The Arts Digital Lab invites you to join us for this celebration of Digital Research in the Faculty of Arts, at which recipients of the 2025 Arts Digital Seed Fund will present on their research.
Monday 9 February, 9.30 am – 12 pm
Meremere 105
Programme:
Digitising and analysing gendered differences in historical shipboard diaries from early British migrants to New Zealand
Sarah van Eyndhoven, Lyndon Fraser and Dorian Ghosh
When British immigrants came to New Zealand’s shores throughout the nineteenth and early twentieth centuries, they faced a long, arduous journey over sea of three months or more. This journey was often documented in shipboard diaries – small, single-sided booklets that could fit in the palm of a hand. These documents provide a unique window into experiences of early international travel, and the interactions between people who would form the new colonial society in New Zealand. In addition, they offer an exciting glimpse into the earliest language use of the settlers who formed the New Zealand English dialect. The Canterbury Museum has a rich collection of shipboard diaries, yet until recently they have existed primarily in manuscript form.
Utilising the capabilities of the historical transcription software Transkribus (Kahle et al. 2017) this project sought to digitise fifty of these diaries, creating training data from 100 manually-transcribed pages to generate a custom-built Handwritten Text Recognition (HTR) model. The HTR struggled with the wide variability in handwriting (error rate 15.78%), but output improved significantly for female-authored diaries. This supported visual observations made during hand-transcription – women’s writing style was more careful, extensive and emotive than men. We argue this likely stemmed from the gendered expectations surrounding literacy – neat writing style and ‘polite’ language use was imperative for women of a certain status (Elsweiler 2022). By utilising digital techniques to inform historically-grounded, qualitative analysis, this project thus offers a two-fold insight into the social and linguistic roles that female migrants played during the Pākeha settlement of New Zealand.
Tracking Islamophobia at Scale: Longitudinal Analysis of Right-Wing Discourse Using Large Language Models
Donald Matheson, Janus Nolasco, and Viet Anh Do
This paper studies how racial prejudice against Islam and Muslims in Aotearoa New Zealand has evolved over 20 years. It focuses on a large right-wing discussion community, Kiwiblog and aims in particular to track change after the country’s worst terrorist attack that killed 51 people at two masjids in the city of Ōtautahi Christchurch, when for a short time there was an outpouring of solidarity with the country’s small Muslim community (##), who had previously been ignored in public discourse (##) or represented in narrow, prejudiced ways.
The paper has three main concerns.
Firstly, it asks how textual practices expressing racial prejudice against Muslims have changed on the site from 2004 to 2025, including both the words and phrases associated with Muslims and Islam, who the focus of Islamophobic prejudice is and the broader discourse practices built through the texts. Racism is a cultural phenomenon, which belongs to historical moments and associated forms of political and cultural power. It is also a discursive phenomenon, not just a matter of individual attitudes, with mediated discourse a central element. Studying mediated racism over the longer term extends our understanding.
Secondly, the paper asks how prejudiced comments changed around the time of the terrorist attack on 15 March 2019 and in the years after. Did the dramatic change in broader public sentiment lead to a change in the way right-wing prejudice was expressed? Was it muted? Did it change in focus or form? Were changes sustained in the succeeding years? In a country with a very small Muslim population distanced from the racialised geopolitics and social issues that Islamophobia feeds off, we explore whether this form of prejudice be dislodged by political leadership or whether it local events have only a temporary impact on a prejudice embedded in a larger western cultural imaginary.
Thirdly, the paper explores how large language models (LLMs) can help to track prejudice. Are these tools as good as human analysts at identifying the presence of racism and at identifying changes in practice? As LLMs enable the study of language in its semantic context at scale, researchers are building tools that offer analysis of big data samples, real-time analysis, moderation and other possibilities. How should they best be trained and how reliable are they?
On the first two questions, the paper reports on some major continuities and shifts in the discursive form of prejudice at this discussion site. While western propagandist discourse equating Muslims with terrorists is sustained, it is joined by more meta-debate around the topic in later years, including greater denial of racism and use of debatability tropes. We find that, when Islamophobia comes under pressure in
Aotearoa, it draws upon other right-wing discourses around free speech and whites as victims.
Methodologically, the paper confirms the robustness of an LLM (in this case Google Gemini) when it is well trained. Gemini was as good as human coders in quality and far exceeded them in work capacity. We contribute to research at the interface of data science and social science by showing the value of training an LLM with talk coded as racist, non-racist and grey, through prompting the tool to give reasons and through using multiple cycles of refinement. The LLM remained weaker, though, in its ability to read for context.
Germany and the Pacific Islands: Media Framing and Discursive Construction
Serena Kelly and Mathew Doidge
German foreign policy has historically been tied to a core set of tenets, rooted in the historical experiences of the Second World War and of the post-1989 reunification process. These have included: an emphasis on soft power mechanisms (e.g. ‘Wandel durch Handel’) and an aversion to the application of hard power; a commitment to multilateralism and a rules-based international order; and a reluctance to engage in geopolitical contestation. Recent events, however, have placed a strain on this values framework. The emergence of the geopolitically-charged Indo-Pacific context, and the 2022 Russian invasion of Ukraine (recognised by Olaf Scholz as a Zeitenwende to which Germany would need to respond), have forced Germany to consider its role and place on the global stage. Since 2020, this has included an apparent rediscovery of the Pacific Island space, a long neglected area in Germany’s external relations.
This project constitutes a first step in a broader project examining Germany–Pacific engagement. Through a study of news media, its intent is to establish a baseline understanding of the way in which Germany is framed and understood in the Pacific Island space (here defined as the membership of the Pacific Islands Forum [PIF], excluding Australia and New Zealand), and correspondingly the way in which the Pacific Islands are framed and understood in German media.
Māori/Iwi Infrastructure Mapping
Garrick Cooper and Hamuera Kahi
Infrastructure development, often discussed and analysed at a nation state level, has received increasing scholarly attention over the last 10 years in what is sometimes referred to as the “infrastructure push” or “ turn”.
That the “infrastructure turn” in the Global South (in decolonial literature indigenous peoples of Global North nations are seen as part of the Global South) is increasingly being discussed, particularly with the greater visibility of BRICS (a grouping of Global South nations) in geo-political discourse, raises the question to what extent can Māori/iwi aspirations for greater autonomy be advanced with more focussed attention on infrastructural needs. By circumstance and perhaps by necessity Māori/iwi aspirations have largely leant on the machinery of the New Zealand government. Yet, as has long been demonstrated, particularly in service delivery areas like health, education and social welfare, yet alone the legal system, Māori have not fared well. On the back of iwi settlements, but not only, the Māori economy has grown exponentially over the last few decades. If internationally infrastructure investment has been demonstrated to aid and stimulate economic development, it would follow that a more tailored and bespoke approach to infrastructure that meets the need of Māori/iwi would contribute likewise to Māori economic aspirations.
This project looks to map out existing activities and services to reveal future infrastructure investment possibilities.