Planning communication access for online conferences

This post was written by Gabrielle Hodge and Lauren Gawne, and originally published on The Research Whisperer blog.

There’s a lot to like about online events. Travel, and the associated cost, are no longer a barrier to participation, making it easier for a wider range of participants to engage with conferences. This has made online events more accessible than physical events, but online events still need a thoughtful approach to communication access. In this post we provide a brief introduction to making conferences accessible to deaf and hard of hearing participants, and our experiences of this as a participant (Gabrielle Hodge) and an organiser (Lauren Gawne) of the 2021 International Conference on Linguistics Communication (aka LingComm21). Making communication access an intentional part of academic event design benefits all participants, yet it does require event organisers to do the work.  

Communication access is about ensuring people can bring their best to the event and for everyone to engage in all directions. We want our academic communities to reflect the same variation in lived experience and expertise as the rest of our lives. Planning for communication access should be the same as planning physical access or catering: you don’t wait until people turn up and tell you they’re hungry to plan catering for an event. Communication access should be built into every event, much like making sure accessible toilets are available, that everyone can get into the building and use facilities with ease and that there’s a range of food, not just egg sandwiches. Here are some common and easy-to-implement communication access options for you to engage with your deaf and hard of hearing colleagues.

Captions are a written representation of the spoken content that appears simultaneously on a screen. Closed captions provide people with a choice to display them or not, while open captions are always running and fixed on the screen for everyone. Captioning can be very effective for pre-recorded video content, or a verbatim transcript can be provided for audio-only content such as podcasts. Captions need to be created by a human to provide access for conference participation. 

Automatic captions are generated using AI, and can be more or less accurate depending on the language, the speaker’s accent and frequency of technical vocabulary. There is a reason we often call them “craptions”. A common assumption is that automatic captions automatically provide communication access for deaf and hard of hearing people. They do not. The automatic caption function should never be switched on and assumed to be providing access: they need to be checked and corrected by a human to ensure they are accurate. You should only make use of automatic caption functions at conferences if you are preparing a pre-recorded resource and are prepared to correct the transcript and time alignment of the captions yourself. 

A 4 image square meme from Star Wars. First panel: A young Anakin Skywalker, text “i’ve added subtitles to make my content accessible. Second panel: Padme Amadala, smiling, text: “but not automatically generated ones, right?”. Third panel: Young Anakin looking serious, no text. Fourth panel: Padme looking slightly concerned, text: “but not or tomato call he gem her eight it ones, write?”.
Meme via Tony B (@Saltbar)

Live captions are human-generated by a specialist stenographer (best accuracy possible) or through re-speaking technology (more prone to error) and occur with speech in near-real time. Professional live captions are much more accurate than automatic captions. At a physical event these can be projected onto a screen or device. At digital events, depending on the workflow, they can be displayed in Zoom, a YouTube livestream, or on a separate webpage. Professional and accurate live captions are generally easy to read and understand. They provide quality access to the spoken language content of the event.

Sign language interpreting at academic events involves qualified sign language interpreters who simultaneously interpret between a spoken and a signed language. For example, between spoken English and BSL (British Sign Language) or between Libras and Brazilian Portuguese. This is an incredibly complex job requiring exceptional proficiency in the working languages, and you’ll often see interpreters swapping out repeatedly over an hour-long event. Interpreters may be deaf or hearing. They might work between a spoken and signed language, or between two signed languages. At some events, there may even be a few different sign language interpreting teams all working between their languages as one big team. Sign language interpreting requires a lot of preparation and therefore must be organised months in advance: ideally as soon as you know the dates and times. One week before the conference is too late. Online conferences are radically changing the working conditions of interpreters (not necessarily for the better), and re-igniting debates in different countries about what signed languages are available at international conferences, and who is responsible for carrying the cost. These discussions might not be happening in your country or research community, but if you’re organising an international event it can be good to know they might be happening in other places.

There are other communication access options that you might also need to consider depending on the event community, such as audio descriptions for blind people. We’ve described the ones above because they’re kind of like vegetarian catering at an event: that is, there is always some background percentage of society that will require vegetarian food as a matter of priority, but vegetarian food options also cater to a range of other participants (in the case of food, this might be people who keep kosher, have pregnancy-based food aversions or just don’t like eggs). In the case of captions, many other people can make use of them: people who have auditory processing issues, people who are listening to a language that is not their first language or who prefer to read, and event organisers who might like a written record of the event. However, these are not the only options for communication access, and many events will benefit from a combination of several options. You should always make space for people to disclose these needs, and build them in as needed. At the same time, if such people are ongoing participants in your community (or are even the subject of study, in the case of sign language linguistics) then they must really become a standard feature of events full stop. It’s the same if a member of your community regularly attends and requires sign language interpreters. 

Of course, it’s rare to go to a conference where everyone is thrilled with the catering, and you won’t always accommodate everyone with the communication access you provide either. The important thing to remember as an organiser is that this is a limitation of the event resources, and not a limitation on behalf of the participants. In the case of LingComm21, we made it clear to everyone from the outset that the registration fees were directly funding live captions for some of the conference, i.e., all of the plenary events and 1-2 of the streamed sessions. We chose to begin with live captioning because we had participants from across the English-speaking world and didn’t want to privilege one sign language from the outset. We also had a clear section on the registration form for other access requests in case someone did want specific sign language interpreting, so that we could work together to arrange it. We ensured our conference platform had technical capacity to include interpreter video feeds if someone wanted to bring their own interpreters, rather than use the captions. Afterwards, many attendees acknowledged that live captions were important, even if they didn’t directly benefit from them. They also said they were willing to pay more in registration in future years to cover these costs. Funding for communication access can be a major budget element, but one of the most important things we can do is normalising it as part of budgeting for your event. 

These guidelines for integrating captions and signed language interpreters from Julia Miller and Gabrielle Hodge at the Centre of Excellence for the Dynamics of Language are a good place to start familiarising yourself with these topics in more detail.

With all of these considerations, we want to discuss how communication access for deaf and hard of hearing people played out at LingComm21 in April 2021. In this section we’ll begin by sharing our individual experiences as organiser and participant in this event, before reflecting on these experiences together. 

Event organiser (Lauren Gawne)

We set out with the intention of designing LingComm21 as an online event that brought the same social energy as a physical event. We held the event in Gather, which is a proximity-based social video platform. This allowed us to have participants move about and use text chat as well as audio/video chat. We had some areas that were specifically flagged as text-only, which was useful for a variety of people who didn’t use audio, including people with low bandwidth internet, and participants who were bravely joining us from Europe at 2am and were in quiet houses. 

As a new platform, Gather still has some way to go to make caption integration smoother. For large sessions we were streaming into Gather from YouTube, but this workflow added enough variables to mean that not everyone was seeing the captions alongside the videos. Videos in Gather are also small unless you click to expand, and when people turn their video off you need to prompt them to turn it back on, which can make it hard to facilitate signed chat. 

We planned live captions rather than interpreting because of the international scope of the audience. The captioning company we worked with were incredibly professional and produced incredibly high-quality transcripts that exceed anything that an auto-captioning program would have managed with our diverse accents and technical vocabularies. As this was the first time we ran this conference and used this workflow, we had to adjust things over the days. This included discovering that while we had tested the transcripts came through to YouTube, and that the YouTube stream came through to the conference, we had failed to test how the closed captions came through in the stream to the conference. It turned out they did not. Thankfully it was a small enough conference we could work directly with affected participants, but it was still mortifying to add additional friction to the event for people. In those moments my main concern was ensuring we got the technology working and not burdening participants with excessive apologies or unnecessary explanations or excuses when they just wanted to get on with the conference (and instead saved the histrionics for my event co-chairs later in the day).

Event participant (Gabrielle Hodge) 

Going into the conference, I was excited because I knew the organisers were conscious about communication access right from the beginning. I loved how it was promoted in terms of communication access and how different people responded to that on social media. The registration form was a dream: it was clear the organisers cared about access in general, that I could ask for different options, and they would work actively to make it happen. It did not feel like the usual event registration form tick box, which is a tiny box down the bottom asking if any additional needs were to be accommodated. I have never received a response from one of these boxes. Organisers either ignore it or address it far too late, like the day before the event begins. So it was great to see this aspect of conferencing addressed properly within the actual registration form. Overall, I found that knowing a lot of effort had gone into planning and including communication access at LingComm21 meant that I was way more invested in the conference. I also felt more understanding and forgiving about how different aspects of the conference played out. 

This event was my first time in Gather, after months of seeing people either rave or complain about this platform. The LingComm21 Gather town was amazing: it looked like a real conference venue, played like an eighties throwback, and even had a couple of ducks hiding in various places that we had to search and find. It was very clearly designed by internet and scicomm nerds, and you wouldn’t want it any other way. A couple of familiar conferencing moments: feeling lonely standing alone by my poster and being rescued with a warm welcome from one of the hosts; a random but interesting conversation with someone working on a completely different topic, where we spent most of the time trying to understand how we each use one specific word; finding my way (late!) to the first plenary talk and seeing a big name rush past in apparently the same flustered state (also late!); the social anxiety of walking into a big room with conference tables and people already sitting down at them (people! sitting down at tables!); and the little golden roped off section near the main stage reserved for the captioners (very sweet and nice touch). 

Not all of the LingComm events were captioned, but there was a range of topics and talks that were available with captions. I believe the organisers allocated the captioning schedule depending on which topics people needing captions had indicated they were interested in attending on the registration form. On entering the Gather space, there were text-only spaces clearly marked out, but there were also fewer people already sitting at these places. Most people were in the regular video space and speaking face-to-face with each other. I did notice people making a beeline for these video spaces, but possibly they already knew the people sitting there and/or wanted to speak with them. When the presenters began, I was ready and waiting for captions to show on the video, but they did not appear. Apparently there was a technical glitch affecting how the captions were streamed, so you could not actually watch the speaker video in GatherTown with the captions. There were two options: (1) watch the speaker live on YouTube with live captions, or (2) read the captions as page text via a separate link. I initially tried to watch them on YouTube, but the timing between speaker and live captions appearing on screen was too delayed, choppy and frustrating to read, so I switched to the page text. This was not ideal, as it is difficult to monitor two different browsers (the Gather and the caption page text), and the Gather browser also had different chat boxes popping up. However, the benefit of reading captions as a page text was that it was like reading a book, easier to skim and absorb quickly. I could then jump back to the main Gather browser and catch up on those chat boxes, while waiting for the next page of captions to load. 

I also attended a couple of workshops where we all watched a brief presentation, such as how to write and pitch a science communication idea to mainstream media outlets, and then talked about how to action our pitch ideas. Everyone in my group was so friendly and thoughtful. We were able to have a few different conversations going on at the same time: one group conversation plus any private direct messaging with different individuals. The speaker expert also came around and gave us concise and helpful feedback on our pitch ideas. Overall, these smaller interactions resulted in a very positive conference experience and I think they are a valuable part of the online conferencing experience. There was no need to think or worry about access, we just communicated via text with videos on and it seemed everyone had a good time. This aspect of conferencing was probably easier than if we were all there in person, where we would need to pull out pen and paper or tech to communicate, plus those who are hearing would have to contend with background noise and so on. I also met a few different researchers who have taken their research skills and interests to the workforce outside academia and that was inspiring too. 

https://theresearchwhisperer.files.wordpress.com/2021/12/screen-shot-2021-10-11-at-8.47.52-pm.png
One of the seminar rooms at LingComm21, includes ‘text only’ tables and the chair reserved for the live captioner.

Event debrief (Lauren and Gab)

Our experience with the technological limitations of live captions at LingComm21 illustrates the challenges of communication access not going right. While the organisation and planning of live captions at LingComm21 were very much within the control of the conference organisers, we failed to anticipate the issue with caption integration, and lacked the ability to find a frictionless solution within the limits of Gather. Above we highlighted how we both immediately dealt with the situation. It was relatively painless compared to the online conferencing experiences that deaf and hard of hearing researchers have most often experienced and discussed online during the pandemic and earlier. LingComm21 was small enough that the organisers had a sense of who was making the most use of the captions. Lauren and Gab also have a long enough professional relationship for Gab to trust that Lauren was earnestly trying to work within the constraints of the tech. 

Unfortunately, there’s a commonly observed pattern at conferences where organisational and/or logistical issues lead to people missing out or even being denied communication access: regardless of whether deaf and hard of hearing people complain privately or publicly, the response from organisers is usually to dismiss the seriousness of the issue and/or respond defensively. When things go wrong with communication access and people are excluded, it’s common nowadays for people to complain on Twitter. This is often extremely effective for the person who has been excluded. It can also be very unpleasant for the organisers and they often feel attacked. However, it is important to recognise that such public complaints are not a scream into the void; rather, they result from the “democratisation of disagreement”. They are a call to action. Quick action. The kind of speed we do not get when we ask politely in private.

Not all people who are excluded due to lack of access consideration will feel comfortable communicating these issues, which is why building space into the planning and registration phase is so important. Minoritised people are used to being celebrated when it’s easy, and told to know their place when they try to challenge things (Erika Stallings refers to this as the ‘pet to threat’ trajectory for Black women in professional contexts). Deaf people are often minoritised in academic contexts, but deaf signing communities are diverse and each individual brings different experiences and needs. Access — and the experience of exclusion — is felt deeply and personally. Emotions can run high, even in well-negotiated situations, as we’ve highlighted above. The person you are excluding is not responsible for managing your feelings when your planning excludes them. This article on the deaf experience of “hearing arrogance” and exclusionary academia captures the frustration many feel when trying to engage with others at academic events. 

The appropriate response to these complaints is humility and action. Anything that even slightly smacks of defensiveness and the need to save face in public, e.g., “why didn’t you first contact us privately so we could resolve this?” is never going to go down well. Such reactions equate to blaming deaf and hard of hearing people for your own failure to ask people what they need way in advance, especially when it is an event where better awareness should be expected. One recent example was an online conference that included presentations about deaf people and sign languages, all presented by hearing non-signers. Their presentations had captions, but the live Q&A sessions afterwards did not have captions or sign language interpreters. This meant deaf signers were locked out of public discussions about our languages and our lives. The only people who could participate were hearing, English-speaking people, most of whom were non-signers. This example illustrates how many conference organisers work against a very important principle: nothing about us without us. We’ve focussed on accessibility for deaf conference participants here, but there are many ways that conference organisers can make events more accessible to more participants. 

When things go good:

It would be really handy if we could just point you to a single check list or set of resources for communication access, but it doesn’t work that way. No two groups of people are the same, needs constantly change and technology changes too (for good and bad). Access is an intersectional activity, and making a conference more physically or linguistically accessible might then require a subsequent rethink of how to make it financially accessible. Thinking of accessibility as an ongoing learning experience can ensure we also do not get complacent about the work that needs to be done. Accessibility requires event organisers to listen, from the planning stages through to feedback, and to take action on what they have learnt. We don’t deny that a good conference takes work for organisers, but people whose access needs are forgotten or ignored are constantly doing the work to even make it in the door. You might only organise a handful of major conferences during your academic career, but some people have to do this work for every single event they want to attend. Even when the conference takes accessibility needs into account, there’s still monitoring and feedback to be done. Even a positive response to a conference can come with lots of additional work. You might have attended a perfectly adequate conference and then found yourself writing a lengthy blog post about communication access at online conferences… 

In this post we’ve tried to illustrate what communication access at an online conference looks like from the perspective of a conference organiser and a participant. This isn’t an ‘us’ and ‘them’ situation. We are, at different times, organisers and attendees. For some of us, our access needs are implicitly considered in standard conference design. For the rest of us, conferences can be a place of additional labour just to attend. Accessibility starts with mutual respect and learning to listen so we can all engage. Accessibility is an ongoing project that requires ongoing conversations, willingness to try new things, and understanding that what might be an occasional consideration for you is an ongoing barrier for someone else. 


Acknowledgments 

Thanks to Maartje de Meulder, Indie Beedie and Gretchen McCulloch for feedback on earlier versions of this post. Thanks to the LingComm21 team and participants for such a great online conference. 

Actions for social justice

Guest author

Gabrielle Hodge

Dr. Gabrielle Hodge is a deaf researcher specialising in the linguistics of signed languages. Her work focuses on the semiotics of face-to-face interaction and the sociology of language use.

She tweets from @gab_hodge, and her ORCID is: 0000-0001-8677-6149.

Online conferencing

This post is part of a 6 part series called LingComm21: a case study in making online conferences more social

  1. Why virtual conferences are antisocial (but they don’t have to be)
  2. Designing online conferences for building community
  3. Scheduling online conferences for building community
  4. Hosting online conferences for building community
  5. Budgeting online conferences or events
  6. Planning accessible online conferences

Budgeting online conferences or events: The case of #LingComm21

In other posts we discuss the importance of creating a meaningful, functional, and accessible conference experience. Running an online conference or similar-scale event is not free of cost. In this post we want to bring some transparency to the costs of organising an online conference. Access to skilled volunteer labour and funding will be different for everyone, and we will try to avoid making presumptions about what kind of budget you have, but we discuss costs, revenue and some of our specific approach to balancing the budget.

Physical conference budgets are massive, and that’s even taking into account that the conferences outsource most of the costs of travel, accommodation and feeding people to the participants. People are used to getting things for free on the internet, and online conferences are much, much more financially accessible than physical events, but for a good conference to be run well, people should expect there to be some cost. 

As with many things, there is a certain amount of exchanging time for money or money for time that can be traded off depending on which one your organizational team has more of, but ultimately some of each is necessary: even if you hand-code every piece of software yourself (we do not recommend this), there are still server and bandwidth costs, and even if you delegate the entire running of the conference to paid assistants (we also do not recommend this), there will still be decisions that can only be made by the organizer(s).

Costs
Infrastructure costs
We chose to run the event in Gather to help foster social interaction. Gather has a public commercial pricing guide, but they also have heavily discounted custom pricing for universities, charities and similar non-commercial entities. An event like LingComm21 with 200 participants will thus need a budget in the hundreds of dollars rather than thousands, and with much more scaling flexibility than a physical space (we were rebuilding walls hours before the conference, which no hotel or convention centre can offer). We recognize reaching out to Gather (or a different platform) well in advance of the conference to confirm rates, as their pricing may have changed since our event (and they were also able to provide us with helpful feedback on our space design based on other people’s experiences).

Many of our other costs were mitigated by our networks. Several committee members had access to institutional Zoom accounts to run some panels, and the Lingthusiasm team had the Slack and YouTube accounts needed to make the co-ordination and streaming work. We ran ticketing through Eventbrite, which charged a percentage fee, but it made the registration process far more easy and flexible than we could have managed with another system.

Paying a third party company to run large-scale events is common for physical conferences and will undoubtedly become the norm for large online conferences as well. Much of the cost of these services includes the digital infrastructure they have access to, which can be costly for high-traffic events, and the labour costs of having people manage processes like livestreaming and captioning. We did the livestreaming in-team for LingComm21 using free software called OBS, which made it more affordable at a strictly dollars and cents level, but it also required a number of members of the organising committee to spend several days before the conference teaching themselves how to use the software as well as taking themselves away from participating in the conference to run the streams, so in future we would prefer to budget for someone else to run streaming. 

Accessibility costs
Although not all possible accessibility needs can be foreseen by conference organizers, and we did have a slot on the registration/participation forms asking about other accessibility requirements, the inaccessibility of audio content is extremely well known and formed part of our planning from the beginning. 

Because LingComm21 was conducted primarily in English, and involved an international community that spans many linguistic areas, we decided on live captioning for the event, rather than interpreting for a specific signed language or several signed languages. Captions also provide an ancillary benefit for attendees who are hard of hearing, have auditory processing difficulties, or who speak English as a second language, in addition to providing a written reference for the organizers after the event. We had live captioning for 11 sessions throughout the conference, including all the introductory and plenary sessions. This cost around $5000 USD and constituted the single biggest expense for the LingComm21 conference. 

We contacted captioning services for a quote as we were still figuring out the structure of the conference, which allowed us to reduce captioning costs by building in breaks for both the captioner and conference participants, meaning that we did not have to pay a second captioner to work in shifts to cover overly-long sessions (and also providing social benefits for participants as detailed in other posts). Reserving captioners early in the planning stages also allowed us to announce that captioning would be provided from the initial advertisement of the conference, rather than waiting for people to request it or invisibly assume that they were not welcome. Around 10% of participants who did the feedback survey said they used the captions, and the majority of the remaining 90% said that although they didn’t personally use them, they were happy to see them there, even though we didn’t solicit this information. (What can we say, the lingcomm community is pretty great.) 

Live captioning is a fixed cost that does not vary whether an event has 10 participants or 1000, and cannot be solved by additional volunteer or organizer labour, as professional captioners have specialized keyboards and training allowing them to type at conversational speed. Captioning labour can be delegated to presenters if presenters are required to upload videos of their talks in advance, at which point they can also be required to furnish them with captions; however, we decided that while this approach may work for some conferences, it was incompatible with the highly interactive conference we were aiming for. Automatic captions, while sometimes useful for individual or small-group purposes, are not of adequate quality for a conference and we did not consider them. 

There are other accessibility challenges that Gather, like all interactional spaces, presents. Proximity chat platforms aren’t the most accessible to blind participants, and people unfamiliar with 2D game navigation can find them difficult too. For the latter, we gave tours in advance to people who expressed reservations about the platform, and had conference volunteers available to provide one-on-one tech help where needed. We did not especially tackle the former challenge except by providing feedback to Gather’s staff that high contrast/screenreader friendly modes would be a good idea.

Labour costs
LingComm21 was organised by a committee of four people with overstretched full-time work commitments, so we knew that the only way the management of the conference would happen coherently was to have a paid conference manager. Our conference manager Liz McCullough spent around 200 hours on the conference, including 150 hours before the event, 30 hours during LingComm21 and 20 hours of post-conference admin (including helping to write these posts). 

We also had 11 volunteers who attended an hour of training before the event and helped people for 3-4 hour-long shifts during the conference, who were compensated with free registration and a certificate of volunteering. Our committee members also volunteered hours of their time to review submissions, build the schedule, and manage sessions during the conference. Many members of the lingcomm community also spend time preparing and moderating sessions, and running meetups on the final day of the conference.

A different organizational committee with a different ratio of time versus money available could have opted for a different balance of paid versus volunteer hours, although note that we’d suggest estimating for approximately twice as many volunteers or volunteer hours compared to paid hours, as volunteers (completely understandably) are more likely to have scheduling complications and last-minute emergencies.

Revenue
Our commitment was to make LingComm21 as fiscally accessible as possible. To that end we kept registration fees very low, and offered free registration upon request (no documentation needed, as we saw no reason to add friction for people asking for assistance). Ticket prices were: $10USD for student/under-employed registration, $20USD for standard registration, and $30USD for standard registration plus subsidising student registration costs. We also had a $100USD personal sponsor registration option, and these sponsors were thanked on our website. We had 192 people register for the conference, with around 70 at each of the two lowest tiers, 39 people who subsidised student costs, and another 13 wonderful personal sponsors. We made it clear that registration costs were being used to cover live captioning for event accessibility. 

Revenue additionally came from sponsors and grants. We received a £300GBP grant from the Linguistics Association of Great Britain (LAGB) and commercial sponsorship from Wordnik. LingComm21 was able to incur upfront costs including staffing and caption reservation before any revenue was available because it was underwritten by Lingthusiasm, as part of our commitment to building lingcomm capacity. Lingthusiasm was capable of sponsoring the conference because of the support of our patrons on Patreon — in 2020, we ran a campaign for new patrons to help us give out more LingComm Grants, and many of them remained patrons afterwards, which enabled us to underwrite this conference, i.e., commit funds to the fixed costs of this conference before we knew whether anyone would pay to attend it. 

Balancing budgets
We expected that Lingthusiasm would underwrite the bulk of the cost of the conference, and indeed this was the case, falling around $5000USD short of running a cost neutral event in 2021. However, we believe that a future LingComm conference could be run in a revenue-neutral way based on two factors. 

First, we underestimated how many people would be interested in registering for the conference. Optimistically, we had hoped we might have up to 100 participants, and instead almost 200 people registered. Although the number of participants increased the expense of our virtual space (Gather charges per user, and we had to redesign the space at the last minute to fit more people), the majority of our costs (captioning and admin support) were fixed costs, which divided by 200 registrants rather than 100 made the necessary revenue per participant much smaller. 

Second, we underestimated how much registrants would be willing to pay for the conference. Many more people selected the subsidize-a-student and personal sponsorships tiers than we had anticipated. We think that with a more expensive fee for standard registration and further encouragement for those with external funding or comfortable employment situations to choose a higher tier, we could have kept the registration fee for students and under-employed individuals low while increasing the rates charged to people who can afford them. Participant feedback also overwhelmingly noted that people would have been happy to pay more to support the event. 

We did not do a very active job of soliciting sponsorship, which will become a more viable option as the conference becomes more established. That said, soliciting more sponsorship would have taken additional hours on the part of the organizers. 

We do not regret underwriting the conference in its inaugural year when we had no baseline from which to create revenue projections and when, due to a global pandemic, we were not spending much on travel. However, for the sustainability of this conference in future years, it is also useful to have a pathway to making it revenue-neutral. Part of doing so is noting that online conferences, while far more affordable than physical ones, are not free to run — even the most affordable online events have some fixed costs. Figuring out those costs before an event is important, as is communicating these costs clearly to participants so that they understand why registration fees are being charged. 

We feel very fortunate to be part of a community of people who care about lingcomm.

Part of a series called LingComm21: a case study in making online conferences more social. Stay tuned for the following posts during upcoming weeks, or subscribe to Gretchen’s newsletter to get the full list of posts sent to you once they’re all out

  1. Why virtual conferences are antisocial (but they don’t have to be)
  2. Designing online conferences for building community
  3. Scheduling online conferences for building community
  4. Hosting online conferences for building community
  5. Budgeting online conferences or events
  6. Planning accessible online conferences

Hosting online conferences for building community: The case of #LingComm21

We hosted LingComm21 on a platform called Gather, in which each user chooses an avatar and navigates it around a two-dimensional space. When an avatar approaches other avatars in the space, the users are automatically connected to one another via video, audio, and text chat. We built a custom Gather space for the conference that featured both session rooms (for scheduled programming) and social rooms (for informal conversations and meetups), as well as a large lobby area, also with informal seating, which connected all the rooms.

Many attendees credited our choice of the Gather platform for their positive experience with the conference, and some have subsequently gone on to use Gather for other events. While there were other factors that contributed to the success of the conference, in particular scheduling that provided lots of opportunities for interaction with varying small groups, using a proximity-chat platform like Gather was certainly an important factor in giving conference attendees choice of who and what to engage with. (Other proximity chat platforms exist and could possibly be adapted along similar design principles.)

The Gather space we used for LingComm21 was highly customized in a variety of ways in light of our goal to foster the repeated, spontaneous interactions that are the groundwork for building friendship. Below we detail some of the design choices that we think influenced attendees’ positive impressions.

Make the space as user-friendly as possible.
Some attendees were understandably concerned about the mental effort involved in figuring out a new online platform. To minimize this source of stress, we did everything we could think of to make the space easy to navigate. All rooms were accessible directly from, and only from, the main lobby area. A pathway to each session room was clearly marked on the floor using subway-style colored lines. Conference volunteers were stationed at virtual “information desks” during the orientation hour at the beginning of the conference and subsequently throughout the week as a resource for anyone needing extra assistance. For attendees who expressed being especially nervous about the tech, we gave them personal tours or one-on-one help either before the conference or on arrival through volunteers.

information desks in the LingComm21 Gather space: two tables in the lobby area

Keep people within the space as much as possible.
We wanted attending this event to be as simple as walking into a conference center and being handed a paper program, rather than regularly leaving the conference platform to check on an informational email, to view a separate video feed, and so on. In addition to being frustrating technologically, frequent program-surfing would increase the number of potential distractions each attendee might face. Thus, as much as possible, we embedded things within Gather, including the programming schedule, the editable list of meetups, and video feeds of larger panel sessions. The physicality of the schedule, meetups, and intros documents also gave people an object of joint attention to use as an excuse to move around the space and interact with fellow attendees.

schedule, meetups, and intros documents in the LingComm21 Gather space: kiosks arranged in a triangle, each of which links to an electronic document, with a plant in the center

Ultimately, due to technological limitations, we did have to send people outside of Gather for live captions, and partway through the conference we ended up making live video feeds available separately, as some users were experiencing difficulty viewing them within Gather. We hope that better accessibility features are something that might change about the platform in the future.

Make it easy to have small conversations.
In a real physical space, people generally expect only to interact with a small group of people at a time. We did our best to recreate this social convention within Gather by designating each seating area (and each poster area) as a “private space,” in which people connect via video, audio, and text chat only with others within the “private space,” not those located outside it. By doing this, we effectively capped the number of people in a conversation to the number of avatars that could fit within a given private space, which had the double advantage of ensuring both that conversations didn’t grow so large that most people became passive listeners and that the number of simultaneous video/audio feeds wouldn’t strain people’s computers. We marked every private space with area rugs or colored rectangles on the carpet so that it was clear at a glance who was inside and outside a given private space. In session rooms, we explicitly labeled some private spaces as “text only” to welcome those who preferred that modality of communication, and provided guidance in briefings that in those spaces users should expect to use Gather’s text chat feature to talk.

To help with the awkwardness of starting conversations with strangers across various formats, we also labelled some of the small conversational spaces as places where people could go when they were explicitly interested in meeting someone new. 

"meet someone new" areas in the LingComm21 Gather space: sets of four chairs arranged around small tables, surrounded by text reading "meet someone new using voice chat" or "meet someone new using text chat"

Give people a place to go when they’re away.
One occasional disadvantage of the Gather platform is that someone can be “in” Gather when they’re not actually paying attention to it; perhaps they’re doing something in a different browser tab, or they’ve just left their computer for a break. This can cause confusion for others who approach their avatar and get no response. Abandoned avatars can also cause “traffic jams” in a sufficiently crowded space. We embraced this traffic metaphor and created an “avatar parking area” (complete with decorative traffic cones) that we encouraged people to leave their avatars in, either during breaks or before exiting out of Gather for the day. Unlike the other non-programming spaces, the avatar parking lot didn’t contain any other interactive or interesting objects — while it would have been tempting to put (say) a livestream of a traffic cam, we didn’t want to encourage people who were actively at their computers to have any reason for hanging out in the parking lot. Participants readily understood the metaphor of the parking lot and used it frequently, and we did not encounter any “zombie” avatars elsewhere in the space. 

avatar parking area in the LingComm21 Gather space: a label reading "Park your avatar here while you're away" above the door, with outlines for avatar-sized parking spaces on the floor and decorative traffic cones in the corners of the room

Humanize the general setting.
Humans respond to aesthetics. Hotels (even cheap ones!) invest in things like art and flowers and landscaping because humans don’t like living in blank boxes. This is no less true in virtual spaces. There were two general design principles we used here. First, we used design assets of a roughly appropriate size and scale for the 32×32 pixel avatars — for example, we want people to use chair and couch images as a cue to “sit” with groups of people there, so chairs need to be of an appropriate size for the avatars to visually sit in them. Second, we added assorted “extras” that contributed to a conference ambience, such as small potted plants on tables, large potted plants and water coolers around the edges of the rooms, and scattered to-go coffee cups everywhere. Both of these tasks were facilitated by Gather’s inbuilt asset library and its interoperability with other pixel art tilesets that people have released online.

Use design to set the mood for various kinds of social spaces.
We had three general-purpose social spaces: a cafe, a boardwalk, and a rooftop bar, which were created as modifications of existing Gather room assets (and thus had much nicer art than we could have made on our own, especially the pixel art skyline at the rooftop bar). The cafe and the bar naturally suggested themselves as social spaces for before and after the conference programming, respectively, while the boardwalk was an “outdoor” space that could be enjoyed at any time.

Further, we had two additional social spaces restricted (by labelling and courtesy, not by technological limitations) to smaller sub-groups within the conference: the Green Room and the Student-ish Lounge. The Green Room (a custom space that was indeed green) was labelled for panelists and moderators only, who were encouraged to meet there 15 minutes before the start of their panels in order to coordinate about technical details and get to know each other a bit, and also to use the Green Room space if at any other point they needed a break from the general conference hubbub. The Student-ish Lounge (created from Gather’s “diner” layout) was labelled for students and others who are socially like students (e.g., recent grads and other junior people), and contained some interactive virtual board games, for more junior attendees to have a low-pressure space away from the general conference hubbub. The organizing committee spent quite a lot of time going in and out of the Green Room during the conference and deliberately did not enter the Student-ish Lounge; conference volunteers, who were largely students, were instructed to keep an eye on it and let us know if there was anything we needed to know about there. We created these spaces because we wanted to recognize that power differentials at conferences are real, even in virtual space, and it can be valuable to have an option where you can remain part of the conference but not risk running into your lingcomm heroes who got you into linguistics (or conversely, where you’re not risking getting surrounded by fans). Although it can be tempting to use technology to limit who can access a particular space, in a relatively high-trust environment like people who were registered for a conference and with a relatively low-stakes outcome of social awkwardness if people did enter a room they weren’t supposed to, we deemed it worth demonstrating to participants that we trusted them by opting for the less friction-filled option. (Had there been any problems here, they would have been dealt with as a code of conduct issue, i.e., by organizers talking to the parties involved.) 

Give people pretexts to spend time in the virtual space.
While the session rooms themselves were mostly business, we added fun interactive details to the social spaces. The unexpected hit of the conference was the “magical duck” that dispensed emojis of snacks or dinosaurs, a fork of a Glitch bot by Alison Stevens that was inspired by an “emoji bar” created by Em Lazer-Walker. There were other Glitch bots, as well, largely inspired by the Gather Glitch bots by Janelle Shane, as well as Gather’s default interactive piano and whiteboard objects. Each day we added a new interactive experience or two so there was always something to discover. These “Easter eggs” motivated people to join early or wander around the space to find things, and sometimes served as convenient conversation starters (“have you gotten a snack from the duck yet?”). For one day, the cafe space was transformed into a “cat cafe” that included several images of cats (including a foreground image so that people could sit “under” the cat) as well as a livestream of kittens, which some people “stood around” watching for some time, thus allowing others to run into them organically. There are many great nature livestreams available on YouTube, and we think that they can be a great solution to the “cheese plate problem” of giving people objects of recurring interestingness to interact around. 

It’s about the space, but it’s not about the space.
Could we have made the Gather space more aesthetically attractive and with even more interactive Easter eggs? Yes. Would doing so have actually made more people use it, or the existing people gain more utility out of it? Probably not. It’s easy to attribute the success of the conference to the Gather space itself, but we’ve seen beautifully designed Gather spaces languish unused when more attention was paid to spatial design rather than temporal design — i.e., providing more and more elaborate rooms and pixel art rather than coming up with events and occasions and programming as a reason for people to keep coming back. It doesn’t matter how beautiful the architecture is in a community center without a calendar of events that provides reasons for people to drop by the community center regularly (and even a shabby building can be much beloved if it hosts warm and welcoming events), and the same thing is true in a virtual space. 

Building elaborate spaces in Gather can be a fun hobby — it appeals to the same parts of our brains that like Lego and Minecraft and The Sims. But if you want other people to actually use your space beyond the initial tour, you need to know where to cut yourself off on the architecture side and direct the bulk of your energy to the people side, prioritizing ease of navigation over esoteric Easter eggs, and especially focusing on events and activities that give people a reason to come and get them actually interacting with each other. Yes, it’s scarier to reach out and invite real living people on the other side of the screen than it is to fuss with virtual furniture solo. But anyone who’s worth being friends with won’t mind if you invite them to your home when it’s still a bit messy, and your digital space doesn’t have to be pixel-perfect, either. In fact, for smaller groups we’ve found that embracing chaos and inviting your guests to help decorate the space with you can be a fun activity! 

Finally, we want to note that while these advice posts may look long and exhaustive, many of them are pulling from things that other conferences, both physical and virtual, are already doing well. We know that many people already recognize that the social parts of conferences are important — hence the oft-repeated advice to go to conferences in the first place rather than just staying home and writing up your ideas. Meeting people at conferences is a way of finding out about news through informal channels, getting to know potential future collaborators, or having a gut-check about whether things in your existing situation are normal. It’s just that the practical implementation of conferences as social spaces can be a huge challenge when it’s something most people are trained to leave to hallways and afterthoughts. 

We hope that putting all of these social design suggestions in one place can help other conference and event organizers take the social function of conferences seriously, and develop concrete ways of making conferences more effective, especially for fostering connections between newcomers, who are the future of any human space. If nothing else, we’d encourage conference attendees to pay attention when conferences are doing a good job at social facilitation, thank their organizers in such cases, and borrow things that previous conferences have done well on the social side when organizing their own conferences.

Part of a series called LingComm21: a case study in making online conferences more social. Stay tuned for the following posts during upcoming weeks, or subscribe to Gretchen’s newsletter to get the full list of posts sent to you once they’re all out

  1. Why virtual conferences are antisocial (but they don’t have to be)
  2. Designing online conferences for building community
  3. Scheduling online conferences for building community
  4. Hosting online conferences for building community
  5. Budgeting online conferences or events
  6. Planning accessible online conferences

Scheduling online conferences for building community: The case of #LingComm21

The goal of LingComm21 was to bring together a community of people who were interested in doing linguistics communication, so we were intentional from the very beginning about the community-building aspects of the program. We hosted LingComm21 on a platform called Gather, in which users communicate by navigating video game-like avatars around a two-dimensional custom map. When an avatar approaches other avatars in the space, the users are automatically connected to one another via video, audio, and text chat. 

Many attendees credited our choice of the Gather platform for their positive experience with the conference. While Gather certainly contributed to this success, and we’ll explain how we set up our custom space there in the next post, there were less obvious aspects of the event’s structure that were also crucial. Below we detail some of the underlying scheduling priorities that we believe resulted in an enjoyable, engaging online experience.

Define a bounded period of time for the conference.
The conference took place over 4 days, with 4 hours of programming per day and the entire last day devoted to meetups. We had a volunteer training the day before conference programming started, and in retrospect it might have also been a good idea to schedule a conference attendee icebreaker event the day before, as a gentle ramp up to the conference like how the meetup day was a gentle ramp down. Generally speaking, we were aiming to create a “magic circle” which requires a defined opening and closing. Inspired by the excellent advice about openings and closings in The Art of Gathering by Priya Parker, we featured opening and closing panels begun by brief opening and closing remarks, rather than starting and closing with logistics. 

Make the conference a shared experience, not solo homework.
When people don’t have to travel for a conference, there’s sometimes a temptation to spread conference events across an entire month, or to assign conference homework of watching talks in advance, which makes it difficult for people to have a shared joint conference experience as an event that’s bounded in time. Pre-recorded talks and/or allowing talks to remain available after the conference may make sense for some conferences, but we’ve observed that watching talks as homework plus a live Q&A part often leads to live Q&A audiences who haven’t watched the talks, making presenters either deliver a short recap of the talk or else suffer in silence, and in any case not accomplishing our goals for this conference of encouraging participants to interact. Instead, we debuted each talk as a live presentation with live breakout groups and Q&A, and recorded only the larger sessions, which were available to attendees for a week following the conference (but not forever, to encourage more candid conversation). Some attendees who were in less convenient time zones reported watching the recorded talks before the next day of programming began, so this limited amount of time shifting helped us give attendees a shared conference experience without creating homework.

Make it possible for people to fit the event into their lives.
As a new, fully virtual conference, we expected that it would be difficult to convince people to set aside entire days for this event. Additionally, we were hoping for synchronous participation from people in many areas of the world, and we knew that longer days would make this more challenging. We settled on a schedule of 4 days of 4-hour conference blocks, beginning each day at 20:00 UTC. Ultimately, attendees in many locations were able to make the timing work, although this choice worked against us in one major way: many people were trying to fit the conference in around a full workday, and felt overwhelmed just a couple of days in. We received approximately the same amount of feedback from attendees who disliked that the conference was during their workday as from attendees who disliked that the conference was not during their workday, so for time zone math purposes we count a time tolerable if it’s during reasonable waking hours, as people clearly have a range of preferences on this issue. For future reference, we note that it is significantly easier to make a conference work for the trifecta of Australia/New Zealand, North America, and Europe when it is Australian Summer Time/Europe Winter (non-Daylight Saving) Time. 

Build interpersonal interaction into the experience.
We had noticed that in many virtual conference environments, socializing and networking were, at best, easily avoidable add-ons. Because facilitating communication among this community was one of our primary goals, we instead made these central parts of the conference-going experience. In most of our session rooms, people “sat” at tables with others, rather than in rows of chairs. Sessions in these rooms included 20 minutes of panel discussion, 10 minutes of small breakout discussions, where people chatted with others at their tables about questions posed by the session’s panelists, and then 10 minutes of questions to the panel again. These breakout tables were self-chosen, rather than randomly assigned, meaning that people could choose to sit at a table where they already knew someone, choose to keep mixing it up, or even move to a different table midway through a session. Our only session room without tables was set up as a “fishbowl,” a style of interactive session that breaks down the dichotomy between audience member and panelist.

Foster conversation…
Within the 4-hour conference schedule, we provided numerous opportunities for people to engage in conversation.  We used time cues, including moving people between sessions every hour and leaving 15 minutes between scheduled events. Additionally, we directly encouraged use of the space before/after scheduled hours, particularly if the official conference programming was at an inconvenient time locally. Many participants commented favorably on the interactivity and several even wanted more, suggesting a half hour coffee break in the middle of the 4-hour programming block. These efforts were part of the overarching goal of building repeated spontaneous interactions over time, which sociologists have identified as the building blocks of friendship. 

…but keep conversations small.
Psychologist Robin Dunbar finds that in the physical world, groups of people tend to split into smaller sub-conversations when they get to be above 4 participants large, but most videochat platforms force everyone to remain part of a single conversational thread regardless of group size. Deliberate planning is necessary to create “normal” conversations of 2-5 people online rather than an endless succession of larger, meeting-style conversations with a few talkers and a lot of listeners. We used physical cues in Gather, including seating people at small tables for talks, keeping the viewing area for each poster small, and building social spaces with areas for small groups, to encourage people to spread out and form both structured and spontaneous small conversations. We also created programming that would simulate small-group conversation by not accepting solo talks — rather, we had prospective participants describe their interests and experience and grouped them together into small panels based on emergent themes, so that even people who were not previously well-networked could get to know others with shared interests.  

Make starting conversations easy.
Before and during the conference, many people expressed concern about entering into conversations, especially if others were already chatting together. While the scheduling helped with some of this, such as the explicit invitation to sit in a “chair” to be part of a breakout group or to go up to poster presenters during the poster session, the informal parts were more tricky. Based on a suggestion from a Gather staffer, we promoted a conference-wide convention of using the “raise hand” reaction (similar to a wave) when approaching people to see if they were open to someone joining them. We explicitly instructed volunteers to go up to new arrivals and proactively greet them on entrance to the space on the first day, and the organizers also tried to keep an eye on the space and social media and introduce people to each other where possible. We also created several seating areas in the lobby explicitly labeled as “MEET SOMEONE NEW” for those who felt more comfortable being approached than doing the approaching, which some attendees reported using. Finally, in making these suggestions to attendees, we acknowledged that these concerns were valid and widely shared, which itself may have helped some people feel more at ease.

In future, we might also consider other social lubricants that adapt approaches we’ve seen work well at physical conferences, such as suggesting that people could put a smile emoji in their names to signal when they’re open to striking up a conversation with someone new.

Give both speakers and attendees the feeling of an audience.
A good public speaker can hold an audience’s attention solo — but they do so by reading the vibe of the room and creating opportunities to get feedback from the audience, such as laughter, gasps, and applause. In an online setting, even the best solo speaker can’t tell how an invisible audience is reacting. So don’t make them give a talk by guesswork — give them some kind of audience! An experienced streamer can keep an eye on a parallel text chat channel, and science comedian Kasha Patel recruited a few audience members to provide a volunteer “laugh track” at a conference one of us attended, but the easiest thing to do is just not have solo talks. If you instead host conversations between two people, or small panels of three speakers and a moderator, then the fellow panelists can be each other’s audience proxy. Audience members also benefit from feeling like they’re part of a communal experience, which we accomplished by livestreaming the talks within Gather and having people watch them while connected to participants at their table via video, audio, and text chat.

Designate roles for attendees.
Many people find it easier to interact with others if they have some kind of existing role, pretext, or shared knowledge to go into the interaction with. In addition to giving people programming roles by being on panels or presenting posters, we also encouraged attendees to organize meetups on the final day, and designated official conference volunteers to greet people at the entry point of the virtual space on the first day and be a first line of contact for basic questions (which they could surface up to the organizing committee as needed). It was more welcoming to attendees to have someone greet them as they entered on the first day, and the volunteers bonded with each other by meeting at the orientation on the day before.

In future, we’d also consider some sort of day-before small group mixer for attendees with an activity to help them meet a few people, such as attendee bingo/a trivia night/small group tours of the conference space with the volunteers as facilitators, analogous to a pre-conference dinner.

Give people some control over the event.
The entire fourth day of our conference was designated as “Meetup Day,” with less formal programming suggested and run by participants on a schedule that we provided and encouraged participants to add to throughout the conference. Meetups included continuations of some conference sessions as well as casual chats about lingcomm on particular platforms, a discussion of possibilities for the next iteration of the conference, and some purely social gatherings for games and crafting. Not everyone attended programming on this day, but at least a third of active participants from previous days did, and many of them commented on it favorably. In addition to helping continue conversations and community-building more generally, Meetup Day provided a gentler re-entry into regular life from the more intense conference experience of the previous three days.

Give people social license to post about the event.
We wanted not only to foster an event for people who communicate about linguistics to broader audiences to talk with each other, but also to help support cross-pollination for people who are fans of the lingcomm materials that many of our attendees already produce. To that end, we also hosted LingFest, a fringe-festival-like series of online events about linguistics (such as podcast liveshows, quiz bowls, Twitch livestreams, etc.), each independently organized but with a centralized website listing all of them where interested people could sign up for notifications about the whole series. LingFest ran during the week following the LingComm conference, which allowed people who’d heard of a project during the conference to subsequently check it out in a streamlined way.

Crucially, LingFest disambiguated who the conference itself was for. Since many of the people we were aiming the conference for have a public presence with fans, rather than having to awkwardly say “no, please don’t come,” both organizers and attendees could point fans to LingFest instead and say “here, this is the event you want.” In this structure, we were inspired by how fan/industry hybrid conferences like VidCon have both fan and creator tracks. 

Many of these schedule-related elements could be designed for regardless of which online platform is used. For example, one could set up an online conference that takes place mostly in a single Zoom call with many breakout rooms, some for talks and some as social spaces, which attendees could move themselves into and out of — thus giving attendees a chance to run into each other in the main “lobby” of the Zoom meeting. One could also organize parallel text-based social spaces on internet platforms that have already proved that they can be social, such as Slack, Discord, or a conference hashtag. At minimum, any conference can manage basic scheduling features like building in breaks, keeping days a reasonable length, considering timezones, and encouraging audience members to use the parallel text chat to a video talk for virtual applause and lightweight interaction. 

In our case, we decided that the community building experience we were aiming for with LingComm21 would be best served by using a relatively new platform called Gather. In our next post, we’ll focus more on Gather itself, and the design decisions we made in our virtual space to encourage a positive virtual conference experience.

Part of a series called LingComm21: a case study in making online conferences more social. Stay tuned for the following posts during upcoming weeks, or subscribe to Gretchen’s newsletter to get the full list of posts sent to you once they’re all out

  1. Why virtual conferences are antisocial (but they don’t have to be)
  2. Designing online conferences for building community
  3. Scheduling online conferences for building community
  4. Hosting online conferences for building community
  5. Budgeting online conferences or events
  6. Planning accessible online conferences