Mapping Beyond Dewey's Boundaries: Constructing Classificatory Space for Marginalized Knowledge Domains.

ABSTRACT

Classifications are bounded systems that marginalize some groups and topics by locating them in ghettoes, diasporized across the system. Other marginalized groups and topics are totally excluded from these systems, being outside of their territorial limits. Because classifications are locational systems, spatial analyses borrowed from various disciplines have potential to identify and address their problems. The philosophical basis for the analysis in this article is Lorraine Code's (1995) conception of "rhetorical spaces" as sites where topics can be taken seriously as legitimate subjects for open discussion. In existing classifications, there is rhetorical space for most mainstream social and scholarly knowledge domains but not for marginalized knowledge domains. Geography offers concepts for building a theoretical framework to ameliorate the biases of classification. This article describes such a framework and how it is applied using techniques such as Gillian Rose's (1993) "paradoxical spaces," which are simultaneously or alternately in the center and at the margin, same and other, inside and outside to develop a more complex and meaningful classification for women and other marginalized groups. The project described here operationalizes these theoretical openings by applying them to the Dewey Decimal Classification as both critique and and as techniques for change.

INTRODUCTION

The problem of bias in classification can be linked to the nature of classification as a social construct. It reflects the same biases as the culture that creates it. Existing literature has critiqued the most widely used classification in the world, the Dewey Decimal Classification (DDC), for its treatment of women, Puerto Ricans, Chinese and Japanese Americans, Mexican Americans, Jews, Native Americans, the developing world (including Africa, the Middle East, and Melanesia), gays, teenagers, senior citizens, people with disabilities, and alternative lifestyles.(1) To look at these biases with a fresh eye, a theoretical construct capable of revealing the complexities of classification and its social construction was sought. The theoretical framework that subsequently evolved draws on the spatial metaphors that have become so prevalent in cultural criticism in recent years.

As Lorraine Code (1995) points out:

   [use of] spatial metaphors picks up a late-twentieth-century concern with
   location: with territories, mappings, positionings where resources are
   variously available, subjectivities are variously enacted, and identities
   are constructed and continually reconstructed in the enactings; and where
   hierarchies of power and privilege always contribute to shaping these
   processes.... (p. ix)

In this spirit, this discussion will move from a description of the construction of classification to the development of spatial imagery as a metaphorical mechanism with the ability to discover the processes by which powerful and privileged discourses shape information and with the potential to inform change. What will then evolve will be a multidisciplinary theoretical framework based on spatial conceptions in the context of a specific project, concluding with suggestions for further research.

THEORETICAL MODELS FOR THE SOCIAL CONSTRUCTION OF CLASSIFICATION

The idea of classification as a social construct is not new. A. C. Foskett (1971) suggests that classificationists are the products of their times. Therefore, since classifications are the products of classificationists, classifications also reflect the biases of their times. Examining the ideological construction and present needs for reconstruction of the former Soviet classification (Sukiasian, 1993) or the Confucian, and later Maoist, classification in China (Studwell, Wu, & Wang, 1994) makes it easy to see that classifications reflect philosophical and ideological presumptions of their cultures and not only the times but also the places. Classifications arrange concepts according to accepted cultural discourses whether those discourses are Leninist or Maoist communisms, the Seven Epitomes of Confucian doctrine, or Dewey's apparent reversal of Francis Bacon's classification scheme.

Allocation of 80 percent of DDC's religion section (the 200s) exclusively to Christianity and the existence of a separate section for American literature (the 810s) when all other literatures are arranged by language is not surprising given the origins of this classification. Finding the topic "concubinage" under customs in 392.6 where it is gathered with topics such as chaperonage and dating or "suttee" in 393.9 and all combined with funerals and wakes has a certain ethnocentric logic. The other major North American classification, the Library of Congress Classification (LCC), exhibits similar biases. For example, the allocation of space and the sequence of development of Class K for law, with separate volumes for individual North American and European countries, was published in the 1960s and 1970s with only one volume appearing in 1993 covering Asia, Eurasia, Africa, Pacific Area, and Antarctica. In each of these cases, there is a tendency to simply accept that these powerful discourses operate, and that change is too expensive and impractical.

IN SEARCH OF A THEORETICAL MODEL

Building a theoretical framework or model to analyze and address the biases of classification in a practical way requires a model capable of revealing the complexities of classification and the discourses that construct it. Developing this framework or model requires examination of the characteristics of classification, testing various conceptions against those characteristics, and reflexively reworking the model.

Two major characteristics of classification are that it gathers similar information together and places it in proximity to related information. If there is to be only one ordering of information, then it is useful for classification to reflect the relationships perceived in the wider society. Because the relationships between concepts can be drawn in a variety of ways, classifications will give more advantageous space in the overall structure to some concepts than to others. As the literature cited earlier suggests, classification tends to reflect the most mainstream version of these relationships. Classificatory structures are developed by the most powerful discourses in a society. The result is the marginalization of concepts outside the mainstream.

Classifications are also closed systems in that they represent some concepts and not others. No classification will ever be all inclusive. Since classifications are notationally controlled vocabularies, these inevitably have limits. Legal scholar Drucilla Cornell (1992) has suggested that any system or structure has limits, and that replacing one system with another will simply define different limits rather than being all inclusive. A system of any kind is defined by what it is not and, because systems tend to be dynamic, like classifications, the definition of what the system's limits are is always deferred (p. 2). It is an instance of Jacques Derrida's deconstructive concept of differance that limits are constructed by their exclusions and are in a state of constant flux because they are socially constructed. The question for classification then becomes, What is left beyond the limit? What is excluded? Given Foskett's observations on classificationists, it comes as no surprise that the limits of classifications are also constructed by the powerful discourses within a society, and that what is excluded is what is further from the mainstream.

Further, library classifications have responded to the needs of libraries to classify published works into a browsable collection. Therefore, what exists in published form will dictate, to a greater or lesser degree, what is included in a classification. Even a classification that does not limit itself to literary warrant will be irresponsible if it ignores the published record. Since what gets published is also limited by powerful social discourses, it too tends to produce a corpus largely representing mainstream thought.

The result of these factors is classification, which might be seen as a dense mainstream core of aptly juxtaposed concepts with marginal concepts scattered around the edges or not represented at all. This conjures up concentric circles of degrees of representation quality forming a distribution--similar to Zipf's, Lotka's, or Bradford's--of: a few core concepts best represented, a middle ground adequately represented, and a large periphery of poorly represented marginal concepts with some concepts outside of the limits not represented at all (see Figure 1). This image of Zipf's core of word occurrences or Lotka's core of published authors or Bradford's core of journals effectively documents the distribution of what currently exists. It also suggests the effectiveness of a spatial conception of classification. However, it does not provide a metaphor for analyzing the historical discourses that have shaped the present nor does it offer suggestions for changing the status quo.

[Figure 1 ILLUSTRATION OMITTED]

This distribution does, however, probably correlate with how representative mainstream views actually are--i.e., they show only what is concentrated at the perceived center and how that puts others in the margins. The core does not consist of many different entities. Putting this distribution into social terms creates a small core surrounded by margins. If one takes all so-called "special interest groups" out of the social equation, there is little left in the mainstream. In North American society, taking away women, African Americans, Hispanic Americans, French Canadians, Native peoples, Asian Americans, lesbians and gay men, people with disabilities, anyone who is not Christian, working class and poor people, and so forth, one is left with a very small "core." An image that shows the complexity of these overlapping categories is that of a huge Venn diagram with many sets limited by Boolean ANDs. The white AND male AND straight AND European AND Christian AND middle-class AND able-bodied AND Anglo mainstream becomes a very small minority (Figure 2), and each set implies what it is not. The implication of this image is that not every person, not every discourse, not every concept, has equal weight. Some discourses simply wield more power than others. Different discourses have different levels of power to construct our realities. The Venn diagram helps to represent the discourses constructing classification. However, Venn diagrams operate on the basis of dualities. Something either is or is not in the circle (fuzzy sets could help somewhat in this respect). Further, as the sets overlap, these estimate only quantity and not the shape or relative location of the dominant.

[Figure 2 ILLUSTRATION OMITTED]

MAPPING AS METAPHOR

Zipf-like distributions and Venn diagrams are descriptive representations of a phenomenon. They do not contain the potential to inform change. A more powerful device that does have this potential is metaphor. Michael A. Arbib and Mary B. Hesse (1986) suggest that:

   Metaphor is potentially revolutionary.... Scientific revolutions are, in
   fact, metaphoric revolutions, and theoretical explanation should be seen as
   metaphoric redescription of the domain of phenomena.... Metaphor causes us
   to "see" the phenomena differently and causes the meanings of terms that
   are relatively observational and literal in the original system to shift
   toward the metaphoric meaning.... Meaning is constituted by a network, and
   metaphor forces us to look at the intersections and interaction of
   different parts of the network. (p. 156)

The diagrams in Figures 1 and 2 are spatial representations but not spatial metaphors. To redescribe classification through theoretical explanation, to explore its network of interactions and intersections, requires a more complex device.

To see classification differently, spatial metaphors were chosen beginning with maps. Whether maps of the earth's surface or maps of knowledge, which we call classifications, these are socially constructed representations. Werner Bies (1996) notes that cartographical and architectural images are particularly prolific as metaphors for classification (p. 4). He suggests that the way we organize knowledge is "an essential part of the cultural memory" just as is the knowledge itself (p. 7). Therefore, analyzing the metaphor as well as the system will reveal the construction with greater clarity. Use of a metaphor is itself part of the social construction. To take this a step further, it is suggested that understanding this cultural metaphor can help us understand the construction of classification.

Mapping is not a new metaphor for classification. To look at the metaphor of mapping and the way it is used in knowledge organization, the discussion will now turn to two of the fathers of knowledge or information organization, Berwick Sayers and B. C. Brookes. In his classic work, Manual of Classification, Berwick Sayers (1926) states categorically that: "A classification scheme is really a map of knowledge.... A general classification is, then, a map of the universe within and without the mind of man; it covers all things we may have known, know or can know. In the language of metaphysics it covers all being" (pp. 65-66). Sayers is using the metaphor of the map to suggest the vastness that classification can cover. He is making a presumption that a boundless system is actually possible and that a universal classification can exist. This presumption is at odds with Drucilla Cornell's idea, described earlier--i.e., that systems are defined by their limits. The contrast between Sayers and Cornell is that of empiricist versus poststructuralist. Sayers implies that there is a single knowable reality that classification can represent in its entirety. Cornell suggests that there are multiple realities such that no system can represent reality in its totality.

In the fourth edition of his work, Sayers (1967) goes further to say that:

   [A classification's] task is to provide for the field of knowledge or part
   of it, as comprehensive and clear a statement as the cartographer is able
   to make of a territory of the earth. For just as a map makes clear the
   relationship between place and place so a classification strives to show
   the relationship of each branch of knowledge to the other branches. (p. 32)

That is, the map as a metaphor works because it is comprehensive (including everything), clear, and shows relationships. In this sense, the map works from a poststructural view as well since the map is just as constructed as the classification.

B.C. Brookes's (1980) perception, expressed in one of his articles on the foundations of information science, adds the characteristic of objectivity. Brookes suggests that a map is an objective representation of a landscape, independent of the perspective one would have in the view from a window. One "objective map accommodates all possible subjective views of the same scene" (p. 270).

Together, then, Sayers and Brookes suggest that classifications can be all-encompassing, accurate, and objective relational representations of knowledge or information--just as maps are. However, there is no reason to think that maps have any of these characteristics. Maps are no more objective or free of perspective than classifications, in spite of their basis in accurate measurement as indicated by Brookes. Maps are just as culturally bound as classifications and classificationists' reliance on cartographic imagery as being neutral and has allowed the continued existence of the illusion that classification can also be neutral. Examining the limitations of maps can help to define what questions we should ask ourselves about classification.

Maps have always determined the limits of our worlds. In 500 B.C., Hecataeus created a map of the earth as a disk representing the world known to him, with the Mediterranean (from the Latin; middle + land or earth) at the center. It is like the Zipf-like diagram in Figure 1. What is at the center is what he knew best. In addition, Hecataeus's map defines the outer limits of the earth's disk as being the ocean beyond Europe, Ethiopia, and India. Present classification schemes are somewhat like maps of the earth as a disk--i.e., limited to a cultural perspective.

A major characteristic of classification is that it is meant to place related concepts in proximity to each other. This factor suggests that classification is a spatial ordering. Certainly, thinking of its use for shelving books in a library, one can recognize its existence in what amounts to one-dimensional linear space. A book on a shelf can be to the left of one book and to the right of another but cannot sit next to a third. Maps also determine the perceived dimensions of spaces. Like the two-dimensional projections of the three-dimensional earth classifications, these must distort all knowledge in its infinite multidimensionality into a linear arrangement suitable for creating a browsable list or locations on shelves. For example, traditional map projections, like the Mercator projection, skewed the size of different parts of the world relative to each other. Therefore, the European and North American colonial powers in the temperate latitudes appear much larger than countries of the South that are closer to the equator (for further explanation and an interesting alternative, see Map of the World, n. d.), in the same way that traditional classifications have allocated more space to mainstream topics and less to marginal topics.

Maps of the same area may be differently constructed depending on cultural discourses. Maps by Native North Americans assisting European explorers included rivers, mountains, and other physical features relative to each other and to the settlements of different tribes (see, for example, three maps on plate 59 in the Historical Atlas of Canada, 1987). These maps did not divide up the entire space into discrete units like euro-settler maps. The latter tradition, which we still follow, divides the "pie" into separate pieces that take up the whole space. There is no common space left. Each piece of the terrain "belongs" to someone, reflecting a particular cultural concept of property. Every inch is part of a jurisdiction. In looking at classification as a sort of mapping, we see how similar discrete domains with boundaries are created. Each concept is limited by its definition, and the definition is the boundary of what is or is not a given concept. The definition marks the territory. Since the different territories or categories in this type of map are ideally mutually exclusive, definitions that criss-cross each other cause problems that need to be dealt with by breaking them into facets and creating hierarchical arrangements--i.e., cities within states, states within countries, and countries within continents. However, going back to the contrast between Native American maps and conventional political maps, it is apparent that even this approach of mapping by dividing up territories is not a universal concept--it is culturally bound.

CONSTRUCTING PARADOXICAL SPACES

This idea of the social construction or mapping of information is very interesting in and of itself, but of what use is it? We know that classifications are not perfect, and that they reflect social biases. Can spatial metaphors of mapping help address these biases? Can these be theoretically revolutionary as Arbib and Hesse (1986) suggest above?

To test the power of the metaphor, an explanation of the development of a theoretical reason drawn from spatial constructions of a range of poststructural and feminist theorists will be given. The specific project that has been a vehicle for evolving these ideas is an effort to map the terminology of a marginalized knowledge domain to a mainstream classification, the Dewey Decimal Classification. The project takes the terms of a widely used feminist vocabulary, A Women's Thesaurus (Capek, 1987), and links these to numbers in DDC. This approach is a way of linking the margins and the center to create a sort of network or web instead of concentric circles with no overlaps. According to Arbib and Hesse (1986), the network creates the meaning. This project constructs a network of intersections different from those in the original DDC so that it creates meaning differently. To enable this network of links, Dennis Ward, a colleague in the School of Library and Information Studies, University of Alberta, is developing Windows and World Wide Web interfaces to reflect the theoretical framework.

DDC is a good representation of mainstream thought. It has not been left to the peculiarities of its nineteenth century origins. For example, it is a long time since it located the status of women in 396 between etiquette and outcast races as it did in earlier editions. However, it does still show the basic structure it inherited from Melvil Dewey's milieu. It is not limited by literary warrant, but revisions are often based on how literature is used in the disciplines represented in DDC such as recent revisions in public administration and the life sciences. Therefore, DDC continues to represent mainstream arrangements effectively and is constructed by various mainstream voices and not just one dominant discourse.

In comparison, the concepts represented in A Women's Thesaurus and the relationships between those concepts cross different disciplines, setting up an alternative structure that is fundamentally different. It is a structure developed in the margins--i.e., in the marginalized knowledge domain of women's studies and feminist thought.

By linking these two modes of representation, some might consider this as trying to fit round pegs into square holes or comparing apples and oranges or some other similar metaphor. However, continuing with these spatial metaphors, it is suggested that what is actually being done is constructing paradoxical spaces. Paradoxical space is a concept developed by feminist geographer Gillian Rose (1993). It is simply a practice that allows existence on both sides of a limit simultaneously or alternately. It is both inside and outside, center and margins. In this way, it does not put a new structure in place of the old but puts a different spin on existing concepts that come to coexist with concepts from the margins (p. 140).

An everyday example of paradoxical space is the concept of separate spheres, public and private, associated with men's roles and women's roles. The private sphere represented by the white bourgeois concept of "home" and the public sphere represented by the paid workplace have been sites of paradoxical space in a variety of ways (Rose, 1993, pp. 52-56; Haraway, 1991, p. 170). Female-intensive professions like nursing, home economics, teaching and, of course, librarianship brought the ethic of care from the private women's sphere to the public male sphere in the nineteenth century. Later in this century, the necessary revisions of white middle-class feminism came to recognize that the private sphere, the home, is a workplace not only for the women who live there, but also for the women who leave their own homes to work in the homes of others. These women are mostly women of color who bring a very different perspective to the idea of the private sphere as a place for women's work. Recognition of the widespread existence of wife abuse also upsets the idea of the private sphere as the place where women are in control. Technology is now reviving the old cottage model of exploitation in the home. The electronic cottage and telecommuting bring the public sphere and its values into the private sphere (see, for example, Fulton, 1997). These examples of the fuzzy boundaries between public and private make both into paradoxical space. It is no longer possible to define the limits between public and private. "Home" is not a simple concept--it never was except in our naive constructions of it. However, we can still understand concepts like "home" because paradoxical spaces can exist.

QUALITY OF CONSTRUCTED SPACE

We can also purposely create paradoxical spaces. In this project, I have worked with research assistants to link the concepts from A Women's Thesaurus to DDC, creating paradoxical space. As we began, the idea worked reasonably well and seemed to have potential but, as we progressed, it became apparent that some concepts mapped to positions qualitatively better than others. In seeking some way to analyze the qualities of the links, what was first considered was their coextensiveness. Coextensiveness is considered here in spatial terms: the "shape" of the topic and the "shape" of the representation are the same, or, as Jessica Milstead (1984) puts it, coextensiveness is "the extent to which the index term reflects the precise content of the item of information ..." (p. 143). Milstead suggests the limitations of coextensiveness for classification when she opposes the predetermined pigeonholes of classification to the potential coextensiveness of thesauri. Most classification constructs pigeonholes, which are preformed without reference to the subjects of particular documents. Therefore, documents are put into the pigeonhole "closest in size to the subject" (p. 144). Coextensiveness is based on the subjects of individual documents while pigeonholing is based on the structure of the system.

Coextensiveness became a useful measure for this project in a sort of Goldilocks-and-the-three-bears way. Some matches between feminist topics and DDC numbers were too broad, others were too narrow, and others were just right. Of course the analogy did not hold up because some were overlapping in an associative, rather than a hierarchical, manner and other concepts simply had no number to represent them at all. The coextensiveness problems were especially acute in the treatment of topics from the gendered perspective implied by a feminist thesaurus that required our assessment of coextensiveness to be split into general and gendered forms (the basic approach of the research and the variables are discussed by Olson & Ward, 1997a, 1997b).

However, even more problematic, and far more theoretically interesting, were problems of gathering and proximity. Classification gathers works on a particular topic or group of topics and places them in close proximity to related topics. What became interesting as we progressed with mapping A Women's Thesaurus to DDC was that the gathering and proximity sometimes created odd, and even unfriendly, environments. To address this idea, feminist philosopher Lorraine Code's (1995) concept of rhetorical space was used:

   Rhetorical spaces ... are fictive but not fanciful or fixed locations,
   whose (tacit, rarely spoken) territorial imperatives structure and limit
   the kinds of utterances that can be voiced within them with a reasonable
   expectation of uptake and "choral support": an expectation of being heard,
   understood, taken seriously. They are the sites where the very possibility
   of an utterance counting as "true-or-false" or of a discussion yielding
   insight is made manifest. Some simple examples will indicate what I mean
   the term to achieve.... Imagine trying to make a true statement about
   whether it is more convenient to fly into Newark or La Guardia airport in
   the year 1600. The statement would not be false but meaningless: it could
   neither be true nor false within the available discursive possibilities. Or
   imagine trying to have a productive public debate about abortion in the
   Vatican in 1995, where there is no available rhetorical space, not because
   the actual speech acts involved would be overtly prohibited, but because
   the available rhetorical space is not one where ideas on such a topic can
   be heard and debated openly, responsively.... what I want this terminology
   [rhetorical space] to do [is], namely to deflect the focus of philosophical
   analysis away from single and presumably self-contained propositional
   utterances pronounced by no one in particular and as though into a neutral
   space; and to move it into textured locations where it matters who is
   speaking and where and why, and where such mattering bears directly upon
   the possibility of knowledge claims, moral pronouncements, descriptions of
   "reality" achieving acknowledgment, going through. Often in such spaces
   discourse becomes a poiesis, a way of representing experience, reality,
   that remakes and alters it in the process. And the making is ordinarily a
   communal process, dependent for its continuance on receptive conditions, on
   engaged responses both favourable and critical. (p. x)

Code proposes a new concept in spatial imagery that helps explain what goes wrong for marginalized topics in classification. Even more than coextensiveness, rhetorical space helps explain why the mapping of classification supports mainstream biases. To demonstrate, key phrases will be taken from Code's defining quotation and will be elaborated on in relation to classification and in light of other feminist and poststructuralist theorists' work.

1. "Rhetorical spaces ... are fictive but not fanciful or fixed...." That is, they are constructed, made-up (fictive), but not arbitrary (not fanciful) and dynamic (not fixed). This is also true of classifications. Classifications are not innate or natural but are constructed. Their construction has some logical basis, and they change as discourses continue to act upon them.

2. "[T]erritorial imperatives structure and limit ...". The spaces have boundaries. They are limited by the way they are constructed and by the imperatives of the discourses that construct them. Again they are like classifications which are limited systems, including some concepts and excluding others. As Cornell suggests, limits define systems and spaces.

3. "[B]eing heard, understood, taken seriously...". What is limited when positive rhetorical space is lacking is voice. Voice is a given for mainstream discourses, but for marginalized discourses it is something more vital. As bell hooks (1989) puts it:

   Moving from silence into speech is for the oppressed, the colonized, the
   exploited, and those who stand and struggle side by side ... a gesture of
   defiance that heals, that makes new life and new growth possible. It is
   that act of speech, of "talking back," that is no mere gesture of empty
   words, that is the expression of our movement from object to subject "the
   liberated voice." (p. 9)

So a positive rhetorical space allows marginalized discourses to be heard as legitimate statements--i.e., to be acknowledged as worth listening to. Cornell (1992) proposes a responsibility for those who control a system to make its limits permeable so that they can approach an ethical relationship with those who are excluded (p. 62). The permeability of the limits allows the voices of the excluded--the oppressed, the colonized, the exploited--to be heard in the system. Those of us with authority must constantly throw the system off balance to maintain this permeability (p. 80).

4. Code intends the idea of rhetorical space "to deflect the focus of philosophical analysis from single and presumably self-contained utterances pronounced by no one in particular as though into a neutral space..." (p. 10). This apparently neutral space is like postmodern theorist Henri Lefebvre's (1991) "transparent space" which denies the existence of anything excluded from its mapping because it appears to be all there is. "Transparent space" is the illusion that location is neutral--i.e., that mapping territory can be a true representation of some essence of reality as B.C. Brookes (1980) suggests. However, there is an implied exclusion defining the transparent space that is hidden by it. Because space has boundaries and always includes and excludes something, it cannot be neutral. Making the exclusions visible means identifying the space's boundaries to allow recognition of what is outside those boundaries. It is identifying the implied opposite of a deconstructive binary opposition. It shows the constructed nature of the space. It moves the discourse "into textured locations where it matters who is speaking and where and why, and where such mattering bears directly upon the possibility of knowledge claims" (p. x). Transparent space is difficult to discern because of its apparent neutrality. The efforts at neutrality that are made in classification can mask exclusions.

5. "[P]oiesis, a way of representing experience, reality, that remakes and alters it in the process." Poiesis is a creative and creating production. In poiesis, we understand that the representation of reality is the construction of reality. The representation of information, through classification, is part of the construction of information. Classification remakes and alters information by constructing a particular context for it--gathering, scattering, and juxtaposing topics in relation to each other. How broadly or narrowly topics are represented will enhance or mask their visibility. In these ways, classification produces information in a creative process. This process of poiesis is a locational one. Feminist sociologist Elspeth Probyn (1990) proposes that:

   Through location knowledges are ordered into sequences which are congruent
   with previously established categories of knowledge. Location, then,
   delineates what we may hold as knowable and, following Foucault, renders
   certain experiences "true" and "scientific" while excluding others. (p.
   178)

She continues on to point out that this act of creation or construction determines not only what is knowable, but whose voices are heard. So the creation of classification creates the space in which some knowledges are central and others are peripheral.

6. This creation of reality is "a communal process, dependent for its continuance on receptive conditions, on engaged responses both favourable and critical." That is, both the context and the process affect the construction of reality. It could be called reflexive or holistic. It is akin to the death of the author and the ascendancy of the reader in literary criticism. The author does not create the text. It is created in the process of reading and depends upon the "receptive" and "engaged" reader for its meaning and existence. It involves interpretation, in the case of classification, by classifiers and users. In this sense it places responsibility for the construction of information not just on classificationists who write classifications but also on the individuals and institutions who use classifications.

OPERATIONALIZING THE THEORY

To operationalize the concept of rhetorical space, our research project examines the DDC context of individual concepts from A Women's Thesaurus by looking at:

1. What other topics share the number?

2. How is the number described?

3. What is the hierarchical context?

4. What topics sit on either side?

Two examples from our pilot study illustrate how the variable of rhetorical space works to reveal whether or not feminist topics can be taken seriously in DDC.

The first topic is colonialism. The following entry from A Women Thesaurus implies the scope of this term:

Entry from A Women's Thesaurus:

   colonialism

   UF   imperialism
   NT   neocolonialism
   RT   apartheid
   cultural imperialism
   decolonization
   developing nations
   ...

Colonialism appears in the index to DDC and points unequivocally to the number 325.3. This concept is not excluded from DDC. However, its rhetorical space is not as neutral as it at first appears. The following entry from DDC shows how 325.3 is defined and what it includes:

Entry from DDC:(2)

   325.3  Colonization
          Class here exercise of political dominion over distant
          territories

   DDC Index Terms:
      Colonialism
      Colonization

The location of colonialism with colonization is an example of transparent space. It seems neutral but is actually one-sided, showing colonization from the point of view of the colonizing power as opposed to the people and culture being colonized. Colonialism is linked to this number not in its caption but only as a reference from the DDC relative index. The entry describes the colonized territories as "distant" from the colonizing powers, not the other way around. Colonies are not distant in the view of colonized people.

The following summary shows the hierarchical and sequential contexts of 325.3 that reinforce the perspective of the imperial power observing the colony:

Summary from DDC:
   300   Social sciences
   320   Political science (Politics and government)
   325   International migration and colonization

      325.1         Immigration
      325.2         Emigration
      325.3         Colonization
      325.4-325.9   International migration to and colonization in
  specific continents, countries, localities in modern
  world

In categories 325.4-325.9, the geographic subdivisions that combine "migration" and "colonization" move from colonizing country "to" colony in its subdivisions and define colonization as "in" the colonized locale and not "by" the colonizers. As with the description of 325.3, the geographic subdivision is entirely from the perspective of the imperial power. Colonization involves movement--being colonized does not. The location of colonialism in the established category of colonization fulfills Probyn's (1990) prediction that the voices of hooks' colonized become peripheral.

A further indication of this marginalization is found by looking higher up the classificatory hierarchy. The DDC principle of hierarchical force(3) dictates that what is true for 325 is also true for its subdivisions including 325.3:

Entry from DDC:

   325  International migration and colonization
        Including involuntary population transfer, population
        exchange
        Class movement of people associated with a specific
        event in history with the event in 909 or 930-990;
        class interdisciplinary works on international movement
        of people in 304.82

While colonialism involves people from the imperial power going to the colonized territory, its disruptive nature often causes other movements of people--most notably the scattering of colonized peoples that form diaspora. For people moving away from the site of colonization as part of the diasporas that often result from colonialism, the instructions under category 325, International migration and colonization, suggest two options. First, movement of people associated with a specific event in history is to be classed with the event in the 900s. This option has at least two problems for representing postcolonial diasporas: first, it is difficult to pin down colonization as a specific event because it tended to happen over a diffuse period of time not conducive to classification and, second, by putting these movements into history, we would take them out of the present day where their results, typically including racism, must be addressed.

The second option is to use the number for interdisciplinary works on international movement of people in 304.82:

Entry from DDC:

   304.82  International movement
           Class international emigration in 304.809; class
           international immigration in 304.83-304.89

The number 304.82 is a subdivision of movement of people under the broader concept of social behaviors, thus taking this issue out of the political realm of colonialism. It does allow for geographic subdivision in either direction--the country to which people went or the country which they left. The latter applies to the postcolonial diaspora, but it uses the less preferred number. Preference will be given, according to DDC's rule of zero,(4) to international immigration rather than emigration. Therefore, people leaving India for other places in general will be classified with movement from India, but people leaving India for specific destinations will be classified with movement toward each of those places (such as movement toward England) thus diasporizing the diaspora.

These details about the siting of colonialism in DDC help to reveal that the apparently neutral transparent space is actually skewed toward a mainstream interpretation because it has been constructed by a mainstream discourse.

A second example is the poor representation in mainstream schemes of the concept of unpaid labor. The heading in A Women's Thesaurus is:

Entry from A Women's Thesaurus:

   unpaid employment

   UF  nonwage labor
   BT  employment
   NT  unpaid household labor
   RT  economic value of women's work
   homemaking
   unpaid labor force
   valuing children
   volunteer work

The term unpaid employment in A Women's Thesaurus is a subdivision of employment. However, when we try to classify it in the same way in DDC, we find that the relative index sends us to 331.125, Labor actively employed:

Entry from DDC:

   331.125  Labor actively employed
            That portion of the total available supply of labor
            employed at any given time
            Including types of employment
            Class here utilization of human resources, employment,
            comprehensive works on employment and compensation
            DDC Index Terms:
              Employment
              Human resources--utilization--economics
              Occupations--active employment

The "class here" note indicates that the number includes "comprehensive works on employment and compensation" implying only paid labor. This implication is confirmed by hierarchical force since 331.125 is a subdivision of 331.12:

Entry from DDC:

   331.12  Labor market
           The activities of and opportunities for buying and
           Selling labor

The phrase "buying and selling" confirms that only paid labor is included. The same is true if we look at the labor force:

Entry from DDC:

   331.11  Labor force
           All who are employed or available for employment
           Class here human resources, manpower and womanpower,
           labor supply, size of labor force

The scope note indicates that this definition of labor force includes "all who are employed or available for employment" which is not likely meant to include people who are available for becoming housewives and househusbands. If the latter are part of the labor force under this definition, it is more likely as people available for paid employment.

Some conventional types of unpaid employment are represented in other places in the classification with the result that they are not treated as labor. Homemaking, one of the related terms to unpaid employment in A Women's Thesaurus, is located in home economics, which is a subdivision of technology:

Entry from DDC:

   640  Home economics and family living
        Class here management of home and personal life,
        domestic arts and sciences
        DDC Index Terms:
          Domestic arts
          Domestic sciences
          Home economics
          Homemaking
          Homes
          Homes--home economics
          Household management

Summary from DDC:

   600      Technology (Applied sciences)
   640      Home economics and family living
   640.4    Specific aspects of household management
   640.41   Helpful hints and miscellaneous recipes
   640.42   Management of money
   640.43   Management of time
   640.46   Household employees
   640.49   Survival housekeeping

While this is an excellent place to put information about the processes and production of homemaking, acknowledging it as appropriate to be adjacent to agriculture or engineering, it does not include the aspect of the people who do this labor except in a subdivision for household employees, the people who are paid to do housework.

Another example of unpaid labor is voluntarism, which is defined as one aspect of social participation along with encounter groups and sensitivity training:

Entry from DDC:

   302.14  Social participation
           Including communalism, competition, cooperation,
           encounter groups, sensitivity training, voluntarism

While anyone who has done volunteer work may appreciate its links to encounters and sensitivity, this location treats voluntarism as "social participation" and not as labor. While it is important not to belittle the value of social participation in a world in which governments are cutting back essential social services which then fall to voluntary agencies, the labor of voluntarism also becomes a key to economic well-being.

A way of addressing this problem is to create a paradoxical space by locating the general concept of unpaid employment alongside paid employment. While these are two different topics, they may sit adjacent to each other to create viable rhetorical space and give legitimate voice to the unpaid labor performed everyday around the world. One option would be to tuck it into other generalizations of the labor force, say in the gap between "qualifications and personal characteristics" and "systems of labor":

Summary from DDC with potential addition:

   300              Social sciences
   330              Economics
   331              Labor economics
   331.1            Labor force and market
   331.11           Labor force
   331.110*           < Zero Subdivisions >
   331[.111]          Geographic distribution
   331.114            Qualifications and personal characteristics
   [right arrow]      Economic basis of labor
   [right arrow]      Unpaid labor
   [right arrow]      Paid labor
   331.117            Systems of labor
   331.1172             Free labor
   331.1173             Compulsory labor
   331.118            Labor productivity
   331.119            Labor force by industry and occupation

Here a section with a title something like "economic basis of labor" with the subdivisions "unpaid labor" and "paid labor" (with the mainstream interpretation coming second to upset the hierarchy a bit) could set this topic alongside basic concepts of the labor force. Because this number would be hierarchically encompassed by "labor force," the definition of the latter would have to be adjusted to include unpaid labor as would 331.1, Labor economics, in a reversal of hierarchical force--the subdivision driving the definition of the dominant concept. Mapping a marginal concept in the midst of a mainstream concept will not alone create positive rhetorical space. However, careful placement of such interpolations will make transparent space visible and will create paradoxical spaces where discussion of issues can continue openly.

CONCLUSION

Following Drucilla Cornell's concept of systems' limits, there would be no point in abandoning our existing classifications in the hope of achieving that objective map described by B.C. Brookes. All systems will exclude and marginalize in some way. However, it is possible to shift between mainstream and margin in our mapping, creating paradoxical spaces and defining the limits differently. With a new theoretical framework, it is possible to make changes in our mapping akin to the changes made when Pythagoras determined that the earth is round and not flat.

The new theoretical framework developed throughout this article is offered as a new way of mapping knowledge in classification. It has potential for both analysis and amelioration. The categories of classification--because they typically reflect a cultural mainstream--appear neutral, objective, and transparent. This makes marginalizations and exclusions difficult to identify. Therefore, to analyze the problems of classification in relation to marginalized knowledge domains, the framework poses three assumptions drawn from feminist and poststructural literature examined in this discussion. First, classification, like any map, is constructed by dominant cultural discourses. Second, classification, like any system, has constructed boundaries or limits that result in exclusions. Third, the construction of classification is a form of location that defines and sequences what is accepted as knowledge, thus marginalizing as well as excluding. Regarding classification as a text and reading it with these three assumptions in mind will make what was transparent and invisible opaque and visible, elucidating the biases and the discourses that construct and enforce them.

To ameliorate the biases of classification, this framework proposes that the limits of a classification be made unstable and permeable to allow the voices of those who have been excluded to be heard. In this way, the classification approaches an ethical relationship with previously silenced voices. Further, to address the marginalizations within classification, this theoretical stance advocates the creation of paradoxical spaces that are neither mainstream nor marginal but are both simultaneously or alternately. By mapping A Women's Thesaurus to DDC, this project creates such spaces. The same concept can offer other ways of deconstructing and reconstructing not only the limits but also the structures of classification.

The creation of paradoxical spaces can become a poiesis--i.e., that alters representation in ways that make boundaries permeable. In the next stage of this project, suggestions will be made for revision, supplements will be devised, and optional practices offered to further develop paradoxical spaces for women's studies and feminist thought in DDC. It will allow for more dimensions and, thus, more creative connections between places/spaces/concepts than have hitherto been available. Further, we hope that it will be a prototype for a poiesis applicable to other marginalized knowledge domains. With care, paradoxical spaces will appear throughout classifications, thereby keeping them from stagnating and keeping them vital and exciting.

NOTES

(1) See, for example, Afolabi (1992) on Africana; Hamdy (1980) on Arabic materials; Iwuji (1989) on Africa; Lochhead (1985) on women; McConnell (1984, 1985a, 1985b) on Melanesia; Milstead Harris & Clack (1979) on nearly all of these groups; Mowery (1989) on Mexican Americans; Pacey (1989) on Africa; Steinberg (1974) on women; and Wolf (1972) on gays and lesbians.

(2) Entries and summaries from DDC are from the electronic version, Dewey for Windows. They have been edited in format. The relative index terms are included only when they are relevant to the discussion and related Library of Congress Subject Headings have been omitted. Other omissions are shown by ellipses. Emphasis is mine to facilitate interpretation.

(3) "Hierarchical force (DDC Glossary). The principle that the attributes of a class as defined in the heading and in certain basic notes apply to all the subdivisions of the class, and to all other classes to which reference is made" (Dewey for Windows).

(4) "Rule of zero (DDC Glossary). The rule instructing that subdivisions beginning with zero should be avoided if there is a choice between 0 and subdivisions beginning with 1-9 in the same position in the notation. Similarly, subdivisions beginning with 00 should be avoided when there is a choice between 00 and 0" (Dewey for Windows, 1996).

REFERENCES

Afolabi, M. (1992). Spiritual matters: Provision for independent African churches in general classification schemes. International Classification, 19(4), 210-213.

Arbib, M.A., & Hesse, M.B. (1986). The construction of reality. Cambridge, MA: Cambridge University Press.

Bies, W. (1996). Thinking with the help of images: On the metaphors of knowledge organization. Knowledge Organization, 23, 3-8.

Brookes, B. C. (1980). The foundations of information science. Part III: Quantitative aspects: Objective maps and subjective landscapes. Journal of Information Science, 2, 269-275.

Capek, M. E. S. (Ed.). (1987). A women's thesaurus. New York: Harper & Row.

Code, L. (1995). Rhetorical spaces: Essays on gendered locations. New York: Routledge.

Cornell, D. (1992). The philosophy of the limit. New York: Routledge. Dewey for Windows. (1996). (Version 1.0). Dublin, OH: OCLC Press.

Foskett, A. C. (1971). Mysogynists all: A study of critical classification. Library Resources & Technical Services, 15, 117-121.

Fulton, C. (1997). The potential pitfall in telework: Control and deskilling issues for work done at home. In B. Frohmann (Ed.), Communication and information in context: Society, technology, and the professions: Canadian Association for Information Science (Proceedings of the 25th Annual Conference/Association canadienne des sciences de l'information, Travaux du 25e congres annuel) (pp. 44-55). Toronto, Ontario, Canada: Canadian Association for Information Science.

Hamdy, N. (1980). Cataloging and classifying Arabic materials. In M. M. Aman (Ed.), Cataloging and classification of non-western material: Concerns, issues and practices (pp. 268-296). Phoenix, AZ: Oryx Press.

Haraway, D.J. (1991). Simians, cyborgs, and women: The reinvention of nature. New York: Routledge.

Historical Atlas of Canada, Vol. I: From the Beginning to 1800. (1987). R.C. Harris (Ed.), G.J. Matthews (Cartographer/designer). Toronto, Ontario, Canada: University of Toronto Press.

hooks, b. (1989). Talking back: Thinking feminist, thinking black. Toronto, Ontario, Canada: Between the Lines.

Iwuji, H. O. M. (1989). Africana in LC and DD classification schemes: A need for an Africana scheme? Journal of Librarianship, 21(1), 1-18.

Lefebvre, H. (1991). The production of space (D. Nicholson-Smith, Trans.). Oxford, England: Basil Blackwell.

Lochhead, I. (1985). Bibliographic control of feminist literature. Catalogue & Index, 76/ 77, 10-15.

Map of the World: Peters Projection. (n. d.). Oxford, England: New Internationalist.

McConnell, F. (1984). Classification of Melanesia: Proposals for revision of DDC, Table 2. Cataloging & Classification Quarterly, 5(1), 53-60.

McConnell, F. (1985a). Languages of Melanesia: Problems and proposals for classification. Cataloging & Classification Quarterly, 5(3), 57-66.

McConnell, F. (1985b). Peoples of Melanesia: Proposals for revision of DDC19, Table 5. Cataloging & Classification Quarterly, 5(4), 47-51.

Milstead, J. L. (1984). Subject access systems: Alternatives in design. Orlando, FL: Academic Press.

Milstead Harris, J. L., & Clack, D. H. (1979). Treatment of people and peoples in subject analysis. Library Resources & Technical Services, 23, 374-390.

Mowery, R. L. (1995). Subject cataloging of Chicano literature. Library Resources & Technical Services, 39, 229-237.

Olson, H. A., & Ward, D. B. (1997). Feminist locales in Dewey's landscape: Mapping a marginalized knowledge domain. In Knowledge organization for information retrieval (Proceedings of the Sixth International Study Conference on Classification Research) (pp. 129-133). The Hague: International Federation for Information and Documentation.

Olson, H. A., & Ward, D. B. (1997a). Ghettoes and diaspora in classification: Communicating across the limits. In B. Frohmann (Ed.), Communication and information in context: Society, technology, and the professions (Proceedings of the 25th Annual Conference/ Association canadienne des sciences de l'information: Travaux du 25e congres annuel) (pp. 19-31). Toronto, Ontario, Canada: Canadian Association for Information Science.

Pacey, P. (1989). The classification of literature in the Dewey Decimal Classification: The primacy of language and the taint of colonialism. Cataloging & Classification Quarterly, 9, 101-107.

Probyn, E. (1990). Travels in the postmodern: Making sense of the local. In L.J. Nicholson (Ed.), Feminism/postmodernism (pp. 176-189). New York: Routledge.

Rose, G. (1993). Feminism & geography: The limits of geographical knowledge. Minneapolis, MN: University of Minnesota Press.

Sayers, W. C. B. (1926). A manual of classification for librarians & bibliographers. London, England: Grafton.

Sayers, W. C. B. (1967). A manual of classification for librarians (4th rev. ed. by A. Maltby). London, England: Andre Deutsch.

Steinberg, F. (1974). Classified women. Librarians for Social Change, 5, 6-7.

Studwell, W. E.; Wu, H.; & Wang, R. (1994). Ideological influences on book classification schemes in the People's Republic of China. Cataloging & Classification Quarterly, 19(1), 61-74.

Sukiasian, E. (1993). Knowledge organization in Russia: Problem-related and historical aspects. Knowledge Organization, 20, 189-191.

Wolf, S. (1972). Sex and the single cataloger: New thoughts on some unthinkable subjects. In C. West (Ed.), Revolting librarians (pp. 39-44). San Francisco, CA: Booklegger Press.

Hope A. Olson, School of Library & Information Studies, 3-20 Rutherford South, University of Alberta, Edmonton, Alberta, CANADA T6G 2J4

HOPE A. OLSON is Associate Professor and Graduate Coordinator at the School of Library and Information Studies, University of Alberta, where she teaches organization of information including cataloging and classification, feminism, and library and information studies and globalization and diversity issues. Ms. Olson's research is on subject access in library catalogs and uses post-structural and other critical theoretical stances. She has published articles on classification and subject headings, on deconstruction as a methodology, and on teaching cataloging.