This repo contains data and code used by Isobel Lester to perform sentiment analysis on two French language corpora - one translated by professional translators, and one translated by machine translation.
This work was guided by four question:
- How accurately can a machine translate emotion and the desired impact of the author compared to a human translation?
- What variations are produced between a human generated and AI generated translation? Do they alter the meaning?
- What limitations are there to AI translation software?
- Should complex human languages be reduced to numbers? How accurately can it do this?
Both these and the project findings are discussed in greater detail in the Project Report.
Anyone wishing to run the code themselves has two options:
- Run everything in this Google Colab notebook
hosted by James Baker.
- To set up a python environment on your machine for local processing, follow the steps described at Zoë Wilkinson Saldaña, "Sentiment Analysis for Exploratory Data Analysis," Programming Historian 7 (2018), https://doi.org/10.46430/phen0079.
This notebook was produced in spring/summer 2022 by Isobel Lester as part of a Digital Humanities Internship funded by the School of Humanities at the University of Southampton.
Code in this repo is released under a CC-BY license.