Mirror of the /src/gist_research url = git@gist.github.com:95870a8e16e30cac6906327fc850b3a0.git

71-Relational_Computing_Innu_AI_Language_Revitalization.md

Autonomous Edge AI for Innu-aimun Language and Culture: A Relational Computing Framework

I. Introduction: The Imperative of Relational Technology

A. Contextualizing Crisis and Opportunity

The global challenge of linguistic extinction presents an urgent imperative for innovative intervention. Approximately 40% of the world's estimated 6,700 languages are at risk of disappearing, disproportionately affecting Indigenous communities worldwide.1 This loss transcends mere vocabulary, representing the erosion of unique cultural worldviews, traditional knowledge, and intellectual heritage.2 For the Innu Nation, the vitality of Innu-aimun is inextricably linked to Innu-aitun—the singular way of seeing the world embedded in the language.3

In Canada, active revitalization efforts are already underway. The Indigenous Languages Act was established to support the reclamation, revitalization, maintenance, and strengthening of these languages.4 Collaborative projects, such as the Innu Language Project (ILP), involving university researchers and Indigenous organizations like Institut Tshakapesh and Mamu Tshishkutamashutau / Innu Education, have successfully compiled comprehensive resources, including a pan-Innu dictionary database containing over 27,000 words, mobile applications, and online lessons.5 Despite these successes, the demand for language resources and trained fluent teachers remains critically high.7 Furthermore, maintaining linguistic heritage shows profound positive correlations with community well-being, including lower teen suicide rates and improved physical health outcomes.1 The proposed Autonomous Edge AI device is designed to meet this urgent need by transforming passive language documentation into an active, fluency-generating tool.

B. Defining the Autonomous Edge AI Device

This project centers on developing an Edge AI system conceptually modeled after the "Language in a Box" concept proposed by Northern Cheyenne computer engineer Michael Running Wolf.9 This architecture is defined by its physical and functional integrity. The device is a secure, offline Edge AI unit, sometimes described as cedar-enclosed.9 Its primary function is to contain the necessary computational elements for a minimal voice-based language curriculum. Utilizing conversational AI technology, similar to commercial smart speakers, the device is intended to help learners improve fluency by providing accurate pronunciation modeling and practice opportunities.9

Crucially, the architecture prioritizes Indigenous Data Sovereignty (IDG). The decision to employ an offline Edge system ensures that all processing and storage of sensitive, low-resource Innu-aimun linguistic data remain local to the community.9 This commitment to autonomous operation is non-negotiable, providing local control and preventing the reliance on centralized, cloud-based infrastructure that inevitably raises ethical concerns regarding data extraction and commodification.

C. Epistemological Grounding: Research as Ceremony and Relational Accountability

The project’s foundation rejects linear, extractive Western technological development paradigms in favor of Indigenous methodologies. The philosophy underlying this work is that relationships are not merely elements that shape Indigenous reality, but relationships are reality.11 Dr. Shawn Wilson defines Indigenous research as the "ceremony of maintaining accountability to these relationships".11

This mandate translates into the rigorous adoption of Indigenous research frameworks:

Table 1: Foundational Indigenous Epistemologies for Decolonial AI

Epistemological Framework	Core Principle	Application to Autonomous Edge AI
Ceremonial Technology 11	Relationships are reality; maintaining accountability to all relations.	Development must be non-extractive and cyclic, prioritizing community-defined outcomes (reciprocity and long-term accountability).
Etuaptmumk (Two-Eyed Seeing) 14	Integrating strengths of Indigenous ways of knowing and Western knowledges.	Integrates Western Edge NLU technology 16 with Innu-aimun oral history protocols (Tshissenitamun) 17 and pedagogical objectives.
Relational Science Model 18	Integrity, respect, humility, and reciprocity as core values.	Dictates co-development, co-authorship of outputs, and community ownership of underlying datasets, ensuring data stewardship and recognitional justice.20
Indigenous Spiral Ontology 21	Time connects nature, human history, ancestors, and descendants.	Governs the iterative project lifecycle, demanding continuous feedback and validation against historical and future community needs, rejecting linear product obsolescence.

Core Indigenous values—integrity, respect, humility, and reciprocity—must shape every phase of the researcher’s responsibilities and methods.18 The concept of reciprocity is particularly crucial here, defined not as simple ethical compliance but as a conceptual intervention that supports biocultural continuity and demands accountability to sustain nature, thus restructuring technological approaches and policy interventions.23

The dedication to Indigenous Data Sovereignty inherently shapes the technological constraints of this project. Standard centralized cloud AI is inherently predicated on the extraction and commodification of data, which violates the core tenets of IDG and relational ethics.24 The mandatory selection of an offline Edge AI architecture is therefore not merely an engineering choice to reduce latency or cost, but a direct, non-negotiable epistemological requirement rooted in maintaining accountability to the Innu community’s relational ethics. This translates a deep philosophical commitment into a strict hardware constraint.

Furthermore, the technology is intended to be integrated into the Innu-aitun worldview. The physical description of the AI as being "cedar-enclosed" 9 leverages a material with deep spiritual and cultural significance in many Indigenous traditions. By building the system into a traditional material, the device is physically aligned with the concept of "Ceremonial Technology".11 This mandates that the relational accountability must extend to the material culture—the cedar itself must be ethically sourced and fabricated, thereby upholding the relational value of environmental respect and people-nature reciprocity.23

II. Philosophical Inquiry: Beyond Preservation to Ontological Creation

A. The Decolonial Turn in Digital Infrastructure

Decolonial thinking emphasizes the need for an alternate set of priorities structured from a new epistemology, ontology, and ethics that rejects Western rationality and dualism.25 The philosophical inquiry underpinning this project must address how AI, if not carefully designed, can reproduce colonial logics. Traditional AI systems often prioritize efficiency and data aggregation.26 When an AI system merely "runs analysis," it risks inadvertently lending authority to existing or past prejudices, rather than possessing the human capacity to challenge them.27

Achieving Epistemic Justice 28 requires a fundamental reorientation of the digital infrastructure. Decolonizing AI goes beyond diversifying training datasets or increasing technological accessibility; it demands a fundamental rethinking of the epistemological and ontological assumptions embedded in AI-driven systems.24 The effort must shift AI development from extractive paradigms toward relational ontologies 24, ensuring that knowledge production remains a pluralistic, situated, and relational process.26 This aligns with global ethical standards, as the UNESCO 2021 Recommendation on the Ethics of Artificial Intelligence calls for a value system that promotes justice, equity, interconnectedness, and autonomous decision-making, setting the bar for responsible IDG.30

B. Ontology of Language and Cultural Integrity (Innu-aimun)

The nature of Innu-aimun provides a specific challenge to conventional computational linguistics. Innu-aimun is understood as a "singular way of seeing the world," where words are not simply arbitrary signifiers but "word-images" formed by juxtaposing two concepts (e.g., Ishkuteutapan—fire-carriage—for "train").3 This highly relational and descriptive linguistic ontology fundamentally challenges the typical computational view of language as merely a corpus of statistical tokens.

The inherent risk is that AI models, optimized for computational efficiency and generic coherence, may inadvertently diminish the richness and authenticity of human speech, thereby overlooking the deep-rooted cultural contexts embedded in the language.31 If the technology is developed in isolation from the full context of Innu-aitun 32, it fails the test of relational accountability. Furthermore, the reliance on algorithms to transmit linguistic knowledge risks the pedagogical abdication of responsibility.26 If the ethical horizon of learning is collapsed into decontextualized, unreflective data flows, the human relationship required for effective pedagogy is endangered. Consequently, the Edge AI must be rigorously designed as a sovereign, culturally scaffolded tool for instruction and practice, never as a substitute for Elders or human educators.

C. The Shift from Archiving (Preservation) to Fluency (Creation)

AI technologies possess considerable utility in language preservation, primarily through archiving functions such as transcribing vast quantities of audio recordings, generating digital texts, and developing spell-checkers.2 However, the core purpose of the Edge AI device is the creation of new fluent speakers.34 This requires shifting the computational focus from passive documentation to active, iterative conversational fluency.

Innu culture is traditionally shared by speaking.3 The device must function as a conversational AI assistant 9, providing learners with an interactive environment where they can practice pronunciation and hear unfamiliar sounds accurately generated by the system. This focus on oral transmission, within a private, low-stakes setting, addresses the critical gap between having documented language assets (dictionaries) and achieving high levels of linguistic mastery.1

The adoption of the Indigenous conceptualization of spiraling time 21 carries direct implications for the AI model’s development and governance. Indigenous accounts of spiraling time connect nature, human history, ancestors, and descendants.21 In technical terms, this ontology mandates that the AI model training cycle cannot be a linear, one-time process. Instead, the model must cyclically and recursively re-validate its linguistic outputs against the ancestral knowledge captured in traditional oral traditions 17, while simultaneously optimizing its pedagogical approach for the needs of future generations (descendants). This necessitates a project management and governance structure featuring version control tied directly to community validation checkpoints, ensuring that technological updates are informed iterations that reinforce the relational core, rather than simply replacing previous versions based on universal, non-contextual metrics.

Furthermore, the principle of reciprocity requires mutual contribution and benefit.23 While the Edge AI provides technological benefits to the community, the reciprocal relationship must manifest tangibly in the academic and professional spheres. Reciprocity demands the co-development and co-authoring of research outputs with community members.18 The implication is that all academic papers, software documentation, and technical components derived from this work must legally and ethically credit the Innu community organizations (such as Institut Tshakapesh and the ILP) as primary knowledge stewards and intellectual contributors, thereby moving the collaboration from potential appropriation to genuine solidarity and transformative justice.24

III. Technical Architecture and Edge Implementation

A. Design Rationale: Autonomous, Low-Resource, and Offline Sovereignty

The foundational technical decision is dictated by the sovereignty mandate.9 Deploying an Edge AI system that operates entirely offline is the only method that guarantees Indigenous Data Sovereignty, ensuring that sensitive data remains physically localized and that the community maintains autonomous decision-making authority over its linguistic assets.10

The architecture must be intrinsically resilient and low-impact. The selection of specialized hardware, such as a low-power, small-footprint Neural Learning Unit (NLU) 16, minimizes dependence on large-scale infrastructure, reduces energy consumption, and aligns with the relational value of sustainability and accountability to the environment.23

B. Neural Learning Unit (NLU) and Low-Resource Modeling

The functional core of the device requires a robust NLU capable of on-chip training.16 This capability is essential for localized adaptation, enabling the system to be fine-tuned to the specific dialectal inputs of its deployment location without relying on external cloud connectivity. The Innu Language Project identifies several distinct dialects, including West, Central, East, Sheshatshiu, and Mushuau.5 A system that employs a generalized, monolithic model risks marginalizing dialectal variations. The requirement for on-chip fine-tuning in the low-power NLU is therefore essential, as it allows the device to be locally customized post-deployment, embodying the principle of epistemic pluralism at the hardware level by validating local linguistic differences.24

The modeling strategy must prioritize Small Language Models (SLMs) fine-tuned specifically on context-specific datasets.37 This approach ensures resource efficiency and high accuracy for the target domain, avoiding the biases inherent in massive, generalized models.27 This project can leverage existing, high-quality linguistic resources developed by the ILP, including the comprehensive dictionary database, specialized workplace glossaries, and existing conversation applications.5 Technical feasibility is established by research validating that specialized NLU architectures can perform on-chip training for complex tasks, achieving high accuracy (e.g., 89.34% for keyword spotting) using minimal training data.16

C. Integration of Innu-aimun Linguistic Assets and Protocols

The existing resources from the Innu Language Project provide a critical linguistic scaffold, including the pan-dialectal dictionary and the Innu Conversation app, which covers 21 conversational topics such as greetings, family, and the weather.5

However, the AI's success depends on integrating this lexical base with the deeper cultural and traditional context, known as Innu-aitun.32 Traditional Innu culture is strongly linked to nomadic life, cuisine, and respect for nature.3 While the ILP provides basic conversational topics, the NLU must be trained beyond mere vocabulary to understand and generate appropriate language regarding traditional topics and cultural scaffolding. This requires leveraging the Innu oral history protocols (Tshissenitamun), which emphasize the "art of oral storytelling" and the complex skills and high level of mastery involved.17 The AI must be trained using archived oral stories 35, not just for accurate pronunciation, but to recognize and validate the distinct narrative structures and rhetoric employed by Elders and knowledge keepers.

The integration of content must, therefore, serve a dual function: the ILP dictionary provides the necessary lexical base, while the oral histories provide the syntactic, rhetorical, and cultural context. This implies a layered model architecture where a fluency module is scaffolded by a cultural narrative module, ensuring the generated language retains authenticity and cultural fidelity.31 Furthermore, any future data collection—such as recording new oral histories or neologisms—must adhere to strict archival protocols, including using quiet locations, collecting contextual metadata (names, dates, interviewer details), and recording at high quality (e.g., 24-bit, 96 kHz) to ensure digital preservation standards are met.38

Table 2: Key Technical Requirements and Edge Implementation Matrix

Technical Dimension	Implementation Focus	Innu-aimun Integration Point	Decolonial Imperative/Risk Mitigation
Platform Sovereignty	Secure, physically isolated Edge processing unit; offline NLU operation.	Cedar Casing (Cultural Integration).9	Ensures Indigenous Data Sovereignty; prevents extractive cloud dynamics.
Model Optimization	Adaptive NLU/SLMs; on-chip training capability.16	Fine-tuning on ILP dictionary/conversation datasets.5	Maximizes accuracy for low-resource language; avoids dependence on generalized, biased large models.27
Pedagogical Output	Conversational AI (Voice-based curriculum); ASR/TTS.2	Prioritizes conversational fluency and accurate pronunciation practice.	Focuses on creating speakers (revitalization) rather than mere archiving (preservation).31
Development Cycle	Iterative design and update pipelines.	Spiraling Time Ontology validation loops.21	Guarantees continuous accountability and integration of Elder/community feedback, rejecting linear product obsolescence.

IV. Outline for the Peer-Reviewed Academic Article

The philosophical grounding and technical novelty of this project warrant publication in a high-impact, peer-reviewed journal at the intersection of computer science, decolonial studies, and Indigenous methodology.

Proposed Article Title: Relational Computing: Operationalizing Ceremonial Technology in the Design of Autonomous Edge AI for Indigenous Language Revitalization.

A. Abstract (200-250 words)

The abstract will frame the design of an Autonomous Edge AI device tailored for Innu-aimun revitalization. It will argue that mainstream AI’s extractive nature is incompatible with Indigenous Data Sovereignty and linguistic preservation. The solution is presented as a model for Relational Computing grounded in the philosophical mandate of Ceremonial Technology 11 and Etuaptmumk (Two-Eyed Seeing).14 The article will present the technical architecture (low-power, offline NLU 16) and the ethical framework that ensures Epistemic Justice 28, shifting the focus from passive language preservation to active speaker creation.

B. I. Introduction: The Epistemic Crisis of Extractive AI and the Case for Innu-aimun

This section will establish the context of linguistic crisis 1 and introduce the ethical challenges posed by conventional digital infrastructure.26 It will define the critical importance of Innu-aimun and summarize the existing foundational work of the ILP, which provides the linguistic assets for this project.5

C. II. Materials and Methods: Relational Science and Two-Eyed Seeing

This core methodological section details the blending of Indigenous and Western scientific approaches (Etuaptmumk).14

• Protocol for Co-Design: Detailed adoption of the Relational Science working model, outlining how the core values of integrity, respect, humility, and reciprocity structure researcher responsibilities.18
• Reciprocity and Authorship: Specific commitments to co-developing and co-authoring research outputs with community members to ensure appropriate credit and knowledge stewardship.18
• Data Governance: Explanation of protocols adherence to Indigenous research ethics and data governance principles, utilizing the offline Edge architecture as the physical manifestation of IDG.24

D. III. The Innu-aimun Context: Oral Tradition and Pedagogical Need

This section grounds the technology in cultural and educational reality.

• Review of existing Innu-aimun revitalization efforts and the critical demand for fluency resources.7
• Analysis of Innu-aitun, emphasizing the role of oral transmission (Tshissenitamun) and the need for the device to validate linguistic mastery beyond simple lexical definitions.3
• The argument for focusing computational effort on building speakers rather than solely archiving data.31

E. IV. Technical Architecture: Autonomous Edge NLU Implementation

A detailed presentation of the engineering solution.

• Hardware Rationale: Justification for selecting a low-power, small-footprint NLU capable of achieving energy savings and fast training times on-chip.16
• Modeling Strategy: Description of the SLM approach, fine-tuned on ILP conversational and dictionary datasets.5
• Sovereignty Implementation: Technical description of the offline operating environment, including the function of the physical cedar casing as an element of culturally responsible design.9

F. V. Results and Ethical Outcomes: Measuring Reciprocity and Creation

This section details how the project measures success across both technical performance and ethical accountability.

• Presentation of technical deployment data (e.g., ASR accuracy for specific Innu dialects).
• Metrics for Relational Accountability: Measuring success by assessing adherence to reciprocity protocols, including evidence of shared knowledge generation, formal co-authorship of outputs, and enhanced autonomous community decision-making authority over the technology lifecycle.18

G. VI. Discussion: Epistemic Justice in Practice and the Spiral of Development

• Analysis of how the project successfully mitigated colonial logics in technology design, specifically by implementing features that prevent the algorithmic usurpation of human pedagogical authority.25
• Extensive discussion of the Indigenous Spiral Ontology 21 as the project's model for research lifecycle and quality assurance, ensuring continuous accountability to both ancestral knowledge and future needs.
• Broader implications for decolonizing digital infrastructure and the potential applicability of the Relational Computing model to other low-resource Indigenous languages globally.

H. VII. Conclusion and Future Work

V. GitHub Issue Catalog: Translating Philosophy into Action

The following catalog translates the ethical and epistemological requirements into concrete engineering tasks for the development team. The assignment of governance tags ensures that ethical compliance is integrated directly into the software development lifecycle (SDLC). The prioritization of these issues reflects the centrality of Indigenous Data Sovereignty and relational accountability. The allocation of specific tags such as Sovereignty, Relationality, and Epistemic Justice to technical issues forces engineers to track the cultural and ethical rationale behind every component decision, ensuring that the decolonial mandate is maintained throughout the development process.24

Table 3: GitHub Issue Catalog for Autonomous Edge AI Development

Issue ID	Title	Description & Rationale	Tags
EDGE-001	Implement On-Chip/Offline NLU Training Module	Configure the NLU hardware 16 to execute all model training and fine-tuning processes entirely on the Edge device. Validate that data logging is restricted to local, sovereign storage. Rationale: This is the non-negotiable architectural requirement for Indigenous Data Sovereignty.9	Technical: Edge, Sovereignty, NLU
ETHICS-002	Define Reciprocity Metrics for Model Update Cycles	Establish a formal governance protocol requiring all model updates and new versions to undergo review, validation, and formal sign-off by designated community partners (e.g., Institut Tshakapesh). Specify legal IP ownership of derivative models. Rationale: Ensures relational accountability and shared knowledge generation.18	Epistemology: Ethics, Governance, Reciprocity
CONTENT-003	Integrate Innu Oral History Protocols (Tshissenitamun)	Develop ASR transcription accuracy and conversational syntax prioritization based on community-defined criteria for oral mastery.17 Train the system to handle the complex narrative structures and rhetorical flourishes found in archived oral stories.35 Rationale: Elevates culturally authentic context over generalized linguistic optimization.31	Content: Innu-aimun, Curriculum, Relationality
TECHNICAL-004	Develop SLM Fine-Tuning Pipeline for Dialect Customization	Utilize existing ILP dialectal data (West, Central, East, Sheshatshiu, Mushuau) 5 to create an adaptive SLM pipeline. The NLU must support local fine-tuning to specific dialectal input. Rationale: Operationalizes epistemic pluralism by preventing marginalization of local dialects.24	Technical: Modeling, Dialect, Epistemic Justice
HARDWARE-005	Design and Prototype Cedar Casing and Environmentally Responsible Peripherals	Finalize the design for the culturally appropriate cedar enclosure.9 Document and execute the sourcing protocol for the cedar wood to ensure alignment with relational values (sustainability and respect for nature).23	Hardware: Design, Sustainability, Relationality
TECHNICAL-006	Establish High-Quality Audio Collection Standard for Future Data Collection	Define strict recording specifications (e.g., 24-bit, 96 kHz, detailed contextual metadata, quiet location) 38 for any future community-led data collection efforts used to refine the model, meeting archival and preservation standards.2	Technical: Edge, Content: Innu-aimun, Archiving
ETHICS-007	Prevent Pedagogical Abdication: Implement "Non-Correction" Protocols	Design the conversational AI to serve as a supportive practice partner. Implement protocols that explicitly prohibit deterministic or critical algorithmic correction of user language errors. Rationale: Maintains the ethical pedagogical relationship and prevents the algorithm from usurping the human teacher’s role.26	Epistemology: Ethics, Pedagogy, Decolonial AI

The inherent need for the ASR/NLU model to accommodate multiple dialects from the ILP 5 reinforces the selection of the on-chip NLU architecture. Since Innu-aimun exists in multiple regional variations, a generalized ASR model risks prioritizing one dominant dialect, potentially accelerating the marginalization of others. The low-power NLU with its capability for on-chip training 16 is critical because it allows the Edge Device to be locally customized and fine-tuned after deployment to the specific dialectal input of its user community. This architectural feature directly resolves the potential ethical conflict of linguistic centralization, serving as the technical implementation of epistemic pluralism.24

The integration of Innu-aimun content must address both the lexical rigor found in the dictionaries and the complex narrative depth of the oral histories. While resources like the ILP dictionaries and conversation apps provide the necessary lexical and basic conversational bases 5, Innu-aitun requires a focus on concepts rooted in cosmology and transgenerational knowledge.3 The NLU must be architected to utilize the dictionary for foundational fluency while simultaneously leveraging the oral history data to provide deep cultural scaffolding within its conversational context window. This layered approach ensures that the language generated and validated by the AI moves beyond superficial communication, guaranteeing that the system contributes to the mastery of the entire ecology of thought represented by Innu knowledge, thereby moving past documenting "just knowledge" to incorporating "a way of looking at the world".12

VI. Conclusion and Future Directions

A. Synthesis of Relational Outcomes

The development of the Autonomous Edge AI Device for Innu-aimun language and culture successfully grounds advanced technological innovation within a rigorous framework of Relational Computing. The project fundamentally adheres to the principles of Ceremonial Technology 11 and Etuaptmumk 14 by establishing Indigenous Data Sovereignty as the primary architectural constraint, manifesting in the offline Edge NLU design.9 This strategic choice mitigates the historical risks associated with AI development, which often involves extractive and colonial logics.26 By integrating core Innu epistemologies—including the spiral ontology for iterative development 21 and oral history protocols for cultural contextualization 17—the project positions the Edge AI not merely as a technological tool, but as a sovereign, culturally scaffolded mechanism dedicated to achieving Epistemic Justice and supporting the intergenerational creation of fluent speakers.

B. Next Steps and Iterative Development (Spiral Continuation)

The immediate next steps involve initiating the first development spiral, guided by the technical and ethical requirements outlined in the GitHub Issue Catalog (Section V).

1. Formalization of Governance: The Data Governance Agreement (DGA) must be finalized and executed with Mamu Tshishkutamashutau and Institut Tshakapesh, legally codifying the principles of IDG, co-authorship, and community oversight (ETHICS-002).
2. Hardware Procurement and Initial NLU Training: The hardware procurement process, including the ethical sourcing and fabrication of the cedar casing (HARDWARE-005), must be completed. This enables the initial foundational NLU training pipeline (EDGE-001, TECHNICAL-004) utilizing existing ILP resources.
3. Community Co-Validation: Crucially, the initial output of the NLU modeling phase will not proceed to deployment without a mandatory community review and co-validation session (ETHICS-002). This session will test the model's accuracy against established oral tradition protocols and dialectal needs, ensuring the technology development remains a ceremony accountable to all Innu relations and knowledge keepers throughout its lifecycle.

Works cited

1. Can A.I. Help Revitalize Indigenous Languages?, accessed October 4, 2025, https://www.smithsonianmag.com/science-nature/can-ai-help-revitalize-indigenous-languages-180987060/
2. AI's Role in the Preservation of Endangered Languages and Cultures - Sure Coded, accessed October 4, 2025, https://surecoded.com/ais-role-in-the-preservation-of-endangered-languages-and-cultures/
3. The Innu - Vacances Essipit, accessed October 4, 2025, https://vacancesessipit.com/en/innu/
4. Indigenous Languages Act in various indigenous languages - Canada.ca, accessed October 4, 2025, https://www.canada.ca/en/canadian-heritage/services/indigenous-languages-act.html
5. Innu Language Project – Innu-aimun, accessed October 4, 2025, https://innu-aimun.ca/english/about/the-innu-language-project-2010/
6. Innu Language Project | Department Of Linguistics | Memorial University of Newfoundland, accessed October 4, 2025, https://www.mun.ca/linguistics/research/indigenous-languages/innu-language-project/
7. What role should universities play in revitalizing Indigenous languages?, accessed October 4, 2025, https://universityaffairs.ca/news/what-role-should-universities-play-in-revitalizing-indigenous-languages/
8. Challenges and Perspectives for Innu-Aimun within Indigenous Language Technologies, accessed October 4, 2025, https://aclanthology.org/2022.computel-1.13/
9. How AI could help safeguard Indigenous languages, accessed October 4, 2025, https://ca.news.yahoo.com/ai-could-help-safeguard-indigenous-121131350.html
10. Preserving the Past: AI in Indigenous Language Preservation - Viterbi Conversations in Ethics, accessed October 4, 2025, https://vce.usc.edu/weekly-news-profile/preserving-the-past-ai-in-indigenous-language-preservation/
11. Research Is Ceremony: Indigenous Research Methods - Fernwood Publishing, accessed October 4, 2025, https://fernwoodpublishing.ca/book/research-is-ceremony-shawn-wilson
12. Research is Ceremony: Indigenous Research Methods : RRC Polytech: Research Partnerships & Innovation, accessed October 4, 2025, https://www.rrc.ca/ar/2014/02/10/research-is-ceremony-indigenous-research-methods/
13. Indigenous Technologies - Resources - Berkeley Center for New Media, accessed October 4, 2025, https://bcnm.berkeley.edu/resources/3997/indigenous-technologies
14. accessed October 4, 2025, http://www.integrativescience.ca/Principles/TwoEyedSeeing/#:~:text=We%20often%20explain%20Etuaptmumk%20%2D%20Two,for%20the%20benefit%20of%20all.
15. Two-eyed seeing: a framework for understanding indigenous and non-indigenous approaches to indigenous health research - PubMed, accessed October 4, 2025, https://pubmed.ncbi.nlm.nih.gov/22894005/
16. (PDF) NLU: An Adaptive, Small-Footprint, Low-Power Neural Learning Unit for Edge and IoT Applications - ResearchGate, accessed October 4, 2025, https://www.researchgate.net/publication/389390845_NLU_An_Adaptive_Small-Footprint_Low-Power_Neural_Learning_Unit_for_Edge_and_IoT_Applications
17. Innu Tshissenitamun, accessed October 4, 2025, https://innueducation.ca/wp-content/uploads/2023/01/Innu-Tshissenitamun-Guide-2023.pdf
18. A values-centered relational science model: supporting Indigenous rights and reconciliation in research - Ecology & Society, accessed October 4, 2025, https://ecologyandsociety.org/vol29/iss2/art11/
19. A values-centered relational science model: supporting Indigenous rights and reconciliation in research | Native Nations Institute, accessed October 4, 2025, https://nni.arizona.edu/publications/values-centered-relational-science-model-supporting-indigenous-rights-and-reconciliation
20. Full article: Exploring Indigenous relationality to inform the relational turn in sustainability science - Taylor & Francis Online, accessed October 4, 2025, https://www.tandfonline.com/doi/full/10.1080/26395916.2023.2229452
21. Indigenous Accounts of Spiraling Time - ResearchGate, accessed October 4, 2025, https://www.researchgate.net/publication/382920403_Indigenous_Accounts_of_Spiraling_Time
22. [PDF] Indigenous Accounts of Spiraling Time - CiteDrive, accessed October 4, 2025, https://www.citedrive.com/en/discovery/indigenous-accounts-of-spiraling-time/
23. Reciprocal Contributions between People and Nature: A Conceptual Intervention | BioScience | Oxford Academic, accessed October 4, 2025, https://academic.oup.com/bioscience/article/72/10/952/6638973
24. Decolonizing Artificial Intelligence: Indigenous Knowledge Systems, Epistemic Pluralism, and the Ethics of Technology - ResearchGate, accessed October 4, 2025, https://www.researchgate.net/publication/393541288_Decolonizing_Artificial_Intelligence_Indigenous_Knowledge_Systems_Epistemic_Pluralism_and_the_Ethics_of_Technology
25. Full article: Towards decolonising the ethics of AI in education - Taylor & Francis Online, accessed October 4, 2025, https://www.tandfonline.com/doi/full/10.1080/14767724.2024.2333821
26. Full article: Decolonizing knowledge in the postdigital era: Pedagogical strategies for navigating AI-driven epistemic transformations - Taylor & Francis Online, accessed October 4, 2025, https://www.tandfonline.com/doi/full/10.1080/00131857.2025.2552764?src=
27. What does the future look like for AI? Caroline Running Wolf - INDIGENOUS AI, accessed October 4, 2025, https://www.indigenous-ai.net/what-does-the-future-look-like-for-ai-3/
28. Ethical Frameworks for AI in Indigenous Knowledge Preservation → Scenario - Prism → Sustainability Directory, accessed October 4, 2025, https://prism.sustainability-directory.com/scenario/ethical-frameworks-for-ai-in-indigenous-knowledge-preservation/
29. Decolonizing Artificial Intelligence: Indigenous Knowledge Systems, Epistemic Pluralism, and the Ethics of Technology - Fringe Global Scientific Press, accessed October 4, 2025, https://fringeglobal.com/ojs/index.php/jcai/article/view/decolonizing-artificial-intelligence-indigenous-knowledge-system
30. Leveraging UNESCO Normative Instruments for an Ethical Generative AI Use of Indigenous Data, accessed October 4, 2025, https://www.unesco.org/en/articles/leveraging-unesco-normative-instruments-ethical-generative-ai-use-indigenous-data
31. Generative AI and Large Language Models in Language Preservation: Opportunities and Challenges - arXiv, accessed October 4, 2025, https://arxiv.org/html/2501.11496v1
32. Discover – Maison de la culture Innue, accessed October 4, 2025, https://maisoncultureinnue.com/en/discover/
33. The Impact of AI in Language Preservation - AutoGPT, accessed October 4, 2025, https://autogpt.net/the-impact-of-ai-in-languages-preservation/
34. Native Language Revitalization - BIA.gov, accessed October 4, 2025, https://www.bia.gov/sites/default/files/dup/inline-files/bia_native_american_revitalization_lit_review_draft_08182023.pdf
35. Innu Oral Stories, accessed October 4, 2025, https://histoires.innu-aimun.ca/
36. Natural Language Understanding on Edge - Vivoka, accessed October 4, 2025, https://vivoka.com/edge-natural-language-understanding/
37. Can small language models revitalize Indigenous languages? - Brookings Institution, accessed October 4, 2025, https://www.brookings.edu/articles/can-small-language-models-revitalize-indigenous-languages/
38. Introduction to Oral History - Baylor University, accessed October 4, 2025, https://www.baylor.edu/content/services/document.php/43912.pdf
39. Oral History Best Practices, accessed October 4, 2025, https://oralhistory.org/best-practices/

---

by: Gemini account ellia url https://gemini.google.com/app/f3703ff0a233e3df

---

SEE ALSO:

• 71-Relational_Computing_Innu_AI_Language_Revitalization.html which is a very interesting HTML page created by https://gemini.google.com/app/f3703ff0a233e3df in which we had the session with within @elliadi account.

I want to plan a session of Autoethnography and Ceremonial Diaries with @avadisabelle soon abotu all that