Methodology

Stuttering Experience is built from public Reddit posts and comments related to stuttering. All content in the database is searchable by keyword, allowing users to look for specific words, phrases, medications, situations, emotions, or experiences that appear in the original posts and comments. This basic keyword search is useful, but it has important limits. People often describe similar experiences using different words, misspellings, slang, abbreviations, or indirect descriptions. A keyword search can only find what it is told to look for.

To move beyond keyword search, Stuttering Experience uses AI-assisted qualitative coding. Each post or comment is analyzed using a deductive coding approach, meaning that the AI applies a predefined framework of themes, subthemes, and codes rather than inventing categories from scratch. These themes and subthemes were developed to reflect both stuttering research and the practical needs of people using the website to explore lived experience. This allows the database to organize posts around broader concepts such as anticipation, avoidance, emotional experience, therapy, parent concerns, medications and substances, school and work, identity, and social participation.

In addition to assigning themes and subthemes, the AI extracts key searchable concepts from each post or comment. These include things such as medication names, medication classes, recreational substances, speaking situations, emotions, therapy methods, contributors to variability, and other concepts that may be useful for advanced searching. This gives the database a richer structure than keyword search alone. Users can still search for exact words, but they can also explore posts grouped by meaning.

For example, a keyword search for posts about psychedelics might look only for terms such as "LSD" or "psilocybin." That approach can miss posts where people use slang, misspell a drug name, refer to a broader drug class, or describe the experience without using the expected keyword. As a proof of concept, Stuttering Experience used medication and substance classes as part of the AI coding process so that posts could be labeled when they described experiences involving a range of medications, supplements, and recreational substances. Because the AI uses semantic and real-world language understanding, it can identify relevant posts even when the wording is inconsistent. This allows the database to capture discussions that may not appear in a simple keyword search.

This approach has value for both the general public and researchers. Public users can explore the database through themes and searchable concepts that reflect real experiences of stuttering. Researchers can use the coded structure to identify patterns, generate research questions, locate relevant examples, and study topics that would be difficult to capture through keyword searches alone. The goal is not to replace careful human interpretation, but to make a large body of public discourse more searchable, organized, and useful.

The data should be interpreted with care. Reddit posts and comments are public online discourse, not clinical records, verified diagnoses, or representative samples of all people who stutter. AI coding also has limitations, and coded results should be understood as a structured discovery tool rather than definitive interpretation. Stuttering Experience is intended to support research, education, and public understanding, not to provide clinical or medical recommendations.