Addressing the Challenges of Group Speaking Assessments in the Time of the Coronavirus

This paper considers the challenges and possible solutions of moving a group speaking assessment from face-to-face mode to online mode during the coronavirus pandemic from February to May in 2020. The assessment involved 168 higher diploma level students from a range of departments at a tertiary institution in Hong Kong taking their second English subject at university. The students generally have a low level of English (Hong Kong Diploma of Secondary Education level 3). They are in their second semester in their first year at university which runs from September to July each year. The students’ first semester was curtailed due to the civil unrest in Hong Kong that led to the closure of the university campus in November 2019. The modified speaking assessment was successfully conducted in early May 2020. Feedback from the English language teachers in the subject was generally positive with the general feeling that students received a more accurate grade as a result of the one-to-one discussion assessment than they had previously received with the group discussion assessment. Grade data appear to confirm this. Student feedback to the modified assessment was mixed with some preferring the original ‘group discussion’ format while others opting for the revised ‘one-to-one discussion’ format. Overall, there were some administrative concerns, and the assessment rubric needs to be further modified if the revised format is continued.


Teaching Online
Teaching online using synchronous platforms was a new experience for nearly all teachers in the ELC. The PolyU uses Blackboard as its Learning Management System (LMS) and its synchronous teaching tool, Collaborate Ultra, was the main platform for teaching. Other platforms included Microsoft Teams and from the middle of the second semester, Zoom. The PolyU and ELC provided teachers with crash courses on the various platforms (English Language Centre, 2020b). Initially the ELC focused on training on how to use the platforms with subsequent sessions looking more at teaching techniques. Even with the training, it would be fair to assume that most teachers were uncomfortable with their ability to teach effectively online. While teachers could learn how to use various applications, techniques which would work in the face-to-face classroom often did not work in the online classroom.
Feedback from teachers after the first few weeks of online teaching included the following comments: "online teaching is exhausting, more so than face-to-face teaching"; "preparation time takes much longer as old materials have to be adapted for the online lesson"; "students are not turning on cameras which makes teaching more difficult"; "my lesson has become much more teacher centred as I get limited feedback from students"; "students prefer to type comments in the chat rather than ask questions using their microphone"; "I'm not sure if students are understanding my input"; "there are 'ghost' students who log in but don't respond to my questions". The comments from ELC teachers at the PolyU were repeated by other faculty in the university, across Hong Kong (HKCPD, 2020) and across the world (Blum, 2020).
While teachers by and large coped with online teaching -the lessons were taking place as scheduled each week and students were logging in -the overwhelming feedback from teachers was that we needed to simplify assessments for students and teachers. This meant reducing steps and procedures so that the assessment was easier to set-up, and that there was less to go wrong as it was not possible to deal with an issue face-to-face.

Challenges of Moving Speaking Assessments Online
Once it was clear that the university was planning to teach online for the whole semester, the ELC had to plan for moving all assessments (written and spoken) online. This planning involved discussion between the Centre's management and the subject leaders. As one of the Associate Directors in the ELC and the subject leader of ELC1008 I was heavily involved in these discussions.
Regarding speaking assessments, our own inclination as well as the university's was to avoid unnecessary complication and move to asynchronous forms of assessment. I identified the following challenges for the ELC1008 speaking assessment.
• Practical: Students not turning on cameras Students were unwilling to turn on cameras, and to a lesser extent microphones, in all subjects at the ELC. This was due to one or more reasons including students not actually owning a webcam, students being concerned about allowing others to see their living environment, and students being unwilling to share their spoken thoughts and/or average language ability with the whole class. If students were not willing to turn on their cameras or use their microphones, teachers had serious concerns about whether there could be sufficient practice opportunities before a group speaking assessment. Students not using cameras also led to an assessment integrity issue. Teachers could not be sure that the student taking the assessment was the student enrolled in the course.

• Practical: Students' living environment
Many students in Hong Kong live in small apartments with multiple family members. Teachers frequently reported students having their microphones on and hearing background noises of babies crying, TVs on and other family members talking. Arranging a discussion with a group of students when one or more had this type of living environment would be challenging. • Practical: Students' equipment and internet connection Many students reported that they did not own a webcam, or that the microphone and/or webcam on their laptop did not work. Internet connection was less of an issue within Hong Kong (Census and Statistics Department, HKSAR, 2020), but there were students joining classes from outside Hong Kong who faced connection issues. • Pedagogical: Differences between online and face-to-face discussions There are clear differences between an online and a face-to-face discussion. For example, in a face-toface discussion, participants can see when a speaker is getting ready to speak as they have an intake of breath that indicates readiness to talk (Blum, 2020). With an online video-based discussion, these signals are often lost. Turn taking appears to be much more rapid in a face-to-face discussion compared to an online spoken video discussion, where the medium can restrict participants from talking over each other. Instead, the discussion might become more formalised with students requesting assigned turn taking (Ockey, Timpe-Laughlin, Davis, & Gu, 2019). All of this is new for our teachers and there was great concern as to whether teachers could provide enough input to students if we were to conduct an online group discussion assessment. • Pedagogical: Input and practice type should match assessment The ELC decided early in the semester that the assessment format should match the input and practice activities. For example, if input and practice were in an online format, the assessment should also be online and not conducted in a face-to-face mode. During the early part of the semester, it was unclear when (or if) we would be allowed to return to face-to-face teaching. To provide some degree of certainty to teachers and students, we decided that as the input and practice were being done online, then the assessment would also be online.

• Pedagogical: Assessment Rubrics
All ELC assessments have detailed rubrics that are shared with students. The rubric for the speaking assessment in ELC1008 covers four components: Content, Participation, Language, and Pronunciation and Fluency (see section 10). The problematic component was 'participation' and while it was possible to make minor changes to the wording in this component, we wanted to avoid a major rewrite halfway through the semester.

• Technical: Blackboard Collaborate Ultra limitations
The PolyU uses Blackboard as its LMS. While the video conferencing system on Blackboard was usable, due to the vast increase in worldwide demand, Blackboard restricted certain features, including limiting the number of videos that could be shown at one time to only two, i.e. only two students could turn on their cameras at one time. This would mean that conducting an online video assessment with four students and a teacher would not be feasible.

• Technical: Synchronous assessment issues
The ELC staff members had numerous online video meetings during February and March 2020. Our own experience showed that during a meeting some participants would experience technical difficulties. For example, not being able to connect the microphone or camera; not being able to hear other participants; suddenly losing a connection. We could foresee that these types of problems would also occur during an online group discussion with students.
• Security: Opposition from students to Zoom; real security concerns The PolyU bought an education license for Zoom mid-way through the second semester.
While this was generally welcomed by teachers, it coincided with security concerns with Zoom (BBC Adam Forrester News, 2020). Students in Hong Kong also expressed concerns about the security of Zoom (SCMP, 2020).With the ELC1008 team of four teachers, two only used Collaborate Ultra, one switched to Zoom, and one tried Zoom for two sessions and then returned to Collaborate Ultra after canvasing student views. In the latter case, students expressed similar concerns to those reported in the South China Morning Post article (SCMP, 2020). • Administrative: Fulfilling the LOs of the subject One of the Learning Outcomes (LO) of ELC1008 is that students can "use appropriate verbal and non-verbal skills in spoken communication in a group context" (English Language Centre, 2020a). Teachers need to assess this LO so that they can confirm that students have fulfilled all the LOs of ELC1008. While the SL could change some of the features of the assessment, students still need to complete the basic LO.

Possible Solutions to Moving Speaking Assessments Online
Taking all these challenges into consideration, I considered the following possible options for the speaking assessment in ELC1008.
Option 1: Online synchronous discussion assessment with 3-4 students (teacher present as examiner) Option 2: Online synchronous discussion assessment with 2 students (teacher present as examiner) Option 3: Online synchronous discussion assessment with 1 student and 1 teacher (as participant and examiner) Option 4: Online asynchronous discussion assessment Option 1 was discounted for practical, pedagogical, and technical reasons. Option 2 was considered, but I felt that practical, pedagogical, and technical reasons still remained, and there were also concerns that the dynamics between two students might affect performance.
Option 4 was a preferred option for the majority of the ELC's speaking assessments, for example presentations and job interviews. We could address practical and technical issues by moving the assessment to an asynchronous mode. While there are some differences between a live face-to-face presentation and one delivered in advance by video, I believed that the students could still complete the LO, i.e. deliver effective oral presentations. Likewise, with a job interview, research showed (e.g. Hudson, n.d) that requiring an applicant to submit a video response to questions is increasingly common in the workplace, so SLs felt that this assessment could also be moved to asynchronous mode. However, SLs also felt that it would be difficult to make a discussion asynchronous as responses would be more like a mini presentation than responding to questions and different points of view.
As the SL for ELC1008, given all the different factors I chose option 3 as the preferred option. A member of the teaching team had initially suggested this option. I thought that practical and technical considerations would be addressed, and while there would still be differences between an online and faceto-face discussion, these would be less pronounced than having a discussion with a group of students. I would need to make minor changes to the assessment rubric (see Section 10). Finally, I believed that the LO could still be achieved by completing this discussion in a one-to-one format between a student and a teacher.

Further Details of Modified Speaking Assessment
The modified speaking assessment (Option 3) is loosely based on IELTS Speaking Part 3. Part 3 of the IELTS Speaking text is called a 'discussion'. The examiner and the test taker discuss issues related to the topic in a general and abstract way and, where appropriate, in depth. This test focuses on the "ability to express and justify opinions and to analyse, discuss and speculate about issues" (IELTS, 2020). This was generally what we wanted to test our students on and therefore after consultations with the ELC1008 teaching team, we believed that using this format was doable in a practicable sense for both students and teachers.
The teacher would tell the student the topic at the beginning of the assessment, and then would modify questions according to the student responses and ability of the student. The modified assessment lasted 5 minutes for each student. Each class group contains up to 20 students.

Preparing for the Assessment
Teachers started preparing students for the third assessment by doing practice activities from around week 8 of the semester, with more intense practice from weeks 10 to 12 after students had submitted their second writing assessment in week 10.
In a face-to-face class, the practice activities would involve group practice and a mixture of peer and teacher feedback. Preferably, students would be able to practice for the modified assessment in a pair 'role play' with one student taking on the role as the teacher. In this scenario, a third student could be an observer and then give feedback to the 'candidate'. For reasons already identified in section 4, I did not think it was feasible in the online classroom during the last semester.
Instead, teachers organised one-to-one practice activities on different topics in both a synchronous and asynchronous mode. The synchronous mode was time consuming for the teacher with a class of 20 students needing 2-3 hours to get through when taking the 5-minute practice task and follow-up feedback into account. However, it enabled teachers and students to go through a 'dry run' and solve most technical problems beforehand such as cameras or microphones not working, or keeping track of time. Teachers also organised practice activities using asynchronous tools including Flipgrid (Flipgrid, 2020). This tool allows students to record short responses to possible questions. While the asynchronous tools meant that students were delivering more of a presentation, it allows them to have several attempts at the task and can therefore make use of the technology as a rehearsal (Gong, Kawasaki, Yeung, Zhang, & Dobinson, 2019), and it also enabled teachers to give feedback on content, organisation, fluency and pronunciation. Flipgrid also allows for peer feedback, although teachers did not use this function in this subject.

Conducting the Assessment
The teachers in ELC1008 conducted the speaking assessment during the final 13th week of the second semester during class time. Teachers with a full class of 20 students were still able to complete the assessment within the 3-hour class. Teachers reported few problems although there were a couple of issues of students not being able to get their microphone and webcam working simultaneously. In these cases, the teaching team decided beforehand to continue the assessment only using the microphone.

Teacher Feedback
The four teachers involved in ELC1008 had a face-to-face meeting after the end of the semester to discuss the subject and online teaching. Regarding the modified online assessment, the following points came up: • Conducting the online one-to-one assessment was physically draining. If using the same format again, teachers suggested that they would assess fewer students per hour so that they would have more time to reflect on the grade and prepare for the next student. Some teachers allocated times for each student. For example, student A would come at 09:30, student B are 09:40, student C at 09:50 and so on. One teacher put students in groups of 3-4 and then allocated each group around 30-40 minutes in total. She said that this allowed her to manage the students better as she did not need to wait for students to arrive for their allocated time. • The one-to-one assessment allowed for a more accurate grading of students. Teachers commented that in the face-to-face group discussion assessment, weaker students could still get a satisfactory grade by limiting their contributions and therefore mistakes. On the other hand, stronger students were often concerned about 'dominating' the discussion and therefore were more hesitant about showing their full potential and might not get the grade that they deserved. Grades analysis confirmed this anecdotal evidence (see section 11 below). • If there is online teaching in the future, teachers were in favour of using the one-to-one discussion.
If face-to-face teaching can return then teachers were more hesitant about which mode of assessment to use -the group of students or the one-to-one with the teacher. This is possibly due to teachers' preferring what they know, and being reluctant to change unless forced to do so. At this point in time, it is still uncertain what mode of teaching the next semester will be in. • Teachers commented that the assessment rubric and the 'participation' component in particular would need further revision if the ELC repeats the one-to-one discussion in future semesters. This would apply for both online and face-to-face formats.

ELC1008 Assessment 3 Rubrics
The criteria for the ELC1008 Assessment 3 rubrics are shown in Table 1. As the assessment had changed from a group discussion with students to a one-to-one discussion with the teacher, small changes were made to the wording to reflect this. The changes are shown below with old wordings crossed out, and new wordings in bold.

Content (30%)
-meets expectations regarding the requirements of the task.
-develops clear opinions/arguments/analyses with relevant support using appropriate (academic) sources

Participation (30%)
-presents ideas clearly in relation to one another. Contributions are logical and discussion follows a logical structure -interacts with others / teacher to facilitate the discussion -makes effective use of verbal and non-verbal techniques to interact with the other members of the group / teacher Language (20%) communicates meaning clearly with accurate grammar and vocabulary -uses language appropriate to the audience and the topic Pronunciation and fluency (20%) communicates meaning through accurate pronunciation -speaks fluently, allowing for ease of understanding Table 1 above shows the descriptors for each criteria from A to F. There are also '+' grades which reflect some of the features in the grades above and below them. The passing grade is D.
The old and modified descriptors for 'A' and 'C' grades for participation are shown in Table 2. The part on 'presenting ideas' did not change so this has been omitted.

Participation 'A' Outstanding 'C' Satisfactory
The participation criteria in the old descriptors had an element of driving the discussion forward and leading the group. As the assessment changed to one that would be teacher led, changes were made to these parts.

Old
-presents ideas clearly in relation to one another. Contributions are logical and discussion follows a logical structure -interacts with others to facilitate the discussion -makes effective use of verbal and non-verbal techniques to interact with the other members of the group Modified -presents ideas clearly in relation to one another. Contributions are logical and discussion follows a logical structure -interacts with teacher to facilitate the discussion -makes effective use of verbal and non-verbal techniques to interact with the teacher

Old
Drives the discussion forward and involves everyone through effective use of interactive strategies (e.g. questions, clarifications and interruptions). Leads others in setting and following a clear, explicit and logical structure.
Shows keen active listening and responds appropriately to others.
Shows confidence consistently through effective use of posture and/or gestures and maintains eye contact with others effectively throughout.

Modified
Moves the discussion forward through effective use of interactive strategies (e.g. questions, clarifications and interruptions). Follows a clear, explicit and logical structure.
Shows keen active listening and responds appropriately to others.
Shows confidence consistently; maintains eye contact effectively throughout.

Old
Occasionally involves others through interactive strategies (e.g. questions, clarifications and interruptions). Generally follows a logical structure.
Shows some active listening and responds mostly appropriately to others.
Shows some confidence with occasional distracting posture and/or gestures and maintains eye contact with others occasionally.

Modified
Occasionally interacts with teacher (e.g. questions, clarifications and interruptions). Generally follows a logical structure.
Shows some active listening and responds mostly appropriately to others.
Shows some confidence; maintains eye contact occasionally.
Instead of leading the discussion, the modified criterion focused more on interactive strategies (e.g. asking questions) and following a logical structure.
In the old face-to-face assessment non-verbal techniques played a bigger part, including posture, eye contact and gestures. In the modified assessment we only had on eye contact as the position of the camera often meant that gestures or posture could not be seen.

Grade Analysis
ELC1008 had its last main run in the second semester of 2018-19 in face-to-face format. I compared grades from the assessment during that semester with grades during the 2019-20 online second semester. See figures 1 to 5.

Total number of students: 146
Total number of students: 168

Figure 1. Content grade (% of students)
In 2018-19 nearly 60% of students received a grade B for content compared to just overall 40% in 2019-20. A greater proportion of students scored B+ or A in 2019-20 compared to 2018-20. A similar proportion of students scored C+ in both years, but more students scored C or below in 2019-20. Overall, there was a wider spread of grades, and a general lowering of the grade. One of the criteria for content was providing sources to support arguments. In a one-to-one assessment, it is easier for students to provide evidence and for teachers to check. It is also easier for teachers to ascertain whether students have completed research and therefore those students who only did limited research received a low grade. These are possible reasons for the greater range and slight lowering of grades in 2019-20.
Total number of students: 146 Total number of students: 168

Figure 2. Participation grade (% of students)
Generally, more students scored lower in 2019-20 for participation than in 2018-19. Over 60% of students scored B or more for participation in 2018-19 compared to only around 40% in 2019-20. As with content, there was a wider spread of grades. Participation was a more difficult criterion for teachers to measure in the one-to-one discussion as the criterion did not fully match the revised assessment. In a group discussion with students, the interaction between students could be graded, while in a one-toone discussion with the teacher, interaction can be assumed as long the student answers the questions. This meant that teachers were concerned about what grade to award for participation in the modified assessment. (see section 10). To alleviate this concern, we gave more weighting to logically sequencing ideas in 2019-20 compared to the previous year.
Total number of students: 146 Total number of students: 168 The language grades also saw a slight fall from 2018-19 to 2019-20, and there was a wider spread. While a similar proportion of students scored B or above (25% and 27%), more students scored C or below in 2019-20 compared to 2018-19. One possible reason is that weaker students are unable to 'hide' in a oneto-one discussion with a teacher, and that teachers can more accurately assess their language ability.
Total number of students: 146 Total number of students: 168 This drop in grades is also likely due to students not being able to 'hide' in a group context. An additional reason might have been the lack of practice by some students before the assessment, although it is doubtful whether this would greatly affect the pronunciation and fluency grade. A few extra hours of speaking practice is unlikely to lead to a significant improvement in this area. One concern before the assessment was that internet connectivity would mean that teachers are unable to hear some students clearly. However, no teachers reported this problem after the assessment.
Total number of students: 146 Total number of students: 168 Figure 5. Overall grade (No. of students) Overall grades showed a wider range. In 2019-20 there were some students scoring an overall D+ or D compared to none in 2018-190. Grades showed a slight fall with around 40% scoring B or above in 2019-20 compared to 53% in 2018-19. Around 23% scored C or below in 2019-2020 compared to only 7% in 2018-19. The fall in overall grades is due to the falls in grades for content, participation, language, and pronunciation and fluency.

Student Feedback
All students were surveyed via a Google form after the end of the semester. The response rate was 50 out of 168 students ( Figure 6). Students were asked to rate their preference for a group discussion assessment or a one-to-one speaking assessment with a teacher, and then give reasons in an open-ended question. Online or face-to-face preferences were not covered in the survey.

Figure 6. Students' Preference for speaking assessment
Prefer group speaking assessment with students (score 1) Prefer individual speaking assessment with teacher (score 10) Figure 6 shows that students had mixed opinions as to their preference for the type of speaking assessment. 25 students (50%) preferred the one-to-one assessment with the teacher (6 to 10) while 23 students (46%) preferred the group speaking assessment with students (1 to 4). Two students (4%) did not have a preference for either. Student reasons (reported verbatim) for keeping the group assessment with other students included: • Opportunity to share different opinions with other student • Similarity to DSE (Hong Kong Diploma of Secondary Education) speaking tasks • Speaking with a teacher makes students more nervous • Easier and more relaxed Student reasons for keeping the group assessment with the teacher included: • Teachers can control the time • More chance to have individual speaking with the teacher • Less stressful to speak with the teacher / easier • Not duplicating options with other students It is interesting to note that students used 'stress' and 'perceived difficulty' as reasons for both assessment types. Generally students who preferred a group speaking assessment with other students stated that they wanted the opportunity to share views with other students, and that the assessment would be familiar as it is similar to what they have done in secondary school. Students who preferred a oneto-one speaking assessment with the teacher indicated that they preferred a more controlled experience, valued the opportunity to speak with a teacher, and acknowledged that group speaking with other students could lead to other issues.
While interesting and potentially useful, the results of this student survey might not be much help in informing future decisions about this speaking assessment. As can be seen from the survey there was support for both the 'group' assessment with students and the modified one-to-one assessment with the teacher.

Reflection
Overall, the modified assessment went well. Our main concern had been technical and practical issues preventing the assessment from running smoothly. We mitigated these issues by modifying the assessment to a one-to-one format, which also allayed the pedagogical and administrative concerns. The PolyU has not confirmed when face-to-face teaching will resume, so it is possible that this online assessment will continue in the next academic year. However, the following concerns remain.

Student grades
Overall student grades fell for the speaking assessment. The reason for this is possibly for one or more of the following factors: the one-to-one assessment better reflected the student's performance than a group assessment; the students lacked speaking practice opportunities in the online classroom compared to the face-to-face classroom; the cohort of students in 2019-20 were weaker than the 2018-19 cohort; the disrupted first semester at PolyU meant students received less input in their first English subject than the previous year.
The rubric was modified only slightly and weightings were left unchanged. Therefore, this is unlikely to be a reason behind the fall in grades, although the rubric not fully matching the modified assessment might be a partial reason.
There is also evidence that group assessments might have lower reliability than one-on-one oral interviews due to the relative lack of control by the examiner (Van Moere, 2006 as cited in Ockey, Timpe-Laughlin, Davis, & Gu, 2019). It is probably wise to combine a group speaking assessment with an individual speaking assessment so that teachers can determine a truer reflection of the students' ability.

Students showing their videos
Getting students to display their video was a major hurdle for teachers and affected the whole semester. Rather than focusing on teaching and getting students to participate in lessons, some teachers might have been over concerned with this aspect of online teaching. Whether there is a need for students to always turn on their camera, or only turn on when asking questions, or not turn on at all is now being widely debated (Reed, 2020). There are assessment integrity issues (e.g. is the student taking the assessment the same as the student taking the course), which we need to address, but the potential for having online synchronous spoken assessments without video is an unexplored area. Having practice tasks which do require the use of video could ensure that integrity issues are addressed, while some of the challenges identified in section 4 of this paper (e.g. student's living environment, technical issues) can be addressed by having an assessment without the need for a video.

Student preference
The results of the survey showed that students had very diverse views as to the format of the speaking assessment This is an area that needs to be investigated more. Associated questions include student preference for face-to-face or online speaking assessments, and if online, synchronous or asynchronous speaking assessments. While it is important that student views should be taken into account when designing assessment tasks, there also needs to be a balance with administrative, pedagogical and integrity concerns.

Type of speaking assessment
In their first semester at the PolyU, the higher diploma students do an individual presentation as part of their first English subject, ELC1007. We discontinued the Q&A part of the presentation when this became an asynchronous online assessment at the end of the first semester. I think it is important that we maintain a speaking assessment with some degree of interaction for the higher diploma students (i.e. in ELC1007 or ELC1008). Current plans are that the online asynchronous presentation assessment will continue in ELC1007, and therefore the question is whether the discussion in ELC1008 should involve active teacher engagement or not.

Skills for the 21st century: Synchronous video conferencing
The workplace environment in the ELC's has changed significantly over the last 6 months. The default meeting is now using Zoom, and colleagues are adept at choosing their own backgrounds, muting their microphones when not speaking and sharing their screens. These working practices have been replicated around the world, and it is highly likely that fresh graduates will also need these skills. As Vurdien (2019) points out, learners are able to enhance "their communicative competence" through video conferencing, and in addition they can become proficient in a tool that will be increasingly used in the workplace. For this reason, even if face-to-face teaching can return, we should maintain some element of video conferencing in the curriculum.

Assessment rubrics
We will need to further revise the participation rubric if the assessment is continued in the one-to-one format. One possibility is to adapt the IELTS speaking descriptors and have the following components: Content; Fluency and coherence; Language; Pronunciation. This would perhaps more accurately reflect student ability than the current modified rubrics.

Feasibility of Wider Use
I think the modified one-to-one assessment worked well with this particular group of students: i.e. generally quite weak in English; in their first year at university; monolingual; unwilling to turn on cameras and microphones. Students with a stronger ability in English, (i.e. those on undergraduate programmes) were more willing to use microphones, and to a lesser extent turn on their cameras. Hong Kong does have excellent internet and there is general widespread ownership of mobile devices. However, some students reported not having a webcam / working microphone on their laptop, and the living environment of many students is not ideal for learning. All these factors would need to be considered when running an online synchronous group speaking assessment.

Suggestions
I think the area of online versus face-to-face assessments, especially for assessments with a speaking component, has a great potential for further research. The ELC is already planning to move its in-class handwritten assessments online, and some speaking assessment will also go online. Benefits include freeing up class time for more input, practice and feedback, and allowing students more opportunity to revise and improve their final product. The major cost is academic integrity -is the student completing the assessment the student enrolled in the course -and how much help is the student getting (if any) during an asynchronous online assessment.
Another unexplored area is the differences between synchronous spoken discussions taking place online and face-to face. While there are many materials available for teaching and practicing face-to-face discussions, I think there needs to be research into the features of online synchronous discussions that can then inform the development of materials to support student learning in this area.
Moving speaking assessments online, especially those involving some form of interaction, was a challenge for the ELC. My advice is to keep things as simple as possible, and generally opt for asynchronous assessments as this will lessen the impact of technical issues. If an online synchronous assessment is needed (e.g. a discussion type assessment), then I recommend it is modified to reduce the number of participants to a minimum. This will also reduce the impact of technical problems.
When conducting assessments online, whether synchronous or asynchronous, it is important to anticipate technical issues and devise policies to cater them. For example, during the ELC1008 modified assessment we allowed the assessment to continue without a student video as long as we were certain that the student speaking was the student registered in the course.
As the ELC gains more experience with online assessments we will become more adventurous with the format. We are already hearing from other universities in Hong Kong about how synchronous writing assessments were 'proctored' using Zoom. I think we are at the start of exploring many new opportunities with online assessments, and the coming years will see some subjects move away from the traditional writing and speaking assessments that have remained unchanged for decades.