The following project information is limited solely to the direct digital scholarship portion of the research.
Project Title: Metaphors, Compound Sentences, and Sentiment
Click Image to Navigate to Interactive Visualization
Description: Parse ‘Save Me the Waltz’, by Zelda Fitzgerald to identify trends in the numbers of metaphors, compound sentences, and sentiment in each section and chapter.
Researchers Involved:
- Joshua Been, Baylor Libraries
- Eileen Bentsen, Baylor Libraries
Procedure:
- Metaphor detection method uses the Python Natural Language Toolkit (NLTK) to measure the similarity of the definitions of noun-noun pairs related by nominal subject. This technique was modified from this Metaphor Detection script. To remove the need for human validation of the correct definition for the noun pairs, the first definition was automatically used.
- Sentiment calculation method uses the Python VADER (Valence Aware Dictionary and sEntiment Reasoner) library.
- Compound sentence detection method uses the FANBOY rule (fanboy = [‘for’, ‘and’, ‘nor’, ‘but’, ‘or’, ‘yet’, ‘so’]). Sentences with a comma followed by a fanboy were identified as a compound sentence.
- Sentence parsing uses the Stanford Lexical Parser.
- Tableau Desktop was used to visualize ratio of sentences per book, chapter, and section, that were identified as either a metaphor or compound sentence, as well as the average sentiment score.
Metaphors, Compound Sentences, and Sentiment