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

This study uses the Palestinian Oral History Archive (POHA) to investigate how Palestinian refugees in Lebanon sustain a cohesive collective memory of the Nakba through shared narratives.Grounded in Halbwachs theory of collective memory, we employ statistical analysis of narrative pairwise similarities, examining the influence of shared gender and geography.We use textual representations and semantic embeddings to represent the interviews.Our analysis yields robust findings.Shared origin emerges as the primary driver of narrative similarity across thematic, factual, and semantic measures.This foundational connection is significantly amplified by a shared residence in Lebanon, with refugees sharing both origin and residence, exhibiting the highest narrative cohesion.Additionally, women's narratives show greater thematic cohesion, particularly in recounting experiences of the British occupation, underscoring the gendered dimensions of memory formation.This research deepens understanding of collective memory in diasporic settings, emphasizing the critical role of oral histories in safeguarding Palestinian identity and resisting erasure. Plain Language SummaryFor over seventy-five years, Palestinian refugees in Lebanon have kept memories of 1948 alive through storytelling.These oral histories describe life before the Nakba, the trauma of displacement, and the hardships of exile.More than personal records, they sustain collective identity across generations.Our research asks: Do refugees from different villages and backgrounds tell similar stories?What shapes those patterns?Drawing on theories of collective memory to form two hypotheses.First, gender would shape storytelling, since men and women experienced the Nakba and exile differently.Second, geography would matter, particularly when refugees from the same village settled in the same Lebanese camp, reinforcing shared narratives through continued proximity.We tested these ideas using the Palestinian Oral History Archive (POHA), one of the largest collections of refugee testimonies, preserving hundreds of Arabic semi-structured interviews documenting displacement, survival, and community life.We analyzed the collection using two complementary computational approaches.First, we examined archival metadata cataloging the people, places, and events mentioned in each interview.Comparing these references reveals whether refugee groups consistently highlight the same historical figures, landmarks, or episodes.Second, we developed a method for comparing narrative themes.We segmented interviews into topic-specific passages (such as Nakba-era violence or camp life), used language models to generate concise summaries, and then converted those summaries into numerical representations called semantic embeddings.This allowed us to measure how similar or different stories are thematically.Together, these methods provide distinct perspectives on collective memory.Our findings strongly confirm that gender structures memory.Women's narratives consistently cluster with other women's accounts, while men's stories show greater similarity to other men's.We also found that geographic continuity (when people from the same village are exiled to the same Lebanese camp) increased narrative similarity.However, shared origin or shared residence alone produced mixed results.This research demonstrates both the endurance of Palestinian collective memory in exile and the potential of computational methods to illuminate patterns in oral history at scale.By combining archival metadata with semantic analysis, we show how digital humanities approaches reveal structure in large collections of testimonies while respecting their complexity and human significance.