Sage Journals: Discover world-class research

Abstract

Online support communities play an important role in providing access to quality social support. Additionally, those who experience health concerns that face social stigma, such as depression, are more likely to seek support online. To identify effective strategies that help individuals seeking support for depression receive high-quality social support online, we explore the role of two linguistic features in encouraging polite responses to their posts. This study employed a 2 (agency: human vs. depression) × 2 (causality: biological vs. nonbiological) factorial design. The politeness package for R was used to identify linguistic indicators of negative politeness and our results demonstrated that attributions of agency and causality did not have a significant direct effect on the prevalence of negative politeness in support responses. This suggests that support seekers’ understanding of the etiology of depression, and how they present it to support providers, has little impact on the quality of support provided in terms of negative politeness.

Keywords

online social support depression politeness linguistic agency locus of causality

Online support forums provide individuals with opportunities for advice and other forms of social support they may not have access to in person (Pan et al., 2019). Research has consistently emphasized the role of online support communities in providing individuals with access to quality social support (Li et al., 2022; Rains et al., 2015). Additionally, online social support allows support seekers to maintain a level of anonymity, which is often desirable for concerns associated with social stigma (Wright, 2016). Thus, it is important to understand if these individuals seeking online help are accessing the quality support that they need, and if there are potential strategies that can be utilized to improve the quality of social support received.

Online support forums vary in their community norms, anonymity, moderation, and user experiences. This heterogeneity generates platform-specific dynamics, influencing how support-seeking messages are received and interpreted. General forums like Reddit may foster varied responses due to their broad user base and looser moderation (De Choudhury & De, 2014), while specialized mental health communities may involve tighter community guidelines and individuals who are better attuned to mental health concerns. While prior literature has widely utilized Reddit as a single source for studying online mental health support (Garcia et al., 2024; Kamarudin et al., 2021; Sit et al., 2024), the current study is designed to focus on major features that are common across a range of existing online support forums. Therefore, the focus of our study models anonymous, text-based interactions to capture the broadly shared characteristics of platforms in which individuals seek support.

The current study aims to fill two major gaps in the literature. First, limited research has examined the impact of linguistic features in support seeking messages on the quality of received support (Khan et al., 2022; Pan et al., 2020). Second, even among the literature that does examine linguistic features, the connection between the linguistic features of the support seeking messages and the politeness of received support warrants further attention. Given that providing support inherently involves face threats (MacGeorge et al., 2002), there is a need to study politeness and its impact on online support forums, particularly when individuals are discussing stigmatized conditions. Linguistic politeness plays a crucial role in online environments, as it serves as an important strategy for managing interpersonal relationships in text-based interactions (Pratama, 2019). The focus on politeness is especially pertinent in online support settings, where verbal language becomes a primary tool for expressing empathy and managing facework. Although a prior study has examined the relationship between linguistic features of support seeking messages, stigma, and support quality (Khan et al., 2022), little is known about how linguistic features in support seeking messages affect interpersonal strategies used by support-providers, particularly in terms of politeness.

Social support is widely understood as a multidimensional construct (Chronister et al., 2006), encompassing a range of interpersonal processes and communicative dimensions that contribute to its effectiveness. Various situational and relational factors may influence the quality of the support individuals receive (e.g., acquaintance, sensitivity; Burleson, 2003). In particular, empirical evidence suggests that support quality is often evaluated through interpersonal features such as person-centeredness (MacGeorge et al., 2017). While such studies have empirically shown that various features contribute to perceived support quality, politeness is particularly relevant in online contexts where interpersonal sensitivity is often conveyed through text or in anonymous settings (Ramli et al., 2019; Taleghani-Nikazm, 2013). When cues such as tone, body language, or relational closeness are absent, politeness serves as a salient marker that can signal care and respectful engagement (Trávníková, 2012). Existing research has extensively studied the impact of politeness in online communities (Burke & Kraut, 2008; Federici, 2025) and have found that messages using politeness strategies were generally considered as higher in quality in terms of appropriateness and message effectiveness (Feng et al., 2017). Messages that are polite include features such as inclusion and liking directed toward the recipient, approbation (or affirmations of the recipient's competence), and tact (MacGeorge et al., 2017).

Politeness theory offers an explanation for why support providers may employ politeness strategies. Brown and Levinson's (1987) politeness theory draws on Goffman's (1967) discussion of face. Face is maintained through the actions of others in interpersonal interactions and refers to one's public self-image (Trees & Manusov, 1998). Face can be further broken down into positive face (desire for acceptance/approval) and negative face (desire for autonomy). Because both positive and negative face are dependent upon recognition from others, individuals are assumed to be interested in helping their conversation partners maintain face so that they will experience the same in return (Trees & Manusov, 1998). Face-threatening acts (such as insults that threaten positive face and orders that threaten negative face) can occur, and the theory posits that when an individual must engage in a face-threatening act they are also motivated to preserve face as much as possible. The more severe the face-threat, the more polite one would need to be to mitigate the impact of the speech act on positive and negative faces (Brown & Levinson, 1987). For this study, we will be examining how two linguistic features of support seeking messages, attributions of agency and causality, threaten the support seekers autonomy. Thus, we will focus on support providers’ methods for mitigating negative face threats by identifying the prevalence of negative politeness strategies in support messages. Negative politeness strategies are means of balancing directness with redress of a negative face threatening act (Brown & Levinson, 1987). In the context of support provision, the aim is to reduce impositions on the support seeker's autonomy when providing support. Brown and Levinson (1987) identify nine negative politeness strategies: be conventionally indirect, use hedges, be pessimistic, minimize the imposition, give deference, apologize, impersonalize, state the face threatening act as a general rule, and nominalize.

Existing work on politeness in online social support has begun to explore predictors of polite support provision online (Feng et al., 2016; Pan et al., 2020; Rains et al., 2019). This research is still in its infancy, and continued experimental work is needed to identify alternative characteristics of support seekers and message features that predict politeness. Linguistic agency and causality have been linked to variables relevant to autonomy (e.g., threat susceptibility, McGlone et al., 2013) and other linguistic indicators of support quality (e.g., Khan et al., 2022). Also, alternative conceptualizations of agency have been explored (e.g., responsibility, Rains et al., 2019). However, to our knowledge, linguistic agency and causality have yet to be studied in the context of negative face threat and polite support provision. Linguistic agency refers to how action and control over specific elements of a situation are attributed (McGlone & Pfiester, 2009). In the context of depression, agency can be attributed to depression or to the individual. Support seeking messages that assign agency to depression may be considered as more face-threatening than messages that assign agency to the individual. McGlone et al. (2013) found that assigning agency to an illness (H1N1) led participants to perceive themselves as more susceptible to the threat. Additionally, assigning agency to the human provides a level of control for the individual. In the context of depression, human agency suggests that there was something the individual could have done to prevent the onset of depression. However, assigning agency to the depression suggests an uncontrollable influence. When agency is assigned to depression, this is a potential negative face threat as support providers may perceive support seekers as having less control over their situation. As a result, we predict that participants will be motivated to mitigate that face threat utilizing negative politeness:
H1: Support seeking posts that assign linguistic agency to depression will elicit more negative politeness compared to posts that assign human agency.

Because depression is a mental illness, its symptoms can be linked to a variety of causes, both biological (e.g., genetic predispositions or chemical imbalances) and nonbiological (e.g., major or traumatic life changes; National Institute of Mental Health, 2024). Biological causes of mental illness can be seen as less under the control of the affected individual and tend to reduce blame toward them (Corrigan & Watson, 2004; Read & Law, 1999). This links causality to the notion of autonomy as well. If individuals are less to blame for their depression in the instance of biological causality, depression would then pose a greater threat to their autonomy than if the cause is nonbiological. For example, if someone attributes their depression to a nonbiological cause, such as work stress, their options for addressing the depression are more in their control (e.g., they can leave their job). Some research indicates that attributing depression to a biological cause was associated with lower personal stigma toward the support-seeker and more favorable emotional tone in support-providers’ responses (Khan et al., 2022). These findings suggest that causal attributions in support-seeking messages may affect the views of, and emotional responses to, individuals experiencing mental illness. Given that attributions shape individuals’ perceptions of control and responsibility, they are also likely to influence how support-providers manage interpersonal sensitivity in responding to support-seeking responses. Considering the threat of biological causality to negative face and higher stigma associated with nonbiological causality, we propose the following hypothesis:
H2: Support seeking posts that assign biological causality will elicit more negative politeness compared to posts that assign nonbiological causality.

In support-seeking contexts, it is typically the support-seeker's negative face that is most at risk due to the desire for autonomy and to avoid imposing on others (Brown & Levinson, 1987). Given that support-seeking involves disclosing personal challenges, these acts can signal vulnerability and dependence on the support seeker's part due to an inability to cope on their own and imposing on the support provider. Recognizing this face threat, support providers often respond with politeness strategies to not only reduce the imposition, but also protect the seeker's autonomy (Cupach & Metts, 1994; Spiers, 1998). For example, research has shown the importance of face-saving and face-threatening acts in patient–physician interactions (Ouma-Odero, 2025), cross-cultural contexts (Yaghubyan & Savoyan, 2025), and online support communities (Federici, 2025). Responses using politeness strategies can affirm the seeker's self-image by assisting their sense of capability or independence and manage the interpersonal dynamics between the seeker and the provider. Considering more severe face-threats elicit greater politeness (Brown & Levinson, 1987), we expect the face threatening nature of support seeking perceived by the support provider will increase in severity when the support seeking messages include additional threats to negative face through attributing agency to depression or biological causality. As a result, support responses will include more negative politeness strategies. In a similar way, assigning agency to the human and nonbiological causality will not increase the threat to autonomy beyond the face threat associated with support seeking more generally. Thus, this condition is expected to elicit the lowest levels of negative politeness compared to the other conditions. What is not clear, however, is if the combinations of human agency/biological causality and depression agency/nonbiological causality will elicit different levels of negative politeness.

We propose the following hypotheses and research question to test the extent to which agency and causality combine to increase the face threatening nature of support seeking, with the support seeking messages eliciting more negative politeness with each additional linguistic feature that threatens autonomy as outlined in the previous hypotheses.
H3: Online support-seeking messages assigning depression agency/biological causality will elicit more negative politeness compared to the other conditions.

H4: Online support-seeking messages assigning depression agency/nonbiological causality or human agency/biological causality will elicit more negative politeness compared to the human agency/nonbiological causality condition.

RQ: Will there be a difference in negative politeness in online support-seeking messages assigning depression agency/nonbiological causality and human agency/biological causality?

In this study, we also examine how stigma toward depression might shape the impact of agency and causality on politeness. Stigma broadly refers to the negative views held toward individuals or groups of people who exhibit characteristics seen as undesirable by society (Blaine, 2000). One type of stigma that is particularly relevant to this study is personal stigma, or an individual's own negative perceptions about mental illness (Vogel et al., 2006). Both biological causes and lack of control over the condition are associated with less blame and stigma toward those with mental illness (Corrigan & Watson, 2004; Goldstein & Rosselli, 2003). Conversely, when individuals were considered responsible for their illness or situation, they faced more blame and less social support (Schwarzer & Weiner, 1991).

Individuals’ stigmatizing attitudes about depression have the potential to influence how they perceive and attend to variations in linguistic agency and causality. Individuals with high stigma may not perceive the same level of threat to autonomy as those with low stigma, because stigmatizing attitudes about depression attribute high autonomy to the person with mental illness (e.g., depression is a sign of personal weakness; Cook & Wang, 2011). We expect that encountering a counterattitudinal message where agency is assigned to depression and the cause is biological (outside of the support seeker's control) will not be sufficient to change preexisting beliefs about autonomy and mental illness for those with high personal stigma. While there is limited work, to our knowledge, in the area of social support provision that evaluates how prior attitudes bias processing of support seeking messages, we draw on extensive existing research on the effect of prior attitudes on message processing more generally (Johnson et al., 2005).

When individuals encounter a message that is inconsistent with their preexisting attitudes, they attempt to reconcile this inconsistency through biased processing (Sekaquaptewa & Espinoza, 2004). Research on stereotypes suggests a number of potential biases. For example, the linguistic ingroup bias where people have a tendency to describe behaviors that are in line with their stereotypes as abstract generalizations (e.g., people with depression are dangerous) and stereotype-incongruent behaviors more specifically (e.g., she has depression, but she hasn’t hurt anyone), and the stereotypic explanatory bias where people are more likely to explain away stereotype inconsistencies (e.g., I wouldn’t hire someone with depression, but he didn’t seem sad during his interview; Vargas et al., 2004). Stereotype-incongruent information is ineffective in changing attitudes when it is explained away in this manner (Vargas et al., 2004). Additionally, past research has demonstrated that linguistic features of messages (e.g., generic nouns, person-first language) do not have a significant direct effect on individuals’ levels of stigma (Williams et al., 2022). Therefore, we predict that individuals who hold high personal stigma will be less likely to provide polite responses to mitigate the face threat, regardless of the severity of the face threat, because the linguistic manipulations of agency and causality will be insufficient to change their beliefs about the role of individual autonomy in managing mental illness.
H5: The relationship between agency, causality, and politeness will be moderated by stigma such that for those with lower levels of stigma, the impact of agency and causality on negative politeness will be stronger than those with higher levels of stigma.

Methods

The current study was an online survey administered via Qualtrics conducted in compliance with protocols and procedures established by the institutional review board at the authors’ institution (Approval Number: 1831442-1). The stimuli and procedures used in this study come from a larger lab experiment (Khan et al., 2022). However, new data were collected in two waves: January–March 2023 and August–December 2024. While the larger project encompasses the influence of agency and causality attribution on perceived stigma and support messages, the current study focuses on a theoretically distinct outcome: the use of negative politeness strategies. One of the future research directions that were suggested in the Khan et al. (2022) was to examine other linguistic features (p. 298). The authors discussed the potential relationship between stigma and the use of linguistic features, specifically suggesting politeness as a potential mechanism. While Khan et al.'s (2022) study explored various linguistic features, such as use of cognitive processing words, positive or negative emotions words, it did not incorporate any analysis of politeness strategies.

Participants

Participants (N = 584) were recruited from a large west coast university, ranging in age from 18 to 42 (M = 19.93, SD = 2.31). A majority of participants identified as women (n = 451, 77.4%), 19.7% (n = 115) identified as men, 2.2% (n = 13) as nonbinary, and 0.7% (n = 4) as other. Approximately 47% (n = 274) identified as AAPI, 23% (n = 136) Hispanic, 19% (n = 112) White, 3% (n = 17) Middle Eastern, 2% (n = 12) Black, 0.2% (n = 1) Native, and 5% (n = 29) other.

Study Design

This study employed a 2 (agency: human vs. depression) × 3 (causality: biological vs. work vs. relationship) × 2 (symptom severity: moderate vs. severe) factorial design. To broaden the generalizability of the results, we included two different nonbiological causes (work and relationship) and two levels of symptom severity (moderate and severe) in the design of our stimuli. Thus, while completing the survey, participants were randomly assigned to one of 12 conditions. Neither nonbiological cause nor severity was part of the theoretical focus of the study, and a preliminary analysis indicated that there was no effect of nonbiological cause or severity on negative politeness. As a result, we collapsed these conditions and nonbiological cause and severity were not included in further analyses,¹ resulting in a 2 (agency: human vs. depression) × 2 (causality: biological vs. nonbiological) factorial design for analysis.

Procedure

First, participants completed a preexposure survey, including measures of issue involvement, experience with mental illness, and stigma. Issue involvement and experience with mental illness were added for the current study to serve as potential control variables and were not included initially in Khan et al. (2022). Issue involvement and experience with mental illness were not significantly correlated with negative politeness and were not included as control variables as a result. Participants were randomly assigned to one of the twelve conditions. After reading the support seeking posts, participants were instructed to: “imagine you decide to respond to this message on the forum website. Please write down the message you would post in response to the writer.” Lastly, participants completed a postexposure survey. The full experiment from which our stimuli and procedures were derived included a manipulation check for the severity conditions, a stimuli realism measure,² personality, individualism, locus of control and communication style measures. None of these measures were included in our analyses for this study. Lastly, participants completed a demographics questionnaire and were provided with national and university-level mental health resources.

Stimuli

Participants were randomly assigned to read one support seeking post. Each post begins with a statement indicating causality: “Because of my hormone imbalance (biological) or work (nonbiological).” For conditions where agency is assigned to depression, the post has indicated the major role that the mental illness plays in their life. For example, “I don’t know how to fight against this depression” as compared to human agency, “I don’t know how to stop being depressed” (see Figure 1 for full stimuli).

Figure 1.
Experimental stimuli.

Measures

Negative Politeness

The politeness package in R was employed to identify linguistic features of negative politeness (Yeomans et al., 2019). Automated tools to identify linguistic features of politeness have been utilized in extant work to assess levels of politeness in contexts such as online discourse (Argyle et al., 2023; Li et al., 2020) and social support (Khan et al., 2022; Pan et al., 2020). We analyzed participant responses for the frequency counts of five linguistic features associated with negative politeness strategies (see Table 1 for examples of each feature for our dataset). The frequency of each indicator was summed to create a total negative measure (M = 1.66, SD = 1.98). The summing of present features is a common method for coding negative politeness strategies (Li & Feng, 2015).

Table 1.
Summary of Five Linguistic Features Used to Identify Negative Politeness.

Linguistic Feature Description Example From Dataset

Please Expressing please “Please reach out to us so that we can help you!”

Apology “sorry,” “oops,” “excuse me,” etc. “I am sorry to hear about your experience.”

Indirect Indirect requests “Would you like to speak to a therapist?”

Hedges Indicators of uncertainty “It might be helpful to reflect on what you need and want from your relationship”

First person plural First-person plural pronouns “Sometimes we stay in relationships due to comfort and a fear of leaving”

Please. While the use of the word please is not associated with a specific negative politeness strategy, using please aids in the function of negative politeness strategies in mitigating impositions. For example, when being conventionally indirect in order to soften requests (Alkhonini et al., 2024; Brown & Levinson, 1987, p. 135; Elmianvari & Kheirabadi, 2013).

Apology. Apologies reduce the negative face threat of the support provider's message by conveying an awareness of the threat to autonomy or reluctance to engage in the face threatening act (Brown & Levinson, 1987; Feng et al., 2016; Schallert et al., 2009).

Indirect. Being conventionally indirect involves questioning the support seeker's willingness or ability to engage in a support provider's recommendation. For example, a support provider may ask “Would you consider speaking to a therapist?” This is a less direct way of conveying advice to seek help from a therapist that acknowledges the support seeker's ability to make the choice for themselves (Feng et al., 2016; Shallert et al., 2009).

Hedges. Hedges are words that indicate the support provider is not assuming the support seeker will agree with them or want to comply with their recommendation (Feng et al., 2016; Shallert et al., 2009).

First personal plural. The use of first-person plural pronouns is associated with impersonalization, so that threats to autonomy are not explicitly directed toward the supper seeker “you,” but toward a more general collective that may include the support seeker (Brown & Levinson, 1987).

Stigma

Stigma was measured with a nine-item personal stigma scale utilized by Cook and Wang (2011) to assess the extent to which participants agree with commonly held stigmatizing beliefs in society about depression (e.g., “People with depression are unpredictable,” “Depression is a sign of personal weakness”). Participants were asked to respond to the items based on the Likert scale ranging from strongly disagree (1) to strongly agree (7). Items were summed in accordance with the original scale and the items demonstrated satisfactory internal consistency (M = 2.39, SD = 0.85; Cronbach's α = .813).

Perceived Agency

Participants were asked to rate the likelihood of six statements being true about the post they read from 1 (very likely) to 9 (very unlikely): The message poster was invaded by depression, depression invaded this person, this message poster's life was infiltrated by depression, depression infiltrated this person's life, this message poster's life was controlled by depression, and the problem was the poster's own fault (reverse coded). Responses to item 6 differed significantly between the biological versus nonbiological causality conditions, and not the agency conditions, and removing item 6 improved the internal reliability of the scale from α = .888 to α = .921. Item 6 was removed, and the remaining items were averaged (M = 3.29, SD = 1.77). There was no significant difference in perceptions of agency between conditions (F_{(3, 580)} = .202, p = .895).

Results

Linear regression was used to assess the effect of linguistic agency and causality on negative politeness and the moderating role of stigma. The first model included stigma (mean centered), agency (0 = human; 1 = depression), and causality (0 = nonbiological; 1 = biological; Adjusted R² = .001, F_{(3, 561)} = 1.23, p = .296), the second model added the interaction terms for the moderating role of stigma (Adjusted R² = −.002, F_{(5, 559)} = .805, p = .547), and the third model added sex as a control variable (Adjusted R² = .005, F_{(5, 558)} = 1.43, p = .202). We predicted that support seeking posts assigning agency to depression (H1) and biological causality (H2) would result in greater negative politeness in support responses. Agency and causality were not significantly associated with negative politeness, thus H1 and H2 were not supported. H5 predicted that the relationship between agency, causality, and politeness would be moderated by stigma. Stigma was not significantly associated with negative politeness, and the interactions between stigma and agency, and stigma and causality were not significantly associated with negative politeness. Thus, H5 was not supported. Sex was the only variable statistically associated with negative politeness, with females demonstrating more negative politeness than males (β = .092, p = .034). See Table 2 for the full results of the regression.

Table 2.
Linear Regression Analysis of the Role of Agency, Causality, and Stigma in Predicting Negative Politeness of Support Responses.

Step 1 Step 2 Step 3

β SE p β SE p β SE p

Stigma −.05 .10 .256 −.03 .16 .624 −.01 .16 .856

Causality −.06 .18 .137 −.06 .18 .137 −.06 .18 .172

Agency .01 .17 .734 .02 .17 .730 .01 .17 .792

Causality * Stigma .01 .21 .781 .02 .21 .764

Agency * Stigma −.03 .20 .604 −.03 .20 .578

Sex¹ .09 .21 .034⁺

F 1.24 .81 1.43

Adjusted R² .001 −.002 .005

1
0 = Male; 1 = Female.

+
Significant at p < .05 level.

A univariate general linear model was used to assess differences in negative politeness between each condition, with sex included as a control variable. H3 predicted that online support-seeking messages assigning depression agency/biological causality would elicit more linguistic indicators of politeness compared to the other conditions. H4 predicted that online support-seeking messages assigning depression agency/nonbiological causality or human agency/biological causality will elicit more linguistic indicators of politeness compared to the human agency/nonbiological causality condition. Additionally, our research question asked if there would be a difference in politeness between depression agency/nonbiological causality and human agency/biological causality conditions. There was no significant difference between conditions in terms of negative politeness (Adjusted R² = .010; F_{(4, 561)} = 1.36, p = .253), thus, H3 and H4 were not supported. Sex was significantly different between conditions (F_{(4, 561)} = 5.05, p = .025) and a follow-up t-test revealed that females (M = 1.74, SD = 2.11) had higher negative politeness than males (M = 1.24, SD = 1.28; t(291.67) = −3.21, p = .001). See Table 3 for a comparison of average negative politeness levels based on condition.

Table 3.
Average Levels of Negative Politeness Based on Condition.

Condition N M SD

Depression Agency, Biological Causality 101 1.34 1.37

Depression Agency, Nonbiological Causality 195 1.85 1.70

Human Agency, Biological Causality 87 1.66 3.56

Human Agency, Nonbiological Causality 200 1.65 1.46

Note: Values are raw means and standard deviations.

Discussion

Our results demonstrated that attributions of agency and causality in support-seeking about depression did not have a significant direct effect on the negative politeness of support responses. Additionally, stigma did not have a significant impact on this relationship. In the context of politeness theory, there are a few important considerations. Depression agency and biological causality may not have been interpreted by support providers as substantive threats to negative face, resulting in similar levels of negative politeness across conditions. A few explanations for this include, one, that support provision can be a threat to the recipient's positive and negative face (MacGeorge et al., 2002), so regardless of linguistic features of support-seeking, participants may have been engaging in politeness strategies to account for the face-threatening nature of support provision. Additionally, these linguistic variations may be too subtle to overcome preexisting attitudes regarding the nature of mental illness and the effectiveness of its treatments. In the case of our sample, levels of stigma were low (M = 2.39, SD = 0.85), so participants tended to disagree with statements over attributing agency to those experiencing depression (e.g., “People with a problem like depression could snap out of it if they wanted”) and claims related to causality (“Depression is not a real medical illness”). It is still possible that linguistic variations in agency and causality are theoretically relevant to other features of quality support. While we found that agency and causality were not significant predictors of negative politeness, extant research has demonstrated the importance of linguistic features in determining quality support provision online (e.g., Khan et al., 2022). Thus, future work is needed before ruling out agency and causality attributions as strategies for receiving high-quality online support.

It is important to note that perceptions of agency did not differ significantly between conditions. In other words, participants did not attribute agency in accordance with the condition they were assigned. Additionally, it is possible that participants did not notice the causality manipulation, thus future work would benefit from asking participants to recall the cause identified in the support seeking message. While conscious awareness of the agency and causality manipulations are not necessarily needed to influence participants’ support messages (Hauser et al., 2018), our agency and causality manipulations may not have been strong enough to impact our participants’ support provision.

Our null findings may also suggest a potential misalignment between our five linguistic indicators and the negative politeness strategies identified by Brown and Levinson (1987). While these five linguistic indicators touch on some of the key negative politeness strategies identified in support provision, including apologies, conventional indirectness, hedges, and impersonalization, there are still some features that we were unable to identify, namely minimizing the imposition and showing deference (Feng et al., 2016). Thus, it is possible negative politeness strategies were more frequent or varied more between conditions than we were able to measure. Additionally, identifying negative politeness based on the frequency of individual lexical terms does not fully capture the features of a message that may be used to convey subtle facework. Therefore, variations in how participants manage face threats may have occurred but could have remained undetected with the current measurement approach. While there is strong empirical evidence for using automatic text analyses tools for politeness, future research may benefit from combining other methods, such as qualitative coding or natural language processing techniques, that are better suited to detect specific politeness strategies. Additionally, negative politeness was our sole outcome variable and primary indicator of support quality. While this was appropriate in the context of politeness theory, attributions of agency and causality may influence other aspects of social support.

Another limitation is the demographic composition of the sample. Our sample skewed toward women and AAPI students, which limits the extent to which our findings apply to other populations. Prior research has found that gender, racial ethnicities, or cultural background affect attitudes toward mental health (Jacob et al., 2025; Ran et al., 2021) and social support behaviors (Wootton et al., 2025; Zaman et al., 2025). Additionally, we found that females demonstrated more negative politeness linguistic indicators than males. Thus, future studies should aim to include a more balanced and representative sample to explore the dynamics across a broader range of groups.

Lastly, we acknowledge that the current study did not account for the differences between various types of online support forums, which can vary in terms of anonymity, moderation, and community norms. While this study design focused on more generalized, anonymous online interactions, there are contextual factors that may influence how support messages are received or interpreted (e.g., Feng et al., 2016). Thus, future studies should examine how platform-specific features or platform types with distinct user norms may interact with politeness of participants’ support responses. Relatedly, there may be relevant differences in the support provided by people who view and are willing to respond to posts on mental health related online forums and the support provided by those who were required to respond for an experiment. Future research would benefit from measuring these support forum engagement behaviors to see if it does lead to differences in support quality.

In terms of practical implications for support seeking, our findings suggest support seekers’ knowledge and understanding of the etiology of their condition, and how they choose to present it to potential support providers, does not seem to impact the quality of support provided in terms of negative politeness. This is beneficial considering the changes that occur in how the causes of depression are framed in medical contexts and among the general public. For example, the prevalence and persistence of the chemical imbalance explanation (Deacon & Baird, 2009), compared to more current emphasis on the interplay between biological and nonbiological causes (Remes et al., 2021). To provide a more complete recommendation for how to discuss the etiology of depression in support seeking, the role of agency and causality should be explored further by examining alternative indicators of support quality, as well as through different populations of support providers. Our use of college students in this study may have influenced our participants’ likelihood of providing polite support regardless of agency or causality. In addition to being a part of the age range most likely to be affected by depression (age 18–25; National Institute of Mental Health, 2023), college students often cite work and school relationships as common stressors (Karyotaki et al., 2020). To receive quality support from a less understanding population, it may be necessary to make specific attributions of agency or causality. Our study focused on responses to an unknown support seeker in the context of online support forums. Thus, the practical implications of our findings are focused on this specific context. The relationship between the support seeker and support provider, and the channel through which they are communicating, have the potential to influence the role of agency, causality, and politeness in supportive communication as well.

Linguistic Feature	Description	Example From Dataset
Please	Expressing please	“Please reach out to us so that we can help you!”
Apology	“sorry,” “oops,” “excuse me,” etc.	“I am sorry to hear about your experience.”
Indirect	Indirect requests	“Would you like to speak to a therapist?”
Hedges	Indicators of uncertainty	“It might be helpful to reflect on what you need and want from your relationship”
First person plural	First-person plural pronouns	“Sometimes we stay in relationships due to comfort and a fear of leaving”

	Step 1	Step 2	Step 3
Stigma	−.05	.10	.256	−.03	.16	.624	−.01	.16	.856
Causality	−.06	.18	.137	−.06	.18	.137	−.06	.18	.172
Agency	.01	.17	.734	.02	.17	.730	.01	.17	.792
Causality * Stigma				.01	.21	.781	.02	.21	.764
Agency * Stigma				−.03	.20	.604	−.03	.20	.578
Sex¹							.09	.21	.034⁺
F		1.24			.81			1.43
Adjusted R²		.001			−.002			.005

Condition	N	M	SD
Depression Agency, Biological Causality	101	1.34	1.37
Depression Agency, Nonbiological Causality	195	1.85	1.70
Human Agency, Biological Causality	87	1.66	3.56
Human Agency, Nonbiological Causality	200	1.65	1.46

Footnotes

Acknowledgments

The authors would like to thank Dr. Bo Feng for her valuable feedback throughout the research process.

ORCID iDs

Rachel E. McKenzie

Lisa Jihyun Hwang

Ethical Considerations

The institutional review board at the author's institution (approval no. 1831442-1) approved our online survey-experiment on 01/05/2023. The project was exempt, thus, informed consent documentation was not required by the IRB.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Notes

References

Alkhonini

O. A.

Al-Shboul

O. K.

Dakamsih

N. J.

(2024). How Americans and Saudis make requests: A politeness theory analysis. Journal of Language Teaching and Research, 15(5), 1683–1689. https://doi.org/10.17507/jltr.1505.29

Argyle

L. P.

Bail

C. A.

Busby

E. C.

Gubler

J. R.

Howe

Rytting

Sorensen

Wingate

(2023). Leveraging AI for democratic discourse: Chat interventions can improve online political conversations at scale. Proceedings of the National Academy of Sciences, 120(41), e2311627120. https://doi.org/10.1073/pnas.2311627120

Blaine

B. E.

(2000). The psychology of diversity: Perceiving and experiencing social difference. McGraw-Hill Humanities, Social Sciences & World Languages.

Brown

Levinson

S. C.

(1987). Politeness: Some universals in language usage. Cambridge University Press.

Burke

Kraut

(2008). Mind your Ps and Qs: The impact of politeness and rudeness in online communities. Proceedings of the 2008 ACM Conference on Computer Supported Cooperative Work, 281–284. https://doi.org/10.1145/1460563.1460609

Burleson

B. R.

(2003). Emotional support skills. In Handbook of communication and social interaction skills (pp. 569–612). Routledge.

Chronister

J. A.

Johnson

E. K.

Berven

N. L.

(2006). Measuring social support in rehabilitation. Disability and Rehabilitation, 28(2), 75–84. https://doi.org/10.1080/09638280500163695

Cook

T. M.

Wang

(2011). Causation beliefs and stigma against depression: Results from a population-based study. Journal of Affective Disorders, 133(1–2), 86–92. https://doi.org/10.1016/j.jad.2011.03.030

Corrigan

P. W.

Watson

A. C.

(2004). At issue: Stop the stigma: Call mental illness a brain disease. Schizophrenia Bulletin, 30(3), 477–479. https://doi.org/10.1093/oxfordjournals.schbul.a007095

10.

Cupach

W. R.

Metts

(1994). Facework (Vol. 7). Sage.

11.

Deacon

B. J.

Baird

G. L.

(2009). The chemical imbalance explanation of depression: Reducing blame at what cost? Journal of Social and Clinical Psychology, 28(4), 415–435. https://doi.org/10.1521/jscp.2009.28.4.415

12.

De Choudhury

(2014). Mental health discourse on reddit: Self-disclosure, social support, and anonymity. Proceedings of the International AAAI Conference on Web and Social Media, 8(1), 71–80. https://doi.org/10.1609/icwsm.v8i1.14526

13.

Elmianvari

Kheirabadi

(2013). The study of EFL students’ requests based on politeness theory. Journal of Language Teaching and Research, 4(2), 375–385. https://doi.org/10.4304/jltr.4.2.375-385

14.

Federici

(2025). The pragmatics of online healthcare communication: Politeness strategies in an anxiety and depression support community. Journal of Pragmatics, 246, 74–89. https://doi.org/10.1016/j.pragma.2025.06.010

15.

Feng

Jang

J. Y.

Kim

Wang

(2017). Explaining the effectiveness of the integrated model of advice-giving in supportive interactions: The mediating roles of politeness and normativeness. Communication Quarterly, 65(4), 402–418. https://doi.org/10.1080/01463373.2017.1279676

16.

Feng

(2016). Is a profile worth a thousand words? How online support-seeker’s profile features may influence the quality of received support messages. Communication Research, 43(2), 253–276. https://doi.org/10.1177/0093650213510942

17.

Garcia

Amador Ayala

Diaz Roldan

Bavarian

(2024). Exploring reddit conversations about mental health difficulties among college students during the COVID-19 pandemic. Journal of American College HealtH, 72(8), 2419–2425. https://doi.org/10.1080/07448481.2022.2115297

18.

Goffman

(1967). Interaction ritual: Essays on face-to-face behavior. Pantheon Books.

19.

Goldstein

Rosselli

(2003). Etiological paradigms of depression: The relationship between perceived causes, empowerment, treatment preferences, and stigma. Journal of Mental Health, 12(6), 551–563. https://doi.org/10.1080/09638230310001627919

20.

Hauser

D. J.

Ellsworth

P. C.

Gonzalez

(2018). Are manipulation checks necessary? Frontiers in Psychology, 9, https://doi.org/10.3389/fpsyg.2018.00998

21.

Jacob

Lavidor

Igra

Hasson-Ohayon

(2025). Cultural aspects of stigma toward people with serious mental illness: A meta-analysis of the association between individualism-collectivism and social distance. Stigma and Health. https://doi.org/10.1037/sah0000641

22.

Johnson

B. T.

Maio

G. R.

Smith-McLallen

(2005). Communication and attitude change: Causes, processes, and effects. In The handbook of attitudes (pp. 617–669). Erlbaum.

23.

Kamarudin

N. S.

Beigi

Liu

(2021). A study on mental health discussion through reddit. 2021 International conference on software engineering & computer systems and 4th international conference on computational science and information management (ICSECS-ICOCSIM) (pp. 637–643).

24.

Karyotaki

Cuijpers

Albor

Alonso

Auerbach

R. P.

Bantjes

Kessler

R. C.

Ebert

D. D.

Hasking

Kiekens

Lee

McLafferty

Mak

Mortier

Sampson

N. A.

Stein

D. J.

Vilagut

Kessler

R. C.

(2020). Sources of stress and their associations with mental disorders among college students: Results of the world health organization world mental health surveys international college student initiative. Frontiers in Psychology, 11, 1759. https://doi.org/10.3389/fpsyg.2020.01759

25.

Khan

Gonzalez

McKenzie

Feng

(2022). The influence of linguistic agency and causality attribution in support-seeking about depression on perceived stigma and support messages. Journal of Health Communication, 27(5), 292–301. https://doi.org/10.1080/10810730.2022.2104412

26.

Hickman

Tay

Ungar

Guntuku

S. C.

(2020). Studying politeness across cultures using English twitter and Mandarin Weibo. Proceedings of the ACM on Human-Computer Interaction, 4(CSCW2), 1–15. https://doi.org/10.1145/3415190

27.

Feng

(2015). What to say to an online support-seeker? The influence of others’ responses and support-Seekers’ replies. Human Communication Research, 41(3), 303–326. https://doi.org/10.1111/hcre.12055

28.

Liao

Kim

Feng

Pan

(2022). Understanding the association between online social support obtainment and coping during a public crisis. Journal of Health Communication, 27(5), 343–352. https://doi.org/10.1080/10810730.2022.2107738

29.

MacGeorge

E. L.

Guntzviller

L. M.

Brisini

K. S.

Bailey

L. C.

Salmon

S. K.

Severen

Branch

S. E.

Lillie

H. M.

Lindley

C. K.

Pastor

R. G.

Cummings

Ryan D.

(2017). The influence of emotional support quality on advice evaluation and outcomes. Communication Quarterly, 65(1), 80–96. https://doi.org/10.1080/01463373.2016.1176945

30.

MacGeorge

E. L.

Lichtman

R. M.

Pressey

L. C.

(2002). The evaluation of advice in supportive interactions: Facework and contextual factors. Human Communication Research, 28(3), 451–463. https://doi.org/10.1111/j.1468-2958.2002.tb00815.x

31.

McGlone

M. S.

Bell

R. A.

Zaitchik

S. T.

McGlynn Iii

(2013). Don’t let the flu catch you: Agency assignment in printed educational materials about the H1N1 influenza virus. Journal of Health Communication, 18(6), 740–756. https://doi.org/10.1080/10810730.2012.727950

32.

McGlone

M. S.

Pfiester

R. A.

(2009). Does time fly when you’re having fun, or do you?: Affect, agency, and embodiment in temporal communication. Journal of Language and Social Psychology, 28(1), 3–31. https://doi.org/10.1177/0261927X08325744

33.

National Institute of Mental Health. (2023). Major depression. U.S. Department of Health and Human Services, National Institutes of Health. https://www.nimh.nih.gov/health/statistics/major-depression

34.

National Institute of Mental Health. (2024). Depression. U.S. Department of Health and Human Services, National Institutes of Health. https://www.nimh.nih.gov/health/topics/depression

35.

Ouma-Odero

(2025). That is somebody’s husband”: Face-saving strategies in doctor–patient interaction in a public health facility in Kenya. Communication & Medicine, e20240021. https://doi.org/10.3138/cam-2024-0021

36.

Pan

Feng

(2019). Online support groups dedicated to specific health problems. Social Support and Health in the Digital Age, 109, 109–128. https://doi.org/10.5040/9781978729698.ch-006

37.

Pan

Feng

Wingate

V. S.

(2020). What to say when seeking support online: A comparison among different levels of self-disclosure. Frontiers in Psychology, 11. https://doi.org/10.3389/fpsyg.2020.00978

38.

Pratama

(2019). Linguistic politeness in online communication. pemimpinmuda. id.

39.

Rains

S. A.

Peterson

E. B.

Wright

K. B.

(2015). Communicating social support in computer-mediated contexts: A meta-analytic review of content analyses examining support messages shared online among individuals coping with illness. Communication Monographs, 82(4), 403–430. https://doi.org/10.1080/03637751.2015.1019530

40.

Rains

S. A.

Tsetsi

Akers

Pavlich

C. A.

Appelbaum

(2019). Factors influencing the quality of social support messages produced online: The role of responsibility for distress and others’ support attempts. Communication Research, 46(6), 866–886. https://doi.org/10.1177/0093650218796371

41.

Ramli

N. S.

Ahmad

U. K.

Mustafa

H. R.

Puteh

Rahman

Azizan

A. R.

Sharif

N. M.

Busri

R. D.

(2019). Politeness strategies used in asynchronous online forum. International Conference on E-Learning Proceeding, 29–40.

42.

Ran

M.-S.

Hall

B. J.

T. T.

Prawira

Breth-Petersen

X.-H.

Zhang

T.-M.

(2021). Stigma of mental illness and cultural factors in Pacific Rim region: A systematic review. BMC Psychiatry, 21(1), 8. https://doi.org/10.1186/s12888-020-02991-5

43.

Read

Law

(1999). The relationship of causal beliefs and contact with users of mental health services to attitudes to the “mentally ill.”. International Journal of Social Psychiatry, 45(3), 216–229. https://doi.org/10.1177/002076409904500309

44.

Remes

Mendes

J. F.

Templeton

(2021). Biological, psychological, and social determinants of depression: A review of recent literature. Brain Sciences, 11(12), 1633. https://doi.org/10.3390/brainsci11121633

45.

Schallert

D. L.

Chiang

Y. H. V.

Park

Jordan

M. E.

Lee

Cheng

A. C. J.

Chu

H. N. R.

Lee

Kim

Song

(2009). Being polite while fulfilling different discourse functions in online classroom discussions. Computers and Education, 53(3), 713–725. https://doi.org/10.1016/j.compedu.2009.04.009

46.

Schallert

D. L.

Chiang

Y. V.

Park

Jordan

M. E.

Lee

Janne Cheng

A.-C.

Rebecca Chu

H.-N.

Lee

Kim

Song

(2009). Being polite while fulfilling different discourse functions in online classroom discussions. Computers & Education, 53(3), 713–725. https://doi.org/10.1016/j.compedu.2009.04.009

47.

Schwarzer

Weiner

(1991). Stigma controllability and coping as predictors of emotions and social support. Journal of Social and Personal Relationships, 8(1), 133–140. https://doi.org/10.1177/0265407591081007

48.

Sekaquaptewa

Espinoza

(2004). Biased processing of stereotype-incongruency is greater for low than high status groups. Journal of Experimental Social Psychology, 40(1), 128–135. https://doi.org/10.1016/S0022-1031(03)00093-3

49.

Sit

Elliott

S. A.

Wright

K. S.

Scott

S. D.

Hartling

(2024). Youth mental health help-seeking information needs and experiences: A thematic analysis of reddit posts. Youth & Society, 56(1), 24–41. https://doi.org/10.1177/0044118X221129642

50.

Spiers

J. A.

(1998). The use of face work and politeness theory. Qualitative Health Research, 8(1), 25–47. https://doi.org/10.1177/104973239800800103

51.

Taleghani-Nikazm

(2013). Politeness in computer-mediated communication. The Encyclopedia of Applied Linguistics, 4858–4863. https://doi.org/10.1002/9781405198431.wbeal0918

52.

Trávníková

(2012). Building rapport in an online community via positive politeness strategies. Discourse and Interaction, 5(2), 67–83. https://doi.org/10.5817/DI2012-2-67

53.

Trees

A. R.

Manusov

(1998). Managing face concerns in criticism: Integrating nonverbal behaviors as a dimension of politeness in female friendship dyads. Human Communication Research, 24(4), 564–583. https://doi.org/10.1111/j.1468-2958.1998.tb00431.x

54.

Vargas

P. T.

Sekaquaptewa

von Hippel

(2004). It's not just what you think, it's also how you think: Prejudice as biased information processing. In Diversity in advertising (pp. 93–119). Psychology Press.

55.

Vogel

Wade

Haake

(2006). Measuring the self-stigma associated with seeking psychological help. Journal of Counseling Psychology, 53, 325–337. https://doi.org/10.1037/0022-0167.53.3.325

56.

Williams

Foulser

A. A.

Tillman

K. A.

(2022). Effects of language on social essentialist beliefs and stigma about mental illness. Proceedings of the Annual Meeting of the Cognitive Science Society, 44(44), 334–341.

57.

Wootton

A. R.

Sterzing

P. R.

Mericle

A. A.

Drabble

L. A.

Cahoon

L. R.

Hughes

T. L.

(2025). Resilience through social support: An intersectional analysis of sexual minority women’s social resources for wellbeing. Journal of Homosexuality, 72(5), 951–971. https://doi.org/10.1080/00918369.2024.2360610

58.

Wright

(2016). Communication in health-related online social support groups/communities: A review of research on predictors of participation, applications of social support theory, and health outcomes. Review of Communication Research, 4, 65–87. https://doi.org/10.12840/issn.2255-4165.2016.04.01.010

59.

Yaghubyan

Savoyan

(2025). Negative politeness strategies in English and Armenian cultures. Foreign Languages in Higher Education, 29(1 (38)), 61–72. https://doi.org/10.46991/FLHE.2025.29.1.061

60.

Yeomans

Kantor

Tingley

(2019). The politeness package: Detecting politeness in natural language. The R Journal, 10(2), 489. https://doi.org/10.32614/rj-2018-079

61.

Zaman

Bibi

Asif

Arif

M. B.

Dino

Khan

(2025). Understanding gender-based differences in coping mechanisms and social support among adolescents in chitral. Journal of Social Signs Review, 3(4), 213–226.

Linguistic Agency and Causality in Support Seeking: Effects on Politeness in Online Support Messages

Abstract

Keywords

Methods

Participants

Study Design

Procedure

Stimuli

Measures

Negative Politeness

Stigma

Perceived Agency

Results

Discussion

Footnotes

Acknowledgments

ORCID iDs

Ethical Considerations

Funding

Declaration of Conflicting Interests

Data Availability Statement

Notes

References