Software Open Access

Supplementary Materials for "Creating a Frequency Dictionary of Spoken Hebrew"

Juan D. Pinto

This repository houses all the scripts and files used to create the Frequency Dictionary of Spoken Hebrew (FDOSH), along with the dictionary itself. This project was created as part of my MA thesis at the University of Texas at Austin in 2018. The thesis itself describes the creation process—and the use of each script—in depth, and can be found in the University of Texas thesis repository. A GitHub repository for the thesis manuscript can also be found at https://github.com/juandpinto/thesis-manuscript.

The scripts make use of OPUS's OpenSubtitles2018 collection, which is a mega-corpus of cleaned, tokenized, and parsed versions of XML files originally obtained from opensubtitles.org. The final frequency dictionary consists of Hebrew lemmas, and is arranged based on a usage coefficient of Gries' (2008) deviation of proportions, or UDP. It also includes frequency and range measures for each entry.

The most important files in this repository are listed below.

- The Frequency Dictionary of Spoken Hebrew (FDOSH): export/frequency-dictionary.tsv
- The main script used for creating the dictionary: create-freq-list.py
- The script used to clean the OpenSubtitles2018 corpus: single_file_extract.py
- The script used to fetch movie metadata for each subtitle file in the corpus: OMDb-fetch.py
- The script used to find the shared entries in two different frequency lists: list_comparison.py

The script used to fetch movie metadata (OMDb-fetch.py) uses Derrick Gilland's omdb.py library, which is a Python wrapper around the OMDb API (Open Movie Database API). OMDb is, in turn, a project that makes use of IMDb (Internet Movie Database) for its data. For each subtitle file in the corpus, the script finds the IMDB ID, title, year, and original language(s). The movies-info/ folder contains extensive lists of the metadata found for the movies used to create the FDOSH.

Each script includes detailed notes within the comments to allow them to be clear and easily customizable. This project is licensed under the MIT License, so feel free to clone and use as you see fit. Suggestions and pull requests are also welcome.

Files (679.9 kB)
Name Size
juandpinto/frequency-dictionary-v1.0.zip
md5:d8d6fdddb7fb53fcec026117925a0380
679.9 kB Download
29
3
views
downloads
All versions This version
Views 2929
Downloads 33
Data volume 2.0 MB2.0 MB
Unique views 2828
Unique downloads 33

Share

Cite as