A Comparison of Human Narrative Coding of Redemption and Automated Linguistic Analysis for Understanding Life Stories: Redemption and Automated Linguistic Analysis
Sara J. Weston, Keith S. Cox, David M. Condon, Joshua J. Jackson
*Journal of Personality*
Abstract
The majority of life narrative research is performed using trained human coders. In contrast, automated linguistic analysis is oft employed in the study of verbal behaviors. These two methodological approaches are directly compared to determine the utility of automated linguistic analysis for the study of life narratives.
Why we asked
Most research on life stories relies on trained human coders. Automated word-count tools, in contrast, are often used to study how people speak and write. We wanted to know whether software could stand in for human coders.
What we did
We ran two studies. The first used in-person interviews (N = 158) and the second used life stories collected online (N = 242). In each, we compared human-coded redemption scores with the output of the Linguistic Inquiry and Word Count program (LIWC). We also looked for patterns of language with an exploratory principal components analysis.
What we found
Redemption scores were modestly correlated with some LIWC categories and unassociated with the language components. The language patterns did not replicate across the two samples, so the structure we found in one did not carry over to the other. Redemption scores and the language components each predicted life satisfaction up to three years later, and they did so independently of one another.
What it means
In other words, human coding and automated analysis were not redundant, and each added information that the other did not. We read them as complementary methods, not substitutes.