Searching for Swedish LGBTQI fiction: the librarians' perspective

Abstract

Purpose

This article aims to help ensure high-quality subject access to Swedish lesbian, gay, bisexual, transgender, queer and intersexual (LGBTQI) fiction, and aims to identify challenges that librarians consider important to address, on behalf of themselves and end users.

Design/methodology/approach

A web-based questionnaire comprising 35 closed and open questions, 22 of which were required, was sent via online channels in January 2022. By the survey closing date, 20 March 2022, 82 responses had been received. The study was intended to complement an earlier study targeting end users.

Findings

Both this study of librarians and the previous study of end users have painted a dismal image of online search services when it comes to searching for LGBTQI fiction. The need to consult different channels (e.g. social media, library catalogues and friends), the inability to search more specifically than for the broad LGBTQI category and suboptimal search interfaces were among the commonly reported issues. The results of these studies are used to inform the development of a dedicated Swedish LGBTQI fiction database with an online search interface.

Originality/value

The subject searching of fiction via online services is usually limited to genre with facets for time and place, while users are often seeking characteristics such as pacing, characterization, storyline, frame/setting, tone and language/style. LGBTQI fiction is even more challenging to search because indexing practices are not really being standardized or disseminated worldwide. This study helps address this important gap, in both research and practical applications.

Citation

Golub, K., Bergenmar, J. and Humelsjö, S. (2023), "Searching for Swedish LGBTQI fiction: the librarians' perspective", Journal of Documentation, Vol. 79 No. 7, pp. 261-279. https://doi.org/10.1108/JD-05-2023-0080

Introduction

When searching for literary fiction, users are looking for plots, characters or themes. However, the professional subject indexing of literary fiction in libraries has been limited to genre while themes have rarely been indexed. At the same time, automatic indexing and retrieval methods are not as successful for literary works as they are for the exact sciences. Finding lesbian, gay, bisexual, transgender, queer and intersexual (LGBTQI) fiction has proved to be even more difficult, as these works tend to be described using overly broad subject index terms or are not being indexed according to LGBTQI themes at all (Author, Year).

That LGBTQI fiction is an important medium for fostering knowledge and self-awareness has been noted by scholars in different contexts: in queer literature studies (e.g. Love, 2007; Wallace, 2016), in literacy and children's and young adult literature studies (Banks, 2009; Smolkin and Yong, 2011) and in library and information studies. Literature may be part of meaning-making processes through which people seek information needed to understand their lives (Ruthven, 2019; Huttunen and Kortelainen, 2021). Access to fictional literature representing LGBTQI experiences supports identity formation and the communication of sexual identities, particularly during early stages of understanding one's own sexual identity (Liming, 2007; Rothbauer, 2013).

To increase the findability of LGBTQI fiction, rather than leave it to serendipity (cf. Pohjanen and Kortelainen, 2016), this study explores librarians' perspectives on LGBTQI fiction searching and indexing and complements findings of an earlier study of end users (Author, Year). Both studies inform the design of a Swedish LGBTQI fiction database. Librarians' perspectives were investigated via a web-based questionnaire distributed in 2022 and completed by 82 participants.

The remainder of the paper is structured as follows. “Background” presents the broader context of the article and covers related research. “Methodology” describes the methods of data collection, sampling and analysis. Then, in “Results,” the results are reported and discussed together with their main implications for the development of the database, search functionalities and thesaurus. Pointers for future research and final remarks are given in “Concluding remarks.”

Background

LGBTQI fiction

Queer literature scholars have emphasized that fiction is part of the collective history of LGBTQI people, serving as an important medium of knowledge and self-awareness (e.g. Love, 2007; Heede, 2015; Wallace, 2016). However, end users consider it challenging to find LGBTQI fiction in online information systems such as library catalogues, online bookshops and social media (Author, Year). Even research overviews of LGBTQI fiction are few, and in many cases limited to specific genres, such as young adult literature (Cart and Jenkins, 2006; Abate and Kidd, 2011) or primarily gay fiction (e.g. Koponen, 1993; White, 1999), lesbian fiction (Faderman, 1981; Griffin, 1993) or both (Stevens, 2011; Herring, 2015). It is only in recent years that scholars have started to outline transgender and intersex literary canons (e.g. Amato, 2016; Chess, 2016; Koch, 2017), although particular themes had been addressed earlier, for example, in the 1990s (e.g. Garber, 1992; Lehnert, 1994, 1997).

From these research overviews we learn that the LGBTQI community has been unevenly represented in fiction, with gay male experiences clearly overrepresented (Cart and Jenkins, 2006; Hicks and Kerrigan, 2020). Transgender characters appear less often and in less central roles than do LGB characters, and their portrayals tend to be less positive (Waite, 2013; Cart and Jenkins, 2018). We have also learned that LGBTQI themes have been more hidden than explicit in fictional literature, at least until the 1990s, when these themes started to become more visible (Cart and Jenkins, 2006). This last matter has also had an adverse impact on subject indexing and, consequently, on finding LGBTQI fiction.

Indexing LGBTQI fiction

While the information profession community has advocated enhancing fiction indexing to capture appeal characteristics such as pacing, characterization, storyline, frame/setting, tone and language/style (Saricks and Wyatt, 2019), the subject indexing practices of fiction in libraries are usually limited to genre, complemented with facets for time and place (Saarti, 1999). Another challenge arises when it comes to “meaning” and “aboutness,” with the former signifying subjective interpretation and the latter an “intrinsic element of a document's intellectual content” (Campbell, 2013, p. 295), which is especially difficult to index when themes are intentionally hidden or when, for example, “homosexuality often appears through implication” (Campbell, 2013, p. 295), as was particularly common in fiction before the 1990s.

Further problems specific to LGBTQI themes pertain to bias and changing terminology. There is a long history of critique of the failure of controlled vocabularies to “respectfully organize library materials about social groups and identities that lack social and political power,” as Drabinski (2013, p. 95) noted, and librarians and scholars have sought to correct and expand subject headings and to correct biases. Edge (2019) identified another problem of providing adequate terms for LGBTQI actions and identities, which is related to changing LGBTQI terminology and the ambiguous use of certain terms such as “gay.” Edge (2019) also mentioned the challenge posed by the term “queer,” signifying fluid identities and resistance to definition, supporting Keilty's (2009) proposition that a queer indexing practice must allow for changeability and rely on an understanding rooted in the community.

Another problem is that general controlled vocabularies are often limited to only rather broad LGBTQI terms. This is the case with the general subject heading system in Sweden, that is Swedish Subject Headings (SAO), in which, for example, only the general term “transgendered people” exists, while more specific ones such as “trans men,” “trans women,” and “transitioning” are absent. At the same time, existing controlled vocabularies for LGBTQI information resources are mostly applied to indexing nonfiction (see, e.g., 15 examples of controlled vocabularies for LGBTQ + materials compiled by Johnson and Colbert, 2007/2017).

Alternative or complementary approaches to the professional subject indexing of fiction have in recent decades mostly been found in social tagging services, while automatic options remain largely unsuccessful in texts characterized by subjectivity, metaphoricity and hidden and subtle motifs. Despite the recognized disadvantages of social tagging such as lack of indexing rules, different users using different tags to refer to the same concept, homonymous tags not being disambiguated, hierarchical and other relationships between tags often lacking, tags written in different forms (e.g. singular/plural and spelling variations) and tags unlimited in quantity or only having relevance for personal use (e.g. “to read”) (see, e.g., Furner, 2010), social tagging is at the same time characterized by the natural everyday language of the end user. Thus, in her review of tagging literature, Rafferty (2018) concluded that while tagging may underperform established subject indexing systems, it can still “complement, enrich, and … enhance conventional retrieval systems” (p. 510), a finding also reached by Rolla (2009), Kipp and Campbell (2010), Golub et al. (2014) and Adler (2009) for transgender books and Bates and Rowley (2011) for LGBTQI works.

QUEERLIT project

In an attempt to address the challenges that LGBTQI literary fiction is in general very sparingly thematically described and that the SAO (Swedish subject headings) LGBTQI terms are broad and lacking in acuity, the Swedish Queerlit Project, 2021–2023, creates a Swedish LGBTQI fiction sub-database of the Swedish union catalogue, Libris (https://libris.kb.se). The Queerlit database is intended to help scholars and the general public discover relevant fiction. In addition to the common Libris web search interface, the Queerlit sub-database is to be searchable through an advanced separate interface supporting subject searching based on a dedicated LGBTQI thesaurus. This thesaurus, the Queer Literature Indexing Thesaurus (QLIT), also developed as part of the project, is largely based on the English-language Homosaurus Vocabulary (https://homosaurus.org/v3). Homosaurus was developed to cover a range of varied information resources, and its first version dates back to 1997 (for more about Homosaurus, see Watson, 2021; Cifor and Rawson, 2022).

QLIT incorporates relevant Homosaurus terms, translated to Swedish, adapted to reflect the Scandinavian context and adapted to fiction. The thesaurus and the search functionalities are further informed by the needs of potential users, including subject experts and the general public (Author, Year) as well as librarians. The remainder of the paper describes the user study of librarians and its implications for Queerlit information system development.

Methodology

To help ensure high-quality subject access to Swedish LGBTQI fiction, the study aimed to determine librarians' perspectives on this topic. These were investigated via a web-based questionnaire comprising 35 closed and open questions, 22 of which were required. The survey was sent to relevant Swedish mailing lists and advertised in associated social media channels on 10 January 2022. By the survey closing date, 20 March 2022, 82 responses had been received.

Results

Demographics

In total 82 respondents completed the survey. Most of them (69 or 84.1%) were 40–59 years old (y.o.). Of these, 25 (30.5%) were 40–49 y.o., 23 (28%) 30–39 y.o. and 21 (25.6%) 50–59 y.o. Some were also younger, 18–29 y.o. (9 or 11%) and some older, 60–69 y.o. (4 or 4.9%).

All but one had completed a university education (81 or 98.8%). Of those with a university degree, most had a four-year master's degree (Swedish “magister”) (32 or 39.5%) or a bachelor's degree (31 or 38.3%). Some had a five-year master's degree (16 or 19.8%), one a two-year degree (1.2%) and one a doctoral degree (1.2%). Most had their highest educational degree in library and information studies (65 or 79.3%); others' highest degree was in comparative literature (4), religious studies (2), cultural studies (1), journalism (1), ethnology (1), music (1), social work (1), theatre (1), as well as science and technology in society (1).

Library work

Most respondents (55 or 67.1%) worked at a large library, either a public library with 20 or more employees (31 or 37.8%) or at a university library (24 or 29.3%). Others were employed by a smaller public library with up to 10 employees (15 or 18.3%) or a medium-sized public library with 11–20 employees (8 or 9.8%). Of the remaining 4 respondents, 2 worked at a special library (2.4%), 1 at a school library (1.2%) and 1 at the Swedish national library (1.2%).

Their typical work tasks involved interaction with end users (e.g. information desk, user training, book presentations and events) (69 or 84.1%) or orders and acquisitions (40 or 48.8%). About a quarter worked in cataloguing and metadata (19 or 23.2%), while only 4 specifically conducted subject indexing (4.9%). The small number working in the last area was likely because most libraries rely on pre-existing metadata from the Swedish Union Catalogue; for example, all adult Swedish fiction is catalogued by the Swedish National Library, and all Swedish children's and young adult literature by the Swedish Institute for Children's Books. A minority also worked in library management and development (6), in book delivery to users unable to come to the library, including bookmobiles (4), in physical collection management (3), as systems librarians, including library website development (3) and in research support, scientific communication and open access (1).

When librarians look for LGBTQI fiction

Most study participants at least sometimes searched for LGBTQI fiction (59 or 72%). Many did it a few times per year (33 or 40.2%), although a good portion of them searched for LGBTQI fiction as often as a few times per month (18 or 22%), weekly (7 or 8.5%) or even daily (1 or 1.2%). Some did not look for LGBTQI fiction at all (23 or 28%); this group left the survey at this point, as the remaining questions specifically concerned LGBTQI fiction searching.

Most librarians searched for LGBTQI fiction on behalf of end users (44 or 74.6%) or to acquire materials for the library (36 or 61%), for example, to build an LGBTQI collection. Searching on behalf of end users entails:

1. giving advice (8), such as reading suggestions (to individual borrowers, school teachers or daycare educators), finding information resources for students (e.g. those enrolled in gender studies' programs) and providing relevant information to colleagues, friends and family;

2. preparing for relevant events (5), such as reading clubs and book presentations; and

3. preparing library exhibitions and book displays (5), such as Pride exhibitions and regular book displays to reflect diversity.

Other reasons for searching for LGBTQI fiction the participants provided include:

1. own personal interest (7), such as pleasure reading and personal development;

2. keeping oneself updated for work (5), that is staying abreast of developments and increasing one's own knowledge of these topics; and

3. adding subject keywords (1).

When users ask for help with LGBTQI fiction

Librarians in the study reported that people looking for LGBTQI fiction came from all age groups – children, youth, adults and elderly. Most of those looking for such fiction often or sometimes were adults, in 37 out of 59 libraries (62.71%), and youth, in 34 out of 56 libraries (60.71%); they were followed by children, in 13 out of 60 libraries (21.67%), and elderly, in 8 out of 57 libraries (14.04%).

Similar to the results reported in the earlier study (Author, Year), the librarians indicated that users tend to prefer printed books (according to 58 or 98.3% of the studied librarians), followed by e-books (31 or 52.5%) and streaming audio books (19 or 32.2%). Also used were talking books (for users with special needs) (12 or 20.3%) and CD audio books (7 or 11.9%).

What information needs do users have when approaching librarians for help in finding LGBTQI fiction? Many respondents only rarely got such questions, some suggesting that this was because it is too revealing for the users, who may prefer to look by themselves.

When librarians were approached with questions about LGBTQI fiction, these were often about themes specific to the LGBTQI community (13), such as particular sexual orientations and gender identities (e.g. non-binary or trans characters), girls who like girls, families who live differently from the heterosexual nuclear family norm, LGBTQI love and lesbian relationships with happy endings. Other questions were about already known items (6), that is specific titles or LGBTQI-themed book series sequels. Some users asked specifically for certain genres (3), such as LGBTQI manga, comic books or picture books.

A fair number of questions came from pre-schools (12) or schools in general (6) about themes of diversity and LGBTQI. Pre-schools might ask for books that “reflect diversity and different ways of being human,” present different family types, discuss gender issues, critique norms or feature norm-creative innovation. This was likely linked to national curricula. For example, the pre-school curriculum promotes the equal value of all people, stating that no child should be subjected to “discrimination on the grounds of gender, transgender identity, or expression” (Skolverket, 2019). Questions also came from students (9) enrolled in teacher training programs, healthcare or nursing sciences and who were asking for norm-critical books or LGBTQI books for children and young adults.

Where librarians find information about LGBTQI fiction

A previous study of the potential end users of the Queerlit database (Author, Year) revealed that most respondents did not identify relevant LGBTQI literature through libraries, but instead relied on social media and got suggestions from friends; they did, however, use libraries to obtain copies of the identified books. This study shows that librarians also look for information about LGBTQI fiction, mostly via social media (38 or 64.4%) or by asking colleagues (33 or 55.9%), but over half also consulted library catalogues (31 or 52.5%). Many also turned to friends and acquaintances (23 or 39%), specialized periodicals and publishers' catalogues (20 or 33%), general periodicals (16 or 27%) and bookshops (14 or 23.7%).

Specifically regarding the internet, most librarians used social media (38 or 64.4%), Libris (25 or 42.4%), bookshops (22 or 37.3%) or other library catalogues (20 or 33.9%). The most popular social media destinations were Instagram (23), including authors' or publishers' channels and specialized LGBTQI channels, such as “Fem Flator” (Swedish blog on lesbian culture) (2). Also popular was Goodreads (19), considered particularly useful as a source of information on youth literature (5); this was followed by Facebook (7), with several mentions of different groups aimed at literature for children and young adults as well as specialized LGBTQI groups. Also used were Tiktok (6), Tumblr (2), StoryGraph (1), Youtube (1), Booktube (1) and Twitter (1).

In addition to their own library catalogue, the participants consulted those of other libraries, including Libris, Kvinnsam (National Resource Library for Gender Studies), the Swedish Agency for Accessible Media and large public libraries (e.g. in Stockholm, Malmö and Umeå). Regarding specialized periodicals and publishers' catalogues, most used Svensk Bokhandels kataloger (triennial catalogue of Swedish-published books) (12), individual publishers' catalogues (11) and BTJ's list (7) (BTJ is a commercial library book seller publishing a catalogue of reviews for newly published books every other week).

The librarians also turned to general periodicals, often for book reviews. Most common were Dagens Nyheter, a daily newspaper with national coverage (6), and Göteborgs-Posten, the largest newspaper in western Sweden (5). Also mentioned were Aftonbladet, one of the largest daily newspapers in the Nordic countries (1), Svenska Dagbladet, a daily newspaper with national coverage (1), and English periodicals – The Guardian (1) and Diva (a magazine for LGBTQI women and non-binary people). The bookshop most often consulted was Adlibris (12), likely because it won the public procurement as the book supplier for most Swedish libraries. Other bookshops used were Bokus (5), SF Bookshop (3), Amazon (1) and Barnes and Noble (1).

Other information channels listed were mostly online, including Google (9), LGBTQI book blogs (9), such as the Swedish “Tekoppens tankar,” library websites (4) with recommended reading lists, such as Umeå library's “Brytiga böcker,” specialized LGBTQI websites (3), such as LGBTQ Reads and YA Pride, and LGBTQI podcasts (2), such as Bookriot and Bögbibblan.

Also used was QX, an LGBTQI newspaper (3), television (3), such as the TV show on literature “Babel” (1), Swedish public radio (2), the literary magazine Vi läser (1), Historiskan, a magazine on women's history (1) and Bokinfo (a database of Swedish books and publishers) (1). Some also mentioned library users themselves as a source of information (3), as well as lectures, special events and book clubs (3). A few also mentioned finding LGBTQI fiction via different mailing lists (1) or by chance (1).

Satisfaction with searching for LGBTQI fiction

Regarding satisfaction with searching for LGBTQI fiction, most librarians were neither satisfied nor dissatisfied (29 or 49.2%); some were satisfied (13 or 22%) and some dissatisfied (11 or 18.6%), with a few very satisfied (5 or 8.5%) and very dissatisfied (1 or 1.7%).

Those who were satisfied reported easy searching using social media such as Goodreads (4), for example, via “Others have liked” or hashtags, or reported that it was easy to conduct known-item searching (3). One person reported being satisfied with their own library catalogue because it has dedicated significant resources to indexing LGBTQI materials. Another librarian felt confident about their search skills, after having gone through the Swedish LGBTQ certification process for institutions, comprising education on LGBTQ issues for all employees.

While known-item searches worked, searching by subject was problematic. Most (26) blamed that on poor subject indexing or unsatisfactory catalogue search functionalities:

1. it was hard to find good search terms (“hard to know if search terms are good”; “it is hard to find subject keywords to search for, books I know are LGBTQI often do not have those keywords”);

2. subject index terms were inconsistent (“an endless number of search possibilities; if one used the same terminology, it would have been easier”; “search words change over time and from one country to another”; “different subject keywords are used, so one needs to do a lot of searches in order to find as many books as possible”; “we have worked on subject indexing in our catalogue … but … newer materials have more subject headings than older materials and … not all books with … similar content always have the same subject headings”);

3. inability to search specifically by theme or identity – often one could only find all LGBTQ books;

4. slow update of subject indexing systems;

5. generally poor search functionalities;

6. even though the librarian might be familiar with the catalogue well enough to help the user, the catalogue would not work for the user; and

7. subject headings were sometimes flawed in library catalogues (a participant cited the example of The Confession by Jessie Burton: in the Swedish edition, the book description is translated to refer to one of the two main characters, Elise and Constance, as “he,” although they are both “she”; because of this error, the library has misplaced the book outside the rainbow shelf, and the book has not received an LGBTQI subject heading).

Others were dissatisfied because they had to consult multiple resources beyond the library catalogue (6), such as blogs, Goodreads and colleagues, or they stated that it was hard to search in general (3) (“many uncertain years of looking for information on the subject”) or that there simply seemed to be a lack of relevant literature. The challenges were perhaps best summarized in the following statement of a study participant: “It is not always easy to find relevant literature: this depends partly on the supply on the market, partly on availability at our library, and partly on poor information in the catalogue – in many cases I see my own expertise as the most useful.”

Metadata and subjects

The metadata elements most often used were subject (by 42 or 71.2% librarians), a combination of two or more elements (27 or 45.8%), author (23 or 39%) and title (20 or 33.9%).

Terms commonly used to search by subject when looking for LGBTQI fiction were:

1. Related to LGBTQI identities and/or practices: LGBTQ (in Swedish: HBTQ) (19), including related word forms such as LGBT* (HBT*), LGBTQ+ (HBTQ+), LGBTQI (HBTQI), LGBTQIA+ (HBTQIA+), as well as phrases such as “LGBQ people” and “LGBTQ + youth”; trans (9), including word forms such as trans people, trans person, as well as closely related terms such as “trans experience,” transsexualism and “life as a trans”; queer (8), including phases such as “queer YA”; homosexuality (8), including related adjectives, phrases and word forms (e.g. homosex*, homosexual, homosexual men, homosexual love and homosexual parents); lesbian (7), including “lesbian women” and lesbianism; nonbinary (5), including “nonbinary persons”; intergender (3); intersex (2), including intersexuality; bisexuality (2), including truncated search word forms such as bisex*; gay (2); asexuality (1); genderfluid (1); love between boys (1); love between girls (1); MLM (men loving men) (1); Pride (1); rainbow family (1); Sapphic fantasy (1); sexuality (1); two girls who become a couple; WLW (women loving women) (1).

2. Related to LGBTQI: love (3), “family types” (2); norms (2); normcreative (2), normcritical, gender (5), including phrases such as “gender identity,” “gender roles,” “gender aspects,” and “gender perspectives”; different (2), including the phrase “being different”; adventure (1); family (1); friendship (1); rainbow (1) and “own voices” (1).

Recent search

The participants were asked to recall their latest subject search for LGBTQI fiction. This was shown to be related to different information needs. Some were looking for books for the library (9), for example, for the physical or digital rainbow shelf, a book talk or an internal training workshop on gender and equality.

On behalf of end users, most information needs were for pleasure (9), including the example of a fantasy novel in which boys like boys; or for work or studies (5), including a student thesis and an academic publication. Other information needs were related to recognizing oneself in the stories (1) or a search for identity (1). Some asked for resources on behalf of children and grandchildren (2), including a grandparent whose grandchildren had two mums and was looking for appropriate children's books. Others mentioned the following types of information resources being sought:

1. fiction on homosexuality in repressive countries (1);

2. picture books for a parent whose child's daycare group consisted of children whose families were white, heterosexual, middle-class families; the parent wanted their child to see other parts of reality (1);

3. youth books with trans characters (1);

4. a graphic novel with an LGBTQI theme for a 12-year-old (1);

5. books about love between boys or between girls, sought by two children in their early teenage years (1);

6. a work of fiction with a love story about boys, sought by youth (1);

7. “something similar” to the Heartstopper graphic novel by Alice Oseman (1);

8. a work of fiction about lesbian relationships with a happy ending (1);

9. works of fiction that would help address a user's general curiosity about the theme (1); and

10. any work of LGBTQI fiction, sought by pupils during a class visit (1).

Teachers have also approached librarians:

1. university teachers (3), in gender studies or teacher education;

2. school teachers (3), looking for books with non-binary characters and perspectives; and

3. daycare workers (3), looking for children's picture books that show different types of families or that more broadly represent gender.

Search terms used in their latest subject search were mostly related to LGBTQI identities and/or practices: LGBTQ (in Swedish: HBTQ) (13), LGBTQI (HBTQI) (4), Queer (4), homosexuality (4), LGBTQ+ (HBTQ+) (3), homo* (3), lesbian (3), gay (2), lesb* (2), as well as one of each of the following: bi*, bisex*, bisexuality, LGBTQ* (HBTQ*), LGBTQI+ (HBTQI+), LGBTQIA+ (HBTQIA+), homosex*, homosexual, intersex, intersexual, LGBTQ, MLM (men loving men), non-binary, pansexuality, pride, rainbow, trans*, transgender, trans people, transsexual; bi* and love*.

Other related terms included: rainbow families (5), gender (3), being different (2), sexuality (2), asex* (1), coming of age (1), gender perspectives (1), gender roles (1), identity (1), Mums (1), norm-criticism (1), same gender (1) and *sexuality (1). Another commonly mentioned term was love (7), including love* (1). Also listed were: relationships (1), romance (1), sex (1), YA (young adults) (1) and youth (1).

How satisfied were they with the search results? Similar to their general search experience reported above, most were neither satisfied nor dissatisfied (17 or 28.8%) or satisfied (16 or 27.1%). Some were somewhat dissatisfied (7 or 11.9%) or very satisfied (5 or 8.5%). One was completely dissatisfied (1.7%). Some could not remember their experience (13 or 22%), likely because they rarely search for LGBTQI fiction on behalf of end users, as seen from earlier replies.

Desirable keywords

What subject terms would the librarians like to be able to search for when looking for LGBTQI fiction?

Again, the desirable search terms can be classified into several different groups. The first one is about LGBTQI identities and/or practices, including both the general term and more specific ones: trans (21), including transfeminine, transmasculine or trans*, “trans person,” “trans people,” and transsexuals; bi (4) and related word forms, such as bisexual or bigender; homosexuality (4); lesbian (5), with the synonym “flata” (2); rainbow family (4); queer (4); LGBTQ (HBTQ) (3); LGBTQI (HBTQI) (2); bisexuality (2); cis and related forms, such as cisperson (2); gay (2); homosexual (2); intersexual (2) and one of each of the following: boy, children, gay, “gender dysphoria,” genderfluid, genderqueer, girl, guy, “homo 2 homo*,” “homosexual men” (when only about men), intersex, non-binary, pan, “queer YA,” “same-gender families,” and woman.

Desired related terms were: asexual (3), gender (3), love (3), asexuality (2), normcreativity (2), including truncated forms such as normcreativ* (1), “gender norms” (2), “star family” (2) (the Swedish word stjärnfamilj is wordplay on the Swedish word for nuclear family and means an extended family with, for example stepparents and stepsiblings), transsexuality (2) and one of each of the following: autosexuality, “biological gender,” cisnorm, dysphoria, empowerment, equality, “family types,” freedom, “gender-conscious books,” “gender expression,” “gender identity,” “gender pronouns,” “gender roles,” heterosexuality, “identity formation for younger children,” inclusion, intersexuality, “legal gender,” “lesbian love,” “LGBTQI family,” “LGBTQI history,” liberation, norm-breaking, norm-critical, norms, polyamorous, polygamy, pride, queer literature, “rainbow representation,” “right of determination,” self-determination, “sexual relationships,” sexuality, society, transition and transphobia.

The librarians would also like to be able to combine these terms with other facets, such as those denoting genre or other themes (e.g. LGBTQI perspective in higher education, LGBTQI for adults, LGBTQI for children, LGBTQI for young adults, LGBTQI easy to read, LGBTQI with genre such as thriller or crime and LGBTQI without explicit description of sexual relationships).

Search interface

In addition to a good subject indexing system and an indexing policy with greater exhaustiveness and specificity than is common when subject indexing fiction (see “Background”), it is important to have a search interface that allows the users to take the full advantage of the intellectual effort invested into creating subject metadata. The majority agreed that good subject metadata accompanied by search functionalities that use the metadata at the interface (e.g. to broaden or narrow the search, or to learn about concepts etc.) would be highly useful for finding LGBTQI fiction (34 or 57.6%) and some thought it would be crucial (9 or 15.3%). Others believed it would help to a certain degree (15 or 25.4%), with one stating that it would not help enough (1 or 1.7%).

When searching for LGBTQI fiction, the librarians would in particular appreciate it if, upon entering a search term, the system would provide a list of related, broader terms in order to get more results (higher recall) (51 or 86.4%) as well as the opposite, a list of narrower, more specific terms in order to get fewer but more specific results (higher precision) (46 or 78%). Many would also appreciate hierarchical subject browsing (36 or 61%) as well as the disambiguation of homonyms or polysems (28 or 47.5%).

Some further justified their choices by explaining that these mechanisms were needed to help in choosing the correct, preferred subject keywords in order to get the best results (i.e. only relevant and all relevant ones). Another participant commented that it was good to retrieve as many resources as possible, but that the results needed to be digestible so that one would not be overwhelmed, which is why additional, narrower searches would be beneficial. One librarian wrote that often the searches yielded too few hits or that it was difficult to find the right term to search for, so getting search term suggestions would be helpful.

Participants also commented that help with the choice of search terms was important for those unfamiliar with the terminology of the subject matter: the gender context is very broad and help in understanding it (e.g. via a browsing tree) would benefit both the librarian and end user. Broader and narrower terms would also help those less familiar with the subject and those whose knowledge is limited to general concepts. There are many possible narrower terms for certain search terms, and if one is not very knowledgeable, one might not be aware of them all or of the best choice. Finally, one participant commented that “library catalogues should be at the forefront of including progressive subjects, not just the traditional and heteronormative ones” and that a browsing overview should help with that.

Other desirable features that the librarians would like to see are: ability to truncate both the beginning and end of search terms, option to limit/filter search results (e.g. by age group, media type, language and year of publication), a list of frequently occurring search terms and a list of newly added subject terms. One librarian identified the problem of some works being indexed as, for example, “bisexuality,” even when the concept is only peripheral to the work, showing that indicating the centrality of index terms would help increase precision.

Regarding final, general comments, the librarians were appreciative of the Queerlit project to help make LGBTQI fiction more visible to librarians and users. One librarian pointed to the need for training in metadata creation (e.g. at BTJ), and another complained about the problems with indexing inconsistencies at the word-form level across systems: for example, a book about gay men may sometimes have the subject term “homosexuality,” while another book also about homosexual men may have the term “homosexual men.” Some titles have the subject terms in English, some in Swedish and some in both languages, the latter mainly because of copy cataloguing and importing records from OCLC (a global library organization that provides metadata and other services) without added Swedish subject headings.

Implications

QLIT and indexing

The desirable and recently used search terms identified in this survey and the previous survey targeting end users (Author, Year) have been compared with existing descriptors in QLIT. Those missing from QLIT were added as new preferred terms or as alternative terms for existing QLIT terms. For example, the term “unga homosexuella” (young homosexuals) has the following non-preferred forms: “homosexuella ungdomar” (homosexual youth), “homosexuella tonåringar” (homosexual teenagers) and “unga lesbiska och bögar” (young lesbians and gays) (identifier: https://queerlit.dh.gu.se/qlit/v1/yp38jg66). The last term was implemented to support searching by as many synonyms as possible. In the future, search logs are planned to be consulted to include further terms in QLIT as synonyms. The interface also invites user suggestions of any missing subject terms or missing works (via email or contact form).

The indexing of major versus minor themes is implemented and reflected on the interface so that major QLIT themes are represented in larger font and dark orange, while minor ones are represented in smaller font and light orange accompanied by “-perifert” (Swedish for peripheral) (see Figure 1). Major and minor themes are not distinguished from one another in Libris, so the distinction cannot be implemented at the level of SAO (in grey in Figure 1). In addition to SAO for general subject terms, the Swedish SAOGF (Swedish subject headings genre/form) system is used to represent genre terms, to allow searching by combining genre with an LGBTQI theme.

Indexing consistency is ensured in three ways: first, by providing a scope note for each descriptor (see Figure 2); second, metadata produced by one librarian are reviewed by another and, third, the team uses regular meetings and chat channels to discuss specific indexing practices, the applicability of subject headings to a specific work, as well as QLIT additions and revisions. The Queerlit team is also preparing an indexing policy, which needs to be aligned with the Libris indexing policy.

The librarians also expressed a need to exclude certain works, such as those on suicide or explicit sex. Subjects such as suicide are addressed at the level of the search interface, by implementing Boolean NOT. Subjects such as explicit sex are addressed at the indexing stage, when such works are indexed as erotic or homoerotic genres, although minor explicit sex themes are not indexed. Also, both librarians and end users touched on the need to distinguish the affective and aesthetic aspects of fictional texts, for example, to search for “feel-good” literature or literature with a “happy ending,” or to search by style, for example, by distinguishing high-brow from popular literature. Some of these matters are related to genre and could be considered for implementation, but not all (e.g. literature with a “happy ending”).

Finally, it is important to plan for retrospective subject indexing, for older books that have been indexed with too few terms, that were not recognized as LGBTQI at the time of indexing, or for which the index terms have changed over the years. Historical LGBTQI terms are denoted by a collective descriptor in QLIT (https://queerlit.dh.gu.se/subjects/qj48lh67), to which relevant terms are added as narrower terms; for example, the term “lotusar (HBTQI)” (lotuses [LGBTQI]) used to refer to people assumed to be lesbians).

Search interface

Regarding search functionalities, both end users (Author, Year) and, in this study, librarians asked for higher recall through suggestions of related terms, and for higher precision, through suggestions of narrower terms. The librarians also reported that the few questions they got from end users usually concerned the need for higher precision, for example, to find literature for a specific age group with a specific theme, such as family constellations outside the norm, or stories about non-binary characters. Consequently, both librarians and end users agreed on the need for more specific terms describing gender identities and sexualities, and that these should be possible to combine with other search facets. Both librarians and end users would appreciate an option to browse hierarchically arranged subjects.

To implement these in the Queerlit project, we have consulted earlier research on how to specifically address the identified user needs at the level of the search interface (Golub et al., 2022, Table 1), and found 18 suggested search functionalities (21 for collections that include images, which Queerlit does not address). The project search interface has implemented or considered all 18 functionalities (https://queerlit.dh.gu.se):

1. Browsing by subjects. This is directly available from the main interface under the term “Ämnen” (subjects) (see Figure 3), offering an alphabetical listing of QLIT terms with a “+” sign allowing the viewing of narrower terms. When one mouses over any of the terms in this browsing tree, an option is offered to directly search for works indexed with that term (see, e.g., “sök på Flickor” [“search for Girls”] in Figure 3).

2. Searching by controlled subjects. This is supported by allowing searching of QLIT entire terms or their component parts (“bisexuella flickor” (bisexual girls) or “bisexuella” or “flickor”).

3. Browsing by individual concepts from pre-coordinated terms. This is supported at the main search interface where the user is provided with a list of terms that contain their search word in the controlled vocabularies used (see Figure 4 for an example).

4. Searching by a combination of controlled subject terms. While Libris software supports searching using different metadata elements, no more than one and the same element can be combined. The software is currently being redeveloped to allow combined searching for controlled subject terms using any number of Boolean AND operators; combining this with OR and NOT operators is also being considered.

5. Searching by major and minor index terms. As the Queerlit indexing policies support this possibility (see Figure 1), Libris is currently developing the possibility of searching by two fields, one for major index terms and one for minor index terms.

6. Presenting and browsing excerpts of concept hierarchies and matching search terms to support disambiguation and broadening or narrowing the search. This is planned to be implemented, but the exact ways to achieve this are challenging as the interface should not be overcrowded with extensive additional features (this is one option that would particularly benefit from UX testing, which we plan to conduct). A related feature has been implemented to increase recall: the search interface allows expanding the search results to also retrieve metadata records containing the search term's narrower terms. For example, entering the term “bögar” (gay men) as a search term would provide metadata records with “bögar” in index terms, but also metadata records that, according to the thesaurus, contain any one of the narrower terms of the descriptor; thus the search for “gay men” will also retrieve results about, for example, older gay men or young gay men.

7. Autocompleting the search term once the user starts typing. The Queerlit search interface has implemented this so that once the user starts typing, the system provides suggestions from the controlled vocabularies immediately after the first character has been entered. This includes suggestions that start with the entered string as well as those that contain the string elsewhere.

8. Auto-suggesting controlled versions of entered search terms. This option has been implemented via alternative terms in QLIT. For example, if one enters the term “tjej” (girl), the system will suggest the preferred term “flickor,” as seen in Figure 5.

9. Suggesting corrections of typos. Since Libris does not support a spelling correction feature, the Queerlit database added common misspellings. These are not available as synonyms in the records describing a descriptor but are hidden from the user. A more systematic spelling correction functionality would be likely more comprehensive and will therefore be prioritized in future developments.

10. Searching by words from various metadata elements and full text. This feature has been implemented.

11. Combining controlled subject searching with searching by other bibliographic fields. This feature has also been implemented.

12. Highlighting search terms in retrieved metadata and resources. This is something that Libris does not currently support and that Queerlit cannot invest in developing at the time of writing.

13. Advanced searching by Boolean and proximity operators, truncation of searches and wildcard searches. Boolean operators AND, OR, NOT are supported but only in free-text searching (cf. Functionality number 4 above). Also available is phrase searching with quotation marks, truncation and wildcard searching. Proximity operators are not supported.

14. Linking each subject access point to its resources. Figure 1 with a retrieved metadata example shows how mousing over an index term allows searching for all other records indexed with the same term (“syskon” – siblings). Also, browsing subject terms as in feature number 1 allows asking for all metadata records with the index term of choice (see Figure 3).

15. Linking subject access points from one controlled vocabulary to corresponding concepts in others. Each QLIT term has exact or near matches in Homosaurus, SAO, Swedish Children's Subject Headings (Svenska Barnämnesord) and Library of Congress Subject Headings (LCSH). For example, preferred term ‘transpersoner’ (transpeople) (https://queerlit.dh.gu.se/qlit/v1/wj27lv47) has an exact match in Swedish Children's Subject Headings (Barnämnesord), Homosaurus, LCSH and SAO.

16. Adding browsing and searching end-user tags. Libris does not allow this and Queerlit is dedicated to high specificity and high exhaustiveness in its indexing policies. Additional user tags from services such as LibraryThing or creating an additional database for user-entered terms has not been considered yet.

17. Combining previous search formulations. This feature has not been implemented, but whether it is possible to do so in the future will be considered.

18. Help with searching. This is available and is being further developed.

Finally, the interface is being evaluated via user studies focusing on user experience (UX), and many of the search interface characteristics and functionalities are expected to further evolve before the project ends in December 2023. Thereafter, the database will be managed and curated by Kvinnsam, although external research funding will need to be sought for any significant further development.

Concluding remarks

This survey focusing on librarians in Swedish libraries was intended to collect data to inform the development of a dedicated subject search interface for Swedish LGBTQI fiction. Of the 82 participants who completed the online questionnaire, most at least sometimes searched for LGBTQI fiction, some as often as a few times per month. Most did this on behalf of end users, often to provide reading suggestions in various contexts (e.g. pleasure reading as well as professional purposes for daycare/school teachers) or for library acquisitions. All age groups asked for search help on this topic, although most requests were from adults and youth, requests from children and elderly being less common.

Some respondents stated that they rarely got LGBTQI fiction questions specifying information needs. They interpreted this as user reluctance to disclose personal interest. Librarians who did get such questions usually described them as about themes specific to LGBTQI, such as specific gender identities, girls who like girls, families who live differently from the hetero nuclear family norm and LGBTQI love. Many questions came from pre-schools and schools where education on diversity and LGBTQI themes is part of the national curriculum, or from students enrolled in teacher training, healthcare or nursing science programs.

The participating librarians looked for information about LGBTQI fiction mostly by searching social media, asking colleagues or consulting library catalogues. Most were neither satisfied nor dissatisfied with online searching. Further analysis showed that while known-item searches worked, searching by subject was problematic; most blamed that on poor subject indexing or unsatisfactory catalogue search functionalities.

When searching for LGBTQI fiction, the librarians would particularly appreciate it if upon entering a search term, the system would provide a list of related, broader terms in order to find a larger number of relevant results as well as the opposite, a list of narrower, more specific terms to get fewer but more specific relevant results. Many would also appreciate hierarchical subject browsing as well as the disambiguation of homonyms or polysems. The results also indicate a need to provide multiple synonyms and word forms as well as automatic spelling correction. Some respondents also emphasized the need to support end users: even if the librarians themselves are able to use the information system even when complex or lacking sufficient search functionalities, end users need much improved systems with the abovementioned functionalities. This is especially important for LGBTQI users, who often prefer not to approach anyone for help.

The Queerlit project was able to implement many, but not all, of these functionalities due to the limitations of the Libris software and the characteristics of the QLIT thesaurus. As to the latter, the QLIT thesaurus was adapted from Homosaurus, which is characterized by a long list of top terms and only a few hierarchical levels for some terms. This prevents visualizing the thesaurus in a browsing tree, which would be possible with some classification systems, such as Dewey Decimal Classification.

Also, as terminology is rapidly developing, LGBTQI information services should be quick to include new terms in their subject indexing systems. User search logs should be regularly monitored for new terms to be included, or simple and quick ways for users to leave terminology feedback should be implemented. Collaborative tagging could also be used as a way to learn about new end-user terms.

Further work is needed to explore search interfaces that make full use of the subject metadata assigned to LGBTQI works, to maximize their findability and use. Also, more research is needed into the user experience (UX) of the specific interfaces. Finally, an important question is how to connect resources in different databases, both within Sweden and worldwide. How can we encourage the uptake of the QLIT thesaurus in other, connected information services, such as BTJ, from which many libraries acquire metadata, or Libris, for cooperative cataloguing in Sweden? How do we move towards the international and multilingual interoperability of index terms for LGBTQI resources? Since Homosaurus is available as linked data, it should be easy to discover other libraries that use the same concept and allow end users to get more results from other libraries around the world.

Figures

Figure 1

Representation of major (larger font, dark orange) and minor themes (smaller font, light orange) in the Queerlit database for a work entitled Sodomsäpplet by Bengt Martin

Figure 2

QLIT: an excerpt of alphabetically listed descriptors, each accompanied by a scope note

Figure 3

Subject browsing interface in the Queerlit database

Figure 4

Searching for “bisexuell” (bisexual) allows the user to choose from subject terms that contain that term. First row (in QLIT): “mothers to bisexuals,” “young bisexuals,” “bisexual men,” and more. Second row (in SAO): “bisexual parents,” “bisexual women,” “bisexual men,” and more

Figure 5

Searching by synonyms (top) and the corresponding thesaurus term entry (below)