Models of Argument-Driven Digital History

In the past half decade, prominent historians have quantified scholarship on women’s history to analyze the state of the field. In a 2004 publication, Gerda Lerner tallied 720 recent U.S. women’s history articles, books and dissertations from the Journal of American History’s list of Current Scholarship by time period and theme. In her 2006 book, Judith Bennett quantified scholarship in various women’s history journals and conferences to show an overemphasis on more recent time periods. In 2007, Merry Wiesner-Hanks analyzed the regional focus of Berkshire Women’s History Conference papers and Journal of Women’s History publications in her essay on the place of women, gender and sexuality in World History.¹ Other scholars have published state of the field pieces that provide qualitative analyses of women’s and gender history.²

Quantitative and qualitative analyses provide these women’s historians with a sense of the developments in much of our field and point to remaining shortcomings and opportunities. This article builds on the desire to understand the scope and achievements of women’s history by providing a more comprehensive quantitative overview of the place of a large segment of women’s history within a large segment of historical publishing between 1985 and 2005. Our analysis examines more than a half million abstracts from two widely used article databases that cover historical study since c.1450. We use these to explore the place of women and women’s history within the historical field in general, and within regional and chronologic subfields in particular. While women’s history reflects some overall shifts in the historical profession, it also has a demonstrably separable trajectory in terms of its expansion, content, and regional/chronological foci.

We then move to exploring the approximately 31,000 abstracts that we identified as women’s history-focused. We look at shifts in article publication patterns between 1985 and 2005 to see how the field has – and has not-- changed. We directly explore the content and range of post-c.1450 women’s history across these two decades, and question whether a seemingly more twentieth-century focus of women’s history is a by-product of overall developments, a more complicated transformation, or less of a shift than it might originally appear. We also track the regional and chronological subject area variations within women’s history abstracts, and suggest that this be a starting point for discussions of the place of women’s history within various subfields. In the paper’s final section, we select a single topic area – sexuality -- for deeper analysis. We identify broad subject areas covered within sexuality histories and trace their variations across time and place of study, suggesting the value of this kind of quantitative exploration.

Thus, as our title suggests, we are addressing several “whats, wheres, and whens” of two decades of post-c.1450 women’s history: what place does women’s history have in the field at large and what kind of subjects are included within women’s history; where in various regional histories does women’s history most frequently appear; when do publications on women’s history increase or decrease in numbers as well as when, chronologically, women’s historians most focus their efforts.

The “sometimes why” is a more complicated venture. At various points we forward possible explanations, grounded in historiographic scholarship, for the trends we have identified. But for the most part, our broad analysis of tens of thousands of abstracts at a time, aims to recognize basic patterns that raise as many questions about the constitution of the field of women’s history as they answer. We mean this article to start a conversation about women’s history as revealed through a large-scale quantitative analysis that provides hard facts, rather than qualitative impressions, about the field. We conclude by proposing questions about how we can benefit from this type of interdisciplinary collaboration, what it might suggest about how we understand the place of women’s history in the field, and the role of the profession and academic institutions in the support of various kinds of historical study.

Methodology

The main source for this article is 513,259 substantive abstracts of articles and essays published between 1985-2005 in America: History and Life and Historical Abstracts databases. America: History and Life (AHL) focuses on the history of the geographic regions that now make up the United States and Canada. Historical Abstracts (HA) covers the history of the world outside of North America from approximately 1450 onward. Some of the most widely used abstract databases in the historical profession, AHL and HA include English and non-English language publications, and gather historically-oriented articles from more than 3,000 scholarly journals, including women’s studies publications. These databases are not perfect representations of the historical profession by any means -- the exclusion of pre-1450 scholarship may be one of the biggest handicaps -- but they still provide a much larger corpus of information than heretofore examined.

We analyze this collection of abstracts using two main methodologies. First, word frequency counting (how often a given word occurs) reveals how and when terms related to women appear in historical abstracts.³ Second, topic modeling, a computer science data mining technology that is arguably the state-of-the-art model for text document collections, allows for a more complex subject analysis.⁴ Topic modeling learns subject categories without a priori subject definitions. Unlike traditional classification systems where texts are fit into preexisting schema (such as Library of Congress subject headings), topic modeling establishes a comprehensive list of subjects through its analysis of the word co-occurrences throughout the corpus. The content of the documents—not a human indexer—determines the topics collectively found in those documents, arguably making topic modeling’s subject indices far more sophisticated than human classification.⁵

Throughout our analysis, we employ a broad definition of women’s history. We include work that focuses on individual women, on women as a group or groups, and on the power dynamics of patriarchy that some have shorthanded with the term “gender.” We do not single out feminist scholarship from that which addresses women’s experiences without an interest in patriarchal power structures. In short: we use “women’s history” expansively, to encompass all kinds of scholarship that addresses women’s lives, experiences, and the societal beliefs that surround them.

We identified this broad field of women’s history by included abstracts that ranked highly in women’s history topics from our initial topic modeling runs; those that were published in women’s studies or women’s history journals; and those that the AHL/HA databases had given the terms of “women,” “woman,” or “gender” as a subject heading.⁶ Together, our combination of subject word, topic model, and women’s studies journals resulted in identification of about 6% of abstracts -- 30,891-- as being substantially focused on women’s history.

This chart, which did not make the final cut for the original article, gives an example of what topic modeling could show about change over time.The topic area that I labeled “Race” has the mostly likely words listed at top of chart and traces its increase from 1985-2005 as a proportion of all historical publications. This chart was created from a 400-topic topic modeling run of all abstracts in order to get to this level of specificity.

See also comment on The Methodological Challenge.

Women’s History in the Field at Large

The amount of scholarship on women and gender within the historical profession has, not surprisingly, grown since 1985. Our analysis shows that this growth has neither been steady over time, nor have particular historical time periods and regions seen equal amounts of women’s history scholarship. Judging by an array of word counts, male figures still remain the overwhelming focus of historical study. In addition, several scholars have suggested that women’s history disproportionately focuses on very recent and on U.S. history topics. Here we show that some of this concentration may actually be more a reflection of general scholarly trends than specific to the field of women’s history.

Publications related to women’s history have been increasing since the 1980s. Figure 1 shows that women’s history articles accounted for about 4% of all abstracts in 1985, and grew to about 8% by 2005. Some of this relative growth resulted from an increase in absolute numbers of women’s history publications and some from a decline in the overall number of historical articles. In 1985, about 27,500 historical articles were abstracted in AHL and HA. Of these, about 1,058 related to women’s history. By the second half of the 1990s, women’s history articles began leveling off at the 1,500 per-year mark, while history articles overall continued to decline -- to about 19,000 by 2005.

Despite overall increases, the growth rate of women’s history abstracts seems to have slowed significantly by the second half of the 1990s; the small relative increases after that are due to an overall decrease in historical abstracts generally, not to an increase in absolute numbers of women’s history abstracts. Do we interpret this as a positive: women’s history is becoming a bigger part of history? Or, given that women’s history-related abstracts still only account for, at most, 8% of all history article abstracts, is the slowdown in absolute numbers of women’s history scholarship a cause for concern? At the very least, we can quantitatively confirm that women’s history is not anywhere near to a majority of publications.

In the article’s original version, this figure was expressed as “Frequency in __ words” which produced what looked like a declining curve. For the revision, we inverted the chart to be a more user-friendly by calculating frequency per 1000 words, which created an upward curve that reflected the increased use of “gender” as a descriptive term. This was one way we tried to make the charts, figures, and math accessible to historians.

I tried to repeatedly remind readers that this was an analysis of abstracts, and so one step removed from the articles themselves. This was particularly important when analyzing the word frequency of “gender.” I suspect that the dramatic rise of the term “gender” in abstracts reflected shifting disciplinary jargon and may overstate a shift in the content and approach to women’s history.

While women’s history articles have been increasing since 1985, their content (or at least the identification of their content) has changed dramatically over time. Most strikingly, the usage of the word “gender” increased exponentially (See Figure 2). In 1985, “gender” occurred less than 1 in every 10,000 words in abstracts. By 2005, “gender” occurred in about every 1,250 words – an eight-fold increase. In comparison, the use of the word “women” grew by only 33% over this period. As most women’s historians know, the discussion of “women” quickly took a back seat in the late 1980s and early 1990s to the use of “gender” in historical scholarship.

Some of this expansive use of “gender” undoubtedly reflects the introduction of the term as a powerful analytic category in the 1980s.⁷ Attention to gender undoubtedly also reflects (and helped to produce) a shift toward cultural histories of power relations, rather than social histories of women’s lives.⁸ Indeed, the frequency of the word “cultural” doubled between 1985-2005, while the word “social” saw a steady (though not as dramatic) downturn. It may also be that abstracters are relying more comfortably on “gender” as a catchall explanation of complex arguments about status and power that might have been classified with other terms a decade earlier.⁹ Word frequency statistics bear this out: words such as “patriarchy” and “sexism” show no significant increase over this time period, and “feminism” shows little increase after the 1980s.¹⁰ Perhaps this reflects less attention to terms that might be seen by mainstream readers as more strident versions of feminist history: “gender” has been naturalized to a generic identifier less associated with activism (witness the regular appearance of “gender” as a checkbox category on institutional forms), whereas “patriarchy” and “sexism” still carry the imprint of a particular activist movement.¹¹

Like the percentage of women’s history abstracts generally, the increasing use of the word, “gender,” also leveled off by the mid-1990s. We certainly hope that this does not mean that women’s and gender histories have already seen their heyday. A more positive interpretation may be that women’s history is becoming more integrated into diverse topics, so that the women’s history content is not fully explained or easily identified in brief abstracts. It may also be possible that as women’s history becomes mainstream history, abstractors see less need to mark such scholarship as focusing on women.

Still, this leveling off also raises several questions for women’s historians and the profession more generally. First, what is an appropriate ultimate level of women’s history publications within historical journals? Can we numerically quantify when women’s history has “succeeded” and is suitably represented in historical publications? Should attention to women appear in half of all historical studies? Should a certain percent of studies include a feminist or gendered analysis? Second are more profession-based questions: do publication levels fairly reflect the percentage of women’s historians? If not, are there material factors – likely related to women’s place in the profession -- that might account for any discrepancy? Women’s historians are, of course, not all women, but they are disproportionately likely to be female, and thus the field is more impacted by professional equity issues than many other thematic fields.¹²

Moreover, the overall increase in work on women’s history does not translate into equitable treatment of women and men in historical studies. Women do receive more attention as a group or subject of study in overall abstracts. The word “women” is used almost six times as often as the word “men.” This likely reflects the scholarly interest in women as a category versus the appearance of individual men as incidental subjects of historical inquiry, without analytic attention to gender identity.

But what about attention to individual male and female subjects? The comparative use of some objective and possessive personal pronouns, “her,” “hers,” “him,” and “his,” makes clear that individual men are still implicitly the focus of the majority of historical scholarship. Together, all of these pronouns are used over 200,000 times in the half million abstracts. Yet only 14% of these uses are for “her” or “hers.” While women, as a category, are a subject (perhaps even an over-essentialized subject) of analysis, historical scholarship still appears to be largely focusing on men’s activities as a matter of course – not necessarily to analyze men as gendered beings, but as default individuals of study.

Nevertheless, analytic studies of masculinity studies have unequivocally been on the rise.¹³ The appearance of the term “masculinity” shows a dramatic trajectory –from a virtually non-existent frequency of`1 in about every 90,000 words in 1985, to appearing more than once every 2000 words by the year 2000. This upswing reflects the ways that historical articles have begun to focus critically on men beyond their incidental appearance. This increase likely suggest a more widespread interest in fully understanding all aspects of gendered power dynamics. But does the increase in masculinity studies relate to the slowing expansion in women’s history in the new millennium? We should be careful not to posit a zero-sum-game of interest in gender-related studies, but it may be useful to consider the relationship between feminist, women-focused, and broadly-constituted gender histories.

Turning to a regional analysis of history abstracts shows that a higher percentage of articles are being published on women in North America than in the rest of the world. Almost twice as many North American abstracts as Non-North American abstracts include either “woman,” “women,” or “gender” somewhere in the entry (almost 10% v. about 5%, respectively). As Table 1 shows, a variety of words relating directly to women are almost twice as likely to appear in North American abstracts as non-North American abstracts.

Table 1. Select Women-Related Word Frequencies in Overall Abstract

These word choices mark the larger presence of women’s history abstracts within North American history overall. Women’s history abstracts are twice as big a proportion of North American abstracts (almost 9%) as non-North American abstracts (slightly more than 4%). This has been fairly consistent over time of publication: in 1985, women’s history-focused abstracts accounted for over 6% of all North American abstracts, but just over 2% of non-North American abstracts. In 2005, women’s history abstracts had expanded to almost 11% of North American abstracts and 6% of non-North-American abstracts. We can view these increases as positive or negative trends: on the negative side, North American women’s history was about the same proportion of it regional field in 1985 as non-North American women’s history was two decades later. On the positive side, the amount of women’s history on non-North American topics has tripled in two decades.

Table 2 compares the regional breakdown of overall abstracts and women’s history abstracts in more detail. North American abstracts account for just over 1/3 of overall abstracts, but more than half of abstracts on women’s history.¹⁴ In contrast, several non-North American geographic regions are particularly underrepresented in women’s history. Work on Eastern Europe/Soviet regions shows the biggest discrepancy, accounting for almost 17% of all abstracts, but less than 7% of women’s history abstracts. Western Europe and, to a lesser degree, South Asia, are both overrepresented in women’s history percentages in their fields.

Source availability, feminist activism, sexism, and acceptance of women’s history as a legitimate field of study all contribute to these regional variations. In the case of Eastern Europe, scholars have already suggested that such underrepresentation results from a significant “infrastructural vacuum and institutional resistance” to women’s history. This is, perhaps, the case in other regions as well. And the reverse may also be true – it would be worth investigating whether North American historians or journal editors are comparatively more supportive of women’s history.¹⁵

Table 2: Regional Distribution in Women’s History versus Overall Abstracts

We might think that the overrepresentation of North American scholarship on women’s history can be explained by chronology. If women’s history tends to be disproportionately concentrated on the most recent centuries, a country with a relatively abbreviated early history would be likely to have an increased proportion of articles related to women’s history. However, when we compare time periods addressed in women’s history abstracts and overall abstracts from c. 1450-20^th century, it does not seem that women’s history is significantly more recently-focused than the profession as a whole.

As Table 3 shows, the twentieth century accounts for the overwhelming majority of all historical abstracts from c. 1450-onward.¹⁶ However, women’s history abstracts are not disproportionately focused there– if anything women’s history is slightly underrepresented in twentieth-century scholarship. Work exclusively on the twentieth century accounted for about 57% of women’s-history-focused articles, and about 63% of overall articles. Still, this analysis supports Judith M. Bennett’s conclusion that women’s history and overall historical scholarship focus primarily on post-1800 time periods.¹⁷

Our larger analysis adds to Bennett’s by teasing out some of the differences between post-c. 1450 women’s history and scholarship more broadly. Rather than unrelentingly focusing on the very recent past, women’s history abstracts focus more on the nineteenth century than does overall scholarship–about 16% of women’s history abstracts but just 12% of general abstracts focus exclusively on the nineteenth century. Likewise, women’s history articles are more likely to focus on turn of the twentieth century period than does general scholarship (6.7% v. 3.7%), which focuses more broadly on the first half of the twentieth century.¹⁸

Table 3: Distribution of Abstracts Over Various Time Periods

*Twentieth-century percentages include 1900-1999 dates, as well as 1900-1945, and post-1945 categories.

Bennett has also explicitly encouraged feminist scholars to avoid taking refuge within our historical era of expertise. However, women’s historians seem reluctant to write across time periods: overall scholarship was much more likely to focus on three or more centuries than was women’s history.¹⁹ It may be that women’s historians, already all-too-often battling the perception of “advocacy” or “ahistoric” scholarship, take great pains to stick closely to our historically-specific evidence. Yet we risk this making us timid about claiming expertise across wide swaths of time or place, and in so doing, we may be limiting the significance of our findings.

As we might expect, non-North American general abstracts are much more likely to focus on earlier time periods than are North American abstracts. Less than 4% of North American abstracts focus on the c.1450-1799 time period, compared to 12% of non-North American abstracts. In addition, non North-American women’s history is actually more likely to focus on earlier time periods than general scholarship: more than 15% of non-North American women’s history abstracts focus exclusively on c.1450-1799. Correspondingly, non-North American women’s history actually focused less on the 20^th-century than did overall non-North American abstracts (51% v. 69%).

The limits of AHL and HA databases restrict us from including ancient and medieval history in this analysis. However, within these limitations, women’s history is not disproportionately more recently-focused than history as a whole. Rather than women’s historians abandoning earlier time periods, non-North American women’s history appears to have focused more on c.1450-1800 topics than has the field at large. Still, because women’s history is a bigger percentage of North American history – which is the biggest overall field -- these chronologically earlier studies may not seem to hold as noticeable a place in women’s history publications.

Focusing on the topics that account for women’s history within all historical abstracts confirms that women’s history publications make up a larger part of North American history abstracts then those focused on the rest of the world. Our topic model run of the half million abstracts identified seven major subject areas focused primarily on women and gender (Table 4). Within every area, women’s history made up a bigger part of North American-focused scholarship. Thus, part of the reason that U.S. women’s history can seem to dominate overall women’s history overall is due to women’s history’s bigger place within North American history than within other regions.

Women’s history-focused topics account for almost 3% of the content of all historical abstracts.²⁰ Because topic modeling identifies the content overall, rather than counting each abstract that mentions a women’s history topic as being exclusively about women’s history, this c. 3% represents the content of all scholarship that focuses on women’s history to the exclusion of other topics. In other words, topic modeling can separate out the parts of women’s history abstracts that simultaneously address other scholarly subjects (e.g.: politics, religion, labor) to focus exclusively on the women’s history language. Thus, 6% of all abstracts may have significant women’s history content, but only about 3% of the text of all abstracts focus explicitly on women’s history – concretely showing the full integration of women’s history with other thematic areas of study.

Table 4 explores these thematic areas of women’s history within historical scholarship. Work on feminism and suffrage is one of the largest topical categories, followed closely by scholarship on more intimate details of women’s lives: families and kinship, and marriage and sexuality. Literary subjects – whether biographies of notable women or literary approaches to women’s history – are also well represented. How are we to evaluate this broad variation of fields that make up women’s history scholarship? Should there be more scholarship overtly on feminism or less on personal lives – however we might define those two categories? Rather than presuming to dictate what women’s history should consist of, we note that even at this high level vantage point, there is a significant spread of topics across multiple aspects of women’s lives. We also note that abstracts likely focus more on methodologies and source than full text, which may account for the prominent percentage of literature and biography-related content.

Table 4: Major Women’s History Topics in Overall Abstracts

There is no doubt that women’s history accounts for a diverse and growing proportion of historical scholarship. Within this growth, however, its rate of increase has been slowing. An increased focus on “gender” in abstracts partly represents changed scholarly interests, but likely also reflects a changed shorthand for women’s history. Even though references to women, as a group, far outnumber mention of men as a category for analysis, the overwhelming majority of abstracts still talk about individual men, not women, as historical subjects. Regionally, by multiple measures, women’s history holds a significantly larger place in the North American and Western European historical fields than in many other regions. However, women’s history’s seemingly-disproportionate focus on more recent time periods appears to be a reflection of the historical profession as a whole. In the next section, we focus more in depth on women’s history abstracts to further analyze the extent to which these trends relate to women’s history rather than larger trends within the profession.

The Field of Women’s History

Here we explore the 30,891 abstracts that focused significantly on women’s history, exploring how the picture changes when we look within women’s history abstracts, rather than at women’s history as part of the historical field as a whole. We show that women’s historians speak with a regionally-shared language, and that some regional variations in content relate more to the place of women’s history within various subfields than the content of women’s history scholarship. We continue to dissect the notion that women’s history has developed a twentieth-century bias and suggest that it is more complicated than a simple shift toward scholarly emphasis on the most recent time periods. In terms of absolute numbers, women’s history has not turned away from c.1450-1799 time periods; rather work on the post-1800 period has increased, therefore changing the overall chronological balance.

Table 1 showed that women-related words appeared far more frequently in North American abstracts than in abstracts on the rest of the world. This regional difference largely disappears within the 30,891 women’s history abstracts (Table 5), suggesting a trans-regionally shared language with which experts talk about women’s history. Thus, the broad regional differences in overall word frequencies appear to have more to do with the place of women’s history in different regions and less to do with the amount of work being done within women’s history. Still, some of this lack of variation is likely due to the flattening of complex issues into abstract-friendly terms, and it does not mean that abstracts that focus on women’s history in different regions and time periods take identical approaches to the field.

Within women’s history abstracts, we can see an array of regional transformations between 1985 and 2005. In the mid-1980s, women’s history was significantly dominated by North American scholarship, which made up close to 60% of all abstracts. But by the mid-1990s and continuing through the first years of the new millennium, women’s history had become equally focused on regions outside of North America.²¹ Of course, equality between one continent, made of two nations, and the rest of the world is not exactly parity, but it does suggest the expansion of women’s history beyond its earlier regional foci.

Western Europe was overwhelmingly the largest non-U.S. regional focus of women’s history abstracts (Figure 3). While both areas' article abstracts increased their absolute numbers fairly steadily over the twenty year period (with some leveling in the new millennium), this was a much bigger proportional increase for Western Europe – it increased about 100% over its starting numbers, compared to only about a 50% increase for North America. So while women’s historians may be turning to non-Western regions, the attention to Western history is still continuing to increase substantially.

Outside of Western Europe and North America, each regional field’s women’s history abstracts likewise increased over time (Figure 4). One of the most dramatic increases occurred in East Asian history: in 1985, it was barely the fourth highest number of abstracts; by the mid-1990s, it had more than doubled to have the highest number of abstracts of any regional field of women’s history outside of North America and Western Europe. Most regions seem to be declining in the first half decade of the twenty-first century, with African women’s history abstracts showing the largest drop. Overall, scholarship still increased in each of these regions, showing the most pronounced gains occurring in the first half of the 1990s and then leveling after 2000.

Table 5: Select Word Frequencies in Women’s History Abstracts

The chronological breakdown of women’s history has recently received a great deal of analytic attention. While abstracts on all time periods of women’s history have increased along broadly similar trajectories since 1985 (See Figure 5), the most marked increase has been in nineteenth – not twentieth-century -- women’s history.²² In fact, the numbers of twentieth-century women’s history abstracts actually increased the least of any time period between 1985 and 2005. Twentieth-century women’s history abstracts grew to 1.4 times their 1985 number, while c.1450-1699 century abstracts grew to 1.7 times; eighteenth-century grew to 1.9 times and nineteenth century grew to 2.1 times their 1985 numbers by 2005. Alternatively, looked at as a percentage of women’s history abstracts each year, twentieth-century women’s history has actually declined – from 75% of all women’s history abstracts in 1985, to 68% in 2005.

Thus, twentieth-century women’s history publications have not been increasingly dominating the scholarship on post-1450 women’s history. But there are several reasons why women’s history may still appear to be disproportionately focused on twentieth-century history. First, abstracts on twentieth-century history overall far outnumber all other time periods put together, meaning far more absolute numbers of women’s history publications on the twentieth century. Second, overall abstracts that are identifiably on the twentieth century have declined fairly significantly since 1985 – from a range of 16,000 abstracts/year for 20th-century scholarship in the late 1980s to a low of below 12,000 by the twenty-first century. In contrast, all other time periods have held steady or declined only slightly. Thus, even though women’s history on the nineteenth century has increased the most, women’s history has become a much larger percentage of overall twentieth century scholarship than of other time periods.

This means that even though the twentieth century remains the largest focus of history as a whole, it does not appear that women’s history has any more disproportionately focused on that time period; the increases in 20^th-century women’s history have not outpaced the increases in other time periods. Indeed, considering the limitations that face historians of periods predating the twentieth century, women’s and gender historians have become increasingly creative in finding and mining primary sources related to women throughout history.

Thus far, we have focused on the amount, not the content, of women’s history abstracts. Table 6 identifies, through topic modeling, the broad subject areas of study within women’s history.²³ In contrast to the broad areas of women’s history identified within the larger historical canon (Table 4), the subjects within women’s history clearly cover many specific thematic fields. The largest topical area focuses, as does much general historical work, on various regional and national specificities. Other topics range from social to political to cultural to intellectual histories. As Gerda Lerner noted in her study of U.S. history scholarship, work on literary subjects (such as much of the Literature and Biography topics) remains a substantial segment of women’s history publications.²⁴

It is also worth remembering that the topic model creates multiple categories per abstract, so these topics measure the percentage of overall content of the abstracts, not individual topics per abstract. For instance, an abstract on labor and family in Latin America would count toward each of those three subject/regional areas in proportion to the number of words in that abstract that were statistically assigned to each of these topics. Thus, rather than marking each abstract with an exclusionary set of labels, each abstract contributes to a variety of topical areas in proportion to the amount that that abstract focuses on each area. This can produce more realistic subject categorizations and challenge the sometimes-mainstream image of women’s history as not intrinsic to other subfields.

Table 6: Broad Topic Areas in Women’s History Abstracts

There are some broad regional variations in these topics: Not surprisingly, non-North American abstracts are more likely to focus on regional issues than are North American abstracts, while Race, Slavery and Civil Rights appear three times as often in relation to North American abstracts. More specific regional differences suggest an array of areas for future analysis. African women’s history focuses less on marriage and family than other regions, but words such as “power” “status” and “traditional” occur comparatively frequently. Commentary on religion was most common in Middle East women’s history (about 19% of the focus of its abstracts, compared to less than 6% for any other region). Middle East history also focused strongly on the study of feminism. East Asian history was most likely to focus on marriage and families. Latin American women’s histories focused more than most regions on political movements, as well as labor and marriage. Eastern Europe/Soviet abstracts were one of the most likely to reference “tradition” and focus on inheritance and households. South Asia scholars frequently addressed feminism and political movements, and were also the most likely to use the word “cultural”-- perhaps partly because of the importance of literary and post-colonial scholars doing South Asian women’s history. Work in North American history was one of the least likely to focus on marriage and family, and among the most likely to focus on sexuality, race, labor, and the term “public.” Finally, Western Europe focused far more on families, households and status than did North America.

Even though, as Table 5 showed, women’s history abstracts use similar levels of women-focused language, this more complicated analysis of what women’s historians address suggests that different thematic emphases predominate in various regional specialties within women’s history. There are undoubtedly numerous reasons for these variations (e.g., source availability, disciplinary guidance), but it does seem that women’s historians are responding to their regions' particular historiographic concerns.

In terms of chronologic comparisons, generalizable topics of study best traversed chronologic distance. Tables 7 and 8 compare the largest individual categories in distant and recent time periods. Chronologically distant women’s history abstracts seemed to focus on more transhistoric topics, such as family and households, women, gender, or religion. In contrast, three of the four biggest topics in post-1945 abstracts focus on issues particularly pertinent to twentieth-century life, such as women’s relation to political parties, civil rights, or employment discrimination. Ironically, the specificity allowed by twentieth-century sources may encourage more parochial topics; distantly-focused women’s historians may be more likely to focus on topic areas that could be of broad scholarly applicability.²⁵

To summarize, the field of women’s history continues to expand in multiple directions. Even though all regional areas of women’s history have increased since 1985, U.S. and Western Europe still dominate the field. How do we begin to explain the complex factors that may have accounted for regional variations in amount and subject of study? To what degree are they about reception to these fields by women’s historians; reception within these regional fields to women’s historians, or artifacts of particular tipping-point moments in training or individual scholarship? While an array of recent reflections on the state of women’s history has endorsed the need for the continued internationalization of women’s history, scholars might want to first more fully explore the limiting factors that produce these variant trends.

In terms of the time periods covered by the AHL and HA databases, the picture is a complicated one: women’s history focuses on more recent time periods, but no more so than the field as a whole. Indeed, by several measures, twentieth-century women’s history scholarship appears to have increased less than some earlier time periods. But the domination of overall twentieth-century scholarship can still leave women’s history looking like a very recently-focused enterprise. Finally, a subject area analysis of women’s history shows coverage of a diverse range of historical subjects, with differing emphases along lines of regional historiographic concerns. Despite these variations, broad topic areas of c.1450-1800 scholarship could easily provide productive intellectual connections to other chronologic times and places, and we hope scholars will increasingly consider chronologically and regionally broad thinking when doing women’s history.

Within Women’s History: Case Study of Sexuality

Historical studies of sexuality have developed, in part, out of work on women and gender, and remain firmly related to the field.²⁶ Here we take sexuality as a case study to further explore the trends and patterns revealed by topic modeling. We break down the broad subject areas of sexuality histories, examine the degree to which studies in sexuality topics vary across time periods and regions, and ask how the topics within sexuality vary along these two axes. We suggest that both the amount and content of sexuality histories appear to vary more by region than by chronology.

Like most women-related subjects, sexuality appeared to be discussed substantially more in North American than in Non-North American abstracts overall. Of more than a dozen explicitly sex-related terms, only “prostitution” appeared more frequently in overall non-North American abstracts– and then, only slightly more often.²⁷ More informal (and less supposedly-scientific) terms on homosexuality show some of the biggest regional differences: “Lesbian” and “gay” occur three-to-five times more often in North American as non-North American overall abstracts.

When we examine sexuality-related words in women’s history abstracts alone, much – but not all -- of these differences begin to disappear (Table 9). For instance, “sex,” “sexual” and “sexuality,” combined, appear in North American and non-North American women’s history abstracts roughly equally, even though they appeared almost twice as often in overall North American abstracts as overall Non-North American abstracts. Likewise, “homosexual” occurred almost evenly in and outside of North American regions within women’s history abstracts, even though the term was almost twice as common in North American as non-North American overall abstracts. Some terms for same-sex sexual behavior remains unevenly distributed within women’s history abstracts: “queer,” “gay” and “lesbian” show some of the most stark regional differences. Future research could explain the degree to which such variations reflect differing historical subject foci and abstracting norms, whether terms such as “queer” carry unacceptable meanings in some contexts or language translations and, more generally, the complexity of differing local and global understandings of same-sex sexuality. In contrast, and following lower discrepancies in overall abstracts, “prostitution” and “rape” appear more frequently in non-North American women’s histories, perhaps suggesting more overall attention to particular kinds of (heterosexual?) transgressive sexual behavior outside of North America.

The sexuality–related subject content of women’s history abstracts more comprehensively shows the distribution of subject areas within history of sexuality. Table 10 summarizes the five main sexuality-related topic areas within a 40-topic breakdown, alongside the keywords that most represent those topics. Together, these topics account for 9% of the text of women’s history abstracts. Work on sexuality generally, is the largest topic, while work on same-sex sexual behavior and reproduction are the smallest. The second largest category focuses on sexual regulation and social control of sexuality, likely reflecting sexuality study’s longtime focus on transgressive behavior and use of legal records for evidence on prohibited sexual practices.

Table 9: Sex-Related Word Frequency in Women’s History Abstracts

Table 10: Sexuality Topics Within Women’s History Abstracts

Figure 6 shows the regional variations in these sexuality topics within women’s history abstracts. Work related to sexuality makes up the biggest proportion of Western European abstracts, and the smallest of Middle Eastern abstracts. Work on same-sex sexual behavior is the smallest portion of every region’s scholarship on sexuality, and also varies greatly – from a low of 0.4% in South Asian abstracts to a high of 1.7% in North American women’s history abstracts. Discussion of sexual regulation was a significant part of all abstracts; in most cases about twice the focus on reproduction. Overall, even regions with similar amounts of sexuality scholarship focused on rather different mixtures of subtopics.

In fact, it appears that there is more clear variation across regional lines of sexuality topics than across time periods. Table 11 shows that the most recent and most distant chronological eras studied address sexuality in near identical amounts (8.5% and 8.6% of each period’s women’s history abstracts, respectively), and that there is no linear pattern of increased work on sexuality as scholarship moves chronologically forward. Even work on same-sex sexual behavior is more common c.1450-1599 than at any time except the 20^th century. The (slight) lowpoint of sexuality work appears to be in the eighteenth century – perhaps scholars interested in the early expressions of sexuality pre-date that century, and scholars interested in the rise of modern sexuality gravitate toward the 19^th and 20^th centuries. Overall, such results suggest that despite much more difficult-to-find sources in earlier centuries, scholars have been publishing on sexuality across time in roughly similar proportions – earlier periods are smaller raw numbers, but reflect roughly similar trans-historic women’s history interest in sexuality studies overall.

This brief case study suggests the ways that we might use data mining technologies to better understand the content of women’s history. In terms of sexuality, specifically, women’s historians might consider focusing more attention on reproduction’s relation to sexuality, moving beyond transgressive (hetero)sexuality, and more fully integrating same-sex sexuality. The diversity of chronologic interest in sexuality also reminds scholars to not presume that sexuality scholarship is most vibrant in recent eras, and might again serve to encourage scholars' in-depth conversations about sexuality across multiple centuries.

Table 11: Percent of Sexuality Topics within Select Time Periods of Women’s History

Conclusion

In June 2009, the New York Times published an article on the decline of “traditional history” courses, and held women’s/gender history largely accountable for this decline.²⁸ The author made these claims using a very limited (and as some have argued, quite flawed) quantitative analysis. We counteract this kind of misrepresentation of women’s history as a behemoth threat to the field with our analysis of a half a million article abstracts to show that women’s history is not even 10% of historical abstracts from 1985-2005. And unlike the New York Times article, topic modeling shows that all of these women’s history abstracts focus on other thematic areas as well – including national, political and military histories.

Underlying our work is not only the desire to better understand the field of women’s history, but also to ask where and when women’s history might best benefit from activist efforts. Armed with concrete information about the progress, gaps, and trends in the field, we can better identify needs and formulate discrete plans of action. It is one thing to encourage internationalization of publications, it is another to ask why a particular regional field seems demonstrably less engaged with women’s history, to directly mentor scholars in those fields for submission to journals, or to engage leaders in those fields with the evidence of their field’s (in)attention to women’s history.

Without doubt, women’s history, like the profession as a whole, is focused on more recent time periods. Women’s historians could certainly have more conversations across centuries and be attentive to the ways that much of the analytic threads running through work being done on pre-1800 time periods may be broadly applicable to other times and places. Women’s historians' can also continue to find productive ways to integrate the outstanding non-historical feminist work done in interdisciplinary-study programs into our chronologically distant historic studies– and to encourage interdisciplinary scholars to better integrate historical work into their own.

We also endorse the suggestion, in a recent AHA Perspectives forum on New Media histories, that digital advances can change how we can do history, not just how we present history. Our collaboration between a women’s historian and a computer scientist has allowed us to manipulate huge quantities of data and apply cutting-edge data mining techniques to women’s history: in other words, not just to access the ever-growing quantity of digital documents, but to find new ways to analyze them.²⁹

This work, however, is just a beginning overview of how we might investigate women’s history’s historiography. Although we have presented a wide array of information about two decades of women’s history abstracts, we have been able to include only rudimentary speculation. The massive amounts of data – starting with more than half a million abstracts – that we were able to analyze lends itself to almost unending tables, figures, and data points. We chose broad areas of analysis based on our own perceptions of women’s and sexuality history, as well as those that spoke to recent historiographic debates in the field. But there are still countless areas open for investigation.

We could easily incorporate forty or fifty years of abstracts to track longer-term trends; for example, seeing whether the slowing increase in women’s history abstracts in the new millennium proves to be a temporary blip. No database is perfect – in this case, the biggest problem with AHL and HA is their chronologic starting point at c. 1450. This limitation means that we have not been able to include ancient and medieval women’s history in our analysis. Moreover, abstracts are a mediated source that may tell us as much about abstracting patterns as about actual scholarship: Are foreign language articles underrepresented? Who are the abstractors, how might their own perspectives affect the content of the abstract, and to what degree do changes in language reflect abstractors' shifting termes d’art as much as shifts in scholarly approaches? Analyzing full-text articles would thus be a logical next step for topic model analysis. Full-text articles would remove the mediation of abstracts and allow for much more sophisticated and in-depth analyses of subfields of scholarship.

Future research might also look more at regional variations – and given the growing interest in transnational scholarship, might productively complicate national and regional identities. Likewise, looking in-depth at specific centuries might help identify the extent to which women’s history is shifting its focus to time periods within particular regions. Should we expect women’s history to map proportionately to overall regional and chronologic divisions, or can we explicitly account for women’s history choosing its own paths of study, sometimes in contrast to the field at large?

Such analysis also has the potential to raise political issues. Should women, as roughly 50% of any given population, expect to be the focus of half of all historical scholarship? Would this inclusion of women necessarily make all history partly women’s history? More specifically, what are we to make of the leveling-off of work on women’s history in the late 1990s? Has women’s history reached a saturation point, where it is incorporated into mainstream histories, and thus unremarkable? Has the shift to cultural history, with its attendant focus on gender and sub-topics, such as masculinity, moved the focus away from publications that are classifiable as women’s history? Or can we identify a backlash against women as subjects of study in what some have called a post-feminist era?

Ultimately, we might ask how women’s place in the academy relates to the expansion of women’s history. If women’s historians are likely to be women, how does equity within the profession relate to the future of the field? Do disproportionate numbers of female adjuncts, who may be less able to publish under heavy teaching loads, decrease women’s history publications? While somewhat far removed from a direct analysis of historical scholarship, these kinds of issues form the institutional background against which scholarship is produced and selected for publication.

To conclude, our analysis had tried to provide a bird’s eye view of roughly two decades of abstracts in post-c.1450 women’s history. Much of our purpose has been to provide useful data, rather than totalizing conclusions. We hope students and scholars will be able to use this data to pursue their own lines of inquiry. Getting a broad sense of where the field has been, how it has changed, and how it relates to the historical field in general will allow us all to better formulate informed opinions about what the next decades of women’s history might accomplish.