Fig 1.
(A) Hypothesized and (B) empirically derived language markers of well-being in healthy aging narratives. (A) Conceptual working model of language indicators based on the preregistration. (B) Empirically derived measurement model of language indicators in N = 701 study participants. All language indicators represent word categories from Linguistic Inquiry and Word Count (LIWC). Residuals were allowed to correlate for conceptually similar language indicators or those that belonged to the same higher-order LIWC category (e.g., “friends” and “family”). Cogproc = cognitive processes, quant = quantifiers, introg = interrogatives, affil = affiliation words.
Table 1.
Sample description (N = 701).
Fig 2.
Combined model with language and questionnaire indicators of well-being (schematic illustration).
N = 701 study participants. Depicted are standardized factor loadings with standard errors; statistically significant paths are presented in boldface. The model is only illustrated schematically for space reasons. For the full results, please refer to S1 File. Language-based latent variables: Affective, evaluative, social; questionnaire-based latent variables: Physical health, psychological well-being. L1, L2, … = language indicator, Q1, Q2, … = questionnaire indicator. * p < .050, ** p < .010, *** p < .001.
Table 2.
Standardized factor loadings in Model 1 (language measurement model).
Table 3.
Associations between language and questionnaire-based latent variables in Model 3 (combined path model).