Linguistics at Senso
Linguistics might not be the first thing you think of when it comes to safeguarding, but it holds incredible potential for protecting users online.
As part of our commitment to always improving our technology, we’re putting considerable investment into our linguistic capabilities here at Senso. This means we’re able to go beyond the typical safeguarding approaches and work towards developing a much more effective and accurate online safety tool.
But what is linguistics? And how exactly does it fit into safeguarding? In this article we explain how and why we’re harnessing the power of language to improve our safeguarding offering.
What is Linguistics?
Linguistics is the scientific study of language and how it works. It helps us understand all of the different aspects of language, including how words are formed, how they’re put together to create sentences, and how meaning is made. Almost everything we do in life relies on language: having a conversation, reading an email, writing a note, using our phones… language is everywhere, and the way we use it can have an incredible impact.
Linguists are trained to analyse how language is used in different contexts, as well as how different languages work. They do this by uncovering the patterns and rules within the grammar, sounds, and vocabulary of a language, and by analysing large amounts of language data. Essentially, linguistics helps us gain insights into how we use language, why language changes over time, and how language influences our thoughts and behaviours. In the simplest terms, linguistics is a kind of detective work into the mysteries and mechanisms of human language.
What does this have to do with Safeguarding?
There are a number of ways users engage with computers and other devices, but the main way is through language. This could be sending and receiving messages, searching online for information, consuming media like news stories, or navigating to websites through URLs. All of these present an opportunity for us to monitor and protect young people online by observing the kinds of language they are either using or being exposed to. Combined with image analysis, accurate linguistic monitoring of school devices is the best chance we have of preventing serious safeguarding risks and keeping children safe online.
Undertaking this kind of task isn’t easy. As we all know, words can have several different meanings. What does the word blow mean? We can blow out a candle, or the wind can blow, but when something else blows that could mean it’s a major disappointment. Then there’s blowing up – if someone blows up it could mean that they become very angry, or, these days, that they’ve gone viral on social media! If we blow something up it could mean to fill it with air, or it could mean to destroy it in an explosion. If something blows over, it could mean that it calms down or that it is overwhelmed by a gust of wind. Blowing our nose is quite different from blowing a trumpet, blowing an opportunity, or blowing a load of cash. Blow, as a noun, can also be a slang term for an illicit drug. In fact, there are more than 80 senses of the word blow. And in a safeguarding context, as with any other instance of language, we need to know which sense is being used. This is where linguistics is key.
We also need to understand exactly what kind of language children and young people are using so that can spot the warning signs of a safeguarding concern. This isn’t something we can just guess or rely on our intuition for; we need to be selecting, processing, and analysing large amounts of appropriate language data. This requires expertise in linguistics, data science, and safeguarding.
Introducing Emma
Emma is our in-house linguist and Head of Safeguarding here at Senso, which makes her the go-to person for all things digital safety and linguistics. As well as helping our customers meet the highest level of safeguarding standards and stay compliant with statutory guidance, Emma also brings her expertise in linguistics and lexicography to help develop new methodologies for processing safeguarding data.
Emma’s day-to-day work involves reviewing and testing our existing processes and resources, consulting with external safeguarding and IT experts to develop new ways forward, and running linguistic analysis on real-time and historical Senso violations to reliably reduce false positive rates. Within two months of Emma joining Senso, our false positive rate dropped by 24% on Microsoft Teams! Since then, Emma has continued to refine our keyword libraries and bring false positive rates down even more. We’re excited to release updated progress figures in the coming months.
Emma came to us from an academic background, having completed a BA, MRes and PhD in Corpus Linguistics, which is the study of large amounts of language data using specialist software and analytical techniques. Prior to her current role at Senso, she was a Lecturer in Corpus-Based Digital Humanities at the University of Wolverhampton, where she was also the designated Safeguarding and Wellbeing Officer for the internationally renowned Research Group in Computational Linguistics. With more than a decade’s experience collaborating with Natural Language Processing and AI experts, and making important contributions to the fields of Computational and Corpus Linguistics, Emma is perfectly placed to lead our efforts on improving Senso’s linguistic capabilities.
“Coming from an educational background myself, I understand the pressures that teachers and institutions are under to keep their students safe in a constantly changing digital world, and with little available time and resources. I could see that Senso was doing really exciting and meaningful work in safeguarding, so it was an easy decision for me to leave my academic career and join the team on this journey.”
Our Fantastic Knowledge Transfer Partnership
As well as pursuing our own in-house linguistic projects, we are also part of an innovative Knowledge Transfer Partnership with linguists at Birmingham City University. The two-year project seeks to use specialist linguistic techniques to revolutionise the way Senso detects and handles user violations, meaning lower false-positive rates, higher quality violation data, and more sophisticated methods of protecting vulnerable users online.
Over the next two years, by continuing to invest in linguistic expertise and through our Knowledge Transfer Partnership project, Senso is poised to make significant strides in the development of novel violation detection methods. By leveraging state-of-the-art corpus linguistic and machine-learning techniques, we are committed to staying at the forefront of technological advancements in the field.
This forward-thinking approach ensures that our safeguarding measures evolve and adapt to the ever-changing world of online harms.
“By leading on innovation and developing new technologies, we have the potential to transform the safeguarding landscape forever. Not only are we passionate about making the internet a safer place for children, but we’re also committed to supporting and empowering our schools and safeguarding leads who shoulder a huge responsibility.”
Follow our Journey
If you’re interested in hearing more about the latest developments at Senso, please don’t hesitate to get in touch. We’d love to hear from you!