Libraries and other memory institutions throughout history have developed a range of methods and tools for transmitting full-text documents between different media. In this sense, library digitization belongs to the same tradition as 20th-century microfilming and the ancient transcribing of manuscripts. The Gutenberg era marked a sharp decline in this full-text transmitting business, and libraries devoted their time to producing bibliographical records for documents rather than reproducing full documents. With digital reproduction technologies, however, libraries have drawn a historic circle. They are yet again dedicating much energy and attention to the full-text transmission they largely abandoned at the dawn of the printed age. In so doing, they take on a much more explicit role of producing and shaping the cultural heritage in addition to its accustomed role of preserving it and making it available. In this paper, we will discuss this new role, i.e., the practices of digitization within the library institutional setting, and, in particular, the national library setting.

When looking at digitization as a practice used by libraries and other cultural institutions to enhance access to their collections, it is easy to perceive the digitization process as a neutral and mechanical process that simply transfers the document or artefact from one type of media to another, i.e., from an analogue to a digital environment. In this article, however, we will follow Dahlström & Hansson and Kjellman and consider library digitization as a knowledge organization practice — comparable to, e.g., descriptive cataloguing — and therefore a signifying practice rather than a neutral, unconditional and mechanical one (Dahlström and Hansson, 2008; Kjellman, 2008).

In line with this, we ask to what extent digitization can be expected to embody, mediate, and document particular discourses and identities within various library and archival institutions. This is in line with recent research in knowledge organization acknowledging how tools and practices that manage and organize documents and artefacts in institutional settings, also carry social meaning. Such research picks up bibliographical tools and discusses them as documents, or texts, each with a particular story to tell, and considers them as biased instruments that potentially embody and mediate particular discourses and institutional identities (Andersen and Skouvig, 2006; Bowker and Star, 1999; Dahlström, 2006; Frohmann, 2004; Hansson, 2006; Hanson, 2010; Hjørland, 2000; Kjellman, 2006; Olson, 2002). This means conceiving of cultural heritage digitization as a remediating practice, affecting the library in several ways. We will here put focus on two aspects:

1. How digitization — when looked upon as a practice related to knowledge organization — contributes to how the collected and organized objects can be understood and in what way they signify meaning in relation to the kind of institutions of which they are parts. In this aspect, digitization is a practice that shapes and reshapes our cultural heritage.

2. How digitization not only transforms cultural heritage itself but also the social character and role of the library. Memory institutions such as libraries develop and change their identities and roles as a result of the development of new and modified professional tools.

By combining research within the three fields of knowledge organization, digitization, and institutional identity, this article points to a way of addressing issues in all of these fields: in knowledge organization and digitization research by using a framework of institutional theory to enhance an understanding of the creation of meaning through these practices, and in institutional theory by pointing out how specific tools and practices help forming identity and value for an institution such as a national library.

Digitization of library collections entails much more than just technically capturing the texts and images of analogue documents. It is rather a complex chain of events (Deegan and Tanner 2002, p. 34f.). In principle, every link in such a chain affects and delimits the nature and quality of the final digital product (Dahlström, 2009a, p. 173f.). How the different links are being implemented and cooperate is in turn dependent on the overall strategy for the digitization project and the general transmission policy of the digitizing institution. These implementations, policies and strategies tangibly affect the extent to which the digitized documents are made publicly and richly available, malleable, and reusable for new research or digitization endeavours (Dahlström, 2009b). They also create a framework for the way the digitized resources may be interpreted. This framework itself, we argue, affects the creation of meaning in a social perspective for the digitizing institution.

In what follows we will first look at the digitizing issue in relation to the library’s institutional practice, and how that practice affects the way documents are managed and presented to the user. In the last section the focus will be on how the new digitizing practice affects institutional identity in, primarily, libraries and museums as memory institutions.

Digitization projects have been a part of library activities for some decades now. With a tentative start in the 1980’s, digitizing activities have increasingly turned into an established practice at almost every national library in the western world. Not only books but also journals, posters, photographs, postcards, letters, microfilms and manuscripts, indeed all kinds of cultural heritage materials collected by the library, have been subjected to digitizing projects of different kinds and converted into ones and zeros. The steps taken have been motivated by a desire to make the items available to a wide range of users on the World Wide Web. As we will see later in this text, however, other purposes have been on the agenda as well.

If we look retrospectively, a general trend has been that cultural heritage digitization started with local in-house initiatives and solutions, developed into experimental phases, and is now increasingly attempting to achieve and adhere to international standardization solutions. Such a development has much in common with the way that digital libraries in general have evolved during the last twenty years, and it has affected technical solutions as well as new and enhanced knowledge organization tools (Besser, 2002). One example is the current development of Web-Dewey, which is becoming increasingly standardized on an international level through its rapid implementation, not least in Europe.[1] Another is the shift of focus that the introduction of FRBR[2] has brought to descriptive cataloguing, from individual documents to the concept of ‘work’ as the formal basis of analysis.

Initially, digitization projects were aimed at developing local tools to facilitate in-house management of the collections. When the web increasingly emerged in the mid-nineties as a tool to be reckoned with, digitization projects began addressing the issue of giving users access to cultural heritage materials and documents online. Many projects investigated how to develop best practices and how to share solutions with other actors. On a European level, for instance, Minerva has been such a joint effort to formulate agreed upon standards and recommendations for both technical formats and work practices, across the institutional borders of archives, libraries, and museums.[3]

This development is not unique to libraries. Since digitization creates opportunities for a more global and general access to our cultural heritage, it brings with it ambitions to adjust local tools and practices to more standardized ones that can be shared by different actors and domains. This has of course been justified for rational and economical reasons — but not only. Shared and standardized knowledge organization tools also enhance opportunities for sharing information between and among different areas and domains (Bowker and Star, 1999; Kjellman, 2008). Using the same kind of metadata and vocabularies in different domains would, in the long run, give the users opportunity to simultaneously co-search different heritage collections on the Internet (Besser, 2002; Hollinka et al., 2004, 601).

This aim accords with the library’s traditional practices. Libraries have a long tradition of collecting and describing their collections in a standardized and levelling way. The library’s mission has been to bring together and present the collective knowledge in a neutral way, and not to emphasize particular views or items at the expense of others (Broome, 1988). In the case of collecting, the library ambition has been to harvest almost all printed knowledge. The legal deposit system, in Sweden dating back to 1661, has also contributed to this ambition. In the case of descriptive cataloguing, the ambition has been the same: all objects in a collection should be managed in the same way, using the same kind of tools, and with equivalent effort.

Looking at the institutional practice of museums, we see a rather different tradition from that of the library sector. When museums collect documents, they also select — much more explicitly and decidedly so than libraries. As museums present their objects in exhibitions and catalogues, there is always a critical assortment of what to show and how. Furthermore, museums (and archives) mainly manage unique objects, and their digitization in consequence regularly concerns single document artefacts rather than the text as a presumed commonality in, e.g., a multiple-book edition. Library collections, on the other hand, are largely made up of printed documents that are mass produced to begin with. Given that the many copies of a published book are normally thought of as identical,[4] digitization projects in libraries are usually more or less indifferent as to whether the one copy or the other is picked (rather than selected) as source document, i.e., as ‘ideal copy’.

Museums and libraries also differ as to where they put their emphasis when selecting, digitizing, and presenting their collection objects. Maroevic identified three central importance criteria for cultural heritage objects within memory institutions: significance, materiality, and form (Maroevic, 1998). He suggested that the balance between these is quite different in libraries and museums. For instance, whereas libraries tend to downplay materiality and form, museums reversely prioritize those criteria. As a result, museums and libraries prioritize differently between these criteria when digitizing their documents. When digitizing a book, for instance, libraries might easily ignore its materiality, spatiality and possibly even form, displaying it primarily as sequences of linguistic text (as the core ‘meaning’ of a work deemed to be significant). A museum, on the other hand, that would be digitizing a spatial artefact such as a sculpture by seeking to reduce it to its bare meaning, significance and perhaps ‘information’ (or whatever is considered to be equivalent to the text of a written object) while ignoring its materiality and form, would be quite unimaginable.

In the text above we have identified two distinguished institutional practices towards the material collected; that of the library and that of the museum. Both practices are historically stable and contribute significantly to shaping the roles of the respective institutions. Digitization changes this to a certain degree. Perhaps this change is most tangible in libraries where the strategies for transmitting, organizing and displaying documents at times approach the more selective practice of museums. However, when libraries digitize their material they show a differentiation in this practice, in particular in their role as cultural heritage institutions. Meeting the possibilities of digitization, libraries are starting to develop views on both developmental and organizational aspects of their collections which turn them into being much more critical and selective in choosing which documents to prioritize as specifically significant from a cultural heritage point of view, and how to exhibit them on the web. At the same time, traditional non-selective practices resurface in the ambition to digitize ‘everything’ in the collections, treating each document in the same way — as has been the aim of bibliographic description for centuries. The differentiation is an important aspect of a distinction between critical digitization and mass digitization.

We will in this part of the paper make a distinction between mass digitization and critical digitization and see to what extent each of them is instrumental to the library’s institutional roles and goals.[5] In doing this, we will consider the ideological dimension of digitization and the rationales for digitization particularly within national libraries. We argue for considering digitization both as a document representation practice (i.e., as a bibliographical knowledge organization activity producing document surrogates and similes) and as a mediating documentation practice (of collections as well as institutional identities) that goes beyond the mere technologies of digitization.

The current trend within national libraries and large research libraries is, no doubt, mass digitization, where huge amounts of documents are digitized by automated means during a relatively short period of time (Coyle, 2006). A much talked-of example is the Google Book Search (GBS) project. It operates on an industrial scale and with as many digitization chain links as possible fully automated. Mass digitization systematically digitizes whole collections with no particular means of discrimination — the idea is to digitize ‘everything’ within the collection. The process by necessity has to avoid manual and labour-intensive work and leaves out intellectual, interpretative and qualitative aspects such as expert commentaries, context descriptions and analyses, descriptive text encoding, or manual proof reading. It cannot afford to produce new metadata and other bibliographic information about the source document. Digitization is flattened out into a linear streamlined affair. Mass digitization gains momentum and value in scale, not in depth. And, as has been pointed out by Deegan & Sutherland, projects such as GBS combine on the one hand commercial agents who are strong in financial resources but in need of content, with, on the other hand, public libraries who are reversely strong in content but in need of financial resources (Deegan and Sutherland, 2009). A marriage made in heaven, it would seem. The result is a gigantic, growing bank of digital texts that can be used as a localizing tool, be free-text searched, and — perhaps more importantly — form the technical base for many kinds of future enhancement using, e.g., data mining or software development. This is an important feature of mass digitization efforts such as GBS: they pour the digitized documents into large silos, pile them up on a flat scale, and treat them as equal items in a large database. GBS is not being modelled on the hierarchical library but on the heterarchical index of the web. This also has a modularizing effect: the items in GBS are discrete and de-contextualized, because the digitizing process has stripped away much of the bibliographical and contextual information about the objects and also because GBS aims for items to be able to be shuffled around, used and re-used in new contexts.

Only a limited number of libraries, however, have the interest, competence and resources to implement mass digitization. Some objects and collections, furthermore, are in need of considerable resource-consuming and manual labour during the digitization process. Such digitization cannot reasonably be referred to as ‘mass’, but is arguably better designated as ‘critical’ digitization.[6] Critical digitization implements several of the links in the long digitization chain in a manual, intellectual, and critical way. At every step one can make choices, select, leave out, and interpret. Mass digitization turns a blind eye to most of these choices, whereas critical digitization acknowledges and makes active use of them. A critical digitization project might, e.g., devote its resources to a single document, or may need to deliberately and strategically select one document from a set of source document candidates. The document text may prove difficult to decipher, or it may need to be considerably edited or commented upon by experts in order to make sense in its new digitized context. And, obviously, it may be out of the question to destroy the source document during the digitization process (as mass digitization sometimes does), but on the contrary it may be crucial to subject the source document to careful preservation or conservation measures. The project may wish to produce a representation that is as faithful and exhaustive as possible in relation to the graphical and material original artefact, providing the representation with rich metadata, indexing, descriptive text encoding, expert paratexts, and full bibliographical and contextual information. In such instances, scholarly research is embedded in the objects themselves.

Therefore, critical digitization is qualitative in the sense that it concentrates on what is unique and contingent in the documents, whereas mass digitization is quantitative in its design to capture what are common, regular, foreseeable traits in large amounts of documents and data. In consequence, then, critical digitization normally has to develop project-specific practices and tools, tailoring them to the qualities of the documents in the particular collection. In mass digitization, the single documents in the digitized collection are, on the contrary, subordinated (tailored, if you will) to more generally, perhaps even universally, standardized practices and tools of knowledge organization.

We can briefly compare the two approaches using a table:

Table 1: Critical digitization vs. mass digitization. Based on Dahlström, 2010

Critical digitization	Mass digitization
Primarily manual	Primarily automated
Critically recognizes the distortion digitization brings about	In effect treats digitization as a cloning process
Undertakes a well informed selective analysis of source copies	Normally picks whatever source copy that praxis or chance happen to present
Maximizes interpretation and metadata	Minimizes interpretation and metadata
Qualitative in its concentration on what makes a document (or set of documents) unique	Quantitative in its concentration on common, regular, foreseeable traits in large numbers of documents
Treats documents as graphical, spatial and material artefacts	Focuses on the linguistic texts of documents
Discrimination: selections/exhibitions	Exhaustiveness: complete holdings
Depth	Scale