Negative language transfer classification

This repository hosts the code used to process and analyse the negative language transfer dataset. This implementation trains two classification models to predict whether English learner errors are related to negative language transfer. Parameter tuning is used to select the best features to model the task. After training and testing, the models' performances on the task are compared and contrasted.

• All scripts in this repository were implemented in Python 3.7

Data

The data used in this project comes from the the learner English negative language transfer dataset. It contains Chinese native speakers English errors annotated with information about their relation to negative language transfer. To have access to this dataset, you need to follow the guidelines provided in its repository.

Processing

The negative language transfer annotated data was preprocessed to extract the part-of-speech and dependency tags associated with the learner errors and their surroundings. The pre_process_data.py script does this syntactic feature extraction and creates a new processed datafile. To run the preprocessing script replace the file name on line 80 with the path to the negative language transfer annotated dataset.

The process_data.py script should be ran after data preprocessing. It creates dummy and encoded versions of the data's categorical features. Then, it splits the processed dataset into train and test files.

Tuning

To select the best parameters for the random forest model used in this classification task, a systematic search over the parameter space was performed. The parameter combinations creation can be found in the create_input_files.py script. The parameter tuning results are available in data/results_tuning.csv.

Analysis

The training and testing of the final classification models are available in a jupyter notebook. Run jupyter notebook on the repository's root to access the NLT_classification.ipynb interactive analysis notebook.

Reference

Leticia Farias Wanderley, Nicole Zhao, and Carrie Demmans Epp. Negative language transfer in learner English: A new dataset. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 3129–3142, Online, June 2021. Association for Computational Linguistics.

@inproceedings{farias-wanderley-etal-2021-negative,
    title = "Negative language transfer in learner {E}nglish: A new dataset",
    author = "Farias Wanderley, Leticia  and
      Zhao, Nicole  and
      Demmans Epp, Carrie",
    booktitle = "Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies",
    month = jun,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2021.naacl-main.251",
    pages = "3129--3142"
}