Gender semantics on the Spanish Wikipedia
in Spanish, female gender is often attributed to objects that are “used by women, natural, round, or light” and male gender to objects “used by men, artificial, angular, or heavy.”
The source is a similar statement in the book Social Psychology of Culture by Chi-Yue Chiu and Ying-yi Hong. No evidence is provided for this claim, as far as I can see.
This project is an attempt to test the above claim by comparing masculine and feminine nouns on the Spanish Wikipedia, specifically the claims concerning natural and artificial objects.
The analysis found no difference between the grammatical gender of nouns in Wikipedia articles to do with nature as compared with manmade objects.
You can check out my code on GitHub.
- The program
download_wikipedia.pydownloads a specified version of the Spanish Wikipedia from a data dump.
- The master list of masculine and feminine nouns comes from the dictionary FreeLing.
- The program
count_all_pages.pyiterates through all the articles and tallies the masculine and feminine nouns, writing the results to a text file.
- The program
sum_counts.pysums the masculine and feminine nouns in a given article, and all the articles that it links to.
- The word de is a preposition and a feminine noun in Spanish.
Every instance of the preposition de would be incorrectly counted as a feminine noun by
count_all_pages.py. There are presumably many other examples of words being incorrectly classed as a noun.
- Certain words occur in Wikipedia articles frequently. For example, Referencias meaning References occurs in virtually every article. These words are not related to the subject matter of the article, but would still be counted.
Here are the results. This table lists the number of nouns in the listed article and all the articles it links to, together with the ratio of masculine nouns to feminine nouns.
|Number of nouns
The articles are just a few that I selected which I thought reflected the natural vs artificial dichotomy. The articles which are artificial seem to have lower masculine:feminine ratios, while the natural articles have a higher ratio. The difference is very small, however, so it’s possible that there is no discernible difference. This is not predicted by the claim we are testing, which would predict a higher ratio for artificial articles compared to natural articles.
- It is possible this negative result is a consequence of the shortcomings listed above.
- It is possible I have incorrectly understood what artificial and natural objects are.
Resources and licenses
- I used this article as a basis for downloading and parsing the Wikipedia articles.
- I used the list of nouns in the dictionary FreeLing. This list was extracted from the Spanish Resource Grammar projects and was developed at the Institut Universitari de Linguistica Aplicada of the Universitat Pompeu Fabra. It is distributed under a “Lesser General Public License For Linguistic Resources” license.
- The rest of my code is licensed with an MIT license.