This repository contains a Jupyter Notebook that focuses on lingustic analysis of the descriptions of psychedelic and non-psychedelic mystical experiences.
The dataset in sav
format is openly available courtesy of Mendeley Data:
Owsiak, M. (2025). Dataset - Psychedelic and Non-Psychedelic Mystical Experiences in Polish Adults. https://doi.org/10.17632/bhkb7zmb9v.1
Languages, frameworks and environments:
> Python
> Jupyter Notebook
> Markdown
Libraries and packages:
> NLTK
> wordcloud
> transformers
> scikit-learn
> SciPy
> NumPy
> pandas
> statsmodels
> Matplotlib
> PIL
> IPython
APIs and pre-trained models:
> DeepL REST API
> EmoRoBERTa by Hugging Face
Using plain Python
along with pandas
and NumPy
predefined functions, the following steps were performed:
- Merging duplicated variables into consistent dataset
- Recoding questionnaires' scorings
- Calculating summed results
- Cleaning and preparing data for analysis
- Calculating word frequency distribution using
NLTK
for tokenization andwordcloud
for stopwords. - Conducting series of correlation analysis using
CountVectorizer
fromscikit-learn
andpearsonr
function fromSciPy
. - Performing emotional sentiment NLP analysis using
EmoRoBERTa
pre-trained model fromtransformers
library.
The results of the study were visualized as bar plots employing Matplotlib
library and as word clouds using wordcloud
library.
The analysis revealed that descriptions of complete mystical experiences often emphasize positive emotions and existential themes, such as love and cosmic concepts, while negative emotions and factual terms are more associated with less intense experiences. Significant differences were noted in the frequency of specific emotions like admiration and fear between the two groups, with complete experiences showing more positivity and less negativity. Overall, the findings highlight how language reflects the intensity and nature of mystical experiences, with positive emotional and existential themes prevalent in more profound experiences.
The original research was performed and presented repository was created by Michał Owsiak. The author declares no potential conflicts of interest with respect to the research, authorship, and/or publication of this repository.