Collective Memory and Narrative Cohesion: A Computational Study of Palestinian Refugee Oral Histories in Lebanon

TL;DR

The paper addresses how Palestinian refugees in Lebanon sustain a cohesive Nakba memory within a dispersed, marginalized community. It adopts a computational narrative-analysis pipeline applied to the Palestinian Oral History Archive (POHA), leveraging BoW metadata and semantic embeddings to quantify narrative similarity across interview pairs, under Halbwachs' group-memory framework. By testing boundary-making identity markers—origin, residence, and gender—the study shows that shared origin and shared residence robustly predict higher narrative cohesion, with amplified effects when both markers align; gender effects are present but more nuanced, with women displaying notable thematic cohesion on certain topics. The findings offer empirical support for diaspora memory mechanisms, demonstrating how oral histories can safeguard identity and counter erasure, and provide a scalable approach for analyzing collective memory in underrepresented archives.

Abstract

This study uses the Palestinian Oral History Archive (POHA) to investigate how Palestinian refugee groups in Lebanon sustain a cohesive collective memory of the Nakba through shared narratives. Grounded in Halbwachs' theory of group memory, we employ statistical analysis of pairwise similarity of narratives, focusing on the influence of shared gender and location. We use textual representation and semantic embeddings of narratives to represent the interviews themselves. Our analysis demonstrates that shared origin is a powerful determinant of narrative similarity across thematic keywords, landmarks, and significant figures, as well as in semantic embeddings of the narratives. Meanwhile, shared residence fosters cohesion, with its impact significantly amplified when paired with shared origin. Additionally, women's narratives exhibit heightened thematic cohesion, particularly in recounting experiences of the British occupation, underscoring the gendered dimensions of memory formation. This research deepens the understanding of collective memory in diasporic settings, emphasizing the critical role of oral histories in safeguarding Palestinian identity and resisting erasure.

Figures (11)

• Figure 1: UMAP of instruction-based embeddings of interview transcript sections for all themes. Related themes are visually indistinguishable in this decomposition, while different themes are distant from each other.
• Figure 2: Model coefficients reflecting the relationship between factual mention-based cohesion and gender. Pairs of interviews where both interviewees are men tend to be less similar than those where both interviewees are women or those where one is a man and one is a woman; interviews where both interviewees are women are usually more similar, especially for the less specific measures.
• Figure 3: Model coefficients reflecting the relationship between thematic cohesion within topics and gender. Pairs of interviews where both interviewees are men may be slightly more similar than the reference class of man-woman pairs, and pairs of interviewees where both interviewees are women are the most similar of all.
• Figure 4: Model coefficients reflecting the relationship between factual mention-based cohesion and shared location. Same origin and resid. consistently predicts the mention of similar entities and (for the keywords) broader concepts. Being from the same place and living in the same place often has a large positive correlation with similarity.
• Figure 5: Model coefficients reflecting the relationship between thematic cohesion within topics and shared location. Similar to Figure \ref{['fig:kw_location']}, being from the same homeplace or residence consistently predicts higher similarities in every major theme. Living in the same place and being from the same location is quite predictive of narrating major historical events similarly