Gallicagram.

In a recent comment, Y wrote:

Recently announced: Gallicagram, an ngram site for French sources and much more. It’s far more advanced and slicker than Google ngrams. Although it doesn’t have access to Google Books, it does have access to MajinBooks, a word statistics database built upon shadow libraries. It’s non-commercial.
Maybe it deserves a posting?

Voilà!

Comments

  1. Image Stu Clayton says

    Although it doesn’t have access to Google Books, it does have access to MajinBooks, a word statistics database built upon shadow libraries.

    Shadow libraries contain pirated text, amirite?
    In this case discarded once the statistics had been prepared. Like LLMs do.

  2. I just tried it on “beur” and got no useful results.

  3. Hm, I see. There are many text collections, but most don’t show the context, only the frequency plot, including Majinbooks français (–2020), Opensubtitles (–2020), and Rap français (–2024). Presse de Gallica does show context, but it only goes to 1950. With that one, I don’t see an obvious way to restrict it to the word alone. It also shows “beurre” etc. Oh well.

  4. Hi, Gallicagram developer here. Thanks for the diffusion!

    Yes it’s quite hard to get the context for most corpus. For Gallica we really on their API, and it’s not too strict, so “beur” will include “beurre”, no way to restrict if further, as far as we are aware. The only one where we do the context ourselves is the French rap, there it is strict. It’s the smallest corpus, which makes it doable. Majinbook is so large that it would require more computational resources than we have, and i’m not sure it’d be legal anyway.

    For recent searches, I recommend Le Monde, it’s continuous since 1945, and the context search sort of work (again, we use their search engine under the hood, and it it not too strict).

    Anyway, hope we’ll be useful to you guys!

  5. Thanks very much for dropping by to explain!

Speak Your Mind

*