Community Wishlist Survey 2022/Bots and gadgets/Readability scores gadget

Readability scores gadget

  • Problem: Many of our articles are (accidentally) written for an academic audience, leaving behind our average reader (see for instance this journal).
  • Proposed solution: Made a gadget that can show (various) readability scores of article(s), and paragraphs within. Most readability scores like the Flesch reading ease score are based on word and sentence length. Others like Dale-Chall readability formula are based on how familiar (or common) the used words are. I think it is good to have at least one of both categories. The weighting of these tests typically varies by language, so maybe it would only be feasible to do this for the biggest 10-20 languages. Smaller wikis that translate would benefit indirectly.

    Available tools are often insufficient. Readability of Wikipedia is English-only, and quite buggy. Copy-pasting it to websites like Hemingway is time-consuming as citations need to be manually removed.

  • Who would benefit: Readers with an average education level. On simplewiki, English-language learners and younger readers.
  • More comments: Add some basic statistics about article/section/paragraph being translated is a similar proposal focused on translated text.
  • Phabricator tickets:
  • Proposer: Femke (talk) 20:08, 20 January 2022 (UTC)

Discussion

Yes "Readability of Wikipedia" does not seem to tell you what is hard to read. And it sometimes comes up with the confusing "Some errors were detected in the mark-up of the Wikipedia article, which could affect the Flesch score" even for featured articles, which seems unlikely. There does not seem to be a way to report bugs. If it is no longer being maintained maybe the creators would give it to us for someone to improve and integrate in as a tool? Chidgk1 (talk) 13:14, 26 January 2022 (UTC)

  • This could also be used to identify articles in English Wikipedia that are already (almost) simple enough to be included in the Simple English Wikipedia. Ottawajin (talk) 05:38, 29 January 2022 (UTC)
  • Readers can easily determine that an article is too difficult. Excess complexity is a real issue. Editors might be interested in a metric that would encourage them to simplify difficult articles. Lfstevens (talk) 05:45, 31 January 2022 (UTC)
  • I just have a very basic question: is this actually portable to other languages easily? - I mean given a list of "difficult words", can it be applied to Arabic, Afrkaans, or Chinese, for example (Note: I am the person who proposed the translation issue linked above)--Eptalon (talk) 22:04, 31 January 2022 (UTC)
    There is some scientific literature on the topic, such as this 2019 paper, and this 2021 paper. The first compares 6 languages (including Basque), the second one is a machine learning application that seems to be extendable more generally. It's not a trivial task, but I believe it is possible to find 'conversions' in the literature. Femke (talk) 21:11, 4 February 2022 (UTC)
  • @Femkemilene: I suggested something a bit like this, a few years ago: phab:T135321 "Vocabulary assistant for Simple Wikipedia, based on Upgoer Text Editors". Hope that helps! Quiddity (talk) 07:39, 10 February 2022 (UTC)
    Thanks. I now see there is phab:T91338 too, which is even closer to this proposal. Femke (talk) 20:33, 10 February 2022 (UTC)

Voting