The Annotative Practices of Graduate Students: Tensions & Negotiations Fostering an Epistemic Practice

This research explores the annotation and note-taking practices of graduate students at a large Canadian university and reports on the sets of activities, habits, objects, tools and methods that define the practice. In particular, this empirical study focuses on understanding the integration of annotation and note-taking practices within larger scholarly processes such as reading, analyzing and writing. This study therefore aims to describe and analyze annotation and note-taking not only as the material externalities of the research process, but also as crucial epistemic practices allowing students to progress from one research activity to the other. Interviews are supplemented by document collection and analyzed using a multi-perspectival framework based on theories of practice. The findings describe an annotation lifecycle and suggest a new model of the scholarly process using annotation and note-taking practices as units of analysis. The study further discusses annotation as a primitive epistemic practice and examines the productive tensions and contradictions fostering the student’s progress towards her goals. Addressing annotation practices in light of the rise of digital textuality, this research finally proposes requirements for future tools supporting scholarly practice. p. 2

Lanzara (2010), echoing Brown & Duguid’s reasoning in The Social Life of Things (1996), highlights how long-established practices can be medium-specific, that is “the objects, tools, routines and representations that constitute the practice are specifically dependent upon the medium in which they have been formed and live” (pp. 1369-1370). This may explain the lack of enthusiasm and widespread adoption of reading technologies and devices for scholarly purposes. p. 13

In his essay entitled “Digital Documents and the Future of the Academic Community”, Peter Lyman warns the scholarly community against such assumption (Lyman, 1999). Lyman questions the continuity of scholarly communication, a term designed to encompass the academic research process, its infrastructures, and its outputs in the context of print publications and digital communications. The impact of new technologies in the academic milieu must be understood not only in terms of increased productivity and efficiency, but also in terms of the creation of new information products, leading to a shift in the core organization of academic work and in the overall academic culture. Lyman calls for a serious re-assessment of the consequences of the modes of production that govern knowledge work, and ponders “whether printed knowledge and digital information are used in the same way” (p. 367). p. 14

recent work by Unsworth (2000) and Palmer & Cragin (2008) hint at the pervasiveness of annotation practices within the scholarship cycle, terming such discrete activities “scholarly primitives”. According to Unsworth, these “basic functions common to scholarly activity across disciplines, over time, and independent of theoretical orientation” (np) form the basis of every higher-level scholarly activity. p. 15

Jackson (2001) also clearly calls for a distinction between notes in books and notes in notebooks. This study therefore considers note-taking separately from annotations. While some of these notes may have similar content and purposes as annotations, their storage on an external medium (such as a notebook) may influence the practice much more differently than annotations. Therefore, our study considers research-related notes not taken on the source material itself: 5 jottings on scraps of paper, comments in a text file, or passage and quotes copied and pasted into a word processing file. p. 16

Annotation is defined in our study as any extra information about another piece of information, attached to the original document without modifying it (Hunter, 2009). Jackson (2001) further distinguishes between marginalia (the descriptive or analytical notes made by the reader and found on the margins of the page), from the non-verbal codes and signs of the reader’s attention (such as asterisks, underlines and highlights). Our study assesses both types of markings. For instance, comments and notes written in the margin of a book and highlighted passages of a journal article are two types of annotations that are considered in this thesis. Additionally, the definition of annotation used for this study does not discriminate between medium. Hence, annotations created on paper using pens and highlighters are assessed alongside the notes and highlights created in a PDF file. Furthermore, this research takes into consideration the process of creation and use of these marks. Annotation is therefore understood as a process and a mark. p. 16

jottings on scraps of paper, comments in a text file, or passage and quotes copied and pasted into a word processing file. p. 17

Cathy Marshall, researcher at Microsoft Research, is often credited for the first modern in-depth studies of annotations in the field of human-computer interaction. Her numerous studies often address the form and functionality of paper-based annotations. Through empirical research, Marshall and her colleagues assess the materiality of paper annotations in order to recreate such practices in a digital environment, especially via the use of reading appliances. Her large study of annotated textbooks (Marshall, 1998) reveals the gamut of notation strategies of students. Highlights, underlines, marginal bars or brackets, symbols (such as asterisks), arrows, short phrases or extensive marginalia are all part of the student’s toolbox. In order to address this diversity, Marshall (1998, 2009) and Agosti, Ferro, Fommholz & Thiel (2004) establish a consistent terminology of annotation, aimed at the development of systems and mechanisms for creating, storing and using annotations in a digital context. Marshall expands on this anatomy in her book entitled Reading and Writing the Electronic Book (2009), building on the work of Agosti, Ferro, Frommholz and Thiel (2004) on the integration of annotation in digital library systems. It should be noted that these studies contain some very specific assumptions tainting their results: annotations are seen as an output of the reading process alone, with little consideration for the future usage of such information. Furthermore, they assess only in situ annotations created using a writing tool, leaving out such practices as bookmarking or annotating using removable media (bookmarks or sticky notes), physical annotations using the affordances of the support material (such as dog-earing) or the practice of taking notes in a separate location (such as in a notebook). According to this terminology, annotations are generally composed of three basic elements: a body, an anchor and a marker. The body of an annotation refers to the content that the user adds p. 22

to the source material, such as a lengthy comment in the margins or an idiosyncratic marking on the text (e.g., an asterisk or a proofreader symbol). In the case of a standalone, highlighted portion of text, the annotation has no explicit body. Marshall explains that these annotations are a common case of null-content. However, we should note that highlights and underlines might contain a very specific meaning or functionality and give an implicit content to the mark. p. 23

The anchor element defines the scope of the annotation and reveals the link between the source document and the added content. Anchors may be explicit (highlights and underlines spanning a precise passage of text) or implicit (a note written in the general area of a portion of the text, without any visual selection of the content to which it is referring). They may also be broad in scope (a note written on the first page of a journal article reviewing the entire content of the article) or narrow (proofreading marks related to a specific). The third element of an annotation is typically the marker. This element indicates how the anchor should be rendered when displayed. For instance, the marker of an underlined passage is the colored line under the selected text, the visual characteristics of the underline itself. While Marshall phrases this last element in a very specific manner in the context of the digital display of annotation, this item is crucial in a print-based context. p. 23

The body-anchor-marker trio revealed in the last section can be combined in multiple ways and to different effect. p. 24

Marshall (1998) examines four kinds of associations 1) Composite/Collection: notes referring to an entire article or chapter (figure 2); 2) Node-to-annotation links: comments or symbols localized within a document part but not visibly anchored (figure 3); 3) Standard association: comments or symbols visually anchored to a portion of the text (figure 4); 4) Word-to-word association: words or shorts phrases written in direct proximity to specific words in the text (figure 5). p. 24

Three mechanisms govern these associations. First, arrows and other lines are used to connect an anchored element in the document (an underlined or highlighted phrase) to its body. Second, brackets and braces, while themselves acting as both a link and an anchor, associate a commentary with the selected text. Finally, proximity is used as a mechanism for association, connecting markings to the text. p. 26

We supplement this analysis by including a collapsed association: some standalone markings such as highlights or underlines can effectively collapse the body-anchor-marker trilogy on itself. A standalone highlight acts as the anchor, body and marker altogether, referring uniquely to the selected highlighted word or passage itself without being linked to any other markings. The same type of association operates for symbolic reader’s marks such as asterisks, stars or other idiosyncratic markings. In fact, these symbolic markings are extremely versatile and may carry more meaning for the reader than highlights and underlines. In his book Used books: marking readers in Renaissance England, William Sherman (2008) explores the role of these marks in the reading strategies of historic readers. He first notes how the concurrent use of multiple symbols and letters in the marginalia of medieval and Renaissance readers appears to serve an indexing function, simultaneously classifying the information contained in the text and pointing out the interesting passages. Hence, these markings effectively collapse the body-anchor-marker trio. First, the symbol acts as an anchor: readers place these marks in the margin next to significant phrases or passages. However, the visual characteristics (the marker) of the symbol are generally not insignificant: these characteristics give meaning to the mark and constitute the body of these annotations. For instance, Sherman describes the annotation strategy of an anonymous reader of the 1569’s edition of Cicero’s De Oratore and states how the reader… p. 26

[…] tried out various symbols but once he had settled on a system he wrote it neatly toward the bottom of the page and surrounded it by a rough decorative border: he lists the signs used for tagging passages on particular topics (a trident p. 26

was used for passages of argumentation or reasoning and the symbol for Venus signaled an interest in love, and so on), and for marking particular rhetorical devices (such as “amp[lificatio], ” “metap[hor], ” and “sim[ile],” each of which is signified by a symbol that looks like a flower). (Sherman, 2008, p. 27) p. 27

Sherman then launches into an exploration of the manicule, these pointing hands drawn or printed in the margins of books (e.g., ☞). Manicules operate in a manner similar to other reader’s marks. However, Sherman asserts “that after a signature and a monogram, the manicule was the most personal symbol a reader could develop and deploy” (Sherman, 2005, p. 20). Hence, for the early modern reader, manicules represented far more than a simple mark. They symbolized their marks and “must have played an important role in the personal process of making a book meaningful” (p. 21). p. 27

The contrasting color of highlights and other marks is the variable responsible for effective emphasis and beneficial annotation strategies for undergraduates according to Fowler and Barker (1974). Several pedagogues in the education field including J. Wesley Miller recommend using this visual characteristic to distinguish between different topics (Miller, 1980). This echoes the earlier recommendations of 16th and 17th century scholars who endorsed the use of color to flag different types of passage as a means to cope with information overload (Blair, 2003). However, color-coded highlighting strategies appear to be rare for contemporary readers, probably because of the number of tools necessary to create such scheme and of the cognitive load involved in switching tools (Marshall, 1997). Annotation is mainly an unselfconscious activity (Marshall & Brush, 2004; Marshall, Price, Golovchinsky, & Schilit, 1999) and therefore, the act of putting down a highlighter pen of one color to grab the next color interrupts the broader reading task. p. 28

The examination of the visual characteristics of annotation opens the discussion to the interesting practices of dog-earing (Blair, 2003, 2004b; Schott, 2007; Daston, 2007), cutting and pasting (Blair, 2003; Sherman, 2008) and marking books with fingernails (Blair, 2003). These practices do not use any intermediary devices (such as pens or highlighters) to create their markings. Rather, they use the materiality of the book and play with the affordances of paper. Whereas p. 28

more visual methods such as commenting and highlighting are described as a separate layer of the document (O’Hara & Sellen, 1997; Miller 1980), these more physical and tactile practices affect the core of the document itself and alter the physical integrity of the book. p. 29

Second, this negative perception brings to light some of the reasons explaining why certain individuals hesitate to digitally annotate documents using standard, multi-purposes reading and writing software such as text editors. In their study of the differences between reading and annotating on paper and on screen, Kenton O’Hara and Abigail Sellen (1997) reveal that this reluctance to annotate on screen is caused by the reader’s perception that “emboldening, italicizing or underlining all alter the original document. Subjects indicated that they wanted to regard annotations as a separate layer of the document, and felt uncomfortable not maintaining this distinction” (O'Hara & Sellen, 1997, p. 137). p. 30

This example, as well as the previous discussion concerning the negative perception of annotation, point to a certain confusion between the text and the medium supporting the text. The dominating discourse here is one where the material qualities of the object (printed or digital) are subsumed into the text. Therefore, affecting or modifying the material qualities of the book is also affecting the text. p. 30

Annotation is thus much more than a simple trace left by the reader on a page: it is a visual- spatial system combining features such as color, shape, proximity, location and medium. Due to the meaning-making nature of annotation, its visual characteristics may best be defined and described in the context of the content and function of annotation. p. 31

Ovsiannikov, Arbib and McNeill (1999) provide one of the most explicit and focused discussion of annotation content in their article entitled “Annotation Technology”. Following their wide review of the literature, they define annotation content as “the semantics carried by the marking” (p. 337). This implies that annotations can be semantically explicit (and thus understandable by others than the annotator herself) or idiosyncratic and very personal in meaning. Also at play here is the notion of semantic distance between the annotation and the original text to which it refers. This semantic distance can be null (such as in the case of a highlighted word) or have a high value (such as doodles or remarks completely unrelated to the original text). Semantic distance is however not necessarily correlated to the semantic explicitness of the annotation: a note in the margin can be clear and spell out precise thoughts, but can be completely unrelated to the text under study. Cathy Marshall examines these two attributes at length in her seminal study “Toward an ecology of hypertext annotation” (1998) terming them “dimensions of annotations”. Marshall’s description of seven dimensions highlights the importance of a contextual understanding of the content of an annotation. It should be noted that these dimensions were devised with the specific intent of shaping design requirements for a subsequent reading device. The characteristics p. 31

discussed in this scheme are thus very explicit and often tied to general patterns of the practice that can be easily integrated in a digital environment. The seven dimensions devised by Marshall are reproduced in Table 2. p. 32

Functions of annotation are generally divided in two categories: higher-order functions and finer-grained functionalities. The subsequent sections address these two perspectives. Higher-order functions tend to be more general and address larger processes at play between the annotative act and the student. Conversely, finer-grained functionalities are derived from smaller processes and are often a direct result from the material characteristics of annotation. p. 33

In their survey of computer science students and professors, Ovsiannikov, Arbib & McNeill (1999) discuss four functions of annotation linked to four wider processes. Annotations are created and used to remember, think, clarify and share. As a means to remember, the authors observe that the visual contrast between the marking and the source document helps draw the attention of the reader and eases the task of re-finding information at a later time. An annotated document can be quickly searched for highlighted keywords in order to remember the text and its main points. Scholars of the history of the book also point to note-taking as a means to remember. Blair’s work on the Jesuit scholar Jeremias Drexel indicate that note-taking and excerpting were the preferred ways to memorize material, therefore supplementing the reader’s own memory (Blair, 2004b, p. 98). p. 33

Annotations are also created to help the reader think critically. This is normally true of marginalia including critical remarks, questions and notes written in the margin. Hence, annotations are an easy way to store new ideas prompted by the text with relatively minimal cognitive overhead. A third function exposed by Ovsiannikov, Arbib & McNeill is the use of annotation to clarify certain ideas in the text. The authors note that, contrary to annotations used for thinking, these clarifying notes do not carry new information and ideas. The benefits of these annotations are not necessarily immediate. They rather often involve a future usage: “[r]ephrased in the person’s own conceptual language, ideas are much easier to work with, which saves the reader time later when reading the paper again” (p. 336). The final function uncovered by the authors does not specifically stem from their survey results. It is rather an hypothesis to be tested in their annotation framework: annotation is used for comment sharing. This function is thought to be more frequent in the review process when the annotator must communicate information related to the source document to her collaborators. p. 34

Ann Blair, a scholar specializing in the cultural and intellectual history of early modern Europe, obtains similar results and discusses similar functions. In her paper entitled “Note Taking as an Art of Transmission” (2004b), she identifies storing, sorting, summarizing and selecting as the four basic “maneuvers” involved in annotating or note-taking. p. 34

Cathy Marshall (1998) reconstructs a considerable number of these functions by studying a wide sample of textbooks. Her results point to the metacognitive aspect of the traces left by readers. The first functionality she surveys is the use of annotation as a procedural signal, distinguishing between what is important for a certain task and what is not. This may be indicated by highlighted higher-level items (such as section headers), or conversely by crossed out paragraphs. As placemarks and aids to memory, annotations are used to store away potentially important information for future usage: emphasis symbols such as asterisks characterize such functionality. Third, annotations seem to be an in situ way of working problems. Explanations and notations found beside equations often demonstrate this function. We should note that this function is highly context sensitive and perhaps a consequence of her textbook sample. Marshall then identifies extended highlighting as a way of working through a complex narrative. Fifth, marginalia and in-text commentaries often function as a record of the interpretive activity of the reader. Finally, she considers markings that may simply be incidental: doodles, notes to self and drawings, however unrelated to the reading, often find their way into textbooks. p. 36

Weintraub’s exploration of the different modes of scholarship of humanists can be considered as a foundational text (Weintraub, 1979). While still very library-centric, Weintraub attempts to characterize the intellectual and material habits and needs of the scholar. Comparing the humanities work processes to the physical sciences, the strength of this paper can be found in the discrete characteristics of humanistic work that Weintraub reveals in passing, thereby igniting a trail of further studies. Most notably, he recognizes the influence of environmental and institutional factors on the work practices of the scholar, qualifies the humanist’s topic of interest as “open-ended”, and links this characteristic to a non-sequential workflow. Fabian (1986) echoes these two latter points, hinting at the constant refinement and readjustment process underlying the scholar’s work. p. 42

The work of Brockman, Neumann, Palmer and Tidline (2001) as well as Palmer and Neumann (2002) is crucial to our understanding of humanist practices. Departing from the traditional metaphor of the library as laboratory, both studies address the day-to-day practices of researchers, based on large-scale qualitative studies of respectively 33 and 25 participants. p. 47

Brockman, Neumann, Palmer and Tidline assess the practices of humanities scholars in light of recent technological developments. They identify four types of scholarly practices - reading, networking, researching, and writing - and decompose each of these practices into core activities. Note-taking and annotation practices appear to be intertwined with every type of scholarly practices. They furthermore allude to the varying function, format, and role of annotation depending on the type of practice deployed by the scholar. The creation, organization, and use of annotation then seem to be closely following the research process. Their findings remain p. 47

grounded in the materiality of practices and reveal the idiosyncrasies and difficulties of such intellectual work. They specifically single out the writing process, which appears to be complex and demanding even for the seasoned scholar. p. 48

Palmer and Neumann (2002) set out their study of interdisciplinary humanities scholars with the broad task of surveying “…the many elements that affect how research is carried out, on the individual, day-to-day level and within the larger work setting that includes institutions, organizations, and researchers’ domains of interest” (p. 87). Building on Case’s (1991) study, they identify the disciplinary norms and the institutional academic context as two influences on the work practices of scholars. The authors identify clear features of humanities research work: 1) while humanities scholars work from a core selection of resources, their path of inquiry is unpredictable, 2) multiple types of reading, often intertwined with the writing process, are practiced by the scholar throughout the project and, 3) technologies are adopted if they fit into established patterns of research (p. 98). p. 48

The work of John Bradley in the field digital humanities is especially worth noting here for its contribution to practice-based studies of the research cycle. Bradley’s work (Bradley 2005; 2008; Bradley & Vetch 2006) addresses the role of annotation in the analysis process of the scholar, based on the findings from Brockmann, Neumann, Palmer & Tidline and Palmer & Neumann. He proposes a tool, Pliny, to support conventional scholarly interpretation, surprisingly distancing himself from the current trend in digital humanities to propose tools transforming scholarly practice. As he grounds his software in practice-based research, he also brings forth the important metaphor of annotation, note-taking and writing as portal mechanisms, allowing the public sphere (i.e., the previous literature) to be brought back into the personal worksphere of the scholar and vice versa (i.e., allowing the dissemination of results back into the public sphere). p. 48

Emerging from the practice-based literature is the notion of the scholarly primitive, first brought forth by John Unsworth (2000). These “basic functions common to scholarly activity across disciplines, over time, and independent of theoretical orientation” (np) form the basis of every higher-level scholarly activity. Unsworth identifies annotating as one of the primitives, among p. 48

his list otherwise comprised of discovering, comparing, referring, sampling, illustrating and representing. As Palmer and Cragin (2008) rightfully pointed out, very few studies fully address scholarly work at this level of granularity, generally painting scholarly activities in broad strokes in the course of widely scoped research (p. 185). Palmer and Cragin define even finer grained primitives termed information work primitives including chaining, annotating and browsing. These primitives are normally associated with the larger information work processes of reading, collecting, searching and writing. They therefore distance themselves from Unsworth’s generalizing view and suggest that these primitives should not be extracted and generalized across domains or disciplines. Rather, and this is especially true for designers of information systems, they should be used as starting points for customization of distinct scholarly applications. p. 49

This recent work on scholarly and information work primitives is reminiscent of the concept of information-seeking features as described by Ellis (1989; 1993). Ellis argues that these features (starting, chaining, browsing, differentiating, monitoring, extracting, verifying and ending) interact and interrelate differently according to the unique situation of the researcher, pointing to the idiosyncratic nature of the individual’s research cycle (Ellis 1989, Wilson 1999). These features conceptually sit between the notion of primitive and research-phases. Indeed, features such as starting and ending point to a certain sequentiality of features in the research work of scholars, while the granularity of the activities brings them closer to the notion of primitive. p. 49

It is important here to distinguish between reading and writing as phases of the scholarship process (or, rather, an accentuation of the reading and writing activities for specific periods of time), and reading and writing as tasks and activities performed by scholars for a variety of purposes. p. 50

Kenton O’Hara undertakes perhaps one of the more thorough investigations of reading as a scholarly practice in his report entitled “Towards a Typology of Reading goals” (1996). His research aims to “deepen our understanding of how the affordances of paper support the task of reading and comprehending documents” (p. 4). O’Hara posits that readers have a variety of reading strategies at their disposal and that the appropriate strategy is determined by the reader’s goals and motivations. Annotation and note-taking are a support activities that may be tailored to different reading styles. These support activities help the reader achieve his reading goals within the navigational and manipulation constraints of the medium. O’Hara mentions that the reader normally selects the appropriate support activity (i.e., underlining, note-taking, outlining or networking) according to the selected reading style and the ultimate reading goal. p. 50

He distinguishes between four reading styles, based on Lunzer’s previous research on reading (1979): receptive, reflective, skim and scanning. According to his typology, a text can be read in a linear fashion or in an incomplete manner (serial vs. non-serial reading) and can be read once or repeatedly. These reading styles imply a different use of annotation and note-taking: a receptive reading (a style approximating listening behavior) will often yield only attentional marks such as underlines, while a reflective reading (a style where reading is often interrupted by reflective moments) may foster the use of note-taking and outlining. Reading goals also influence the choice of a specific support activity. O’Hara lists twelve reading goals: to learn, to self inform, to search/answer question, for research, to summarize, for discussion, proof-reading, to write and revise documents, for critical review, to apply, for p. 50

problem solving and for enjoyment. This typology hints at the pervasiveness of reading throughout the research cycle (e.g. to write and revise documents). This is demonstrated by our mapping of O’Hara’s reading goals to the research-phases model of Chu (1999) as seen in table 5. As with reading styles, different reading goals necessitate different support activities: reading to write and revise document may entail outlining and note-taking, while reading to search/answer question may only produce highlights and underlines. Consequently, this research provides two important aspects for annotation: 1) annotation and note-taking practices are performed throughout the research cycle, 2) different reading styles and reading goals generate different annotating and note-taking strategies. p. 51

Other studies refine and expand O’Hara’s typology by looking at specific disciplines and areas of scholarship. Brockman, Neumann, Palmer and Tidline (2001) expand O’Hara’s classification by tailoring the typology of reading to the humanist’s needs and purposes for reading. Findings indicate that scholars do background reading, comprehensive reading and continual reading. They read around a person or period and relate to primary materials by reading for details and to become immersed. Here again, the authors understand reading and annotating as intimately connected. p. 51

source materials often necessitate different annotation strategies due to their content (i.e., primary materials are analyzed differently than secondary materials) and their format. The latter highlights the impact of materiality on the choice of annotation strategy. For instance, the humanist’s primary material (e.g., books, documents) may often be fragile or only briefly accessible through the services of a librarian. Consequently, scholar must use removable media to take notes, or, are required to photocopy the source in order to highlight or write comments. p. 52

Many authors address the role of annotation in the natural sciences (Knorr Cetina, 2001, Latour & Woolgar, 1986; Rheinberger, 1997; 2003). While these markings fall outside of our research scope, Rheinberger’s insights concerning epistemic things and the materiality of research may be beneficial for our own study (see section 3.1.2.3 Objectual and Epistemic Practice for a discussion of epistemic objects). According to him, research scribblings are different from any other types of annotation or note-taking strategies. He states that they are closer to the “the materialities of scientific work than are research communications, insofar as the scribbles are quasi parts of the research objects, and therefore have a share in what will become the passive voice of those objects” (1997, p. 314). p. 52

Palmer and Neumann (2002) identify three distinct modes of reading for interdisciplinary humanities scholars (scanning, rereading, and reading for writing). The latter two generate a different type of written record of the intellectual work. According to the authors, rereading is facilitated by note-taking. Annotations and notes guide further reading and are conversely refined by subsequent reading. This is especially true of reading for writing, when original sources are reread. In this instance, notes are created to “fix the intellectual work of reading in a primitive form for future development” (p. 100). The annotations and notes created when reading for writing are therefore more than the documentation of an idea. They are formative in themselves, helping the scholar see the relationships between concepts and ideas. p. 52

Perhaps a key concept that has garnered sustained attention in this last decade is the notion of “active reading”. Originally described by Mortimer Adler in 1940, the notion of active reading opposes the prevalent assumption that reading is by nature a passive activity fueled by consumptive behaviors. Rather, Adler discusses the dynamic nature of reading and implores readers to actively engage with the text by annotating it: “reading, if it is active, is thinking, and p. 52

thinking tends to express itself in words, spoken or written. The marked book is usually the thought-through book.” (Adler, 1940a). p. 53

Active reading recently re-emerged in HCI studies of reading devices and interfaces. Active reading in most cases (Schilit, Golovchinski & Price, 1998; Kopak & Chiang, 2009) is the key concept justifying the integration of reader’s tools in screen-based reading interfaces designed for scholars. While some of these devices and interfaces eventually proved unsuccessful, the researchers nevertheless succeeded in building on Adler’s work by refining and augmenting the concept of active reading. Recent research has shown that active reading is highly goal-oriented (Kopak & Chiang, 2009; Wolfe, 2000; Marshall et al., 1999; Marshall and Brush, 2004), gives rise to a “critical engagement” (i.e., the “interplay between information as encountered and the analysis and use of that information” [Kopak & Chiang, 2009, p. 115]), and also involves chaining, chasing, and finding related material (Schilit, Golovchinski & Price, 1998). p. 53

O’Hara and Sellen (1997) directly link the failure of previous studies to uncover any significant difference between paper and screen to the measures used. They echo Dillon’s concern (1992) that a misconception of the nature of reading often taints the methodology and that researchers “seem to concern themselves with the control of so many variables that the resulting p. 54

2 See Dillon, 1992 for a thorough review of this type of literature, which, even if almost 20 years old, still provide valuable insights. p. 54

experimental task bears little resemblance to the activities most of us routinely perform as ‘reading’” (1997, p. 335). Their approach then involves an experimental task closer to the everyday activities of their participants: summarizing from multiple texts. Their findings unveil major differences in support activities such as annotation and note-taking. Paper affords multiple types of annotation to be made quickly, concurrently with reading. This is predominantly linked to the materiality of the document and to the interplay between multiple documents (or multiple pages of a single document), which may be displayed and rearranged quickly on a work surface. Marking on paper allows the reader to extract the structure of the text while providing the wider context for the marking. Conversely, taking notes away from the text (e.g., in a notebook) has a purpose of its own, which is to re-structure ideas and collate information from multiple sources. These notes normally take the form of outlines or plans, continuously modified during the course of reading and writing. The authors observe that even when the notes are taken away from the original source, the process is done without disrupting the main reading task, due to the spatiality afforded by paper (i.e., the spatial layout of documents on a work surface). p. 55

Conversely, reading online documents (in a browser) appears to be very different, especially when considering annotation and note-taking. Graphic elements inherent to the reading interface such as toolbars and scrollbars tend to hinder smooth reading and note-taking. Marking the document itself seems to be complex, if not impossible for many test subjects: the limitations and inflexibility of interaction techniques via mouse and keyboards disrupt the flow of reading. Moreover, annotating the original source (if the text was editable) is perceived as altering the integrity of the document. This differs from annotations on paper, perceived as an additional of layer on top of the original document. p. 55

The restricted field of view caused by the limited real estate of the screen is one of the main constraints of reading on screen. In their study, O’Hara and Sellen found that, since only a page or part of a page is displayed at once on a computer screen, participants tend to feel lost and that “much of the necessary contextual information for developing a sense of text and location lay beyond the window boundaries” (p. 340). Participants have to make a conscious choice when selecting and displaying specific documents since, unlike an actual workspace where one can lay out multiple document pages (Marshall, 1998; Ovsiannikov, Arbib & McNeill, 1999; O’Hara, Taylor, Newman & Sellen, 2002), the computer screen denies any periphery. p. 55

In her research on hypertext fiction reading, Anne Mangen (2008) investigates the influence of haptics on the sense of immersion a reader obtains when reading a text. She observes how different types of materiality impact our embodied reading experience differently. She claims that it is the intangibility and the volatility of digital reading that concretely affects the reading experience and fosters a shallow reading style. The materiality of the reading appliance (e.g., book or a screen) limits the scope of possible interactions. Materiality consequently affects the distance between the text and the reader and the location of the interaction. According to Mangen, interaction with digital texts “is experienced as taking place at an indeterminate distance from the actual text, whereas when reading print text we are physically and phenomenologically (and literally) in touch with the material substrate of the text itself” (p. 405). This distance created by the materiality of the text and of its reading appliance directly shapes the annotation practices of the reader. Hillesund reiterates the importance of bodily and material aspects of reading in his study describing various digital reading spaces: reading “in addition to decoding and comprehension, is also physical handling of reading appliances and that the way we read is dependent on technologies, implement design and text composition.” (np). p. 56

These arguments for the reconsideration of the bodily and material aspects of reading are reminiscent of the early discussions on the conventions and orders of the book brought to the fore by French historian and historiographer Roger Chartier and his lineage (Blair 2003; 2004a; 2004;, Topham, 2004). According to Chartier (1994), texts cannot be separated from their support or their materiality: the structure of the text supports a specific delivery format. Hence, two mechanisms operate in the creation of meaning for the reader: the strategies of writing and the author’s intentions, as well as the formal qualities of the book, often the result of a long series of publishing decisions and technological constraints. From this perspective, it is not the practice of reading that changes and shifts with the rise of digital media, but rather the support for meaning-making. The meaning of the text is altered when moving to digital texts: “one must state that forms produce meaning and that a text, stable in its letter, is invested with a new meaning and status when the mechanisms that make it available to interpretation change” (Chartier, 1994, p. 3). In sum, readers read more than the words (Topham, 2004). p. 56

Rowlands et al. (2008) claim that the short time spent viewing webpages or online article indicates that new forms of reading are emerging. Perhaps more intriguing is the “squirreling” behavior uncovered by Rowlands et al., and the numerous CIBER studies (Nicholas, Rowlands, Clark, Huntington, Jamali & Ollé, 2008). Academic users exhibit strong consumer instincts and “squirrel away content in the form of downloads, especially when there are free offers” (Rowlands et al., 2008, p. 295). p. 57

Writing has been described as the “complex interplay between the knowledge in the writer’s head and the physical and informational characteristics of the external source documents and compositions” (O’Hara, Taylor, Newman & Sellen, 2002, p. 280). p. 59

The authors dedicate a large section to annotations created and used while writing from multiple documents. They reveal that the creation of annotations on source documents also occur in the writing activity, particularly on printouts of digital documents and photocopies. Annotation and note-taking might be essential in the writing stage: the cognitive activity of writing, which entails a complex interplay between the knowledge in the writer’s head, the source material and draft documents, may only be manageable by breaking it into discrete sub-tasks such as annotating, note-taking and outlining. This externalization of knowledge prevents crucial information from being lost. p. 59

The authors take a distributed cognition approach to the types of links and notes created in such an instance: navigational links (such as “see notes in the red book”) are a fitting example of how meaning is distributed across internal (what and where is the red book?) and external representation (the navigational link), and reconstructed when necessary (p. 290). p. 59

Chartier posits that the relationship between the reader and the book is indeed a mediated one (Chartier, 1994; Slights, 2001). He thus departs from the widely accepted notion of “horizons of expectations” as advanced by Jauss “where each texts a reader encounters automatically evokes previous texts and their expectations and rules” (as described in Colclough, 2007). This encounter between the signal emitted by the text and the horizon of expectations of the public is pure and unmediated by materiality or technology. Chartier (1989) and Cavallo & Chartier (1999) reject this notion when addressing the changing materiality of the book throughout the history of reading in the West. Reading, here, is a practice that is realized in acts, places and habits. Meaning does not only emerge from the readable space, but also from the form through which texts are received: texts shift in meaning with every change of material support. p. 61

The concepts of social areas (Roger Chartier), reading communities (Ann Blair, Stephen Colclough) and interpretative communities (Stanley Fish) are crucial to our understanding of annotation as a social practice influenced by communities and their prevalent modes of thoughts. Daston’s concept of “on-paper communities” addresses this explicitly. She describes readers perpetuating social codes and conventions by creating and using annotations for specific meaning-making purposes and to demonstrate their membership to the community. Daston speaks of the necessity of the note-taking process to not only retrace the arguments advanced by authors, but also, and speaking specifically of the habits of readers annotating Descartes, “to participate in his radical doubt and gradual restoration of belief. The apparatus of annotations, citations, and footnotes enmeshed both author and readers in a web of other works, with their respective authors and readers” (Daston, 2004, p. 447). p. 61

Interpretative communities are at the core of Grafton (1991) and Jardine & Grafton’s (1990) interpretation of Gabriel Harvey’s Livy. Their study unearths a new type of reader (the reader as facilitator) and reveals the importance of the centrifugal mode of reading. By studying his heavily annotated Livy, Jardine and Grafton revealed that Gabriel Harvey (c. 1545 – 1630), notable scholar whose reputation suffered from various quarrels, acted as a facilitator for noble men and politicians, “easing the difficult negotiations between modern needs and ancient texts” (Jardine & Grafton, 1990, p. 35). The annotations remaining in his book demonstrate that Harvey read his Livy multiple times and for very different purposes, but always as a “trigger for action” (p. 40). p. 61

His annotations show that his readings were connected to other texts, and that he even perhaps read (often alongside his colleague/employer) with “the appropriate books open on the table before them” (44). From the amount of works and books referenced, Grafton and Jardine posit that Harvey might have read using a book wheel (see figure 7). This demonstrates how Harvey was part of a wider community of readers, as the book wheel “belongs to Harvey’s cultural moment, in which collation and parallel citation were an essential, constructive part of a particular kind of reading; it allowed the imbedding of text in context, after the fashion that Harvey and (we would argue) many of his professional academic contemporaries practiced” (p. 48). The potential presence of the book wheel in his reading practice suggests that Harvey might have been perceived (or wanted to be perceived) as a “skilled reader”, just like other individuals may be skilled woodworkers. In sum, the annotations found in Harvey’s Livy are indicative of a centrifugal mode of reading, where the reader pulls information from multiple texts. This centrifugal reading practice is supported by the appropriate tools such as the book wheel, which situate Harvey as part of a wider community of readers. p. 62

Data was analyzed using a strategy similar to grounded theory. Audio transcriptions and visual documents were analyzed first using an open coding technique. Open coding lets categories emerge (Babbie, 2001, p. 366) and is thus less directly related to the researcher’s preconceived social constructs. As coding progressed, a constant comparative method was used in order to saturate the categories by constantly referring the fresh data to emerging categories (Creswell, 1998, p. 57). Tams Analyzer (Weinstein, 2008), an open source qualitative analysis tool, proved to be useful for this first analytical step. p. 88

Once the open coding ended, the researcher engaged in axial coding, looking for “conditions, strategies, interactions and consequences” (Strauss, as cited in Berg, 1998, p. 240) and assembling the emerging categories in new ways. Here, evidence of recurring themes were gathered, data was searched for the various strategies and aspects of the annotation and note- taking practices, and emergent themes were compared to themes and constructs gleaned from the literature review. p. 88

Our study reveals that quick jottings are very frequent in the early days of the research project as the student attempts to establish the theme, scope and boundaries of the research. It is also at this stage that participants such as Maxine create visual overviews of their current knowledge and of their prospected research through mindmaps. p. 100

Michael Bernstein’s team in their impressive study of knowledge workers and personal information scrap (2008). p. 131

However, there is also a need for these services to also be integrated with other scholarly activities such as bibliographic management. We should note the efforts of the Open Annotation Collaboration (http://openannotation.org/) project currently underway. This project, while focused on web annotation, attempts to link multiple information spaces such as vendors, bibliographic management software and annotation software, in order to integrate these and p. 194

create a hybrid space offering tools for annotation interoperable with these different services. This would let the scholar use the appropriate tools for her research without worrying about information fragmentation or loss of information and annotation when moving back and forth across services. p. 195

Adler, M. J. (1940a). How to Mark a Book. Saturday Review of Literature, 6, 250-252. Retrieved from http://bhs.cc/ciswriting/wp-content/uploads/2007/02/how to mark a book.pdf. Adler, M. J. (1940b). How to read a book; the art of getting a liberal education. New York, NY: Simon and Schuster. p. 204

Bazerman, C. (1988). Shaping written knowledge: The genre and activity of the experimental article in science. Madison, WI: The University of Wisconsin Press. p. 204

Bernstein, M., Van Kleek, M., Karger, D., & Schraefel, M. C. (2008). Information scraps: How and Why Information Eludes Our Personal Information Management Tools. ACM Transactions on Information Systems, 26(4), 1-46. p. 204

Bradley, J., & Vetch, P. (2006). Supporting Annotation as a Scholarly Tool–Experiences From the Online Chopin Variorum Edition. Literary and Linguistic Computing, 22(2), 225-241. p. 205

Brittain, J. M. (1984). Internationality of the Social Sciences: Implications for Information Transfer. Journal of the American Society for Information Science, 35(1), 11-18. p. 205

Dillon, A. (2010, June). As We May Have Thought, and May (Still) Think. Keynote Address presented as the ACM Hypertext 2010 Conference, University of Toronto, Toronto, Canada. p. 206

Fowler, R. L., & Barker, A. S. (1974). Effectiveness of highlighting for retention of text material. Journal of Applied Psychology, 59(3), 358-364. p. 206

Hillesund, T. (2010). Digital reading spaces: How expert readers handle books, the web and electronic paper. First Monday, 15(4). p. 206

Hunter, J. (2009). Collaborative Semantic Tagging and Annotation Systems. Annual Review of Information Science and Technology, 43(1), 187-239. p. 207

Hutchinson, D. (2007). Teaching practices for effective cooperative learning in an online learning environment (OLE). Journal of Information Systems Education, 18(3), 357-367. p. 207

Kopak, R., & Chiang, C. (2009). An interactive reading environment for online scholarly journals: The Open Journal Systems reading tools. OCLC Systems & Services: International Digital Library Services, 25(2), 114-124. p. 207

Mangen, A. (2008). Hypertext fiction reading: haptics and immersion. Journal of Research in Reading, 31(4), 404-419. p. 208

Nicholas, D., Huntington, P., Jamali, H. R., Rowlands, I., Dobrowolski, T., Tenopir, C., et al. (2008). Viewing and reading behaviour in a virtual environment: The full-text download and what can be read into it. Aslib Proceedings, 60(3), 185-198. p. 208

Tutty, J. I., & Klein, J. D. (2008). Computer-mediated instruction: a comparison of online and face-to-face collaboration. Education Technology Research and Development, 56(2), 101-124. p. 211