Adam Worrall

Site Navigation

To conduct research on a given topic and build our collective understanding of it, significant knowledge of what has come before is necessary. This chapter provides this necessary context, reviewing the literature of importance and relevance to the study of social digital libraries as boundary objects within and across social and information worlds. It first looks at digital libraries and differing conceptions held of them by researchers and practitioners. Next, conceptions of communities and collaboration, two phenomena of great contextual importance to social digital libraries, are examined. Further discussion of social digital libraries illustrates their place within this greater context. Previous research on social digital libraries, including studies, models, and frameworks with varying degrees of success, is reviewed next. The chapter then presents a brief review of literature on virtual book clubs followed by an extended review of boundary object theory, the central part of a well-grounded, context-sensitive, flexible theoretical framework for studying social digital libraries. The development of this framework based on boundary object theory, the social world perspective, and the theory of information worlds is also presented. The chapter concludes with a brief discussion of the implications of this framework for social digital libraries.

2.1. Digital Libraries

2.1.1. Pre-History

Bearman (2007) dates the term digital libraries to 1991, with common use dating to 1993. The research stream and concept can be extended further back (see Lynch, 2005) to Kahn and Cerf (1988); earlier work on databases, online public access catalogs, and information retrieval systems; and as far back as Licklider (1965), Bush (1945), and Otlet (see e.g. Rayward, 1997). Bush’s (1945) memex concept may not have been a digital library as we would see it today, but one can consider it a forerunner of digital libraries, information systems, and the Internet as a whole. Smith’s (1981, 1991) citation analysis of documents citing Bush’s ideas found citations picked up after 1980, which she attributes “at least in part to the association of Bush with concepts similar to those underlying hypertext” (p. 264), a key element in most modern digital libraries. She used content analysis to break citations into five categories; the majority fell into the first, historical background or perspective. This indicates Bush’s work as an important milepost in and beginning of the history of information systems and the understanding of the problems surrounding information organization, seeking, and use; it was seen as “the starting point of modern information science” (Smith, 1981, p. 352). Hypermedia (as discussed by Smith, 1991) and digital libraries (see Kahn & Cerf, 1988; Lynch, 2005) build on much of this work in information retrieval; Smith (1991, pp. 269–270) alludes to what would become digital libraries for specialized communities. Bush is an important founding influence on information science and a key originator of the concepts and ideas behind digital libraries.

“By the mid–1980s,” according to Lynch (2005, para. 4), “there were systems … that might reasonably be considered digital libraries at least by some definitions” and early concepts of what digital libraries might be. Kahn and Cerf (1988), providing the earliest modern conception, defined a digital library as “a rich collection of archival quality information … of current and possibly only transient interest” which blended “the conventional archive of current or historically important information and knowledge … with ephemeral material such as drafts, notes, memoranda and files of ongoing activity” (p. 3). Kahn and Cerf focused on the idea of a digital library as a collection of varied data, information, and knowledge, and as part of a broader Internet-based network of digital libraries.

2.1.2. Emergence: Two Differing Conceptions

When the term became common in the early-to-mid 1990s, U.S. federal agencies engaged in major funding efforts, under the Digital Library Initiative (DLI) banner, to spur the development of digital libraries. In these still relatively early years, “already the term … [was] used to describe [many] entities and concepts” (Borgman, 1999, p. 228). Reviewing these, Borgman found two differing conceptions of what a digital library is. One definition, followed by many researchers at the time, considered digital libraries as “[digital] content collected on behalf of user communities” (p. 229). This view developed out of a National Science Foundation (NSF)-sponsored “Social Aspects of Digital Libraries workshop” in 1995 (p. 234); it was a broad extension of Kahn and Cerf’s (1988) definition. Another conception, followed by many librarians and other practitioners, considered digital libraries as “institutions or services” (Borgman, 1999, p. 229). This conception stemmed from the Digital Library Federation’s (DLF) definition, first given by Waters (1998, as cited in Borgman, 1999, p. 236) who considered digital libraries as “organizations” which provided “resources” and various services—paralleling traditional libraries—surrounding “collections of digital works” that could be used “by a defined community or set of communities.” As Borgman stated (p. 236), this conception “captures a much broader sense of the term ‘library,’” but required there to be an institution offering services alongside the collection of digital content placed on the Internet.

Borgman (1999) believed much of the divide between and within the research and practice communities came from the wide variety and stages of work being done on digital libraries, in multiple disciplines and from multiple perspectives. Such digital library research and practice, from about 1995 to 2005, varied from theory- and model-building, to module development, to prototype construction, to entire system implementations, to studies of digital library use. Borgman found definitions had expanded beyond “enabling technologies” to include the contexts of digital library use—“social, behavioral and economic”—within the full cycle of information behavior and information resources ( p. 240), paralleling similar changes in the library and information science (LIS) field as a whole and in information behavior research (see e.g. Case, 2012; Courtright, 2008; Raber, 2003).

Digital library textbooks have pulled from both sides of the divide, but often retain bias towards content- and technology-centric conceptions. For example, Arms (2000) defined digital libraries as “a managed collection of information, with associated services, where the information is stored in digital formats and accessible over a network” (p. 2). This includes content- and service-based elements, but emphasizes the collection and the technology used to store, access, and organize it, as shown by Arms’s claim that “data … when organized systematically, becomes a digital library collection” (p. 2). Lesk (2005), writing five years later, employed a similar definition: “a collection of information which is both digitized and organized” (p. 2). While showing similar bias to Arms, Lesk listed content, access, and services as key elements of a digital library and discussed the “social effects” (p. 2) digital libraries can have on their users, user communities, and cultures. Despite this, both authors’ textbooks emphasized content, organization, and access over services, community, and culture.

2.1.3. Present-Day

While many of the tensions and divides between researchers, practitioners, and disciplines are still present, most in the digital library field have adopted broad definitions, including the ideas of both collections of content and services to users. The ACM/IEEE Joint Conference on Digital Libraries (JCDL), to give a notable example, “encompasses the many meanings of the term ‘digital libraries’” from information retrieval system and digital content collection to digital library institutions and digital content in social, cognitive, and organizational contexts (“About JCDL,” 2012), including each of the camps Borgman identified. This broad-based and flexible approach is seen in Bearman’s (2007) recent review of the field.

Bearman (2007) was not without views of what digital libraries are, how they should be conceived, and which contexts and aspects are most important. He maintained digital libraries are “not mere technical constructs” (p. 251), not simply information retrieval systems or databases; instead, they are inherently social organizations and environments, socially constructed (Talja et al., 2005; Tuominen & Savolainen, 1997) by users, communities, organizations, and other key stakeholders. His view required the inclusion of each of the camps Borgman identified: content collected for a community and services provided to the community by an organization or institution, with a focus on the social contexts and aspects of digital libraries. Such a consideration was far from new, as will be returned to in section 2.4. First, consideration of the two concepts of communities and collaboration is necessary, as important context for social digital libraries.

2.2. Communities

The concept of community is important in many different fields, but it is not one with a universal definition across or within disciplines. Sociology, concerned with human society and the communities they form, is often the source for conceptions of community adopted in LIS and cognate fields. Sociology featured 94 different definitions of community by the 1950s (Hillery, 1955); a common thread in all of these definitions was people, the human society sociology takes as its main interest. A majority of the definitions included elements of social interaction (97%), tie(s) between individuals (78% with both of these), and an “area” (73% with all three) (p. 118). Hillery concluded a core definition of community would include social interaction between individuals, within an area, with common ties.

Total agreement on these core areas is impossible to reach, even within the field of sociology. The 1970s were a period of tension for defining community; while common elements were similar to Hillery’s (see Jones, 1995, p. 21), there were tensions emerging underneath. The “ecological” approach popular in the first half of the 20th century, theorizing interpersonal ties, social structures, and social norms were weakened due to anomie caused by large aggregations into communities, had been called “into great doubt” (Fischer, 1975, p. 1321). Great numbers of personal ties and smaller groups were being found in large urban communities, leading to a “nonecological” approach to communities becoming prevalent (see also Wellman, 1982, 1999). Fischer (1975, p. 1321) presented an influential counterargument to these two approaches, feeling they had “a serious flaw” in either assuming the presence of anomie or ignoring ecological factors in communities. He proposed a “subcultural theory” of urbanism (p. 1323)—one which can be extended to all types of communities—drawing from both ecological and nonecological approaches. His theory concluded larger communities, such as cities, are composed of numerous smaller subcultural communities, each with somewhat different norms, social structures, and ties.

Others, influenced by Fischer’s theory, took a similar approach. Wellman (1982) argued for an emphasis on the “structures of communities” and “the larger institutional contexts in which … [community] networks were embedded” (p. 63). Wellman echoed Fischer in noting “community ties are not bound up in solidary clusters” and multiple community ties give “members ramifying, indirect connections to other social circles” ( p. 79). Besides a subcultural, multi-level, and social network approach to studying community, Wellman argued for studying the context of ties, given “ties link persons and not specific strands” (p. 79).

The following overview of community conceptions focuses on concepts having seen recent use in the LIS field or fields cognate to it. It draws on Ellis, Oldridge, and Vaconcelos’s (2004) review article but adds additional concepts found in the literature on information behavior, digital libraries, and knowledge management (KM).

2.2.1. Communities of Practice

Communities of practice (often abbreviated as CoPs) originated in the early 1990s out of Lave and Wenger’s studies of situated, organizational learning (Brown & Duguid, 1991; Cox, 2005), first published in 1991 (Lave & Wenger, 1991). CoPs are groups of people who share a practice—a broad activity all in the group are engaged in, often a profession—and who are learning this practice through situated, social learning and interaction (Brown & Duguid, 1991; Wenger, 1998). Emphasis on the situated learning element was strongest in Lave and Wenger’s (1991) initial formulation, where they focused on the process of legitimate peripheral participation (LPP) and on the reproduction of communities through it: newcomers to a community become acclimated and learn the practice from experienced members, similar to apprenticeship. Later perspectives and uses of CoPs (Brown & Duguid, 1998, 2001, 2002; Wenger, 1998, 2006; Wenger, McDermott, & Snyder, 2002) have been more ambiguous and less grounded in theory, focusing on the role such communities can play in knowledge management, sharing, and transfer within the broader organizations they are part of. These later views share affinity with Fischer’s (1975) subcultural theory and include at least two of the three characteristics discussed by Hillery (1955).

Many are critical of CoPs, arguing their true theoretical core has become lost in translation and ambiguity (Cox, 2005; Handley, Sturdy, Fincham, & Clark, 2006; Hughes, 2007; Murillo, 2011; Storberg-Walker, 2008), they lack a strong consideration of power issues (S. Fox, 2000; J. Roberts, 2006), and face difficulty in balancing emergent vs. managed communities (Chanal & Kimble, 2010; J. Roberts, 2006). Key originators of the CoP concept have criticized the shifts in its use over time (Duguid, 2005, 2008; Lave, 2008). The biggest shift, from an initial theoretical basis as an emergent phenomenon of situated learning to a prescriptive and pragmatic approach believing communities can be created and managed for the purpose of knowledge sharing, is a key limitation of CoPs. Most views of CoPs are limited in their definition of engagement and by focusing on practice.

Brown and Duguid’s (1998, 2001, 2002) revised version of CoPs has the most potential benefits for studying social digital libraries: it eliminates the strong limitations in learning, apprenticeship, LPP, or reproduction earlier versions have; includes a multi-level view with the introduction of networks of practice; grounds CoPs in KM concepts; and incorporates relevant research on common ground (cf. Bechky, 2003; Davenport & Prusak, 2000, p. 98; Olson & Olson, 2000, pp. 157–161; Wasko & Faraj, 2000), boundary spanners (cf. Brown & Duguid, 1998; Leonard-Barton, 1995; Levina & Vaast, 2005), and boundary objects (see section 2.7 below). Unfortunately, it relies on limited research and observation, is vague on the topic of engagement, (like all versions) struggles with the debate between emergent and designed / managed CoPs, and is not well-grounded in theory.

2.2.2. Virtual Communities

The concept of virtual communities is used in many disciplines, with researchers using different definitions but maintaining common characteristics (Ellis et al., 2004; Preece & Maloney-Krichmar, 2003). Rheingold (1993, republished in 2000) was an early user of the term, defining virtual communities as “computer-mediated social groups” ( p. xv) and as “social aggregations that emerge from the [Internet] when enough people carry on … public discussions long enough, with sufficient human feeling, to form webs of personal relationships in cyberspace” ( p. xx). Such a definition features social interaction between individuals who form social ties, two of the three areas suggested as core to sociological definitions of community by Hillery (1955), and implies such communities are emergent social constructions (cf. Tuominen & Savolainen, 1997). Geographical boundaries were not a part of Rheingold’s (2000) conception. His conception falls within the subcultural view proposed by Fischer (1975): “there is no such thing as a single, monolithic, online subculture … it’s more like an ecosystem of subcultures” (Rheingold, 2000, p. xviii). Although Rheingold’s book focuses on the virtual community known as the WELL (Whole Earth ‘Lectronic Link), he stressed the different subcommunities present within—and often across—its boundaries, a view in clear agreement with Fischer (1975) and Wellman (1982).

Virtual communities have, of course, shifted since 1993. Ellis et al. (2004) discussed the concept’s relations with other community concepts, indicating the “growing up” of virtual communities as a concept and as a social phenomenon. Kraut, Wang, Butler, Joyce, and Burke (2008) defined online communities as “large, persistent collections of individuals with common or complementary interests whose primary method of communication is over the Internet” (p. 1); social interaction and social ties were included, but again a geographical area was not. Common replacements for the latter as motivation and support for the community to organize—and as potential limits to the concept’s applicability—include a “shared goal, interest, need, or activity”; “repeated [and] active participation”; “strong emotional ties”; “access to shared resources”; “reciprocity of information, support, and services”; and “shared context” in the form of “conventions, language, [and] protocols” (Whittaker, Isaacs, & O’Day, 1997, p. 29). Haythornthwaite’s (2007) focus on social interactions and ties between members fell closer to prior sociological definitions of community, discarding most of Whittaker et al.’s (1997) additional characteristics. Others associated with LIS, sociology, science and technology studies (STS), and computer-supported cooperative work (CSCW) have stressed the boundaries of virtual communities and how they are crossed (or not) as an important factor (Burnett et al., 2001; Star et al., 2003). Burnett, Dickey, Kazmer, and Chudoba (2003) considered virtual communities as “consistent with the concept of a community of practice” (Virtual Community section, para. 3), but virtual communities breach most of the limitations of the latter discussed above.

2.2.3. Arenas

The concept of arenas was developed by Strauss, Schatzman, Bucher, Ehrlich, and Sabshin (1964), who found existing models of hospitals to be lacking for explaining “psychiatric philosophies and … associated daily practice” (p. 3). They developed arenas to examine the “social organization” and “social process[es]” around psychiatric philosophies (p. 14), focusing on levels of agreements and negotiations, including of norms, rules, and values often taken for granted. External and internal interactions and negotiations between smaller subgroups of the organization were included. Strauss et al. argued the arena concept could be extended beyond hospitals to looser and more informal “organizations” such as professions, hobbyists, and enthusiasts. Arenas parallel later work on CoPs (Ellis et al., 2004) and foreshadow later work in organizational science, STS, and KM, albeit each of these placing different spins on similar broad ideas. From a sociological perspective, arenas provide an approach to communities similar to Fischer’s (1975) subcultural theory; using the arena concept, one could apply multiple levels of analysis in examining an organization, its subunits, and any superordinate units. It includes all three of Hillery’s (1955) core criteria: social interaction, ties between individuals (via the organization), and a form of area (the arena), although not always a geographic one. Arenas include contextual factors of interest—agreements, norms, rules, commitments, and interactions—which are similar to the processes of key interest in CoPs or the concepts within the theory of information worlds (discussed in sections 2.2.6 and 2.8.2). Arenas have broad application, but their focus on organizations, agreements, and commitments limits their flexibility.

2.2.4. Social Worlds

Strauss’s (1978) later concept of social worlds is more flexible than his earlier development of arenas. Strauss built on the work of Shibutani (1955); each studied sociology at the Chicago School with Blumer (see Becker, 1999, p. 8; Clarke & Star, 2008; Wacker, 1995, p. 146). Shibutani examined the concept of “reference groups” and argued it should be restricted to the perspective of a group from which the actor sees him/herself as acting, constituting the actor’s frame of reference. He stated these groups “arise through the internalization of norms … [and] constitute the structure of expectations” belonging to a group which an actor chooses to relate to, but may not belong to (p. 565). An “amazing variety” of these norms and beliefs led to a similar “variety of social worlds,” each with its own “organized outlook” based on the interaction and communication of a given group of people (p. 566). Their boundaries were based on communication, not “territory” or “formal group membership” (p. 566). Shibutani never clarified the relationship between the reference group and social world concepts, appearing to equate them in places while maintaining vague differences elsewhere.

Strauss (1978) proposed “a social world perspective” (p. 121), moving beyond his focus on organizations as arenas and Shibutani’s on communication. Strauss argued social worlds consist of social organizations, processes, communication, “activities, memberships, sites, [and] technologies” in relation to social change (p. 121). He further stressed the differences in size, spatial ties, visibility, privacy, abstractness, permeability, and structure of social worlds. A Straussian social world was theorized to include (a) “at least one primary activity … strikingly evident”; (b) “sites where activities occur”; (c) “technology … [for] carrying out the social world’s activities”; and—in established social worlds—(d) “organizations … to further one aspect or another of the world’s activities” (p. 122). Strauss believed social worlds would “intersect … under [various] conditions” (p. 122), and could segment into smaller subworlds given sufficient analysis.

Strauss’s social world perspective relates to other conceptions of community, including his earlier concept of arenas (Strauss et al., 1964), now defined as organizations where representatives of subworlds would negotiate conflicts and issues between them, impacting on the broader social world (Strauss, 1978). Arenas focused on the negotiation and agreement aspects of organization and social practice. Given the similarities to CoPs, one could consider substituting the latter for arenas in the social world perspective, but CoPs have a more restricted perspective; social worlds are not restricted to practice, online environments, or organizational negotiation. There are still a few limitations provided by the framing concepts of activities, sites, technologies, and—in established cases—organizations, but Strauss intended for broad and flexible interpretation of these. Unlike Lave and Wenger’s learning-focused practice, Strauss’s activities could be almost anything that could be labeled as such. The social worlds perspective falls within the social constructionist (Talja et al., 2005; Weinberg, 2009) and social constructivist (Clarke & Star, 2008) perspectives, given its focus on “universes of discourse” (as termed by Mead; see Clarke & Star, 2008, p. 115).

2.2.5. Social Networks

Social network analysis and the social network perspective have been used to describe and study communities of all kinds; they are most often used when patterns of interaction, information exchange, and knowledge transfer are of interest. Part of the goal of the social network approach is to “tease out the prominent patterns” and “trace the flow of information (and other resources) through” the ties and relations of people and organizations (Garton, Haythornthwaite, & Wellman, 1997, The Social Network Approach section, para. 3). Freeman (2004, p. 2) termed it a “structural approach … based on the study of interaction among social actors.” Many researchers study what effect relations between people, organizations, and the associated networks have on the behavior and activities of these individuals and groups (Garton et al., 1997). Wellman (1999, p. 15) has argued a network view of community reflects that “the paramount concerns of sociologists are social structures and social processes—and not spatial groupings.” Communities under this perspective can overlap and interact, given people at the far ends of a network will not be aware of each other (see also Milgram, 1967) and should be considered members of different communities. As discussed above, Wellman’s thinking fits with Fischer’s (1975) subcultural theory, and many social network analysts share this view. The focus on social ties represents one of Hillery’s (1955) three core criteria for conceptions of community, and social interaction is implicit—a social network without interaction is a set of isolated nodes—but social network analysis does not require a bounded “area.” Conceptualizing communities as social networks falls close in line with other sociological concepts, remaining compatible with the concepts of CoPs, virtual communities, arenas, and social worlds. It has one major limitation, requiring community to be viewed through the “social network” lens; this is still less constraining to many than a basis in situated learning and practice, a required common goal, or a focus on negotiation and norms.

2.2.6. Information Worlds

While its origination differs, a related concept to social worlds is information worlds, developed as a theory by Burnett and Jaeger (2008; Jaeger & Burnett, 2010) and discussed at length in section 2.8.2 below. As a conception of community, it includes social interaction, ties between individuals, and a bounded—albeit non-physical—area, the information world. While it includes all three of Hillery’s (1955) core elements, it treats each one in line with the subcultural theory proposed by Fischer (1975). The normative behavior in and mutual influence between information worlds (see section 2.8.2 below) are indicative of how such worlds are socially constructed (cf. Talja et al., 2005; Tuominen & Savolainen, 1997).

Due to Burnett and Jaeger drawing on Chatman’s (1991, 1992, 1996, 2000) small worlds and Habermas’s lifeworlds, the concept and theory of information worlds are compatible with Strauss’s (1978) conceptions of arenas and social worlds at multiple levels and scales of analysis. The theory’s focus on information behavior echoes Shibutani’s (1955) earlier focus on communication, albeit Burnett and Jaeger’s work has stronger grounding in previous theoretical and practical work in LIS. This grounding is stronger and firmer than that of CoPs in KM, given the latter did not start out as a KM theory (as seen in Lave & Wenger, 1991; see also the discussion by Cox, 2005). Virtual communities can be considered “public information worlds rooted in group-based social interaction … [and] almost textbook examples of small information worlds,” as discussed by Jaeger and Burnett (2010, p. 93); a social network perspective to studying one or more information worlds would be compatible (albeit not required). The theory and concept are compatible with and can incorporate many of the other concepts reviewed above.

The main weaknesses of information worlds are its foci on norms, types, values, information, and information behavior; the theory and concept have less direct applicability in studies where knowledge, learning, or other phenomena are of greater interest. Their strength for LIS studies and studies of digital library communities are in their flexibility and compatibility with other conceptions, theories, and perspectives, while focusing on specific analysis of norms, types, values, behaviors, and boundaries. Combined with other theories and concepts, information worlds can serve as part of a strong theoretical framework for social digital libraries, discussed later in this chapter.

2.3. Collaboration

Collaboration is another concept important to social digital libraries, one “as common and natural a form of information behavior as individual seeking” (Talja, 2002, p. 9). Although kinds of and contexts for collaboration are well-defined—e.g. scientific collaboration (Sonnenwald, 2007)—it is rare for the root concept to be specified in great detail and “there is no widely accepted definition of collaboration” (Hansen & Järvelin, 2005, p. 1102). Nevertheless, there is significant literature discussing the concept of collaboration in multiple research circles.

The literature most germane to collaboration in and around digital libraries is from information seeking and retrieval, computer-supported cooperative work (CSCW), and scientific collaboration research. Hansen and Järvelin (2005) provided a useful review of collaboration in the first two of these areas, noting “the importance of personal contacts and discussions” (pp. 1102–1103), the key role of “gatekeepers in organizations” (p. 1103; cf. Brown & Duguid, 1998; Leonard-Barton, 1995; Levina & Vaast, 2005), and research on the communication patterns and networks in an organization. Research had found information sharing takes place in multiple ways (including formal and informal), using multiple source types, and at multiple levels (Hansen & Järvelin, 2005). Talja’s (2002) in-depth study found sharing and collaboration could be (a) strategic, consciously “maximizing efficiency”; (b) paradigmatic, establishing a new research area, field, or practice; (c) directive, “between teachers and students”; or (d) social, building relationships and communities without a strict goal or task (p. 4); sharing could also be nonexistent. Social interaction is more important in sharing “interpreted information” than factual information (Hansen & Järvelin, 2005, p. 1104), similar to findings on the difficulty in sharing tacit knowledge (Nonaka, 1994). Hansen and Järvelin’s (2005) review of CSCW studies found cooperation and collaboration fell on a set of continua: asynchronous vs. synchronous activities, traditional communication vs. CMC; loosely vs. tightly coupled activities; and different degrees of awareness of information and information sources. This view provides a large amount of flexibility, but without detailing what occurs during collaboration beyond sharing information. An extensive review of CSCW by Mills (2010) included human-to-human communication, coordination of common activities, access to information (topical and to support the collaboration), and support for interactions—and the resulting collaboration—across space and time under a view of collaboration.

Sonnenwald’s (2007) definition of scientific collaboration, presented in the context of a thorough literature review, is similar to the CSCW-based definition given by Mills (2010): “interaction taking place within a social context among two or more scientists,” with this interaction facilitating “the sharing of meaning and completion of tasks,” tasks part of “a mutually shared, superordinate goal” (Sonnenwald, 2007, p. 645). Hansen and Järvelin’s (2005) “broad and preliminary definition” for collaboration (p. 1102)—synthesized from the information retrieval, information seeking, and CSCW literature—was similar; they required information access, a specific problem or task at hand, and human beings as direct or indirect information sources. They stressed flexibility in the information seeking environment and the specifics of the source. This definition, much like Mills’s (2010), is still restricted to collaboration taking place around a specific problem or task. Sonnenwald’s (2007) definition—if “scientists” are replaced by “people”—lacks the information access requirement of Hansen and Järvelin (2005) but is otherwise compatible.

All four authors appear to consider serendipitous information sharing—e.g. happening to come across a research article on a topic which a colleague is researching, choosing to send the link to them—to be “less” than collaboration. Serendipity is common as individual information behavior (see Case, 2012, p. 101; Erdelez, 2005; Foster & Ford, 2003) and “an integral part of the creative process” across fields (p. 321); it is accounted for in Talja’s (2002) social information sharing category. There is no obvious reason why it would not occur or should be ignored in the context of collaborative information behavior, if a broad view of the latter is adopted.

Talja’s (2002) separation of serendipity into a separate category implies collaboration may differ by type, level, or degree, an idea present in Gunawardena, Weber, and Agosto’s (2010) classification of coordination, cooperation, and collaboration as separate activities. Drawing on Maienschein and Mattesich’s work, they argued groups who do not choose their goal are engaged not in collaboration but in cooperation. All three concepts included information exchange, activities of mutual benefit, and a common purpose; coordination did not include sharing of resources, and only collaboration included “enhancing the capacity of another” (Gunawardena et al., 2010, p. 212). When moving from cooperation to collaboration, relationships became more integrated; risk increased; new structures were created; commitment to common goals increased; planning was more comprehensive; and resources, responsibilities, and authority were shared and pooled more often. Gunawardena et al. stated such distinctions were not often made in studies, leading to different definitions in different contexts. In trying to build such a consensus, they adapted other definitions and defined collaboration as “human behavior that makes a substantial contribution toward the advancement of a research project … with respect to a mutually shared superordinate research goal and which takes place in a research setting” (Gunawardena et al., 2010, pp. 213–214). This definition again requires a common goal, task, or activity to be completed through collaboration; while focused on the research setting, it provides a useful summary of the broad consensus of Hansen, Järvelin, Mills, and Sonnenwald. There is still disagreement on if common access to information sources is required and if serendipity is true collaboration or a form of coordination.

2.4. Social Digital Libraries

2.4.1. The Social Paradigm

Having explored communities and collaboration as important social phenomena, a discussion of considering digital libraries in social context—a third view alongside the two conceptions Borgman (1999) found—can now take place. While Smith (1981, 1991) discussed critiques of Bush’s (1945) introduction of memex as focusing too much on the technical and engineering challenges of the information problem and not enough on the human side, Bush (1945) was an early believer in the social paradigm, as it might be called today (cf. Raber, 2003, pp. 201–223). He did not consider the scholar’s memex to be a single-user system, used by one individual for simple storage and retrieval of information (although it was capable of this). Instead, he intended the information stored in a memex to be socially exchanged, constructed, and discussed by and with other scholars and scientists within and beyond the scholar’s social network.

The social paradigm sees information as a broad phenomenon having “social significance” (N. Roberts, 1976, p. 249) which “must be examined in the context of its social nature” (Raber, 2003, p. 222). Information is subjective within this context, differing in meaning and interpretation between different individuals, groups, communities, organizations, cultures, or societies (Jaeger & Burnett, 2010; Talja et al., 2005). Data, information and knowledge may be within information systems, outside these systems, in the heads of users, or be socially constructed by groups and communities (Tuominen & Savolainen, 1997). Research examines the “production of knowledge” (Karamuftuoglu, 1998, p. 1071), where “information acquiring and processing capabilities are not individual but social manifestations” (Brookes, 1975, as cited in N. Roberts, 1976, p. 254). Users are active participants in the social construction and production of information and knowledge and may take on multiple roles in the resulting discourse (Talja et al., 2005). Early calls for a social paradigm (N. Roberts, 1976; Rosenberg, 1974; T. D. Wilson, 1981), combined with the recent research and advocacy of Chatman (2000), Kling (1999; see also Horton, Davenport, & Wood-Harper, 2005), Savolainen (1995; Tuominen & Savolainen, 1997), Sawyer (Sawyer & Eschenfelder, 2002; Sawyer & Tapia, 2007), and others have led to broad—but by no means exclusive—adoption of the social paradigm among information science researchers.

Despite this expressed need—as far back as 1945—for social contexts to be considered under a social paradigm, many early information retrieval systems focused on the technology, as is evident in portions of Smith’s (1981, 1991) analysis and in Raber’s (2003) discussion of the key problems of information science. Echoes of paradigmatic unrest are seen in the divisions on how digital libraries should be seen (see above and Borgman, 1999) and Brown and Duguid’s (2002) rejection of technology-centric solutions to information and knowledge problems.

2.4.2. Calls for Social Digital Libraries

Nevertheless, many have repeated and restated the call for consideration of digital libraries-as-information-systems in social context. As early as the first academic conference on digital libraries, Ackerman (1994, p. 198) argued it is “unwise … [and] unnecessary … [for] social exchange and interaction” to be ignored by digital libraries. Levy and Marshall stated digital libraries should be considered social as early as 1995, feeling conceptualizing their use as solely individual—in alignment with early information seeking research (see Case, 2012; Ellis, 1992; Raber, 2003)—was a mistake (Levy & Marshall, 1995). Digital libraries, they believed, should instead be modeled after traditional libraries, serving as “meeting places where joint research is carried out” in “a highly collaborative” way, supporting their user communities, and facilitating the communication and collaboration taking place between users (p. 80).

Neuhold, Neiderée, and Stewart (2003, p. 1) stated digital libraries must support “community efforts to capture, structure, and share knowledge,” an approach similar to that taken by KM and CoPs but compatible with the views of Bearman, Levy, Marshall, and others. Marchionini, Plaisant, and Komlodi (2003, p. 121) believed digital libraries “are embedded in many different communities … [and] contexts” which are “inescapable,” a view stressing these contexts and communities must be considered throughout the lifecycle and study of a digital library. Van House (2003, p. 272) argued digital libraries must support collaboration: “cognitive or knowledge work” that is “situated, distributed, and social.” Her thinking has strong echoes of Levy and Marshall and the placing of digital libraries—and the information behavior of their users—within a situated social context.

Marshall wrote a further paper on sharing encountered information via digital libraries with Bly (Marshall & Bly, 2004), wherein they concluded digital libraries must allow for the finding and sharing of information that strengthens communities and social ties, placing emphasis on supporting tie-strengthening collaborations. Frumkin (2004) argued digital libraries are “content, provided through digital services” (p. 155), but had to provide traditional “value-added information experience[s]” desired by patrons and users, be they technical, individual, or social (p. 156). Lynch (2005, para. 21), in looking at the future of digital libraries after their first ten years of funded research, maintained they must be “connect[ed] and integrat[ed] … with broader individual, group, and societal activities” and must support collaboration in these contexts. Gazan (2008, Introduction section, para. 2) argued content in digital libraries is a natural part of “an ongoing conversation among a community” and digital libraries should not ignore this conversation; ignoring the contexts, community, and collaboration would be a grave mistake.

These calls relate to all of the conceptions of community discussed above: Fischer’s subcultural theory (all, but especially Neuhold et al., 2003), communities of practice (Gazan, 2008; Levy & Marshall, 1995; Marchionini et al., 2003; Van House, 2003), virtual communities (Ackerman, 1994; Gazan, 2008), arenas and social worlds (Levy & Marshall, 1995; Lynch, 2005; Neuhold et al., 2003; Van House, 2003), social network analysis (Marshall & Bly, 2004), and information worlds (Lynch, 2005; Neuhold et al., 2003). Most authors fall under the consensus view of collaboration summarized by Gunawardena et al. (2010), with Marshall and Bly (2004) and Lynch (2005) allowing the most for flexibility and serendipity. Of these calls, many show a belief that digital libraries, the information they contain, and the communities that use them are socially constructed; Levy and Marshall (1995), Neuhold et al. (2003), Van House (2003), Lynch (2005), and Gazan (2008) make the strongest arguments for such a view.

2.4.3. Defining a Social Digital Library

A social digital library, then, is of clear importance as a goal for all digital libraries. Adapting Borgman’s (1999) definitions, it can be defined as (a) having one or more collections of digital content collected on behalf of a user community; and (b) offering services, relating to the content, by or through the digital library to the user community. It is, or is part of, one or more formal or informal organizations that manage these content and services, focusing on facilitating information and knowledge creation and sharing (after Lankes, 2009, 2011) and excluding different primary motivations ( e.g. selling products). These characteristics should be considered in light of the various contexts they inhabit, most of all the social contexts; the social digital library is socially constructed (cf. Tuominen & Savolainen, 1997).

This definition and conception of a social digital library parallels the roles of physical libraries (Pomerantz & Marchionini, 2007, p. 506), which are not just physical collections and technical services but physical and conceptual spaces “link[ing] people to ideas and to each other.” It further parallels the definition found in the DELOS Reference Model and associated Digital Library Manifesto (see Candela et al., 2007, sec. 2, para. 3), which included (a) an organization; (b) the collection and management of digital content; and (c) functionality and services associated with the content. The DELOS definition includes additional elements of quality, policy, and preservation; these are important parts of many, but not all digital libraries, and are not made explicit in the definition considered here (but would fall under services).

Collaboration and community are major factors in this definition. A user community may consist of smaller communities or groups, adopting the subcultural view pioneered by Fischer (1975) and incorporating flexible use of one or more of the conceptions discussed above and used in calls for social digital libraries. The content collected by the digital library is intended to be used by these communities; the services it offers are for the communities; and the organization(s) it is associated with are a form of community. A major objective of digital libraries is to support and help construct these differing kinds of “knowledge communities” that use their content and services (Bearman, 2007, p. 245). Digital libraries can and should improve and build these communities by supporting their “internal workings … and their links to the rest of the world” (Agre, 2003, p. 227), becoming tightly bound to these communities (Star et al., 2003). Supporting these workings and links requires supporting social contexts: collaboration within and across communities, the building and construction of new and existing communities, and individual and collaborative information behavior. Such collaboration centers around a common overall project, goal, interest, or practice (Gunawardena et al., 2010), although serendipitous opportunities for collaboration within and between communities should not be dismissed, given the commonality and value of serendipity as information seeking behavior (see Case, 2012, p. 101; Erdelez, 2005; Foster & Ford, 2003; Talja, 2002) and its importance “for its role in connection building, discovery and creativity” (Foster & Ford, 2003, p. 323) by and between individuals and groups (see also Marshall & Bly, 2004).

2.5. Previous Social Digital Library Research

Once the why question has been answered, the next big question is, of course, how can we support the social contexts in and around digital libraries? Many approaches, perspectives, models, and theories have been applied to studying and supporting the communities served by digital libraries and the collaborations their members engage in, with varying and mixed degrees of success. These can be placed into two groups: earlier, experimental efforts; and perspectives with more long-term success and promise.

2.5.1. Experimental

Many experimental models and perspectives proposed by researchers showed great promise at first, but have not been as successful in practice over time. Despite this, these projects and models can help better inform the conceptualization, design, and development of social digital libraries and their support for collaboration and community-building behaviors.

2.5.1.1. Sharium

Marchionini’s (1999, p. 1) sharium model for digital libraries stressed collaboration and sharing within communities and networks “to facilitate communication and distribute the load of solving information problems.” Proposed features included (a) experts sharing knowledge and time in digital reference, question-answering, and recommendation services; (b) easy contribution and sharing of digital content by the community; and (c) better support of collaborative, self-directed learning. Marchionini applied the model within three digital library projects, but none were unqualified successes: the American Front Porch (AFP) project (Sonnenwald et al., 1999) was too ambitious and never implemented many collaboration services; the Baltimore Learning Community (BLC) (Marchionini et al., 2003) faced low adoption of social features and technical difficulties by its user base of middle-school teachers; and the Open Video Digital Library (OVDL; Marchionini, Wildemuth, & Geisler, 2006) had many of its sharium features stripped, e.g. videos to be added were accepted only as part of already existing university or government collection, not from end-users of the digital library. While the OVDL is still in existence, its current support for collaboration and interaction appears minimal at best, and no contributions are being accepted (Open Video Project, n.d.). The sharium model provided an excellent theoretical view of a social digital library—at least at first glance—but not one achieved in practice. Too strong a focus on its ideals and on the technology required to support them (cf. Brown & Duguid, 2002), at the expense of the theory and concepts behind community—Marchionini (1999) included no citations of any major conceptions or theories of such—appear to have been major contributions to its low level of success.

2.5.1.2. CKESS

The CKESS model and project proposed by Bieber et al. (2002) drew on research into knowledge management, CoPs, and virtual communities. To say it was ambitious would be an understatement; it included communication, concept and process mapping, hypermedia, and decision analysis support among its many planned features. This ambition, combined with a high degree of specificity and a lack of realistic consideration of context—expert knowledge of a modeling language and semantic concept maps was expected of users—restricted its ability to be adopted by other projects and stunted its own progress beyond Beiber et al.’s thought exercise. Its publication in a management information systems (MIS) journal and not a common source of digital library literature may have contributed (K. Burnett, personal communication, September 16, 2009). CKESS’s lack of success may not have been the fault of the conceptions of community and collaboration applied, but greater consideration of their context was necessary. Bieber et al. may not have understood the limitations of the conceptions they drew on, or felt such limitations could be worked around through sheer ambition and use of technology.

2.5.1.3. CYCLADES

Unlike CKESS, the CYCLADES project (Candela & Straccia, 2003; Renda & Straccia, 2005) did make it to the prototype stage in its development of a folder-sharing, community-based digital library for supporting collaboration. Collaboration was supported via folder-sharing features and “discussion forums and mutual awareness” (Candela & Straccia, 2003, p. 159). Their conception of community was as “a set of users sharing a common (scientific, professional) background or view of the world” (Renda & Straccia, 2005, p. 9), perhaps most akin to social worlds but also compatible with CoPs and information worlds. The prototype is no longer online nor is a final system; its defunct web site (CYCLADES, n.d.) implies it ran out of funding and could not secure more. As with CKESS the conceptions used may not have been to blame, but greater understanding of them and less emphasis on technology may have led to greater success.

2.5.1.4. Alexander

The Alexander project undertaken by Kolbitsch, Maurer, and their colleagues (Kolbitsch & Maurer, 2006a, 2006b; Kolbitsch, Safran, & Maurer, 2007) aimed to build “a community around an encyclopaedic body of knowledge” (Kolbitsch & Maurer, 2006b, p. 185) combined “with contemporary news articles” (Kolbitsch, Safran, & Maurer, 2007, Prototype Implementation section, para. 1). As shown by Kolbitsch and Maurer’s (2006a) review, they were aware of multiple conceptions of communities, but focused on virtual communities—emphasizing blogs and wikis—and social networks. Alexander’s proposed feature set was almost as ambitious as CKESS’s. A small subset of these made it into a prototype, which—as with CYCLADES—is no longer online and the project concluded in 2007 (Institute for Information Systems and Computer Media, n.d.) without any further publications as planned per Kolbitsch, Safran, and Maurer (2007). Ambition, a lack of funding, and a narrow tunnel-like focus on technology appear to have doomed Alexander.

2.5.1.5. 5S

Fox first proposed his 5S model for digital libraries in 1999 (Fox, 1999). The five S’s are streams, structures, spaces, scenarios, and societies, corresponding to content, organization, retrieval and interface, services, and community (Fox & Urs, 2002, p. 519; Fox & Gonçalves, 2009). Early use of the 5S model applied it to existing digital libraries, as seen in Fox (1999). The societies component, of most interest here, took the form of an informal analysis of stakeholder communities and their role within and around the digital libraries in question. Gonçalves, Fox, Watson, and Kipp (2004) formalized the model in mathematical, set-theoretic terms, defining a society as “a set of entities”—human and technical—“and the relationships between them” (p. 275). Unfortunately, they abstracted communities away from “social communities” and towards “digital communities … instantiated by the adoption of a particular architecture and interface” (p. 279). Although Gonçalves et al. included content, services, and communities in their conception of a digital library, they were focused on technical and information retrieval contexts. Recent work has continued this focus while developing a framework for evaluating the quality of a digital library (see e.g. Moreira, Gonçalves, Laender, & Fox, 2009). 5S as it stands is inappropriate for modeling social, community-based digital libraries under the conceptions of community reviewed above. Bearman’s (2007) review places 5S within a section on technology, and not under services or social impacts; he considers societies to be representative of “clients” and not communities (p. 236). If a more appropriate conception of community was substituted for that of Fox, Gonçalves, and their colleagues, the 5S model could have potential for research on and practice in social, community-based digital libraries. Substantial further work would be necessary to test this potential applicability.

Brief mention should be made of a related framework for digital libraries that may have more potential, proposed by Fox and other colleagues at Virginia Tech and elsewhere. Akbar et al. (2011) have labeled this “DL 2.0”; it is based on testing of the Ensemble educational digital library portal, a National Science Digital Library (NSDL) project (see also Brusilovsky et al., 2010). This framework consists of resources, services, and users, with the model supporting services that provide relations between resources and other resources (tags, links, ratings), resources and users (ownership / authorship, reading / downloading, contribution of comments or ratings), and users and other users (membership, contact / tie formation, co-authorship). The framework is couched in language that could—if elaborated on, developed further, and tested in one or more digital libraries—lead to successful social digital libraries; at this point, limited to the single article of Akbar et al. (2011), full assessment of its promise is not possible.

2.5.2. Successful, Promising

Other approaches have been more successful and/or have appeared quite promising, including in contexts outside the usual scope of digital libraries.

2.5.2.1. Wikis

Wikis seem a natural fit for supporting collaboration around digital libraries, given their nature as social and collaborative constructions. They have been discussed in the context of communities formed around bodies of knowledge (Kolbitsch & Maurer, 2006a, 2006b), collaborative learning within these (e.g. Chu, 2008), and how these communities organize themselves and their practices (e.g. Stvilia, Twidale, Smith, & Gasser, 2008), but there is little known literature directly applying wikis to the design and development of digital libraries. Krowne (2003) is an exception, developing a successful digital library called PlanetMath (planetmath.org) using a wiki-like approach he called “commons-based peer production” (Introduction section, para. 1). This approach defines the community as having the goal of creating an “intellectual work” (Introduction section, para. 1) and collaborating to do so; this matches Gunawardena et al.’s (2010) summary of the consensus on collaboration, while restricting the concept of community to one with a purpose, most similar to CoPs and some views of virtual communities. Wikis may prove useful for social digital libraries, but further research and application is necessary to see if they are applicable under other conceptions, models, or theories of community.

2.5.2.2. Social annotations

Social annotations—“enrichment[s] of information object[s] with comments and other forms of meta-information” (Neuhold et al., 2003, p. 10) that are shared with the public and can be annotated or “enriched” themselves by other users (p. 11)—have been used with success to support social digital libraries and community-based collaboration. While related to social tagging, social annotations include other user-contributed additions to digital content, allowing digital library users and user communities to “take a more active part” in socially constructing the digital library and providing “a valuable medium for collaboration” within and beyond these communities (p. 11). Social annotations have been used to a degree of success in the DEBORA (Nichols et al., 2000) and COLLATE (Frommholz et al., 2003) projects, albeit they did not make it past the prototype stage. The Digital Library for Earth Science Education (DLESE; dlese.org) and the Multimedia Educational Resource for Learning and Online Teaching (MERLOT; merlot.org) are more successful examples of social annotations as evaluative metadata in production digital libraries (see also You, 2010), albeit DLESE faced limitations and usability issues (Arko, Ginger, Kastens, & Weatherley, 2006). Brusilovsky et al. (2010) have included social annotations as part of personalization techniques in the Ensemble educational digital library portal, leading to what they called “social navigation,” facilitating users’ use of Ensemble and adding an element of social construction to its interface (p. 2890). García-Crespo, Gómez-Berbís, Colomo-Palacios, and García-Sánchez (2011) developed a digital library prototype, CallimachusDL, that incorporated social tagging and folksonomies within a semantics-based design. Their proof-of-concept was not explicit in addressing many social phenomena such as communities or collaboration, but tested well in use by final-year undergraduates in computer science. Ensemble and CallimachusDL, while promising, require further testing and may need additional adaptation to prove their potential as applications of social annotations as one element of a social digital library design.

Social annotations have been employed with success in contexts similar to digital libraries. These include Web 2.0 social question-and-answer site AnswerBag (answerbag.com), which faced many of the same community-building challenges digital libraries face (Gazan, 2008); it has been successful, with over a million users in late 2009 (Answerbag, 2009) and more than a million unique visitors in April 2014 per Compete, Inc. (2014). While Pinterest does not share all the characteristics of a social digital library, it uses many of the same type of social annotations (Zarro & Hall, 2012), and might provide some guidance for social digital libraries wishing to incorporate social tagging, linking, and sharing features. Fringe, a prototype of a folksonomic contact manager which expanded social annotations and tagging from documents to people, was studied by Farrell, Lau, and Nusser (2009); they found tagging and folksonomies provided incentive for community-building and concluded further study of this form of social annotations in this and other contexts would be useful. The Steve project (www.steve.museum) is intended to explore “the role of social tagging” and to study “the resulting folksonomy” around digital art museums (Trant, 2006, p. 1), which have many similarities with digital libraries. Trant’s study of a preliminary tagging prototype concluded “social tagging seems a promising way to” improve access (p. 22), and the prototype appeared successful at building community through increasing engagement and motivating contributions (see also Bearman & Trant, 2005). A production version of Steve has been used in experimental studies of image tagging behavior (Landbeck, 2013; Stvilia, Jörgensen, & Wu, 2012). Steve, Pinterest, Fringe, and AnswerBag provide good examples of the potential of social annotations for supporting emergent community and socially constructive collaboration around and within social digital libraries.

2.5.2.3. Social constructionism

While the ideas inherent in social constructionism are present in the work of others, Tuominen, Talja, and Savolainen (2003) directly applied social constructionism to the concept of social digital libraries, focusing on “discourse” and multiple expressed perspectives (p. 564; see also Talja, Tuominen, & Savolainen, 2005; Tuominen & Savolainen, 1997). This specific approach to social constructionism is most compatible with information worlds; Shibutani’s (1955) version of social worlds and/or reference groups is quite similar. Situated learning-centric views of CoPs—but not as much later management-centric conceptions—feature elements of social constructionism. Social constructionism has been applied to a digital library prototype, ScholOnto (pp. 565–567), which now appears moribund (Knowledge Media Institute, 2004) much like CYCLADES and Alexander. The CallimachusDL digital library prototype of García-Crespo et al. (2011), through its incorporation of semantics, could be considered to include social constructionism elements, although this is not claimed by the authors. Alemu, Stevens, and Ross (2012) argued for a social constructivist approach to improve metadata interoperability across digital libraries through social annotations, an approach similar but not identical to social constructionism (see also Talja et al., 2005).

Social networking sites have focused on the centrality of social discourse to users’ information behavior, although each has chosen a different set of features and different elements of social constructionism are present in each. Facebook (facebook.com) and Twitter (twitter.com) are the best-known examples, but Google Wave (wave.google.com) and FriendFeed (friendfeed.com)—neither in current, active development—in many ways took the idea further. Elements of social constructionism are present in social blogging and pinning services, such as Tumblr (tumblr.com) and Pinterest (pinterest.com; see also Zarro & Hall, 2012). These services share in their overall flexibility, implying flexible and broad conceptions of communities and collaboration would represent them best. Given mixed-to-good levels of success in the Web 2.0 world, it would be useful to see further exploration of how the ideas inherent in social constructionism could—through appropriate theories and conceptual constructs—be applied to digital libraries.

2.5.2.4. Social network approach

Two measures from social network analysis (discussed above) could be used as measures of community-building and collaboration activity and of a successful social digital library. These are range—the size and heterogeneity of the network—and density—how many relations and ties occur compared to the theoretical maximum number possible. There may be community-building and collaboration activities not caused by the digital library or its content that confound some measures of range and density; these would have to be controlled for. Social network analysis is—as discussed above—somewhat limited by its focus on social ties. While others have suggested the social network approach can and should be applied to all communities (see e.g. Haythornthwaite, 2007; Wellman, 1982) and digital libraries’ communities (Farooq, Ganoe, Carroll, & Giles, 2009; Neuhold et al., 2003; Star et al., 2003), there are no known studies—besides a pilot test by the present author (Worrall, 2010)—directly applying social network analysis or the underlying concepts of range and density to examine the social networks of digital library users or community-building and collaboration activities by members of such networks. Further exploration of the potential of social network analysis in studying the communities of social digital libraries and community-building and collaboration behaviors is needed.

2.5.2.5. Situated context

Bishop’s framework of situated context (Bishop, 1999; Bishop et al., 2000) proposed to examine how users use a digital library in the context of individual information uses and the social interactions taking place between individuals and others in the community. She and her colleagues focused on virtual communities supporting real, physical communities, which became sub-communities (cf. Fischer, 1975) within the broader geographic location. They studied two early digital libraries, DeLIver and Prairienet, and found consideration of the context of users’ individual and group information behaviors on multiple levels was lacking, with minor usability flaws and other minor issues becoming “insurmountable molehills” (Bishop, 1999, p. 97) to successful individual and group use of the digital libraries and to the building of community around them. Bishop et al.’s view of collaboration was similar to the consensus, and a focus of much of their study was on the deliberate (i.e. goal-driven) information seeking behavior of their users.

2.5.2.6. Boundary objects

Star’s boundary object theory (Star, 1989; Star & Griesemer, 1989) is discussed at significant length in section 2.7 below. Based around Strauss’s (1978) social worlds and first used in science and technology studies, it has been applied to and extended in studying social digital libraries by Star, Bowker, and Neumann (2003) and Van House (2003), with similar application proposed by Fleischmann (2007a, 2007b). In the context of CoPs and Chatman’s view of information worlds (Burnett et al., 2001; Chatman, 1992, 1996), Star et al. found digital libraries could sometimes be successful at integrating existing overlapping communities and information worlds and building new communities, with the potential to support collaboration across boundaries. The translation between these different communities is difficult; it requires digital libraries to “fit with … [existing] practices” across multiple existing communities while supporting emergent work processes and the formation of new communities (Van House, 2003, p. 290).

2.5.3. Potential Implications for Social Digital Libraries

No one approach, method, theory, or model has already been determined to be the way to support collaboration and community-building in and around digital libraries. Social annotations are promising; wikis might prove useful, given further study of their advantages and disadvantages. Social constructionism theory, situated context, boundary object theory, and social network analysis show significant promise as approaches to studying and improving social digital libraries. The most common theme in the literature is the need to study the cognitive, institutional, cultural, organizational, and—above all—social contexts of digital libraries, to gain a complete picture of their roles in communities of and collaboration between users. Communities, networks, and collaboration within and between them cannot be supported in a digital library that ignores the socially constructed conversation it is a natural part of (Gazan, 2008). Studies and methods applying well-grounded, context-sensitive, and flexible theories and conceptions of communities and collaboration provide the most insightful findings and the highest chances of success.

2.6. Virtual Book Clubs

Before considering appropriate theoretical grounding for a study of social digital libraries, another area of research requires brief discussion. As explained later in Chapter 3, this dissertation study focuses on LibraryThing and Goodreads, two digital libraries and online communities for lovers of books and of reading, and on nine groups from the two sites. These groups can be conceived of as virtual book clubs, and so a short review of the literature in this area provides additional appropriate context. Most of this literature has focused on using virtual book clubs in K–12 or public library-based education and reading programs (e.g. Scharber, 2009; Scharber, Melrose, & Wurl, 2009) or on the practical implementation of virtual book clubs by libraries ( e.g. AuYeung, Dalton, & Gornall, 2007). These topics are not the focus of this dissertation, but a few other selections from the literature have considered users’ participation, interaction, or information behavior in the context of a virtual book club as community, as is of interest here.

Rehberg Sedo’s (2003) survey of 251 international readers and members of face-to-face and virtual book clubs, focused on reading practices, found that members participated in virtual book clubs to connect with others like them, for intellectual stimulation, for fun, to read books they would not read on a regular basis, and to talk with others about the books they were reading. The topic or choice of books the virtual book club was reading was a factor for a plurality of members. Discussion of the meaning and interpretation of books occurred for about half of the members, with many of these discussions bringing in personal opinions, feelings, and values. Differences in these, and the opportunity presented to learn new ideas and share new experiences, were valued by a majority of the club members in Rehberg Sedo’s study. She called for further research on virtual book clubs using mixed methods and to explore community phenomena (e.g. power hierarchies, identity and practice-making) in this context.

Rehberg Sedo edited a later book (Rehberg Sedo, 2011b) which includes, as part of the introduction, a brief review of additional studies that pay “critical attention to the social aspects of reading and [focus] on the discussions that take place within groups, their reading lists, and the contexts in which the reading takes place” (p. 10). Many of these are of face-to-face book clubs, including most of the historical and present-day studies reported in the book. Nevertheless, when taken as a whole they provide a clear view of how this subset of book club research sees the concept of community: it “take[s] many forms and serve[s] many purposes,” it can provide “emotional gratification … [and] a sense of belonging,” it can help one feel “part of something larger,” it can fulfill a “need for emotional connections with other people,” and it “is constructed and maintained socially” (p. 11). These are similar to the characteristics associated with online communities (see section 2.2 above).

Two chapters of the Rehberg Sedo (2011b) edited book examined online book discussions. Kiernan (2011) focused on the positive role of television media in encouraging reading, the impact of new readers and book club members on the culture of reading, the view of publishers and reviewers in this context, and the influence of book clubs associated with television media. Rehberg Sedo’s (2011a) own study (conducted in 2002 and 2003) of a book club dedicated to young adult (YA) literature, where the members served as “cultural intermediaries … between the online [YA] literature reading community and the off-line communities in which young adults live” (p. 102), explored issues of agency, power, norms, rules, and authority uncovered through the club’s discursive practices, simultaneously “oppress[ing] and giv[ing] voice to individual readers” (p. 102) and “establish[ing] and reinforc[ing] cultural authority” for the intermediaries (p. 118). The dynamics of norms, rules, values, and practices present and socially constructed within the book club were impacted by, and impacted on, the broader communities and society that intermediaries were part of. This finding is in clear agreement with the scholarship established around social informatics, social constructionism, the social worlds perspective, and the theory of information worlds (the latter two mentioned in section 2.2 and discussed in greater depth below).

Fister’s (2005) article provides a deep description of one e-mail list-based “reading list” and virtual book club, focusing on those features that made it an online community. Reviewing her article, one sees that participants shared humor and in-jokes, discussed their everyday life often, had established many social ties with others, felt solidarity in their reading “addiction,” and valued the group for providing emotional support, all common characteristics of strong, established online communities. Fister described the practices by which the group read books together, led by volunteer “question maestros” who would lead and moderate discussions and provide additional material and guidance as necessary and desired (p. 306). Many of the characteristics Fister identified in this group are in common with those Rehberg Sedo found in her study and discussed in the introduction to the book she edited.

Elsayed (2010) surveyed the moderators of seven online Arab book clubs, augmenting this data with observation of the book clubs’ online spaces. His study found low participation rates and discussions of mixed depth, many being superficial at best; this proves not all virtual book clubs are strong online communities with many social ties like those identified by Fister and Rehberg Sedo. Elsayed nevertheless concluded online book clubs were “a promising environment for promoting reading” and for motivating the contribution and exchange of ideas, information, and knowledge (p. 246). Elsayed explored the processes at work in the book clubs for choosing which books are of interest or should be read as a group, with some variety in these appearing across the groups but member voting being most common; and for discussions of books once chosen and read, which in some groups was moderator-led and in others was more open.

While Foasberg (2012) focused her study of reading challenges and the social dynamics and processes behind them on book blogs instead of book clubs, it shares many similarities with the work of Rehberg Sedo, Fister, and Elsayed. Foasberg observed and described three specific case studies of book challenges in depth. She found such challenges allowed virtual book club members to discuss what they were reading and form bonds and social ties with readers who shared interests, values, or beliefs. Many challenges, while blog-based, linked to LibraryThing, Goodreads, or other social media venues (e.g. Twitter) as an additional venue for discussion. She stated that generalizing “about the communities formed by reading challenges is difficult” (p. 40), but that common elements existed: different levels of participation, updates and discussion via social media, and—for about a fourth of the challenges—a form of prize or reward for completing the challenge. Foasberg believed the communities that emerged were quite different from those formed from social network-based sites (which LibraryThing and Goodreads can be considered): “the challenge [did] not form a boundary to the community,” since blogs were open to others who were not participating in a given challenge and allowed for a broad, “content-focused” community to develop around them (p. 50). Foasberg suggested further ethnographic research on reading challenges, with emphasis on those taking place via book blogs.

Research by Greene (2012) explored how youths’ identities were constructed in a virtual book club on Facebook. While not examining a broad range of information behavior, Greene was explicit in positioning virtual book clubs as online communities and considered the influence of a sense of community on identity construction. She found emotional safety was present through emoticon and acronym use as normative behavior; users took on roles, such as a caregiver expressing a mothering stance as part of discussions. Greene further examined the roles played by facilitation and gender perspective in constructing identity, but these parts of her study are of less direct relevance to this review.

In sum, the existing virtual book club literature has provided thick description of cases of such groups and shown the emergence of general characteristics of users’ participation, interaction, and information behavior, and of the communities that have been formed around and within such groups. None of the known literature has placed such a study in the context of well-grounded theory (with the exception of Greene’s work, not yet published in journal article or book form). Except for some hints in Foasberg’s study of book blogs, none of the literature has examined the roles played by a given venue or technology (e-mail list, social network, blog, etc.) in the communities and information behaviors of members of a virtual book club, or the roles that might be played by a social digital library in such a group.

2.7. Boundary Object Theory

Boundary object theory, first developed in science and technology studies (STS) by Susan Leigh Star, can be a strong element of a well-grounded, context-sensitive, flexible theoretical framework for studying social digital libraries. It has been applied to and extended within many disciplines, including library and information science (LIS), to study the interactions taking place and the objects people create and use in the context of crossing the boundaries of different social worlds and communities. Within LIS and cognate fields, it has been applied to studying the information behavior of users, communities, and organizations; information systems that serve multiple social and information worlds; and interrelations and interactions between these two elements. This section reviews boundary object theory as developed by Star and Griesemer (1989; see also Star, 1989), how Star and others have developed it over time, and how researchers have applied it within LIS and related fields. It analyzes and evaluates the key concepts and propositions of the theory and its resulting strengths and weaknesses.

Star’s PhD education was in ethnographic sociology under Anselm Strauss at the University of California, San Francisco, earning her degree in 1983 (Zachry, 2008, p. 439). She had become interested in artificial intelligence and computer science after working with Carl Hewitt of the Massachusetts Institute of Technology (MIT) (p. 438). At the time of her development of boundary object theory, Star was an assistant professor of information and computer science at the University of California-Irvine (Star & Griesemer, 1989), working with Rob Kling and others in STS and the area since labeled “social informatics” (see Kling, 1999). In her career, she worked as faculty and conducted research in sociology, LIS, STS, social informatics, computer-supported cooperative work (CSCW), communication, and feminist studies before her unfortunate passing in 2010 (Clarke, 2010).

2.7.1. Core Concepts

There are four core concepts used in boundary object theory: social worlds, translation, boundary objects, and coherence / convergence (given separate names but the same concept). All were derived from the work of others, but the degrees of derivation and modification by Star and Griesemer (1989) vary.

2.7.1.1. Social worlds

Strauss (1978) developed his concept of social worlds within the symbolic interactionist school of sociology originating at the University of Chicago (see also Clarke & Star, 2008). This concept and the framing concepts within it were discussed in section 2.2.4 above. It was an abstract concept, but one concrete enough to be operationalized and reliably applied in research. The framing concepts of activities, sites, technologies, and organizations provide limitations, but Strauss intended for broad and flexible interpretation of these. Star and Griesemer (1989) used Strauss’s concept as-is; the remainder of the theory also fell within the social worlds framework (Clarke & Star, 2008).

2.7.1.2. Interessement and translation

Boundary object theory draws on the derived concepts of interessement and translation, the latter adapted for use in the theory. Interessement was first developed by Latour, Callon, and Law from multiple case studies of the sociology of scientific practice, as part of actor-network theory (ANT). Star and Griesemer (1989) defined it as follows:

[T]o create scientific authority, entrepreneurs gradually enlist participants (or in Latour’s word, ‘allies’) from a range of locations, re-interpret their concerns to fit their own programmatic goals and then establish themselves as gatekeepers (in Law’s terms, as ‘obligatory points of passage’) [(Callon & Law, 1982; Law, 1987; both as cited in Star & Griesemer, 1989, p. 389)] …. Latour and Callon have called this process interessement, to indicate the translation of the concerns of the non-scientist into those of the scientist. (p. 389)

Translation—also derived from the work of Latour, Callon, and Law on ANT—was defined in the context of multiple social worlds as “the task of reconciling [the] meanings” of objects, methods, and concepts across these worlds (Star & Griesemer, 1989, p. 388) so people can “work together” (p. 389), drawing on Strauss’s social worlds.

Translation and interessement as defined by Latour and his colleagues were limited to the study of scientists. Star and Griesemer (1989) felt an ecological approach to analysis within the social worlds framework (see Clarke & Star, 2008; Strauss, 1978) considering all possible viewpoints was a better approach, one which had no limitations on which individuals and social worlds could be studied. They extended the concepts to allow for multiple translations, gatekeepers, or “passage points” to exist between different social worlds (p. 390), corresponding to and going beyond Latour and Callon’s conception of interessement.

2.7.1.3. Boundary objects

The conception of boundary objects themselves, while a unique element of Star and Griesemer’s theory, was still derived from the work of Strauss on social worlds and Latour, Callon, and Law on interessement. Boundary objects were theorized to exist where social worlds intersected with each other, requiring a translation process or interessement to occur. As such, they could be considered an extension of the “passage points” of Law, although they do not serve an identical role (see Star & Griesemer, 1989, p. 390).

Boundary objects were defined as objects crossing the boundaries between multiple social worlds, used within and adapted to many of them “simultaneously” (Star & Griesemer, 1989, p. 408) and “‘sit[ting] in the middle’ of a group of actors with divergent viewpoints” (Star, 1989, p. 46). They “adapt to local needs” within a social world but are “robust enough to maintain a common identity across sites” (p. 46). The boundaries of boundary objects themselves may vary in permeability and fixedness. Boundary objects can be either abstract, concrete, both simultaneously, or somewhere along a continuum between these extremes (Star & Griesemer, 1989, p. 408). For example, Star and Griesemer’s original case study identified maps of life zones in California as boundary objects. These were concrete for biologists because they were familiar with the ecological concepts they portrayed, but “highly abstract” for the public and professionals from other worlds who were not as familiar with the concept of life zones (p. 411). This conception was developed inductively based on the boundary objects identified in Star and Griesemer’s (1989) case study and on logical inferences from the work of Latour, Callon, Law, Strauss, and others. While the boundaries of the concept were at first restricted to studies of scientists—much like interessement—further applications and extensions of the theory have proved its generalizability to other settings.

2.7.1.4. Coherence and convergence

Another concept used in boundary object theory is coherence, derived from the concept of intersecting social worlds (Strauss) and the results of translation (Latour and colleagues). Star and Griesemer (1989, p. 390) stated the “coherence of sets of translations depends on the extent to which entrepreneurial efforts from multiple worlds can coexist,” with “an indeterminate number of coherent sets of translations” possible. While they never gave coherence an explicit, glossary-style definition, it is easy to determine coherence is the degree of consistency between different translations and social worlds. Boundary objects play a critical role “in developing and maintaining coherence across intersecting social worlds” (p. 393).

Convergence is a more recent development—at least in name—extending coherence. It considers to what degree the “tools, systems, interfaces, and devices for storing, tracking, displaying, and retrieving information”—conceptualized as “information artifacts”—“are fitted to” the communities of users that create and work with them (Star et al., 2003, p. 244; see also Bowker & Star, 1999, pp. 46–49). Star and colleagues applied the concept of information worlds, as used but ill-defined by Chatman in her theory of normative behavior (see section 2.8.2.1), to the result of this convergence process. This view of convergence is a restatement of coherence, but from a different perspective. The starting point is the boundary objects themselves—in the guise of information artifacts—and not the social worlds. The focus is less on the translation process and more on the consistency between the results of the process and the communities—or social worlds—that are part of it. The difference between coherence and convergence is in perspective; the two as used by Star and her colleagues are not truly separate concepts.

2.7.2. Propositions

While many theories have explicit propositions explaining the relations between their concepts (see Grover & Glazier, 1986; Meleis, 1991), the propositions of boundary object theory are implicit, requiring analysis to derive. Such an analysis identifies five relational propositions in the statements made by Star and Griesemer (1989): two propositions discussing the relation between boundary objects and social worlds, and three propositions discussing the role boundary objects play in the process of translation and in determining coherence. The following analysis draws from Meleis’s (1991) discussion of how theories should be evaluated.

2.7.2.1. Boundary objects across social worlds

Star and Griesemer (1989, p. 393) stated boundary objects “both inhabit several intersecting social worlds … and satisfy the informational requirements”—or what LIS terms information needs—“of each of them.” They “are weakly structured in common use, and become strongly structured in individual-site use” (p. 393). These statements can be restated as two formal propositions which relate boundary objects and social worlds:

  • P1: Boundary objects are weakly structured enough to inhabit and be used across multiple social worlds, but become strongly structured when used within individual social worlds.
  • P2: Successful boundary objects satisfy the informational requirements (needs) of each of the social worlds they are used within; more successful boundary objects should satisfy more requirements from more social worlds.

These two propositions have been made probabilistic in practice, and could be reversed, although the degree of reversibility is low. They imply coexistence, not sequential cause-and-effect relationships; other variables could come into play as contingencies in determining the success of boundary objects, and the degree to which a boundary object is successful will vary.

2.7.2.2. Boundary objects’ role in translation and coherence

Star and Griesemer (1989, p. 393) further stated boundary objects’ structure “is common enough to more than one world to make them recognizable, a means of translation. The creation and management of boundary objects is a key process in developing and maintaining coherence across intersecting social worlds.” Three related claims are made:

  1. “the central cooperative task of social worlds which share the same space but different perspectives is the ‘translation’ of each others’ perspectives” (p. 412);
  2. translations are “perform[ed] … to craft objects containing elements which are different in different worlds,” i.e. boundary objects ( p. 412); and
  3. mismatches between the overlapping meanings and representations of these objects across social worlds “become problems for negotiation,” which to be successful (and maintain coherence) must take care to manage the boundary objects, their meanings and representations, and the interfaces they provide between social worlds (p. 412).

These statements, taken together, can be formally stated as three further propositions:

  • P3: Boundary objects, due to their recognizability across social worlds, should facilitate translation and support a level of coherence between these worlds.
  • P4: A successful translation and negotiation process is one supporting and maintaining a high level of coherence between social worlds.
  • P5: A high level of coherence should result from careful management of the creation, crafting, meaning, and representation of boundary objects and the interfaces they provide between and across social worlds.

These propositions are what Meleis (1991) terms stochastic, because successful translation and high levels of coherence should occur, but are not guaranteed. Proposition 3 is not reversible, and other factors may correlate with the facilitation of translation and support of coherence, implying coexistence and substitutability. Contingencies impacting on Proposition 1—and the degree of success of boundary objects—may carry over to Proposition 3. Propositions 4 and 5 are reversible and coexisting, since a successful translation process, careful management of boundary objects, and a high level of convergence should correlate with each other. Proposition 4 is contingent on other potentially substitutable factors that may influence the success of the translation process, although these are not included in the theory. Proposition 5 is sufficient and necessary because all factors which could be considered would be placed by Star and Griesemer (1989) under the banner of careful management of boundary objects and their interfaces, at least within the probabilistic and stochastic confines of the proposition. This is despite Meleis’s (1991) statement that sufficient and necessary propositions are uncommon in social science; in this case the breadth of factors considerable under the careful management of boundary objects allows these limitations to be overcome.

2.7.2.3. Convergence

As mentioned earlier, convergence and coherence can be considered the same concept from different perspectives. Social worlds and CoPs, while conceptualized by different researchers and having different definitions, were considered interchangeable by Bowker and Star (1999, p. 294). As such, Propositions 3, 4, and 5 about coherence could apply to convergence and CoPs, through substituting these terms for coherence and social worlds respectively. Having “use and practice fit design and access” (Star et al., 2003, p. 244) refers to the careful managing of how boundary objects and their interfaces are created, crafted, and represented, included in Proposition 5. Any relation between boundary objects, convergence, and information worlds as concepts is, however, a relation between boundary object theory and Chatman, Burnett, and Jaeger’s theories, which used the concept of information worlds; it is not a proposition within boundary object theory.

2.7.3. Theory as a Whole

Boundary object theory has many interrelations between its concepts, as seen above. These relations act in what Meleis (1991, p. 227) termed “a chain-link fashion”: Proposition 1 links boundary objects to social worlds; Proposition 2 explains the success (or not) of that link; Proposition 3 links boundary objects to coherence; Proposition 4 links successful translation and successful coherence; and Proposition 5 explains if the link between coherence, boundary objects, and social worlds is successful. The theory is based on and around these concepts, which are “concatenated” together (p. 227); boundary object theory is thus a concatenated theory. It takes a field approach in its construction, focusing “on the relationships between the phenomena and thus … [on explaining] the phenomena by the relationships” (Meleis, 1991, p. 227); it is also an explanatory theory (p. 229). Boundary object theory is an inductive grounded theory that takes advantage of the strengths of deductive and constructive reasoning to further its generalizability, applicability, and adaptability.

Star and Griesemer’s (1989) original application of boundary object theory was to an applied problem—translation by scientists between different social worlds—in the STS field. Boundary object theory has become a grand theory (as defined by Glazier & Grover, 2002) through its wide application in and extension by other disciplines, but could be placed within or near the category of middle-range theories due to its “substantive focus” (Meleis, 1991, p. 228) on the role of boundary objects in crossing social worlds; it does not try to explain everything about social worlds or boundary objects as a true grand theory would. This apparent duality shows in the propositions: they are not restricted to scientists and show high levels of substitutability and stochasticity, allowing for many and varied uses, interpretations, and extensions. The ecological approach taken by Star and Griesemer (1989), considering the view of every social world, allowed for wide application of their theory and for it to serve as a macrotheory (as defined by Meleis, 1991).

2.7.4. Further Development and Application

While Griesemer has not contributed further to boundary object theory, Star applied her theory and extended the scope of the problems it addresses. Her most cited work is the book Sorting Things Out: Classification and its Consequences, written with Geoffrey Bowker (Bowker & Star, 1999); they conceived of classifications as boundary objects used across multiple CoPs, new classifications being created when existing systems come into conflict. Bowker and Star introduced the concept of boundary infrastructures: sets or “regimes of boundary objects”—such as information objects and systems—which support multiple CoPs and social worlds (p. 313). Star applied boundary object theory in the context of CoPs (Bowker & Star, 1999, p. 294; Star, 2002, p. 118) and information worlds (Star et al., 2003), and in studying artificial intelligence (Star, 1989), electronic community systems (Star & Ruhleder, 1996), and digital libraries (Star et al., 2003), the latter introducing the revised concept of convergence. Her work explored the difficulties faced by designers and developers of information systems in supporting, in social and organizational contexts, successful translation between and coherence across the multiple social worlds and communities the systems are intended to serve.

Star believed boundary objects are “interpretively flexible” in many but not all locations and differ in their nature and shape “depending on the work or informational needs of the different social groups” part of their creation and use (Zachry, 2008, p. 452). Before her passing (see Clarke, 2010), Star was working on a book extending boundary object theory and exploring “the idea of different types of boundary objects” (Zachry, 2008, p. 454). In one of her last published papers (Star, 2010), she argued further consideration of types of boundary objects and their “material/organizational structure” (p. 602) was needed. She was interested, building on her work with Bowker, in what one might call the lifecycle of standardization of boundary objects: how they progress through standardization attempts, “the back and forth between ill structured and well structured,” and the eventual creation of “new boundary objects” by so-called “residual categories”—outsiders and others on the periphery (p. 614).

Other researchers have applied boundary object theory or developed it further; citation analysis using Web of Knowledge shows such usage has continued to increase, with about three times as many citations of Star and Griesemer’s (1989) paper in 2013 (150) as 2005 (53). The majority of use of boundary object theory and/or its concepts has been in the fields of management, history and philosophy of science, LIS, computer science, and sociology. Applications of boundary object theory have included the study of communities, collaboration, organizations, information systems (in various guises), and information behavior.

While reviewing all 1,446 total citations (as of June 2014) is beyond the scope of this chapter, the studies of Bødker and Christiansen (1997), Bannon (1997), and Lutters and Ackerman (2007) in CSCW are worth noting; the latter provided a brief, but useful summary of boundary object theory’s application in CSCW research (pp. 344–345). In management and knowledge management Pawlowski has applied the theory alongside CoPs (Pawlowski & Robey, 2004; Pawlowski, Robey, & Raven, 2000), and many others have applied the concepts from boundary object theory to KM concerns. This includes the work of Brown and Duguid (1996, 1998); Wenger (1998, 2000); Bechky (2003); Hislop (2004); Kimble and Hildreth (2005); Levina and Vaast (2005); and Kimble, Grenier, and Goglio-Primard (2010) on boundary objects, gatekeepers, and boundary spanners in KM. The LIS field has examined the information behavior of users, communities, and organizations and how it relates to information systems—including databases and digital libraries—serving as and containing boundary objects. A notable study is Van House’s (2003) conceiving of the Calflora subset of the University of California–Berkeley Digital Library as a boundary object. She found trust to be a major issue, arguing to build such trust, developers of digital libraries should conceptualize them and have them act as boundary objects: converging users’ social and information worlds together and supporting their current and emerging work processes, while retaining a common system identity. Fleischmann (2007a, 2007b) proposed a theoretical framework he termed “boundary objects with agency” to examine the embedded values in digital libraries, building on social worlds, boundary objects, and the concept of nonhuman agency present in actor-network theory.

Further developments of the theory have come from many authors. Henderson (1991) added her concept of “conscription devices” (p. 452): deliberately constructed objects that “enlist group participation” and contain “knowledge created and adjusted through group interaction … [with] a common goal” (p. 456). Akkerman and Bakker (2011) conceptualized transformation as one of four learning processes taking place around boundaries and boundary objects. They identified five potential stages: (a) confrontation, where “intersecting worlds [are forced] to seriously reconsider their current practices” (p. 15); (b) recognizing and building a shared problem space, including boundary objects as a part of this; (c) hybridization, where a new world begins to form around boundary objects; (d) crystallization, where the new hybrid world becomes “embed[ded] … in practice so that it has real consequences” (p. 17); and, in some cases, (e) “maintaining uniqueness of the intersecting practices,” where the established worlds reinforce themselves as distinct (p. 18). Akkerman and Bakker’s stage view of transformation is similar to Star’s (2010) recent discussion of the life cycle of boundary objects and the social worlds they interact with.

Researchers have further developed the theory by exploring the degree of flexibility boundary objects have, an area Star argued needed further study (Zachry, 2008). Fujimura’s (1992) work provides an early example of this: she argued boundary objects are too flexible, conceiving of “standardized packages” (p. 176); they combined multiple boundary objects with “standardized methods … in ways which further restrict and define each object,” reducing the possible interpretations without “entirely defin[ing] them” (p. 176). Gal, Yoo, and Boland (2004) looked at the flexibility of boundary objects, taking the concept of information artifacts (also seen in Star et al., 2003) as ever-changing boundary objects that changed the social worlds, information worlds, and information behavior of the various stakeholders in a construction project. They developed a model of “the dynamic interplay” among changes and adjustments in social identity, infrastructure, community, and the boundary objects themselves (p. 198). Lutters and Ackerman (2007) argued, based on empirical evidence, that flexibility for boundary objects in the situated context of negotiation and deviations from official process was necessary. They believed boundary objects should be viewed through a temporal lens: “boundary objects existed within a history greater than themselves” (p. 366).

Carlile (2002) was interested in the categories of boundary objects identified by Star and Griesemer (1989), arguing successful boundary objects—producing high levels of coherence and convergence—have three key characteristics: (a) “a shared syntax or language” for knowledge representation (Carlile, 2002, p. 451), shortened to “representing” ( p. 453); (b) a way for “individuals to specify and learn about their differences and dependencies across a given boundary” (p. 452), shortened to “learning”; and (c) a facilitation of “individuals … jointly transform[ing] their knowledge” (p. 452), shortened to “transforming” (p. 453). His work has been often cited in management and knowledge management, where he has continued his research and developed a framework describing “three progressively complex boundaries … syntactic, semantic, and pragmatic,” alongside “three progressively complex processes … transfer, translation, and transformation” (Carlile, 2004, p. 555).

2.7.5. Limitations and Criticisms

While boundary object theory has been praised by most, a few researchers have criticized the theory—either as a whole or elements of it—and pointed out its limitations. First, Fujimura (1992) and Lee (2007) have questioned if a fully ecological approach can be taken, Fujimura contending whichever viewpoint or social world there was the most data about would be unavoidably central to any study of boundary object(s), despite Star and Griesemer’s best efforts. Second, the degree of flexibility of boundary objects has been questioned. Fujimura (1992) believed they had been conceptualized as having too wide a “margin of negotiation” (p. 175), and their weak, flexible structure across social worlds would hamper their possible success in encouraging continuing coherence between social worlds. In contrast, Lee (2007) believed boundary objects were not flexible enough. Gal, Yoo, and Boland (2004) and Lutters and Ackerman (2007) stressed at least some boundary objects must remain flexible in situated and temporal contexts, but the processes of standardization discussed by Fujimura (1992), Bowker and Star (1999), and others must be considered. Third, Lee (2007, p. 313) did not believe boundary object theory accounted for “active negotiation of shared understanding,” arguing “the boundary objects concept is not incorrect … [but] it is incomplete” and concluding the active and chaotic negotiation processes she identified in her study were major elements missing from boundary object theory. Other factors may impact on and influence boundary objects, such as Van House’s (2003) identification of barriers to information sharing: fear over potential misuse, need to make data presentable, conflicts with professionals from other fields and worlds, and concerns over trust. Contextual factors must not be ignored in studies of boundary objects within and across social worlds and communities.

The introduction of convergence by Star, Bowker, and Neumann (2003; see also Bowker & Star, 1999, pp. 46–49) is a potential liability, due to their use of multiple conceptions of community. If social worlds, CoPs, and information worlds are fully compatible with each other, then convergence implies a cycle where social worlds, CoPs, and information worlds converge with information artifacts to create information worlds. However, Star et al. (2003, p. 244) used a different definition of an information world than the concept’s originators (see section 2.8.2 below), as “the collection of information resources employed by an individual, organization, institution, or other group to solve [information] problems, learn, play, and work.” They removed from the equation shared activities (as in social worlds), information behaviors, and the concepts of social norms and social types originating in Chatman’s theory of normative behavior. Star, Bowker, and Neumann’s idea of information worlds as the result of convergence should not be considered to be the same concept as Burnett and Jaeger’s version of information worlds or the conception used but ill-defined by Chatman in her work. Because coherence has no required cycle and is compatible with other theories and conceptions of community, it is a better and more flexible concept to apply to many research problems.

2.8. A Theoretical Framework for Social Digital Libraries

Two other theories are also important in the context of social digital libraries: Strauss’s social worlds perspective and Burnett and Jaeger’s theory of information worlds. While these theories were introduced in section 2.2 as including relevant concepts of community, their important relations with boundary object theory suggest further analysis and synthesis is appropriate. Such a synthesis with boundary object theory can serve as a well-grounded theoretical framework for social digital libraries.

2.8.1. Social Worlds Perspective

Strauss’s social worlds perspective (Strauss, 1978; see also Clarke & Star, 2008) was discussed in sections 2.2.4 (as a concept of community) and 2.7.2.1 (as a component of boundary object theory). As a brief review, Strauss (1978) built his social worlds perspective on the work of Shibutani (1955), who argued there is a great “variety of social worlds,” each with its own “organized outlook” based on the norms, beliefs, interaction, and communication of a given group of people ( p. 566). Strauss (1978) proposed social worlds consist of “activities, memberships, sites, [and] technologies” in relation to social change ( p. 121). A social world included

  • “at least one primary activity … strikingly evident”;
  • sites where activities occur”;
  • technology … [for] carrying out the social world’s activities”; and
  • in established social worlds, “organizations … to further one aspect or another of the world’s activities” (p. 122; emphasis added).

To these four key concepts, Strauss added social worlds could and would “intersect … under [various] conditions” and segment into smaller subworlds given sufficient analysis (p. 122).

2.8.2. Theory of Information Worlds

2.8.2.1. Chatman’s small worlds research

Burnett and Jaeger’s (2008; Jaeger & Burnett, 2010) theory of information worlds built on the work of Chatman, whose work culminated in her theory of normative behavior (Burnett et al., 2001). She focused on an interest in what she called small worlds, ascribing (in Chatman, 1991) her use of this term to her dissertation chair, Patrick Wilson (1983). Wilson stated that, based on one’s social location, the information and knowledge one would need or be interested in would differ: “what one needs to know … depends in part on what others expect one to know” (p. 150). Wilson labeled worlds where people were selective in their choices of information, knowledge, and topics of interest “small worlds.” While neither Chatman or Wilson had formal sociological training (Fulton, 2010, p. 239; Maclay, 2003), they were well-read in sociology and anthropology (evident by the citations in P. Wilson, 1983, and in Chatman, 1983, p. 104). Small worlds to Wilson (1983) and Chatman (1991) were social worlds (Shibutani, 1955; Strauss, 1978) where people experienced a limited view of the broader social world; a “small world life” was one “played out on a small stage … characterized by commonness or routineness” (Pendleton & Chatman, 1998, p. 733). Wilson (1983) focused on information and knowledge sources and their perceived authority, while Chatman took a broader approach to information (see Burnett et al., 2001; Chatman, 1991, 1992, 1996, 1999, 2000; Pendleton & Chatman, 1998). Jaeger and Burnett (2010) restated her recent views of small worlds as moving from “definable localized social groupings of people” towards a broader view of “social environments in which an interconnected group of individuals live and work, bonded together by common interests, expectations, and behaviors” (p. 21).

Chatman developed three theories in succession as she found existing theories lacking in explanatory power; these were information poverty (Chatman, 1992, 1996), life in the round (Chatman, 1999), and the culmination of her work, her theory of normative behavior (Burnett et al., 2001; Pendleton & Chatman, 1998). The latter examined small social worlds in the context of “normative behavior,” that “viewed by inhabitants of a social world as most appropriate for that particular context” and socially constructed within that context (Burnett et al., 2001, p. 538). Chatman and her colleagues focused on four key concepts, first presented in Pendleton and Chatman (1998): (a) social norms, or the “standards of ‘rightness’ and ‘wrongness’ in social appearances”; (b) worldview, or the “collective perception held in common by members of a social world regarding those things that are deemed important or trivial”; (c) social types, “the [social] classification of a person”; and (d) information behavior, “a state in which one may or may not act on available or offered information” (Burnett et al., 2001, p. 537). Theoretical propositions were proposed (p. 538), but Chatman and Pendleton’s (1998) and Burnett et al.’s (2001) studies focused on the concepts. The broadening of Chatman’s thinking, moving beyond information poverty and even small worlds themselves to an extent, can nevertheless be seen in these two works.

2.8.2.2. Burnett and Jaeger’s theory

Burnett and Jaeger (2008), while building on much of Chatman’s theory of normative behavior, wanted to move beyond its limitation in small worlds. Chatman had used the term information world as early as the 1980s (see Chatman, 1983, 1987), but left it ill-defined, requiring its meaning to be interpreted based on her views of small worlds and social worlds. In developing their theory of information worlds, Burnett and Jaeger (2008) saw to be more explicit, combining Chatman’s work with Habermas’s on lifeworlds and the public sphere. Habermas’s work conceptualized lifeworlds as “the collective information and communication environment—the social tapestry—of a society” (Burnett & Jaeger, 2008, “Public sphere” section, para. 7). Burnett and Jaeger believed many of Chatman’s concepts could apply to both the smaller and larger worlds people are part of, and borrowed the concepts of social norms, social types, and information behavior from the earlier theory. Their version of information behavior was couched in somewhat broader terms than Chatman’s, as “the full spectrum of normative [information] behavior … that are available to members of a … world” (“Small worlds” section, para. 8). In Jaeger & Burnett (2010) they included a refined version of worldview, termed information value, which related to the value judgments of different information within and across worlds. They preferred to term the resulting worlds information worlds, stressing the ability to examine the context of lifeworlds, small worlds, and other worlds of varied sizes, settings, or shapes.

Jaeger and Burnett (2010) further refined their theory in a book, adding a fifth concept of boundaries, “the places at which information worlds come into contact with each other and across which communication and information exchange can—but may or may not—take place” (p. 8). Information worlds are social spaces, contiguous or overlapping, where individuals can “cross between the different worlds to which they belong” and interact with individuals from other worlds (p. 9), allowing for “multi-leveled” analysis of “the interactions between social norms and values, information, and community, particularly in situations in which multiple small worlds overlap” (p. 30).

2.8.2.3. Propositions

No theoretical propositions were offered in either publication by Burnett and Jaeger (2008; Jaeger & Burnett, 2010), but Burnett has since written five propositions focused on defining the five core concepts and their roles within and across information worlds (Burnett, Burnett, Kazmer, & Hinnant, 2012, p. 9), albeit these are as of yet unpublished. Further relational propositions can be extracted from the first two chapters of Jaeger and Burnett’s (2010) book and grouped into three themes as follows:

  • P1: Mutual influence

    1. “Information behavior is simultaneously shaped by immediate influences,” including “friends, family, co-workers, and trusted information sources … as well as larger social influences” such as “public sphere institution, media, technology, and politics” (pp. 7–8).
    2. The broader information lifeworld is influenced by the perspectives of the smaller information worlds that make it up and by public institutions such as the media, government, libraries, and schools.
    3. In return, these promote, constrain, or influence—to varying degrees—the movement of information between information worlds and, in some cases, within them.
    4. Information worlds can be seen on multiple levels—small, intermediate, and the broader lifeworld—which shape and are mutually influenced by each other and by their participants’ behavior (p. 31); “social contexts are not, in most cases, isolated from one another” ( p. 30).
  • P2: Normative behavior

    1. Each information world develops its own “normative ways in which information is accessed, understood, and exchanged both within the [information] world and with others outside that world” (p. 8).
    2. The common (i.e. social) norms, types, values, and behavior of each information world require individuals to “generally conform to the norms and expectations” of the world they are interacting with at any one time (p. 8).
  • P3: Boundaries

    1. Points of interaction with members of other information worlds “serve as the boundaries between different worlds” (p. 9).
    2. These boundaries are the conduit through which information flows between worlds. This information is understood in the context of each world’s social norms, social types, information value, and information behavior.
    3. Such interactions and information flow may lead to new information worlds being created or existing ones changing or being eliminated.
    4. Information from the broader lifeworld “can be conceived of as a kind of ‘boundary object’” (in the sense intended by Star and Griesemer); it exists apart from individual information worlds and as part of them, seen in the context of the world’s norms, values, types, and behavior (p. 31).

2.8.3. Synthesis

Having presented Strauss’s social worlds perspective and Burnett and Jaeger’s theory of information worlds above, this section synthesizes these two together with boundary object theory to produce a well-grounded theoretical framework for studies of social digital libraries. The synthesis proceeds in three steps: (a) comparing the social world and information world lenses; (b) considering the results of coherence and convergence, under boundary object theory, in relation to information worlds; and (c) synthesizing these together into a rich, theoretical view of boundary objects in relation with existing and emergent communities and worlds.

2.8.3.1. Social and information: Two lenses on worlds

The framing concepts of the social worlds perspective and the theory of information worlds differ: the former focuses on activities, sites, technologies, and organizations, while the latter is concerned with mutual influence, normative behavior, and the five core concepts of social norms, social types, information value, information behavior, and boundaries. Nevertheless, Chatman (1991) had thought of small worlds as a special subset of a broader social world. While one could extend this to think of a small information world as a special subset of a broader social world, in the context of the higher-level information worlds added to the theory by Burnett and Jaeger (2008) this analogous reasoning breaks down; social worlds and information worlds were considered at multiple levels of analysis by their originators (Jaeger & Burnett, 2010, p. 30; Strauss, 1978, p. 121).

A different analogy is better for synthesis. Since both offer a multi-level understanding of social groupings, an information world should be seen not as a subset of a social world, but as a different lens or perspective on one. Strauss (1978) argued for focusing on his framing concepts, while Burnett and Jaeger (2008; Jaeger & Burnett, 2010) argued for theirs. Nevertheless, at its core each is concerned with the same base phenomenon of a world of people, of various size, who are engaged in interaction, communication, and social construction of this and other worlds. Strauss, Burnett, and Jaeger all held these worlds or communities differ in one or more characteristics: size, shape, visibility, abstractness, permeability, and structure. They may intersect and overlap, or be contiguous; social and information worlds can also segment into smaller subworlds. Each theory takes a broad view of its framing concepts; Strauss (1978) construed activities, sites, technologies, and organizations as flexible, while Burnett and Jaeger (2008; Jaeger & Burnett, 2010) considered a wide range of social norms, social types, information value, and information behaviors, defining information to encompass data and knowledge.

Burnett and Jaeger’s theory has stronger grounding than the views of social worlds taken by Shibutani (1955) and Strauss (1978) in previous theoretical and practical work in LIS. Its focus on information behavior, information flow, and the roles of information in the context of the worlds people are part of has clear applicability to multiple research tracks within the LIS field (many discussed by Jaeger & Burnett, 2010). The alternative, but compatible lens offered by Strauss and his social world perspective (see also Clarke & Star, 2008) can and should be applied where and when appropriate.

2.8.3.2. Information worlds meet boundary objects

As discussed in section 2.7.2.3, Star, Bowker, and Neumann’s (2003) idea of information worlds as the result of convergence is not the same concept of information worlds as used by Chatman, Burnett, or Jaeger. Nevertheless, Star et al.’s work indicates how boundary object theory could expand beyond social worlds to other forms of community. Burnett and Jaeger’s (2008; Jaeger & Burnett, 2010) theory of information worlds, given its compatibility and similarities with social worlds and its incorporation of boundaries as a core concept, is an obvious place where such expansion can take place in LIS contexts. The two theories have been used together in Burnett, Subramaniam, and Gibson’s (2009) study of Latina IT professionals’ social construction of gender as a boundary object in context of the IT industry as an information world, but solely as a framework for coding and analyzing data; a fuller theoretical synthesis is required.

Per Proposition 5 (as derived) of boundary object theory, coherence comes from careful management of boundary objects and the interfaces they provide between and across social worlds. The characteristics of information worlds impact the interface boundaries—the points of interaction—between them (Jaeger & Burnett, 2010). Information from the broader lifeworld is seen “as a kind of ‘boundary object’” (p. 31). It is not much of a stretch, applying boundary object theory further, to conceive of information from a different information world having a boundary with this one as a potential boundary object. Such boundary objects serve as partial, cohered expressions of multiple information worlds. They represent a portion of their socially constructed normative behavior, common meanings, and common ground of information and knowledge (cf. Bechky, 2003; Davenport & Prusak, 2000, p. 98; Olson & Olson, 2000, pp. 157–161; Wasko & Faraj, 2000).

This partial-but-common expression of norms, types, values, and behavior by boundary objects relates to the revised process of convergence discussed by Star, Bowker, and Neumann (2003). Per Jaeger and Burnett (2010), the multi-leveled interactions and information flow present in information worlds may lead to new information worlds being created or existing ones changing or being eliminated. Per Star et al. (2003, p. 244), a community has a better fit with an information artifact “when use and practice fit design and access” (Star et al., 2003, p. 244). Combining the two theories, for this to be true the artifact must be compatible with the community-as-information-world’s norms, types, values, and behaviors in-use and in-practice. A good fit implies the artifact shares at least a subset of the characteristics of the information world. If the artifact is used by other information worlds, subsets of their characteristics will exert mutual influence on the artifact, turning it into a boundary object.

The combination of these different subsets—with some norms drawn from one information world, others from another, and yet others from still another—could then be seen as the norms, types, values, and behaviors of a new information world, emergent and socially constructed around the artifact-as-boundary-object as it converged the different information worlds together. The information flow allowed by the boundary object between these information worlds would have created a new world, one sharing sufficient elements of—common ground (cf. Bechky, 2003; Davenport & Prusak, 2000, p. 98; Olson & Olson, 2000, pp. 157–161; Wasko & Faraj, 2000) from—each of the original worlds to communicate, interact, and share information and knowledge with them. It would have coalesced, cohered, and converged into its own world and community surrounding the information artifact that began life as a boundary object, but has now become a local standard for the new information world.

2.8.3.3. All together now

Returning to the view through the lens of Strauss’s (1978) social world perspective, boundary objects should act as common sites and technologies for people from different social worlds to engage in common information-based activities. As a new community coalesces and establishes around the boundary object, it may form an organization—formal or informal—to further these activities. The full theoretical view established here can be summarized in propositional form:

  • P1: Boundary objects serve as partially cohered expressions of the socially constructed characteristics—social norms, social types, information value, information behaviors, activities, and organizations—of the multiple information and social worlds they are used within and across.
  • P2: Boundary objects, maintaining a common identity, act as common sites and technologies for information-based activities—including information and knowledge sharing—within, between, and across these worlds.
  • P3: Over time, new worlds may coalesce, cohere, and converge around boundary objects, sharing a combination of norms, types, values, behaviors, activities, and (potentially) organizations, as the boundary object becomes a new, localized, and socially constructed standard.

This rich theoretical picture echoes Star’s (2010, pp. 614–615) comments on the cycle between ill-structured and well-structured boundary objects and Akkerman and Bakker’s (2011, pp. 15–19) conceptualization of the process of transformation. It further grounds the process of convergence in theory from sociology, philosophy, and information science; and synthesizes the interrelated work of Strauss, Star, Chatman, Habermas, Burnett, and Jaeger on social worlds, boundary objects, small worlds, normative behavior, lifeworlds, and information worlds.

2.8.4. Resulting View of Social Digital Libraries

Under the theoretical framework developed above, social digital libraries are used by and cross the boundaries of multiple social worlds, information worlds, and communities; they are socially constructed boundary objects (Van House, 2003), and should adapt to the “local needs” (Star, 1989, p. 46) of as many of these worlds and communities as possible. Serving as an interface and translation device between social and information worlds, they should reconcile the “meanings” and understandings across these worlds to allow users to “work together” (Star & Griesemer, 1989, pp. 388–389), collaborate, and interact. The translations they provide should be coherent and consistent for and with as many of the social and information worlds that use them as possible.

Social digital libraries should support the emergence of localized and common social norms, social types, information values, and information behaviors shared—to varying and potentially overlapping extents—by the different information worlds using them (Burnett & Jaeger, 2008; Jaeger & Burnett, 2010). Social digital libraries should act as common sites and technologies for users to engage in information-based activities (Strauss, 1978), including collaboration, information sharing, and knowledge sharing. In doing so, they should support the possible emergence of a broader community (information world, social world) as the social digital library converges portions of the multiple communities it serves.

2.9. Conclusion

This chapter has established the need for research on social digital libraries—content collected for a user community, services offered to the community, and one or more formal or informal organizations managing these content and services—as inherently social organizations and environments, socially constructed by users, communities, and organizations. While different conceptions of communities and collaboration may be included, a theoretical framework based in Star’s boundary object theory and incorporating elements of Strauss’s social worlds perspective and Burnett and Jaeger’s theory of information worlds provides a well-grounded, flexible basis for studying social digital libraries as boundary objects. Research is needed to test part or all of this framework as a theory, using operationalized definitions of its concepts and propositions, but research should explore and describe what roles one or more digital libraries play, as socially constructed boundary objects, in supporting and facilitating collaboration, communities, and other socially constructive behaviors within and across social and information worlds. This latter task is more achievable at the present time, and so the following chapter begins the presentation of a study—this dissertation—that helps satisfy this latter need.