1

I am working on a deep learning translator for latin and I need a dataset of about 140,000 latin sentences in a text file and their corresponding english translation in another text file. I have done quite a bit of research and I have not been able to find a dataset like this that I can use. Does anyone know where I can find something like this, all help is appreciated.

2
  • Welcome to the site! I'll point you to two older questions for starters, on Google Translate with Latin and online text corpus options. I'll also have to mention that translations are not unique, so you should build in the option to have multiple translations of the same sentence. And also that sentences in isolation tend to get translated surprisingly differently than in context.
    – Joonas Ilmavirta
    Jan 3, 2023 at 16:37
  • The Latin Vulgate is pretty cleanly divided into verses, although there are far fewer than 140,000 of them. There are English translations of the Vulgate freely available, but they in Elizabethan English, not 21st century English.
    – Figulus
    Jan 14, 2023 at 22:35

1 Answer 1

1

Classical Latin was not originally written as a series of separate sentences; periods and the like are a modern innovation. When such punctuation is added in, the "sentences" thus created are often unusually long compared to those in English, meaning that there is no 1:1 correspondence.

So I don't think that "sentence-by-sentence" translation is an ideal goal. Instead, translating section-by-section may work better; you can get a large dataset of Latin works and translations, sometimes "chunked" into relatively small pieces, at Perseus.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.