Fig. 1. Gallica homepage
Source: Gallica, Bnf, 2019
- 1 Up until 31 May 2018, Gallica offered access to 4.3 million digitized archive documents of the Bibl (...)
- 2 This principle of interconnecting approaches has guided the work of Bibli-Lab, a research partnersh (...)
1Gallica1 users, as well as the volume and range of its content, provide an ideal space for the observation of types of appropriation related to accessing digital archive material. A comparison of two major self-administered questionnaires (in 2011 and 2016) helped identify changes in the profiles of Gallica users and their reported practices, and, via this interesting example, in the wider relationship to online archive content in the context of scholarly research (with the purpose of knowledge acquisition). Despite the wealth of information gathered, these studies also have a limitation for the social science researcher: their initial objective was to know the users and usages of digital objects considered to be fully formed, without questioning their technological functioning, implications, or the wider practices to which they belong. Such an approach can be explained by the primary need of a public institution such as the Bibliothèque nationale de France (BnF) to regularly assess the quality of its collections, and also the number and diversity of its users. However, beyond Gallica and the relationship to online archives, I hope that this narrow framework can shed light on “ordinary scholarly practices” (Le Marec & Mairesse 2017) in terms of digital research, even if this would require additional information and comparisons which are not in the purview of this article. I focus on two notable developments in usage of Gallica in the space of five years: the growing role of personal research—in a context in which the reasons for use are increasingly varied and interlinked—and the evolution of close reading online. I will remain cautious in my presentation of these observations owing to issues regarding methodology that we will discuss first of all: in view of the explosion of digital practices, building an overall vision of the uses of online content that can be freely viewed and exported has become very difficult. For this reason, the results of questionnaires will be compared with other analysis: interviews; ethnographic video observations; and analysis of access logs2.
- 3 The advert was extended to be visible on all pages of the site, rather than just the home page that (...)
2Despite the ambiguity of the term “digital library” (Papy 2016), I use it in two (inseparable) senses: an organized collection of digital documents and the means to access these documents. Following that of 2011, a quantitative survey of Gallica users was conducted in October 2016 by the “TMO regions” market researcher (TMO 2017), within the framework of a study carried out by the BnF, the Obvil laboratory, and Télécom ParisTech] “Making heritage available online: The development of usage and transformation of knowledge?” The procedure used in 2011 and 2016 was roughly the same: an invitation to respond to a survey was visible for a month on the Gallica website, with a link to an online questionnaire. So as not to distort the results by approaching certain groups of users, contact lists and social networks were only used two weeks in to the survey, so as to control their impact on the original results through the comparison of subset data (before and after the campaign). The 2016 survey collected 7,625 responses in total compared with 3,828 in 2011. This increase is bigger than the increase in Gallica users between the two surveys and can partly be explained by the greater visibility of the advert on the site3. At the same time, it proves users’ strong attachment to the digital library, even though in contrast there has been a fall in response rates to this type of online questionnaire, which does not appear to be about to slow.
- 4 Figure given by Xiti, cf. note 3.
- 5 A research project on Gallica server access records thus showed that the average length of a sessio (...)
3However, the excellent response rate—calculated at 1.2% of the estimated number of unique visitors during the survey period4—does not in any way preclude bias (Loosveldt, Carton, Billiet 2004). The problem increases when we aim, as I do in this article, to compare data from two samples obtained with five years in between using a procedure that has itself evolved, even if only marginally in the case at hand. Unlike with surveys carried out in the BnF’s physical spaces, according to the survey procedure followed, it is not in fact possible to control the online sample, either by comparison with actual data, or by ensuring random selection from the target population. One solution to this problem is to consider that this type of inquiry allows us to gather responses from the most committed users, who have both a certain degree of knowledge of the site and a personal interest in its existence, as Ganassali & Moscarola’s study does (2004). It would not make sense to discuss whether the data are representative or not outside this restriction. Therefore I chose to call respondents to both surveys “Gallicanauts” to distinguish them from other users, including a significant proportion who only spend very short periods of time on the site, sometimes without even knowing exactly what it is5.
4Beyond the problem of how representative the data are, which is resolved by avoiding inferring general relevance from what the most committed users say, we encounter the issue of how adequate the responses are—the majority of which are “closed” within the framework of a quantitative questionnaire—in light of real users in a digital environment in permanent evolution. It is difficult not to project, owing to the way the responses are “coded” in advance by the researcher, presumed usages, linked in particular to librarians’ representations relating to legitimate scholarly practices, in particular those of reading. Updating representations of “Gallicanauts” and hearing the way they talk about their online activity were the main aims of the qualitative interview stage, which preceded the drafting of the 2016 questionnaire: fifteen semi-structured interviews were thus carried out with various profiles of Gallicanaut who used the digital library regularly (Beaudouin, Garron, Rollet 2016). The interviews lasted between thirty minutes and an hour and a quarter, and took place preferably in the environment in which respondents were most used to using Gallica (office, home, etc.). Among other adaptations of the online questionnaire that were dictated by the interviews, the possible reasons for using Gallica were reformulated and developed, allowing us to move beyond the reading paradigm as it stood before online media: users may “view a document closely” on Gallica, but also “explore a domain of interest or theme,” “collect documents,” or “compose a bibliography,” or use it “out of curiosity, as casual browsing.” The 2016 questionnaire strove to take users’ sporadic reasons for using Gallica into account, distinguishing what they logged on to do “every time, or almost every time,” from what they “often” logged on to do, and even from what they “occasionally” logged on to do.
5Despite these efforts, using a self-assessment approach supposes a “regular” reason for users logging on (since they are asked why they log on “mainly,” “often,” etc.) and that they are aware of this reason—two presuppositions that are increasingly difficult to verify when it comes to multi-activity and micro-temporalities. It is more and more difficult for users to “separate out their activities and have a clear idea of the time spent on [each one]” (Beaudouin & Denis 2014). Comparison with results from other methods of observation and analysis of Gallica users is therefore necessary to support my two main assumptions. These take the form of semi-structured interviews with users, which help to show the rationale behind choices and motivations; ethnographic video observations, which put online research inquiries into the materiality of their own environment; and finally analysis of access logs, which account for even the briefest visits. This collation is necessary since it allows us to vary the “unity of action observed and therefore the position of the observer” (Denis 2009). At the same time it very quickly reaches its limit since it combines ultimately incomparable data: we cannot observe sample groups of users over time and via multiple channels (logs, interviews, videos, etc.)—which represents an ideal but costly and, in fact, rarely achieved situation. We must rather consider that the two observations formulated come amid a swathe of presumptions which appear worthy of being shared.
- 6 26 years old in the Haut-de-jardin fully public library; 40 years old in the research library that (...)
- 7 In order to better quantify the answers relating to the reason for the visit, the statement was cha (...)
6The first observation made between the two surveys was that the average age of Gallicanauts—traditionally higher than that in the reading rooms of the BnF6—increased sharply after the 2011 survey, rising from 48 to 54 years. The ageing of the Gallicanaut population of 2011 and the increase in audience do not sufficiently explain such a difference. But it is easier to understand this picture of an older audience when we put it into perspective with the role of consulting the archives when carrying out personal research. Among the reasons for use proposed by the questionnaire, alongside the traditional responses of “for studies” and “for work,” we now find “for personal research” distinguished from the simple “for leisure.” This concept of “personal research,” inspired by older qualitative surveys (Amar & Chevallier 2014, for example), allows the identification, outside an academic or professional context, of strong personal engagements within research with a precise objective, a long time scale, and a form of self-discipline (regularity, adherence to a method, etc.) that is far from the meaning now attached to “leisure” in industrial society (in other words, farniente rather than otium). A projection of Gallica7 visits shows that personal research now represents 45% of visits, which is a long way ahead of “for work” (20%), “for leisure” (20%), and “for studies” (15%); 64% of Gallicanauts also said they “always, or almost always,” used Gallica for the purposes of personal research, and 15% said they “often” used it for this. In 2011, although the question was worded slightly differently, just 52% of respondents said that they consulted Gallica “mainly” for “research relating to personal work” (GMV 2012).
7This increase in forms of personal engagement with Gallica may of course refer first of all to a redistribution of—or even explosion in—users’ online activity, including that of “historic” users—students and professional researchers. As with the entire sample of “Gallicanauts,” students and professional researchers are less and less “mono-motivated” (those that remain so represent just 28% of respondents) and thereby alternate between visits for a study project and visits for leisure or personal research. This variety of reasons for use, which increased from 2011, is not only played out between sessions, but also within the same session, interwoven with frequent bifurcations, as shown by a video-ethnography of ten Gallicanauts carried out by researchers from Télécom ParisTech within the framework of the Obvil laboratory (Rollet, Beaudouin, Garron 2016). The objective was to re-record the questionnaire results in the thick of human activity, observing the paths of frequent Gallica users up close. The audiovisual recording of screen activity was accompanied by self-confrontation interviews, which helped give the observations meaning. In particular, these reveal “users” capacity within one research sequence to, on the one hand, swing from one objective to another and, on the other, handle several objectives at the same time, thereby adopting different postures simultaneously’ (Rollet et al. 2016). Rather than diversity, we should rather talk about “porosity” of uses.
8While this interlacing of reasons for logging on to Gallica has been a factor in the rise of personal research, in particular regarding personal research by students and professional researchers, it is not the only explanatory factor. The most frequent intersection of reasons is in fact now that of leisure and personal research (19% of respondents), which is way ahead of all other combinations. This figure provides the outline of a unique group of users that is increasing rapidly, which can be characterized first off as using Gallica a) exclusively for research or alternating between “research” and “leisure” outside the professional and academic spheres; b) being older than the average “Gallicanaut”; c) having a higher representation of men. Since it bears similarities to the traditional figure of amateur that has gone hand in hand with the evolution of natural sciences (ornithology, zoology, botany, etc.) and history, I call this group “Senior Amateurs.” It is significant that the term “amateur” does not historically refer to having less skill or taking research less seriously than a professional would, but rather to having an emotional connection to the object of knowledge, supported by an expert personal practice and a certain degree of commitment to it (Guichard 2008). The now very visible presence of amateurs amongst Gallica users can be considered alongside the increased perception of the nature of Gallica as being “for everyone”—an opinion that increased from 71% to 85% in agreement in 2016 (with an even greater increase for “completely agree”). Research into the Great War provides a unique but measurable example of digital archive use that will give a clearer picture of this user.
- 8 A referrer URL is any site that links to a section of Gallica and thus becomes a source of traffic. (...)
- 9 Accessed on 29 August 2019.
9The appearance since 2011 on the list of Gallica’s top fifteen referrer URLs8 of a discussion forum for amateur researchers on the First World War called Pages 14-189 reveals the breadth of the phenomenon and appears to illustrate the results of the survey, as we will see. Created in 2001, today this forum has twenty-three thousand registered members and five hundred thousand exchanges of messages. This is not a down to chance: its appearance in 2011 on the radar of referrer URLs coincided with the end of the first wave of mass digitization of BnF printed material (2007-2011, at a rate of one hundred thousand volumes a year), which included the 1914-1918 period. Essential sources for the study of this period were thereby made available, freely and remotely, to a wide audience via Gallica. The presence of the First World War theme among referrer URLs, after the wider theme of genealogy—which may often include the former—thus convinced the BnF to launch research into the “online future of digital archives” around this period, so as to better understand its appropriations by communities of amateurs on the internet. This project, carried out between 2013 and 2016 in partnership with La Contemporaine (ex-BDIC) and Télécom ParisTech within the framework of Les passés dans le présent laboratory, proved via interviews and a cartography of five hundred French internet sites on the Great War, not only the vitality of exchanges between amateurs on this theme, but also the role of digitized documents from the digital library archives or document portals Gallica, Mémoire des hommes, L’Argonnaute, etc.). After analysis (all the following results are excerpts from Beaudouin, Chevallier, and Maurel 2018, in particular Valérie Beaudouin’s chapters entitled “La carte du web consacré à la Grande Guerre”, the online map of the Great War, and “Le forum Pages 14-18 et sa communauté de pratique”, the Pages 14-18 forum and its community of practice), it became clear that in fact more than half (54%) of French websites about the Great War collected by the BnF under legal deposit were sites created and run by internet users for personal interest or by associations. Observing their content also revealed that digitized archival documents are frequently published, commented on, captioned, and even transformed there for knowledge purposes. On a forum such as Pages 14-18, where data is analysed automatically, more than one in two messages contain an image or hyperlink to a document; and Mémoire des hommes, the site of France’s Ministry of the Armies that offers access to 1.3 million death records for France, is the most frequently cited website, followed by Gallica. Citations of links to documents and especially images also experienced a remarkable increase in discussion threads on the forum between 2005 and 2012, in line with the increase in the number of documents available online during this period.
Fig. 2. Map of the French web of the Great War
November 2014 data collection with at least thirty incoming and outgoing links (with legend “typology of sites”)
Source: Valérie Beaudouin, Zeynep Pehlivan, Télécom ParisTech, Labex Les passés dans le présent
10The free provision of a critical mass of documents on the internet by the institutions that hold them supports or even encourages the development of online communities of interest where a careful exchange of documents can be of great value to individuals. As Valérie Beaudouin points out, these forums “produce both knowledge and a social link” (2016) in an online context that allows individuals to participate regardless of their social status. The possibility of sharing their research and discoveries freely online reinforces the amateur researcher’s desire to look for and increase their personal resources via the exchange of good practices. The Pages 14-18 discussion forum is thus testament to both the high level of specialization and competence and the high degree of personal commitment that these historical studies encourage on the internet, in relation to the issue of collective or individual memory. Such forums play a mediator role between resources that have been digitized by heritage institutions and their online readership.
11If we take a look at their stated identity via the form they are asked to fill in when they sign up to the forum, registered users stand out from the average French population in terms of two characteristics that connect them, rather, to the Gallicanauts who carry out personal research: the overrepresentation of men (80% in the forum), and above all their age (43% in the forum are 60 years old or over). Analysis of the forum and the Gallicanauts questionnaire produce coherent results that invite speculation about what it is commonly referred to as the “democratization” of access to knowledge thanks to the internet, particularly in favour of the young. While diversification efforts have worked well in terms of “senior amateurs,” we cannot help but notice the small proportion of students using Gallica (which falls from 21% to 10%), even resulting in a reduction in the estimated number. Consulting “specialist” or “heritage” digital libraries, and no doubt even understanding what they offer, in effect requires a familiarity with the concept of archive material and a minimum knowledge of the issues involved in its digitization that some teachers lament to be lacking in their students (Amar & Chevallier 2014). A European qualitative study confirms this difficulty of convincing 15-24 year olds of the merits of a specialized digital library such as Europeana, which is measured against Google and Wikipedia (Dobreva, McCulloch, Birrell, Ünal, Feliciati 2010).
12In addition to the low representation of young people, the group of those aware of Gallica stands out from the general French population in that its members are more likely to have a postgraduate education, high socio-occupational status, and value reading and cultural activities, as found by a recent Ifop study (Ifop 2018). The invisible boundary surrounding knowledge of the BnF and its work in French society is almost identical online: knowledge of Gallica (one in four French people) does not reach an audience that is significantly more diverse in social terms. It is therefore unsurprising that 38% of Gallicanauts in 2016 said they had visited the physical space of the BnF: this figure, which is high in relation to the common perception that leads to users being too-hastily labelled “remote,” is a reminder that the physical and the digital aspects of the library fuel one another. The overrepresentation of Parisians among Gallicanauts (one in four lives in Paris and its surrounding suburbs) proves that proximity to the BnF’s buildings has an impact on online usage.
13The second significant development from 2011 is the increase in close reading online when it comes to use of documents: 66% say that they “often” or “always” do this, compared with 31% in 2011. This type of viewing is now on the same scale as downloading, which was the main practice in 2011 with regard to documents found on Gallica and a practice that is more generally representative of “digital collectionism” observed among researchers today (Roustan 2016; Le Marec & Mairesse 2017). With the likely improvement of user interfaces and internet connections (as high-speed broadband spreads) and owing to the limits of “collectionism”—in the sense that the researcher still comes up against materiality and time constraints (problems of classification, storage space, and security for a critical mass of digital documents held) in a digital universe which initially appeared to offer immediate and limitless access—a new relationship to the interface emerges.
14In 2011, audiences motivated by personal or professional research preferred to bring the content of Gallica into their own world—the digital library on their computer, a paper archive, etc.—with low usage of the advanced research and document manipulation functions available. Among the new features and services that users said they would like to see, the ability to download the entire corpus came first (61%), emphasizing a “consumerist” relationship to Gallica which was seen as a document bank before being an access tool in its own right and a part of the activities of research, reading, and interpreting documents. This “denial of the interface” manifested itself first and foremost in the practices and representations of users motivated by specific research, who were categorized into two groups using agglomerative hierarchical clustering: “those with specialist hobbies” (28%) and “pros” (30%). Reading online, on the other hand, was characteristic of users who came out of simple curiosity for casual browsing or amusement. As interviews with intensive users of Gallica had previously shown (Amar & Touitou 2013; Amar & Chevallier 2014), the fact of keeping the digital document offline, usually by printing or downloading it in PDF format, was primarily motivated by a fear of not being able to find it quickly owing to a complex interface which they were not good at handling, whose search engine seemed to produce fluctuating results; but also through a desire to keep control of “one’s own” library and—in the case of personal digital libraries—where necessary add other digital sources to it, including those they created themselves (digital photographs). As for paper printing, which was still done “sometimes” or “often” by one in two Gallicanauts in 2011, this seems to happen when thorough reading is required, in accordance with the widely held opinion that reading online is not really reading (Amar & Touitou 2013; Amar & Chevallier 2014).
15In contrast, downloading the complete corpus fell by 61% to 41% in terms of user wishes in 2016; and this was the same for use as a tool for annotating documents (52% to 32%), and as a collaborative workspace (28% to 13%). This last decrease should be viewed against the rise in sharing Gallica documents on the internet: 40% now do this at least occasionally, first and foremost on Facebook, compared with 13% in 2011. The potential improvements presented in the questionnaire were also well received: they concerned document conversion tools (cutting, extraction, etc.) (chosen by 49%); and hyperlinks to other resources, of the BnF (47%) or its partners (44%). Those who welcomed the potential improvements already possessed a good level of knowledge of Gallica and made good use of its basic features: viewing the table of contents (58% do this “often”), changing the document display (43%), and, above all, zooming in (81%). It could be argued that although we observe good use of the interface, advanced features (such as advanced search, copying the document code, retrieval via an API, etc.) remain largely under-used, confirming a phenomenon regularly observed on internet search interfaces. But we should not be too quick to attribute this under-use to a deficit in user training in computer search techniques (Papy 2016) and thus to inadequate competence regarding interfaces. As the 2016 survey shows, Gallicanauts’ “expertise”—measured by frequency of use of the full range of advanced features—is not in any way correlated to statistics of age, socio-occupational status, or level of education, but exclusively to frequency of visits. Expertise is not, therefore linked so much to general competence but to a certain type of activity.
16This more positive perception of the interface recorded in the 2016 survey was certainly aided by the ergonomic development of the site: appreciation of the navigation and accessibility of document manipulation features thus saw an increase of seven points (from 27% to 35% “totally agree” responses), while general satisfaction saw a remarkable improvement, with the number of “completely satisfied” responses going from 36% to 48%. But this vision which focuses on the technical object and its developments, which go through stages to which the user adapts to varying degrees while remaining on the outside, benefits from being viewed from a different angle—one of human activity and interactions, whereby the technical object is seen as a partner and not as an obstacle (Denis 2009). The video-ethnography of regular Gallicanauts (Rollet et al. 2017) allowed us to observe usually open research (ie, not restricted to a document or precise piece of information, but, for example, on a theme or into a period) in its natural environment, shedding new light on this positive perception by drawing attention to two dimensions of Gallica use which could not feature in the questionnaire and very rarely in interviews. The first dimension is the role of the interface in structuring research and in the construction of a line of inquiry, in particular through the means of viewing and evaluating search engine results: the latter, which are open to surprises and disappointments, can lead to the user changing their initial strategy—by suggesting, for example, interim goals, or producing “fluctuating classifications”—and even, in certain cases, to “completely changing the direction of their argument.” “In this idea, search bar queries can be read as steps in an evolving thought process” (Rollet et al. 2017). The second dimension is the ecology of use of Gallica, that is to say its place in a wider context of activity that includes the body, other digital or material artefacts, but also other people (physically present or in the memory of the user): there is not only activity on Gallica, but also with Gallica, whereby the interface and the consultation tool (for e.g., a computer or smart phone) are not erased but fit into “an open, material and digital universe (from the notebook lying by the keyboard to Evernote or YouTube)” within activities which are mutually articulated and structured (Rollet et al. 2017).
- 10 It would be interesting to study how this “outdated” use has been completely overhauled today by th (...)
17Of course, using Gallica as a document repository—implying a heavy use of downloads via the interface, which was very significant in the surveys carried out in 2010-201210—does not rule out either changes of direction in research, nor the interplay of tools and activities; but the relationship to the technical object is experienced primarily negatively: the interface is that which we seek to free ourselves from as quickly as possible to carry out, away from it, the real activity of reading, thought, or even the simple organization of knowledge. What the interface finally reveals of itself to the “collector” or “collectioner” is its capacity for resistance (its slowness to respond, search engine “noise,” etc.). It acts only as a brake on collecting, just as it acts as a brake on document viewing when, for example, in the words of a user of the NumDam digital library in 2011, we “kill our eyes looking at a screen for two hours for every page turned” (Amar & Touitou 2013).
18In sequences of the video-ethnography, despite being restricted to a certain type of use and user (open research, regular use), the website becomes, in contrast, a partner in the research activity in an environment rich in interactions and the presence of various others (other people, and sometimes the “stubborn presence” of electronic messages). This recognition of the interface’s own contribution, which gives opportunity for complex activity in which action and cognition constantly collide and bounce off one another, facilitates the sustainable integration of the researcher online. For example, it is striking that the apparent impression of disorder in search engine results, which is a subject of complaint commonly referred to as “noise,” can be in some cases sought by the user, conscious instead not to “close” his or her research too soon (like Didier in the video-ethnography, who makes sure not to put brackets around his query). Regarding scrolling through a long list of results, which is just as potentially stressful, this is not necessarily perceived as a waste of time—a tedious and unnecessary activity—but rather as an activity to be perfectly mastered that gives a “course-grained” assessment of a domain or a type of document (Rollet et al. 2017).
19The Gallica digital library gives a chance to observe, through comparative analysis of different surveys and methodologies, the development of the relationship to digital archive material, both from the perspective of users’ reasons for usage (such as a strong commitment to carrying out personal research, as in the figure of the amateur) and that of their concrete activity (the interface as research partner and no longer as an obstacle to overcome). The newfound ease before the interface –whereby well-known flaws of the document portals or digital libraries become opportunities, whereby the user develops their own strategies to control the mass of material and noise, and whereby the digital document, seen in its original technical form as it was on publication, appears to lose its scope—could be a sign of an extension of what academic discourse has established as the “digital.” Nothing more could get in its way, and, as ease brought speed and productivity, other media and mediations of knowledge would be replaced. But this is not the case, and this pacified relationship with interfaces, which are no longer mysterious or troubling objects—not to mention the fears of superficiality, of a “flattening out” of knowledge, or of a loss of context or materiality so often heard in the surveys—is perhaps a paradoxical sign that they can find their rightful place in intellectual practices that are never simply “digital” as Joëlle Le Marec and François Mairesse have suggested. In fact their recent ethnography of ordinary scholarly practices reveals the opposite phenomenon, even—or indeed especially—with regard to the users most invested in the new tools: “Technological devices (still) seem to emphasize divergent trends more than they homogenize them” (Le Marec & Mairesse 2017). Rather it is the great heterogeneity of practices that is observed by the anthropology of knowledge, whether in the form of different moments in documentary research where we yearn for the object or make do with its image (Roustan 2016), or that of an open work space filled with things (Rollet et al. 2017).