The following is a guest post* from Andrew Higgins, who recently received his PhD in philosophy from the University of Illinois, Urbana-Champaign. He spent some time combing through data at the Open Syllabus Project (previously) and in the post presents some information that should be of interest to fellow philosophers. Thanks, Dr. Higgins!
Philosophy Data from the Open Syllabus Project
by Andrew Higgins
The Open Syllabus Project has collected data on which texts are taught in college courses in English speaking countries based on the information provided in over one million syllabi (see this NYT article for a summary). One can search through their syllabi and filter based on discipline, institution, state, country, or source. Searching for philosophy syllabi, you’ll see the most popular texts, going as deep into the 47,920 texts as you’d like. However, it’s hard to see the big picture wading through a list of thousands of authors and texts. Below, I’ve summarized what I’ve found from analyzing all texts which were assigned in three or more syllabi. I start by presenting the results, but note that these results should be taken with a grain of salt, for the reasons listed at the end.
As one can also find on the project website, these texts are most frequently assigned:
|2||944||Utilitarianism||Mill, John Stuart, 1806-1873|
|4||658||Meditations on First Philosophy||Descartes, René, 1596-1650|
|9||489||The Metaphysics of Morals||Kant, Immanuel, 1724-1804|
|10||468||Existentialism||Sartre, Jean Paul, 1905-1980|
Ancient Greek philosophy appears to be overrepresented, but the ranking of texts can be a little misleading. For example, Heidegger’s book on Nietzsche is higher ranked than any of Nietzsche’s own works, but overall Nietzsche has been assigned almost twice as often as Heidegger (947 vs. 518). We should dig deeper and shift our focus to the list of most significant authors.
From an initial set of 70,393 assignments of 7,038 texts, I cleaned the data by merging duplicate or near-duplicate names, resulting in a list of 4,007 authors (the actual number is likely around 3,900, as co-authored items have not yet been addressed). Below you can see the 20 most frequently assigned authors.
|3||1934||Kant, Immanuel, 1724-1804|
|4||1532||Mill, John Stuart, 1806-1873|
|5||1465||Descartes, René, 1596-1650|
|6||1056||Sartre, Jean Paul, 1905-1980|
|7||994||Hume, David, 1711-1776|
|8||947||Nietzsche, Friedrich Wilhelm, 1844-1900|
|9||759||Rawls, John, 1921-2002|
|10||535||Marx, Karl, 1818-1883|
|11||519||Locke, John, 1632-1704|
|12||518||Heidegger, Martin, 1889-1976|
|13||497||Russell, Bertrand, 1872-1970|
|14||483||James, William, 1842-1910|
|15||472||Hobbes, Thomas, 1588-1679|
|16||408||Nagel, Thomas, 1937|
|17||398||Rousseau, Jean-Jacques, 1712-1778|
|18||372||Singer, Peter, 1946|
|19||340||Ayer, A. J. (Alfred Jules), 1910-1989|
|20||339||Leibniz, Gottfried Wilhelm, Freiherr Von, 1646-1716|
This list indicates that Ancient Greek philosophy is significant, but we also give attention to the Modern period and both Anglophone and Continental philosophers. It appears that non-Western philosophy is severely underrepresented. Only two non-Western authors, Confucius and Suzuki, are found in the top-200. In this same group of 200, 60.2% of post-Kant readings go to Anglophone authors, 35.4% to Continental authors, and the remainder to figures I judged to be neutral between the two (e.g., Darwin).
I then expanded the search to the top 1,317 entries (all with 10+ uses) and looked instead at the most used journals. From this set, the following journals are most often assigned:
|1||923||The Journal of Philosophy|
|2||851||The Philosophical Review|
|5||479||Philosophy & Public Affairs|
|7||181||Revue Internationale De Philosophie|
|9||111||Proceedings and Addresses of the American Philosophical Association|
|10||111||The Philosophical Quarterly|
This information could be valuable for department librarians deciding whether it’s worth purchasing access to a journal.
In my final analysis thus far, I categorized all 4,007 authors by number of represented publications, sex, and date of death. 74% of authors have only one text taught in this sample, 1% have more than ten, and Plato has 51 (various permutations of the dialogues). In the graph below, you can see how well each time period is represented.
Not shown are the 45,104 assignments of texts published from 1911 to the present. In this graph you can see that, as a field, we largely ignore the medieval period. More worrisome, though, is how we are ignoring women. Of the 70,393 assignments given by the philosophers represented here, only 5,470 (7.8%) of those were assignments to read a female author. From Ancient up through the 19th century, only about 2% of assignments from each period were written by women. Of all texts written by authors who died after 1910, only 11.4% were authored by women.
The take home message, for me, is that we should consciously aim to better represent non-Western philosophers, female scholars, and pre-Modern thinkers other than Plato and Aristotle.
I don’t know whether this data is accurate. My original data consists of syllabi collected from the web; the sample set is large, so I don’t worry about the representativeness of their original information. However, I have only looked at sub-sets of that original data, with the largest sub-set being all texts with 3+ uses (7,038). That means I’m ignoring 40,882 texts used only once or twice. My set has 70,393 assignments, and the remaining 40,882 texts probably have at most around 60,000 assignments, so I can at least say that this data represents over half of all assigned texts in their database.
Even if these 7,038 texts are representative, my methods weren’t ideal. In sorting male/female, some names were removed because sex could not be determined within 1 minute of web searching, and I counted a text as 100% female even if only one of five co-authors were female. My judgments on Anglophone vs. Continental may have been off, as I only really know the Anglophone tradition.
Furthermore, we should remember that the numbers here are the numbers of texts assigned; since these records do not distinguish between a one page Analysis article and a three-hundred page book, we should be extra cautious in inferring any general conclusions about which figures and traditions are taught more frequently.
Automated processing is probably the way to go if one wanted to analyze the full set of 47,920 texts, but it has its downsides. Without some machine learning, you won’t get automated Anglophone/Continental breakdowns, and the entries don’t reliably show dates, so you can’t just rely on automatic date extraction. One could use scripts that sort based on sex, but they aren’t perfectly reliable, and you’d still have to manually check the names with abbreviated first names.
On the bright side, there’s low hanging fruit if one wants to further filter the data by institution or geographic locations. I’m planning to develop a network of text-demographic relations extracted from this data to see which institutions and regions are teaching what, but that’ll have to wait for another day.
If you’d like to see more, I’ve posted various other data-driven pictures of the field, here.