Language Use Differences among Socially Isolated and Depressed Elders On Elder Tree Online Support System

Background:

Depression and social isolation are associated with increased mortality and morbidity (Cornwell & Waite, 2007; Steffens et al, 2000). Increasingly practitioners are investigating computer support systems as possible solutions to bridge the gap in social isolation and loneliness faced by many older adults. However, scholars have found that not all elders receive benefits and positive effects of these systems have been inconsistent (Chen & Schulz, 2016). Additional research is needed to understand how elders with depression and/or social isolation may be utilizing these systems differently and whether this contributes to benefits received. Multiple studies have indicated that the linguistic features of messages can provide important insights into psychological states and potential benefits to well-being (Rude et al, 2004; Holmes et al, 2007; Tausczik & Pennebaker, 2010). Thus, this study investigates differences in message linguistic features among elders with depression and social isolation utilizing an online support system, Elder Tree.

Program background: We utilized data from 387 participants (Nov. 2013 – May 2015) in 5 Wisconsin counties from a randomized, longitudinal controlled trial of the web-based support intervention, “Elder Tree” (Gustafson et al., 2015). Elder Tree includes information about healthy aging as well as discussion forums and private messaging. Participants posted 14,285 messages across 9 topic-specific discussion boards, including Health & Wellness, Just Chatting, and Religion/Spirituality (Gustafson et al., 2015).

Evaluation Methods and Results:

Depression status was assigned utilizing baseline PHQ-8 ≥10. Content analysis revealed that elders utilized the Elder Tree system to obtain and relay health information, share personal experiences, and to solicit and provide advice. Using LIWC (Linguistic Inquiry and Word Count, a computational text analysis tool) and independent sample t-tests, we found that across all forums, non-depressed elders were more likely to express words with negative emotions while depressed elders were more likely to express words associated with positive emotions. We also found that depressed elders’ messages on the Health and Wellness board were more likely to contain sad words while messages associated with anxiety and death were more likely to be written by non-depressed elders on Just Chatting. In addition, messages by non-socially isolated elders contained more words related to power and health while isolated elders‘ messages contained more sad words.

Conclusions:

Linguistic analysis revealed that elders who are depressed may be expressing more positive emotions on online support systems compared to those who are not depressed. Positive emotion expression may contribute to greater benefits received from these systems while negative emotion expression may be indicative of the opposite effect. However, the effects of emotional expression may depend on the context of the forum. Health-specific forums garnered more messages associated with sadness while others garnered more positive emotions among the different elderly groups.

Implications for research and/or practice:

This work adds to the body of literature seeking to understand how online support systems may be improving well-being among different elders by demonstrating differences in language use among elders poised to benefit from these systems, socially isolated and depressed elders. Future research should investigate whether differences predict benefits received and whether specific forums are better able to facilitate improved well-being.