Table 2. Operationalization of communication competence across studies.
| Communication competence (number of studies), research papers | Operationalization |
|---|---|
| Empathetic response (n=18) | |
| Trzebiński et al. (2023) [73] | Express empathy and autonomy support by showing understanding of the users’ concerns, acknowledging their knowledge, supporting their autonomy regarding vaccination, showing interest in their situation, and offering comfort |
| He et al (2022) [39] | Incorporate reflective listening to acknowledge the user’s health concerns, affirm their points and efforts, validate their personal choice, and express compassion for their suffering |
| El Hefny et al (2021) [74] | Use empathetic expressions extracted from the “EmpatheticDialogues” dataset to approach the users in a friendly manner and create a bond |
| Gotthardt et al (2022) [64] | Use empathetic techniques such as mirroring, empathetic listening, cheering up, or calming in response to the user’s emotions and answers to the psychoeducational quiz |
| Ho (2018) [67] | Provide responses validating the conversational partner’s feelings |
| Meng and Dai (2021) [71] | Provide emotional support that communicates empathy, emotional validation, and encouragement to the conversational partner |
| Liu and Sundar (2018) (Study 1 and Study 2) [38] |
Express sorry for the user in the beginning of each response (Sympathy) Recognize and acknowledge the user’s feelings and situations in the beginning of each response (Cognitive empathy). Express understanding of the user’s feelings in the response (Affective empathy) |
| Kraus et al (2021) [75] | Show empathetic reactions by expressing understanding of the user’s negative mood shared during the daily check-in |
| Rains et al (2020) [40], (2020) [76] Rains and High (2021) [72] |
Provide responses that explicitly acknowledged and elaborated on the conversational partner’s feelings and offer suggestions on how to reframe negative affect |
| You et al (2023) [77] | Provide emotional support to the user by using caring language and encouragement as well as offering potential treatment suggestions |
| Beattie (2023) [66] | Provided highly person-centered responses that focus on the user’s emotional and stressors, including reflecting on, acknowledging, and confirming the challenges related to the topic, and emphasizing the normality of the stressful feeling |
| Pecune et al (2020) [42] | Use acknowledgments to show understanding of what users just said |
| Ghandeharioun et al (2019a) [60], (2019b) [61] | Acknowledging the user’s emotional state after receiving their mood report |
| Lin et al (2023) [78] | Show understanding of the users’ input and generate empathetic responses based on the user’s emotions |
| Contingency (n=5) | |
| Meng et al (2023) [41] | Repeat the conversational partner’s self-disclosure and refer to the conversational partner’s specific situations mentioned in the previous conversation |
| De Boni et al (2008) [79] | Preserve the records of previous conversations and refer back to the issues (eg, barriers and solutions) discussed in the previous conversations |
| He et al (2022) [39] | Summarize previous conversation |
| Liu et al (2022) [69] | Embed user’s personal information asked in the previous conversation (eg, gender, age, and eating and living habit) in the response when providing diagnostic suggestions |
| You et al (2023) [77] | Provide a personalized summary of the symptoms mentioned by the user in the previous conversation to explain the diagnosis |
| Humor (n=3) | |
| El Hefny et al (2021) [74] | Induce humor with GIFs to create a friendly and cheerful atmosphere |
| De Boni et al (2008) [79] | Use self-contained jokes at the end of each session and self-deprecation |
| Lopatovska et al (2022a) [62], (2022b) [63] | Tells jokes to the user |
| Small talk (n=5) | |
| Kraus et al (2021) [75] | Deal with casual topics such as music preferences, personality, daily feeling, daily plans, and weather |
| Kobori et al (2016) [80] | Generate small talk utterances by choosing an appropriate response from the database based on the preceding user response, such as utterances about food preference, taste, or fun facts about specific food |
| Pecune et al (2020) [42] | Engage in small talk in the introductory phase by asking the user’s name, whether they are doing good, typical food for dinner, and reasons behind their food choices |
| De Boni et al (2008) [79] | Incorporate small talk elements during the greeting, which became more personal over time |
| Lee et al (2020) [70] | Build a small-talk session to discuss topics such as favorite holidays and zoo experiences before moving on to sensitive questions |
| Emotional expressiveness (n=2) | |
| El Hefny et al (2021) [74] | Add positive emojis to the response to convey affection |
| Ghandeharioun et al (2019a) [60], (2019b) [61] | Use emotionally expressive texts and emojis to convey appropriate emotions in response to the user’s mood and during the delivery of interventions |
| Self-disclosure (n=6) | |
| Meng and Dai (2021) [71] | Respond with its past experiences, thoughts, and feelings related to stressful situations |
| Lee et al (2020) [70] | High self-disclosure: Reveal its deep feelings, thoughts, and experiences in the past in the small-talk session. Low self-disclosure: Revealed less frequent and less intense feelings, thoughts, and past experiences in the small-talk session |
| Pecune et al (2020) [42] | Disclose information about itself to the user (eg, their eating preference and habits) during the introductory and information-gathering phases |
| Mai et al (2021) [68], (2022a) [81], (2022b) [82] | Talk about its own experience and feelings in similar situations before asking about the user’s experiences |
| Personalization (n=1) | |
| Albers et al (2022) [83] | Provide persuasive messages considering the person’s current state (eg, barriers or resources), future states, and the effectiveness of persuasive strategies for other similar people. It also updated the persuasion algorithm based on the user’s involvement in the recommended activity |
| Social etiquette (n=4) | |
| Li et al (2023) [49] | Use expressions of self-introduction, greetings, farewells, thanks, and tips and advice |
| You et al (2023) [77] | Use friendly addresses and greetings in the beginning of the conversation |
| Pecune et al (2020) [42] | Use reciprocal appreciation to give feedback to the user’s response |
| Kraus et al (2021) [75] | Express appreciation to the user for sharing personal topics during the daily check-in |
| Explanation (n=4) | |
| Woodcock et al (2021) [65] | Provide an explanation for the disease diagnosis using input influence (mentioned 2 symptoms most likely to indicate the disease), social proof (stated that a large number of people with similar symptoms have the disease), or counterfactual explanation (provided the symptom most likely to change a clinician’s opinion) |
| Pecune et al (2020) [42] | Use personal opinions as explanations of recipe recommendation |
| You et al (2023) [77] | Explain the rationale of each probing question and potential diagnosis using verified medical information |
| Buzcu et al (2023) [84] | Generate health-related (ie, nutritional values, such as protein, calorie, vitamin, and cholesterol information) and preference-related (eg, user’s preferred cuisine and ingredient) explanations to users |
| Open-ended question (n=1) | |
| He et al (2022) [39] | Ask open questions to encourage people to reflect on the reasons for quitting smoking |
| Partnership (n=1) | |
| He et al (2022) [39] | Emphasize shared understanding between the user and the chatbot and ask for consent before moving on |