adding her has htag and signature. She accessed the tweet app beginning and end twice, first when she started tweeting and again when she submitted his tweet. Accessing the camera to add her previously taken photo was employed one time while composing. Writing prompt, al ready accessed photo, and scrolling were the only task environment code not evidenced in his composition moves. Finally, the resource level section of Figure 23 (colored blue), shows there were nine occurrences of rereading the text. The first reading occ urred after Irene typed her entire first draft. The additional rereadings were spaced out around the middle of her writing time after pausing to think or after making corrections, and again before submitting to publish. Two pauses were recorded to account for the time before the tweeting started and once midway through our time together. Attention diverted Complete tweet analysis: In thirds. time periods (beginnin g, middle, and final) as shown in Table 8 and Figure 24. Dividing the timeline into three equal time periods provides a clustered temporal view of the four composition moves: composing, revision, task environment, and resource level . Table 8 Summary of Ire Beginning Third Middle Third Final Third Total % Composing 10 12 8 30 Revision 1 4 5 10 16 17 21 54 133 Task Environment Resource level 1 3 2 6 Total % 28 36 36 100 Figure 24 her composing as multi - letter typing with some single - letter and spacebar typing revision work included one edit task environment work included accessing the camera to add her photo, a high frequency of autocomplete use, and some keyboard searching resource level work included the pause waiting for the Twitter a pp to load. concentration of total composition moves as her final third. During this time both typing multiple 134 letters, single letters, and typing spacebar were numerous with som e typing punctuation middle third revision work included both editing and deleting in equal amounts. The task environment moves were keyboard autocomplete, keyboard searching with some cursor repositioning/magnifying glass and task materials . Fina lly, during this time Irene paused once and read her text written so far four times. composition moves. When composing , in addition to typing multiple letters , sing le letters , and typing spacebar this time also included the use of typing punctuation and typing hashtags . For revision , all of the editing and deleting moves were similar to the middle third editing and deleting with an increase in editing by two. The tas k environment composition moves included a high increase in task materials and use of the cursor reposition/magnifying glass, a single autocomplete and her most frequent use of keyboard open or close . Finally, for resource level , the final third included five different readings of the text written so far. Emergent Features use of tweet - composing technology. This element was categorized as e mergent findings associated with her statements about various word tools ( red underline , blue highlight, magnifying glass ) and what were observed in her use of the autocomplete word suggestions, which were short, single, unelaborated responses. Word tools : Red Underline, Blue Highlight, Magnifying Glass and Autocomplete The data indicated Irene has a basic understanding of the red underline, blue highlight , and magnifying glass features. When asked to tell me about the red underline, Irene replied, 135 rene did not say much more to connect the blue highlighted words with the autocomplete suggestions given her high frequency of use with this feature. Asking Irene to explain the magnifying glass the bubble is kind of like a magnifying gla ss so it can help you see what you need to, it can help a close analysis of the screen capture video indicated that Irene used the autocomplete feature for most of the words in her tweet. This included words she had already typed out completely. For suggestion bar. Making this selection produced a space following the sel ected word. It was not clear why Irene made these selections. Unfortunately, I did not notice she was doing this at the time of the video stimulated recall interview thus did not ask her to tell me what she was thinking. Summary of Irene -- Overall The da ta interpreted for Irene resulted in a profile that indicates she is a confident writing processes she seemed motivated by the enjoyment of the task and sharing her work with others. Her tweeting goals included sharing her learning an d creating interesting content following the teacher - established expectation to tweet about the what and how/why of her learning. Irene planned her tweet content before writing and made in - the - moment plans while writing. She used genre knowledge to compose short - form writing including both a photo and hashtag within a fixed topic text structure. She also showed a growing composition moves were predominately task environment components a ssociated with the keyboard and task materials 136 followed by composing (transcription component skills). Specifically, for transcription, Hope often typed multiple and single letters and used the spacebar. For revision Irene had multiple edits and deletes. T ask environment components included a high frequency of autocomplete, keyboard searching and repositioning the cursor. She also spent time looking at the twitter checklist paper. Irene was not distracted, paused only twice and read from her text nine times . Finally, Irene has an emerging understanding of the red underline, blue highlight, and magnifying glass features with an interesting approach when using the autocomplete feature. Kip: Average - Low Scoring Male General Description Kip appeared nervous whe n he first started, taking a big breath as he approached the iPad to start tweeting. When I asked questions he would pause a second, take another deep breath, then answer in one - word responses. For instance, toward the beginning of our time together I aske d Kip to tell me his thinking after deleting part of a word underlined red. He stopped, looked . He started talking more, using hand gestures as he explained. In the end, analysis of his writing processes and composition moves suggested a more relaxed and confident engagement with tweeting. Writing Processes Based on the writing process model of Hay - aloud and video - stimulated recall data were analyzed in terms of motivation , goal setting , planning , and writing schemas . 137 Writing Processes: Motivation motivation to tweet appeared positive. He seemed motivated by the enjoyment of the task and sharing with about something, and people around the world can see [the tweets], so it feels like you are an Writing Processes: Goal Setting Kip expressed two goals for tweeting: sharing or teaching his readers and creating meaningful messages. When asked about his first goal , Kip talked about sharing what he is do ing and that the purpose of tweeting is to communicate with others. T his intent was most visibly expressed in his awareness of the teacher - established tweeting guidelines. Writing Processes: Planning planning included both in - advance and in - the - moment planning. In both cases, he in - advance planning occurred when he selected an object and took a photo of it to go with his tweet, using the object and photo as tools for advanced planning . Evidence of in - the - moment planning appeared when he generated specific ideas and words as he created his message. For example, when writing his second sentence Kip started listing reasons for liking math with someone. He could not think of a third reason, thus adjusting his plan. in - the - moment planning may have been 138 influenced by the autocomplete suggestions. This is evidenced by a word choice edit, sel ecting Writing Processes: Writing Schemas writing schemas: genre knowledge , context ual elements , and text structure . Writing schemas: genre knowledge. Kip displayed knowledge about the fundamental and secondary elements of the Twitter genre: 280 character limit, student signature, photo, hashtag, and emoji. He successfully composed a c omplete message using short - form writing, including meaningful multimodal elements, (a photo and a meaningful hashtag). Although his published tweet did not include an emoji, Kip talked about careful emoji selection. The hashtag and photo Kip included in h is tweet matched the meaning of the text, with the assumption the reader shared in his classroom experience. For instance, the photo is an image of the math rotations for the day. This includes papers magnetically held to a classroom whiteboard, each paper associated with instructions written on the whiteboard. Writing schemas: contextual elements. Kip displayed knowledge of four contextual elements as he tweeted: followers , likes , cybersafety , and networking . For instance, without follo wers likes ple with a] Twitter account, you could like theirs, like sometimes, and you could see around the whole cybersafety protocols, Kip acted in ways consistent with the If you put your name at the end of it or 139 his concrete understanding of this abstract concept. When it came to the contextual element of networking d about sharing and receiving ideas. For example, Kip explained that he th Writing schemas: text structure. fixed topic text structure . His topic was an aspect of a class activity called math rotations. Math with ence did provide a reason why he likes Also, while following the teacher - established guidelines to tweet about what he is learning and why or how this learning is help ful, Kip did not explicitly say that he was learning a math concept. Instead, he told his readers why he liked a particular math activity, assuming his readers share a common knowledge of the math rotations and math with someone experience. Composition Mo ves - aloud and screen - capture video data were analyzed at two levels: the letter/word/phrase level and the complete tweet level. Composition Moves: Letter, Word and Phrase Level Analysis Bas ed on the screen - capture and talk - phrases captured his real - time composition moves, which includes all the text composed, by either typing or using tools made available by the iPad to generate words, (i. e., autocomplete and 140 word, and phrase composition moves follow. Figure 25 - of - composing As I adjusted the screen capture to enlarge the recorded iPad screen, Kip sat patiently. ord I and am. He quickly re - added the space and then I interrupted asking that he tell me what he was thinking. He replied e what he was going to type before he changed his mind. He 141 saying a loud the words he was typing or intended to type. For example, when typing the word number keyboard. He added the apostrophe. The keyboard automatically returned to the lowercase letters but he did not notice this change. Kip tapped the button to change the keyboard moving to the numeric keyboard, then quickly tapped the button again, returning to the alphabet keyboard. He then adde word t understanding of autocorrect or limited language to explain autocorrect. time the screen showed his attempt underlined in red. After looking at the screen he continued with his message saying the words automatic. To add his ending punctuation Kip changed to the numeric keyboard, typed the period and then paused to think. Returning to his typing he changed keyboards back to the alphabet board and added th 142 words he deleted the letter. Without adding a space he pressed and held his finger on the letter t the screen, Kip continued to problem solve deciding that he needed to use the capital letter keyboard to type Ki ncing at the iPad was thinking. He said as if a question, s head, no explanation suggests a limited understanding of the keyboard tools and his attempt to invent meaning with something that does not make sense. When I asked him what he would do to help himself with word later. Even though we paused from his writing to talk about the underlined word, Kip did not return to his writing by rereading first. As if mentally holding on to his ideas he jumped right back into his work saying 143 xt where he had typed mark. This is interesting because before accessing the magnifying glass the cursor was already in that position. This makes me wonder if he did not realize he could start typing from that position ri ght after deleting or if he simply did not notice the cursor was already there. Finally successful with his cursor position Kip stopped, rubbed the right side of his forehead with his right hand and started to reread the first part of his text up to the po int of his followed by a space. This moved Kip's end punctuation a space a way from the word. He did not correct this space error. Instead, he positioned his cursor to the end of his text and then read aloud his entire tweet. After reading he switched to the numeric keyboard and typed a period a space away from his final word and then looked up at me. Kip glanced at his screen, deleted the period and space then typed his end punctuation next to the last letter of his last word. This shows me that Kip notices when he has spaces or punctuation marks inaccurately placed. Why did he n ot notice the spacing error after his first sentence? Also, when typing the expected tweet ending he did not add a space after the punctuation mark. Immediately after the end punctuation, 144 mainly talking about math with Kip did not change keyboards, rather he used the alphabet keyboard. Kip opened the photos, selected the i (me) he published his tweet. From start to finish, including a few moments of conversation, it took Kip 11 minutes and 40 seconds to compose his tweet. The prominent patter composition moves indicate that he maintained focus the entire time, made a strategic word easily typed al l of his words using the autocomplete tool only when completing the word o add underline was inconsistent showing correctly spelled words (homophones) with a red underline. Kip was conscientious with his spaces between words with one er ror after his first sentence. This spaces between words in his hashta g. 145 Composition Moves: Complete Tweet Analysis As previously stated, four composition move elements were used for this analysis: composing , r evision processes , task environment , and resource level . As can be seen in Figure 26, the complete tweet analysis - by - second real - time composition moves over the course of composing a tweet. Taken together, this view of the data Figure 26 osition Moves Frequency of Use Timeline Analysis Complete tweet analysis: Overall. As the composing section of Figure 26 indicates (colored green), the most frequently occurring composition moves observed were typing multiple letters typing spacebar after a word (n=17) and typing single letters , often letters at the beginning and medial points of words and to correct ccurrences of typing punctuation (three periods, two commas, and one apostrophe) and one typing hashtags in which 146 the hashtag text included spaces between the words. Finally, typing backspace and planning in - the - moment were each observed once. His planning in - the - moment occurred when writing his second sentence. Kip started listing reasons for liking math with someone. He could not think of a third reason, thus adjusting his plan. The composing code typing emoji was not evidenced in The revision processes section of Figure 26 indicates (colored yellow) the most frequently occurring revision processes move observed was deleting to correct mistyped letters Edit ing occurred three have been influenced by the autocomplete suggestions. Thi s is evidenced when he selected the four revision processes codes ( revise , add new text , rewriting , reorganizing ) were not evidenced For the task environment section of Figure 26 (colored red), the most frequently occurring composition moves were keyboard searching to find needed letters (n=23) using task materials when looking at the iPad screen (n=18) and keeping the keyboard open or close to navigate between the alphabet, numeric, and special character keyboards (n=11). Kip accessed the keyboard autocomplete usually after multiple letters in a word had already been typed (n=6) and the cursor reposition/magnifying glass to relocated is cursor five times. Kip accessed the tweet app beginning and end twice, first when he started tweeting and again when he submitted his tweet. He also accessed the camera twice, once to add his previously taken photo and once tapping the camera access butto n on accident. Similarly, scrolling was accessed once by mistake, 147 attempting to scroll the tweet text up to view his photo, realizing he had not yet added the photo to his tweet. Already accessed photo and writing prompt were the only task environment code not evidenced in his composition moves. Finally, the resource level section of Figure 26 (colored blue), shows there were seven pause occurrences, once before the tweeting started, and multiple times throughout the composition moves. Similar to Hal, thes e pauses appeared to be moments of thinking. Reading text three times, once around the beginning of his writing time, after making his first edit, once after making corrections before adding his tweet closing signature, and again before submitting to publi sh. Attention diverted Complete tweet analysis: In thirds. time periods (beginning, middle, and final) as shown in Table 9 and Figure 27. Dividing the timeline into t hree equal time periods provides a clustered temporal view of the four composition moves: composing, revision, task environment, and resource level . Table 9 Beginning Third Middle Third Final Third Total % Composing 21 15 10 46 Revision 2 3 2 7 Task Environment 10 10 21 41 Resource level 3 2 1 6 Total % 36 30 34 100 148 Figure 27 composing as multi and single letter typing and spacebar typing . Kip also did some revision work, deleting made a word choice edit task environment work included keyboard searching , keyboard open or close and task materials (looking at the iPad screen). resource level work included the pausing waiting f or the Twitter app to load and three other pauses with one time reading his text after making revisions. composing as multi - letter typing with a few single - letter and spacebar moves. This part of the timeline also included punctuation , (commas in a sequence) and in - the - moment planning revision moves included six deletes . These deletes were to re task environment time is similar to his beginning third with keyboard searching and task materials , 149 the difference being a reduction in keyboard searching and an increase in task materials (looking at iPad s creen, thinking). Kip also used the autocomplete resource level moves included three pauses . He paused after completing his first sentence and after the beginning phrase of his second sentence. The third pause was during a transition from answering my question and getting back into his writing. composing time with multi - letter typing , only three single - letter typing moves, two short mome nts of adding punctuation and a short moment adding the hashtag revision work included two edits and two deletes , for similar reasons as previous edits and deletes. The task environment work included an increase in the variety of moves. Keyboard sea rching and task materials remained similar to the beginning and middle thirds. The final third included accessing the camera , scrolling , and using the cursor reposition for the first time. There was also increased use of keyboard open or close and autocomp lete resource level work included two readings of his tweet while making final changes before tweeting. Emergent Features use of tweet - composing technology. This element was categorized as emergent findings associated with his statements about various word tools ( red underline , and autocomplete ). Similar to Irene, during the video - stimulated recall interview, Kip provided short responses without much elaboration. Word tools: Red Underline and Autocomplete The data indicated Kip has a basic understanding of the red underline and autocomplete features with some misconceptions. Kip talked about the red underline and misspelled words, 150 beyond the obvious misspelling to activate the red underline. Similarly, during the video - stimulated recall interview, I asked Kip about an autocomplete action he did not notice while typing. The close video showed "becuse" highlighted in blue change to " be use" (a choice provided by the autocomplete) when Kip pressed the spacebar. After watching this video segment Kip responded, to spell a word you could see it up there [pointing at the top of the keyboard] if it's up there, you could press on it, and it puts the word up there. I did it [for the helpful autocomplete feature as a convenient tool for gen Summary of Kip -- Overall The data interpreted for Kip resulted in a profile that indicates Kip is a confident tweeter writing processes he seemed motivat ed by sharing his learning with others and feeling like an author. His tweeting goals included sharing what he was doing in class and following the teacher - established expectation to tweet about the what and how/why of his learning. Kip planned his tweet content before writing and made in - the - moment plans while writing. He used genre knowledge to compose short - form writing including a photo and a hashtag within a fixed topic text structure and showed a growing understanding of followers, likes, cybersafety composition moves were 151 predominately composing (transcription component skills) and task environment components typing errors based on s ide - by - side keyboard location and accidental typing moves. For this reason, Kip had multiple edits and deletes to correct these errors and adjust for word choice changes. Task environment components included frequent keyboard searching and looking at the i Pad screen doing what looked like thinking of what to write next. He used the autocomplete as a faster way to write big words and the cursor repositioning as an efficient tool when navigating the text. Finally, Kip has an emerging understanding of the iPad keyboard affordances inventing explanations to how and why things work without fully understanding the technology. Kayla: Average - Low Scoring Female General Description Kayla appeared less confident than her classmates when meeting with me. This is evide nced by her immediate request for help both for an initial idea and spelling the fourth word of her tweet. She also experienced several keyboard - related errors during our meeting. The analysis of her writing processes and composition moves, however, provid ed evidence of a more resilient problem - solving engagement with tweeting as time went by. Writing Processes - aloud and video - stimulated recall data were analyzed in terms of motivation , goal s etting , planning , and writing schemas . Writing Processes: Motivation motivation to tweet appeared positive. When asked if she likes to tweet she shook her head with 152 a big smile, W riting Processes: Goal Setting Kayla expressed two goals for tweeting: sharing with her readers and creating meaningful messages. When asked about her first goal , Kayla talked about sharing what she is doing so other people on Twitter can see her tweets. When asked about her second goal she explained why she made it the same way it was before, because, I'm like, it didn't make sense. If I added another word, it w Writing Processes: Planning planning included both in - advance and in - the - moment planning. In both cases, she followed the tea in - advance planning occurred when she selected an object and took a photo of it to go with her tweet, using the object and photo as tools for advanced planning . in - the - moment planning appeared when she generated specific ideas and words as she created her message. For example, before typing, Kayla paused, unsure of what to write about. After loading her previously taken photo she quickly start in - the - moment planning add another word but I didn't know a 153 Writing Processes: Writing Schemas of writing schemas: genre knowledge , contextual elements , and text structure . Writing schemas: gen re knowledge. Kayla displayed knowledge about the fundamental and secondary elements of the Twitter genre: 280 character limit, student signature, photo, and hashtag. She successfully composed a complete message using short - form writing, including meaning ful multimodal elements, (photos and a meaningful hashtag). The hashtag and photo Kayla included in her tweet matched the meaning of the text, with the assumption the reader shared in her classroom experience. For instance, Kayla selected the same photo co ntent as Kip, an image of the math rotations for the day. As stated previously, this included papers magnetically held to a classroom whiteboard, each paper associated with instructions written on the whiteboard. Writing schemas: contextual elements. Kay la displayed knowledge of four contextual elements as she tweeted: followers , likes , cybersafety , and networking . For instance, in a statement explaining the importance of a coherent message she evidenced her awareness of followers and likes The people who follow and want to retweet [your tweet] and like cybersafety protocols, Kayla acted in ways consistent with the protocols established by her teacher. When it came to the networking 154 Writing schemas: text structure. fixed topic text structure . Similar to Kip, her topic was math rotations. When asked how she knew her Similar to other students, she did not explicit ly say what she is learning, she tells the reader about similar to other students, this statement did not explain the details of her math learning. Composition M oves - aloud and screen - capture video data were analyzed at two levels: the letter/word/phrase level and the complete tweet level. Composition Moves: Letter, Word and Phrase Level Analysis Based on the screen - capture and talk - and phrases captured her real - time composition moves, which includes all the text composed, by either typing or using tools made available by the iPad to generate words , (i.e., autocomplete and underline of text followed by a bright green circle. The line of text toward the bottom of the ithout realizing she was typing in the hashtag search bar. 155 Figure 28 - of - composing Kayla started her tweet by immediately asking for help, saying she did not know what to e, and "are" without the caps lock automatically engaged on the keyboard without Kayla noticing. The caps lock disengaged once the "A" was typed. Kayla paused a moment, looked at the letter, then backspaced to deleted "A." Next, the numeric keyboard opened and switched to the special 156 characters keyboard, then switched to the alphabet keyboard. This all happened in seconds as if s witching through the keyboards has become a common composition move. Next, Kayla keyboard, adjusted in her seat and smiled. Looking back at the iPad screen she removed the space, typed a period and pressed the spacebar. At this point, Kayla had been working two minutes and forty - seven seconds. After reading her first sentence she looked up, scratched her neck, and continued to her nd selected "this" from the autocomplete choices. She searched the autocomplete selection, the cursor automatically spaced for the next word. When Kayla typed t he spacebar the cursor blinked but stayed in its already spaced position. The next three words, hout noticing the automatic space Kayla typed her end punctuation. She looked at the iPad screen, deleted the punctuation mark and space then retyped the punctuation next the last word in her second sentence. At this time, three minutes and fifty - seven sec onds had passed. her second sentence, put her left fist to her mouth and continued to think. Unable to come up with more words to add on, she retyped her end punctuation. Kayla pressed buttons causing happening. She returned to the twee t space, smiled, took a deep breath and returned to her 157 hashtag choices searching the keyboard between single and multi - are math roteation. This will help and "m." At this point, the entire tweet highlighted in blue. Confused, Kayla commented, of "This" in but she was resilient and problem solved her way through the confusion. he She added a space between the words 158 from the autocomplete choices. After talking to me for a moment she returned to her work, searching the keyboard for end pun sound out the sentence] it d the word. take. This moved her text to the second line. Again the keyboard minimized and reopened. This is when I noticed every time Kayla did this keyboard switch it delet help returning the second line text back to the first line. I asked if she had any ideas to solve the problem. She made a movement that looked like she was trying t o drag the text where she wanted it. Realizing this approach would not work she looked to me for help. I instructed Kayla 159 line. With a huge smile of reli ef, Kayla looked at me and swayed in her seat. She reread her t published her tweet. This tweet session ended at 14 minutes and forty - two seconds. composition moves indicate th autocomplete or autocorrect for twelve of the nineteen words typed. Inefficient technology use i s found in less prominent patterns with text deletion. For instance, Kayla deleted sections of text then immediately retyped this same text. Most prominent was the inefficient frequent keyboard switching and accidental button pushing (all associated with t he keyboard task environment). Kayla successfully repositioned her cursor eleven times and reread sections of her tweet three times. 160 Composition Moves: Complete Tweet Analysis As previously stated, four composition move elements were used for this analys is: composing , r evision processes , task environment , and resource level . As can be seen in Figure - by - second real - time composition moves over the course of composing a tweet. Taken together, this view of the data Figure 29 Complete tweet analysis: Overall. As the composing section of Figure 29 indicates (colored green), the most frequently occurring composition moves observed were typing single letters as Kayla typed most words a letter at a time (n=41). When typing words found a second typing multiple letters for composition move was typing spacebar five typing pun ctuation occurrences and three each of typing backspace to remove a space after a 161 word and to correct text placement that was accidentally moved to the second line of her text box and typing hashtags. T yping hashtags occurred twice in the confusion of keyb oard switching and button pushing and a final time to create her tweet closing. Finally, planning in - the - moment was observed twice. First after adding her photo and again when thinking of additional words to add to her second sentence. The composing code t yping emoji composition moves. The revision processes section of Figure 29 (colored yellow) shows six deleting such as the # and @ symbols. Editing occurred five times, correcting spelling errors in the word Revising occurred twice. First when Kayla took a photo of her pre - selected object, inspected the photo for quality, and decided to retake the picture revision processes codes ( add new text , rewriting , reorganizing n moves. For the task environment section of Figure 29 (colored red), the most frequently occurring composition moves were keeping the keyboard open or close many times appearing unintentional (n=22), and keyboard searching to find needed letters (n=14). T he use of both keyboard autocomplete and cursor reposition to access different parts of text occurred eleven ties. Task materials and seven times to look at the iPad screen, appearing to think. Kayla used scrolling twice. First for a hashtag and again to scroll her screen from looking at her picture back up to look at her text box. Kayla accessed the tweet app beginning and end twice first when she started tweeting, and again when she submitted his tweet. Accessing the camera also occurred twice, once to take a 162 picture of her previously taken photo and to add her photo to her tweet. Kayla looked at her already accessed photo once to think of what to write and writing prompt as the only task environment Finally, the resource level section of Figure 29 (color ed blue), shows there were three occurrences each of reading text and pausing . The first reading occurred after completing her first sentence. The remaining readings occurred once while making edits and again before submitting her tweet. The pausing occurr ences recorded to account for the time before the tweeting started and twice midway through our time together. Attention diverted was not part of Complete tweet analysis: In thirds. hree time periods (beginning, middle, and final) as shown in Table 10 and Figure 30. Dividing the timeline into three equal time periods provides a clustered temporal view of the four composition moves: composing, revision, task environment, and resource l evel . Table 10 Beginning Third Middle Third Final Third Total % Composing 13 18 12 43 Revision 1 5 2 8 Task Environment 13 21 11 45 Resource level 2 0 1 4 Total % 29 44 27 100% 163 Figure 30 photo, thinking of what to write, and typing her first two sentences. During the beginning third, both composing and task environment elements were given equal attention. D uring this time Kayla did a lot of single letter typing , keyboard searching and used the keyboard autocomplete . Revision elements and resource level elements were of low use, deleting and editing each once, reading once and pausing three times. The compos text and assigned tweet number. This time is also when frequent errors occurred with random and often unintentional button pushing and keyboard changes. Once again, composing and task environment elements had a similar frequency of use. Kayla employed her highest frequency of typing single letter, keyboard searching, cursor reposition, and task materials during this time. She also used keyboard open and closed often during this tim e. It makes sense that the largest 164 worked to fix the errors created from her unintentional button pushing. She stayed focused without pausing or stopping to read her te xt during this middle third of her composing. single letter typing and some multi - letter typing . Kayla finished with some editing, deleting, to continue fixing errors and revising, making a change to her hashtag text. Her task environment moves were mostly keyboard open or close with some cursor reposition and task materials . Kayla ended her composition moves reading twice once while making final corrections and again before publishing her t weet. Emergent Features There were no emergent features (i.e., unexpected or noteworthy patterns that extended beyond the a priori analysis of writing processes and composition moves ) associated with this case. Summary of Kayla -- Overall The data interpreted for Kayla resulted in a profile that indicates she is a resilient problem writing processes s he seemed motivated by the enjoyment of the task and sharing wit h others. Her tweeting goals included sharing with others and writing a coherent message. Kayla planned her tweet content before writing and required some support for in - the - moment plans while writing. She used genre knowledge to compose short - form writing including a photo and hashtag within a fixed topic text structure and showed a growing understanding of followers, likes, cybersafety, and networking. composition moves were dominated by composing (transcription component skills) and task environm ent components associated with the keyboard and task materials. Kayla navigated the iPad keyboard at first with slow and methodical movements, which eventually turned into 165 random keyboard switching and unintentional button pushing. She edited misspelled wo rds and worked diligently to address all errors. She considered revising her text to enhance meaning with additional words and used the cursor reposition and autocomplete tool often. Luke: Low Scoring Male General Description At first, Luke appeared nervo us when meeting with me. He frequently rubbed his forehead, sometimes pulling at his eyebrow when pausing to answer my questions. While comfort - level seemed to incr ease as his quiet voice grew louder, easily answering questions and ok okay and returned to typing. The analysis of his writing processes and composition moves suggest Luke is a tentative yet capable tweeter. Writing Processes Ba - aloud and video - stimulated recall data were analyzed in terms of motivation , goal setting , planning , and writing schemas . Writing Processes: Motivation Drawing on both the verbal and body langu motivation to tweet appeared positive but conflicted. He seemed motivated by sharing with others and improving writing skills but often felt conflicted when he had an opportunity to tweet because it 166 took time away from com pleting his reading goal. When asked about this conflict Luke explained, Well, when I don't meet my reading goal and um it's kind of a hard choice because, reading and tweeting, is kind of hard so sometimes, some days when Mrs. Hammer gives us a choice, we tweet or we read to self. Sometimes I choose tweeting and sometimes I choose read - to - self. Writing Processes: Goal Setting Luke expressed three goals for tweeting: sharing with his friends, becoming a better writer, and constructing meaningful messages. When asked about the first goal , Luke explained making sense. Writing Processes: Planning planning included both in - advance and in - the - moment planning. In both cases, he in - advance picture I think what I'm going to take a picture of, and then I look at in - the - moment planning occurred when he considered writing his tweet about using iPads and then changed his mind to wr 167 Writing Processes: Writing Schemas writing schemas: genre knowledge , contextual elements , and text structure . Writing schemas: genre knowledge. For starters, Luke displayed knowledge about the fundamental and secondary elements of the Twitter genre: 280 character limit, student signature, photo, hashtag, and emoji. He successfully composed a complete message using short - form writing, including m eaningful multimodal elements, (a photo and a meaningful hashtag). Although his published tweet did include two emojis, Luke could not explain why he selected selected. Writing schemas: contextual elements. Luke displayed knowledge of four contextual elements as he tweeted: followers , likes , cybersafety , and network ing . For instance, without followers, some people read our tweets and some peo likes cybersafety protocols, Luke acted in ways consistent with the protocols established by his teacher. He also want your footprint big. You want it small. And you do not want to put evidencing his concrete understanding of this abstract concept. When it came to the contextual 168 networking escribing how a Writing schemas: text structure. ile following the teacher - established guidelines to tweet about what he is learning and why or how this learning is helpful, Luke did not write about how this learning, assuming his readers share a common knowledge of the class vocabulary instruction. Conv ersely, I think the reader can assume Luke is learning about new words. Composition Moves - aloud and screen - capture video data were analyzed at two levels: the letter/word/phrase level and t he complete tweet level. Composition Moves: Letter, Word and Phrase Level Analysis Based on the screen - capture and talk - phrases captured his real - time composition moves, which includes all the text co mposed, by either typing or using tools made available by the iPad to generate words, (i.e., autocomplete and 169 a point where he revised is shown as a horizon tal black mark strikethrough. A detailed account Figure 31 - of - composing space. Noticing his apprehension or nervousness, I tried to ease Luke into the work by reminding was the most important thing you learned in second grade and why was it the most important or how is it going to help important thing they learned in second gr reminding him of the prompt, causing him to start tweeting about iPads rather than the word 170 working on. H e smiled and said that it was fun to look at the larger screen then as if immediately going to type something about learning how to use iPads but his picture did not match that idea and he was not sure how he could take a picture of an iPad. I offered to help him either take a photo or find a photo of an iPad but he quickly decided to stick with writing about his original t end punctuation mark. From the time Luke started typing his tweet to the completion of this first sentence, including short moments of our talking was one minute an d fifty - seven seconds. Continuing with his second sentence, Luke added a space after his end punctuation Again, like the first sentence, Luke deleted the space, added his end punctuation mark and typed the spacebar. Composing t wi 171 e beginning time to add his photo, was six minutes and twenty - one seconds. time his head no, for an emoji character he liked from watching the Emoji Movie. Unable to find this particular emoji he selected an open book to represent being smart. He also added the American flag. I eplied, reminder. I may have been providing these reminders in response t o his apparent nervousness. ver. Clever means you are smart. It helps me learn. 172 time at this moment was twelve minutes and twenty - six seconds. There were few prominent patterns observed in t he letter, word, and phrase level analysis composition moves , based on his simple and mostly efficient composing. Within the three short sentences, Luke wrote his composing consisted mostly of typing multiple letters, using the spacebar between w ords, random emoji selection, and moments of planning. When typing multiple letters he usually quickly typed the first three or four letters of the word, slowing down to add any remaining letters for longer words. Pressing the spacebar was an automatic com position move often requiring a backspace to add punctuation at the end of his sentence. a cat - face emoji that he seemed locate initial letters in words and to search emojis, and task materials to look at his iPad screen. few, only reading once when prompted before submitting his tweet and pausing three times. Finally, Luke had only two text errors associated with spacing. Composition Moves: Complete Tweet Analysis As previously stated, four composition move elements wer e used for this analysis: composing , r evision processes , task environment , and resource level . As can be seen in Figure - by - second real - time composition moves over the course of composing a tweet. Taken together, this view of the data 173 Figure 32 Complete tweet analysis: Overall. As the composing section of Figure 32 indicates (colored green), the most frequently occurring composition moves observed were typing multiple letters used for almost all of the words typed and usually the beginning chunks (n=24 ). The second most frequent compos ing move was typing spacebar (n = 18). Luke added a space after every word as if a routine composition move. This space bar routine caused a need for typing backspace for each of his three sentences, removing the space to then use typing punctuation (n=3). single letter typing moments may be associated with the visual si Typing hashtag and in the moment planning were each employed once. The revision processes section of Figure 32 (colored yellow) shows very view revision moves. Deleting occurred twice, each time f 174 - face emoji. Luke employed revising once when starting to type revision proc esses codes ( add new text , editing , rewriting , reorganizing ) were not evidenced in For the task environment section of Figure 32 (colored red), the most frequently occurring composition moves were keyboard searching to locate lett ers and emojis (n=19), keyboard open or close (n=7), and task materials when looking at the iPad screen (n=7). Luke accessed the tweet app beginning and end twice, first when he started tweeting and again when he submitted his tweet. Accessing the camera t o add his previously taken photo, scrolling for a hashtag, and keyboard autocomplete were employed one time each while composing. Cursor reposition , already accessed photo, and writing prompt were the only task environment code not evidenced in his composi tion moves. Finally, the resource level section of Figure 32 (colored blue), shows there were three moments of pausing and one of reading . The pausing occurrences recorded to account for the time before the tweeting started and twice midway through our time together. The reading occurred after being prompted to check his text before submitting his tweet. Attention diverted tion moves. Complete tweet analysis: In thirds. time periods (beginning, middle, and final) as shown in Table 11 and Figure 33. Dividing the timeline into three equal time periods provides a clustered tempo ral view of the four composition moves: composing, revision, task environment, and resource level . 175 Table 11 - of - use in thirds Beginning Third % Middle Third % Final Third % Total % Composing 21 23 11 55 Revision 2 1 0 3 Task Environment 15 12 11 38 Resource level 1 1 2 4 Total % 39 37 24 100 Figure 33 photo, thinking of what to write, and typing his first sentences and part of his second sentence. During the beginning third, composing was most frequent followed by task en vironment elements. During this time Luke did a lot of multi - letter typing , typing spacebar , and keyboard 176 searching . Single - letter typing and typing punctuation also occurred. Revision elements and resource level elements were of low use, deleting , readin g, and pausing each one time. third with similar amounts of multi - letter typing , typing spacebar , and keyboard searching . The middle third also included typing punctuat ion , searching for emojis looking at the keyboard screen ( task materials ) and pausing once. multi - letter typing, typing emoji, typing spacebar, and typing hashtag . Luke did not employ revision processes during this time and task environment moves were mostly keyboard search and keyboard open or close Luke ended his composition moves pausing once and reading once before publishing her tweet. Emergent Features There were no emergent features (i.e., unexpected or noteworthy patterns that extended beyond the a priori analysis of writing processes and composition moves ) associated with this case. Summary of Luke -- Overall The data interpreted for Luke resulted i n a profile that indicates he is a tentative yet capable tweeter. Nervous and unsure of his abilities yet competent in his knowledge of Twitter writing processes he seemed motivated by sharing with ot hers and improving his spelling, conversely, Sometimes he does not like to tweet because it conflicts with time reading. His tweeting goals included sharing with others, becoming a better writer, and writing a coherent message. Luke planned his tweet conte nt before writing and made in - the - moment plans while writing. He used genre knowledge to compose 177 short - form writing including a photo, hashtag, and two emojis. The photo, hashtag, and one emoji clearly matched the text. Luke used a fixed topic text structu re and showed a growing composition moves were dominated by composing (transcription component skills) and task environment components associated with the keyboard and task materials. L uke navigated the iPad keyboard easily, typing most text with multi - letter typing, using the spacebar, keyboard searching and looking at the iPad screen. He deleted typing errors and considered revising his text once. Finally, Luke paused three periodicall y during our time together and had to be reminded to read his tweet before publishing. Lori: Low Scoring Female General Description Lori appeared confident while working with me, but this confidence was occasionally juxtaposed with moments of dependence. For instance, Lori showed confidence when beginning to tweet, activating the Twitter application, typing her first sentence and inventing a strategy to eliminate the red underline. But, during a moment of problem - solving, she displayed a sense of dependenc writing processes and composition moves are peppered with elements of conf idence and moments of dependence. Writing Processes - aloud and video - stimulated recall data were analyzed in terms of motivation , goal setting , planning , and writing schemas . 178 Writing Processes: Motivation motivation to tweet appeared positive. She seemed motivated by the enjoyment of the task and sharing with others. When asked if she likes to tweet, Lori said enthusiast Writing Processes: Goal Setting Lori expressed two goals for tweeting: creating meaningful messages and sharing what she has done in school. When asked about her first goal Lori could recall from memory the teacher - going to write about and say how it helps you and say what you can do with i about the second goal she talked about sharing her learning and what she is doing in school. Writing Processes: Planning planning included both in - advance and in - the - moment planning. In both cases, she ormat from memory and added ideas as she created her message. in - advance and in - the - moment planning first appeared when she explained, build up her Lego structure] figure out what to write about, that's why I brought all of these things for Writing Processes: Writing Schemas writing schemas: genre knowledge , contextual elements , and text structure . 179 Writing schemas: genre knowledge. For starters, Lori displayed knowledge about the fundamental and secondary elements of the Twitter genre: 280 character limit, student signature, enough r oom they can put a hashtag but if they don't want to put a hashtag they don't have to put Writing schemas: contextual elements. Lori displayed knowledge of four contextual elements as she tweeted: followers , likes , cybersafety , and networking . For instance, without followers bout followers at a global level but did say, family, and friends that y likes cybersafety protocols, Lori acted in ways consistent with the protocols established by her teacher. And when it networking Writing schemas: text structure. ic of a fixed topic text structure . Her topic was STEM. Following the teacher - established guidelines to tweet about what she was learning and why or how this learning would be helpful, Lori explicitly d not explain the contents of her photos, similar to Hal, assuming her reader would understand how building an object with Legos will help her learn about STEM - related concepts. 180 Composition Moves talk - aloud and screen - capture video data were analyzed at two levels: the letter/word/phrase level and the complete tweet level. Composition Moves: Letter, Word and Phrase Level Analysis Based on the screen - capture and talk - aloud data, the analysis of - time composition moves, which includes all the text composed, by either typing or using tools made available by the iPad to generate words, (i.e., autocomplete and autocorrect). To ill delete marks and the gray circles represent student - activated cursor relocation. A detailed Figure 34 rase flow - of - composing 181 screen. Because she had not talked while composing this first part of her tweet I took advantage e think her teacher provided tweet suggestions for Lori as part of her in - advance - planning . I asked if she knew what words she was reminded Lori to talk out loud as she worked and to tell me what she was thinking. She shook her head yes as she leaned over the iPad to continue typing. corner of her twee thought she would make a comment about the spelling or the blue highlight. Instead, she told me each letter at a quick pac e than when typing previous words. Lori tapped the spacebar, deleted 182 After this first sentence, Lori looked up at me without talking. I waited two seconds xt or what Then, quickly noticed her error and button eliminating the space between she posit ioned the cursor to the end of her sentence. Again, she looked at me and I again re but I d her right index finger under her lip and gazed down at her chair. Finally, with a shy smile, finger now rocking against ts until she gets an idea. I assured her that it was okay if she needed to wait and think. After twenty - three seconds of wait time, Lori looked up at me, still 183 yes After I asked Lori who would be reading her tweet and what she wanted her readers to suggestion. autospace and added her end punctuation mark. Our time together thus far was seven minutes and fifty seconds. Similar to when Lori finished her first sentence, she looked up at me. Again, I prompted, - friendly version of the engineering design process she brought from class. She looked at her paper again and again That's okay! So, let's see. you've got 2nd graders and 3rd graders that are going to read your tweet. And, so now they know, that STEM helps you because STEM helps you by how to build. Anything else that you want y our readers to know about STEM? 184 I as yes Stem helps me by This time for in - the - moment planning with su pport ended at eleven minutes and thirty seconds, for a total of three minutes and forty seconds. With renewed energy, Lori returned to the iPad, quickly deleted her end punctuation mark, added a space and began to type in a sing - automatically followed the autocomplete aut end punctuation mark. Following her now familiar pattern, once she comp leted her sentence she looked up at me 185 how you minutes and thirty - four seconds passed. know? That's something you can teach me because I'm not sure how that works. Can you tell me re [pointing to the screen] then And, that's something you don't want to do? Tell me more about that; I'm still trying to understand what you meant when you said if you pu t more right here on the next line you'd have a whole bunch more room and you'd have to write more. Is that what you said or did I not hear it right? tell me more a 1 86 I sucking on the charm of her necklace, inching toward the end of her chair, she nodded her head, yes . composition moves were varied but concentrated with typing multiple and single letters , the spacebar , searching the keyboard , and looking at her task materials deleti ng and editing , occurred as she worked through sentences. Her composition moves also included typical second - emove blue highlights and red underlines. If a word highlighted in blue she would delete the final letter, retype that letter and then press the spacebar . If a word was underlined red she would press the spacebar adding two spaces after the underlined word and then delete the extra space. Her use of the autocomplete technology was minimal, only three times. Finally, Lori paused often after completing a sentence or phrase, as if not sure what to do next and only read her text when prompted. Composition Move s: Complete Tweet Analysis As previously stated, four composition move elements were used for this analysis: composing , r evision processes , task environment , and resource level . As can be seen in Figure 35, the complete tweet analysis used a timeline to - by - second real - time composition moves over the course of composing a tweet. Taken together, this view of the data 187 Figure 35 equency of Use Timeline Analysis Complete tweet analysis: Overall . As the composing section of Figure 35 indicates (colored green), the most frequently occurring composition moves observed were typing spacebar to separate words, (n=28), typing multiple letters typing single letters (n=22). Single letters were often typed when correcting miscues and for longer words. For example, typing the ped slowly searching for letters. Also, when addressing typing errors caused typing backspace codes, in response to the automatically created space after using the autocomplete feature and to remove extra spaces, and three typing punctuation . The other three composing codes ( typing emoji, typing hashtags, and in - the - moment planning sition moves. Codes for in - the - moment planning were not added because Lori only added more to her text with my support making it unclear if this would be a composition move she would have used on her own. 188 The revision processes section of Figure 35 (color ed yellow) shows deleting occurred Deleting also occurred when Lori deleted a word highlighted in blue only to retype that same word as a strategy to remove the highlight. Also, edi ting revision processes codes (revising, add new text , rewriting , reorganizing For the task environment section of Figure 35 (colored red), the most frequently occurring composition moves were keyboard searching to find needed letters (n=19) and task materials such as looking at the iPad screen or external objects (n=14). Employing keyboard auto complete, usually after multiple letters in a word had already been typed, keeping the keyboard open or close to navigate between the keyboard and camera when taking and loading photos, and accessing the camera occurred four times each. Lori accessed twice each the cursor reposition tweet app beginning and end , first when she started tweeting and again when she submitted her tweet. Scrolling , writing prompt, and already accessed photo, were the only task environment code not evi denced in her composition moves. Finally, the resource level section of Figure 35 (colored blue), shows there were eight occurrences of pausing, once before the tweeting started, and multiple times throughout her composing. As I stated with the other case s, these pauses may have been moments of thinking but I did not collect evidence to support this claim. There were four occurrences of reading text , all from my prompting. Lori read twice around the beginning of her writing time, after her first sentence a nd again twice after her second sentence. Each of these moments was times when Lori publish. Attention diverted 189 Complet e tweet analysis: In thirds. time periods (beginning, middle, and final) as shown in Table 12 and Figure 36. Dividing the timeline into three equal time periods provides a clustered temporal view of the fou r composition moves: composing, revision, task environment, and resource level . Table 12 Beginning Third % Middle Third % Final Third % Total % Composing 18 15 22 54 Revision 2 1 4 7 Task Environment 8 10 12 31 Resource level 4 3 1 8 Total % 32 29 39 100 Figure 36 190 composing as multi - letter, single letter and spacebar typing, including a punctuation mark to complete her first sentence. This part of the timeline also included backspacing to remove the automatic space provided after an autocomplete suggestion or other extra spaces. Lori also did some revision work, editing the wo deleting last letters in two task environment work included keyboard searching , task materials (glancing at the iPad screen), and cursor reposition twice to acc ess the resource level work included pausing four times, first waiting for the Twitter app to load and again while working on and after completing her first sentence. After being prompted, Lori read her first sent ence twice to help her think of what to write next and to hear how the sentence would sound after making a correction. single letter typing with some multi - letter typing and s pacebar moves. Similar to the beginning third, this part of the timeline also included backspacing to remove the automatic space provided after an autocomplete suggestion or other extra spaces, and a punctuation mark to end her second revi sion task environment time included searching the keyboard, task materials, and one autocomplete. Lor resource level moves included pausing three times and reading her text once. third, included composing as multi - letter, and spacebar typing, with some single letter typing mixed in. Also, this time included a punctuation mark to complete her third sentence and backspacing to remove the automatic space provided after an autocomplete suggestion or other 191 revision work included one edit and six deletes , for similar reasons as previous edits and deletes. The task environment work included some keyboard searching , keyboard open or close, accessing the camera with some task materials and autocomplete resource level work include d a final pausing and a reading after her delete work but she did not read her tweet before publishing. Emergent Features There were no emergent features (i.e., unexpected or noteworthy patterns that extended beyond the a priori analysis of writing process es and composition moves ) associated with this case. Summary of Lori -- Overall The data interpreted for Lori resulted in a profile that indicates she is a dependent yet confident tweeter, or similar to Luke, she is a tentative yet capable tweeter. Reluct ant and unsure of her abilities yet competent in her knowledge of Twitter and ability to compose a complete writing processes she seemed motivated by the enjoyment of the task and sharing with others. Her tweeting goals included sharing w ith others and writing a coherent message. Lori planned her tweet content before writing, bringing her Lego structure and - the - moment plans before writing each of her sentences with significant teacher - support (my support). She used genre knowledge to compose short - form writing including two photos. Lori used a fixed topic text structure and showed a growing understanding of followers, likes, cybersafety, and networking. composition moves were largely composing (transcription component skills) and task environment components associated with the keyboard and task materials. Lori typed most text with multi - letter typing, using the spacebar, keyboard searching and looking at the iPad screen. 192 S he deleted typing errors and edited spelling twice. Finally, Lori paused often during our time together and had to be reminded to read her tweet. Cross - Case Comparisons In this second section, I present comparisons of the eight cases in three parts. First, I present comparisons based on the writing processes data (i.e., thoughts made known through talk - alouds and video - stimulated recall). Second, I present comparisons based on the composition moves data (i.e., actions recorded by the screen - capture video an d further understood by talk - alouds). Finally, I present comparisons based on the emergent features data (i.e., thoughts and actions from all data sources). Cross - Case Comparison Writing Processes Based on the writing process model of Hayes (2012), codes f rom talk - aloud and video - stimulated recall data were compared across cases in terms of motivation , goal setting , planning , and writing schemas . Table 13 summarizes the individual findings for each of the eight tweeters, highlighted to indicate patterns. 193 Table 13 Cross - Hal Hope Inez Irene Kip Kayla Luke Lori Motivation For fun To develop technical skills To share work with others For fun To share work with others For fun To share work with others For fun To share work with others For fun To share work with others For fun To share work with others For fun * Sometimes not motivated to tweet To share work with others For fu n Goal Setting To share about their learning To ensure messages make sense To develop technical skills To share about their learning To ensure messages make sense To teach others To share about their learning To ensure messages make sense To share about their learning To ensure messages make sense To share about their learning To ensure messages make sense To share about their learning To ensure messages make sense To share about their learning To ensure messages make sense To share about their learning To ensure messages make sense Planning In - advance To pre - select topic & take photo In - the - moment To think of each sentence while writing To use autocomplete suggestions To seek teacher support In - advance To pre - select topic & take photo In - the - moment To think of each sentence while writing To use autocomplete suggestions In - advance - select topic & take photo In - the - moment To think of each sentence while writing In - advance - select topi c & take photo In - the - moment To think of each sentence while writing To use autocomplete suggestions In - advance - select topic & take photo In - the - moment To think of each sentence while writing In - advance - select topic & take photo In - the - moment To think of each sentence while writing To use autocomplete suggestions To seek teacher support In - advance - select topic & take photo In - the - moment To think of each sentence while writing To seek teacher support In - advance - select topic & take photo In - the - moment To think of each sentence while writing To seek teacher support Writing Schemas - genre knowledge 2 elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter demonstrate elements of a 2nd - grade expert tweeter demonstrate elements of a 2nd - grade expert tweeter Writing Schemas - contextual elements 3 elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter elements of a 2nd - grade expert tweeter element s of a 2nd - grade expert tweeter demonstrate elements of a 2nd - grade expert tweeter Writing Schemas - text structure Fixed Topic Giraffe Text of Tweet This is my giraffe It will help me become a better Builder. Giraffes live in the rainforest. by531#Giraffe Fixed Topic Twitter sheet Text of Tweet This helped me in 2nd grade by showing me the steps how to Tweet. This is called the Twitter sheet . By 310#Twittersheet Fixed Topic Reading a book Reading log Text of Tweet I am reading a book. It is called Penguins! I marked it doun on a peas of paper so I coled keep track of what I am reading . This will help me with reading. #reading by875 Fixed Topic Word of the day Text of Tweet The word of the day helps me with my words that I need some chunking. Words that are hard words I can look up and see what needs to be chucked. by610 #wordoftheday Fixed Topic Math with someone Text of Tweet I like math rotations because we have math with someone. I like math with someone because we get to w ork and play with someone . By 105 #math with someone Fixed Topic Math rotations Text of Tweet This is are math rotations. This will help me with math. by972#mathrotat ions Fixed Topic The word clever Text of Tweet We learned the word clever. Clever means you are smart. It helps me learn. By 411 #Clever Fixed Topic STEM Text of Tweet I am going to tweet about stem. Stem helps me by how to build and we can build things made of legos . By 202 2 See Figure 11 for the flow chart used to conduct this analysis. 3 Also see Figure 11. 194 As can be seen in Table 13, seven of the eight students expressed the same two types of motivation for tweeting: (a) to share their work with others, and (b) to have fun. In addition, Hal expressed motivation to develop better technical skills as a result of his tweeting. Similarly, all eight students expressed goals that were focused on following the teacher - established guideline for tweeting (which was to create a message that would make sense to their audie nce about content from their learning). Table 13 indicates that an additional goal was omething new and give ideas to other teachers about learning experiences they might want to try in their classroom. in - advance planning followed the same pattern: selecting an object and then taking a pho to of that object for their tweet, using the object and photo as tools for advanced planning . in - the - moment planning indicated their thoughtful generation of each sentence (using specific words and ideas to use) while wr iting about their pre - in - the - moment planning for generating letters, words and ideas was of three types: used autocomplete suggestions to add letters, words and ideas (Hal, Hope, Irene, and Kip), sought teach er support for letter, word and idea generation (Hal, Kayla, Luke, and Lori), and worked independently on letter, word and idea generation (Hope, Inez, Irene, and Kip). aspects of writing schemas : genre knowledge , contextual elements , and text structure . Writing schemas: genre knowledge. All eight students provided evidence of including and/or explaining the fundamental (i.e., 280 character limit, student signature) an d secondary (i.e., photo and hashtag) elements of the Twitter genre. In addition, Hope and Luke added the 195 optional emojis to their tweets. Hope included an emoji matching the meaning of her text after her first sentence. Luke included two emojis as part of his tweet closure. One of the two aligned with the meaning of his text. Lori, interestingly, was aware that a hashtag was needed in her limit had not been reache d; she had 176 characters remaining). Writing schemas: contextual elements. Table 13 indicates that all eight students displayed knowledge of the four contextual elements as they tweeted or talked about tweeting: followers , likes , cybersafety , and network ing . Specifically, all eight students mentioned something about others ( followers ) reading their tweets and liking their tweets. When it came to cybersafety protocols, all eight students acted in ways consistent with the protocols established by their teac networking, with and learning from others. Writing schemas: text structure. As the bottom row of Table 13 displays, all eight ed topic text structure. KEY: Each part of their guideline begins with the direction to: Write 2 sentences. Then it prompts students with 3 questions: W hat are you learning about? (highlighted in yellow) Why are you learning about it? (highlighted in turquoise) How will you use this information? (highlighted in green) The pink highlight accounts for elaboration. Each respective topic aligned with the i tem in the pre - selected photo (highlighted in yellow). Each student followed the teacher - established guidelines to tweet about: (a) what information they were learning (highlighted in yellow), (b) why they were learning about it (highlighted in turquoise) , and/or (c) how they would use it (highlighted in 196 of the prompt (highlighted in green). Hal, Kip, and Luke added additional information about their topic (hig hlighted in pink). Cross - Case Comparison Composition Moves Based on the writing process model of Hayes (2012), a cross - case comparison was conducted at two levels: the letter/word/phrase level and the complete tweet level. Four composition moves elements were used for analysis at each of the two levels: composing , r evision processes , task environment , and resource level . To both manage the large amount of data and to see if a similar number of coded composition moves generated a similar flow - of - composing, comparison pairs with identical or a similar number of codes for a particular composition move were conducted. Cross - Case Comparison of Composition Moves: Letter, Word and Phrase Level Analysis A cross - case comparison was made of the letter, word, and phrase moves enacted by the eight students. The comparison analyzes flow - of - composing figures that have been paired. Each pair represents one of the four composition moves : composing , r evision proc esses , task environment , and resource level . Pair selection was based on similar quantities of codes for each of the four composition moves elements. Luke and Irene were compared based on their same quantity of codes from the composing category, (n=55). In ez and Kayla were compared based on their similar quantity of codes from the revision processes category, (n=12) and (n=13) respectively. Luke and Hal were compared based on their similar quantity of codes from the task environment category, (n=38) and (n= 42) respectively. And, Inez and Irene were compared based on their similar quantity of codes from the resource level category, (n=9) and (n=11) respectively. 197 Composing . While the number of composing codes for Luke and Irene were identical (n=55) their flo w - of - composing figures for the complete tweet letter, word, and phrase analysis displayed very different story lines for their composition moves . As seen in Table 14 and Figure 37, Luke typed every letter of every word in his tweet, except one (when he use d the between every word, backspacing on three occasions. Additionally, he typed three complete emojis (leaving two in his published tweet), three punctuation marks, and one hashtag. In short, Luke was a do - it - yourselfer when it came to composing a tweet. In other words, Luke did not take advantage of the variou keyboard tools such as the autocomplete. In contrast, Irene typed very few complete words, opting ins tead to key in a letter or two spacebar less frequently than Luke because the autocomplete feature automatically placed a space after the word selected by h - of - composing figure indicates, she did not use the backspace at all, she typed six punctuation marks (with only two ending up in her published tweet), and she typed two hashtags (with only one in her published tweet). Hence, Irene was a n autopilot - user when it came to composing a tweet. In other words, Irene used the keyboard tools. Thus, the identical number of composing codes for Luke and Irene could suggest similarity in the pattern for their composition moves, but their flow - of - comp osing figures indicate otherwise. Luke was a do - it - yourselfer while composing, Irene was an autopilot - user . 198 Table 14 Composing code totals for Luke and Irene Composing: Luke Irene Typing emoji 3 0 Typing single letter 2 11 Typing multiple letters 24 25 Typing spacebar 18 11 Typing backspace 3 0 Typing punctuation 3 6 Typing hashtags 1 2 Planning inline/realtime 1 0 Total 55 55 Figure 37 Side - by - - of - composing figures Composing Codes (n=55) Composing Codes (n=55) 199 Revision Processes . While the number of revision processes codes for Inez and Kayla were similar (n=12 and 13 respectively), their flow - of - composing figures for the complete tweet letter, word, and phrase analysis displayed very different story lines for their composition moves . As seen in Table 15 and Figure 38, Inez was deleting letters and spaces at nearly every turn: two and one for forgetting a space after deleting letters. Her repeated editing of words appeared four times, three of which were associated with correcting and re - revising flip - flopper when it came to revision processes in her tweet.In other words, Inez would type letters before realizing changes were needed. She flipped and flopped between typing and deleting., fixing spelling errors based on what looked right or suggestions provided by autocomplete. num ber. In a business - meticulously excised mis - typed characters such as the # and @ symbols. Her editing occurred allenging 2nd grade revising occurred twice, once when she took a photo of her pre - selected object, inspected the photo for quality, and decided to retake the pictur e for improved quality; and again when she changed her chaos - cleaner when it came to revision processes in her tweet. In other words, Kayla would type her message without 200 noticing accide ntal cursor movement or accidentally typing random characters, then deleting to clean up the message. Thus, the similar number of revision processes codes for Inez and Kayla could suggest similarity in the pattern for their composition moves, but their flo w - of - composing figures indicate otherwise. Inez was a flip - flopper while composing, Kayla was a chaos - cleaner . Table 15 Revision Processes code totals for Inez and Kayla Revision Processes: Inez Kayla Add new text 0 0 Editing 4 5 Rewriting 0 0 Revising 1 2 Reorganizing 0 0 Deleting 7 6 Total 12 13 201 Figure 38 Side - by - - of - composing figures Revision Processes Codes (n=12) Revision Processes Codes (n=13) Task Environment. While the number of task environment codes for Luke and Hal were similar (n=38 and 42 respectively), their flow - of - composing figures for the complete tweet letter, word, and phrase analysis displayed very different story lines about the ir composition moves. As seen in Table 16 and Figure 39, Luke spent considerable time searching the keyboard to locate letters and emojis (n=19), opening, closing or toggling among keyboards (n=7), and looking at the iPad screen (n=7). He tapped the iPad s creen on two occasions, once when he started tweeting and then again when he submitted his tweet. And unlike other students in this study, he dawdled with the camera while adding his previously taken photo, got lost while scrolling for a hashtag, and marve led at the keyboard autocomplete feature in a star - struck composing moment. 202 Thus, Luke was a keyboard - clunker/lost - in - the - labyrinth when it came to task environment influences on his tweet. In other words, Luke was not fluent and confident with the task en vironment, specifically the keyboard, often taking extended amounts of time to complete typing tasks. focused on the shiny and new. He toyed happily with the microphone and keyboards (n=10) and such as the speech - to - text microphone. In fact, his fixation with the speech - to - text feature was his lodestar. He first used it to spel all of the words to the second sentence in his tweet (n=14). In addition, he engaged in quick keyboard searching to find needed letters seven times and used the keyboard autocomplete tool nearly as often. To a lesser degree, Hal tapped around the iPad screen, once when he started tweeting and again when he submitted his tweet. Finally, he popped open the camera to add his photo to the tweet, amped - up his scrolling swipes in search of an emoji, ru mmaged for the photo kid - in - a - candy - store/inspector gadget when it came to task environment influences on his tweet. In other words, Hal confidently navigated the task environment, trying new typing supports offered by the keyboard technology. Thus, the similar number of task environment codes for Luke and Hal could suggest similarity in the pattern for their composition moves, but their flow - of - composing figures indica te otherwise. Luke was a keyboard - clunker/lost - in - the - labarynth while composing, Hal was a kid - in - a - candy - store/inspector - gadget . 203 Table 16 Task Environment code totals for Luke and Hal Task Environment: Luke Hal Camera 1 1 Tweet app beginning and end 2 2 Keyboard Searching 19 7 Keyboard open or close 7 10 Keyboard autocomplete 1 5 Cursor reposition 0 1 Scrolling 1 1 Task materials 7 14 Already accessed photo 0 1 Writing prompt 0 0 Total 38 42 Figure 39 Side - by - letter, word, phrase flow - of - composing figure Task Environment Codes (n=38) Task Environment Codes (n=42) 204 Resource Level . Finally, while the number of resource level codes for Inez and Irene were similar (n=9 and 11 respectively), their flow - of - composing figures for the complete tweet letter, word, and phrase analysis displayed a different story lines for their composition moves. As seen in Table 17, and Figure 40, Ine z was reading right away, first at the beginning of her writing time, multiple times spaced out around the middle of her writing time after a thinking or editing moment and again before submitting to publish. Inez paused only once before tweeting. In short , Inez was a right - away reader when it came to resource level engagement while composing her tweet. In other words, Ineze employed rereading DURING writing. In contrast, Irene delayed reading, waiting to type her entire first draft before her first read. Similar to Inez, the additional rereadings were spaced out around the middle of her writing time after a thinking or editing moment and again before submitting to publish. Irene paused twice, once before tweeting and again midway through tweeting. In short , Irene was a slow - to - go reader when it came to resource level engagement while composing her tweet. In other words, Irene employed rereading AFTER writing her complete message. In sum, the similar number of resource level codes for Inez and Irene could su ggest similarity in the pattern for their composition moves, but their flow - of - composing figures indicate otherwise. Inez was a right - away - reader while composing, Irene was a slow - to - go reader . 205 Table 17 Resource Level code totals for Inez and Irene Resource Level: Inez Irene Reading text 8 9 Attention diverted 0 0 Pause, unknown reason 1 2 Total 9 11 Figure 40 Side - by - - of - composing figures Inez's Resource Level (n=9) Resource Level Codes (n=11) In sum, the flow - of - composing figures from selected cases provides insight into the similar in many ways, the letter, word, and phrase flow - of - composing figures t ells a different 206 story. In other words, a similar quantity of composition moves codes for one of the four elements does not equal a similar composition moves experience. The cross - case comparison work in the next section identifies the similarities across cases with the four composition moves. It will be important to keep in mind how these surface - level similarities may have multiple meanings when it comes to how each tweeter experienced a particular composition move. Cross - Case Comparison of Composition M oves: Complete Tweet Analysis: Overall The cross - case comparison of composition moves initially examined the overall patterns for the four codes: composing , r evision processes , task environment , and resource level . Table 18 displays the eight students at the top of each column, with the four main composition moves in the far left column. The sub - codes for each composition move are entered inside each cell and highlighted to indicate prominence. The highlights indicate the most frequently occurring sub - code s for composing (green), revision processes (yellow) and task environment (light red) . The shades of blue highlights indicate patterns in the resource level section. 207 Table 18 Cross - highlighted to indicate patterns Hal Hope Inez Irene Kip Kayla Luke Lori Composing Typing emoji Typing single letter Typing multiple letter Typing spacebar Typing backspace Typing punctuation Typing hashtag In the moment planning Typing emoji Typing single letter Typing multiple letter Typing spacebar Typing backspace Typing punctuation Typing hashtag In the moment planning Typing emoji Typing single letter Typing multiple letter Typing spacebar Typing backspace Typing punctuation Typing hashtag In the moment planning Typing emoji Typing single letter Typing multiple letter Typing spacebar Typing backspace Typing punctuation Typing hashtag In the moment planning Typing emoji Typing single letter Typing multiple letter Typing spacebar Typing backspace Typing punctuation Typing hashtag In the moment planning Typing emoji Typing single letter Typing multiple letter Typing spacebar Typing backspace Typing punctuation Typing hashtag In the moment planning Typing emoji Typing single letter Typing multiple letter Typing spacebar Typing backspace Typing punctuation Typing hashtag In the moment planning Typing emoji Typing single letter Typing multiple letter Typing spacebar Typing backspace Typing punctuation Typing hashtag In the moment planning Revision Processes Add new text Editing Rewriting Revising Reorganizing Deleting Add new text Editing Rewriting Revising Reorganizing Deleting Add new text Editing Rewriting Revising Reorganizing Deleting Add new text Editing Rewriting Revising Reorganizing Deleting Add new text Editing Rewriting Revising Reorganizing Deleting Add new text Editing Rewriting Revising Reorganizing Deleting Add new text Editing Rewriting Revising Reorganizing Deleting Add new text Editing Rewriting Revising Reorganizing Deleting Task Environment Camera Tweet app Keyboard searching Keyboard open close Keyboard autocomplete Cursor Reposition Scrolling Task materials Already accessed photo Writing prompt Camera Tweet app Keyboard searching Keyboard open close Keyboard autocomplete Cursor Reposition Scrolling Task materials Already accessed photo Writing prompt Camera Tweet app Keyboard searching Keyboard open close Keyboard autocomplete Cursor Reposition Scrolling Task materials Already accessed photo Writing prompt Camera Tweet app Keyboard searching Keyboard open close Keyboard autocomplete Cursor Reposition Scrolling Task materials Already accessed photo Writing prompt Camera Tweet app Keyboard searching Keyboard open close Keyboard autocomplete Cursor Reposition Scrolling Task materials Already accessed photo Writing prompt Camera Tweet app Keyboard searching Keyboard open close Keyboard autocomplete Cursor Reposition Scrolling Task materials Already accessed photo Writing prompt Camera Tweet app Keyboard searching Keyboard open close Keyboard autocomplete Cursor Reposition Scrolling Task materials Already accessed photo Writing prompt Camera Tweet app Keyboard searching Keyboard open close Keyboard autocomplete Cursor Reposition Scrolling Task materials Already accessed photo 208 Table 18 Resour ce Level Focused Rereading at end of thought/sente nce Attention diverted once Focused Rereading at end of thought/sente nce Rereading (noticed unwanted autocorrect) Focused Rereading at end of thought/sente nce Rereading (noticed unwanted autocorrect) Focused Rereading at end of thought/sente nce Focused Rereading at end of thought/sente nce Rereading (noticed word to change) Focused Rereading at end of thought/sente nce Focused Rereadin g at end (with (promptin g) Focused Rereading (w/ prompting) Pausing/ thinking/stu ck For the composition move of composing , two patterns were evident. One was the pattern of multi - letter , spacebar and backspace moves made by Hal and Luke. These sub - codes suggest that their typing was proficient enough to key in multiple letters and spaces, punctuated occasionally with a tap o n the backspace to delete an unnecessary space. The other pattern was composed of single - letter , multi - letter and spacebar moves made by Hope, Inez, Irene, Kip, Kayla and Lori. Their typing was a mix of single and multiple letters peppered with periodic pr essing of the spacebar. The other four sub - codes for composing appeared infrequently. Three patterns were evident in the composition move of revision processes . One was formed by the editing , deleting and revising moves made by Hal, Hope and Inez. This com bination of sub - codes suggested that their revision was proficient enough to address misspelled words, delete occasional typing mistakes and make changes to tweet content to enhance meaning. The second pattern was made up of editing and deleting moves made by Irene, Kip, Kayla and Lori. Their revision was proficient enough to address misspelled words and delete occasional typing mistakes. And the third pattern was composed of deleting and revising moves made by Luke. His revision was proficient enough to de lete occasional typing mistakes and make changes to tweet content to enhance meaning. 209 For the composition move of task environment , three patterns are evident. One was the pattern of keyboard searching , keyboard open or close , keyboard autocomplete , and ta sk materials moves made by Hal, Hope, Inez, Kip, Luke and Lori. This combination of sub - codes suggested that their task environment moves were a combination of scanning the keyboard for letters, toggling among keyboards, selecting autocomplete suggestions, and spending time looking at the iPad screen or other assignment materials. The second pattern was made up of keyboard searching , keyboard open or close , keyboard autocomplete , and cursor reposition moves by Kayla. This combination of subcodes suggest her task environment was scanning the keyboard for letters, toggling among keyboards, selecting autocomplete suggestions, and moving the cursor from one location to another. And the third pattern was co mposed of keyboard searching , keyboard autocomplete , cursor reposition and task materials moves by Irene. This combination of sub - codes suggested that their task environment moves were a combination of scanning the keyboard for letters, selecting autocompl ete suggestions, moving the cursor from one location to another, and spending time looking at the iPad screen or other assignment materials. Three patterns were evident in the composition move of resource level . One was the pattern of focused , rereading at end of thought/sentence and rereading (noticed unwanted autocorrect) moves made by Hope, Inez and Kip. Their combination of sub - codes suggested that resource level use was focused with an eye toward noticing miscues, rereading often. The second pattern wa s made up of focused and reader at end of thought/sentence moves made by Hal, Irene and Kayla. Their resource level moves were focused with reading often. And the third pattern was composed of focused and rereading (with prompting) moves by Luke and Lori. Their resource level moves were focused without a concern for rereading unless reminded. 210 Cross - Case Comparison of Composition Moves: Complete Tweet Analysis: In Thirds The cross - case comparison of composition moves was also divided into three time periods (beginning, middle, and final). Dividing the timeline into three equal time periods provided a clustered temporal view of the four composition moves: composing, revision, task environment, and resource level . Table 19 Most Frequent code occurrence divided into three time periods, highlighted to indicate patterns Key: Composing, Revision Processes, Task Environment Resource Level Hal Hope Inez Irene Kip Kayla Luke Lori Beginning third Composing Composing Composing Composing Composing Composing Composi ng Task Environment Task Environment Task Environment Middle third Composing Composing Composing Composing Composi ng Task Environment Task Environment Task Environment Final third Composing Composing Composing Composi ng Task Environment Task Environment Task Environment Task Environment Task Environment 211 Table 20 Summary of All Cases Composition Moves frequency of use in thirds Beginning Third % Middle Third % Final Third % Total % Composing 17 16 13 46 Revision 1 3 2 6 Task Environment 13 14 15 42 Resource level 2 2 2 6 Total % 33 35 32 100 The o verall pattern across Tables 20 and 21 are the prominence of composing and task environment moves in the compositions of the students across all three time periods. Simply put, this meant that the eight students were largely focused on typing and technology tools, assignment materials, and text written so far in the task environment while compos ing their tweets. The beginning third of the tweets was largely shaped by composing moves (7 of the 8 students in Table 19; 17% vs 13% in column 1 in Table 20). The middle third was shaped by a mix of composing and task environment moves, with a larger num ber of students using composing moves (5 of the 8 students in Table 19; 16% vs 14% in column 2 of Table 20). And the final third was shaped by a re - mix of task environment and composing moves, with a slight increase in the number of students using task env ironment moves (5 of the 8 students in Table 19; 15% vs 13% in column 3 of Table 20). In sum, the prominence of composing moves by students decreased slightly from beginning to middle to final (i.e., from 7 to 5 to 4), while the task environment moves incr eased slightly from beginning to middle to final (i.e., from 3 to 3 to 5). Such patterns suggest that student moves represented a shift from getting words on the screen to manipulating the words once on the screen. The composition moves for revision proces ses and resource level were used much less frequently by the students across all three time periods. 212 Cross - Case Comparison of Emergent Features capture recordings, talk - aloud transcripts, video stimulated recall interview transcripts, student written artifacts, and field notes), eight elements associated with the tweet - composing technology and context were summarized and re - analyzed using Table 21. The eight elements were : (a) autocomplete , (b) red underline , (c) blue highlight , (d) , (e) magnifying glass activation , (f) black - box words , (g) copyright , and (h) reading URLs . To scaffold the analysis of emergent features, I designed a conceptual spectrum that indicated the degree of student understanding of features used or mentioned. The spectrum is informed by the concept of learning progressions in education (i.e., Mosher, 2011). The degrees of understanding were determined from student explan ations and/or observable use of features. The spectrum included six conceptual bands ranging from an accurate understanding on one end to a complete lack of feature awareness ( incognizant ) on the other. These conceptual bands were used to analyze each of t he eight emerging feature elements across the eight cases in Table 21. The conceptual spectrum band was defined and color coded for visual analysis. The first spectrum level, accurate conception , highlighted green in Table 21, indicates a student correctl y understands and uses the associated element consistently. The second spectrum level, patchy conception , highlighted yellow, indicates a student mostly understands, but utilizes the associated element inconsistently or excessively. The third spectrum leve l, partial conception , highlighted purple, indicates a student has limited or incomplete understanding of the associated element. The fourth spectrum level, false conception , highlighted red, indicates a student misunderstands the associated element. The f ifth spectrum level, strategy - creating conception , highlighted blue, indicates a student has varying degrees of understanding and invents/creates a 213 strategy/response to work with the associated element. Finally, the sixth spectrum level, incognizant concep tion , highlighted white, indicates a student is not aware that the associated element exists. The comparison of each student, for each element, along the conceptual spectrum, follows. Table 21 Cross - highlighted to indicate patterns Hal Hope Inez Irene Kip Kayla Luke Lori Autocomplete Quicker to spell words than typing Words you might be trying to spell Sometimes forgets to use this feature words forgets to use this feature Words suggested by who made Single letter activation strategy Multi - letter activation strategy Possible strategy to add space between words Strategy to type longer words faster Red Underline Indicates spelling mistake Indicates spelling mistake Indicates spelling mistake unnoticed underlines correct spelling a word is too close to another word strategy: double spaces, deletes a space 214 Blue highlight Tells you something is spelled wrong real word Indicates Changes the word from what she wants to write Clicking on blue highlight tells computer there and she will go back and fix it A reminder to go back to that word So you know where you are with your typing strategy: double spaces, deletes a space Technology as ism of technology ism of technology Magnifying glass activation To view small areas To view small areas punctuation activation 215 Black box words Provides words with similar spelling pattern students are not allowed to use those words blindness (feature not noticed) blindness (feature not noticed) Copyright students should not copy tweet ideas Reading URLs important word Word to sound out 216 Autocomplete The autocomplete feature was available on the iPad keyboard used by each student. When a student typed on the keyboard the autocomplete noticed these suggested words and thought one fit his/her writing intention, s/he could simply tap a word on the wordbar displayed across the top of the keyboard and the complete word would appear in the text box followed by an automatic space so the next word could be entered. The cross - case comparison of autocomplete findings is displayed in the first row of Table 21. Th e row is subdivided into six sub - rows that represent the bands of the conceptual spectrum, starting with the top sub - row, with each successive band below it (i.e., from accurate conception across the top sub - row, moving to the next band in the next sub - row , with incognizant conception across the bottom sub - row). The autocomplete feature was identified as emergent for five of the eight students: Hal, Hope, Inez, Irene and Kip. Two students, Hal and Inez, were found to have an accurate conception of the aut ocomplete capability for displaying words the writer is trying to spell and to add words to a tweet quicker than typing a full word letter by letter. Three students were found to have a patchy conception , with Inez and Kip admitting they sometimes forgot a bout the autocomplete feature, while Irene used the autocomplete excessively, selecting an autocomplete word suggestion for all but two of the words in her tweet, including words she had already typed out completely. One student, Inez, provided a false con ception when she explained that the autocomplete students expressed strategy - creating conceptions for the autocomplete feature. Hal consistently typed a single letter before selecti ng an autocomplete suggestion where as; Hope consistently 217 typed multiple letters before making a selection; Irene appeared to use the autocomplete feature to add spaces between her words; Kip reported using the autocomplete feature only when typing long wo rds. No evidence was found to indicate students had partial or incognizant conceptions of the autocomplete feature. Red Underline The red underline feature was available on the iPad keyboard used by each student. When a student typed on the keyboard the r ed underline appeared under the text which did not follow typical English spelling and/or syntax, such as homophones. If the student typed a string of letters that did not match an autocomplete suggestion a red underline would appear once the spacebar was tapped, notifying the student of a possible spelling error. If the student tapped the spacebar a second time the red underline would disappear. If the student tapped on the underlined text it would become highlighted pink and a black box would appear above the pink The cross - case comparison of red underline findings is displayed in the second row of Table 21, which, like the autocomplete row, is subdivide d into six sub - rows that from top - to - bottom represent the conceptual spectrum (from accurate conception to incognizant conception ). The red underline feature was identified as emergent for five of the eight students: Hal, Inez, Irene, Kip and Lori. Three students, Hal, Irene and Kip, were found to have an accurate conception of the red underline feature for indicating a spelling error. Three students were found to have a partial conception. Inez explained that the red underline because the spacebar is pressed. Kip explained that sometimes correctly spelled words are 218 limited understanding of homophones. Finally, Lori explained that the red underline appears when a space is not placed between words. This makes sense given that the combination of two words without a space is more li kely to produce misspellings rather than compound words. One student, Lori, provided an strategy - creating conception when confronted with a word marked by a red underline after pressing the spacebar. She proceeded to press the spacebar a second time to remove the red underline, then deleted the extra space before typing her next word. No evidence was found to indicate students had patchy, false or incognizant conceptions o f the red underline feature. Blue Highlight The blue highlight feature was available on the iPad keyboard used by each student. When a student typed on the keyboard the blue highlight appeared once an unexpected letter was added to a string of letters typically found in a word. For example, if the string of letters typed possible mistake. In response to this possible mistake, the autocomplete wordbar would then highlight in white the word located in the center of the three suggestions. If the writer pressed the automatically replace the blue highlighted blue highlight would not reappear over ressed the word would display a red underline . 219 The cross - case comparison of blue highlight findings for each student is displayed in the third row of Table 21. The blue highlight feature was identified as emergent for five of the eight students: Hope, In ez, Irene, Kip and Lori. One student, Hope, was found to have an accurate conception of the blue highlight partial conceptions . Irene said the blue highlight given that the highlight may appear before the writer has finished typing a word. But, it was coded a partial conception because Irene did not realize th e blue highlight of the unfinished word was a signal to the writer to consider the autocomplete suggestions. Relatedly, Lori explained that the blue highlight her understanding was a partial conc eption of why the word changed. Three students -- Hope, Inez and Kip -- provided data indicating false conception. Hope, for example, explained that clicking on the blue highlight blue highlight ] is there and she wi Lori provided evidence of strategy - creating conception for the blue highlight feature that paralleled that which she gave for the red underline feature. For instance, when confronted with a blue highlight word after the initial spacebar press, she retyped the word originally typed, then pressed the spacebar again. At this point the red underline appears. She pressed the spacebar a second time to remove the red underline , then deleted the extra space before typing her next word. No evidence was found to indicate students had patchy or incognizant conceptions of the blu e highlight feature. 220 Technology as Knowing Other is a feature of the language used by students to explain - case comparison of is displayed in the fourth row of Table 21. This feature was identified as emergent for two of the eight students. In both cases, the students used language that anthropomorphized a feature of the digital technology being used. Such a conception of technology was evidence of a false conception because the students misunderstood ez asked about the students had accurate, patchy, partial, strategy - creating, or incognizant conceptions of Magnifying Glass Activation The magnifying glass activation feature was available on the iPad used by each student. To activate the magnifying glass when typing, the writer had to firmly press a finger on or between letters and words on the iPad screen. The location was usually where a typing error existed or a pl ace where a letter or word needed to be added or deleted. The firm finger press activated a magnified circular view (approximately an inch in diameter) of text on the screen. While maintaining the firm finger press, the writer slid her finger across the te xt. As her finger slid, the cursor followed. The magnification remained as long as the finger remained firmly pressed on the iPad screen. When the writer removed her finger from touching the screen, the cursor was inserted at the last touch point. 221 The cros s - case comparison of the magnifying glass activation findings for each student is displayed in the fifth row of Table 21. The magnifying glass feature was identified as emergent for two of the eight students: Hal and Irene. Both students explained that th e purpose of this feature was to view small areas, which was an accurate conception of the feature. In addition, Hal explained that end - of - sentence punctuation was necessary to activate the magnifying feature, which is a false conception . Furthermore, he m ade an attempt at one point to demonstrate how the magnifying glass would not appear unless a period was added to the end of his sentence. After adding the period, he tried the magnification feature, which activated the magnifying glass. No evidence was fo und to indicate students had patchy, partial, strategy - creating, and incognizant conceptions of the magnifying glass feature. Black Box Words The black box words feature was available on the iPad keyboard used by each student. Three variations of the blac k box words were observed. First, if the writer firmly pressed a finger to the iPad screen at the location of a word, a black box with three command choices appeared, bla ck box autocomplete suggestions, then deleting the space to position the cursor next to the last letter of the autocomplete word, a black box would appear with additional word choices. The cross - case comparison of the black box words findings for each student is displayed in the sixth row of Table 21. The black box words feat ure was identified as emergent for three of the eight students: Hope, Irene and Hal. Hope, for instance, reasoned that the black box words accurate conception . 222 black box , which was evidence of a false conception . Finally, both Hope and Hal admitted that they had never noticed the black box word feature until it was pointed out during their video stimulated recall interview. both students provided evidence of an incognizant conception of the black box words feature. No evidence was found to indicate students had patchy, pa rtial, strategy - creating, or incognizant conceptions of the black box words feature. Copyright Copyright was a feature of the composing process expressed by a student when peers poached ideas from her tweet to use in their own. The cross - case comparison o f copyright is displayed in the seventh row of Table 21. This feature was was identified as emergent for one of the eight students . Hope brought up the issue of copying when asked which parts of tweeting and some people actually copy me and then composing with technology was evidence of partial conception because of Internet issues related to open access, fair - use and co pyright. No evidence was found to indicate students had a ccurate, patchy, false, strategy - creating, or incognizant conceptions of copyright . Reading URL Reading URL is a feature of the process used by students to indicate their understanding of what a fea ture is. The cross - case comparison of reading URL is displayed in the eighth row of Table 21. This feature was identified as emergent for one of the eight students: Hal. While reading a tweet aloud, Hal sounded out all the letters in the embedded URL as if decoding an unknown or challenging word, indica ting his false conception of a URL as a word that signified 223 spoken or written meaning in the conventional sense. No evidence was found to indicate students had accurate, patchy, partial, strategy - creating, or incognizant conceptions for reading URL . Summa ry of Cross - Case Comparison -- Overall The data interpreted for the cross - case comparison of the eight cases provided categories of similarities and differences in each of three parts. First, the writing processes data (i.e., thoughts made known through ta lk - alouds and video - stimulated recall) suggests each of the eight cases were similar in terms of motivation , goal setting , planning , and writing schemas . Second, the composition moves data (i.e., actions recorded by the screen - capture video and further un derstood by talk - alouds) suggests both similarities and differences across cases. At the complete tweet level it is easy to notice various commonalities based on quantity of codes for each of the composition moves elements: composing , r evision processes , t ask environment , and resource level . But, when evaluating a composition move elements at the letter/word/phrase level experience. Finally, the emergent features data (i.e ., thoughts and actions from all data sources) suggest a spectrum of conceptual knowledge (from accurate conception to incognizant conception ) related to eight elements associated with the tweet - composing technology features and the tweeting context: (a) a utocomplete , (b) red underline , (c) blue highlight , (d) technology , (e) magnifying glass activation , (f) black - box words , (g) copyright , and (h) reading URLs . Chapter Summary The results of this study provided some answers to the researc writing processes and composition moves made by second graders when composing tweets for 224 - case descriptions provided a look into the nuances of n moves, and emergent findings. A summary of each case then a summary of the cross - case comparisons follow. Individual - Case Summaries Hal is a resourceful and efficient, just - get - it - done type tweeter, who was motivated and goal oriented with an in - advance plan and the ability to plan in - the - moment with support, displaying a 2nd - grade expert level writing schema composition moves in composing involved quick multi - letter, spacebar and backspace typing that autocomplete suggestions for the retyping of several text sections. He easily edited a word underlined in red and considered revising task en vironment composition moves were mostly keyboard open or closed and task materials which included moments looking at the iPad screen and time spent using the speech - to - were few, maintaining a fo cus with limited pausing and rereading before adding on and before publishing his tweet. Finally, Hal revealed an emerging understanding of the iPad keyboard affordances inventing explanations to how and why things work without fully understanding the tech nology. Hope is a conscientious and confident reader - pleaser type tweeter, who was motivated and goal oriented with an in - advance plan and the ability to plan in - the - moment , displaying a 2nd - grade expert level writing schema knowledge of Twitter. co mposition moves in composing involved quick multi - letter, and single - letter typing - strategy to activate the autocomplete suggestions for the typing of several text sections. She easily noticed and edited typing miscues and revised her text by adding a word to an existing 225 task environment composition moves were mostly keyboard searching, autocomplete, and keyboard open or close. resource level composition moves included limited pausing and mul tiple rereadings at the completion of each sentence, after fixing a miscue, before adding on, and before publishing her tweet. Finally, Hope revealed an emerging understanding of the composing technology affordances and a growing concern for copyright. Ine z is a confident tweeter even though she struggles with spelling. Inez was motivated and goal oriented with an in - advance plan and the ability to plan in - the - moment , displaying a 2nd - grade expert level writing schema knowledge of Twitter. compositio n moves in composing involved quick multi - letter , spacebar , and single - letter typing . Her revision work involved deleting single letters and editing misspelled words and revising her text to make is task environment composition moves were mostly task materials (looking at external documents and the iPad screen) and keyboard searching resource level composition moves were few, maintaining a focus with a single beginning of tweet pause and rereading af ter making changes, before adding on and before publishing her tweet. Finally, similar to Hal, Inez has an emerging understanding of the iPad keyboard affordances inventing explanations to how and why things work without fully understanding the technology. Irene is a confident tweeter. She was motivated and goal oriented with an in - advance plan and the ability to plan in - the - moment , displaying a 2nd - grade expert level writing schema knowledge of Twitter. composition moves in composing involved multi - letter and single - letter typing with limited spacebar use. Her revision work involved editing and deleting to task environment composition moves were mostly keyboard a utocomplete (generating most of her text), task materials (looking at external documents and the iPad screen), and keyboard 226 searching resource level composition moves were few, maintaining a focus with a beginning and mid - tweet writing pa use and rereading after making changes and before publishing her tweet. Finally, Irene has an emerging understanding of the red underline, blue highlight, and magnifying glass features. Kip is a confident tweeter even though, at first, he seemed nervous . H e saw himself as an author, was motivated and goal oriented with an in - advance plan and the ability to plan in - the - moment with support, displaying a 2nd - grade expert level writing schema knowledge of Twitter. composition moves in composing included quick multi - letter and single - letter typing , and - autocomplete suggestions for the typing of the edited for word choice and typing miscues in conjunction wi th multiple deletes task environment composition moves were mostly keyboard searching , task materials , and keyboard open or close resource level composition moves included multiple pausing and three rereadings after fixing miscues, a nd toward the end of his tweeting. Finally, Kip has an emerging understanding of the iPad keyboard affordances inventing explanations to how and why things work without fully understanding the technology. Kayla is a resilient problem solver type tweeter wi th some difficulty navigating the keyboard. Kayla was motivated and goal oriented with an in - advance plan and the ability to plan in - the - moment , displaying a 2nd - grade expert level writing schema knowledge of Twitter. composition moves in composing involved a high number of single - letter typing , with some multi - letter typing and spacebar . Her revision work included editing spelling miscues and deleting unwanted text and her task environment composition moves were mostly keyboard open or close , keybo ard searching , autocomplete and cursor reposition resource 227 level composition moves included limited pausing and multiple readings at the end of her first sentence and before publishing her tweet. Luke is a tentative yet capable tweeter. N ervous and unsure of his abilities yet competent in his knowledge of Twitter and the ability to compose a complete tweet. Luke was motivated and goal oriented with an in - advance plan and the ability to plan in - the - moment , displaying a 2nd - grade expert leve l writing schema knowledge of Twitter. Conversely, Sometimes Luke did composition moves in composing included a high number of multi - letter typing and spacebar with two single - letter typing occurrences that may have been associated with visually similar letters. His revision work included deleting unwanted text and one idea revision task environment composition moves were mostly keyboard searching , keyboard open or close , and task ma terials when resource level composition moves included limited pausing and one reading before publishing his tweet. Lori is a dependent yet confident tweeter, reluctant and unsure of her abilities yet competent i n her knowledge of Twitter and ability to compose a complete tweet. Lori was motivated and goal oriented with an in - advance plan and the ability to plan in - the - moment with significant support, displaying a 2nd - grade expert level writing schema knowledge of Twitter. composition moves in composing included frequent typing spacebar , multi - letter , and single - letter typing . She noticed and edited spelling errors and easily deleted multiple typing task environment composition moves were most ly keyboard searching and task materials. resource level composition moves included multiple pausing and readings only when prompted. 228 Cross - Case Comparisons Summary The data interpreted for the cross - case comparison of the eight cases provi ded categories of similarities and differences in each of three parts. First, the writing processes data (i.e., thoughts made known through talk - alouds and video - stimulated recall) suggests each of the eight cases were similar in terms of motivation , goal setting , planning , and writing schemas . Second, the composition moves data (i.e., actions recorded by the screen - capture video and further understood by talk - alouds) suggests both similarities and differences across cases. At the complete tweet level it is easy to notice various commonalities based on quantity of codes for each of the composition moves elements: composing , r evision processes , task environment , and resource level . But, when evaluating a composition move element at the letter/word/phrase level emergent features data (i.e., thoughts and actions from all data sources) suggest a spectrum of conceptual knowledge (from accurate t o incognizant ) related to eight elements associated with the tweet - composing technology features and the tweeting context. 229 Chapter 5: Discussion, Limitations, and Implications Evidence indicates that young children are writing online (e.g., Clark & Dug dale, 2009), but little consideration has been given to the actual writing processes and composition moves made by these children, especially when they compose short form writing online. Such a paucity of research is a problem for theoretical development a nd pedagogical design. This study addressed both these problems by posing the question: What are the writing processes and composition moves made by second graders when composing tweets for online publication? by using (and adapting) the most comprehensive model of offline writing (Hayes, 2012) as a lens to closely examine the online writing processes and composition moves of eight second - grade ) in situ as they compose short form writing for their class Twitter account. Through this lens I conducted a multi - case study, analyzing data from field notes, written artifacts, screen capture, talk aloud transcripts, and video - stimulated recall intervie w transcripts. The result is an account of the "why" (writing processes) and "how" (composition moves) of young writers writing online. writing by describing one approach to short - form writing online and illustrating how it shapes the writing processes and composition moves of young children (cf., Maggio, Lété, Chenu, Jisa, & Fayol, 2012; or Merchant, 2005). While other approaches could have resulted in different accounts o f what young writers do as they write online, the pedagogical approach utilized in this study provides another lens for understanding how one particular approach shapes the online composing of children. 230 I begin this chapter by discussing how my study resul ts suggest Twitter writing is similar to other writing according to the modified Hayes (2012) model. I then discuss how Twitter writing is unique in some ways, addressing elements not accounted for by the Hayes model. I writing is shaped by the curriculum. I conclude this chapter first identifying limitations of the study and then offering implications for future instruction and research. The results of my analysis i - form (2012) comprehensive model of the writing process (see chapter 2) -- so it also included the processes an -- I examined the Twitter writing using a two - part analytic framework: writing processes (i.e., thoughts made known through talk - alouds and video - stimulated recall), and composition moves ( i.e., actions recorded by the screen - capture video and further understood by talk - alouds). I discuss these similarities in the following sections, first with writing processes . Writing Processes: Motivation The writing - processes results on motivation indi cated that, in general, the eight tweeters technical expertise and Luke occasionally not motivated to tweet. Having fun is similar to prior research on young children composing online (McGrail & Davis, 2011; Merchant, 2005b). And composing (Burnett & Mayers, 2006; McGrail & Davis, 2011). Thus, the results of this study 231 confirm pre vious scholarship and extend it by providing evidence that motivation is a part of the Writing Processes: Goal Setting The results on goal setting twin goals. While prior research on online composing does not report goal setting as a writing process voiced by children (e.g., Burnett and Mayers, 2006 and Merchant, 2005b), there is evidence in the literature that offline composing does so (e.g., Graham, McKeown, Kiuhara, and Harris, 2012). Thus, the results of this study confirm previous scholarship and extend it by providing ev idence that goal setting is an integral part of the writing process when children write online with Twitter. Writing Processes: Planning The results indicated that two forms of planning online composing: in - advance an d in - the - moment . All eight tweeters voiced in - advance planning by first selecting an object as the focus for their tweet, then taking a photo of it. Other forms of in - advance Niiya, & Wa rschauer, 2015). Thus, the results of this study confirm previous scholarship and extend it by providing evidence that in - advance planning is a part of the writing process that In addition, all eig ht students voiced in - the - moment planning, indicated by their thoughtful composing of each sentence and the generating of each letter and word. Prior research in - the - moment planning (e.g., Burnett & Mayers, 2006; Bereiter & Scardamalia, 1987), but the general description of 232 in - the - momen t planning is indeed an integral process of their writing. Thus , the results of this study confirm previous scholarship and extend it by providing evidence that both forms of planning are an integral part of the writing process when children write online with Twitter. Writing Processes: Writing Schemas Finally, the w riting - processes results generally indicated that writing schemas were represented in three ways: (a) used genre knowledge to compose short - form writing including multimo dal elements, (b) showed an awareness and growing understanding of followers , likes , cybersafety , and networking , and (c) developed a fixed topic text structure using their photo as a knowledge in children's online writing is alluded to (e.g., Dowdall, 2009; Lindstrom & Niederhauser 2016). Of particular note, though, is recent work by Takayoshi (2018) which incorporates writing schemas into a model of adult online composers who are simu ltaneously engaging with multiple online spaces. Thus, the results of this study confirm previous scholarship and extend it by providing evidence that rt form. Composition Moves: Composing The composition - moves results for composing indicated that all eight tweeters generated multi - letters and single letters as well as using the spacebar and backspace when initially composing at the character - level is limited (Dow dall, 2009; Lindstrom & Niederhauser, 2016; McGrail & Davis, 233 character - confirm previous scholarship and extend it by providing evidence that composing at the character - previous studies reported. Composition Moves: Revision Processes The results for r evision processes indicated that the eight tweeters recursively used editing and deleting at the character level when manipulating existing text on the screen. Prior es when manipulating existing text such as the conventions of punctuation and spelling (e.g., McGrail and Davis, 2011; Merchant, 2005b). This result is similar to the attention children paid to surface - level convention problems found by MacArthur, Graham, writers. Correspondingly, Takayoshi (2018) found that young adults recursively used editing and deleting behaviors for both surface - level conventions and deeper - level craft decisions at the character level in her study of online writing. Thus, the results of this study confirms previous scholarship by providing evidence that recursive revision processes are a factor in young Moreover, the recursive edit and delete behaviors used by the eight tweeters were mostly in response to the tactical cues provided by the transcription technology to detect problems (i.e., the red underline and blue highlight (1983) findings that tactical cues in offline writing support the increase of student revision processes (e.g., correction of misspelled word). Compared to their findings, my results showed that the eight tweeters had differing responses to the tactical cues ( red underline , blue highlight ). F or example, tactical cues were misunderstood, misused, purposefully ignored, or unnoticed, 234 similar to what is represented in the model of Flower and Hayes (1986). Thus, the results of this study confirm and extend the range of possible responses young writ ers have to available keyboard tactical cues and the knowledge used during revision processes . Composition Moves: Task Environment The results for task environment indicated that the eight tweeters used various combinations of keyboard searching , keyboard open or close , keyboard autocomplete , cursor reposition , and task materials to produce their short - form writing. In other words, the eight task environment was comprised of scanning the keyboard for letters, toggling among keyboards, selecting autocomplete suggestions, moving the cursor from one location to another, and spending time looking at the iPad screen or other assignment materials. These detailed res task environment elements implied in previous research (e.g., Merchant, 2005b; Wollman - Bonilla & Carpenter, 2003). Notably, Takayoshi (2018) found task environment elements integral to the online composing of adults (i.e., engaging with multiple online space notifications). Thus, the results of this study confirm and extend previous scholarship by providing nuanced evidence that the task environment is a factor in young Composition Moves: Resour ce Level Finally, the composition - moves results for resource level indicated that nearly all of the eight tweeters read and reread their Twitter text to (a) check that the message makes sense (b) think of what else to add, and (c) check for mechanical and local problems (e.g., spelling and & Myers, 2006) and offline writing (e.g., Bereiter and Scardamalia, 1983). Thus, the results for these aspects of resour ce level are confirmed by previous research. 235 In addition, pausing was a resource level composition move made by all eight tweeters, with patterns of pause duration and timeline location unique to each tweeter. Previous research found that both types of pau sing are associated with more thoughtful planning and better text quality in children writing offline (Limpo & Alves, 2017). Other studies have found that pauses in offline writing are associated with the length and qualities of words previously written b y children (Maggio et. al., 2012). In sum, the results of this study confirm previous scholarship by providing evidence that pausing behaviors constitute the resource level composition moves in g is Unique in Some Ways - form - and long - form writing. Based on their distinctive explanations of iPad and Twitter attributes, the tweeters voiced eight writing processes and composition moves that were not elements of the modified Hayes (2012) writing model: (a) autocomplete , (b) red underline , (c) blue highlight , (d) , (e) magnif ying glass activation , (f) black - box words , (g) copyright , and (h) reading URLs . I located the conceptual veracity of these elements, called emergent features, along a spectrum (ranging from accurate to partial, patchy, false, strategy - creating, and incogn izant conceptions ) so a non - binary band of codes could be used for visual analysis. These emergent eight emergent features are discussed next. Emergent Features : Autocomplete, Red Underline, and Blue Highlight The results for emergent features indicated most tweeters expressed a combination of accurate, patchy, false, and strategy - creating conceptions for the autocomplete , red underline , 236 and blue highlight features. Because these features are not accounted for in the Hayes (2012) model, and because a range of understandings about these features were voiced by some of the eight tweeters in the present study (and unexamined or unreported in previous studies), my results provide a preliminary basis for claiming that emergent features such as autocomplete , red underline , and blue highlight are not uniformly understood -- or used -- by young children composing online. Consequently, a more contemporary writing model f or young short - form social media writers would account for the conceptual spectrum that these three emergent features could have on children's online composing. Emergent Features: Magnifying glass activation, Black Box Words, Copyright and Reading URLs In addition, there were other emergent features expressed by some tweeters that were a combination of accurate and false conceptions for the magnifying glass activation feature, black box words , copyright , and reading URLs . Because these features were also n ot accounted for in the Hayes (2012) model, and because a range of understandings about these features were put into words by tweeters in the present study (and unexamined or unreported in previous studies), my results provide a preliminary basis for claim ing that emergent features such as magnifying glass activation , black box words , copyright , and reading URLs are not uniformly understood -- or used -- by young children composing online. Consequently, a more contemporary writing model for young short - form soc ial media writers would account for the conceptual spectrum that these four emergent features could have on children's online composing. Em ergent Features: Technology as Knowing - O ther Finally, some tweeters expressed what I initially coded as false concep tions about knowing other 237 words, these tweeters used anthropomorphic language when talking about technological features of the iPad and Twitter. This finding is not found in writing but is well represented in other research fields (e.g., Airenti, 2015, says computers are most often anthropomorphized). Moreover, research shows that anthropomorphization of computers is not the result of "false conceptions" per se, but rather the results of talking about interactive processes (between people and computers) with language structures that make it natural to talk about interaction in human terms. In short, our language makes it convenient to talk about technology in this way (Papert, 1988). Therefore, a more contemporary writing model for young short - form social media writers would include anthropomorphic features as they emerge featuring the symbiotic interactions between writer and digit al technologies. Finally, the results of my individual and cross - case analyses indicate similar patterns writing processes (motivation, goal setting, planning, and writing schemas). These similarities Twitter curriculum shaped the tweeting of all eight children as they wrote (Bazerman, et al., 2017). This curriculum, represented in the teacher - created handout in Figure 41, described the procedures (i.e., prewrite, draft, revise, edit, publish), structure criteria composing a tweet. Over time, the teacher modeled the elements of this curriculum, gradually releasing responsibility for tweeting in this way to the students, so they eventually used the teacher - created handout as a tool to guide their tweeting (Pearson & Gallagher, 1983). As a resul tweeted. 238 Figure 41 Teacher - created handout for Tweeting Curriculum and Motivation The evidence for this curriculum - shaping - the - tweets conclusion is visible when the writing processes (motivation, goal setting, planning, and writing schemas) were used to examine the teacher - lum allowed tweet. As previous research indicates, choice and autonomy - enhancing experiences are powerful an, & Roth, 2002). Moreover, authentic Murray, 1985; Graves, 1994). And teaching students to tweet for a real purpose, to a real audience, with authentic feedb 239 Curriculum and Goal Setting When asked about goals for tweeting, all eight tweeters reported similar goal setting processes: to create a mess age that would make sense to their audience about content from their learning. This goal setting intent has a direct link to the criteria for tweeting listed in the teacher - created handout, which was to tweet about their learning and to check that their se ntences make sense. The effect of this criteria was to teach students to reread their tweet before publishing it and to ask themselves if they had achieved the goal of making sense in their tweet. Previous research has indicated that setting goals for read ing to achieve a particular writing task is an integral part of the writing task environment (Hayes, 2012; Hayes, Flower, Schriver, Stratman, & Carey, 1987). The consistent reminders to check for meaning when reading and writing was another way in which th Curriculum and Planning In terms of planning in - advance planning as they composed a tweet. The planning process began with guidance to: (a) think about what they might want to tweet about, (b) share their ideas with a classmate, and (c) take a picture of an object best representing their selected topic. This planning aspect of the curriculum included time for students to talk about their plans and for the teacher to provide direct instruction on how to use the technology to take a photo and add the photo to a tweet. Previous research indicates the powerful influence th at planning has on the form and structure of 240 Curriculum and Writing Schemas writing schemas by all eight tweeters, which resulted in their: (a) using genre knowledge to compose short - form writing including multimodal elements; (b) showing an awareness and growing understanding of followers, likes, cybersafety , and networking ; and (c) developing a fixed topic text structure using - shaping process occurred during daily discussions of their class t witter feed, which was projected as an enlarged shared - reading document for all the class to see. During these discussions the teacher directly taught about using genre knowledge to create meaningful messages with a fixed topic text structure limited to 28 0 characters (short form writing). Her two - step method was to first model her own tweeting process, and then follow it with shared writing of a class tweet. The teacher and children also talked from time to time about followers , likes , connecting with othe rs in a network and cybersafety during these daily discussions. These daily, whole - class These consistently implemented curricular element s were part of daily writing instruction with Twitter and over the course of the school year became an integral part of everyday practice curriculum shaped the way s tudent motivation , goal setting , planning and knowledge of writing schemas was enacted in their Twitter composing. This finding is similar to research that - form offline writing (i.e., Harris & Graham, 2016). This study extends that finding to short - form online writing. 241 Limitations Although actions were followed to increase trustworthiness, the results of this multi - case study are limited to the particular circumstances from which it was conducted. To increase trustworthiness , both composition moves and writing processes were clearly defined based on existing models and theories of off - line, long - form writing (e.g., Bereiter and Scardamalia, 1987; Hayes, 2011; 2012; Hayes & Olinghouse, 2015) and short - f orm online writing (e.g.,Takayoshi 2015, 2018). Ultimately, for this case study, the Hayes (2012) Cognitive Writing Processes Model was selected as a theoretical lens, with necessary modifications based on seven clearly described criteria. Also, the custom ary three - to - five case - study participants (Creswell & Poth, 2018) was increased to eight participants. Finally, multiple sources of evidence, thick description of each case and recursive, constant comparative analysis were used. With each subsequent analys is, any overlooked action in the screen capture was noted, amending the data to accurately reflect the composition moves of each tweeter (Yin, 2018). Even with these measures to strengthen trustworthiness, certain concerns remain. Second - by - second writing behaviors may have been overlooked in the data analysis phase, given the volume of video data. Data saturation may not have been achieved, especially given the diversity of possible composition moves at the character level of composing and various task en vironment writing to children and the established relationship with the classroom teacher, providing es that were not obvious to an outsider, may have limited the results. 242 Implications This study heeds the largely unanswered call to use moments of composing as objects of study (Takayoshi, 2018). Consequently, it has methodological, theoretical, and pedagogical implications. Methodologically, this study captured those moments of composing by using screen - capture video on an iPad screen while simultaneously capturing video of the iPad user and corresponding audio. The result is a study that illustrates a method worth refining so that other detailed analyses of composing moments can be condu cted. Additional theoretical and pedagogical implications follow. Theoretically, the results of this study suggest that the lens used for data analysis was - form writing experiences in online spaces. Fur ther study -- with a wider variety of participants, curricula, settings, and technologies -- would develop the modified Hayes (2012) model even further as a tool for theorizing the processes and moves of short - and long - form writing in off - and on - line spaces . Such a project would be accrual of studies that expand and refine ou r models of composingwould, in the long run, lead to a re - Pedagogically, the results of this study indicate that the linear writing process espoused by many teachers -- start with drafting/p lanning, composing, revising, editing, and finish with publishing (Hayes & Olinghouse, 2014) -- is out of step with the practice of composing online. Even previous scholarship on composing offline has suggested that teaching writing in a stepwise manner doe s not represent the actual work of writers (e.g., Murray, 1985; Graves, 1994). The actual work of young writers tweeting in this study shows the iterative and 243 idiosyncratic ways their writing processes and composition moves play out online, just as they wo uld offline, but in a wider range of ways because of what the technology affords. In particu lar, the data from this study imply several pedagogical possibilities for supporting the writing development of children composing short - form prose online using a t echnology like Twitter. Two examples are provided below in Figure 42. In each, one of the composition moves observed in the data serves as the basis for: formulating a tweeting criteria (stated in the left column), that if mis - composed by a child tweeting (illustrated in the middle column), the teacher can provide pedagogical support focused on (stated in the right column). Figure 42 Pedagogical possibilities for supporting children composing short - form prose online Composition Moves -- Editing If your tw eeting criteria include ... And students type something like Then your instruction should focus label summary by310#read [instead of By 310 #read ] Concept of hashtag Composition Moves -- Task Environment If your tweeting criteria include ... And students include something Then your instruction should focus represents unrelated content Selecting an image that of focus Future Research Results of the present study suggest when composing a tweet to publish on their class 244 writing, as well as some new elements, which were made possible by the afford ances of adult writers in an off - line context, it was necessary to modify his model for use as a theoretical lens for data analysis. The effect was to extend and r elemen ts divided into two categories: (a) writing processes : motivation, goals setting, planning, and writing scheme; and (b) composition moves : composing, revision processes, task environment, and resource level. In addition, an additional set of features emerged that did not fit neatly into either of these categories. These were labeled emergent features , and provided an evidence - tha t spirit, I discuss future research possibilities that emerge from the adapted mo del and results of this study. A number of new questions could be posed by subsequent research on writing. For example, at what age do writers composing in online spaces expe rience the layering distractions described by Takayoshi (2018)? What type of instructional interventions for children might best prepare them for these layering distractions? What are the theoretical implications for this type of writing context in terms o f motivation , goal setting , and planning ? For instance, do contemporary composers set goals or create in - advance plans while experiencing simultaneous layers of online spaces and frequent notification distractions? In addition, methodological refinements c ould be made. For instance, further research might design a leaner way to describe the flow of composing while also illustrating the complex and many times almost simultaneous composition moves. Relatedly, an improved approach for preserving and accounting for the complex and individually unique nature of the writing experience might be developed to map out writing profile criteria for qualitative comparative 245 analysis QCA (Ragin & Davey, 2016). And finally, an approach to systematically identify profiles of young writers composing short - form writing for an online networked space may help identify unique learning needs of writers. Regarding elements of the Hayes (2012) adapted model, future research could examine an element such as the task environment to und erstand what aspects of an environment assist the writer best or least, or not at all (Clark & Salomon 1986). With this particular element, the question remains, were the task environment features experienced by the eight tweeters a candidate for internali zation? Does the technology captured in the task environment subcategories of the modified Hayes (2012) model somehow overtly model the writing process? If so, how? If not, why? In addition, Takayoshi, (2018) discussed task environment related experiences such as the immediate response to the frequent notification, interrupting the writing process, yet adult online writers developed effective responses. Considering these adult online writing experiences, what are the implications for teaching children self - regulation in anticipation of this possible constant disruption by notifications? The eight tweeters in my study task environment. Finally, future research may c onsider already existing data -- such as that from this study -- through differing conceptual lenses. One fruitful lens might be the issues initially raised by Fayol - might be one more fin ely attuned to the cognitive strategies related to writing (problem - solving, decision making, searching, questioning) and the work of processing information competing for limited space in short term memory (Jonassen & Reeves, 1996). A final approach, sugge sted by 246 Hayes and Berninger (2014), is to examine writing for social media (such as Twitter) as having more in common with conversational features than formal school writing by considering the data from the present study through a conversational process th eoretical lens 247 APPENDICES 248 APPENDIX A: Assent Script CHILD ASSENT TEMPLATE Verbal assent script for below minimal risk activities (young children) : Hi, I am Mrs. Marich from Michigan State University. I am trying to learn something about what children think and do when they write a tweet and post it on Twitter. I would like you to help me with this research. You do not have to do it, if you do not want to. Nobody will be upset with you if you decide you would rather do so mething else. Does this sound like it would be something you would like to do? Thank you. Assent Template for Ages 8 to 12 : Young Tweeters Tweeting Person leading the study: Holly Marich Why are we doing this research? The reason I am doing this research is to learn something about what children think and do when they write a tweet and post it on Twitter. Why are you being asked to participate in this research study? I am asking you to participate in this study because I know you twee t all the time for your class Twitter page. What will happen during the study? During the study the first thing we will do is meet in the next room where I have my laptop and iPad set up to video and audio record. We will sit down at the table together a nd I will tell you about all of the neat technology I will be using for the study. Then, I will video and audio record what you are doing and saying while you write an idea for a tweet on this notecard. After you are finished writing your idea I will conti nue watching you while you write your tweet. As you are working I will want you to talk to me about what you are doing. After you publish your tweet we will watch the video of you working. While we watch this video I will pause the video to ask you questio ns about what you were doing. After we watch all of the video and you have told me all that you think is important to teach me about what you do when you tweet, we will be done and you will go back to class. Risks and Benefits? 249 The good things about being in the study are that you will help me learn something new about children writing and this new information may help other teachers when they teach writing. A problem with being a part of this study is that you may get tired as we work together and yo u will be out of your classroom for about 45 min. But, we will take breaks if you need them and you will not miss recess or lunch with your class. Who will be told the things we learn about you in this study? The information I learn from this study will be shared with other teachers and parents that let their children tweet. It will also be shared with other researchers. Do you know how you use a number when you tweet instead of adding your name? I will do the same thing with the study. I will not use yo ur name at all in the study. What if you or your parents do not want you to be in this study? You can only participate if both you and your parents agree for you to be in the study. Nobody will be upset if you do not want to be in the study. It is yo ur decision. If you decide to be in the study, and later change your mind that is okay too. You can stop being in the study anytime you like. If you have any questions about the study, you can either tell your parents and have them talk to me, or talk t o me yourself. Here is my phone number and address: 775 - 293 - 0186 , hmarich@msu.edu , Holly Marich HC33 Box 33200 Ely, NV 89301 Documentation of Assent Would you like to participate in the study? Student Name _________________________________ Date __________________ If you sign your name on this page, it means that you agree to take part in this research study. You may change your mind at any time. 250 APPENDIX B: Talk Aloud and Interview Analysis Example Table 22 Talk Aloud and Interview Analysis Example Control Level W riting Processes Talk Aloud & Interview Analysis: Writing Processes: Thoughts made known through think aloud and interview Control Level Writing Processes Characteristics Note Motivation Disposition Expresses enjoyment in the task - body language (pointing to screen) then there will be tons more room When asked if she likes to tweet she said Yes!. I asked tweet about like what we learned in school and um tweeting Expresses enjoyment in the task - verbally Expresses interest in doing a good job at the task 251 APPENDIX B: Talk Aloud and Interview Analysis Example Control Level Writing Processes Characteristics Note Goal setting Purpose (follow teacher - generated tweeting expectations) Can articulate the purpose of the writing task: to communicate with others. going to write about and say h ow it helps you and say To help others learn about what they are learning. To help others learn about what he/she is doing. Planning Idea Generation Generating Tweet ideas before planning the text out and then... umm, I said and we can build things out of Legos...Well, while we were talking and thinking about things. And with the lego it helped me (picking up her lego structure) figure out what to write about. ...umhumm that's why I brought all of these things for me to help Planning Making a plan before writing 252 APPENDIX B: Talk Aloud and Interview Analysis Example Control Level Writing Processes Characteristics Note Writing Schemas Audience Can identify audience as known others some people actually read our tweets they actually like Can identify audience as unknown others Can identify audience as global, unknown other, and known other Genre Knowledge of General Tweet defined as a message posted to Twitter containing text, photos, a GIF, and/or video. Hashtag (summarize tweet content) (I'm not sure what she said there) if there is enough room they can put a hashtag but if they don't want to Mentions concept of Tweet Like multimodal including links, photos, GIFs, videos, surveys, emojis Two Photos Followers Networking 253 APPENDIX B: Talk Aloud and Interview Analysis Example Control Level Writing Processes Characteristics Note Cyber Safety Text Structure Knowledge telling, Flexible Focus Fixed Topic Strategy Table 23 Control Level Aspects of Genre Knowledge and Text Structure Control Level Aspects of Genre Knowledge and Text Structure Elements Code Notes Matches meaning of text Extend s meani ng of text Repeat s existin g text Assumes audience knows context (temporal) No explanation in the tweet to give context. Provides a picture of the Engineering design process with the assumption that readers know how this connects with her message about STEM. # hashtag 254 APPENDIX B: Talk Aloud and Interview Analysis Example Table 23 Control Level Aspects of Genre Knowledge and Text Structure photo X X emoji Table 24 Control Level Goal Setting Control Level : Goal Setting, The goal is dictated by the assignment expectations. This may also dictate text structure. Prompt Expectations Text from Tweet Notes *Elaborates Clarity of message What am I learning about As a reader, I can easily infer that she is learning about STEM. why am I learning it or how is it going to help me how to build and we can build thing X Extends with details - building with Legos. 255 Published Tweet Analysis: Composition Moves : What can be observed - screen capture and published tweet Figure 43 Published Tweet Example Table 25 Talk Aloud and Interview Analysis Example Composition Moves Process Level: Composing and Revising Transcriber composing such as Frustration Statements or obvious body language indicating the task is perceived as too hard Evaluator editing or revising such as 256 APPENDIX B: Talk Aloud and Interview Analysis Example Tab le 25 Inline planning planning what to write while writing Task environment the immediate social and physical surroundings such as keyboard technologies or the text written so far Meaning making Collaborators and Critics the tweeted message making sense Table 26 Talk Aloud and Interview Analysis Example Editing for Spelling, Spacing, Capital Letters, and P unctuation Process Level Editing for spelling, spacing, capital letters, and punctuation code Text explanation spelling adout B and d reversals she did not catch this one but caught another when typing build. This noticing may be attributed to letter position in word. spelling lagos This spelling was offered in the autocomplete and she selected it. It matched her initial spelling spell ing thing Should have been things spacing lagos.By202 No space between by and 202 257 APPENDIX B: Talk Aloud and Interview Analysis Example spacing lagos.By202 No space after end punctuation End punctuation used at the end of sentences Capital letters used for proper pronouns Other Interesting Notables : Spontaneous use of strategies Automation of task operations Technology features that assist the writer Table 27 Other Interesting Notables Keyboard Technology C onfusion and U nderstanding Keyboard technology confusion and understanding Ya, you have to plan it out, hu. ummhum Okay, so let's watch some of the video and if you see a part that you want to talk to me about you tell me to stop, okay. So we are not going to just play the video I'm going to fast forward until I see something that I want you to tell me about. 258 APPENDIX B: Talk Aloud and Interview Analysis Example What's going on right there? Mostly I went back because when it gets outlined it kind of like changes the word that I use and what I wrote so I go back and rewrite it and then I like double like go forward then go back one more then it won't be all messed up She has created a strategy for dealing with the blue highlight without recognizing how the blue highlight might help her. So, is that what happened right here? ya Let's watch I'm going to back up just a little bit while you create this word um, what word are you doing here? do you remember? about Okay, so let's see what's going on. So, do you notice, what's wrong with the word so far? Anything? I'm missing the u and the t the word is not all typed yet. It says ado. She does not notice the d in place of the b Okay, let's keep watching. So, there you go. You have your u Look! It's going to show up like that. (She points to the screen) This is around 2 min. 20 sec of the screen capture video. The words across here? (pointing to the autocomplete selections) Ya, but, the one that's actually white and out lined blue like that, it's , that word is going to show up like that. And, is that a good thing? Is that helpful, or... It's kind of frustrating. She is using our interview language "frustrating" 259 APPENDIX B: Talk Aloud and Interview Analysis Example Tell me, tell me what you mean, why is it frustrating? Because whenever like I write the other one I like push the space and it shows up with that and I have to erase the whole word. So, it automatically puts that word up ther e? ya and it's not the word you wanted ya Does this mean there is partial knowledge, using but confusing? what's going on here? Let's see, does it happen here in our video? Alright, so you deleted the t in the word about and then it went back to normal. So, you deleted that, so tell me that, tell me waht your strategy was there. Um, go back and redo it. So, do you delete the whole word or part of the word or what? part of the word and then you fix it and then I push spacer twice and then I go back one. So, do you delete the whole word or part of the word or what? part of the word and then you fix it and then I push spacer twice and then I go back one. 260 APPENDIX B: Talk Aloud and Interview Analysis Example Ohhh, okay. Why do you press space twice? Because then umm, the red line under it kind of means, don't get the word too close to this one, so I push the spacer twice and then I push the back button and then the red line under it is not there. Interesting, we were talking about the blue highlight which showed up first but while watching the video, after she used her strategy to remove the blue highlight, and then fix the word (in this case, put the t back) the red underline appeared. She also has her own und erstanding of what the red underline means. strategy invention tell me what you are thinking, why was that a good spot to stop and tell me what's going on? It's underlined, and whenever it's underlined blue it kind of gets me frustrated and it makes me a little mad because, umm, I'm trying to write and if I like push the space button it will show up with a different word sometimes. Okay, teach me about that. How do you know or how does that work? Mostly on twitter, umm, everybody has twitter accounts so everybody can see our tweets Okay, okay, who's everybody? Like everybody in the school, everybody's family, Okay, what about the lady at the grocery store? She looks at me confused. 261 APPENDIX B: Talk Aloud and Interview Analysis Example The lady that checks out the groceries? Can that person read my tweets? Or your tweets? maybe only families or friends So, it's not like, just anybody just family and friends that you know. Table 28 Other : Possibly Related to Social Norms Possibly Related to Social Norms Umm, because some people already tweeted about that and some people actually copy other people but, then some people read both tweets and they usually notice they are the same and then they learn the same thing. this chuck is about issues of duplicate tweets - kids copying what others have tweeted. This makes me wonder about the social pressures in the classroom, the tacit social norms among the children. 262 APPENDIX C: Screen Capture & Think Aloud Data Example Table 29 Screen Capture & Think Aloud Data Example Screen Capture & Think Aloud Data Example Seconds start Seconds end Description of Activity code Talk Aloud text Notes 0 12 setting up Pause 12 13 opens Twitter space APP 13 14 taps screen for cursor CR 15 16 KS 16 17 The KA 18 19 deleted automatic space BS 20 22 i s TML this the word "this" popped up as a black word cloud above the "the" but 531 was not looking at the screen to notice. He was focused on looking at his keyboard for letters. 23 24 SB 23 24 TM oh looks up at screen to see what he's typed. He looks at the laptop, not at the iPad 24 26 deleting to correct spelling E 263 APPENDIX C: Screen Capture & Think Aloud Data Example 28 30 ST "It's confusing me." laughing a little His statement is based on looking at laptop screen rather than iPad screen. This is causing some confusion. 30 35 RT "You'll have to keep your eye down here, hu (pointing to iPad screen) Pretend like this isn't here (pointing to laptop screen) 33 34 T TSL 34 36 KS 36 37 h i s TML saying the word he is typing "this" 37 40 KS 40 41 SB 41 42 i s TML is 42 43 SB 43 46 KS 46 47 m y TML my 47 48 TM Looks up at laptop screen 48 49 SB 50 54 SAH Umm, how do you spell giraffe? 54 61 RAQ Oh, okay, so let's think about it. I'll help you. Tell me if you umm.. 264 APPENDIX C: Screen Capture & Think Aloud Data Example 60 62 ST Or there's a microphone. 62 63 RQ Have you used that before? 63 64 SAQ nods head no 64 65 RQ Do you want to try it? as I ask this, he is already typing the microphone icon 65 66 taps microphone icon KM 67 69 talks into the microphone KM, ST "giraffe" 69 70 taps done under audio stream KM, Done 70 73 watches screen, waiting TM He rests his chin in the cup of his virtical fist and watches to see the result of his initial use of the microphone. 73 76 rereads his text so far Read This is my giraffe 76 77 SB 78 81 i t TML /i/ /t/ sounding out as he types 82 83 SB 84 87 w i l l TML will saying word as he types 87 88 SB 89 91 ST "help" says word as he gets ready to type it 91 92 h TSL 265 APPENDIX C: Screen Capture & Think Aloud Data Example 92 93 auto complete the word help KA 94 95 auto complete the word me KA "me" 96 99 b e TML "be" 100 101 auto complete the word become KA "become" 101 102 a TSL "a" 101 102 SB 103 104 auto complete the word better KA "better" 105 108 ST "a better" restates what he's written before adding next word 109 110 KM 110 112 "builder" into microphone KM, ST "builder" 112 113 KM, Done 113 117 TM watches with chin in fist 117 118 period TP "period" 119 123 Read "This is my giraffe it will help me become a better builder." 123 124 TM ummm thinking of next word 124 125 SB 266 APPENDIX C: Screen Capture & Think Aloud Data Example 126 128 delets space BS wait, and then, capitalize (as he taps up arrow for capital letters) 128 130 TSL cap button 131 134 Pause thinking of what to type next 135 136 ST "by?" Says this like a question, looking up at me. 136 139 RAQ If you are ready for that. 139 140 B y TML "by" 141 142 SB 142 143 BS It's like the SB is an automatic movement, then it is deleted because the writer did not mean to put it there. 143 144 number keyboard KO 144 147 531 TML "five hundred thirty one" 147 149 RQ Then what? 150 152 special character keyboard KO 152 153 TH "hashtag" 153 157 KS 157 158 KM 158 159 KM, ST "giraffe" 267 APPENDIX C: Screen Capture & Think Aloud Data Example 159 160 KM, Done 160 163 watches screen TM sits back in seat, knowing he is done with the task 164 165 RQ Now what? 165 170 thinking SAQ Ummm... we are done (shrugs shoulders) 171 176 RQ Okay so is there anything else you want to add to your tweet? 176 177 SAQ shakes head, "no" 177 179 RQ How do you know that it's ready to go? 180 190 SAQ Um, you do, "by 531" and then you can do a hashtag, or something. Oh, I can do one more... (starts deleting his text) 190 193 deleting all text after "By" R He is revising, changing the text to add value, deleted too far on accident 193 194 number keyboard KO 194 196 531 TML putting his number back, he must not have meant to delete that part. 197 198 opens to emojis KO 198 203 searching for a giraffe KS they have a giraffe 268 APPENDIX C: Screen Capture & Think Aloud Data Example 203 222 scrolling, looking for giraffe Scroll I don't see a giraffe...nope 222 224 RQ "so what are you thinking?" 224 228 SAQ "ummm, I'm done." 229 232 back to letters KO 232 235 special character KB KO I need a hashtag 236 239 TH hashtag 240 241 KM 241 242 KM, ST "giraffe" 244 246 Pause to look at me Pause 247 249 RQ Ummm, did you add your picture? 250 251 SAQ no (moving toward keyboard to add pic) 251 259 opens space to access pic C getting picture he took before meeting with me. 254 256 RQ Have you forgotten your picture before? 257 258 SAQ No 259 261 watching as picture shows up TM 269 APPENDIX C: Screen Capture & Think Aloud Data Example 261 265 looks out window at sound of kids ATT 265 268 We start talking here. RQ Let's see, this is your giraffe? We start chatting about his work at this point. SAQ ummhumm, nodding yes RQ Okay, are there any words you need help with in your tweet? SAQ Ummm, no, (looking at text) RQ Do you have any questions or do you want me to check anything before you send it off? SAQ "check" as he pushes iPad in my direction RQ what do you want me to check? SAQ If I spelled... become, no, ya I spelled it. RQ how do you know that it's okay? SAQ Because I can sound it out, then he spells it, b e c o m e RQ Alright, okay, what else would you like me to look at or check? SAQ Ummm, if Ihave to put a period anywhere. 270 RQ Do you have more than one sentence? SAQ nods no RQ So, what should you do? SAQ make another sentence. (grabbing iPad) 331 334 taps screen, opens up image KO 335 336 I close image and help him get to text tech error 337 338 opens letters KO 339 342 deletes text ending with cursor next to Builder D 341 342 RQ So, why are you deleting those things? 342 352 SAQ Because I have to delete them, then if we do that, it's at the end of the tweet. 353 354 . TP adds a period that he accidentally deleted 355 357 SB and then, no 358 359 hit return button Enter his return button by accident 27 1 360 361 hit BS button BS 362 363 BS 363 367 RQ So, why don't you put a space after the period? You put a space, then you backed up. 367 369 SAQ I don't know. shrugs shoulders 370 381 Thinking Pause I can't remember 381 383 RT Go ahead and reread 383 387 rereads text so far Read This is my giraffe it wil help me become a better Builder. 388 400 thinking Pause ummm, I can't figure out anything else. RQ Hmmm, is there anything about being a builder that you could add? SAQ shakes head no RQ how do you come up with more to write, you know, when you have to write more? What are some of the things you do to help yourself think of ideas? 272 SAQ I don't know, shrugs shoulders RT I know a lot of kiddos say they look at the picture, umm, and that helps them get ideas. Do you want to look at your picture? 432 438 opens image TM 438 439 opens letters KO 438 439 RQ Do you already have an idea? 439 440 SAQ shakes head, yes, as he looks at keyboard 440 446 I help him RT Okay, make sure you are in the right place, becuase builder is down here. the cursor was at the end of the first line rather than at the end of the writing. I'm not sure why I quickly told him to check his cursor location. Maybe I did not want to allow him to make that error and start all over. 447 449 opens microphone KM 449 453 talks into microphone KM, ST Giraffes live in the rainforest" 453 455 KM, Done 455 459 watches screen for text to appear TM 460 462 RQ Is it what you want it to be? 273 462 463 SAQ shakes head yes RQ Why did you decide, giraffes live in the rainforest, and how do you know that? SAQ Ummm, cuz I've read about girraffes and it tells you where they live. RQ Who's going to read this tweet? SAQ everyone. RQ So, should you tell your readers how you know this is true? Do you think they will want to know more, or... SAQ we don't usually do that , though. this makes me laugh. He's quick to answer. RQ Why not? SAQ shrugs shoulders, I don't know. RQ Well, you can do it if you want. SAQ I don't want to. 460 494 RT Okay and that's okay. You can be done. 494 496 looking at screen TM 274 496 497 RQ Do you have your end punctuation? Why did I ask this? I am leading here! 498 499 types a period TP 500 503 b y TML 504 505 numbers KO 505 507 531 TML 507 511 KS 511 512 special character KB KO 512 513 TH hashtag 514 515 KM 515 516 KM, ST giraffe 516 517 KM, Done 517 521 watches text appear TM 521 522 RQ Is it ready to go? 522 523 SAQ shakes head yes 275 APPENDIX RQ So, 531, while we wait, can you tell me, how do you plan what you are going to tweet about? SAQ We mostly just think then we take a picture of what we want to do and then we start tweeting about it. RQ I noticed you did that. When I went to pick you up at your class I asked you, "Do you know what you want to tweet about?" and you said no, but then you quickly came up with an idea. How did you come up with your idea? SAQ Cuz, I remember that we built a girraffe with Legos. RQ How do you decide that that's something that should be tweeted about 276 RQ How does it make you feel, knowing everyone is going to read your tweet? SAQ Shrugs shoulders, smiles, I don't know. RQ You don't know? Who reads your tweet? SAQ Mrs. Hammer, my friends, RQ I've read some of your tweets. SAQ You've liked some of them! (moving to scroll through feed to find it) ST I know where one that is super funny. RQ And, how does that make you feel when somebody likes a tweet? SAQ Good. (scrolling through feed) there's one where Solly is eating books. ST Who tweeted about sea otters? (stops to look at someone's tweet) Now what are you going to do? I want to read it. 277 Okay. Go ahead. He reads the tweet aloud. Do you know who that is? (by 461) umhumm, nodding head yes who do you think it is? I think it's L --- or T___ Why do you think that's who wrote it? Because they've told me their number. So, you just liked that. Can you tell me about that? Why do people like tweets? So, like this one (shows me a tweet with a 1 next to the heart) has been liked. So someone likes their sentence. (scrolling) and like that one, so people liked the sentence, so there's that thing right there. (pointing to heart) okay 278 Then they click it and it shows that red thing. (talking and scrolling) hey (scrolling) What are you looking at now? Another book? umhumm, nodding head yes what made you stop to look at that book? It's cuz I've already read it before. Do you get ideas for what to read about when you read other peoples' tweets? No, I've already had this book. unhuu. So, what made you decide to stop and look at that tweet? Because I really wanted to look at the book to make sure it's the right one and it is. Okay. now what are you looking for? What is this (looking at a pic of flooring) What does it say? 279 Reading tweet aloud: This is the school ground. It is red, gray, and white. It is a cool pattern. by sixhundred fifty one, hashtag ground. What do you think about that? Shrugs shoulders super high, I don't know! Laugh, It's kind of funny! Ya, did it surprise you? shakes head yes (keeps scrolling through feed) Now, why did you decide to click on this one? Cuz, um, it's one of the ones that we did when we were Skyping. He reads the tweet aloud. It's ends with "hashtag emoji" but what kind of emoji? happy does that make sense? 280 APPENDIX C: Screen Capture & Think Okay, so, ____, do you like to tweet? Because it's um like fun like um, helps you, become a better like typer Okay, say more about that. Ummmm.... so when you go into high school you if you have work that you have to do on your computer you can like type it fast. So you get it done. Right. and you don't have to do it like after school. Does it matter to you that other people read your tweets? shakes head no why not? I don't know (he starts looking at the class Twitter Feed again) 281 you are r eading other peoples' tweets. Ar en't you. we are allowed to. of course you are. As he scrolls through the tweets he notices he can slide the screen left to right and when he does this the same screen of the class tweeter feed shows up. Have you ever seen that before? No (smiles) he keeps sliding the screen up down, left, right, just exploring What are you thinking? (he stopped at a tweet) He reads the tweet aloud "my kind of teaching day! MASD informational (it actually says environmental) center field trip, field trip. hashtag mas . masssd proud (for #hasdproud) 282 does that make sense? No what would you do to.. chunk it up. okay. he starts chunking up the hashtag, sounding out each letter. Is that a d? I think so. I'm wondering if that, look right up here, (pointing to the text that says MASD) Hold on. I know how to make sure it's a d (opens up a blank space to type a tweet) How are you going to make sure it's a d? okay, it's a d (looking at the keyboard then closing the keyboard to go back to the tweets) What did you do? How did you help yourself? 283 Becuase, the b is supposed to be down there (pointing to a space on the keyboard) and the d is supposed to be up there (pointing as if he where looking at the keyboard) He goes back to read the hashtag, masssd proud Ya, I think that might be the name of their school or um the initials fo rthier school, MASD so, MASD proud. Does that make sense? Ummm maybe. So, what do you think this tweet is about? This was just tweeted. How do you know that? Because mine is right here and if I go up to see more, I could. Right here is says right and the letter m which means eight minutes ago. that was tweeted eight minutes ago. 284 Oh, mine was tweeted nine minutes ago. That means it tweeted before mine. Why, what do you mean? Cuz, nine minutes is before eight minutes. Eight min. is after nine (starts scrolling through feed) that was one hour So, one hour ago this happened and (scrolling up) nine minutes ago, this happened (scrolling up again) and, eight minutes ago this happened. Oh, wait (the time changed) nine minutes ago that happened. It's not eight. (he looks at the tweet confused) Why do you think it changed? Ohhhh cuz it's counting on the clock!!! (he's excited, like he solved a puzzle.) So, do you think that number will hcange again? Look at this one, right here. umhumm and starts scrolling again 285 APPENDIX C: Screen Capture & He scrolls by a Downy ad "whoa, what's that?" Oh, it's just a commercial. Then goes back to scrolling, stopping at a tweet with five emojis a horse, car, truck, motorcycle, house, he says, "hey, those emojis aren't supposed to be there." Then he starts reading the tweet aloud, "I am learning about getting my day on schedule. World (for Earth emoji) it is great to have a scuttle to do that I need to be on time. (he looks confused) "horse, bike, garbage truck, motorcycle, by sixhundred fifty three hashtag skittle" looking confused. What are you thinking? Looking at tweet for a while, then, "It doesn't make sense." So, when you see a tweet and it doesn't make sense what does that make you think as a writer? I don't know. They were trying to do it fast. ohhh. okay. (he starts scrolling again) So, let's move forward, I need to ask you some more questions and we need to watch the video of you tweeting. are you ready? 286 APPENDIX D : Eight Tweeters Tweeting Code Book Table 30 Writing Processes Codebook Writing Processes: Thoughts made known through talk aloud and interview Control Level Codes: Writing Processes Action/ Indicator Code Statements about: Example Motivation M Statements or obvious body language expr essing enjoyment/willingness to tweet something, and people around the world can see them, so it feels like you about like what we learned in school Goals Goal MM Reasons for tweeting, purpose, goals, why he/she tweets And statements about the tweeted message making sense Like, you don't need to have a picture but it needs to have what makes Planning In - advance Planning IAP Generating tweet ideas, planning tweet content before writing picture of what we want to do and then 287 APPENDIX D : Eight Tweeters Tweeting Code Book Planning In - the - moment Planning IMP Generating ideas, next words, while tweeting Writing Schema Fundamental elements TS Teacher - created guidelines, writing prompt, 280 character limit why, Writing Schema Genre knowledge Multimodal Fundamental elements Multimodal The online - space use of images, emojis, GIFs, URLs The tweet might include one or more multimodal elements: The reader can see that clever is a word of the day and is part of what they do in the classroom Writing Schema Genre knowledge Hashtags Fundamental elements Hashtag The online - space social aspect of hashtags um, hashtags um are kind like a title but it goes at the end and it's a few words stuck together there are no spaces in it and it's about Writing Schema Genre knowledge Followers Contextual elements Followers or A - audience The online - space social aspect of following, who reads the tweet the world. Writing Schema Genre knowledge Likes Contextual elements Likes The online - space social aspect of likes actually read our tweets they actually 288 APPENDIX D : Eight Tweeters Tweeting Code Book Writing Schema Genre knowledge Cyber Safety Contextual elements Cybersafety The online - space safety/ digital citizenship or like your phone number or whatever... you will.... they will maybe Writing Schema Genre knowledge Networking Contextual elem ents Networking The online - space social aspect of networking tweet comments to us and their own tweets and so and sometimes they get funny little videos on it on comments so it tells us, they type something below and then about. And then, you get an idea from Writing Schema Genre knowledge Text Structure Text Structure Text Structures: what, why, how flexible focus, fixed topic, topic elaboration, assumes reader context knowledge Text Structures: what, why, how flexible focus, fixed topic, topic elaboration, assumes reader context knowledge 289 APPENDIX D : Eight Tweeters Tweeting Code Book Table 31 Composition Moves Code Book Composition Moves : What can be observed - screen capture A letter, Phrase, and Word Flow of Composing Composition Moves - Flow of Composing Composing Code Action/Definition Types emoji Types spacebar Types backspace to delete space Revision Processes Code Action/Definition Types backspace to delete text Revision, word changes based on original thinking made known from talk - aloud or from changes in the text Task Environment Code Action/Definition Adds words to tweet using the speech to text microphone feature on the iPad keyboard 290 APPENDIX D : Eight Tweeters Tweeting Code Book Pressing a finger to the iPad screen where the text has been typed and corresponding text is highlighted pink Word typed by the child is changed by the autocorrect feature and the child does not notice this change. Word typed by the child is spelled incorrectly, the autocorrect feature then suggests words the child may have been trying to type, the child selects one of these words. Selects a word from the autocomplete suggestions. Types an entire word and before typing the next word or typing a spacebar the student taps on the matching autocomplete suggestion. Switching between any of the available keyboards: alphabetic, alphan umeric, special character, emoji keyboards. Moves cursor to a new location in the text. This may include activating the magnifying glass. Resource Level Code Action/Definition The student stops typing and reads what has been typed so far. This is either a part of the text or the entire text. 291 APPENDIX D : Eight Tweeters Tweeting Code Book Table 32 Process Level Codes: Composition Moves related to Transcription Composition Moves (27 codes) : What can be observed - screen capture Process Level Codes: Composition Moves related to Transcription/Composing Component Skills Composing Elements Code Action/Definition Example found in screen - capture Typing emoji Emoji Adds emoji to message Types book emoji Typing a single letter TSL Types one letter followed by keyboard search < 2 sec. Typing multi - letter TML Types many letters without stopping, > 2 sec between letters between letters Typing spacebar TSB Types the spacebar to move forward space In some cases, it's like the space bar is an automatic movement, then it is deleted because the w riter did not mean to put it there. Typing backspace TBS Types the delete button to move back a space presses delete to remove the automatic punctuation. 292 APPENDIX D : Eight Tweeters Tweeting Code Book Typing punctuation TP Types a form of punctuation or special character types end punctuation. Typing hashtags TH Types a hashtag In - the - moment planning P While writing, thinks of what to write Student says, "I like math with someone because" then thinks of what to type next. Table 33 Process Level Codes: Compos ition Moves related to Revision Process Level Codes: Composition Moves related to Revision/Evaluation Skills Revision Processes Revision Elements Code Action/Definition Example found in screen - capture Addition of new text NT Identifying area needing detail and adding text n/a 293 APPENDIX D : Eight Tweeters Tweeting Code Book Editing E Changing word choice, grammar, spelling Rewriting RW Rewriting a sentence from scratch n/a Revising R Altering a sentence to add value without rewriting Reorganizing RO Moving text around, including copy and paste n/a Deleting D Deleting text from the document to s on the keyboard. 294 APPENDIX D : Eight Tweeters Tweeting Code Book Table 34 Process Level Codes: Compos ition Moves related to Task Environment Process Level Codes: Composition Moves related to Task Environment, Interacting with the media, composing technology, assignm ent materials Task Environment Task Env. Element Code Action/Definition Example found in screen - capture Camera C Taking or accessing photo(s) opens photos to access pic Twitter app App Opening space to work on iPad and submitting a tweet Taps twitter app on iPad to start tweeting. Taps "tweet" button in top right of screen to submit tweet Keyboard searching KS Pausing to look for letter/item on the keyboard(s) Keyboard open or closed KO Changing, disappearing and reappearing keyboard changes to number keyboard Keyboard autocomplete KA Selecting word suggestion at top of the keyboard 295 APPENDIX D : Eight Tweeters Tweeting Code Book Cursor move CR/MG Touching screen to position cursor, including magnifying glass (MG) Moves cursor location to the end of first "somebody" in message Scrolling Scroll Swiping the screen to view content below or above the working area Scrolling through emoji keyboard, then Task materials TM Looking at environmental print - external sources - iPad screen - assignment pages Looking inside the book for the title Looking at picture and papers Thinking, looking at screen saying words Says "builder" into the microphone Already accessed photo AAP Looking at the image captured before meeting with me Posts photo before tweeting. Pauses from tweeting to look at photo for ideas. Writing prompt WP Looking at the writing prompt reminder Looks at the white paper the teacher provided to remind the students of their writing prompt. 296 APPENDIX D : Eight Tweeters Tweeting Code Book Table 35 Process Level Codes: Compos ition Moves related to Resource Level Resource level Codes: Composition Moves Resource Level Resource Elements Code Action/Definition Example found in screen - capture Reading Read Reading followed by a continuation of writing or deciding to submit a tweet Reads, "called penguins" Attention diverted ATT Attending to something other than the tweeting task looks out the window at students walking through the hallway Pausing Pause Cannot observe or identify from the interview the reason for pause Pausing, takes a breath, pulls hair away from face Pausing, thinking about the next letter 297 APPENDIX D : Eight Tweeters Tweeting Code Book Table 36 Emergent Features Codes Emergent Features : What can be observed screen capture, and what is made apparent from any other data Process Level Codes: Writing Processes/Composition Moves Emergent Features Emergent Features autocomplete autocomple te the autocomplete feature displayed three initial letters keyed in and the previous syntax. different, like different spell ing, so, if you're trying to spell a word you could see it up there, if it's up there, you could press on it, Red underline Red underline When a student typed on the keyboard the red underline appeared under the text which did not follow typical English spelling and/or syntax, such as homophones. Blue highlight Blue highlight When a student typed on the keyboard the blue highlight appeared once an une xpected letter was added to a string of letters typically found in a word. or a word isn't spelled that way it highlights a feature of the language used by students to they composed their tweets. 298 APPENDIX D : Eight Tweeters Tweeting Code Book Table 3 Magnifying glass MG To activate the magnifying glass when typing, the writer had to firmly press a finger on or between letters and words on the iPad screen. Black Box Words Black box words if the writer firmly pressed a finger to the iPad screen at the location of a word, a black box with additional w ord choices. spelling pattern, like the words that spell like have umm just like pushed that but I didn't Copyright copyright feature of the composing process expressed by a student when peers poached ideas from her tweet to use in their own. actually copy me and then write some other things down, instead of exactly the same URLs URLs a featur e of the process used by students to indicate their understanding of what a feature is. Reading a posted tweet aloud, at the URL he read: "msftsokial/gnavfa" sounding out the letters, as if decoding a hard word. 299 REFERENCES 300 REFERENCES Ackerman, John M. (1994). Collaborative Planning in Busin ess and Architectural Training. Technical Report for the Center for the Study of Writin g and Literacy at UC Berkeley and Carnegie Mellon University. Ackerman, J. & Oates, S. (1996). Image, text, and power in architectural design and workplace writing. In A.H. Duin & C.J. Hansen (Ed.), Nonacademic Writing: Social Theory and Technology (pp. 81 - 120). New York: Routledge. Airenti, G. (2015) . The cognitive bases of anthropomorphism: from relatedness to empathy. International Journal of Social Robotics, 7(1), 117 - 127. Allal, L., & Ducrey, G. P. (2000). Assessment of or in the zone of proximal development. Learning and instruction, 10(2), 137 - 152. Andrews, R., & Haythornthwaite, C. (Eds.). (2007). The Sage handbook of e - learning research. Sage. Andrews, R., & Smith, A. (2011). Developing writers: Teaching and learning in the digital age. McGraw - Hill Education (UK). Assor, A., Kaplan, H., & R oth, G. (2002). Choice is good, but relevance is excellent: engagement in schoolwork. British journal of educational psychology, 72(2), 261 - 278. Atwell, N. (1987). In the middle: W riting, reading, and learning with adolescents. Heinemann Educational Books, 316 Hanover St., Portsmouth, NH 03801. Bazerman, C., & Prior, P. (2005). Participating in emergent socio - literate worlds: Genre, disciplinarity, interdisciplinarity. Multidiscipl inary perspectives on literacy research, 2, 133 - 178. Becker, A. (2006). A review of writing model research based on cognitive processes. Revision: History, theory, and practice, 25 - 49. Berdanier, C. G., & Trellinger, N. M. (2017, October). Development of a method to study real - time engineering writing processes. In 2017 IEEE Frontiers in Education Conference (FIE) (pp. 1 - 9). IEEE. Scardamalia, M., & Bereiter, C. (1987). Knowledge telling and knowledge transforming in written composition. Advances in appl ied psycholinguistics, 2, 142 - 175. 301 Bereiter, C., & Scardamalia, M. (2013). The psychology of written composition. Routledge. Berninger, V. W., Cartwright, A. C., Yates, C. M., Swanson, H. L., & Abbott, R. D. (1994). Developmental skills related to writin g and reading acquisition in the intermediate grades. Reading and Writing, 6(2), 161 - 196. Bernstein & Crowley. (2008). Searching for Signs of Intelligent Life: An Investigation of Young Children's Beliefs About Robot Intelligence, Journal of the Learning Sciences, 17:2, 225 - 247, DOI: 10.1080/10508400801986116 Boren, T., & Ramey, J. (2000). Thinking aloud: Reconciling theory and practice. IEEE transactions on professional communication, 43(3), 261 - 278. Brand, A. G. (1989). The psychology of writing: The a ffective experience (No. 13). Greenwood Press. Brandt, D. (2014). The rise of writing: Redefining mass literacy. Cambridge University Press. Adolescent & Adult Literacy, 59(3), 271 - 274. Burnett, C., & Myers, J. (2006). Observing children writing on screen: Exploring the process of multi - modal composition. Language and Literacy. Calkins, L. M. (1986). The art of teaching writing. Heinemann Educational Books Inc. , 70 Court St., Portsmouth, NH 03801. \ Il l.: Heinemann. Calkins, L. M. (2003). The nuts and bolts of teaching writing. Firsthand Books. Chenoweth, N. A., & Hayes, J. R. (2001). Fluency in writing: Generating text in L1 and L2. Written communication, 18(1), 80 - 98. Chenoweth, N. A., & Hayes, J. R. (2003). The inner voice in writing. Written communication, 20(1), 99 - 118. Clark, C., & Dugdale, G. (2009). Young People's Wri ting: Attitudes, Behaviour and the Role of Technology. National Literacy Trust. Clark, R. E., & Salomon, G. (1986). Media in teaching. Handbook of research on teaching, 3, 464 - 478. Coiro, J., & Dobler, E. (2007). Exploring the online reading comprehension strategies used by 302 research quarterly, 42(2), 214 - 257. Coiro, J., Knobel, M., Lankshear, C., & Leu, D. J. (2008). Handbook of new literacies research. Cooper, M., & Holzman, M. (1983). Talking about protocols. College composition and communication, 34(3), 284 - 293. Corbin, J.M., & Strauss, A.L. (2008). Basics of Qualitative Research (3rd ed.): Techniques and Procedures for Developing Grounded Theory. Co instruments. Technology, Pedagogy and Education, 16(3), 337 - 356. Creswell J (2007) Qualitative Inquiry & Research Design: Choosing among Five Approaches (2nd Edition). Thousa nd Oaks, CA: Sage. Creswell, J. W. (2009). Research design: Qualitative and mixed methods approaches. London and Thousand Oaks: Sage Publications. Creswell, J. W., & Poth, C. N. (2018). Qualitative inquiry & research design: Choosing among five approache s (pp. 181 - 223) Thousand Oaks. Cutler, L., & Graham, S. (2008). Primary grade writing instruction: A national survey. Journal of Educational Psychology, 100(4), 907 - 919. Davidsson, P., & Findahl, O. (2016). Internetstiftelsen i Sverige. Iis. se. Stockholm: Internetstiftelsen i Sverige. Cognitive models of writing: Writing proficiency as a complex integrated skill. ETS Research Report Series, 2008(2), i - 36. Dowdall, C. (2009). Impressions, improvisations and compositions: Reframing children's text production in social network sites. Literacy, 43(2), 91 - 99. Dudley - Brown, S. L. (1997). The evaluation of nursing theory: a method for our madness. International journal of nursing studies, 34(1), 76 - 83. Edwards - Groves, C. J. (2011). The multimodal writing process: Changing practices in contemporary classrooms. Language and Education, 25(1), 49 - 64. Eliason, N., (2017) The Step - by - Step Guide to Go From Novice to Expert in Any Skill Published or Updated on Jun 26, 2017 accessed February 20, 2018. https://www.nateliason.com/blog /become - expert - dreyfus Emig, J. (1971). The composing processes of twelfth graders. Ericsson, K. A., & Simon, H. A. (1984). Protocol analysis: Verbal reports as data. the MIT Press. 303 Fayol, M., Foulin, J. N., Maggio, S., & Lété, B. (2012). To wards a dynamic approach of how children and adults manage text production. Writing. A mosaic of new perspectives, 141 158. Fitzgerald, J. (2013). Constructing instruction for struggling writers: What and how. Annals of Dyslexia, 63(1), 80 - 95. Retrieved from http://www.jstor.org/stable/23764728 Fitzgerald, J., & Shanahan, T. (2000). Reading and writing relations and their development. Educational Psychologist, 35(1), 39 - 50. Flower, L. (1989). Problem - solving strategies for writin g. Orders, Harcourt Brace Jovanovich, 7555 Caldwell Ave., Chicago, IL 60648. Flower, L. (1994). The construction of negotiated meaning: A social cognitive theory of writing. SIU Press. Flower, L. (1996). Negotiating the meaning of difference. Written Communication, 13(1), 44 - 92. Flower, L., & Hayes, J. R. (1981). A cognitive process theory of writing. College composition and communication, 32(4), 365 - 387.doi:10.1207/S15326985EP3501_5 Fuller (1995) writing structures Glaser, B., & Strauss, A. (1967). The discovery of grounded theory. 1967. Weidenfield & Nicolson, London, 1 - 19. Goodman, K. S. (1973). Miscue Analysis: Applications to Reading Instruction. Graham, S. (2006b). Writing. In P. Alexander & P. Winne (Eds.), Handbook of educational psychology (pp. 457 477). Mahwah, NJ: Erlbaum Graham, S., & Harris, K. R. (2016). A path to better writing. The Reading Teacher, 69(4), 359 - 365. Graham, S., & Harris, K. R. (2017). Evidence - based writing practices: A meta - analysis of existing meta - analyses. In De sign principles for teaching effective writing (pp. 13 - 37). Brill. Graham, S., McKeown, D., Kiuhara, S., & Harris, K. R. (2012). A meta - analysis of writing instruction for students in the elementary grades. Journal of educational psychology, 104(4), 879. Graves, D. H. (1973). Children's Writing: Research Directions and Hypotheses Based Upon an Examination of the Writing Processes of Seven Year Old Children. Graves, D. H. (1980). Research update: A new look at writing research. Language Arts, 57(8), 304 913 - 9 19. Graves, D. H. (1983). Writing: Teachers and children at work. Heinemann Educational Books, 4 Front St., Exeter, NH 03833. Graves, D. H. (1994). A fresh look at writing. Heinemann, 361 Hanover St., Portsmouth, NH 03801 - 3912 (ISBN - 0 - 435 - 08824 - 6, $20).. Greenhow, C., Robelia, B., & Hughes, J. (2009). Web 2.0 and classroom research: What path should we take now? Educational Researcher, 38, 246 260. Greenhow, C., Sonnevend, J., & Agur, C. (2016). Education and Social Media: Toward a Digital Future. MIT Press. Howe, K. (2017). Personal phone communication, October 12th, 2017 Howe, K. (2018). Personal email communication, January 23rd, 2018 Harris, K. R., & Graham, S. (2016). Self - regulated strategy development in writing: Policy implications of an evid ence - based practice. Policy Insights from the Behavioral and Brain Sciences, 3(1), 77 - 84. Harris, K. R., Graham, S., Brindle, M., & Sandmel, K. (2009). Metacognition and - 153. Hayes J. R. (19 96). A new framework for understanding cognition and affect in writing. In Levy C. M., Ransdell S. E. (Eds.), The science of writing: Theories, methods, individual differences, and applications (pp. 1 - 27). Mahwah, NJ: Erlbaum. Google Scholar Hayes, J. R. (2011). Kinds of knowledge - telling: Modeling early writing development. Journal of Writing Research, 3(2), 73 - 92. Hayes J. R. (2012). My past and present as writing researcher and thoughts about the future of writing research. In Berninger V. (Ed.), Past, present, and future contributions of cognitive writing research to cognitive psychology (pp. 3 - 26). New York: Psychology Press. Hayes, J. R., & Berninger, V. W. (2014). Cognitive processes in writing: A framework. Writing development in children with he aring loss, dyslexia, or oral language problems, 3 - 15. Hayes J. R., Flower L. S. (1980). Identifying the organization of writing processes. In Gregg L. W., Steinberg E. R. (Eds.), Cognitive processes in writing (pp. 3 - 30). Mahwah, NJ: Erlbaum. 305 Hayes J. R., Flower L. S., Schriver K. A., Stratman J., Carey L. (1987). Cognitive processes in revision. In Rosenberg S. (Ed.), Advances in psycholinguistics: Reading, writing, and language processing (Vol. 2, pp. 176 - 240). Cambridge: Cambridge University Press. Hayes, J. R., & Olinghouse, N. G. (2015). Can cognitive writing models inform the design of the common core state standards?. The elementary school journal, 115(4), 480 - 497. Heider, F., & Simmel, M. (1944). An experimental study of apparent behavior. The American journal of psychology, 57(2), 243 - 259. Hodges, T. S. (2017). Theoretically speaking: An examination of four theories and how they support writing in the classroom. The Clearing House: A Journal of Educational Strategies, Issues and Ideas, 90(4), 139 - 146. Hunt, K. W. (1965). A synopsis of clause - to - sentence length factors. The English Journal, 54(4), 300 - 309. Hyland, K. (2015). Teaching and researching writing. Routledge. Jessica S. Mitchell, Rachael F. Thompson, Rebecca S. Anderson (2016), Unde rstanding a Digital Writing Cycle: Barriers, Bridges, and Outcomes in Two Second - Grade Classroom s, in Evan Ortlieb , Earl H. Cheek, Jr. , Wolfram Verlaan (ed.) WritingInstruction to Support Literacy Success (Literacy Research, Practice and Evaluation, Volu me 7) Emerald Group Publishing Limited, pp.137 - 155 Johnston, P.H. (2012). Opening minds: Using language to change lives. Portland, ME: Stenhouse Publishers. Jonassen, D. H., & Reeves, T. C. (1996). Learning with technology: Using computers as cognitive tools. In D. H. Jonassen (Ed.), Handbook of research for educational communications and technology (pp. 693 - 719). New York: Macmillan Kellogg, R. T. (1999). The psychology of writing. Oxford University Press. Koro - Ljungberg, M., Douglas, E. P., Therriault, D., Malcolm, Z., & McNeill, N. (2013). Reconceptualizing and decentering think - aloud methodology in qualitative research. Qualitative Research, 13(6), 735 - 753. Kurtz, J. (2009). Twittering about learning: Using Twitter in an elementary school classroom. Horace, 25(1), n1. Kuusela, H., & Pallab, P. (2000). A comparison of concurrent and retrospective verbal protocol analysis. The American journal of psychology, 113(3), 387. Lacina, J., & Silva, C. (2010). Cases of successful literacy teachers . Sage Publications. 306 Lankshear, C., & Knobel, M. (2003). New literacies: Changing knowledge and classroom learning. Buckingham, UK: Open University Press. Lankshear, C., & Knobel, M. (2007). Researching New Literacies: Web 2.0 practices and insider pers pectives. E Learning, 4(3), 224 240. Lankshear, C., & Knobel, M. (2008). Digital literacies: Concepts, policies, and practices. New York: Peter Lang. Lawrence, J. F., Niiya, M., & Warschauer, M. (2015). Narrative writing in digital formats: Interpretin g the impact of audience. Psychology of Language and Communication, 19(3), 201 - 221. Leggette, H. R., Rutherford, T., Dunsford, D., & Costello, L. (2015). A review and evaluation of prominent theories of writing. Journal of Applied Communications, 99(3), 37 - 53. Leu, D. J., Kinzer, C. K., Coiro, J. L., & Cammack, D. W. (2004). Toward a theory of new literacies emerging from the Internet and other information and communication technologies. Theoretical models and processes of reading, 5(1), 1570 - 1613. Leu, D. J., Kinzer, C. K., Coiro, J., Castek, J., & Henry, L. A. (2013). New literacies: A dual level theory of the changing nature of literacy, instruction, and assessment. In D. E. Alvermann, N. J. Unrau, & R. B. Ruddell (Eds.), Theoretical models and proces ses of reading (pp. 1150 1181). Newark, DE: International Reading Association. - Cacopardo, H. (2009). Comments on Greenhow, Robelia, and Hughes: Expanding the new literacies conversation. Educational researcher, 38(4), 264 - 269. Leu, D. J., Slomp, D., Zawilinski, L., & Corrigan, J. A. (2016). Writing research through a new literacies lens. Handbook of writing research, 2, 41 - 53. Leijten, M., & Van Waes, L. (2013). Keystroke l ogging in writing research: Using Inputlog to analyze and visualize writing processes. Written Communication, 30(3), 358 - 392. Lindstrom, D. L., & Niederhauser, D. S. (2016). Digital literacies go to school: A cross - case analysis of the literacy practices used in a classroom - based social network site. Computers in the Schools, 33(2), 103 - 119. Limpo, T., & Alves, R. A. (2017). Written language bursts mediate the relationship between transcription skills and writing performance. Written Communication, 34(3), 306 - 332. Maggio, S., Lété, B., Chenu, F., Jisa, H., & Fayol, M. (2012). Tracking the mind during writing: Immediacy, delayed, and anticipatory effects on pauses and writing rate. Reading and 307 Writing, 25(9), 2131 - 2151. Maher, D. (2010). Using the Interne t as a communication tool to develop primary school students' social awareness. The Social Educator. Marich, H. (2016). Twitter in the Elementary Classroom: A Teacher's Journey. Language Arts, 94(1), 67 Marsh, D. (2014). The secrets of great headline writ ing. The Guardian, 9. McCarthy, T. (1994). On the idea of a critical theory and its relation to philosophy. In D. Couzens Hoy & T. McCarthy, Critical theory (pp. 7 30). Cambridge, MA: Blackwell. McGrail, E., & Davis, A. (2011). The influence of classroom blogging on elementary student writing. Journal of Research in Childhood Education, 25(4), 415 - 437. McGeown, S., Goodwin, H., Henderson, N., & Wright, P. (2012). Gender differences in reading motivation: Does sex or gender identity provide a better accou nt? Journal of Research in Reading, 35(3), 328 336. McKee, H., & Porter, J. E. (2008). The ethics of digital writing research: A rhetorical approach. College Composition and Communication, 711 - 749. Mellon, J. C. (1968). Transformational sentence - combining: A method for enhancing the development of syntactic fluency in English composition (No. 10). National Council of Teachers of English. Merchant, G. (2005). Digikids: Cool dudes and the new writing. E - learning and Digital Media, 2(1), 50 - 60. Merchant, G. (2005). Electric involvement: Identity performance in children's informal digital writing. Discourse: Studies in the cultural politics of education, 26(3), 301 - 314. Merchant, G. (2007). Writing the future in the digital age. Literacy, 41(3), 118 - 128. Mills, K. A., & Chandra, V. (2011). Microblogging as a literacy practice for educational communities. Journal of Adolescent & Adult Literacy, 55(1), 35 - 45. Mitchell, J. S., Thompson, R. F., & Anderson, R. S. (2017). Understanding a Digita l Writing Cycle: Barriers, Bridges, and Outcomes in Two Second - Grade Classrooms', Writing Instruction to Support Literacy Success (Literacy Research, Practice and Evaluation, Volume 7). Miyashita, K., & Knezek, G. (1992). The young children's computer inv entory: a Likert scale for assessing attitudes related to computers in instruction. Journal of Computing in Childhood Education, 3(1), 63 - 72. 308 Moffett, J. (1968). A Student - Centered Language Arts Curriculum, Grades K - 13: A Handbook for Teachers. Morgan, H. (2014). Focus on Technology: Enhancing Instruction and Communication With Twitter: Hani Morgan. Childhood Education, 90(1), 75 - 76. Mosher, F.A. (2011). The role of learning progressions in standards - based education reform. (CPRE Policy Brief RB - 52). Re trieved from http://www.cpre.org/role - learning - progressions - standards - based - education - reform Murray, D. M. (1968). A writer teaches writing: A practical method of teaching composition. Houghton Mifflin. Murray, D. (1984). Teach Writing as a Process. The Essential Don Murray: Lessons from America's Greatest Writing Teacher. Myhill, D. (2009). Children's Patterns of Composition and Their Reflections on Their Composing Processes. British Educational Research Journal, 35(1), 47 - 64. Retrieved from http://www.jstor.org/stable/40375555 Thomas, S. (2016). Future Ready Learning: Reimagining the Role of Technology in Education. 2016 National Education Technology Plan. Office of Educational Technology, US Department of Education. North, S. M. (1987). The Making of Knowledge in Composition: Portrait of an Emerging Field. Portsmouth: Boynton Cook Northwest Evaluation Association. (2013). Measures of academic progress (MAP). Lake Oswego, OR: Author. Nystrand, M. (2006). The social and historical context for writing research. Handbook of writing research, 11 - 27. Onwuegbuzie, A. J., & Leech, N. L. (2007). Validity and qualitative research: An oxymoron?. Quality & quantity, 41(2), 233 - 249. Papert, S. ( 1988). The Conservation of Piaget: The Computer as Grist. Constructivism in the computer age, 3 - 14. Pearson, P. D., & Gallagher, M. C. (1983). The instruction of reading comprehension. Contemporary educational psychology, 8(3), 317 - 344. Perl, S. (1979). The composing processes of unskilled college writers. Research in the Teaching of English, 13(4), 317 - 336. Perry, K. H. (2012). What Is Literacy? -- A Critical Overview of Sociocultural 309 Perspectives. Journal of Language and Literacy Education, 8(1), 50 - 71. Pillow, W. (20 03). Confession, catharsis, or cure? Rethinking the uses of reflexivity as methodological power in qualitative research. International journal of qualitative studies in education, 16(2), 175 - 196. Prensky, M. (2001). Digital natives, digital immigrants. On the horizon, 9(5). Pressley, M., & Afflerbach, P. (1995). Verbal protocols of reading: The nature of constructively responsive reading. Routledge. Prior, P. (2006). A sociocultural theory of writing. Handbook of writing research, 54 - 66. Reflector Director developed by Squirrels LLC ( http://www.airsquirrels.com ). Ranker, J. (2007). Designing meaning with multiple media sources: A case study of an eight - year - search in the Teaching of English, 41(4), 402 434. Rideout, V. (2017). The Common Sense census: Media use by kids age zero to eight. San Francisco, CA: Common Sense Media. cognitifs Rohman, D. G. (1965). Pre - writing the stage of discovery in the writing process. College composition and communication, 16(2), 106 - 112. Routman, R. (1995). Donald Graves: Outstanding educator in the language arts. Language Arts, 72(7), 518. Saldaña, J. (2016). Ethnotheatre: Research from page to stage. Routledge. Schumacher, G., Klare, G., Cronin, F., & Moses, J. (1984). Cognitive Activities of Beginning and Advanced College Writers: A Pausal Analysis. Research in the Teaching of English, 18(2), 169 - 187. Retrieved from http://www.jstor.org/stable/40170987 Schürer, O. (2006). Buil ding: Cognitive agent. In K. Oosterhuis & L. Feireiss (Eds.), The Architecture Co - Laboratory: Game Set Match II, On Computer Games, Advanced Geometries, and Digital Technologies (pp. 553 - 559). Auflage: Episode. Seipel, B., Carlson, S. E., & Clinton, V. E. (2017). When Do Comprehender Groups Differ? A Moment - by - Moment Analysis of Think - Aloud Protocols of Good and Poor Comprehenders. Reading Psychology, 38(1), 39 - 70. Shaughnessy, M. (1977). Some needed research on writing. College Composition and 310 Communication, 28(4), 317 - 320. Shepherd, R. P. (2015). FB in FYC: Facebook use among first - year composition students. Computers and Composition, 35, 86 - 107. Silby, A., & Watts, M. (2015). Making the tacit explicit: children's strategies for classroom wri ting. British Educational Research Journal, 41(5), 801 - 819. Simons, D. J., & Chabris, C. F. (1999). Gorillas in our midst: Sustained inattentional blindness for dynamic events. Perception, 28, 1059 1074. Standage, T. (1998). The Victorian Internet: The r emarkable story of the telegraph and the nineteenth century's on - line pioneers. London: Weidenfeld & Nicolson. Takayoshi, P. (2015). Short - form writing: Studying process in the context of contemporary composing technologies. Computers and Composition, 37, 1 - 13. Takayoshi, P. (2018). Writing in social worlds: An argument for researching composing processes. College Composition and Communication, 69(4), 550 - 580. The New London Group. (1996). A pedagogy of multiliteracies: Designing social futures. Harvard Educational Review, 66(1), 60 92. Tierney, R. J., & Shanahan, T. (1996). Research on the relationship: Interactions, transactions, and outcomes. Handbook of reading research, 2, 246. Tom Standage, 2005; 1998. The Victorian Internet: The Remarkable Story of the Telegraph and the Nineteenth Century's On - Line Pioneers Van Waes, Leijten, Lindgren, & Wengelin, (2011) Logging tools to study digital writing processes Taylor & Francis Van Waes, L., Leijten, M., & Van Weijen, D. (2009). Keystroke logging in wri ting research: Observing writing processes with Inputlog. German as a foreign language, (2), 41 - 64. Vygotsky, L. S. (1980). Mind in society: The development of higher psychological processes. Harvard university press. Waller, M. (2010). 'It's very very f un and ecsiting' - using Twitter in the Primary Classroom. English 4 -- 11, (39). Warnock, S. (2009). Teaching writing online: How and why. Urbana, IL: National Council of Teachers of English. Wason, P. C. (1968). Reasoning about a rule. Quarterly journal of experimental psychology, 20(3), 273 - 281. 311 Wengelin, A., Torrance, M., Holmqvist, K., Simpson, S., Galbraith, D., Johansson, V., & Johansson, R. (2009). Combined eye tracking and keystroke - logging methods for studying cognitive processes in text production Behavior Research Methods (2009) 41: 337. doi:10.3758/BRM.41.2.337 Wilkinson, A., Barnsley, G., Hanna, P., & Swan, M. (1980). The development of writing. English in education, 14(3), 1 - 12. Wollman - Bonilla, J. E., & Carpenter, M. (2003). E - mail as genre: A beginning writer learns the conventions. Language Arts, 81(2), 126. Wood, D., Bruner, J., & Ross, G. (1978). The role of tutoring in problem solving. Journal of Child Psychology and Psychiatry, 17, 89 - 100. Yin, R. K. (2014). Case study research: Design and methods. Sage. Yin, R. K. (2018). Case study research and applications: Design and methods. Sage publications. learning. The Journal of Educational Enquir y.