Systematic Review: Where is Current Research on Conversation Analysis?

As social media and other new communication technologies are integrated into teaching and learning environments, educators and researchers continue to be interested in the discussion that takes place in such spaces. This paper reports the findings and the research gaps grounded from current research articles on conversation analysis (CA). The data is collected from international and highly reputable journal publishers, namely Science Direct, Elsevier, Sage, and Wiley. From the resulting 49 articles collected, the screening excluded 24 articles. Therefore, it is the remaining 25 articles that are systematically reviewed. The results indicate several themes throughout the current research field, i.e. digital CA, theory and methodology construction, body language or nonverbal conversation, atypical interaction, usage of specific phrases, and novel settings and objects. The practical implication of this systematic review is a collection of research gaps and recommendations that researchers can take note of and tackle in future studies.


Introduction
Before, humans could only interact by face-to-face, written means, gestures, symbols and pictures. Now, information and communication technologies (ICT) enable people to interact remotely and asynchronously with massive disposal of online features available in multifarious platforms (Hamuddin, Rahman, Pammu, Baso, & Derin, 2020;Manca, 2020). With the insurgence of online communications due to the physical distancing issued worldwide to deal with the COVID-19 pandemic, this study finds it important to review the state-of-the-art research during the last five years to provide an overall picture for future studies on human interactions. One of the sciences that investigate how people communicate is known as conversation analysis (CA).
CA is an investigation on the interaction between two or multiple persons in any kind of context. According to Hutchby (2017), CA may seem to be a type of science that is "obsessed" on "obvious" details of a talk, but this science"s importance lies in discovering the variety of aspects in people"s interaction, the diversity of people who are interacting, the range of nonverbal signals people do when they interact and the how any type of context can influence the ways in which people interact (Marwa, 2015). Beyond the principal purpose of uncovering tacit reasoning behind the sequences of a conversation, CA contributes heavily to the continuing development of instructional and teaching techniques, persuasive techniques, medical therapy, court procedures, helplines and anthropology (Paulus, Warren, & Lester, 2016;Rancew-Sikora & Remisiewicz, 2020). American sociologist Harvey Sacks (1935-1975 is generally credited with founding the discipline. At the time of its conception, CA was used as a method in the sociology discipline. It is sometimes described with the term "talk-in-interaction" analysis or "ethnomethodology" that focused on casual conversation. Nowadays, CA"s purpose is clearer and more complex than merely identifying the underlying organisation of a moment-bymoment evolution of a particular conversation in a given context. This realisation of CA"s complexity is because conversation is now understood as more than just talking, but as a social act that reflects an individual"s ability and willingness to interact with others (Derin, Nursafira, Yudar, Gowasa, & Hamuddin, 2020). Additionally, with the fact that a conversation or interaction is a rather permanent feature in nearly every "phase" of human life (e.g. childhood, school life, work life, online/virtual life), so many disciplines conduct CA to fill their respective research gaps. This study aims to review CA studies that have been published from 2016 to 2020 to generate a focused view on the current findings and research gaps.
There have been two previous studies that also reviewed the literature on CA. Paulus (2016) claims to be the first study to conduct a systematic review on the application of CA to understand online discourse. The study drew on 89 previous studies since 1994. The themes forwarded in this previous systematic study include the contexts of the data, the fundamental structures of conversation, the research aims of using CA, the mechanics of how online talk is coherent to participants, the comparison of face-to-face with online talk, the ways participants accomplish social actions in asynchronous environments, and the techniques participants use to deal with trouble in online talk. The other literature review study on CA research is Meredith (2019), although this study did not detail how many articles they screened and reviewed. Rather, Meredith (2019) discussed the current understanding of the core organisation features of conversation, e.g. turn-taking, sequence organisation, repair, openings, and embodied conduct.
Both of the existing literature reviews on CA studies have been carried out with a focus on solely online talks, with face-to-face talks coming into consideration only if they are being compared with online talks. The contribution of this current systematic review is the inclusion of contexts beyond solely online talks. This study finds it important to discuss how CA has been used on all types of talks to emphasise the variety of research gaps that future studies can address.

Method
This study aims to report the findings and research gaps that have been addressed by current Conversation Analysis research, leading to identifying areas for future research. The method that is used to reach this objective is a systematic review, which is a type of literature review that is considered the gold standard way to synthesise the findings of several studies from different disciplines. Systematic review locates, appraises, and synthesises the best available evidence relating to a specific research question, resulting in an informative, evidence-based answers (Boland, Cherry, & Dickson, 2017). This method allows researchers to critically evaluate and integrate the findings of all relevant, high-quality individual studies addressing the research question (Munn, Peters, Stern, Tufanaru, McArthur, & Aromataris, 2018), in this instance, "what has been shown in current research on Conversation Analysis?" In conducting the systematic review, this study followed the guideline provided by Boland, Cherry, & Dickson (2017) namely scoping, planning, identification, screening, and eligibility. Scoping is to formulate one or more research questions. Research questions can be obtained from previous studies or future research conducted on articles that have been read by researchers. After formulating the research questions, proceed with clarifying carefully whether an approved systematic review has been carried out. Planning can be done by compiling research questions that have been formulated into several concepts to create search terms. The researchers formulated and then reviewed the initial inclusion and exclusion criteria in the initial phases of the literature search and selection process. In planning, researchers also make a clear recording system and keep careful records using systematics.
In the discussion, the researcher uses the search term to search for different electronic (relevant) pairs, examine the search results carefully and make additional searches to make sure the researcher has found all the works that he wants to be published and not supported. Screening is a place where researchers enter references into citation managers to arrange search results. Screening also involves reading the titles and/or abstracts of the identified works. The final stage is the feasibility where the researcher Take the full text version of the paper that can be qualified and extract the relevant information. Followed by tabulation and summary of the results, as well as analysing and synthesis.
This study adapted this guide from Boland, Cherry, & Dickson (2017) to suit with the research question. Eventually, the method of this study"s systematic review is illustrated in Figure 1. Figure 1 illustrates the methodological framework that this study adapted from the guide book.

Analysis and Synthesis
First of all, the research question has been formulated as "what has been shown in current research on Conversation Analysis?" Thereby, it is easy to decide that the research terms this study will use during the keyword-based search are "Conversation Analysis" and "conversation analysis".
The selection criteria are pinpointed on the fact that this systematic review intends to focus only on current research. So, this study only included articles that have been published in 2016, 2017, 2018, 2019, and 2020. Studies outside of these ranges are excluded. The criteria also includes only original research articles that deal with primary data, thereby excluding book reviews, editorial articles, and possibly other literature review-based articles. During the combing of the collected articles, this study would also exclude articles that weren"t actually about CA, using CA, even if they were first obtained due to containing the relevant keywords.
Using four internationally-accredited electronic databases for journal articles, namely Science Direct, Elsevier, Sage, Wiley, the keyword-based search yielded as many as 49 research articles. Fine-tuning the articles during the analysis on the full-text versions, this study included only 25 articles for the systematic review.

Findings & Discussion
This study found 49 articles from the electronic databases, and excluded 24 articles that did not make the cut. The remaining 25 articles that passed the screening are then tabulated. This study discusses the research gaps addressed in the selected articles and what these current studies have recommended for future researchers to pursue. This discussion frames the studies" recommendations as potential areas for future CA studies.

Digital Conversation Analysis
One of the first, and much explored, area in current CA studies in the comparison between face-to-face and video interactions. Not so little is known about the similarities and differences between these two types interactions nowadays as publication in this area continues to increase. A consequence of this prolific area of study is the development of CA methodology centred on digital conversations. Digital CA, which has been brought to recognition by Giles, Stommel, Paulus, Lester, & Reed (2015), is an enticing field for CA researchers to explore. Other terms have been used in place of "digital conversation analysis", such as "computer-mediated communication" (CMC), "computer-mediated discourse", "electronic discourse", and "online talk" (Derin, Putri, Nursafira, & Hamuddin, 2020).

Meredith (2017) emphasised how digital Conversation
Analysis is still an infant compared to spoken Conversation Analysis. The study explored the concept of affordances that is brought up due to the presence of technology influencing human interaction. This 2017 research focused on text-based interaction, but future studies can continue it by exploring the platforms that afford multi-modality of technologised interactions, allowing for pictures and videos to be included in CA. A particularly interesting implication for future studies to explore is the observation on how screen-capture data might seem to be a better form of CA data compared to timed log files due to how the former provides insights into how the interaction actually unfolds as it occurred for the participants.
A feature of video interaction, that is the otherattentiveness people display when they communicate through the screen, is recommended to be explored further by Stommel, Goor, & Stommel (2019), particularly when it comes to patient consultations. The study suggested that future research could have surgeons display otherattentiveness by stating rather than asking, or asking patients to choose between providing self-report first or receiving the pathology report. Gredel (2017) has integrated digital CA with Foucauldian discourse analysis to create a new analytic framework for analysing online interaction. The study used Wikipedia as the research object, so they mainly focused on text-based online interaction. Therefore, they similarly recommended future studies to conduct multimodal analysis on text-picture convergence. An exciting implication found by Gredel (2017) is the uniqueness of online interaction using strikethrough text, a feature that marks people "changing their minds" in a highly controversial or precarious context. This typographic peculiarity of online interaction is neither found in oral nor in written conversation, so future CA studies may benefit highly by studying digital discourses that contain this typographic feature.
Michel & Cappelini (2019) conducted a considerably thorough exploration on conversational alignment in computer-mediated interactions. To address the gap of studying naturally occurring L2 interactions in the face of the numerous highly controlled lab-based experimental L2 interaction studies, the study considered ten different types in four dimensions. In the first level, the study looked two different types of computer-mediated communication (CMC), i.e. video conference and text chat. In the second level, the study explored both two conversational alignments, i.e. lexical and structural alignment. In the third level, the study collected data on naturally occurring conversations in three different languages, i.e. Chinese, French, and German. Lastly, the study involved different statuses of the interactional partners, i.e. L2 peer, L1 peer, and L1 tutor. Recommendations by Michel & Cappelini (2019) include comparing the CMC between interlocutors with other different statuses and the difference between lexical and structural alignment. Bimodal alignment, bimodal turns, and disfluencies in video chats which will need to be distinguished by CA researchers from genuine self-alignment. The study also recommended to use lemmabased approach when coding the lexical alignment rather than using the exact overlap of N-grams, data-driven approach to establish distance between primes and targets rather than setting an arbitrary threshold, and to restrict to precise structures when comparing structural alignment across languages.

Theory and Methodology Construction
There are still some problems that are facing CA researchers for many years. Researchers who wish to focus on practices where there are actions in interaction often face difficulty in identifying them from their datasets because such research object is sensitive to context. This problem is exacerbated when CA practitioners have to deal with large datasets of spoken interaction. Therefore, Haugh & Musgrave (2019) proposed an explicitly combinatorial approach to search can more readily find potentially relevant examples of these phenomena. Their combinatorial approach is a way to overcome the practical challenge that CA researchers face when attempting to build collections for analysis from large tracts of data because it assembles a preliminary collection of a relatively infrequent conversational phenomenon. This approach may allow CA researchers with variability and generalisability.
For more specific purposes, such as cognitive therapy, Cannon, Meredith, Speer, & Mansell (2019) used CA to investigate the interaction between therapists and their patients to improve therapists" "stocks of interactional knowledge" (SIK). The study revealed incredible detail on when and how therapists ask about and clients recognising their own disruptions, e.g. changes in speech or mannerism. Disruptions in conversations is a regular occurrence, but pointing the occurrences shifts patients" awareness to a "mindful awareness" and the conversation to a "metalevel commentary", two things that can be explored further in future studies.
One study observed an overlooked aspect in many previous CA studies, namely "regrading". Its presence and consequences is a frequent occurrence in CA studies, but have been taken for granted in conversational practices. Bilmes (2019) argues that researchers should view the upgrading and downgrading as an object of study, laying evidence of its complexity, prevalence, and role in interactions. According to the study, regrading is a common move in conversation that has not been given sufficient study, and Bilmes (2019) points toward more studies on primary scales, secondary scales, and word choices as scaling choices.
Hall (2019) also points to another missing body, namely the jointly constructed actions and courses of action that comprise social contexts of use in the shaping of language. The study reconsiders the usage-based understanding of transdisciplinary perspective on second language acquisition (SLA). Offering the new terms repertoire, semiotic resources, and register as alternative terms to competence and grammar to better capture current understanding on SLA, Hall (2019) recommends charting new directions in L2 learning research by exploring specialised interaction metalanguage. Mondada (2019) expands the multimodal approach into a multi-sensorial approach to social interaction, showing that people engage with their bodies in not only communicating with each other but also in sensing the world. Their results invite future studies to explore the interactional conception of multi-sensoriality as a phenomenon to understand what makes embodied details accountable in shaping people"s actions while they communicate.

Body Language or Nonverbal Conversation
People also often delay in responding, but CA studies have only demonstrated that, in responding, recipients have two response options, i.e. immediately give a preferred response or delay to give a dispreferred response. Stokoe & Attenborough (2020) investigated how delays may actually produce preferred responses on four high-stakes interaction, e.g. police negotiations to suicidal persons, emergency calls to suicidal persons, mediators talking to potential clients, and salespeople talking to potential customers.

Atypical Interaction
Atypical interaction is a term that describes an interaction involving a speaker with communication impairment. This is a research area with a long history, but Wilkinson (2019) observed that the vast majority of existing studies have focused on one form of atypical interaction, that is the type involving autistic patients or people with aphasia. Meanwhile, there is a paucity of studies on the atypical interaction between participants who are still developing their communication abilities. With high interest to explore the overall existing knowledge on this body of research, Wilkinson (2019) highlighted three forms of atypicality, namely the delay, the understandability problems, and the actions in these types of interactions. The study ended with a note of advice for future researchers to prioritise studying developmental disorders because the current intervention programs predominantly lean on acquired disorders. The study also suggested further exploration on how communication impairment impacts the interface between talk and conduct, i.e. neural, cognitive, motor, and sensory structures. Tuccio & Garcia (2020) is a study on the linguistics of aviation flight instruction, specifically focusing on the personal hypothetical "I would" that is used frequently in recent instructional interactions. Traditional pilot training have mainly focused on testing students" technical skills, but contemporary focused on aeronautical decision making (ADM) which includes risk management, situation awareness, and resource management. Compared to a normal classroom instructor, a pilot instructor has overlapping roles and goals. They simultaneously act as a teacher, potential co-pilot, and safety monitor, and they have multiple goals that include giving directions, instruction, advice, corrections, and necessary physical interventions. It is found that the utterances containing the words "I would" firmly establishes the instructor as the expert role model, making the words identifiable as part of an interactional competence. The study recommends further exploration on the hypothetical use of "I would" by students or in other instructional contexts, particularly coaching and tutoring interactions that are mediated by technology. Interactional competence between native and non-native English speakers is also potential avenues for future studies, and future research may explore the instructor"s role as a customer servicer caused by the competition in the career.

Usage of Specific Phrases
Other CA studies looked into the usage of non-English phrases. Chen & Barnes (2020) explored the interactional functions of the Mandarin response token dui. The study researched this response in the context of selling cosmetic products, and recommends future studies to explore it in a wider variety of contexts beyond marketing. Kaneyasu (2020) studied the Japanese epistemic modals 'daroo' and 'deshoo'. Both of them are often described as plain/polite variants in formal conversations, but the study collected evidence on how the phrases are actually a frequent occurrence during informal conversations. The study argues to reclassify them as interactional resources for negotiating and achieving intersubjective stance. The study recommended future studies to study the phrases" usage in more formal contexts such as interviews, speech, essays, and see how other related forms such as 'ssho' used. One limit in the study is its exclusion of demographic factors, so future studies may provide a more thorough insight by taking into account of participants" age and gender.

Novel Settings and Objects
Monologues such as speeches are not considered an object of much interest for conversation analysis studies. One example is a religious sermon. However, Akhimien & Farotimi (2019) identified high frequency usage of conversational features such as call-response, adjacency pairs, openings, closings, repairs, and next-speakerselections, enough to argue the reclassification of religious sermons as a type of conversation that should be further explored.
At least a couple of recent studies introduced the field of linguistics to novel areas that might have been overlooked or have not been considered in previous research. Simone & Galatolo (2020) looked at the sequential and temporal coordination in the instructional interaction for indoor climbing sessions that involved blind and partially-sighted athletes. Few studies the instructional chain, cooperative engagement, intercorporeal attunement, and the distribution of action, and this study focuses on the sessions involving participants who heavily rely on verbal instructions. There may be other activities that use assistance instructions like indoor guided climbing where the climbing route plays a crucial role as it offers multisensory resource for the trainer who has visual access and the athlete who has tactile access. This novel setting also points future studies to look into assistance instructions in other activities, since this one is purposefully shaped to enable the participation of persons with impaired vision.
Moreover, Rancew-Sikora & Remisiewicz (2020) brought the attention to a global family ritual that hasn"t been under a lot of systematic scrutiny, namely child birthdays. This study looks at the family ritual of celebrating first birthdays systematically as a novel research because festive rituals have not yet been studied using methods of multimodal analysis of interaction. "Routine" is frequently used in CA as a regular, collaborative interaction, but "ritual" is relatively rare and is more of a synonym for routine. This study claimed that studying family rituals is important to be studied more by academics because it treats family as a collective unit and focuses on family processes, thereby providing meaning to repetitive group activities. In the end, Rancew-Sikora & Remisiewicz (2020) recommended future studies to recognise more "child involvement techniques" in multiparty ritual interactions by distinguishing verbal, vocal, and non-vocal (haptic, kinesthetic) actions, and examine more on how parents determine the extent of the child's participation, whether as a partner or as a topic for an interaction that's dominated by adults.

Conclusion
There are many studies that this study selected, but only the most interesting gaps and recommendations are discussed to highlight the ones with the biggest potential for the field of research on/using Conversation Analysis (CA). Firstly, digital Conversation Analysis that still has methodological issues, but offer novel conversational features such as the strikethrough text. Secondly, multisensoriality of human interaction. Thirdly, interactional competence (IC) which is the ability to jointly communicate in setting-specific ways; it is about using communicative resources to co-construct understanding and co-accomplish context-specific goals. Finally, family rituals in which the interactions treat family as one collective unit.