This runs a SpaCy implementation of the Stanford NLP dependency parser and writes the output to .conll files in the Universal Dependencies format. It was originally made to parse the Corpus of Contemporary American English (COCA).
Create and activate a virtual environment:
ENV_NAME=
python3 -m venv $ENV_NAME
source $ENV_NAME/bin/activate
Run pip install -r requirements.txt to download the dependencies. To run the parser, you need to specify input and output directories, and whether you want to use a GPU (default false, add the --use_gpu flag to make it true.
INPUT=
OUTPUT=
python run_parser.py --input_dir $INPUT --output_dir $OUTPUT
The output files can be read by a parser like conllu or pyconll.