David Howcroft

David Howcroft is an Advanced Research Fellow in Natural Language Generation at the School of Natural and Computing Sciences, University of Aberdeen. He has held prior research positions at Edinburgh Napier University, Heriot-Watt University, and Saarland University, focusing on NLG, human evaluation, and low-resource language processing.

His research interests include:

• Natural Language Generation (NLG)
• Low-resource and under-resourced language processing
• Human evaluation methodologies in NLP
• Crowdsourcing for dataset creation and evaluation
• Psycholinguistics and linguistic representations
• Bayesian statistics and statistical modeling

The recent publications highlight a strong focus on methodological rigor in NLG evaluation, addressing low-resource settings, and ethical participatory design. His work spans corpus creation for minority languages like Scottish Gaelic, statistical analysis of human judgments, and improving evaluation practices through better metric reporting and participatory frameworks.

Scientific contributions include:

• Developing the first NLG datasets for Scottish Gaelic
• Advocating for proper statistical treatment of ordinal evaluation data
• Surveying and improving automatic metric usage in NLG
• Promoting participatory design in NLP for positive societal impact

David has led and contributed to multiple research projects involving stakeholder engagement, including the ASICA project for melanoma patient support and the NLG for Low-Resource Domains project. He has secured small grant funding and developed tools for data collection, evaluation, and model deployment. His work bridges computational methods with human-centered design and psycholinguistic insights.

He is actively involved in the NLP community through publications, talks, and outreach, including tutorials on tools like OpenCCG and Treex, and maintaining a technical blog on research practices and website development.