Tired of reading documents that use masculine nouns and pronouns for talking about people in the abstract, especially when referring to scientists, programmers, and engineers? I know I am! Here's a script I wrote to help me get through Thomas S. Kuhn's The Structure of Scientific Revolutions (1962) [1]. Maybe it's useful for you, too! Works by re-gendering all references to men in PDFs & text-based files.
- Get a PDF or text copy of the document you want to convert. HTML and LaTeX should probably also work, but I didn't test.
- Install dependencies (see Setup section).
- Run the
scripts/feminize.pyscript in this repository to output a converted version of your document.
$ python scripts/feminize.py --help
Usage: feminize.py [OPTIONS]
Read content from INPUT file (.pdf or .txt), replace all male nouns and
pronouns with female ones, and write to OUTPUT file.
Options:
-i, --input TEXT Input file as .txt or .pdf [required]
-o, --output TEXT Output file as .txt [default: feminized.txt]
--help Show this message and exit.
- Check out the printed statistics to see how many words were converted in your document.
- If you want to read the converted document as a PDF, you can copy the text from the output file into a Microsoft Word document (or equivalent) and choose "export to PDF". (let me know about any easy-to-use python libraries for generating PDFs! Wait do I really want that... PDFs are a terrible format. revised: tell me about your favorite way to read plaintext!)
The PDF parsing requires the apache tika library, which you can install like a normal python library along with the other (very minimal) requirements.
pip install -r requirements.txt
However, if you don't already have a JDK on your machine (my Mac did not), you should first install it from here. See my stream of consciousness thoughts in the Background & documentation section if you get stuck. Note that you only need to worry about this if you want to convert a pdf. (Of course, you probably want to convert a PDF ;P)
$ python scripts/feminize.py -i ../documents/Kuhn_Structure_of_Scientific_Revolutions.pdf
Grabbing content from documents/Kuhn_Structure_of_Scientific_Revolutions.pdf...
...parsing as PDF
Running m[]n to wom[]n conversion...
* 107 m[]n are now wom[]n
Running he to she conversion...
* 189 he are now she
Running him/his to her conversion...
* 251 him/his are now her
Writing text to output file: feminized.txt
* Total 547 word changes out of 79992 (0.68%) to feminize your text!
* 12.33% of sentences changed (assume one change per sentence for est. 4436 sentences)
The study of paradigms, including many that are far more specialized than those named illustratively above, is what mainly prepares the student for membership in the particular scientific community with which he will later practice. Because he there joins men who learned the bases of their field from the same concrete models, his subsequent practice will seldom evoke overt disagreement over fundamentals. Men whose research is based on shared paradigms are committed to the same rules and standards for scientific practice. That commitment and the apparent consensus it produces are prerequisites for normal science, i.e., for the genesis and continuation of a particular research tradition.
(excerpt from the chapter, The Route to Normal Science; see [1])
The study of paradigms, including many that are far more specialized than those named illustratively above, is what mainly prepares the student for membership in the particular scientific community with which she will later practice. Because she there joins women who learned the bases of their field from the same concrete models, her subsequent practice will seldom evoke overt disagreement over fundamentals. Women whose research is based on shared paradigms are committed to the same rules and standards for scientific practice. That commitment and the apparent consensus it produces are prerequisites for normal science, i.e., for the genesis and continuation of a particular research tradition.
(based on an excerpt from the chapter, The Route to Normal Science; see [1])
For best results, use python 3.7.
PDF parsing is a real nightmare, but tika worked for me basically straight out of the box once I installed the JDK (see Setup section).
One note is that your converted document might have a ton of newlines before you get to the actual content, so before you freak out at an empty output file, try scrolling down. If you still don't see any output, you can check out the two log files that get written: the "_content.log" and the "_metadata.log". If these files are both empty, something went wrong with the parsing & you're on your own :)
You can read the Background & documentation section to follow my development process, which might help.
- Why not "they"?
I think if the "default" is a male scientist, programmer, or engineer, your mind will still conjure a male when reading "they", even if it's a "gender-neutral" word. The idea of this script is to re-train defaults. If it doesn't matter, let's re-paint the world female!
- What happens to pronouns that refer to human (real or imaginary) males?
They also get re-gendered! Everyone gets re-gendered! At first I started thinking about ways to fix this, but then I decided it doesn't need fixing :)
- Why not call it "womanizer.py"? Isn't that a better pun?
Yeah :D
Here was my name brainstorm for this repo:
- regenderization
- transcription
- transcrypt
- womanizer
- his_and_hers
- _he
- m2f
- feminizer
- text_that_is_easier_on_the_eyes
Note: this is stream of consciousness
2h project including thinking time and file downloading time.
I keep getting distracted by all the times Kuhn says "men in science". I might have an easier time reading The Structure of Scientific Revolutions if I replace all instances of "men" with "women".
I found this apache tika library for parsing PDFs. https://github.com/chrismattmann/tika-python
The tika library kicked off a download of a jar file that ended up here:
$ python scripts/feminize.py -i ../texts/Kuhn_Structure_of_Scientific_Revolutions.pdf
2020-02-08 11:06:05,685 [MainThread ] [INFO ] Retrieving http://search.maven.org/remotecontent?filepath=org/apache/tika/tika-server/1.23/tika-server-1.23.jar to /var/folders/9h/ttx3gqgd6jbfxzybwz7mqq7r0000gn/T/tika-server.jar.
Still I didn't have a JDK on my machine, so I had to download it here: https://www.oracle.com/technetwork/java/javase/downloads/jdk13-downloads-5672538.html
And the last thing to do is remind tika where the jar is (but I think it might already know):
TIKA_SERVER_JAR=/var/folders/9h/ttx3gqgd6jbfxzybwz7mqq7r0000gn/T/tika-server.jar
The readme for the tika project explains all the env vars to set, but that's the only one I needed (and I probably didn't because I didn't move the jar from where it was downloaded).
And it's done! Below script counts the number of instances of men & women (in different forms), replaces all of the former with the latter, and returns a text file with the changes.
python scripts/feminize.py -i ../texts/Kuhn_Structure_of_Scientific_Revolutions.pdf
I then copy-pasted the text into a word doc and exported as a PDF (somehow it wasn't easy to find a text-to-pdf python library... I couldn't ever imagine why...). The spacing on the result isn't great, but i's not horrible either, and I'll finish reading the book like this.
I'm satisfied with the output for now. In a second iteration I could work with the pronouns (spoiler: I later ended up including changing the pronouns). For instance, right now there are sentences like:
"Intellectually such a woman has made his choice, but the conversion required if it is to be effective eludes him."
Right now it's amusing enough to me that I could leave it, OR, also not a lot of work, change ALL the pronouns into feminine pronouns. Changing all the pronouns would leave us with:
"Maxwell herself was a Newtonian who believed that light and electromagnetism in general were due to variable displacements of the particles of a mechanical ether."
Actually in the process of looking for when "his" actually refers to a specific person, I noticed so many generic ones, and I think i'll go ahead with the pronouns.
Ok, I changed the code to by default swap the pronouns as well. I like this even better, due to the actually not insignificant number (~370), compared to instances of "men" and "woman" (~100).
(edit: This snippet is from an older version of this code; current counts in the Example section earlier in this document. Thank you, regular expressions, for detecting more 10 more noun changes and 70 more pronoun changes than my original hacky code could find.)
python scripts/feminize_text.py ../texts/Kuhn_Structure_of_Scientific_Revolutions.pdf
../texts/Kuhn_Structure_of_Scientific_Revolutions.pdf
Men = 97
Women = 1
Male pronouns est. = 370
Men = 0
Women = 98
But it also occurred to me along the way that it might be less of a hassle to just change them all to neutral pronouns (e.g. they, their). One reason I'm going to read the text with the female pronouns is because I want to feel the impact of it. "They" still conjures a "male image", where "men" are still the default. So if pushed completely in the other direction we can see the world in a different light.
I can't comment at all about who was in Kuhn's world, and who were the scientists in history. If women were written out or they weren't there at all, it's immaterial now to the (somehow still existing) uphill battle to get in and stay in technical and scientific domains due to issues orthogonal to desire and ability to do the work. In order to unburden women who are already here, this is a small act to test a subtle effect.
The other reason I'm doing all the pronouns also, instead of randomly spattering femaleness around is because sometimes the feminine pronouns are clearly used in a text more for the "ignorant student", and I want to not give randomness the chance to conjure that image.
[1] Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago: University of Chicago Press.