Methods for documenting and preserving endangered languages Automatic translate

Endangerment is generally defined as a situation where children and young adults no longer use their ancestral language in everyday life. While the language itself may still be spoken within the family, at religious or ceremonial events, it is not passed on as a first language to the next generation.

UNESCO identifies several levels of distress: vulnerable, when children use the language only at home, then definitely, seriously, and critically endangered, all the way to the language’s demise after the death of its last speaker. UNESCO’s online atlas included data on approximately 2,500 such languages out of approximately 6,000 to 7,000 existing ones.

Many languages lack a written tradition, standardized orthography, or stable teaching system. They survive through oral transmission, and as speakers’ linguistic habits shift, entire layers of grammar, vocabulary, and discourse practices disappear. Documentation and preservation in these conditions requires a combination of fieldwork, digital technologies, and collaboration with the communities themselves.

The difference between documenting, describing, and preserving a language

In modern linguistics, three related but distinct approaches are often distinguished: documentation, description, and language maintenance or revitalization. Documentation aims to create a voluminous digital corpus of recordings: oral histories, dialogues, rituals, everyday conversations, as well as annotated texts, dictionaries, and grammar notes.

Descriptive work forms a more abstract level: grammars, lexicographic works, studies of phonology, syntax, and semantics. These works rely on documented material and enable comparisons between languages.

Support and revitalization are linked to an increase in the number of speakers and the expansion of language use. These include full-immersion kindergartens, school programs, adult courses, media projects, and legal recognition of the language’s status. Many communities prioritize these tasks, viewing documentation as a supporting activity.

Principles of documentary linguistics

Documentary linguistics emerged as a distinct field by the end of the 20th century. The goal was to create a durable corpus of recordings with the maximum diversity of genres and communicative situations.

Basic principles typically include: a focus on natural speech, multi-level data annotation, a thorough description of the recording context, and transparent access conditions for users and researchers.

Another important principle is community participation. Native speakers act not only as informants but also as co-authors of the project: they determine priority topics, choose which texts can be published and which should remain confidential, and participate in transcription and translation.

Field data collection methods

Field research remains the basis for documentation. Specific methods depend on the social situation, the size of the settlement, people’s attitudes toward recording speech, and the extent to which the language is still widely spoken.

A combination of free notations and deliberate elicitation is typically used. Free notations capture stories, dialogues, folklore, everyday scenes, and native speakers’ own comments about the language. Elicitation helps glean examples of specific grammatical phenomena or vocabulary items that rarely occur spontaneously.

Choice of media and social context

When documenting, it’s important to consider age, gender, level of proficiency in multiple languages, and the extent of participation in traditional activities and rituals. A language community can contain active and passive speakers, as well as people with different dialects.

Researchers note that recording the "last speaker" alone is often insufficient. Data is needed on how languages are distributed in a region, what types of mixing exist, and how multilingual family and neighborhood networks are structured. This perspective allows us to understand the causes of language decline and assess the feasibility of language revitalization programs.

Ethical aspects and consent for recording

With the consent of the recipients, the terms of recording, storage, and distribution of materials are negotiated. A number of projects create access levels, ranging from completely open records to files intended only for members of a specific group or family.

Intellectual property issues are discussed separately: who owns songs, tales, and ritual texts; who may benefit materially or symbolically from their use. Archives are developing standard license forms and agreements, adapting them to the expectations of local communities.

Audio and video recordings of endangered languages

Digital audio and video recording is a central tool in modern documentary linguistics. With good quality, the camera’s angle of view and clear audio make it possible to revisit data decades later and analyze new aspects: gestures, gaze, and the spatial arrangement of participants.

Experts recommend recording audio in uncompressed formats with a sampling rate of at least 44.1 kHz and 16-bit resolution, and video in common high-bitrate codecs. This facilitates long-term storage and subsequent conversion.

Setting up the equipment itself is part of the methodology. In multilingual villages, cameras sometimes raise suspicion, so the researcher begins with audio recordings and gradually familiarizes themselves with the site, sharing copies with the family. In other cases, people willingly agree to video recordings of rituals, crafts, hunting, or fishing, seeing them as an archive for their descendants.

Genre diversity of recordings

The goal is to cover a variety of genres: stories about the past, fairy tales, songs, prayers, dialogues in a store, children’s games, household instructions, and explanations of grammatical forms. This set provides material for grammar, vocabulary, and sociolinguistic analysis.

Particular attention is paid to spontaneous, everyday speech. It allows one to identify frequent constructions, discourse markers, pauses, and self-corrections rarely found in traditional grammars.

Transcription, translation and annotation

After recording, the long process of transcribing and annotating the material begins. For many languages, this requires simultaneously creating a convenient alphabetical notation, developing rules for conveying length and tone, and reflecting the differences between closely related consonants and vowels.

Orthographic policy often seeks a compromise between phonetic accuracy and convenience for speakers, especially when schools are involved. Researchers discuss options with teachers, elders, and activists, taking into account existing writing traditions in neighboring languages.

ELAN, FLEx tools and integrated workflows

The most widely used platform for working with multi-layered annotations is the ELAN program: it synchronizes audio and video recordings with multiple lines of annotations, where you can enter transcription, literal translation, free translation, grammatical notes, and comments.

FieldWorks Language Explorer (FLEx) is widely used for morphological analysis and dictionary maintenance. Using these tools together allows for a streamlined workflow: transcription and initial translation in ELAN, then export to FLEx for morphological annotation and dictionary expansion, after which the updated data is returned to ELAN for refinement.

Additional tools have recently been developed to overcome technical barriers to file exchange between ELAN and FLEx. These solutions preserve metadata, speaker data, and multiple writing systems, and facilitate native speaker participation in transcription and editing.

Ontologies and search in marked corpora

To enable more flexible searching of multimedia corpora, ontological annotation systems are being created, where each gesture, action, or grammatical phenomenon is associated with an ontological element. The newly developed OntoELAN tool demonstrates how such concept dictionaries enable searching by semantic categories, not just by text strings.

Researchers also discuss the use of ELAN as a search engine for hierarchically tagged corpora. This reveals the technical limitations of standard search algorithms, stimulating the development of specialized tools for corpus work with resource-poor languages.

Lexicography for endangered languages

Dictionaries for languages with a small number of speakers serve several purposes: scientific, educational, and cultural. Unlike large national languages, they often need to combine information about dialectal differences, cultural realities, written transmission patterns, and usage examples.

Modern projects emphasize the role of corpora: dictionary entries are linked to audio and video examples, morphologically annotated texts, and illustrative material. This allows us to trace the word’s use in real speech, not just in artificially selected examples.

A separate area of research is the creation of bilingual dictionaries with an "intermediary language." Algorithms for automatically transferring lexical relationships from large networks (for example, WordNet) make it possible to create dictionaries even when only one established bilingual dictionary exists for the more widely spoken language.

Grammars and text collections

A grammatical description establishes a system of categories: word types, ways of expressing tense, aspect, voice, case, word order, and the structure of complex sentences. For endangered languages, grammar is typically based on a corpus of documented texts, not just on responses to individual questions in a questionnaire.

Collections of texts — stories, songs, dialogues, folklore — traditionally occupy a special place. They provide material for the analysis of stylistics, discourse markers, code-switching mechanisms, and for the study of oral tradition.

A number of projects are creating parallel publications: a text in the community language, a literal and free translation into the national language, and detailed grammatical commentary. These publications serve native speakers, linguists, and school curricula.

Archiving and digital preservation

The long-term preservation of linguistic material depends on high-quality archiving. Digital media are subject to format obsolescence and physical wear, so data is stored in specialized archives with a policy of regular format migration and backup.

Among the well-known archives is the Endangered Languages Archive (ELAR), founded in the 2000s and now housed at the Berlin-Brandenburg Academy of Sciences. The archive contains audio and video recordings, transcripts, dictionaries, and educational materials on more than 500 languages. User access is provided via a web interface, with access settings selected by speakers and researchers.

Other major initiatives include the DOBES, PARADISEC, and AILLA projects, as well as a number of national archives that are adopting collections based on the Open Linguistic Archives (OLAC) standards. These support unified metadata, facilitating the search and reuse of data for research and educational purposes.

Metadata and access rights

Metadata describes not only the technical parameters of a file but also the social context: who is speaking, where and when the recording was made, the language and dialect in which it is spoken, the topics covered, and who owns the distribution rights. Rich metadata increases the value of a collection for future research.

Archives are developing guidelines for access levels, license types, and linking methods to collections. This allows them to balance the demands of open science with respect for community expectations and privacy regulations.

Community as a participant in documentation and preservation

Experience from many projects shows that sustainable results are achieved with the active participation of native speakers themselves. These people serve not only as sources of material but also as field assistants, translators, transcribers, teachers, and archival collection managers.

Archives and grant foundations’ training programs include training in recording, annotation, metadata creation, and archival materials preparation. These courses are taught by staff from ELAR, PARADISEC, AILLA, and other organizations, combining online and in-person seminars.

In some cases, documentation is initiated by the communities themselves, with external specialists joining in as technical consultants. This is particularly noticeable in projects related to the rights of indigenous peoples and the legal recognition of languages.

Enrichment programs: language nests and immersion schools

One of the most well-known approaches to language revitalization is the "language nest" model, first implemented in Māori kōhanga reo kindergartens in the 1980s. In these institutions, children hear only their ancestral language from a very early age, and classes are taught by native speakers — often elderly relatives.

The success of the Māori model inspired other communities. "Language nests" became part of broader programs: full- or partial-immersion schools, camps, family clubs, and evening classes for parents.

Documentation is closely linked to such initiatives. Recorded stories and songs are used as teaching materials, dictionaries and grammars provide the basis for school curricula, and collaborative transcription efforts strengthen the language’s status as a resource for future generations.

Grant programs and international initiatives

Major grant foundations are developing targeted programs to support the documentation and preservation of languages with few speakers. In the United States, the Documenting Endangered Languages (DEL) program, implemented by the National Science Foundation and the National Endowment for the Humanities, funds field projects, archival collection development, and community outreach.

At the international level, UNESCO programmes have played and continue to play a significant role: the development of an atlas of endangered languages, the holding of conferences and increased attention to linguistic diversity in cultural policy.

Private philanthropic foundations such as Arcadia supported the creation of archives and branch campuses. For example, it was precisely this kind of contribution that made it possible to develop the Endangered Languages Documentation Program and the associated ELAR archive.

Modern digital tools and language technologies

Advances in automatic speech recognition, machine translation, and natural language processing have opened up new opportunities for working with underresourced and endangered languages. However, these approaches require careful implementation and the ongoing involvement of native speakers.

Research shows that automatic speech recognition can reduce the workload of transcriptionists. Experimental systems have been developed for some languages, such as Neo-Aramaic dialects or the Mixtec language Yoloxóchitl, that accelerate corpus creation.

At the same time, initiatives are emerging to create tools for lexicography and thesauri construction based on existing bilingual dictionaries and large lexical networks. Such solutions provide additional resources even for languages with extremely limited data sets.

Infrastructures for low-resource and Uralic languages

Some projects are building complex infrastructures for groups of related languages. Electronic dictionaries in XML format are being created for Uralic languages, which then serve as the basis for morphological analyzers and other tools.

These infrastructures combine traditional fieldwork methods with modern neural network models. The quality of the source data remains central: competent annotation and accurate metadata increase the value of every minute of recording.

Artificial Intelligence in Documenting Pragmatics and Semantics

Several studies demonstrate how machine learning methods can help identify pragmatic markers and semantic structures in languages with very few texts. For example, regional languages in Pakistan, for which written corpora are almost nonexistent, were studied. Combining fieldwork with analysis using modern models helped systematize markers that regulate the flow of conversation and express speaker attitudes.

However, the authors of such studies themselves emphasize that linguistic analysis, native speaker participation, and cultural context remain indispensable. Technology serves as an accelerator, not a substitute, for fieldwork and collaborative discussions.

Documenting prosody and intonation

For many languages, especially those with tonal or complex intonation systems, it is important to capture not only the sound sequence but also the melody of an utterance. Research on Dene-Athabaskan languages shows that comparing data from different types of tasks — reading, retelling, and free speech — helps identify intonation patterns associated with utterance types and information structure.

Such studies utilize high-quality recordings, precise ELAN alignments, and specialized phonetic analysis programs. This results in corpora that allow for the study of the interaction of intonation, morphology, and syntax, which is unlikely to be possible with text data without audio.

Multilingualism and language contact in documentation

In many regions, endangered languages coexist with several more widely spoken languages. People freely switch between them, borrow constructions, and change code depending on the topic and interlocutor.

Some researchers believe that to truly capture the life of a language, it’s necessary to document the multilingual environment, not just "pure" monolingual texts. Specialized corpora focused on language contact and multilingualism help trace how language shift occurs, which domains remain in the native language, and which are transferred to the state language.

When annotating such materials, it is necessary to take into account not only the linguistic affiliation of each statement, but also social factors: the speaker’s status, his age, attitude towards language and towards research.

Documentation of sign languages and bimodal bilingualism

Sign languages are also threatened with extinction. High-quality video recording and tools that allow for the identification of multiple channels — hands, face, body, and parallel or alternating spoken language — are particularly important for their documentation.

There are projects studying children growing up in families with deaf parents and simultaneously acquiring both sign and spoken language. For such corpora, ELAN is developing special annotation conventions, where each modality receives its own annotation strings, and the relationships between them are recorded with precise time stamps.

Methods for working with such data are then transferred to other communities where sign language also comes under pressure from dominant languages and practices.

Automating recording and assisting field linguists

Current research is exploring whether a machine learning model can suggest which forms have not yet been recorded and what questions to ask the speaker to more effectively collect morphological paradigms.

Systems are proposed that analyze existing data and offer examples for clarification, minimizing repetitive questions and filling gaps in paradigms. This approach allows for better use of limited fieldwork time and reduces the burden on researchers, who often become fatigued by lengthy elicitation sessions.

At the same time, the authors emphasize that the models are trained on existing data, so the richness of the collected corpus still depends on the initial stage, where the intuition of the field researcher and joint planning with the community are important.

Examples of project methods: Moklen, Komi, and Megrelian languages

The Moklen language documentation project demonstrates how a specialized system, LangDoc, helps organize work with a language without an established written system. Researchers use a word list as the basis for recording, then attach audio, transcription, phonetic, and cultural annotations to each lexeme.

The system integrates project management, recording, quality control, and annotation, and prepares data for subsequent dictionary and grammar creation. This approach reduces the number of disparate files and facilitates monitoring of vocabulary coverage.

The Izhem Komi language project focused on automated annotation: a script was created linking ELAN with morphological analyzers and syntactic taggers developed for Uralic languages. This allowed for faster annotation of a large corpus of spoken and written texts and brought work with this resource-poor language closer to the level available for national languages.

For Mingrelian, a member of the Kartvelian family, lexicography relied on documentation data and a rethinking of priorities: attention shifted from a simple list of translations to reflecting dialectal differences, examples from living speech, and connections with other Kartvelian languages.

Digitization of printed dictionaries and "obsolete" resources

Over the decades, many missionaries, educators, and researchers created dictionaries on paper cards, printed them on machines, and published them in small print runs. These works often remain the only recorded evidence of the vocabulary of a number of languages.

Digitization projects for such dictionaries use optical character recognition, then automatically or semi-automatically convert the dictionary entry structures into a machine-readable format. This requires developing rules for identifying lemmas, translations, examples, grammar notes, and style notes.

Once structured, the data can be linked to new corpora, compared with other dictionaries, and used as a starting point for further expansion. Thus, decades of work done in the pre-digital era are given new life in modern infrastructures.

Education and training of specialists

Field documentation and archival work place special demands on researchers. They must master recording techniques, sound engineering fundamentals, annotation principles, ethical standards, and have an understanding of information standards and licenses.

A number of universities and archives offer specialized courses and summer schools that combine theoretical lessons with practical training in working with ELAN, FLEx, archival interfaces, and grant proposal writing.

Digital courses and open learning materials make it possible to engage not only linguistics students but also language activists, teachers, and community representatives in such training, which enhances the practical impact of documentation.

Methods for evaluating the effectiveness of language preservation projects

When discussing language preservation, an important question is assessing the impact of a specific project on the language’s vitality. Some studies suggest taking into account the dynamics of the number of speakers, changes in age distribution, the expansion of language use, and the emergence of new domains such as media, digital platforms, and official events.

From a documentation perspective, one indicator is the completeness and availability of the corpus: the presence of audio and video recordings of different genres, grammars, dictionaries, teaching materials, as well as the degree of community participation in their creation and use.

Researchers emphasize that there are no universal scales for evaluating such projects. Approaches must be adapted to local conditions, demographics, the political status of the language, and the expectations of the speakers themselves.

Anchoring languages in the digital space

Documentation opens the door to the digital presence of endangered languages. Corpora are used to develop keyboard layouts, fonts, spelling standards, and electronic dictionaries. Archival collections are becoming a source of audio material for podcasts, video channels, and mobile apps.

Research projects to create generative models for underresourced languages raise questions about data protection and the ethical aspects of training models on materials created and owned by specific communities. Initiatives are being developed to use new technologies to help speakers control the use of their languages and knowledge.

With a well-designed access rights architecture and transparent collaboration conditions, digital tools become another means through which documentation connects with revitalization initiatives and everyday language use practices.

Legal framework and language rights

Documentation is closely linked to the legal recognition of languages. International documents from UNESCO and the UN emphasize that the use of one’s native language is a human right, and cultural multilingualism is described as a resource requiring protection.

National laws define the status of languages differently. Some countries guarantee instruction in local languages, while others permit their use only in cultural contexts, without official recognition by courts and government agencies. These differences affect access to funding and the scale of documentation projects.

Legal norms also affect archives. Licensing agreements are emerging that stipulate who can reproduce recordings, under what conditions commercial use is permitted, and what forms of attribution are required. Archives are developing their own consent models to take into account the collective rights of communities, not just the individual rights of speakers.

Interdisciplinary documentation links

Materials on endangered languages are of interest not only to linguists. Anthropologists use them to analyze rituals, kinship systems, and behavioral norms. Ethnographers study economic practices and spatial concepts through oral histories. Musicologists study song genres and speech rhythms.

These disciplines contribute their own methodologies. For example, a detailed description of the ritual context clarifies the meaning of forms of address, while a musical analysis of a ritual song reveals recurring syllable structures important for phonology and morphology. Collaborative work helps harmonize terminology and annotation formats so that the materials can be used in various studies.

Musical and poetic material

Songs, chants, and poetic forms require special documentation methods. They are often associated with sacred practices, and permission to record must be agreed upon with a group of elders or religious leaders. Sometimes, only audio recordings are permitted, without video, or limited archival distribution.

When annotating such materials, researchers work with speakers knowledgeable about the tradition: they clarify the structure of verses, the functions of repeated lines, and the relationship between melody and accent patterns. For songs, parallel layers of annotations are created: lyrics, melodic line, rhythmic markings, and comments on the content and performance situation.

Musical material is often used in educational projects. Recorded songs become the basis for school concerts, radio broadcasts, and compact compilations for family listening. It is important to coordinate distribution methods with those who hold the tradition, so as not to violate local access regulations for certain genres.

Folk knowledge and environmental terminology

In many communities, knowledge of local flora and fauna, landscape features, and seasonal phenomena is linked to the native language. Documentation encompasses a list of plant, animal, and landform names, as well as descriptions of their uses and associated stories.

Ethnobiologists and linguists document which characteristics are considered important for classification: color, shape, behavior, taste, and medicinal properties. Records of conversations, field trips, and collaborative work demonstrate how these terms are enshrined in set expressions and proverbs.

This material is subsequently used in regional educational programs and environmental projects. It’s important to avoid romanticizing it: for those who hold it, this knowledge is connected to everyday survival and economic strategies, not just symbolic meanings.

Urban and diaspora communities

Some endangered languages survive not in rural communities, but in large cities and diasporas. Here, documentation faces different challenges: families may switch between several countries daily via phone and messaging apps, and the ancestral language is heard only in certain communication scenarios.

A field linguist records conversations in apartments, at celebrations, and at public organizations. Multilingualism is particularly evident: code-switching occurs within a single phrase, children incorporate elements of the official language into conversations with their grandmothers, and adults adapt their vocabulary to urban realities.

Documentation in such conditions requires flexible ethical decisions: people may be wary of being recorded due to their immigration status, conflicts within the diaspora, or the political situation in their home country. It’s important to discuss in advance where and how the materials will be stored, who will be able to access them, and how to organize the return of the recordings to the participants themselves.

Methodological disputes in documentary linguistics

Several persistent issues are debated in the professional community. One concerns the balance between natural speech and conventional elicitation. Some researchers emphasize free dialogue and folklore, while others believe it’s necessary to systematically collect samples through questionnaires to avoid missing rare grammatical constructions.

Another issue concerns the volume of accompanying data. Some projects devote considerable attention to describing cultural context, economic practices, and genealogies, while others focus on linguistic structure and limit themselves to minimal annotations. The debate revolves around what priorities are appropriate given limited resources and time.

Quality standards are also discussed: should we strive for maximum technical accuracy in recordings if this reduces the spontaneity of communication? What level of phonetic detail is justified in transcription? How much time is acceptable to spend on checking each text when native speakers and researchers are overloaded with other tasks.

Data standards and resource interoperability

For long-term work with corpora, standard formats and descriptions are essential. OLAC initiatives and other consortia are developing metadata sets that allow collections to be described using standard parameters: language, region, genre, technical characteristics, and access conditions.

Common text and annotation exchange formats based on XML and related standards are used. This facilitates the transfer of collections between archives, software updates, and the development of new search and visualization tools. Each community and project can also introduce its own additional fields if there are local needs.

For lexicographic data, entry description standards are used, allowing for linking different dictionaries together and matching them with corpora and machine translation tools. Such solutions enhance the value of each individual dictionary, even if it covers a limited number of lemmas.

Educational materials based on documentation

Many projects aim to use the results of recordings and annotations in teaching children and adults. Corpora are used to create reading books, audio lessons, flashcards for games, and materials for clubs and schools. These resources are based on real speech, not fictitious examples.

Documentation helps identify the most frequent words and expressions, as well as typical constructions useful for beginners. Teachers and activists select short stories, dialogues, and songs from the corpus, adapt the spelling, and create illustrations. This approach narrows the gap between the "academic" corpus and everyday language use.

An important task is training the community itself to work with the materials. Training is needed on how to use the archival interface, corpus searching skills, and the ability to adapt texts to the students’ age and language proficiency.

Media and digital content in endangered languages

Documentation stimulates the emergence of media projects. Podcasts, short videos, radio programs, and sometimes even local-language series are created based on recorded stories and songs. These formats attract a young audience accustomed to the digital environment.

Speech corpora facilitate the creation of subtitles and dubbing. Native speakers record their own stories, and linguists assist with spelling, markup, and technical aspects. This creates a product that simultaneously entertains and strengthens listening and reading skills in the native language.

Some projects are experimenting with interactive applications: vocabulary-based games, phrase memorization trainers, and local audio guides. In these cases, documentation provides the foundation without which such products could not exist.

Working with archival historical records

In addition to new field expeditions, the digitization of old collections is of great importance. These include phonograph cylinders, magnetic tapes, and early video recordings made by anthropologists and musicologists in the 20th century. For a number of languages, this is the only available material.

Restoration processes include transferring audio to modern media, filtering noise, and improving speech intelligibility. A transcription, translation, and annotation are then created, as for modern recordings. It is important to preserve the original files and document the processing methods used.

Comparing old and new recordings of the same language allows us to trace changes in vocabulary, phonetics, and speech tempo. This is not only a historical source but also a benchmark for modern revitalization programs, which sometimes attempt to restore lost vocabulary or grammatical forms.

Documenting the "last carriers"

In extreme cases, researchers encounter situations where only a few elderly speakers, or even just one, are still alive. Here, the research methodology changes: the emphasis shifts to maximizing the speaker’s comfort, searching for old recordings, letters, and notes that could provide further insight.

The workload for such a person is high, so recording sessions are divided into short segments, alternating between conversations, reading old texts, discussing photographs, and other visual stimuli. Family and friends are often involved, even if they are no longer fluent in the language, to support the conversation and reduce emotional tension.

The ethical aspect is particularly perceptible: one must avoid the feeling of being the "last witness" or reducing one to the status of "last bearer." Jointly planning the recording, discussing the desired themes and forms of use of the material, helps to alleviate these tensions somewhat.

Financial and organizational difficulties of projects

Documenting endangered languages is often carried out with limited resources. Travel to remote areas is costly, and grant programs compete with other humanitarian efforts. Short-term contracts make long-term planning difficult.

Project organization includes coordination with local authorities, obtaining permits, equipment logistics, and recruiting translators and assistants. For the project’s sustainability, it is important to establish collaboration with local schools, community organizations, and cultural centers, which can continue their work after the grant expires.

Additional challenges arise during political instability, natural disasters, epidemics, and border closures. In such circumstances, some work is transferred to online formats, with media acting as independent data collectors using available recording devices.

Criticism, risks and responses to them

Some researchers and activists criticize documentation for its potential "extractive" practices, where external specialists receive data, grants, and publications while communities themselves see no benefit. In response, archives and programs support the principles of collaborative planning, fair compensation for archivists, and shared ownership of materials.

Issues of privacy and sensitive information are being discussed. Flexible access settings are being implemented in the archives, including time limits and user restrictions. Ethical codes are being developed that require researchers to share their findings with the community whenever possible and to incorporate feedback.

The academic community is also raising the issue of quality: not all collections are equally detailed, and metadata doesn’t always meet high standards. Continuing education courses, the exchange of experience between archives, and the publication of methodological guides and examples of good practice can help.

Youth participation and training of research speakers

In recent years, increasing attention has been paid to the participation of young media in documentation projects themselves. Schoolchildren and students are trained in the use of voice recorders, cameras, annotation software, and the basics of linguistics and archival science.

This approach achieves several goals simultaneously. Young people gain skills that can be applied in other fields, communities gain people capable of independently leading new projects, and researchers gain partners well-versed in the community’s cultural context and social networks.

Some programs offer scholarships and mini-grants specifically for native speakers to conduct their own research: recording family histories, researching local toponymy, and collecting craft terminology. Archives provide technical and methodological support for such initiatives.

Practical guidelines shared by many experts

Despite the diversity of projects and approaches, several principles can be identified that are often encountered in descriptions of successful initiatives to document and preserve endangered languages:

• Respectful and cooperative attitude towards the community, jointly defining the goals and topics of the recording.
• The desire to capture natural speech, not just responses to questionnaires, while maintaining the grammatical completeness of the collected data.
• Long-term storage is a priority: choosing reliable formats, detailed metadata, and storing them in a specialized archive.
• Maximum possible involvement of native speakers in all stages of the work – from recording and transcription to the creation of dictionaries and educational materials.
• Focus on data reuse: open formats, clear descriptions, accompanying documents explaining the structure of the collection.

These guidelines do not cover all the different situations, but they are often used as a starting point when planning new projects and discussing work already completed.