Researchers in the School of English have developed a new open access platform comprising of millions of tweets relating to Covid-19, in order to capture public opinion, moderate misinformation and provide a resource to other researchers.
Covid conversations
Andrew Kehoe,听Matt Gee,听Mark McGlashan,听Tatiana Tkacukova,听,听补苍诲听Robert Lawson听have created听, otherwise known as TRAC:COVID.
The team were inspired to develop this resource due to the volume of social media updates posted on Twitter from a broad cross section of people.
鈥淭witter has seen journalists, politicians, doctors, nurses, and members of the general public share their thoughts about lockdown, furlough, quarantine, testing, masks, vaccination, social distancing, recovery, and, in the worst cases, loss and bereavement,鈥� Robert says.
鈥淲e wanted to make sense of the kinds of Covid-related conversations people have been having 鈥� what worried them, how they viewed lockdown 鈥� as well as charting how these conversations shifted over time.鈥�
TRAC:COVID will also, the team hopes, serve as an important tool in addressing misinformation, which has been identified as a key concern ever since Covid-19 became prominent.
For example, studies have found that tweets with completely false claims spread faster than those with partially false claims and unverified personal Twitter accounts feature the highest rate of Covid-19 misinformation.
鈥淎lthough these studies have been essential in tackling the problem of misinformation, it鈥檚 not always easy for people to explore the underlying datasets,鈥� Robert says.
鈥淲e wanted to develop something that was publicly accessible to researchers, academics, and members of the general public 鈥� hence the creation of TRAC:COVID.鈥�
Exploring the data
The TRAC:COVID dashboard enables the analysis of over 84 million UK tweets which contain words and hashtags related to the pandemic.听
The dashboard is built on methods from a discipline known as corpus linguistics, which helps people quickly analyse millions of words. Corpus linguistic methods have recently been used to analyse healthcare communications, including听.
The dataset currently covers January 2020 to April 2021, and there are hopes to extend this in the future.
鈥淭hrough the dashboard, users can chart how language use changed during the pandemic and how particular words have acquired new meanings, to when certain words stop being used altogether, all without requiring specialist knowledge or language analysis skills,鈥� Robert says.
Understanding pandemic perspectives
The team hopes that TRAC:COVID will also prove beneficial in tackling ongoing issues related to the pandemic, such as engagement with the vaccination programme.
鈥淪tudies have shown vaccine hesitancy is fuelled by misinformation, particularly on Twitter,鈥� explains Mark McGlashan, the author of the team鈥檚 case study about COVID-19 misinformation.
鈥淯sing TRAC:COVID, we 鈥� and the public 鈥� have a good idea of the scale and diversity of 鈥榓nti-vax鈥� stances on Twitter.鈥�
The team also charted public reaction to the range of public health measures implemented over the past 18 months, including masks, social distancing, and the vaccination programme.
鈥淧eople on Twitter were generally supportive of these public health measures, in some cases arguing for stronger restrictions to drive down rates of infection quicker鈥�, said Tatiana Tkacukova, who led the public reaction strand of the project.
The team鈥檚 work has already resulted in听听补苍诲听, while parts of the case study on public reactions were submitted to the call for evidence on Initial Learning from the Government鈥檚 Response to the COVID-19 pandemic and, more recently, to the call for evidence by the听All-Party Parliamentary Group.
Findings from the research form part of a newly published report from the House of Commons Committee of Public Accounts titled听Initial lessons from the government鈥檚 response to the COVID-19 pandemic.听听
Written evidence submitted to the Committee found that messaging from the government and public health bodies:听
- Lacked clarity about who the messages were directed to听
- Used long sentences with complex vocabulary, grammar and syntax, which made it difficult to understand听
- Used language which was ambiguous听
- Used terminology which was likely to elicit negative reactions听
- Used terms which could have excluded some intended recipients of messages (e.g. using the term 鈥榟ouse鈥� rather than 鈥榟ome鈥�)听
Tatiana added that, "We hope that these reports will contribute to planning the communication strategy on the safety measures required to address the ongoing risks to public health."
鈥淭weets about Covid-19 represent an important cultural artefact,鈥� Robert says. 鈥淲ith our new resource, people will be able to explore an archive of pandemic perspectives, deepening their understanding of what preoccupied UK Twitter users during this time.鈥�
Find out more via the听TRAC:COVID project page.