Spatial Humanities 2024

Dr. Ignatius Ezeani

I have a Bachelor (1st) in Computer Science, a Masters in Advanced Software Engineering, and a PhD in Natural Language Processing (NLP). I am interested in the application of NLP techniques in low-resource domains. My interests span other related areas like spatial narratives corpus linguistics, distributional semantics, information retrieval and extraction, machine learning, data science, deep neural models, and general AI. Currently, I am the lead software developer on the £814k SBE-UKRI project trying to understand imprecise space and time in textual narratives through qualitative representations, reasoning, and visualization.


Session

09-25
16:00
30min
The Geography of Emotions in the Holocaust Survivor’s Testimonies
Dr. Ignatius Ezeani

In World War 2, the Nazi regime systematically persecuted and murdered millions of Jews and other targeted groups – an event commonly referred to as the Holocaust. While previous literature has explored what – drawing on Agnew and Duncan – the locations and locales of the Holocaust, less has been written on sense of place. One key source for understanding victims’ sense of place is post-war interviews with survivors. These provide valuable sources of historical and cultural knowledge, as well as emotional and psychological insight into the human condition under extreme circumstances. Of particular interest to us, given the focus on sense of place, survivors’ narratives contain references to the emotions experienced when describing memories of people, places, and events.
One aspect that can be explored in Holocaust survivors' testimonies is the spatial and temporal dimensions of the emotions expressed about people, places, and events, otherwise known as the geography of emotion. As Guy Miron signals in the case of German Jews, individuals experienced Nazi spatial control "both as a feeling and as a physical reality". Just as spatial experiences had an emotional dimension, so too did emotions have a spatiality or geography. Emotional geography is a concept that helps us understand how people feel about and react to, their environment, and how their environment influences their identity and memory. It also allows us to examine the interplay of different emotional experiences, e.g. fear, anger, surprise, sadness, disgust, and even joy which were originally proposed by Ekman and Friesen. We hypothesise that the analysis of these combinations of emotions expressed by multiple individuals at different places and times and in different situations during the Holocaust provides a much richer understanding of the geography and physicality of these emotions.
With this work, we aim to develop a computational framework for understanding the emotional landscapes of a textual narrative in a more nuanced form beyond the classification into positive and negative sentiments. We applied our method to a sample of Holocaust survivors' testimonies with a focus on Ekman and Friesen’s 6 emotion classes. We approached this study by posing the following fundamental research questions:
• Can we use natural language processing techniques - possibly leveraging large language models - to effectively extract and analyse expressions of emotions in Holocaust testimonies?
• If so, how can we quantitatively and qualitatively represent the interplay of different emotions in a survivor’s testimony?
• How does the expression of each of the emotions change across the narrative sequence of each testimony?
• Do different spatiotemporal elements (toponyms, geographical features, events, date and time, etc.) particularly relate or interact with specific emotions in any way?
For this work, we will adapt an extraction pipeline that is a version of the framework originally proposed by Ezeani et al., for extracting place names, and geographical feature nouns from text through named entity recognition for the Lake District corpora in the United Kingdom. The framework includes processes for fine-tuning an off-the-shelf named-entity recogniser. The fine-tuned model is subsequently applied to similar texts to perform surface-level extraction of spatial elements and even sentiment-bearing words for basic sentiment analysis. This pipeline will be improved by leveraging large language models for emotion classification as well as modifications appropriate to Holocaust testimonies.

AI and Spatial Humanities (Chair: Joana Vieira Paulino)
MG1/02.05