Best Practices in the Creation and Use of Emotion Lexicons

Emotion lexicons are powerful tools with real-world consequences: These word-emotion association databases are used to analyze sentiments about products, policies, and health conditions, meaning errors or misuse can directly harm people through flawed decision-making based on inaccurate emotional analysis.

There is no universal emotion lexicon: The paper emphasizes that appropriate lexicon selection depends on context—what works for analyzing social media posts about products may not work for medical research or literary analysis, requiring practitioners to thoughtfully match their tool to their specific application.

Cultural and linguistic biases are embedded in emotion lexicons: Since these lexicons are human-created, they reflect the values and perspectives of their creators, potentially leading to biased or inappropriate inferences when applied across different cultures, languages, or communities without careful consideration.

Responsible use requires transparency about limitations: The paper advocates for practitioners to understand and communicate the constraints of their chosen lexicon, including what emotions it captures, what populations it was designed for, and how it might fail—rather than treating it as a neutral, objective tool.

Thoughtful design matters at every stage: From deciding which emotions to include, to creating the lexicon, to interpreting results, the paper calls for deliberate ethical consideration throughout the process to avoid causing harm through seemingly technical decisions.