The traditional “lit review” bears more than a passing resemblance to a landscape painting. In both cases, the artist or scholar scans the horizon to trace the outlines of what appear to be the major summits, filling in the hills and valleys below to a varying and limited degree. In both cases as well, the selection of prominent features is largely dependent on perspective: one artist’s mountain is another’s molehill. But there are still some constants, particularly in the scholarly world. Those figures who loom over the others tend to be at the peak of their careers. As for the rest, whether or not they make it into the picture at all is to a considerable degree a matter of chance and space. And, of course, of time: we can read only so many books and articles. Sometimes excellent works simply don’t cross our desks (or, these days, our desktops).

But the rise of scholarly reference databases, either of citations or of full-text documents, has made it possible to identify, and harder to ignore, many sources that might previously have flown beneath our radars. These databases are not limited to articles, either: JSTOR now includes some books; and the MLA International Bibliography catalogs monographs. In itself, this newfound access does not per se solve any problems. If anything, it can make matters worse, by calling our attention to the inconvenient fact that an impossibly large number of sources could be read and evaluated before we write our review. No matter how well intentioned we are, it is often unfeasible—and possibly unnecessary—to plow through this forest. But the availability of all this metadata (and data, in the case of full-text works) also raises another possibility: that of using quantitative measures to assess the impact of certain arguments and ideas. Just as literary scholars have adopted text-mining techniques to explore the “great unread” of literary history, humanists can now use data-mining tools to structure their lit review.[1]

Now, I am not suggesting that thoughtful analysis and engagement with individual works can be replaced with number crunching. Because the lit review is ultimately about assessing the quality of other people’s arguments, it will and should remain a fundamentally qualitative exercise. The question is, are we always assessing the right works? Are we missing important trends? Might we be granting more weight to certain authors than they deserve, in light of their limited impact on the scholarship? And could we be ignoring parallel debates and arguments that are taking place in neighboring fields? What data mining can offer, I suggest, is a broad yet detailed backdrop that helps guide our analyses of secondary sources. It can reveal trends about the evolution of a field that might in turn lead us to pay more attention to a particular discipline, time period, theorist, or argument.

It is also a tricky exercise, which can easily be misleading. In this article, I experiment with the potential of JSTOR’s “data for research” portal (hereafter abbreviated as DfR) for discovering trends in scholarship. Data-mining JSTOR yields rows and columns of neatly structured metadata, which any Excel user can turn into scientific-looking charts. But the data themselves are often quite messy. Consider the following example: a scholar interested in the place-names mentioned in documents discussing “the Enlightenment” might be surprised to learn that “Princeton” appears more often than “Paris.” This is not because Princeton was a hotbed of the Enlightenment (which would have been a curious finding, since the university was still named the College of New Jersey in the eighteenth century), but because Princeton is a place where many twentieth-century books have been published on the Enlightenment. Full-text searches simply do not distinguish between footnotes and text.

More problematically still, the JSTOR holdings have a strong Anglo-American tilt. The United States is mentioned in about 20 percent of the total number of documents in JSTOR, and England in 14 percent; by comparison, 8 percent include a mention of Germany, 5 percent of China, 4 percent of Russia, and 2 percent of Brazil.[2] In addition to this geographical bias, there is also a linguistic one: 86 percent of the documents in JSTOR are in English.

The metrics one can produce using DfR thus vary significantly from other types of data-mining results. In the same spirit in which we speak of “dirty OCR,” I would dub this method “dirty quantification.” [3] The numerical results are not meant to offer clear, definitive answers to particular questions, but simply, and in the best of cases, to provide a general, fuzzy sense of how, where, and when certain scholars have engaged with certain ideas. These are dirty data, and they’re done dirt cheap: anyone, without any technical know-how, can download a mother lode of information. The trick, of course, is making sense of what it might mean. That is the challenge I take up in this article, using as my test case the vast expanse of scholarship on the Enlightenment.

1. Some Methodological Considerations for Using “Data for Research”

DfR is a search portal that, since 2008, lets users analyze the results of search queries by separating out the multiple dimensions of documents, including year of publication, language, article type, journal, discipline, and subject. To give an example that I return to later in this article, a search for “the Enlightenment” in DfR returns over 50,000 documents. Under the heading “Narrow results by,” one can select any of the given dimensions, analyze the values in the browser itself (by clicking on the “Charts View” icon), or download the results as a separate file. If one follows this process for “Year of Publication,” it is simple to visualize the results in the form of, say, a line graph, as can be seen in Figure 1.

Figure 1  Yearly pattern of usage for “the Enlightenment” in JSTOR holdings. Source: http://dfr.jstor.org/

This chart presents a fair amount of information: most notably, it suggests that interest in the Enlightenment as a scholarly topic has increased dramatically, and at times even exponentially, over the course of the twentieth century. It also seems to indicate that this interest peaked in 1998 and has dropped significantly since that high. But before we move on to interpreting these observations, it is important to specify what exactly we are looking at. A first point of clarification is that the sums presented in the chart refer, not to the total number of occurrences per year, but rather to the total number of documents containing this text string (i.e., there were not 1,862 mentions of “the Enlightenment” in 1998; this expression can be found in 1,862 documents). The implications of this distinction are that all documents are placed on the same footing: an article on the economic policies of the late Ming emperors that happens to contain a fleeting mention of “the Enlightenment” counts equally as a study of Diderot and the Encyclopédie. Unlike the Google Ngram Book viewer, these results do not provide data about frequency of use.

Because DfR counts documents, this means that any trends in the data must take into account variations in the size of the total holdings. A rise in the number of articles mentioning “the Enlightenment” could mean that more scholars are showing interest in this event; but it could also simply reflect the fact that there are more JSTOR documents in toto for a given period. What’s more, data from the past five to ten years should be treated with particular caution, since many journals are embargoed for a period of years. By “normalizing” the data (a service that DfR provides under “relative statistics”)—that is, by comparing the number of documents that mention the Enlightenment with the total number of documents contained, per year, in JSTOR—we can verify that the trend evident in Figure 1 is not just a factor of volume (see Fig. 2).[4]

Figure 2  Normalized yearly pattern of usage for “the Enlightenment,” divided by the total number of documents in JSTOR holdings. Source: http://dfr.jstor.org/.

This is the kind of pattern that calls out for interpretation (and is the subject of section 5). Can it be explained by disciplinary trends? Is this pattern more pronounced in certain journals? Are book reviews or are research articles more responsible for this change? We can test each of these hypotheses by studying the values of the various dimensions. Or we can run searches within the set of documents that mention “the Enlightenment” to identify the patterns of usage for other key terms, such as the names of major theorists. This is the kind of “drill[ing] down into subgroups” that makes “big data” qualitatively, as well as quantitatively, different from most data samples.[5] It can also help us distinguish between meaningful trends in the data and spikes that are due to sporadic events (such as the publication, and subsequent reviews, of a single book). This is particularly true when we’re dealing with fairly small subsets of data.

Even when results do not seem to depend on sample size or special circumstances, it can still be very hard to make sense of them. If more scholars use expression A rather than expression B, does that mean that the former is more important? Perhaps, but numerical values on their own can tell us only so much. The point here is not to divine the “wisdom of the crowd” through statistical measures. Indeed, one could very well use preponderance of usage to argue that everyone else is flat-out wrong or has missed a central point. Given that scholarship is driven by the need to produce original arguments, it would be absurd to posit that conventional usage is the vox dei to be respected at all costs.

At the same time, traditional lit reviews are littered with qualifiers that indicate a degree of quantitative importance (“seminal,” “influential,” “pathbreaking,” “classic,” etc.). We readily recognize that some works, ideas, and approaches have more weight than others, but our means of evaluating this weight tend to be impressionistic (scholars in the humanities, for instance, rarely use citation indexes).[6] Data-mining the literature on a topic can thus enable scholars to move from an impressionistic account of the scholarly landscape toward a more quantitatively driven view. Of course, not every historiographical question can be turned into a meaningful search query. But in many cases, data mining serves as a helpful, preliminary process for identifying those areas or periods that merit closer attention.

2. Enlightenment Debates

The scholarship on the Enlightenment is particularly well suited as a case study for this data-driven approach. First, it is a field spread across many disciplines: historians, literary scholars, political scientists, sociologists, and economists, among others, have all adopted it as an object of study. These different groups of scholars do not share the same methodological concerns, and they are often unaware of each other’s work. But they are all well represented in JSTOR, making data mining one of few options for studying Enlightenment scholarship across multiple disciplines. Second, this is a field with a very long history, and while much of that history is written in monographs, which are unfortunately not included in the JSTOR archive, the influence of those works can usually be detected in the articles (including book reviews) that JSTOR holds. Finally, much of the scholarship on the Enlightenment is preoccupied with definitional questions, some of which (if not all) can be identified through search queries and then measured over time.[7]

The first debate I focus on in this article comes from the field of early-modern historiography and concerns the singularity versus multiplicity of the Enlightenment.[8] Up until the 1970s, it was uncontroversial to speak of “the Enlightenment.” [9] This singular definition was forcefully presented by Peter Gay in his two-volume study entitled, precisely, The Enlightenment.[10] His work ranged widely across Europe and North America but considered the aims of the Enlightenment to be more or less identical in Florence, Vienna, Paris, and Philadelphia. The Enlightenment was also, in Gay’s account, a predominantly French affair, with the philosophes setting the tone for the arguments and campaigns to be led.

But this singular view has been challenged, first and foremost by scholars who wished to introduce national, confessional, or regional differences. In 1981 Roy Porter and Mikuláš Teich edited a volume called The Enlightenment in National Context, in which contributors examined how the Enlightenment played out differently in different countries. This approach has led scholars to focus more specifically on an Enlightenment—for instance, the British, German, Scottish, Russian, or (still) French. This preponderance of Enlightenments has been interpreted in different ways. For some, these national variants are still refractions of a single event: in their view, the “Scottish Enlightenment” is simply a handy term for describing how the Enlightenment (in the singular) played out in Scotland.[11] But others have argued that there is a degree of incompatibility between Enlightenments, and that one is only really entitled to use this term in the plural.[12]

This fracturing of the Enlightenment into smaller pieces raises new questions as well. Did everyone have an Enlightenment, or did some nations and cultures miss out? And by “everyone,” whom do we mean—Europeans and North Americans, or non-Western cultures as well? [13] And if there was a multiplicity of Enlightenments, were some more developed than others?

These various historiographical arguments can each be debated on their own terms, and a complete lit review would engage with them on that qualitative level. In this article, I mainly explore how far data mining can get us. My focus is accordingly less on the strength of individual arguments and more on the adoption of these arguments by others. The primary indicators I use to gauge such adoption are keywords: I start by looking at how the different qualifiers of the Enlightenment have been used over time, by discipline, and in conjunction with other terms (such as place-names). The following sections discuss the results.

3. Not All Enlightenments Are Equal: On National Differences

These searches prompt two immediate observations. First, scholars have, for a considerable time, recognized national, confessional, and conceptual differences within the broad category of the Enlightenment. But, second, a great majority of scholars seem to consider the “the Enlightenment” in the singular to be a valid category of historical analysis, as we can see in Figure 3a, which compares mentions of “the Enlightenment” in JSTOR documents with references to the French, European, and Scottish Enlightenment (respectively). At its peak, in 1998, the phrase “the Enlightenment” was used in the singular in 1,862 documents—that is, more than ten times as often as “French Enlightenment” (169 documents).

This discrepancy was somewhat less pronounced when I conducted the same search in the Google Books Ngram viewer: at its peak, in 1999, “the Enlightenment” was used slightly over twice as frequently as “French Enlightenment” (see Fig. 3b).[14]

Figure 3a Normalized patterns of usage for “French Enlightenment,” “Scottish Enlightenment,” and “European Enlightenment,” compared with “the Enlightenment.” Source: http://dfr.jstor.org/

Figure 3b Google Books Ngram viewer graph for “The Enlightenment, French Enlightenment,” 1900–2000 (English language only).

These varying results can be explained by the fact that the Ngram viewer graphs frequency of use. The JSTOR results suggest that it is common for many scholars to include at least a few references to “the Enlightenment” in their articles, whereas the graph created by Ngram can be read as showing that some works on an Enlightenment topic use the expression “French Enlightenment” very often. In other words, there is probably a concentration of uses within single documents that will appear only in a word frequency count. (And, of course, the data samples are different: there are comparatively far fewer books in JSTOR and, conversely, fewer journal articles in Google Books.) It is not evident which measure (number of documents vs. number of uses) is more meaningful: passing references can be dismissed as less important, but frequency counts can also simply reflect a small number of heavy users.

While we can certainly debate what precisely it entails for scholars to use the expression “the Enlightenment,” these graphs do suggest that most scholars have no qualms in referring to the Enlightenment as a single, identifiable phenomenon. At the same time, a few national Enlightenments do also receive considerable attention, and have done for quite some time (well before 1981 in any case: see Fig. 4). The three Enlightenments that stand out are the French (4,637 documents up through 2011), the Scottish (3,814), and the German Enlightenment; the latter tends to be discussed, even in English-language articles, under its German name, Aufklärung (3,275).[15]

Figure 4 Normalized patterns of usage for “French Enlightenment,” “Scottish Enlightenment,” and “European Enlightenment,” as well as “Aufklarung” (English-language journals only). Source: http://dfr.jstor.org/.

For those familiar with the history of the Enlightenment, these numbers are of course not surprising. Indeed, the strong showing of these particular Enlightenments tells us, more than anything, that this method of mining the JSTOR data returns plausible results. We can corroborate them by looking at the city names most commonly found in articles that reference “the Enlightenment.” As we saw with the Princeton example, this type of querying yields artificially high results for place-names that are also common places of publication for modern works (e.g., London and New York).[16] Leaving these names aside, out of all the documents in JSTOR that mention “the Enlightenment,” a third also mention Paris; Berlin comes in second, followed by Philadelphia and Edinburgh (see Fig. 5). In other words, with the exception of Philadelphia, the top cities in this (nonexhaustive) list correspond to the top three national Enlightenments.

Figure 5 References to city names in JSTOR documents that mention “the Enlightenment” (excluding cities that are currently major centers for publishing English-language books). Source: http://dfr.jstor.org/.

These numbers are meant to supply only a very general idea about the level of interest that scholars writing about the Enlightenment might have taken in these different national versions. One reason they should be treated with a few servings of salt is because of the aforementioned Anglo-American sample bias of JSTOR. In this regard, the high showing of the French and German Enlightenments is significant precisely because it goes against this bias, whereas the popularity of the Scottish Enlightenment probably benefits from it. Indeed, if we were to include scholarship written in other languages, the standing of the Scottish Enlightenment would likely decline slightly.

But the prominence of the French and German Enlightenments also reflects the fact that French and German authors were among the first to think self-consciously about the Enlightenment—indeed, they were already doing so in the eighteenth century.[17] In English, by contrast, it only became common to refer to the intellectual and social developments of the eighteenth century as “the Enlightenment” toward the end of the nineteenth century.[18] Even the expression “Scottish Enlightenment” began its rise to prominence only in the 1960s, as Figure 6 confirms.

Figure 6 Google Books Ngram viewer graph for “French Enlightenment” and “Scottish Enlightenment,” 1900–2000 (English language only).

Simply looking at these figures, can we venture any hypotheses about how scholars thought about the Enlightenment? In other words, do these graphs tell us anything about whether scholars adopted a pluralistic understanding of the Enlightenment (in which, say, the Scottish Enlightenment is a different beast altogether from the Aufklärung), or do the national qualifiers simply reflect a desire for precision (where “French Enlightenment” is just shorthand for “the Enlightenment in France”)?

This is a tricky exercise, as the data can be read in different ways and do not offer a clear-cut answer to this question. On the one hand, the fact that there are twice as many English-language documents that mention Aufklärung than mention the “German Enlightenment” might suggest that scholars see this philosophical movement as more specific than, and distinct from, the general Enlightenment.[19]

On the other hand, the fact that the plot line for “European Enlightenment” parallels that of other top-performing Enlightenments could be read as an indication that scholars thought about these movements as part of a single, broader Enlightenment (see Fig. 4). This is particularly striking with respect to the peak in scholarship relating to all the top-tier Enlightenments around 1997–99. I return to this curious peak in section 5—for now, I would simply note that it is also found in Figure 3, where references to “the Enlightenment” peak in 1998. While the reasons for this pattern are ambiguous, a plausible explanation would be that this period witnessed a burst of interest in the Enlightenment, and that scholars writing about specific Enlightenments were investigating aspects of a single phenomenon. Indeed, it would be more likely for interest in specific, separate Enlightenments to exhibit a different pattern of usage (since this would indicate that a smaller field of scholars, probably under a common methodological thrust, had adopted a concept). This is precisely the pattern we see with some of the other, less commonly referenced, national Enlightenments (see Fig. 7).

Figure 7 Normalized patterns of usage for “first-” and “second-tier” Enlightenments. Source: http://dfr.jstor.org/.

What is most striking in this graph is the fairly clear division, starting in the 1980s, between the “top-tier” Enlightenments discussed above and a “second tier” of Enlightenments. In this second tier, we find the Haskalah (the common term for the “Jewish Enlightenment,” present in 1,639 documents up through 2011), as well as the American (1,234 documents) and English (707) Enlightenments; slightly below (not shown on the graph) are the Counter- and Radical Enlightenment (688 and 685, respectively).[20] One can further distinguish a “third tier” of Enlightenments: here we find the Catholic (346), Spanish (332), British (299), Russian (209), and Italian (202) Enlightenments (see Fig. 8). Below this lies yet another tier, which consists of Enlightenments mentioned by fewer than 150 documents in JSTOR.

Figure 8 Top national/confessional Enlightenments in JSTOR, measured by number of documents. Source: http://dfr.jstor.org/.

While these numbers indicate that scholars are willing to append a wide variety of qualifiers to “Enlightenment,” they also reveal stark disparities in the attention afforded them—there is a hundredfold difference between the number of documents referencing the French Enlightenment and those referencing the Florentine (46 documents). To some extent, these differences in usage can again be explained by sample bias—one would presumably find more articles written in Italian that mention an illuminismo fiorentino. So obviously, the “poor” showing of certain national Enlightenments, in this graph, cannot be read as a sign of their “lesser” importance or as suggesting that the Enlightenment did not really reach those places.

What can we learn, then, from these numbers? Taken as a whole, they reinforce the impression outlined above: namely, that most scholars who discuss the place or effects of the Enlightenment in and around Europe seem to view it as a transnational phenomenon, one that is sufficiently homogeneous that it does not require qualifiers. Only 17 articles mention a “Venetian Enlightenment,” yet Venice is cited in over 3,000 documents referencing “the Enlightenment.” A similar story can be told for the “Russian Enlightenment” (209 documents) and Moscow (3,667), the “Spanish Enlightenment” (332) and Madrid (3,091), the “Polish Enlightenment” (113) and Warsaw (1,717), the “Dutch Enlightenment” (52) and Amsterdam (5,064). I list these numbers only to provide a sense of scale; many of the articles on the Enlightenment that mention Amsterdam, for instance, will not in fact be about the Dutch Republic at all and will simply be referring to a major site of publication for French books. But the numerical differences here are so many orders of magnitude apart that they can sustain high margins of error without modifying the trend.

Digging down into this data, we can uncover other interesting patterns of use, notably for “American Enlightenment.”[21] Back in the 1970s, the American Enlightenment was referenced in as many documents as the Scottish Enlightenment or the Aufklärung, placing it in the first tier for that period (see Fig. 7). My colleague Caroline Winterer has suggested an explanation for why documents referencing the “American Enlightenment” peaked at this time: it was in the context of the Cold War that American historians began promoting this category.[22] But after a peak in 1976 (which has yet to be surpassed, surprisingly, given the overall increase in the total number of JSTOR holdings since then), scholarly mentions of the American Enlightenment flagged for a good twenty years, before picking up again in the mid-1990s. At the same time, Benjamin Franklin and Thomas Jefferson are among the most cited individuals in documents that mention the Enlightenment, a discrepancy that suggests that scholars of colonial and early-republican America tend not to qualify their references to the Enlightenment (see Fig. 9; for a fuller list, see the appendix).

Figure 9 References to major figures (in conjunction with “Enlightenment”) in JSTOR documents (full names entered with quotes). Source: http://dfr.jstor.org/.

What does it mean for a label to drop or to rise (as in the case, say, of the Haskalah)? There are multiple possible explanations, which the graphs alone are ill-suited to disentangle. But when we compare more closely related patterns of usage, a narrative may start to emerge. Consider the trend lines for “English Enlightenment” and “British Enlightenment” (Fig. 10). They exhibit some similarities: both terms see an increase in 2006 and a drop-off after 2009. But overall, the fortunes of these two expressions seem inversely proportional: between 1992 and 2009, “British Enlightenment” witnessed considerable growth, whereas in the same time period, “English Enlightenment” greatly declined. Could this be an instance of one term simply being replaced by another? In the following section, I take a closer look at the English Enlightenment, to see what data mining can teach us about scholarly “branding.”

Figure 10 Patterns of usage for “British Enlightenment” and “English Enlightenment.” Source: http://dfr.jstor.org/.

4. What’s in a Name? The Case of the English Enlightenment

Why does interest in some national Enlightenments take off and not in others? There are a number of reasons why an expression might catch on: influential scholars might propose a new research agenda that other scholars then carry out; overlooked documents can come to light, through the release of critical editions, inspiring further research; or revisionist scholars might mount a challenge to existing historiographical categories and propose a different set of terms.

There are also less lofty reasons. Some expressions might be viewed as more attractive (i.e., better branding) than others. Describing your research as an investigation into the “early Enlightenment” may appear sexier than a study on “late Scholasticism”; as Simon Grote recently noted, using the term “Enlightenment” “can add an instant luster of modernity and relevance to anything and everything in the eighteenth century.”[23] To be sure, there can be intellectually valid reasons for connecting our work to one field of scholarship rather than another. But we all crave relevance, and that sometimes means marketing our work in ways that may have little to do with the content.

In the case of the Enlightenment, there are also subcurrents of nationalism at play. An example of this tendency can be seen in the late Roy Porter’s 2001 book, Enlightenment: Britain and the Creation of the Modern World.[24] Here Porter, himself British, explicitly sought to “reclaim” the Enlightenment for Britain. As the graphs above show, England and Britain have long been the poor cousins of the French, German, and Scottish Enlightenments in the secondary literature. So why shouldn’t England (since the Scottish part of Britain has long been recognized for its contributions) have an equally respectable share of the Enlightenment? I shall not rehearse here the reasons why, historically, England has not loomed large on the map of Enlightenment studies, despite being home to three of its patron saints (Bacon, Newton, and Locke).[25] What I wish to do instead is ask whether the JSTOR data can shed any light on the practices of Enlightenment naming.

One of the first insights that the data yield is that, up until recently (and, to some degree, still today), scholars used a different term when referring to eighteenth-century English literature: they characterized it as “Augustan,” calling this period the “Augustan Age.” The origins of this expression are to be found in the period itself, though it was only in the twentieth century that it began to serve as a period label.[26] It is of course more challenging to chart the fortunes of this expression over time, as it also refers to a very different period (the reign of the Roman emperor Augustus). To get around this ambiguity, I restricted my queries to journals that JSTOR characterizes as falling in the “British Studies” discipline and then added other journals (from the “History” and “Language & Literature” disciplines) that focus on modern literature or eighteenth-century studies.[27]

Because these queries mine a far smaller number of journals, the total numbers they yield are far smaller: the expression “English Enlightenment” appears in 157 of these documents (compared with 712 in the full JSTOR database, up through 2011), whereas “Augustan Age” is cited in 396. In themselves, these numbers may not be statistically meaningful, and their small volume should serve as a caution against reading too much into them. That said, when one graphs the two plot lines together, they do reveal an interesting pattern—a pattern that suggests that “English Enlightenment” tends to be used in lieu of “Augustan Age” and may be in the process of replacing it as the term of choice (see Fig. 11).

Figure 11 Normalized patterns of usage for “Augustan Age” and “English Enlightenment” in select JSTOR journals. Source: http://dfr.jstor.org/.

It’s obvious from this graph that “Augustan Age” is the more traditional of the two expressions, appearing in 79 documents before 1950 (inclusively). “English Enlightenment,” by contrast, does not appear in this dataset before 1950 and is not used in more than 4 documents per year before 1992 (when it occurs in 15). This corresponds more or less to the pattern of usage for this expression across the whole JSTOR database (see Fig. 10), where we similarly find a peak in 1992 (39 documents) and very few uses before 1950 (15 documents, 1882–1948). Despite the smaller sample size, the plot line we observe in Figure 11 thus seems to be fairly consistent with the general performance of this term.

Between 1951 and 1991, “Augustan Age” is used in seven times more documents in this dataset than “English Enlightenment” (197 vs. 28 documents). But after 1991, the trend changes, with both expressions appearing in roughly the same number of documents between 1992 and 2011 (120 and 121, respectively). This is where the comparison of plot lines becomes interesting. In 1994, two years after “English Enlightenment” spikes for the first time, this newer term suddenly drops down to a low level (4 documents; a comparable drop is evident in Fig. 10, as “Augustan Age” spikes back up to a new high, 15 documents). The return of “English Enlightenment” in 1995–97 marks a relative decline of “Augustan Age” (compared with its 1986 and 1994 peaks), whereas the years 2001, 2005, and 2008 reveal a remarkable contrast: the fortunes of each expression are neatly reversed, with “English Enlightenment” spiking (and “Augustan Age” plunging) in 2001, and the opposite trend occurring during the other years. These divergences suggest that scholars who use one term tend not to use the other. Indeed, there is very little overlap in the usage of these expressions: only 10 documents in this select dataset mention both “Augustan Age” and “English Enlightenment.” In all of JSTOR, only 20 documents (up through 2011) contain both expressions.

What might be driving this substitution? There could be generational reasons; “Augustan Age” is also not very politically correct. What we can say with some certainty is who is driving it: about half the documents that mention “Augustan Age” in this dataset were published in journals that JSTOR places in the “Language & Literature” discipline (203 documents out of 396); around the same number (214) are categorized as “History.” [28] In the case of “English Enlightenment,” by contrast, 83 percent of the documents featuring this expression appeared in “History” journals (125 out of 150), as opposed to only 20 percent in “Language & Literature” (30 documents). So the rise of “English Enlightenment” seems to have largely been driven by historians. This pattern holds, albeit less starkly, when we consider the breakdown by discipline in the full JSTOR database of the documents up through 2011 that mention “English Enlightenment” (707 in total): those that fall into the “History” discipline (295 documents) account for 42 percent of the overall uses, compared with 22 percent of the uses categorized under “Language & Literature” (155).[29] In comparison, there is a smaller discrepancy between the percentages of both for articles mentioning the “French Enlightenment” (4,637 documents): “History” (1,887) accounts for 41 percent of all uses, and “Language & Literature” (1,286), for 28 percent. (I discuss the breakdown of results by discipline more broadly in the following section.)

In and of itself, the observation that historians, rather than literary scholars, seem to be the ones pushing the label “English Enlightenment” over “Augustan Age” can be interpreted in many ways, and data mining can get us only so far in resolving this question. It can, however, draw us down into the data to a remarkable level of precision. The first time that “English Enlightenment” peaks in usage is in 1992, when it appears in 39 documents. But this is still a small volume of uses, which can accordingly be swayed by a single publication. In this case, it would appear that the publication of Peter Harrison’s “Religion” and the Religions in the English Enlightenment in 1990 is largely responsible for this surge, since 12 documents appearing in 1992 mention his book. What’s more, not all of these are even articles or book reviews: many simply fall into the “miscellaneous” article type category and consist of front or back matter, “volume information,” or lists of “books received.” [30] At the same time, it is not clear whether these documents should be excluded from queries, as they, too, can contribute to the dissemination of new terms.

Finally, it helps to step back from the data for a moment and recall why the expression “Augustan Age” was introduced in the first place. The adjective “Augustan” was originally affixed to literary authors; in the words of Oliver Goldsmith, who penned “An Account of the Augustan Age in England” (1759), it was a matter of recognizing “when language and learning arrived at [their] highest perfection.” [31] Goldsmith considers philosophers such as Locke, Bolingbroke, and Shaftesbury in his account, but it is mostly to praise their style. In general, his essay focuses on authors who were renowned for their literary prowess, such as Pope, Swift, Addison, Steele, and Walpole. These are still among the best-known (and most studied) authors of this period, but they are mostly studied by scholars in English departments. So it is curious that it was a different group of scholars who set about “rebranding” this period. What’s more, given the comparatively small volume of documents (in the full JSTOR archive) that mention the “English Enlightenment,” one suspects that this term has not found much resonance beyond the field of English studies.

5. The Glory years of enlightenment scholarship

Many of the figures above suggest a surge of interest in the Enlightenment, along with its principal subgroups, toward the end of the 1990s. Do the data offer an explanation for this? Again, there are a number of possible causes: it might be linked to the popularity of one or more theorists; to the publication of key works; to the emergence of new fields around this topic; or to the redirection of a discipline. As it turns out, multiple factors appear to have been in play.

The Theorists

If one graphs the references to some of the major Enlightenment theorists (in documents that also mention “Enlightenment”), we find the pattern shown in Figure 12, with the same peak between 1997 and 1999.[32] Again, the discrepancies between theorists are impressive. They’re also somewhat surprising. Few reviews of Enlightenment historiography by intellectual historians would fail to mention Robert Darnton, yet he hardly registers on this graph: Foucault, Habermas, Adorno, and (to a lesser extent) Pocock and Arendt overshadow all the rest. The footprints of these giants are remarkable for their sheer size. In 1998 alone, out of 4,249 documents that refer to “Enlightenment,” 730 (or 17 percent) also mention Foucault. Jürgen Habermas, for his part, was mentioned 13 percent of the time a scholar wrote “Enlightenment” (548 documents). Taking into account the documents that mention both theorists (289), nearly a quarter of the texts mentioning “Enlightenment” in 1998 include a reference to either Foucault or Habermas or both (989 documents, or 23 percent). By comparison, for the same year, the number of documents referencing the Aufklärung or the French or the Scottish Enlightenment is far lower (110, 169, and 216 documents, respectively).

Figure 12 Normalized patterns of references to major theorists in conjunction with “Enlightenment.” Source: http://dfr.jstor.org/.

The place of these theorists is remarkable not only in itself but also in contrast with that of earlier Enlightenment scholars. In 1978, for instance, Peter Gay (then at his height in JSTOR documents) was mentioned in a mere 42 of the 1,772 documents containing the term “Enlightenment” that year (3 percent). But the surge of interest in these later theorists in conjunction with the Enlightenment raises an interesting interpretative question: are we dealing here with an instance of causation or of correlation? Indeed, the years 1997–98 also marked the high point of scholarly interest in both Foucault (references to whom, across the full JSTOR database, peaked in 1998, at 2,719 documents) and Habermas (who peaked in 1997, with 1,396 documents mentioning him). What these data do not tell us, then, is whether scholars were writing about Foucault and Habermas, and happened to mention the Enlightenment because these authors do, or whether they were writing about the Enlightenment and mentioned these theorists because they were all the rage.

In the case of Foucault, we can attempt to tease apart these alternatives by considering the context in which he was brought up. Of the documents that, in 1997, mention both “Foucault” and “Enlightenment,” 19 percent also reference “discipline and punish,” the title of Foucault’s famous 1975 study, first translated into English in 1977 and reissued in 1995—precisely three years before the peak of his influence on Enlightenment scholarship.[33] The Enlightenment, in this work, lies at the center of Foucault’s thesis: “The ‘Enlightenment,’ which discovered the liberties, also invented the disciplines,” he argues in his chapter “Panopticism.”[34] The strong showing of this particular work indicates the extent to which the scholarship from this period took a rather dim view of the age of Enlightenment (as the success of Adorno, on display in fig. 12, also attests to).

For Habermas, the source of interest in his work can also be identified fairly precisely: of the 548 documents that, in 1998, reference him and “Enlightenment,” 304 (or 55 percent) also mention “public sphere.” By contrast, only 28, or 5 percent, reference the “philosophical discourse of modernity,” the title of his 1990 essay collection.[35]

Even if it was the craze for Foucault and Habermas that was driving the surge in interest in Enlightenment topics, this would still leave 75 percent of the documents unaccounted for. What else was going on in those years? It turns out that documents referencing women or gender, in conjunction with “the Enlightenment,” were at their peak during this period (see Fig. 13). In 1998 a little more than half (55 percent) of the documents that mention the Enlightenment also refer to “women” (1,016 documents out of 1,862). There is some overlap between these documents and those that cite Foucault, who is mentioned in about a third of them. Other than Foucault, I was unable to identify a particular feminist/gender theorist who might have been behind this surge.[36] It would appear that the rise of women’s studies on the whole may have been responsible.[37] This hypothesis leads us to a second way of interpreting the success of Enlightenment studies in 1997–99, which is to examine the data by discipline.

Figure 13 Normalized patterns of references to “women” or “gender” and “the Enlightenment.” Source: http://dfr.jstor.org/.

The Disciplines

As we saw in the previous section, it is not always possible to know why a trend is occurring, but we can sometimes get a sense of who is behind it. The two best clues for assessing this question are disciplines and journals. To better comprehend what exactly was occurring during the peak moment of interest in the Enlightenment, I compared it with another three-year interval shortly thereafter (2002–4).

What this method of analysis reveals first is that, once again, the surge of interest in the Enlightenment seems largely to have been driven by historians (see Fig. 14). In 1997–99 articles from the Journal of Modern History (JMH) accounted for 10 percent of the documents referencing “the Enlightenment.” In 2002–4, JMH articles contributed only 2 percent, and the journal containing the most documents on the Enlightenment, the American Historical Review, still provided only 4 percent of the total. The overall number of history documents also dropped significantly, from 1,820 to 1,502 (i.e., a decline of 16.5 percent). That drop alone accounts for 40 percent of the difference between the total number of documents mentioning the Enlightenment in 1997–99 (5,455) and in 2002–4 (4,680, for a difference of 775). By contrast, articles classified as Language & Literature declined more modestly, from 1,212 to 1,123 (7 percent).

Figure 14 Normalized pattern of references to “the Enlightenment,” broken down by the top two disciplines. Source: http://dfr.jstor.org/.

But historians were not alone in turning their attention to the Enlightenment during this time—sociologists did as well. The American Journal of Sociology published a high volume of documents mentioning the Enlightenment in 1997–99 (4.6 percent of total), whereas in 2002–4, it published none.[38] This disciplinary pattern was not limited to that one journal. In 1997–99 History was responsible for 33 percent of the references to the Enlightenment, with Sociology in third place, at 9 percent (Language & Literature contributed 22 percent). In 2002–4, by contrast, History remained on top (still close to a third), but Sociology contributed only half as many documents (4.7 percent), dropping down to eighth place. In total, there were 267 fewer sociology documents mentioning the Enlightenment. So the decline in both history and sociology documents mentioning the Enlightenment between these years is responsible for three-quarters of the overall drop.[39]

Conclusion

This experiment in data mining leaves one with an overriding sense that many debates about the Enlightenment take place in isolated corners of academia. What is accepted as scripture in one field may be heresy in another. A recent article claimed that “only a small—if vociferous—minority of historians maintain the unity of the Enlightenment project,” [40] but in fact, the numbers suggest quite the opposite: a small, if vociferous, minority of historians debate the unity of the Enlightenment passionately, while the vast majority of scholars (historians, but also economists, literary critics, sociologists, political scientists, etc.) take its unity as a given. This is because they are interested in exploring other questions, which modern European historians, conversely, do not address. For instance, a major topic in economic history is the relation between the Enlightenment and the Industrial Revolution, a topic that often goes entirely unmentioned in the scholarship among social, cultural, and intellectual historians.[41] Each subfield has its own historiography; data mining these parallel literatures calls attention to the divides between them and to the blind spots in each.

But a quantitative analysis of scholarship also draws a whole range of other scholars into the mix, particularly when the period under study is one of the big-ticket items of world history. These are the scholars who do not specialize in this period per se, but who reference it as the backdrop, beginning, or bookend of their own research. Accordingly, they are likely to be less well versed, and possibly less interested, in the more specialized (and, particularly, recent) scholarship. At the same time, their number is legion, and so they will sway quantitative measures strongly in one direction or the other.

This exercise in quantification thus forces us to consider a somewhat awkward question: should everyone get a vote? If we discover, for instance, that “Enlightenments” is used in twenty times fewer documents than “Enlightenment,” is this evidence of anything more than the fact that most scholars have not read the latest Enlightenment scholarship? And should nonspecialists get a say in the historiographical debates of a particular field?

This a difficult question to answer, though it may help to consider the effect of time on a scholarly consensus. While information travels slowly in our congested Republic of Letters, in some cases news does eventually manage to spread. Until the 1980s, for instance, Marxist historians dominated the historiography of the French Revolution and were accordingly referenced in the largest number of documents (see Fig. 15). After François Furet published Penser la Révolution française, in 1978, however, he soon overtook them, eventually becoming the reference point even for scholars outside the field.[42] What this example suggests is that some scholarly messages do eventually get across. So why some and not others?

Figure 15 Documents mentioning “Furet,” “Lefebvre,” or “Soboul” in conjunction with “French Revolution.” Source: http://dfr.jstor.org/.

There is no doubt an element of chance, and of academic fashion, in these examples of “breakthrough” scholarship: consider the phenomenal rise of interest in Foucault, as seen in Figure 12. But presumably these success stories also have something to do with the work itself. The question then becomes one of knowing whether there are any common features that could explain the success (or, conversely, the failure) of certain kinds of scholarship. In the cases of Foucault and Furet, one might say that both authors offered new interpretative frameworks for making sense of history. This type of “methodological” contribution can be contrasted with scholarship that brings to light new or overlooked evidence: in this category we could place recent works by Steve Pincus and Tim Harris on the Glorious Revolution, which emphasize the violence that accompanied this supposedly bloodless revolution.[43] These studies transform our understanding of an event mainly by means of archival discoveries. And both of these approaches, it would seem, have the potential to reach audiences beyond their narrow fields.

Then there is a third kind of scholarship, one that focuses more on defining terms. Scholars writing in this vein concentrate on such questions as What is liberalism? What is Romanticism (or Realism or Modernism)? And what is Enlightenment? Indeed, many of the recent debates about the Enlightenment fall into this category. Was the Enlightenment national, global, radical? These are primarily arguments about the terms of the argument, and only indirectly about the content of the Enlightenment itself. It could well be that these kinds of questions are less likely to appeal to nonspecialists, who may not always see the point (or what is at stake) in such arguments.

In singling out this last category of scholarship (among other types—the list is not intended to be exhaustive), I do not mean to suggest that defining terms is unimportant. But to end on a more qualitative, rather than a quantitative, note, it does seem fair to point out that recent debates about the Enlightenment have focused excessively on terminology. The problem here is not restricted to scholarship on the Enlightenment. The labels for any historical period face the same challenges: they must be able to account for multiplicity, variety, evolution, and sometimes even contradictions in the corpus they designate. These are, after all, features of almost every period of human activity, and even of a single life. The point of the historical label is to identify the forest, not to describe the trees.

The danger that lurks in these terminological woods, however, is the temptation to read too much into a label. We may readily grant that there is some value in distinguishing the French from the Scottish Enlightenment, but is there, fundamentally, such a thing as “the French Enlightenment”? Voltaire and Diderot may well have less in common than, say, Diderot and Hume. But maybe we should not expect historical labels to perform such heavy lifting. If by “French Enlightenment,” we mean only “the collection of ideas, works, and practices produced in France by individuals associated with what is generally recognized as the Enlightenment movement,” then we’re not raising expectations too high. It is only when we assume that a historical label can be unpacked to reveal an entire philosophical, political, and religious program that we run into problems. As humanists, we tend to value “thick description,” those webs of meaning that our sources spin around their worlds and their works. It is precisely these webs that we can unravel when we’re engaging closely with a source. But the primary usefulness of a historical label is to offer a “thin description,” that is, a minimalist introduction to what may well be a heterogeneous set of texts and practices. This is also what quantitative analysis is best suited to deliver.[44] 

Appendix

The following is a list of names associated with “Enlightenment” in JSTOR documents up through 2011. When a first name is included, the search was conducted by placing the full name in quotation marks.