Another way to explore of the question whether Nietzsche's Ecce Homo is his autobiography is to use topic modeling. There are multiple approaches to topic modeling, but the general idea is to find clusters of related words or topics. If a particular text is an autobiography, or even a biography, you would expect certain word clusters. In the case of an autobiography, you'd expect words connected to childhood, or perhaps words related to whatever the person became famous for, and so on. Of course, there will also be many common words and they may cluster in ways that could be revealing, but also may not be. For example, present tense forms of "to be" might be present in all clusters, but that's not typically informative. Common but largely uninformative words are called "stop words" and traditionally one eliminates them from the text when performing these sorts of analyses. There are off-the-shelf lists of such words readily available. I decided to roll my own because some traditional stop words like "I" and "you" are important for my analysis and I did not want them eliminated. I added words to the list only as necessary to improve clustering. I ended up with this list of stop words:
against, all, among, an, and, any, are, as, at, be, because, been, before, but, by, can, come, does, even, ever, for, from, had, has, have, he, here, him, his, how, in, into, is, it, man, men, more, most, much, must, no, of, on, one, only, saw, say, than, than, that, the, this, those, to, too, was, were, what, when, which, who, whom, with, would
For my initial foray, I used BERTopic and discovered that when I set ngram_range = (1,1),[code] the algorhythm would reliably return up to five topics. One of the indicators of a good spread is that the topic clusters are neither too close nor too distant from each other. Here is the intertopic distances from one particular run:
So what are these topics?
What can we infer from these clusterings? First, while it's not surprising that "my" occurs in four of the five topics, it is interesting that it does not occur in the topic concerned with morality, at least (one assumes) in the sense of moral ideals. This is unsurprising given Nietzsche's trenchant critiques of western morality. That the third topic pertains to books and writing is to be expected since much of Ecce Homo concerns Nietzsche's writings. I found it interesting that in Topic 0, concerning all things deutsch, seemingly excludes Nietzsche directly (i.e., "me" is absent although "my" is included). Perhaps this reflects Nietzsche's own ambivilance, best captured in this line: "I am perhaps more German than present-day Germans, mere citizens of the German Reich, could possibly be—I, the last anti-political German. And yet my ancestors were Polish noblemen" (EH Wise 3). Also interesting is when "not" appears and when it is absent. Since Nietzsche frequently describes western morality as life negating, it's understandable that "not" appears in second place in topic 1. The Germans in general and Wagner in particular are also often described as negating life in various ways. Perhaps Nietzsche's friendship with Wagner and esteem for him keeps "not" in a relatively low position in that topic. I suspect that "not" belongs in the Zarathustra cluster because he "says No and does No to an unheard-of degree, to everything to which one has so far said Yes" (EH Z 6). More simply, it might be that Nietzsche not infrequently describes Thus Spoke Zarathustra using the word "not" (e.g., "Is not Zarathustra in view of all this a seducer? […] Not only does he speak differently, he also is different" [EH Preface 4], "My whole Zarathustra is a dithyramb […] not on pure foolishness" [EH Wise 8], and "Suppose I had published my Zarathustra under another name […]—the acuteness of two thousand years would not have been sufficient for anyone to guess that the author of Human, All-Too-Human is the visionary of Zarathustra" [EH Clever 4]).
It might be helpful to compare this analysis of Ecce Homo with a similar analysis of a text that is unquestionably an autobiography. I have chosen John Stuart Mill's Autobiography. While Nietzsche was certainly no fan of Mill or Utilitarianism more generally, Mill was roughly from Nietzsche's father's generation, so the temporal distance is not particularly great. Mill also indicates to the reader the genre of his book both in the title and in the first sentence: it is a "biographical sketch," so we needn't guess. Perhaps this goes to the heart of Nietzsche's criticism of Mill in The Twilight of the Idols as "insulting clarity [die beleidigende Klarheit]" (Skirmishes, §1). In any event, I followed the same procedure as I did for Ecce Homo. To achieve a similar result required a longer list of stop words, namely:
again, against, all, among, an, and, any, are, as, at, be, because, been, but, by, can, come, could, did, does, even, ever, first, for, from, had, has, have, he, here, him, his, how, however, in, is, it, its, man, men, more, most, mr, much, must, of, on, one, only, or, saw, say, seemed, should, than, than, that, the, their, them, they, this, those, to, too, very, was, were, what, when, which, who, whom, with, would
If I were analyzing Mill's Autobiography in isolation, I would have also removed "not" but retained it here since I kept it for Nietzsche. Here is the intertopic distance map:
Again, I wanted to achieve relatively tight but non-overlapping topics, which I was able to do on this run. Let's look at the topics.
Topics 1-3 are unsurprising as Mill's father and Jeremy Bentham were the greatest influences on his life, and his term as a Member of Parliament was important to him. Topic 4 likely concerns Harriet Taylor, his long-time friend and later wife. Topic 0 is interesting given its similarity to Nietzsche's Topic 1 above. Despite the differences in the two philosophers approach to ethics, both clusters connect morality, life, and various negative words.
So far, the analyses have been for looking at single word topics. What happens if we increase the size of the topics to 3 or 4 words? To do that, I initially ran the analysis without stop words. After all, a commonly used word that perhaps has little meaning in isolation can be more important when looked at in phrases. The results for Mill were generally what one might expect. For example, one of the topics was "principles of political economy," the title of one of Mill's most widely-read books. Other examples of more to-be-expected phrases include "member of parliament" and "house of commons." At best, the only mildly surprising phrase was "my father made me," which isn't that surprising given that Mill was educated by his father. What is impressive, at least to me, is that BERTopic indentified this as an important topic, which it is, even though the phrase "my father made me" occurs only three times in Mill's Autobiography (lest anyone imagine anything particularly horrible, twice the next word is "study" and the remaining time it is "read"). The results from Ecce Homo were even less helpful. Included in the topic of Germans was now "herr von treitschke" but likely only because his name is three words. BERTopic correctly identified "human all too human" (the title of one, or two depending on how you count, of Nietzsche's books) and the "revaluation of all values."
Since Wagner is the only person who warrants a topic, I also investigated The Case of Wagner. Perhaps because it is more sharply focused than Ecce Homo, I had to reduce the number of topics to three and often two. For the present question about Ecce Homo, the results were unhelpful, but it was interesting that the topics tended to be Wagner and various masculine pronouns against morality, passion, and various second person plural pronouns.
I wanted to see how Ecce Homo would compare with an autobiography of a non-philosopher, so I ran a similar analysis on Anthony Trollope's An Autobiography. I chose Trollope because like Mill, he was British and because he lived during much the same time (1806-1873 for Mill, and 1815-1882 for Trollope). I'm not conversant with Trollope's work and life, but it would seem that BERTopic did an excellent job of identifying the salient topics after a little fiddling (including a much longer list of stop words). One consistent topic was the post office. Initially I assumed that this had to do with the mailing of manuscripts. Since Trollope published over a hundred works, I assumed that he was a regular postal customer. While this might be the case, it also turned out that he worked in the post office for seven years. Another topic included various political words (e.g., "parliament" and "liberal" ), but also "beverley." In my ignorance, I thought this was an error, but closer inspection revealed that it names, not a woman, but a borough where Trollope ran unsuccessfully for a seat in the House of Commons. The remaining two topics concerned his family (including Harrow and Winchester, where he was a student) and the economic side of his work as an author.
For all three writers, at least one topic concerned their work, even if in the case of Trollope it was as a means of income rather than something more "noble" in some sense. Although Nietzsche mentions his father in Ecce Homo (Why I Am So Wise §§1 and 3-5), he is not mentioned in any of the topics. In contrast, both Mill and Trollope have a topic that includes their fathers. Similarly, BERTopic never seized upon Nietzsche's decade long tenure as a professor as being cluster-worthy, even though Nietzsche mentions it in several places (Why I Am So Wise §1, Why I am so Clever §9, and "Human, All-Too-Human" §3). In contrast, both Mill's and Trollope's political experiences, as well as Trollope's postal job, are mentioned.
The analysis of Trollope's An Autobiography is helpful in the following way. Suppose that we ask a slightly different question than "Is Ecce Homo an autobiography?" Instead, let us pose the following question: On the basis of the identified topics, can we say that the analyzed work is an autobiography and not a retrospective discussion of the author's own works? In the case of Nietzsche, I don't think we can. All of the identified topics fit nicely into either genre. In contrast, we can in the case of Mill. Many of the topics, such as his father and his role in Mill's education, as well as Mill's civil service, I do not think appear in his published works. If I've overlooked some such instance, I do not think they play a vital role (e.g., they could be the subject of an example, but not one that could not be modified without changing the import of the discussion). Like the case of Nietzsche, the case of Trollope seems ambiguous. For example, Trollope tells the reader that "Percycross [in Ralph The Heir] and Beverley [in Trollope's life] were, of course, one and the same place" (Chapter XIX. Ralph The Heir--The Eustace Diamonds--Lady Anna--Australia.) At the same time, some of the topics identified are absent, or at least not importantly present, in Trollope's writings, such as working in the post office. Similarly, the topic concerning authorial income could be just as much part of a retrospective as an autobiography. This insight in turn brings with it another. It is arguable that the line between authorial retrospective and autobiography is crisper in the case of Mill than in the case of Trollope because philosophy offers significantly fewer opportunities for autobiographical connections than fiction does.
Readers familiar with Nietzsche will be quick to point out his oft-quoted remark in Beyond Good and Evil that:
Gradually it has become clear to me what every great philosophy so far as has been: namely, the personal confession of its author and a kind of involuntary and unconscious memoir [Allmählich hat sich mir herausgestellt, was jede grosse Philosophie bisher war: nämlich das Selbstbekenntnis ihres Urhebers und eine Art ungewollter und unvermerkter mémoires] (BGE §6).
Although rarely quoted, the sentence continues:
likewise that the moral (or immoral) intentions in every philosophy constituted the real germ of life from which the whole plant had grown [insgleichen, dass die moralischen (oder unmoralischen) Absichten in jeder Philosophie den eigentlichen Lebenskeim ausmachten, aus dem jedesmal die ganze Pflanze gewachsen ist].
Although "germ of life" is a good, if not literal, translation of Lebenskeim, I suspect that modern readers are more likely to think of a product in an organic or bio store than what Nietzsche had in mind. It is not a coincidence, I would argue, that "Lebenskeim" is used to translate the Stoic “seminal principle [σπερματικοὺς λόγους]" Marcus Aurelius uses in his Meditations (VI: 24). Either way, I don't think that Nietzsche has in mind the sort of correspondence we find in Trollope, which is not to say that he is denying that such things occur. Nietzsche has in mind something more subtle, more psychological, than would be picked up by topic modeling. It's also a more interesting claim than merely saying that it is somehow biographically revealing if, say, a philosopher uses a dog as an example.
One could object that my alternative question is a leading one since what I described is exactly what Nietzsche does in the largest chapter of Ecce Homo. So I re-ran the analysis using only the first two chapters, "Why I am so Wise" and "Why I am so Clever." This omits the chapter "Why I Write Such Good Books," where Nietzsche goes through his books one by one. I also omitted the final chapter, "Why I am a Destiny," since it is not particularly autobiographical. After much fiddling, I found that BERTopic gave the best results when limited to four topics. As was the case with the analysis of Ecce Homo in toto, one topic always included the terms wagner, german, germans, and culture. The term my was included in at least three of the four topics, and occasionally all four. Another topic usually included me, my, myself, and life. This gives the best evidence so far that Ecce Homo is at least partly an autobiography. Curiously, one topic always included Paris. Although Nietzsche mentions the city four times in the two chapters under consideration, never does he make a direct personal connection with it. Lastly, there was always a topic whose underlying theme eluded me. For example, here are the terms generated on the last run of the analysis: oneself, not, cannot, attack, ressentiment, bad,things, pity, great, and my.
Perhaps another potentially useful comparison would be between Ecce Homo and another of Nietzsche's books. I chose Dawn and The Gay Science since they seem to be among his most personal. Indeed, Michael Ure recently claimed "We can therefore mark out [The Gay Science] as a philosophical autobiography."[1] I also tried Twilight of the Idols as it was written in the same highly productive timeframe as Ecce Homo.
Despite much fiddling, I could never get a decent analysis of Dawn. Interestingly, even though all of the books analyzed are made up of consecutively numbered sections, this is the only one were BERTopic included numbers in the clusters. In other contexts, I would have added "or" to the stopwords. I didn't in this case because Nietzsche seems to think that it's important that it is the last word of Dawn (EH "Dawn" 1). I likewise kept "or" off the stopword list for the other books analyzed. Here are some of clusters:
- Once gold was identified as belonging to a cluster. The word occurs by itself seven times, and four additional times in words like "golden" or "goldsmith."
- solitude, alone, 177 [This section is entitled "Learning solitude"], fairer [occurs only in 485, which is about solitude], 249 [This section is entitled "Who is ever alone?"].
- vita [Nietzsche uses both 'vita activa' and 'vita contemplativa'], praise, duel, spirit, making, misunderstanding, brave, different, and good.
After a great deal of experimentation, combining some of the best results on The Gay Science would produce clusters of the following. In every case, these clusters would also have seemingly unrelated words, and there would be entire clusters of seemingly random words. So here are the edited best results:
- There was invariably a topic concerning Christianity. The terms were typically: god, faith, church, christianity, christian, and religion.
- A common, but not constant, topic involved: pain, pity, distress, and suffering.
- Equally as often was a cluster of terms like this: love, women, woman, and sex.
The better results included clusters like these, but they were never guarenteed:
- german, germans, germany, wagner, jews, schopenhauer, and europe.
- art, artists, artist, and works.
- music, poet, poetry, poets, musician, and rhythm.
As with Dawn, the results with Twilight of the Idols, even after much fiddling, were disappointing. Here are portions of the best clusters:[2]
- socrates, plato, greek, greeks, athens, thucydides.
- morality, christian, moral, religion, life,[3] christianity.
- german, germans, spirit, germany
- art, artist, pour,[4] poet, tragic, life.
Disappointed in these results, I added the following specification to the code: embedding_model = SentenceTransformer('all-MiniLM-L12-v2'). Initially I tried the all-MiniLM-L6-v2 model but I found that this one worked a little better. Interestingly, the clusters now included seemingly random numbers between 32 and 45. Unlike Nietzsche's other books, the sections in Twilight of the Idols are not consecutively numbered so these numbers do not uniquely correspond to particular sections. When I added those numbers to the stopwords list, the clusters began including numbers in the 20s. Once I added those to the stoplist, numbers no longer appeared. Even after these change and adding more words to the stoplist, the results were only incrementally better. The model now included beer in the German cluster (see "What the Germans Lack" §2) and english in the morality cluster (see "Skirmishes" §5). The model did hit on one theme of the book, which might be summarized as the mistakes humans have made concerning causes. Specifically, the following terms were clustered: cause, causes, will, causal, world, explanation, and error.
One cluster that appeared in multiple runs concerned marriage. Specifically, marriage, institutions, decline, institution, modern, equal, society, liberal, and power. This is interesting for two reasons. First, all of Nietzsche's eleven uses of "marriage" occur in one section ("Skirmishes" §39). Secondly, even in that section, Nietzsche presents the topic of marriage as an example to illustrate his thesis in that "The whole of the West no longer possesses the instincts out of which institutions grow, out of which a future grows." That is to say, marriage is not one of the main topics of Twilight of the Idols.
Overall, I do not think this analysis brought me any closer to deciding whether Ecce Homo is an autobiography, although it did confirm that Ecce Homo contains less traditional biographical material than one would expect, but that's not enough to disqualify it from the genre. The results I think were nonetheless impressive, especially since I had not trained the model in any way. The next step is of course to graduate from zero-shot learning to something requiring more work on my end, but hopefully also producing more robust results.
References
[1] Michele Ure (2019). Nietzsche's The Gay Science: An Introduction, p. 1.↩
[2] Each complete cluster was ten terms.↩
[3] For those unfamiliar with N's "attack" (his word) on Christianity, his primary objection is that it is hostile to life.↩
[4] Of the six times "pour" appears in Twilight of the Idols, four occur in "Skirmishes of an Untimely Man" §24 and in the French expression "L'art pour l'art." The fifth is a translation of the French répandre ("Skirmishes of an Untimely Man" §46), and the final occurance is in "What I Owe to the Ancients" §1.↩