Tutorial on Modeling Language Change

This repository holds notebooks, data, and slides related to tutorials on computational models of language change at NAACL and IC2S2.

• The data for both the tutorials should be found in the data directory. In addition to the data here, two larger data files can be found at bitbucket: https://bitbucket.org/jacobeisenstein/language-change-tutorial-data

• The notebooks for the NAACL tutorial are under naacl-notebooks and the notebooks for the IC2S2 tutorial are under ic2s2-notebooks

• These slides are used for the NAACL tutorial and these slides are for the IC2S2 tutorial.

Installation

All notebooks are based on Python 3. To ensure that you have all the necessary packages, create a new conda environment and then activate it (update conda if it asks you to).

pip install --user conda #if necessary
conda env create -f environment.yml # create the environment. update conda if it complains.
source activate time_tutorial # activate the environment

You can then launch a notebook as follows:

cd ic2s2-notebooks
jupyter-notebook HappierOnTheWeekend.ipynb

The FollowTheLeader.ipynb notebook requires you to install pyhawkes using pip install --user pyhawkes autograd (Linux) or ~/miniconda3/envs/time_tutorial/bin/pip install pyhawkes autograd (Mac). If you have trouble, you may want to clone the repo directly here.

Data

To get the additional data files, run:

git clone https://bitbucket.org/jacobeisenstein/language-change-tutorial-data.git

Alternatively, you can run

bash download_data.sh

which will download all the data and setup the symbolic links for you.