Lancaster University researchers have helped create a free bilingual online platform to analyse ‘free-text’ survey data.

Led by academics at Cardiff University, and working with partners at National Trust Cymru, Amgueddfa Cymru - Museum Wales, WJEC, National Centre for Learning Welsh and Cadw, the project involving researchers from Lancaster University’s School of Computing and Communications has delivered the FreeTxt | TestunRhydd toolkit, which allows large organisations to quickly respond to opinions from both Welsh and English consumers.

Free-text qualitative comments, which are captured in feedback from surveys and questionnaires, pose a particular challenge to a range of private and public companies and institutions, who may not have the skills and expertise to process and act on these comments with ease.

Project lead Dr Dawn Knight, based at Cardiff University’s School of English, Communication and Philosophy, said: “Surveys and questionnaires are used in many walks of life. They can be targeted at consumers, used in staff development and professional training, in product design and testing, and in various forms of service provision across the public and private sector.

“But making sense of the qualitative, text-based responses that are often included at the end of surveys is often difficult and costly to analyse. FreeTxt provides a quick an easy and accessible way for doing this. Our online platform is already being used by companies, including Cardiff University, Amgueddfa Cymru - Museum Wales and Cadw, allowing them to more easily assess this kind of feedback.

“We hope that developing this free service will allow the views of employees and consumers across both the public and private sectors to be heard and acted upon more quickly. The fact that this platform can assess feedback in both English and Welsh also ensures that companies in Wales are fulfilling their obligations under the Welsh Language (Wales) Measure 2011.”

This tool is the first to fully support the use of Welsh language responses to surveys. Users can enter their data, with the platform then quickly processing the text to provide an easy-to-interpret visualisation of the words, as well as themes and patterns of sentiment that appear most prominently. Analyses can be easily downloaded in different visualisations, from charts to word clouds and tables.

Academics have worked closely with external project partners to co-design, co-construct and test FreeTxt | TestunRhydd to ensure that the resource is fit-for-purpose and fairly and consistently meets the needs of Welsh and English-language responses.

Dr Mo El-Haj, Senior Lecturer in Natural Language Processing (NLP) at Lancaster University’s School of Computing and Communications, along with Lancaster’s FreeTxt team, which includes Professor Paul Rayson, and researchers Dr Nouran Khallaf and Dr Ignatius Ezeani, said: "Lancaster University has played a crucial role in the FreeTxt project, focusing on the technical aspects of NLP. Our team's expertise has been instrumental in advancing the analysis of free-text survey comments, with a particular emphasis on the Welsh language, which deserves more attention in the field of NLP.

"We've applied a range of NLP methods, including sentiment analysis, summarisation, meaning analysis, word cloud generation, and through the understanding of word usage and relationships, to enhance the comprehension of feedback from surveys and questionnaires."

“Our collaboration with key partners has been central to the project's success. By working closely with organisations like National Trust Cymru, Amgueddfa Cymru - Museum Wales, WJEC, National Centre for Learning Welsh, and Cadw, we've ensured that FreeTxt | TestunRhydd meets the needs of both English and Welsh responses, through applying and developing the tools and techniques required for effective processing in each context."

Dr El-Haj added: "FreeTxt | TestunRhydd represents a significant step forward in NLP, particularly in its support for Welsh language survey analysis. We anticipate it will enable organisations to respond more effectively to feedback from diverse audiences."

Rhys James, Insights Officer at Amgueddfa Cymru said: “We were delighted to support such an important project, which champions the Welsh language and brings us closer to what our communities think and feel. Amgueddfa Cymru is committed to making sure that everyone is represented, and the work we do in our insights team is important in making sure everyone is heard.

“The FreeTxt | TestunRhydd tool has helped make that process more cost effective as well as bring added sophistication, ease, and consistency to our research. In turn, this will help us better reflect the voice of our visitors in the decisions we are taking.”

The FreeTxt | TestunRhydd project was funded by the Arts and Humanities Research Council (AHRC). It builds on previous work by Dr Knight, the CorCenCC (National Corpus of Contemporary Welsh), which is a freely accessible collection of multiple language samples, gathered from real-life communication.