predict-idlab / gssp_analysis

Analysis notebooks and scripts of the gssp web app data collection

Analysis of the GSSP

This repository contains the analysis code for the Ghent Semi-spontaneous Speech Paradigm (GSSP). The GSSP is a picture description task that is used to capture (near) spontaneous speech in a controlled setting.

• The data is collected via a web application and can be found on kaggle
• The paradigm is described in detail in this preprint manuscript.
• The supplementals can be found here
• The notebooks README contains a thorough description of the speech parsing and analysis notebooks.

In a nutshell the r-scripts folder performs a thorough statistical analysis of the arousal & valence scores for the audio files. The outcome can be observed in a shiny html file. All speech data transformation and analysis is performed in the notebooks folder.

The utilized python packages are listed in the pyproject.toml file and the utilized R packages are listed in the scripts/r_packages.txt file.

---

Folder structure

├── docs
│   └── cgn             <-- CGN related documentation
├── GSSP_utils          <-- Python functions shared across notebooks (and CGN parsing)
├── loc_data            <-- Local data shared across notebooks
├── notebooks           <-- the analysis Jupyter notebooks
├── reports             <-- Generated figures from the notebooks
└── scripts             <-- R scripts for statistical analysis & shiny app

Cite

• A preprint manuscript is available on psyArxiv.

@misc{van_der_donckt_2023,
 title={Ecologically Valid Speech Collection in Behavioral Research: The Ghent Semi-spontaneous Speech Paradigm (GSSP)},
 url={psyarxiv.com/e2qxw},
 DOI={10.31234/osf.io/e2qxw},
 publisher={PsyArXiv},
 author={Van Der Donckt, Jonas and Kappen, Mitchel and Degraeve, Vic and Demuynck, Kris and Vanderhasselt, Marie Anne and Van Hoecke, Sofie},
 year={2023},
 month={Mar}
}

---

👤 Jonas Van Der Donckt, Mitchel Kappen