Towards Digital Histories of Women’s Suffrage Movements A Feminist Historian’s Journey to the World of Digital Humanities

During the past decade, the amount of digitised material has exploded,1 but they are not available to all researchers in an equal manner. The entrance to the field of digital humanities requires cultural and technological capital which excludes or marginalises researchers who do not have the skills to conduct digital analyses by themselves or do not have access to the organisational support. According to Matthew K. Gold, it is research-intensive universities containing sufficient financial and human resources that have been able to embrace the digital turn.2 Again, this ability to focus on digital research and hire personnel to carry out the analysis has formed a ‘circle of good’ stabilising their status within the field. Other universities, not to mention individual researchers, have been less fortunate, but simultaneously digital analysis tools increase the expectations that we as scholars are expected to accomplish.3 Gender is one of the factors that seems to affect the ability of researchers to take part in the digital turn. Farida Umrani and Rehana Ghadially discuss the


Introduction
During the past decade, the amount of digitised material has exploded, 1 but they are not available to all researchers in an equal manner. The entrance to the field of digital humanities requires cultural and technological capital which excludes or marginalises researchers who do not have the skills to conduct digital analyses by themselves or do not have access to the organisational support. According to Matthew K. Gold, it is research-intensive universities containing sufficient financial and human resources that have been able to embrace the digital turn. 2 Again, this ability to focus on digital research and hire personnel to carry out the analysis has formed a 'circle of good' stabilising their status within the field. Other universities, not to mention individual researchers, have been less fortunate, but simultaneously digital analysis tools increase the expectations that we as scholars are expected to accomplish. 3 Gender is one of the factors that seems to affect the ability of researchers to take part in the digital turn. Farida Umrani and Rehana Ghadially discuss the aspect of empowerment that is connected to the access of computers for women by using the division between 'information rich' and 'information poor' people. 4 Even though their approach is connected to third-world countries and the use of computers in general, I find the division useful also in the context of digital humanities. Similarly, researchers within humanities and social sciences are nowadays divided into those who have skills and access to digital research methods and those who don't. Based on stereotypical gender role expectations (such as viewing technology as masculine coded) with which most generations of current historians have been raised, men have often better opportunities to explore the field even if the starting point is the same with their female colleagues. The reason for this, as Miriam Posner has pointed out, is that middle-class white men are more likely to have been encouraged to explore with computers at a young age than women or other marginalised groups. 5 At the same time, most present-day researchers are regardless of gender already to some extent participating in the digital turn as the mixing of traditional and new (digital) research practices have become a self-evident part of our presentday work as scholars, in the form of digital voice recorders, digital cameras and the use and analyses of digital texts and images. 6 In this chapter, I will discuss what is needed when a historical scholar with limited digital skills wants to take a step towards learning how to conduct digital analyses, towards becoming a digital historian. As a feminist historian, I will combine this approach with a discussion of the relation of feminist research and digital humanities. In line with practice in feminist research, I will be using a self-reflexive approach and asking how the increase in the understanding of digital methods influences our research questions in feminist history. Do digital humanities tools transform our work as feminist historians? How can digital analyses develop the field of gender history in general and the history of feminism in particular? Can a scholar who has limited technological skills engage with an informed and critical discussion with digitised materials?
Even though the main points of my chapter apply to all historical research, a focus on gender analysis is worth making as gender seems to have remained a rather limited category of analysis among digital historians. And although not all gender historians identify their work as feminist, there is a strong connection between the two, 7 which makes the discussion of the relation between feminism and digital analysis a valid starting point for this chapter. The intention is not, however, to claim that there is a clear difference between feminist history and historical research in general, but to participate in the discussion of the meanings of feminist approaches to digital humanities and ponder why in particular feminist historians should be part of these discussions.

Feminism and Digital Humanities
At first, the combination of feminist research and digital analysis may seem as strange bedfellows, but for the past decade feminist digital humanities research has been conducted in various fields in the Anglo-American world, in particular, and scholars have engaged in critical discussion of the relation between feminism and digital humanities. Some scholars have problematised this relation by deconstructing the gender-neutrality of digital analysis tools in order to find ways to overcome the divide between male producers and female users of computational tools. 8 Others have asked how the use of a large amount of data fits with feminist research that relies on gender-sensitive reading. 9 Scholars have also found similarities between feminist research and digital humanities approaches, such as collaborative research. According to Janine Solberg, feminist research and digital humanities may even form a fruitful pair since the idea of an ethical (feminist) researcher encourages the scholar to be open for multiple viewpoints and to position oneself as a researcher and conduct the research in a transparent manner (features that are also valued in digital humanities). 10 Discussion on a feminist approach to digital humanities has mainly focused on pondering how gender, race and other marginalising factors can be taken into account when compiling datasets and digital archives. In addition, the responsibility of feminist scholars to unsettle digital humanities' 'retro-humanist' practices that maintain a canonical understanding of what is relevant to digitise have been pointed out. 11 However, less has been written about the actual methodological practice of conducting a feminist digital humanities project. Nevertheless, there are some exceptions. In her insightful article on the US suffragette Frances Maule, Solberg, for example, points out how the new technology made it possible for her to find information about Maule, whose life was relatively unknown when she discovered her. At the same time, new information that she was able to find thanks to digitised material changed the interpretations of Maule's texts used by Solberg in her work. 12 Thus, digitised data can help us to find sporadic information of our research subject and combine these pieces of information more easily than previously. Mass digitisation may also widen our opportunities to find traces of people who have been marginalised in the past or purely forgotten, which is consistent with the core ideas of feminist research.
In spite of the existing literature on feminism and digital humanities, feminist digital history seems to be an under-discussed area of study. For instance, scholars of feminist historiography of rhetoric, Jessica Enoch and Jean Bessette, argue that feminists have used digitally born materials to study women's lives, but historians have rarely pondered how digital methods could widen their scope of study. 13 A few exceptions have appeared in feminist literary history, in particular, but the field is still narrow. This seems problematic because the digital turn has already started a revolution in history which will potentially profoundly change our scholarship and require us to learn new tools, as Alexis Lothian and Amanda Phillips have formulated. 14 Also due to this, gendersensitive historians should start to pay attention to the challenges and opportunities that the digital turn will cause in our field.
The argument follows Solberg as well as Enoch and Bessette, who have demanded more discipline-specific discussions on the role of digital analysis tools and digital research materials. 15 According to my understanding, a focus on gender history is particularly important because, as an already marginalised sub-field of a traditionally masculine discipline, it may not otherwise be able to answer the demands of financers that have started to highlight the importance of digital methods. In addition, gender sensitivity is needed to guarantee that digitisation of archival sources and other material such as print media or books do not focus solely on canonical pieces or notable people of history. Even though gender has been written into history increasingly since the 1960s, digitised collections often maintain gender bias when men's history tends to be viewed as more important. Furthermore, a move towards more digitally aware research may require that gender historians, alongside gender scholars in other fields, start to discuss how women's and other minority groups' abilities to conduct digital humanities research could be supported. However, the intention of this chapter is not to strengthen the essentialist notion of women as less capable than men of conducting digital analysis, but by using my own research field as a starting point to instead problematise what is needed to support scholars such as myself who do not have the basic digital skills, making it difficult to start on their own.

Taking the First Steps in the World of Digital Humanities
In Finland, the first computer started to operate in 1958 and a relatively rapid computerisation has taken place in the country since the 1960s. This has also had its effect on research. Historian Viljo Rasila was already writing about computer-assisted research in 1967 and used these kinds of methods in his work. 16 However, when I began my studies at university as a fresh undergraduate student in 1999, I did not own a computer, and neither did many other students at this time. Computer-assisted methods did not belong to the curriculum and in spite of the accelerating computerisation of the Western world, for many they remained primarily a tool for writing and for using publishing or photo programs. This applied also to my relationship with computers, which explains why I never learned to understand properly how computers work as operational systems. For me, computers remained tools that I used to write and I knew only as much of them as was needed to complete that task.
As a scholar who began her postgraduate studies in the mid-2000s, I was even able to conduct my PhD studies without ever hearing the words 'digital humanities' . I first became familiar with the field as late as in 2015 when editing an article on that matter for a Finnish peer-reviewed journal as part of my duties as a sub-editor. I became instantly intrigued; but for a person with limited IT skills; it felt overwhelming to even try to figure out how I could use the approach in research. As for many, I assume, the first push towards this took place after a year while I was writing a research proposal for a major financing body and tried to figure out how I could elevate the state of art so that my project would be successful in receiving funding. I started to read digital humanities literature and tried to understand what all this could mean for my project.
By reading the texts, I started to realise that digital humanities projects were often collaborative initiatives (that is, not everyone needs to know how to code). However, at that time, I was working as a visiting scholar at a Swedish university with a Finnish research grant, which meant I found it difficult to start looking for collaborative partners. How could I even start to look for them, I was asking myself. In short, I lacked the institutional support as well as technical skills that would have helped me to explore the field on my own.
For a year, every now and then, I read articles on digital humanities, some of which made more sense to me than others. I was, for instance, exhilarated to find out that distant reading could be combined with close reading, the latter of which is the method I am most familiar with. However, the basic question remained the same: How could I start to understand the process of the analysis? Simultaneously, I changed universities and my new colleagues helped me to find pages that offered guidance for different analysis tools, such as The Programming Historian, but it still felt difficult to start learning on my own. Luckily, in the spring of 2018, I was selected for the practical course on digital humanities organised by the project 'From Roadmap to a Road Show' led by Mats Fridlund. It turned out to be the first step towards understanding what digital humanities could actually mean for my historical research. As my project, I selected a smallscale case study of the use of the word 'naisasialiike' (women's suffrage movement) in the biggest national newspaper Päivälehti (1889-1904) / Helsingin Sanomat (1904-) at the turn of the 19th and 20th centuries.

Learning by One's Mistakes
Janine Solberg has argued that digital environments can be used as safe spaces to test our research ideas. 17 In her work, Solberg did not rely on big data, but used digitised material to trace pieces of information of her research subject. However, her argument also seems suitable for a scholar who combines a big data approach with close reading of a relatively small pool of data, as is the case in this chapter.
Previously, I had used the National Library of Finland's digital newspaper archive from time to time to look for information. However, I had only used the search option, without trying to familiarise myself with the platform. Due to this, I had the habit of writing down the texts that interested me and it was only when preparing the data for the course that I realised that an OCR view of the text would make it easier to gather the data. However, from the literature, I had learned that not all letters would necessarily appear the same in the OCR text as they were in the original. 18 This was painfully clear concerning the material from Päivälehti, in which the articles were published by using fraktur, an old German font type. For instance, the machine was unable to recognise the letter 'w' as such, but it was often written with 'm' or with a combination of two letters such as 'n' and 'i' . Also similar kinds of errors took place with other letters such as 's' , which had different kinds of typographical variations in the original text. Furthermore, wider spacing of words which was used to highlight certain words in the newspaper texts, such as surnames, caused trouble for the machine. More importantly, the machine was unable to recognise the newspaper columns which ran in an uneven manner at that time. This meant that the OCR view did not include texts in the order they were mentioned, and I had to copy the paragraphs manually. This included removing paragraphs that were not part of the article I was interested in and ascertaining that all the paragraphs of my article had been copied to the file. Due to these problems, the OCR view made it only slightly easier to prepare the material than my original manner of writing everything manually. While the newest articles might had had only a couple of errors, older texts were often impossible to understand based on the OCR. This meant that I had already read through some of the texts while correcting the OCR, making me familiar with the material. The same kinds of problems have also been noticed by other scholars who have problematised the idea of digital analysis as a rapid way to conduct research with newspaper material. 19 My first experience with any kind of a data analysis software took place in May 2018, when I participated in the earlier-mentioned digital history course. Based on my project abstract, in which I had suggested that I would use a statistical natural language processing tool called MALLET for carrying out topic modelling analysis of the data, I had been assigned to a MALLET group led by the digital historian Mila Oiva. In addition, Juho Savela was providing technical support. My first challenge on the course was to learn to understand what can be made when using the command prompt of my computer. After that, I learned some basic commands for MALLET which helped me to start playing around with the material. Thus, the course gave me a basic understanding of how the command prompt functions worked and what I as a researcher could do with the material by using MALLET. However, the process of writing this chapter has been a test in which I have used MALLET and my own computer as a safe space for learning more by using the 'learn by your mistakes' method. Gradually, this has deepened my understanding of the process, even though there are still many things I do not understand.
One of the major revelations during the process has been that combining digitised material with technologically assisted analysis needs suitable research questions. According to Solberg, digital tools change our ways of discovering, accessing and making sense of the past. To be more specific, digital environments can reorientate us 'both physically and conceptually' if we choose to be active technology users instead of remaining as passive users of them. 20 Similarly, Jacqueline Wernimont has defined the division of male creators and female users of digital tools as one of the critical questions that feminist digital humanities needs to address. 21 Furthermore, other feminist scholars have engaged in critical discussion of what is enough to make the field more diverse and whether the ability to code is a necessity for all digital humanities scholars. 22 In my case, the move towards a more active user of digital tools meant that my original research questions changed during the process. I realised that MALLET would not be the best option to trace the transnational influences of feminist ideas as I had originally thought. Instead, it offered a window towards the variety of topics connected to which the word naisasialiike was used. In the following, I will outline the process of carrying out the analysis as well as ponder whether digital analysis of a relatively small pool of newspaper articles can bring new information concerning the early feminist movement in Finland.

The Importance of Search Words
In the project plan, I outlined the research period to cover the years between 1889 and 1929. I used Päivälehti's first publishing year as a starting point for the search period because the first women's organisation Finsk Kvinnoförening -Suomen Naisyhdistys had been established five years earlier in 1884. 23 The period ends in 1929, when the New Marriage Law was approved in Finland, forming one kind of an end point for the early feminist movement. 24 When outlining the period like this, I assumed that it would consist of a reasonable amount of data that I could use in my analysis. Surprisingly, the number of texts using the word naisasialiike proved to be relatively low. The search brought 51 results, one of which was a list of literature for Christmas presents. Because the list contained only one item that was relevant for the theme, I did not take it into account. Other hits were mentions as part of larger articles dealing with various women's organisations and their gatherings or books that dealt with women's issues. Additionally, the pool of data consisted of few notifications for meetings organised by the Women's Feminist Union (Naisasialiitto Unioni), among others.
The search also brought to light other astonishing revelations. At first, the word naisasialiike seems not to have belonged to the newspaper's vocabulary at all, since the first hit was from the year 1896 (12 years after the first women's organisation had been established). Before 1900, the word naisasialiike had been used only five times and continued to be used quite rarely until the 1920s: between 1900 and 1920, it appeared 12 times. Thus, it seems that the word made its breakthrough in the 1920s, even though it was still used only occasionally. This is slightly surprising because the 1920s was a relatively quiet period in the Finnish women's movement compared to earlier decades.
One explanation for the concentration of the use of naisasialiike may be that, during the 1920s, it was used as a retrospective term to look back on the history of the women's movement. However, throughout the studied period, it appears also as an umbrella term that was used to refer to women's emancipatory demands of its own time. In other words, naisasialiike seems to have become a label that was used both by the women's movement activists and their opponents, and it was accepted by the newspaper's editorial office. Thus, the question remains: Why did the most active years of early feminism as a movement in Finland not cause wider coverage in the pages of Päivälehti / Helsingin Sanomat? The question is particularly interesting because the word naisasialiike is the one that is commonly used in the research to refer to the early feminist movement in Finland. Due to this, it could have been assumed that the word had appeared in public discussion at the turn of the century.
When looking for answers to the above-mentioned question, it is worth taking into account that the results would have been different if I had used different search words. For instance, Hieke Huistra and Bram Mellink have reminded us that a digital humanities scholar needs to choose the right search words to receive reliable results. 25 However, the problem is that a topic can be described with a variety of words that appear at different times, the meaning of which may change in different contexts and throughout the studied period. In Päivälehti / Helsingin Sanomat, the words 'feminismi' (feminism) and 'femi nisti' (a feminist) received 11 hits respectively between 1889 and 1929, but the usage of them took place mostly before the 1920s. The word 'naisliike' (women's movement), which refers to all kinds of women's organising (both feminist and non-feminist), received 255 hits, which suggests that issues relating to women's status in society were not yet directly connected to feminist organising at the turn of the 19th and 20th centuries. The most comprehensive word was 'nais asia' , which received 897 hits. It was used for the first time in 1890, after which it appeared continuously throughout the research period, suggesting that women's issue as a theme was part of the public discussion of its time, but it was not connected to a specific movement per se. An analysis of the usage of the word naisasia would, therefore, give us a more comprehensive understanding of the early feminism in Finland, but the cleaning of the material would also require a considerable amount of work, which was not possible in the scope of this chapter. Furthermore, such big data would have made it difficult to use this process as an opportunity to reflect on the relation between feminist research and digital humanities. Comparison of the usage of various words dealing with women's issues nevertheless reveals the development of terminologies which has been shown to be one of the strongest sides of big data analysis. However, as Alex Mold and Virginia Berridge remind us, these kinds of results also need to be contextualised and triangulated with other sources/traditional research methods in order to receive a more accurate understanding of the results provided by the digital analysis. 26 As pointed out earlier, a close reading of research material is one of the corner stones of feminist research, and digitised computer reading of big data seems to be in contradiction to this. One solution to overcome this conflict is to combine computer-assisted analysis with close reading of the material or parts of it, as Johan Jarlbrink, Pelle Snickars and Christian Colliander have suggested, among others. 27 Based on my small-scale project, a combination of distant and close reading is not only necessary to validate the results, but the combination also gives new perspectives on the material-one example of which is the connection between national and transnational discussion in Finnish feminism.
Previous scholars who have worked with archival material or used media texts in a more traditional manner have extensively shown how the so-called firstwave feminism was committed to national issues in Finland even though the early feminists, at the same time, had wide transnational networks. For instance, Alexandra Gripenberg saw women's emancipation as necessary for human progress and therefore it had to be strived for universally. Simultaneously, her work towards women's emancipation was tied with nationalism. 28 Both sides of early feminism also appear in my data, which I recognised while cleaning the material for the analysis. However, the digital analysis also revealed different nuances in the dynamics between national and transnational aspects of early Finnish feminism, as will be shown in the last section of this chapter.

How the Analysis Was Made
The project was started by preparing the dataset of 50 texts for the analysis, after which I conducted topic modelling with 10, 15 and 20 topics. At first, the topics produced by MALLET seemed like a foreign language to me and the fact that every round of analysis could bring different kinds of word lists was puzzling. Even though I was mechanically able to make the right commands, the ability to start the analysis required a new way of interpreting the lists produced by the computer. This I could not have done without the guidance of Mila Oiva, who patiently used her own research as an example to walk me through the process of shifting my way of thinking. Learning a new way of interpreting the word lists was not the only challenge: I also had some problems with stop-words. Some of them kept popping up in the topics even though I had added them to the list. However, the final round of topic modelling offered satisfactory clean topics even if they still contained some of the listed stop-words, such as the foreign words 'del' , 'und' and 'des' , as well as abbreviations such as 'klo' . Because the pool of data was small, I chose to make the analysis based on 10 topics, which brought the clearest image of the data (see Table 9.1).
Three out of 10 topics pointed out to transnational exchange of ideas with words that referred to foreign countries in general or by name and to nationalities or countries in plural. However, there were differences between these three topics. While the first one clearly referred to international connections in media in forms of news reports, the second one attached internationalism to the past of the women's movement and the third topic connected international connections and women's movement congresses, pointing to the transnational nature of the women's movement. Other topics were clearly national by nature, but nationalism became a marker for only one of them which included the word 'isänmaan' (nation's), for instance.
Otherwise, the topics emphasised meetings of various women's organisations and particular individuals such as Maikki Friberg. Four of the topics include words referring to men. Two of them seem to point to the negotiation between women's and men's roles, while others bring men's point of view to the women's question to the centre. This can be explained with the variety of writers and their relation to the women's movement. Texts that appeared in the regular column of the Women's Feminist Union were most probably written by women's movement activists themselves, as were some other texts published in the paper. However, other texts presented opinions of prominent men. Motherhood is part of three topics, whereas temperance appears to be part of only one topic. The emphasis on women's maternal roles seems accurate because, during the early 20th century, bourgeois women argued on behalf of a social motherhood locating motherhood as women's most important task in the society. According to Sulkunen and others, this bipartisan citizenship was supported by a majority of Finnish women by the 1920s. 29 On the contrary, the marginal role of temperance within the topics is slightly surprising. What do the above-mentioned outlines tell us about the data? First, as an unexperienced user of MALLET, as pointed out earlier, I began the process by staring at the list of keywords offered by the software without really knowing how I could interpret them. Due to this, I used the scattered ideas I had gained of the texts while carrying out the above-presented classification, even though I had not read all the texts with a similar intensity. Thus, a combination of digital analysis and a close reading of the material helped me to pinpoint the topics that might have otherwise remained unnoticed. 30 Second, the strength of the digital analysis is in its manner of presenting the data in a different form, which highlights certain patterns. In this case, it is evident that the discussion on the women's movement was conducted as part of the Finnish public sphere and nationally topical issues. Simultaneously, foreign countries served as a standard reference point and the women's movement appeared as a transnational phenomenon even though it was connected to national discussions. Third, the public discussion offered room also for men to define their stance towards the women's issue. Fourth, rather surprisingly, certain milestones in the development of women's status were not connected to the women's movement in the public debate. For instance, themes such as women's suffrage (1906) and the New Marriage Law (1929) did not raise discussion in which the word naisa sialiike had been used.
Thus, it is evident that digitised material has the potential to show us surprising results, features that we don't expect to find from the material, as Mold and Berridge have pointed out. 31 However, to be able to understand these results more profoundly, they need to be contextualised both thematically and journalistically. That is to say that computer-assisted analysis also needs a human to contextualise the results (an example of which are media texts that should not be seen as a number of separate articles, but instead as part of the publication context of their time). 32 For instance, the length of articles had great variation at the turn of the 19th and 20th centuries. Päivälehti / Helsingin Sanomat published short news and reports as well as extensively long congress reports, which were often several pages in length. Potentially, this affects the results as I assume has been the case with the discussion on temperance. Based on the topic modelling, temperance had a relatively minor role in the public discussion, but based on close reading of the data it was a recurring theme within the material. However, it was not discussed as part of the longer articles, but only mentioned briefly in other texts. Elsewhere, I have also argued the need to bear editorial practices in mind while historians use media texts to make interpretations of past phenomena. This is particularly important while using digitised materials that easily shadow the journalistic processes behind the texts by taking them out of the context. 33 In my opinion, contextualising may also form the bridge that brings digital humanities and feminist history closer together, moving them from being strange bedfellows to being a functional pair.

Conclusion
In this self-reflexive chapter, I have discussed my own road to digital humanities, a journey which has actually only just begun. I believe that my reflections correlate with those of many of my fellow historians and other humanists who have started their scholarly work before the increasing digitisation of the society and are now trying to figure out how the digital methods can be used. What did I learn when conducting my small-scale case study?
Based on my experience, I agree with Solberg, who has argued that digital environments create 'new ways of interacting with' the material. 34 I would like to add that, at least for scholars with limited digital skills, they offer an opportunity to conduct the research in a more self-aware manner, when every step of the process needs more thought than a traditional research day working with paper archives, for instance. For a feminist scholar, digital humanities may also serve as a channel for emancipation if the scholar chooses to actively participate in the process of analysis instead of relying on the results produced by IT support. However, to be able to do this, we need the support from our universities to focus on this kind of a large-scale project that also requires time for learning new skills.
My experience clearly demonstrates that conducting a basic digital analysis is possible even for a beginner if she receives sufficient support to carry it out. Additionally, the practice is the best way to increase one's understanding of digital analysis. When the understanding increases, the research questions become more accurate at the same time. Within the limits of this small-scale project, the results were not mind-blowing, but they merely strengthened the results of other scholars focusing on the intertwined relation between national and transnational in the history of early feminism. However, the data also reveals new and previously unresearched questions, such as the use and development of vocabulary relating to women's issues in Finland. Furthermore, results of big data analysis expose new ways of perceiving the material which may revolutionise gender history by revealing gender in places that previous research has been unable to grasp. By also challenging feminist scholars to take a step back and examine the material from a distance, digital humanities has the potential to change our understanding of gendered patterns in the past.
These questions require more sophisticated digital analysis than has been possible in this chapter, but it is an inspiring direction towards which I hope to be able to move, alongside other feminist historians in the future. One way to do this is to develop grassroots digitisation projects in which gender, race and other marginalising factors could be taken into account when selecting the objects of digitisation. These kinds of projects have the possibility of developing the field by producing more localised and situated data collections that challenge the history we are writing and offer a broader participation in digital history work also for those with basic technological skills.