Talent Development Awards 2021

Funded by

Dr David Bartram


From cross-sectional multi-level modelling to longitudinal analysis of country-level variables

University of Leicester


To investigate the impact of a country-level variable on an individual-level variable, many researchers would use cross-sectional multi-level modelling. We advocate a longitudinal analysis of time-varying country-level variables as a useful alternative in many situations. The dependent variable can be constructed via aggregation (e.g. averaging) of repeated cross-sectional survey data containing individual responses. There is no need for individual-level controls, because these are very unlikely to be antecedents of the (country-level) independent variable. We can then implement a longitudinal (‘within’/‘fixed-effects’) analysis of country-level variables (including country-level controls), which is more effective in minimising bias from omitted confounders.

Our project will employ a research assistant to conduct ‘scoping research’ on the practices of UK-based quantitative researchers and teachers in the social sciences. We will then hold two engagement events (one focused on research, the other on teaching), where participants will be invited to explore the relative merits of the alternative approaches.

Dr Kathryn Brown


Sensing the Surface: A Computer Vision Analysis of Claude Monet's Water Lily Paintings

Loughborough University


This project combines critical methods of inquiry from the humanities with advanced data science to address a gap in understanding the late works of Impressionist painter, Claude Monet. A machine learning framework will be developed to offer accurate insight into the chronology and dating of the artist's water lily paintings created between 1890 and 1926, to investigate stylistic patterns across works, and to resolve debates about the point at which Monet considered a work to be ‘finished’. The research leads to a larger inclusive design project that will build three dimensional models of the surfaces of Monet’s paintings. This will facilitate a tactile experience of Impressionist painting's complex brushwork and offer new outreach strategies to visually impaired audiences. In addition to unlocking fresh insight into Monet's creative practice, the project will generate innovative quantitative methodologies that can be transferred to projects in other branches of art history and cultural heritage.

Dr Ruijun Bu


Dissecting Systemic Risks in Large Economic Sectors: A Semiparametric Big Data Approach

University of Liverpool


Systemic risks in larger economic sectors, such as financial markets and manufacturing sectors, is a crucial source of economic uncertainty, posing potentially critical threats to key economic issues such as financial stability, economic growth, and social welfare. Sufficient understanding of their dynamics and driving forces is one of the most central and challenging tasks in modern economic studies and risk management. This project aims to propose and investigate a ground-breaking hybrid big-data approach for measuring and tracking the vital dependence structure of the influences of potential economic fundamentals on the performances of individuals in large economic sectors. This hidden and time-varying structure is responsible for a significant portion of sectoral systemic risks and thus holds the key to successful economic risk management for decision and policy makers. The methodology exploits the power of several cutting-edge big-data techniques, which potentially carries fundamental importance and profound implications for both theory and practice.

Dr James Cummings


JUMPSTART: Joining Up Methodological Practices for Studying Texts with Advanced Research Training

Newcastle University


This project seeks to study advanced research training through provision of a Humanities training workshop on advanced methodologies for research with pre-modern text. This joins up training that is often STEM-focussed with the needs of Humanities researchers, but simultaneously studies the form of training itself. This project will jumpstart the development of advanced quantitative and data science skills for Humanities academics. Training will include: an introduction to the fundamental technologies of data science (version control, a general purpose programming language, techniques of data wrangling), standards and methods specific to digital research in the humanities (TEI, TEI Publisher, Text Analysis, HCR and Transkribus) and also introduce advanced methods such as the use of Artificial Intelligence and Machine Learning for Humanities’ research. The skills exchange provided by this workshop will provide training for up to 40 academics as well as building a corpus of re-usable training materials.

Dr Filippo Dionigi


Text-Mining and International Relations: a Methodological Assessment through the Analysis of Islamist Discourse

University of Bristol


International Relations has only recently begun to fully explore the potential of text-mining methodologies. Whereas this methodology is promising, it is also challenging for its reliance on complex computational and programming skills applied to a field of enquiry wherein more conventional research methods have traditionally prevailed. This project, therefore, facilitates a research and training cooperation between an IR scholar and data scientists to develop further text-mining research skills, produce new research, and critically assess the potential of text-mining applied to International Relations. The research will consist of a term-frequency analysis of the speeches of Hezbollah’s secretary-general that will illustrate the shifts in the political identity of this movement across time. During the research process, data scientists will supervise and support the computer-based component of the research and provide ad hoc training in text-mining research techniques and programming.

Dr Lucy Donaldson


Videographic Scholarship: innovation and collaboration in research practice

University of St Andrews


The videographic essay is an emerging mode of scholarship in the disciplines of film, television and media studies, bringing with it new possibilities for research and pedagogy. The objective of this project is to facilitate participation in an international workshop, held at Middlebury College, which provides training in digital techniques which inspire the production of innovative research and teaching. The workshop creates a space for individual exploration and collaborative practice in digital scholarship, preparing participants to produce original research in a pioneering form of digital humanities. Moreover, the workshop promotes leadership in the field through further dissemination of videographic methodologies in research and teaching practice, empowering participants to contribute to the development of a ground-breaking scholarly form. Engagement with the workshop raises the quality of digital research skills responsive to the increasing demands of open access, digital publishing and public engagement with strong potential for cross-disciplinary skills exchange.

Professor Gabriel Egan


Quantitative Methods for Literary and Historical Scholarship -- In Theory and Practice

De Montfort University


Traditionally the study of literary and historical documents has largely relied on techniques of close inspection and reading, and qualitative judgement. New quantitative methods of analysis have recently become possible because computing power has become widely available and large bodies of texts (particularly those from the past) have been published in digital form.

Take-up of these new methods is uneven. Many linguists use computers to count particular features of large bodies of writing, but scholars in other humanities fields may be left behind because they do not know how to get started in using quantitative methods and have not seen examples of the benefits.

This project will run two 2-day workshops -- one in Leicester and one in Leeds -- to develop the skills of scholars wishing to explore quantitative methods for the first time, and those wishing to broaden their knowledge of these methods and what they offer.

Dr Harry Farmer


In Another’s Skin: How Embodiment and Context Affect Gender Stereotypes

University of Greenwich


Recent advances in technology have massively increased both the quality and affordability of Virtual Reality (VR) systems increasing its viability as both a novel form of popular entertainment and a research tool for the social sciences. The proposed project will investigate how embodying a character of a different gender in virtual reality interacts with the virtual setting experienced to challenge or entrench stereotypical views. It will consist of a research study that places male participants within a female avatar in either a science lab or domestic kitchen setting. Participants implicit and explicit associations between gender and STEM subjects will be assessed both before and after the virtual experience allowing for an assessment of whether embodying female characters affects stereotypical attitudes. In addition, the proposal will fund a workshop that will draw together academics and workers in immersive media to explore the opportunities and risks of challenging stereotypes via VR.

Dr Paul Hufe


Inequality and Perceived Unfairness – An Experimental Validation

University of Bristol


We apply to the Talent Development Awards Scheme to fund the piloting of a survey module for the elicitation of fairness preferences and perceptions.

Fairness in economic systems is widely debated in societies across the globe. Yet, evidence on between-country differences in fairness preferences and perceptions is scant. Therefore, we have developed an innovative and low-cost survey instrument to elicit fairness preferences and perceptions in representative population surveys.

In this pilot, we test whether the module uncovers preferences and perceptions that are consistent with the “gold standard” of incentivized experiments. In particular, we will assess whether fairness preferences and perceptions that are elicited in our module differ when respondents’ decisions are incentivized by monetary rewards or their decisions have monetary consequences for real-life people.

The pilot is an important input for a long-term research agenda and builds UK-based capacity for quantitative analyses of (perceived) unfairness in different societies.

Dr Christian Jones


Learner conversations as models of spoken language in second language German

University of Liverpool


The number of students studying modern languages in the UK, in particular German, has been in decline for many years. This project aims to address the problem by developing motivating conversational materials which are authentic to learners of German in the UK and which respond directly to their conversational needs. Although conversational skills are often prioritised by learners, there is a lack of practice materials for the specific requirements of UK HE learners. The materials will be for use by speakers and teachers as an open-access online resource. In addition, a template will be developed to allow teachers to work with their more advanced learners to produce similar materials for their peers who are studying at a lower level.

Dr Thomas Larkin


Mapping American Socio-Commercial Networks & Mobility in Nineteenth-Century China

University of Bristol


My goal is to develop the skills in integrative database construction and advanced quantitative research methods that will enable me to pilot a digital humanities project demonstrating the extent of socio-commercial networking and migration between China and the West in the nineteenth century. This award will equip me with the knowledge, skills, and opportunities to foster cross-institutional networks between scholars at Bristol, Marseille, and Vancouver that will support the project’s development and the construction of a proof of concept necessary to secure AHRC Early Career research funding to mount an online platform. This training and collaboration will guide the creation of a database of foreigners in nineteenth-century China, with an aim to producing a user-friendly digital platform for visualising socio-commercial networks and transnational mobility. Aimed at researchers and students, the platform will both serve as a reference tool and address the need for alternative pedagogical approaches to teaching global history.

Dr Alex Mangold


Artistic Research and Creative Assessment in Modern Languages

Aberystwyth University


An application is made for the employment of a PG research assistant to establish an online resource hub for ‘Artistic Research and Creative Assessment in Modern Languages’ – following on from the impressive contributions and results at our recent 'Artistic Research in Modern Languages’ conference held in April 2021.

The award will enable us to work within the BA's Europe's Futures and Knowledge Frontiers themes and address several of the key issues formulated in Aberystwyth and Bangor's research strategies for 2019-24 (integration, ambition, collaboration, inclusivity, and language use across communicative contexts – as results would be available freely online). The PG research assistant will create a substantial innovative teaching and learning hub for teachers and students of Modern Languages in the UK. This hub will be a first step towards the establishment of a productive and collaborative UK-EU research network for ‘Artistic Research in Modern Languages’.

Dr Ori Ossmy


Integrating Developmental Science and Engineering to Study Natural Human Play

Birkbeck, University of London


Object play is ubiquitous across every age and culture and produces experiences that underlie human behavioural development. Yet, theories of play are limited to artificial laboratory environments. In the era of Covid-19, when lab examinations were kept on hold, researchers could only resort to online studies. Therefore, interdisciplinary exchange of ideas and tools is required to test natural, everyday play and catalyse discovery about human behavioural development.

By combining perspectives and methods from developmental science, computer science, and engineering, this proposal aims to innovate automatic monitoring of children's everyday play at home using "intelligent toys"—3D-printed objects with embedded sensors that capture children's actions. Funds are requested for training the PI to develop this technology. A secondary aim is to establish a cross-site, collaborative society of young scholars who will benefit from integrating developmental science, computer science, and state-of-the-art engineering. All resources will be publicly shared on a dedicated website.

Dr Christos Pliatsikas


Implementing advanced quantitative statistics in linguistic research

University of Reading


All fields of scientific enquiry need to constantly evolve by considering and incorporating recent advances in research methods and statistical interference. This is true of linguistics as it is other sciences. However, keeping up to speed cannot be achieved without regular and comprehensive training, especially for the newer generations of researchers. For these reasons, we propose to hold a series of workshops on the application of advanced statistical methods (Bayesian inference, Generalised Additive Mixed-Models) to research questions in linguistics, which will be offered to researchers in linguistics and related fields. These workshops will be followed up by publications that will act as methodological primers. We aim for these workshops to develop expertise in the field in order to form the basis of comprehensive regular training programmes around the UK and beyond, to equip linguistics researchers with the necessary analytical tools to enable new insights in different subfields of linguistics research.

Dr Paul Rauwolf


Understanding the psychological risk factors for believing political misinformation by merging neural network and multiple regression models

Bangor University


In the last few years, work has begun to diagnose the individual risk factors associated with believing political misinformation. However, many of these factors have been considered in isolation. Since many psychological factors covary, it is vital to understand which factors predict vulnerability to misinformation above and beyond other factors. Historic techniques for evaluating the most important predictors in a dataset have some limitations. Multiple regression is prone to overfitting, reducing the likelihood of replicability. Machine learning techniques (i.e. neural networks) are criticized for their lack of transparency. This work will use state-of-the-art techniques to merge the outputs of neural networks and multiple regressions to better understand the combination of features which best predict vulnerability to misinformation. This project will find a model with the (out-of-sample) predictive power of a neural network and the transparency of a multiple regression model.

Dr Kathrin Thomas


Enhancing quantitative analysis skills in Politics and International Relations using digital delivery

University of Aberdeen


The proposed programme speaks to three core themes of the talent development scheme: (1) development of innovative teaching courses, (2) training in advanced quantitative skills, and (3) support for conferences, workshops and other activities. It aims to develop digital modules for advanced quantitative data analysis in politics and international relations for research students at postgraduate level. Furthermore, to conduct political research using big data, but also to teach big data harvesting beyond the funding period, the proposed programme allows the PI to train in webscraping methods. Lastly, the proposed programme will host a training specifically targeted at women across disciplines to learn coding and apply the learned skills in their research. Outputs include: a full digital course on advanced statistics offered at the University of Aberdeen; a concept paper on developing a digital course on webscraping; and a report on the success of the women's coding training.

Dr Justin van Dijk


Exploring GPU-based analysis for social and geographic applications

University College London (UCL)


This project aims to explore the extension of current computationally intensive quantitative models in the field of social and geographic data science with GPU-accelerated analyses. We will first configure a NVIDIA Tesla V100 GPU within a local computer cluster. After configuration, the usability of GPU-accelerated analytics in the context of three different geographic applications will be investigated through three case studies. The first of these case studies will explore the matching and joining of large textual datasets comprising millions of addresses. The second case study will explore the performance of deep learning algorithms for GeoAI applications. The third case study will look specifically at the opportunities for improving the performance of disease risk prediction using raster data. The final deliverable of the project will be a workshop alongside a freely available, shareable digital resource with detailed instructions on how to implement GPU-based analysis in social and geographic research and teaching.

Mrs Kate Weir


The Conversation Application

University of Westminster


Being able to hold a conversation in a foreign language is a challenging skill - whatever the level of the language learner. The aim of our research is to create a conversation application that generates context- and level-specific stimuli for both parties in a 2-person conversation with each person only having access to their part of the conversation. Language learners will be able to search for appropriate role play/information-gap stimuli by context, language function and level. All stimuli will initially be produced in English. The application will provide vocabulary support in the target language and feedback on information gap exercises. All stimuli/role plays/information-gap exercises will be benchmarked against the Common European Framework of Reference (Languages) so that language learners can practise conversations appropriate to their level.

Dr Kaitlyn Zavaleta


The future is multilingual: Developing language skills and expanding research engagement

De Montfort University


Language skills are crucial for the UK’s economy, inter-cultural relationships, diplomacy, security and research. However, there is a persistent decline in the number of individuals who choose to study languages in higher education. This may be due to a lack of awareness about the importance of language learning, a belief that language learning is too difficult, or a lack of research-informed language teaching. How to overcome this disconnect? We propose a conversation: we produce podcasts aimed at the general public and in turn, conduct primary research to provide insight into the specific barriers to language education. Interviewing experts and professionals, we plan to highlight: 1) the importance of language learning, 2) how to facilitate language learning, and 3) how to conduct language learning research. By distributing the podcasts to the general public and engaging with their views, the Principal Investigators will develop their public engagement, dissemination and digital research skills.

Dr Ying Zheng


An Investigation of the Statistical Alignment of A-Level Mandarin Chinese to the Common European Framework of Reference for Languages (CEFR)

University of Southampton


Learning Mandarin Chinese is strategically important for the current and future generations of UK students (British Council, 2017). Along with the emerging recognition of the importance of grasping a modern foreign language in the post-Brexit era, there is a lack of understanding of how Chinese as a foreign language is aligned to a common benchmark of proficiency and how best to integrate the teaching, learning and assessment of Mandarin into foreign language curricula at post-secondary school level in the UK.

This proposed study will conduct a systematic investigation into the alignment of A-level Mandarin Chinese to the Common European Framework of Reference, using focus group interviews, quantitative standard setting exercises and corpus-based linguistic analyses. This research has the potential to link UK initiatives with the global teaching of modern foreign languages, including Mandarin Chinese, and to help develop shared international understanding of standards and proficiency levels.

Dr Mimi Zou


Computational legal studies: integrating data analytics in legal research and teaching

University of Reading


There is nascent interest among legal scholars and practitioners in relation to applications of new data-driven technologies, particularly artificial intelligence (AI) and machine learning, in the legal sector. In this context, computational legal studies is an emerging interdisciplinary field of research, which is embracing novel uses of computational methods to undertake legal analysis and the investigation of new research questions in legal scholarship. Nevertheless, law has generally lagged behind other social sciences in adopting and adapting computational methods. My project will examine the state of the art in the application of computational research methods in legal scholarship to date and experiment with new approaches in my own research. I plan to use the Award to acquire and strengthen my computational/data analytics skills, establish a network of UK-based computational legal scholars and an online portal to promote new scholarship in this field and the teaching of computational skills to law students.

Dr Roi Zur


How Parties Die? Political Parties Change and Mortality in Western Democracies

University of Essex


The importance of political parties for representative democracy cannot be overestimated. Parties represent citizens’ ideologies in national parliaments, they are the organizations that translate voters’ preferences to policy outputs, and the institutions that govern nations. However, over 70% of parties disappear or change, whether via dissolution, splits or mergers. Thus, understanding the dynamics of party change and mortality, and the strategic incentives related to these changes, have important implications for democratic representation. This project will combine advanced quantitative data science approaches and archival work to build a data set that is suitable for statistical methods usually used in biostatistics and cancer research. The data set will allow us to shed new light on the causes and consequences of parties’ change throughout their lifespan.

Sign up to our email newsletters