Intersubjectivity within asynchronous learning discussions is the progression from independent student posts to interdependent ones, in which students build on one another’s ideas. Conversational repair is the practice of clarifying misunderstandings and answering questions. In this study of 516 discussion posts across 31 participants, the total frequency of conversational repair predicted the level of intersubjectivity. Intersubjectivity was measured using the Intersubjectivity Repair Dyad (IRD) of Whye (2019), and conversational repair was measured using the repair types identified by Schegloff et al. (1977). After a brief review of the study, implications for practitioners and researchers are considered.
Intersubjectivity is the representation of knowledge construction achieved through a synergistic progression from individual contributions to sequences of interdependent contributions. Repair is a conversational strategy comprising interaction sequences that contribute to the development of intersubjectivity, thereby establishing mutual understanding. Repair is used when there is an interruption or a need for further clarification by interactants in the dialogue. The research question in this study asked, “What relationship, if any, exists between intersubjectivity and the use of conversational repair in asynchronous online learning discussions?” The complete paper was originally published in Hall and Whye (2023).
The study examined intersubjectivity and conversational repair within the theoretical framework of social constructivism. Human cognition is a collective, distributed, cultural process involving the interpretation of linguistically encoded utterances (Morin et al., 2020). Mead and Schubert (1934) argued that there is no thinking outside of social processes. The coding tool used in this study—the Intersubjectivity Repair Dyad (IRD) (Whye, 2019)—is based on the work of Lev Vygotsky (1978) and the idea that social interaction facilitates the development of higher cognition. Cognitive development results from social interaction and negotiation strategies through communication and collaboration in asynchronous computer-mediated communication at fixed time points.
The study used a non-experimental, correlational design. The purpose of correlational research is to determine what relationship, if any, exists between two or more measures (Creswell, 2008). A correlational design was appropriate because the study aimed to determine the extent of the relationship between intersubjectivity and conversational repair in online learning discussions. The researchers addressed the research question by comparing the levels of intersubjectivity as measured by the Intersubjectivity Repair Dyad (IRD) (Whye, 2019), illustrated in Table 1, with the total frequency of the four types of conversational repair sequences.
Table 1
Intersubjectivity Repair Dyad
Level | Label | Indicators |
|---|---|---|
1 | Sharing of meaningful information through interaction, leading to the identification of a problem or state of agreement | Each party contributes information to the conversation using symbols as a means of communication |
2 | Negotiation and co-creation of knowledge using the external resources available | Discussion of dialogue that emerges from reviewing the information provided by each interactant |
3 | Review of the subject matter and the development of new psychological systems | Review of cultural understandings |
4 | Co-creation of knowledge is internalized, and the application of new knowledge creates developmental and cultural change | Applying the knowledge developed, contributing to ongoing cultural changes in social interaction, and continuing the collective intersubjective process within dialogue |
The four types of repair sequences, illustrated in Figure 1, are self-initiated self-repair, other-initiated self-repair, self-initiated other-repair, and other-initiated other-repair (Schegloff et al., 1977).
Figure 1
Types of Conversational Repair
Data were collected through conversation analysis. The data were taken from the transcripts of threaded discussions in a wholly online, master’s-level introductory course on instructional design offered at a regionally accredited university in the southeastern United States. Neither author was affiliated with the research site. The transcripts included the first five weekly course discussions, totaling 516 individual discussion posts across 31 participants. Both researchers coded all discussion posts. The researchers coded each discussion post for intersubjectivity using the Intersubjectivity Repair Dyad (Whye, 2019). The researchers also coded each post for type and frequency of conversational repair using the codes established by Schegloff et al. (1977). The researchers coded the discussion posts separately and then compared their scores. The initial measure of inter-rater reliability (IRR) was 86% agreement; after discussion, it was 98% agreement. The researchers used ordinal logistic regression to analyze the data.
The researchers addressed the research question by comparing the levels of intersubjectivity as measured by the Intersubjectivity Repair Dyad (IRD) (Whye, 2019) with the total frequency of the four types of conversational repair sequences (Schegloff et al., 1977). Total repair frequencies by IRD level are shown in Table 2. The IRD records the discussion prompt as Level 1, so there are no peer responses coded at Level 1.
Table 2
Total Frequency of Repair by Level of IRD
IRD Level Total Frequency of Repair
IRD Level | Total Frequency of Repair |
1 | 0 |
2 | 28 |
3 | 13 |
4 | 6 |
A cumulative-odds ordinal logistic regression with proportional odds was conducted to assess the effect of repair frequency on IRD level. There were proportional odds, as assessed by a full-likelihood ratio test comparing the fitted model to a model with varying location parameters (χ2(1) = 3.184, p = .074). While both the Pearson and deviance goodness-of-fit tests indicated the model was a good fit to the observed data, 67% of cells had zero frequencies. The final model statistically significantly predicted the dependent variable over and above the intercept-only model, χ2(1) = 52.241, p < .001. The total frequency of repair has a statistically significant effect on the prediction of the level of IRD, meaning an increase in repair was associated with an increase in the odds of scoring higher on the IRD, with an odds ratio of 9.559, 95% CI [5.524, 16.541], Wald χ2(1) = 65.096, p < .001. The total frequency of repair appears to decrease because there are fewer posts at higher IRD levels.
This study has implications for practitioners such as instructors and instructional designers. Practitioners are encouraged to reflect on their own practices, understand the constructs of intersubjectivity and conversational repair, and seek divergence of ideas rather than convergence. Other implications can be broken down into subsections based on specific tasks.
There are several implications for the design of asynchronous online discussions. Practitioners are encouraged to avoid discussion prompts that ask students to summarize course materials. All discussion responses will be similar, leaving little room for interesting replies from peers. An alternative is to design discussions that prompt students to apply course materials to their own unique experiences. There should also be sufficient, explicit instructions to help students synthesize course materials and fellow students' posts in their peer responses. This guidance might include a requirement that peer responses include a relevant question directed back to the student who made the initial post, as asking questions is one way to engage in conversational repair.
It is important to balance presence in the discussion with dominance. Too much instructor engagement in the discussion might stifle students' ideas and participation. Facilitators are encouraged to model expected behavior in their own posts, weaving together salient points from several student posts and connecting them back to the course materials. Asking questions back to students by name is another way to invite engagement in the discussion. When students are expected to facilitate the discussion, these student facilitators should have at least some training in engaging in the expected discussion behaviors. Of course, to adequately train student facilitators, the instructors need to engage in those discussion behaviors that promote conversational repair and intersubjectivity.
What gets counted gets done. Students are likely to meet expectations that affect their grades, so it is important to grade behaviors that promote conversational repair and intersubjectivity. Of course, tell students the expectations, as noted in the design section, and model those behaviors, as noted in the facilitation section. Use a checklist or rubric that measures the synthesis of ideas and provides students with the specific performance criteria. While it may seem obvious, it is worth stating that it is important to then measure students – grade them – against those clearly stated expectations and performance criteria.
There are also implications for researchers of asynchronous online discussions. First, researchers would benefit from recognizing the differences between simple interaction and the synergistic outcome of intersubjectivity. Focus more specifically on intersubjectivity than on similar yet distinct concepts like interaction, engagement, or collaboration (see Dennen et al., 2023). Researchers might also benefit from recognizing how the established technique of conversational repair manifests somewhat differently in asynchronous online communication than in synchronous, face-to-face communication. For example, in an asynchronous, text-based conversation, there are no non-verbal facial expressions that might indicate misunderstanding or conflict that needs to be repaired. There are several future research questions worth exploring:
What types of repair and repair sequences are correlated with the presence of intersubjectivity?
Does intersubjectivity correlate with social presence or cognitive presence as measured by the Community of Inquiry framework (Garrison et al., 2000)?
Can the Intersubjectivity Repair Dyad be used to measure intersubjectivity in other contexts, such as undergraduate education, different disciplines, or clinical contexts?
To what extent does discussion facilitation – by teachers or peers - enhance the use of conversational repair and intersubjectivity?
What is the role of the discussion prompt in guiding repair of the subsequent conversation and the development of intersubjectivity?
What role might time play in the development of conversational repair and subsequent intersubjectivity?
What affordances of the LMS might correlate with higher or lower levels of conversational repair and intersubjectivity?
The results of this study indicate that increased use of conversational repair in student posts is associated with higher odds that those posts show intersubjectivity, suggesting that conversational repair facilitates intersubjectivity. Encouraging the use of conversational repair in discussion posts might lead to greater mutual understanding (intersubjectivity) among the topics under discussion. Detailed operationalization of the component processes of intersubjectivity in social interactions is limited (Stevanovic & Koski, 2018); the findings of this study are significant in advancing the operationalization of these processes, a critical component of which is the use of conversational repair.
EdTech Archives