Analysis and classification of oral literature of resistance using emerging technologies

Hassan Beheshtipour
Translated by Kianoush Borzouei

2026-1-22

Abstract:
The analysis and classification of oral literature of resistance in the age of emerging technologies is not merely a technical or archival undertaking; rather, it is inseparably bound up with fundamental methodological, epistemological, ethical, and political questions. The central issue concerns not only the capacities of these technologies, but also the limits and legitimacy of their intervention in narratives that have taken shape within the domains of speech, memory, emotion, and human relationality. Adopting a critical and interdisciplinary perspective, this article articulates the requirements of analyzing and classifying texts of oral resistance literature across three levels—analysis, classification, and shared imperatives—and demonstrates that the use of new technologies can contribute to a deeper understanding of this heritage only insofar as they operate within the framework of human interpretive authority, methodological transparency, and cultural–ethical sensitivity.

Keywords: Oral history; resistance literature; emerging technologies; artificial intelligence; digital ethics; shared authority

Introduction

The analysis and classification of oral resistance literature through the use of emerging technologies is not merely a technical or archival matter; it is deeply entangled with fundamental methodological, epistemological, ethical, and political concerns.(1) The pivotal question is not simply what emerging technologies are capable of doing, but rather how far—and under what constraints—they are entitled to intervene in oral narratives: narratives that are inherently formed within the domain of speech, memory, affect, and human relationships.(2)

Oral resistance literature is not merely a corpus of historical data; it constitutes a field in which individual and collective memory, lived experience, and the politics of memory converge. Consequently, any technological intervention in the analysis and classification of these texts inevitably carries epistemic and ethical implications. The central argument of this article is that the analysis and classification of oral resistance literature in the age of emerging technologies can be considered valid and ethical only when technology remains an interpretive instrument under human authority, rather than serving as the ultimate judge of meaning. On this basis, the requirements of this field are reorganized and explained across three principal domains—“analysis,” “classification,” and “shared and integrative imperatives”—without subjecting the content of the texts to deletion or reduction.

1. The Domain of Analysis in Oral Resistance Literature

1.1. Technical–Technological Imperatives

At the technical level, the use of advanced Automatic Speech Recognition (ASR)^[1] systems to convert speech into text represents a significant yet problematic opportunity. The accuracy of these systems—particularly when confronted with dialects, local accents, and marginalized voices—consistently requires human review and cannot fully replace the attentive listening and transcription performed by the researcher.(3)

Preserving vocal characteristics such as intonation, pauses, hesitations, and vocal fractures in oral texts is not merely a technical concern; it constitutes an integral part of the semantic layers and narrative authenticity of the text. Accordingly, the process of speech-to-text conversion must adopt policies that reflect these elements, as far as possible, through annotation and paratextual markers, rather than eliminating them entirely during the stages of “cleaning” and editing.(4)

While tools designed to identify and differentiate dialects and local accents can enrich linguistic analysis, they simultaneously entail the risk of unintended homogenization and the erasure of indigenous subtleties. In this context, the selection of models and software configurations must be carried out in collaboration with linguists and scholars familiar with local cultures, and the resulting outputs must be subjected to continuous critical evaluation.(5)

Technologies aimed at noise reduction and audio enhancement in older or field-based recordings are essential for the preservation of auditory heritage; at the same time, they raise the crucial question of where the boundary between “restoration” and “manipulation” lies. Any form of audio processing must therefore be accompanied by meticulous documentation of the procedures applied, enabling subsequent users to assess the degree of intervention and, where necessary, to access less-processed versions of the recordings.(6)

Protocols for detecting emotion and affect in audio recordings can provide a preliminary understanding of the emotional layers of a narrative; however, they must not supplant the researcher’s interpretive grasp of suffering, fear, hope, and resistance. Such tools should function as complements to qualitative analysis rather than as final arbiters of human emotion.

The use of generative artificial intelligence to reconstruct historical voices, simulate accents, or produce “cleaner” versions of narratives—despite its appeal—constitutes one of the most sensitive areas of this field. When deployed without transparency and explicit labeling, these technologies risk generating imaginative representations of the past and producing a form of “artificial authenticity,” thereby entailing the danger of unconscious historical distortion and a rupture between narrator and narrative. Consequently, any form of audio reconstruction must be explicitly identified as such and rigorously documented.(7)

1.2. Methodological Imperatives

The analysis of oral resistance literature necessitates the development of interdisciplinary frameworks that critically—rather than merely aggregatively—integrate oral history, linguistics, cultural studies, memory studies, and computer science. This integration must be operationalized through the design of explicit methodologies, coding schemes, and narrative analysis protocols.(8)

The formulation of qualitative criteria for narrative content analysis is essential for moving beyond purely quantitative and frequency-based analyses, which typically prove inadequate in capturing the complexity of human experience embedded in resistance literature. The combination of quantitative indicators (such as word frequency, conceptual networks, and thematic patterns) with qualitative measures (such as narrative depth, polyphony, ambiguity, and silence) can offer a more balanced representation of the material.

The integration of automated analysis with human interpretation is not an optional choice but a methodological necessity. This integration can be clearly illustrated in practice. For instance, AI-based sentiment analysis tools may classify a segment of a narrative—delivered in a mournful tone and recounting suffering—simply as “negative affect.” Yet a human researcher familiar with the historical context of resistance may recognize this tone as conveying a complex combination of collective mourning, protest, and, simultaneously, an expression of unbroken resilience. In other words, the machine measures “emotion,” whereas the human decode the “meaning” of that emotion within its cultural–historical constellation.

A similar issue arises in automated thematic classification. Large language models may categorize various portions of a narrative under headings such as “everyday life in prison” based on keyword detection. However, only the interpretive intervention of the researcher can illuminate the metaphorical connections between these apparently dull descriptions and concepts such as “passive resistance,” “the preservation of human dignity,” or “the creation of autonomy-enhancing spaces,” thereby preventing the reduction of lived experience to a list of abstract themes. Scientific protocols, however precise and powerful, lack historical lifeworlds, lived experience, and ethical and local sensitivities. Their role must therefore be confined to preliminary stages—such as transcription, initial indexing, and pattern suggestion—while interpretive decision-making remains firmly in the hands of the researcher.

Ethical principles—including informed consent, transparency regarding the use of narratives, the possibility of access restrictions, confidentiality in sensitive cases, and respect for personal experiences—must be embedded within the very text of the analytical methodology, rather than added as a ceremonial appendix at the conclusion of a project. In this regard, technological transformation and the emergence of new modes of dissemination intensify the need to revisit consent forms and agreements.(9)

Attention to the socio-political layers of narratives is likewise of fundamental importance, particularly in terms of how notions such as the enemy, resistance, suffering, victory, and collective identity are represented. At this level, narrative analysis becomes inseparable from questions of power, ideology, and the “politics of memory”: whose voices are recorded and amplified, which voices are marginalized, and who holds the authority to interpret and represent them.

It must also be acknowledged that technological development has posed unprecedented challenges to traditional notions of informed consent. Older consent forms—which often broadly authorized “future educational and research uses”—are inadequate in the face of new practices such as public online uploads, processing by artificial intelligence algorithms, or visualization within interactive digital maps. As highlighted in legal debates surrounding digital oral history, such ambiguities may generate both ethical concerns (violating the spirit of consent) and legal risks (including potential claims based on “defective consent”). Moreover, advanced relational database architectures that enable the linkage of an individual narrative to vast arrays of metadata (networks of persons, places, and events) introduce new levels of threat to privacy. A narrator may never have imagined that a personal memory could later be automatically connected, within an entirely new context, to data enabling identification or reinterpretation in unintended directions. Consequently, revising consent forms and providing precise transparency regarding all possible modes of access, digital processing, and data linkage is not a matter of choice but an ethical and methodological imperative in the age of emerging technologies.

1.3. Content–Linguistic Imperatives

Oral resistance literature embodies distinctive linguistic and narrative elements that must be identified and preserved: repetitions, chants and local songs, prayers and curses, proverbs, mournful compositions, and meaningful silences within speech. These elements constitute part of the affective and discursive logic of the narrative and must not be disregarded or eliminated during transcription and editing as mere “linguistic errors” by automated systems.

Analyzing the structures of oral narration—including modes of opening and closure, the role of the interviewer, the position of the audience, and the swinging between past and present—is essential to grasping the internal logic of the narrative. These structures differ fundamentally from those of classical written narratives. Consequently, methodological guidelines must acquaint programmers and system designers with these distinctions so that computational processes do not distort oral narrative structures.

A precise understanding of the cultural–historical context of each narrative is a prerequisite for rigorous analysis. An oral narrative stripped of context is reduced to impoverished data that carries only a faint residue of its original meaning. The connections between the narrative and its spatial, temporal, macro-historical events, and collective experiences must therefore be reflected both in metadata and in analytical frameworks.(10)

Identifying the relationship between collective and personal narratives—and the dynamics of their interaction—is particularly crucial in societies that have experienced violence, war, or repression. In such contexts, individual narratives must not be reduced to mere instances of a generalized “master pattern”; rather, attention must be paid to divergences and tensions between personal accounts and official discourse.

Attention must also be given to the temporality and mutability of narratives within collective memory. Multiple versions of a single narrative across different time periods should be documented and analyzed, rather than privileging one version as the definitive or “final” account. This process-oriented perspective enables the study of the evolution of collective memory and the continual rewriting of the past.

2. The Domain of Classification in Oral Resistance Literature

2.1. Conceptual and Thematic Imperatives

The classification of oral resistance literature requires the design of multidimensional classification systems—systems that simultaneously take into account thematic, geographical, temporal, narrator-centered, and event-centered dimensions. Such systems must enable scholars to move beyond conventional one-dimensional taxonomies.(11)

Within this framework, the identification of master narratives and counter-narratives becomes significant, without imposing official narratives upon marginal ones. Master narratives may illuminate dominant structures of collective memory, while counter-narratives possess the capacity to reveal ruptures, acts of resistance, and suppressed voices.

The differentiation of oral genres—such as memoir, tale, oral text, local poetry, funeral poem, and resistance-related songs—must be undertaken with cultural sensitivity and in collaboration with local experts. This differentiation should be grounded not solely in literary form but also in social and affective function.

Classification based on the social roles of narrators—from combatants and civilians to women, children, medics, prisoners, and veterans—must avoid reinforcing stereotypes or simplistic identity frameworks. Instead, it should enable the representation of experiential diversity and the disruption of reductive categorizations.

Attention to translation and intercultural transmission is likewise essential, as every act of translation carries the potential loss of emotional, cultural, and religious resonance. The classification of translated narratives must acknowledge this distance and maintain transparent links between original and translated versions.

2.2. Technological Imperatives

Intelligent tagging systems^[2] and automated classification mechanisms are valid only when they are capable of learning from experts and allowing for human correction. The design of such systems must be undertaken in close interaction with scholars of resistance literature, oral history, and cultural studies, ensuring that tags reflect culturally embedded concepts rather than mere translations of globally standardized terminology.

Multimodal classification models that analyze text, audio, images, and metadata concurrently are more effective than unidimensional models, yet they still require human oversight. While such models may suggest unused patterns within narrative networks, the validity and meaning of these patterns must be assessed through human interpretive processes.

Relational databases, particularly when integrated with spatial or schematic visualization tools, enable networked representations of relationships among narratives, narrators, locations, and events. When properly designed, such databases can provide fertile ground for uncovering unarticulated connections and unexpected pathways within collective memory.

Data visualization tools^[3]—such as interactive maps, narrative graphs, and timelines—can facilitate comprehension of overarching patterns, but they must not supplant historical and narrative analysis. Visualization should serve to generate new questions and clarify emerging patterns, rather than to oversimplify complex realities.

2.3. Cultural–Ethical Imperatives

Any classification based on formalized guidelines must remain acutely sensitive to the risk of imposing external conceptual frameworks upon indigenous narratives. Respect for intra-cultural classifications, local languages, and indigenous knowledge constitutes a fundamental cultural requirement in this field.(12)

Avoiding simplification and stereotyping—particularly with respect to marginalized groups such as women, ethnic and religious minorities, or victims of violence—is imperative. Classification should not become a mechanism for reproducing power ranking or erasing difference.

Transparency in classification criteria—both within methodological documentation and digital platform interfaces—is a prerequisite for scholarly and ethical accountability. Users must be able to understand the logic and indicators through which narratives have been grouped.

Questions of intellectual ownership and rights of publication and use—especially in narratives of suffering, violence, and trauma—must be addressed explicitly. Any secondary use (in film, games, literary, or artistic works) must respect narrators’ rights and operate within clearly defined agreements.

3. Shared and Integrative Imperatives

3.1. Authenticity and Credibility

maintaining the authenticity and credibility of narratives requires the establishment of systems capable of tracing the chain of transmission—from the moment of recording to digital archiving and subsequent uses. This chain should include information on the narrator, interviewer, time and place of recording, transcription processes, editorial stages, and forms of secondary use.

The linkage between narratives and their geographical and historical contexts must be preserved and reflected in metadata; severing this connection abstracts the narrative from its background and reduces it to decontextualized data. Analytical and classificatory processes must be documented in ways that enable scholarly review, critique, and reproducibility.

Legal and institutional mechanisms to prevent deliberate distortion, fabrication, or instrumental exploitation of narratives—particularly in political and media conflicts—are essential. Such mechanisms may take the form of guidelines, ethical codes, and explicit agreements among researchers, archives, and local communities.

3.2. Access and Participation

The design of participatory digital platforms that enable users and scholars to contribute annotations, interpretations, and corrections can enhance narrative dynamism and analytical richness. At the same time, the boundaries between raw data, scholarly interpretation, and public readings must remain transparent.

Striking a balance between preservation and open access constitutes one of the core challenges of this field. Restricting access in sensitive cases does not equate to concealing history; rather, it may represent a form of respect toward narrators and affected communities.

Multilingual utilization of narratives must preserve the integrity of the original language. Providing parallel access to original texts alongside translations—and documenting semantic and cultural divergences—can help prevent distortion and cultural homogenization.

Facilitating interaction between researchers and living narrators for the completion, correction, or updating of narratives forms a central component of the “shared authority” approach in oral history and must be incorporated into the design of digital systems.

Respect for indigenous knowledge and resistance to imposed external frameworks require that the principle of shared authority extend beyond the interview stage into the domains of analysis and classification. This can be achieved through structured participatory mechanisms within digital platforms—for example, enabling community representatives to review AI-generated tags, propose culturally grounded terminology, or add contextual interpretations that clarify local meanings. Such participation not only reduces stereotyping and unintended distortion but also significantly enriches metadata and generates multiple layers of meaning. This process transforms the archive from a static repository into a sustained dialogical space between collective memory, technology, and academic knowledge, providing an ethical safeguard for the more faithful representation of marginalized voices.

3.3. Sustainability and Development

Metadata standardization, the use of scalable technologies, and the integration of traditional archival knowledge with digital archiving capabilities are prerequisites for sustainability in this field. The adoption of open formats, avoidance of dependency on fully closed digital platforms, and long-term planning for data migration are integral to this sustainability.

The long-term preservation of this digital heritage requires attention to two interrelated layers: technological sustainability and semantic sustainability. At the technological level, the use of open formats and international metadata standards^[4] ensures that data are not confined within transient commercial platforms but remain transferable and retrievable over time. Experiences from projects employing XML-based^[5] tools for oral history archiving demonstrate how reliance on open standards can provide durable infrastructures for organizing and conducting complex searches across multimedia resources.

At the semantic level, sustainability depends on meticulous documentation of analytical and classificatory processes. Storing raw data or final outputs alone is insufficient; a “history of interpretation” must also be preserved—detailing which criteria informed each tag or classification, by whom (human agents or human-designed protocols), and within what conceptual context. Such transparency enables future review, critique, and reinterpretation as scholarly paradigms evolve.

The development of foresight models can help anticipate and manage the future trajectories of oral narratives and their modes of access across generations. These models may outline scenarios involving technological change, discursive transformation, and shifts in the loci of collective memory power.

Conclusion

The analysis and classification of oral resistance literature through emerging technologies demand a delicate balance between technical precision and human sensitivity, between the operational efficiency of computational protocols and deep content-based reflection, and between digital accessibility and the preservation of oral authenticity. This process must not sacrifice the most inborn quality of oral literature—its existence within the realms of speech, memory, and human relationship—to digital accuracy. Rather, through a critical engagement with technology, it should open new horizons for understanding, preserving, and transmitting this heritage.

Oral resistance literature is not merely a cultural legacy; it is a living document of history and the politics of memory. Technology—when properly understood and governed—must therefore serve history and its narrators, not replace them.

References:

(1) https://www.tandfonline.com/doi/full/10.1093/ohr/oht028
(2) https://oralhistoryreview.org/2021-virtual-issue/
(3) https://www.dsgains.pitt.edu/doing-oral-history-slow-work-can-artificial-intelligence-help-part-2
(4) https://arxiv.org/html/2508.06729v1
(5) https://wisprflow.ai/post/speech-recognition-challenges
(6) https://www3.dcc.fc.up.pt/~nam/elml/xata08.pdf
(7) https://oralhistoryreview.org/technology/chatgpt-oral-history/
(8) https://calenda.org/446300?lang=en
(9) https://ohda.matrix.msu.edu/2012/06/major-legal-challenges/
(10) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5652210
(11) https://dl.acm.org/doi/full/10.1145/3686397.3686420
(12) https://www.taylorfrancis.com/chapters/edit/10.4324/9781003122166-18/digital-oral-history-ethical-dilemmas-dealing-three-kinds-public-public-history-sugandha-agarwal