Part of our mission at Yakalo Collective is to increase public understanding of spoken languages from around the world that inform people’s relationships with land and animals. Cross-lingual language documentation has been practiced by linguists for centuries, but its efficiency could be improved with the provision of collaborative tools and the ability to store highly structured metadata. We believe that releasing a language agnostic tool that members of language communities can use to create structured annotations of spoken audio clips will enable them to curate datasets that help keep their languages in spoken use.

We have been developing Dito: Dialect interpretation and translation (online), a web app for creating collaborative annotations of audio data. We have nine principles for Dito:

1. Be usable wherever there is Internet–without requiring software downloads or specialized equipment–and function with any available Unicode script
   • Since downloading and installing software can be difficult and require access to specific platforms and fonts, we have developed Dito as a browser-based, Unicode-compatible web app that can be used with as many different linguistic scripts, audio file formats, and hardware devices as possible.
2. Maintain both accessibility and reasonable privacy of information, so that creators of audio files and their individual transcriptions/translations can intentionally hide them from or share them with other users and the public
   • Languages and dialects are inherently very personal and represent the landscape of the mind. Dito’s data is stored with Amazon Web Services (AWS) to take advantage of state-of-the-art security features; however, prospective users should consider whether they are willing to let their data be stored with AWS before contributing to Dito. Dito users are able to upload audio recordings and then annotate them privately, share them with others for collaborative work and learning, publish them for the general public, as well as switch between these options.
3. Center the epistemologies of orators such that their knowledge remains contextualized as much as possible; extracting or reducing their knowledge into small pieces should be discouraged by the format of the tool
   • Dito‘s annotation tools make it intuitive to link written quotations with their original soundtrack as well as contextualize them within larger quotations
4. Reduce the redundancy of mental labor asked of each separate person who transcribes, translates, listens to, or studies from an audio file within Dito
   • Different members of cross-lingual audio data annotation teams contribute in different ways. Dito makes any person’s advances in data annotation shareable with other members of their team; if one person annotates the audio file, then the text or timestamps of their annotations may be shared with other annotators, who can adjust them or reuse them in alternative annotations with just a few clicks.
5. Be compatible with, and build on, existing data (e.g. audio files, transcriptions, and translations)
   • Dito can be used to display and continue editing annotations from .eaf (ELAN Annotation Format) files, .srt (SubRip) files, Audacity Timing Files, and Pangloss/LACITO files–and we are open to requests for compatibility with other file types!
6. Allow descendants of expatriated speakers of dialects to study and learn from speech recorded in ancestral homelands making use of unique vocabulary that is not necessarily revealed through discussions in a city setting [use case]
   • Dito converts annotated audio data into beginner, intermediate, and advanced language-learning games and can be used to help users comprehend and internalize any annotated recordings of natural speech.
7. Create opportunities to practice and teach existing spoken dialects and any new orthographies that are designed to represent them [use case]
   • Dito converts annotated audio data into beginner, intermediate, and advanced language-learning games and can be used to help users learn to speak natural phrases and type using new orthographies.
8. Create a simple interface for transcribing audio directly into Dito [community request]
   • Dito allows data annotators to specify a length of phrase that they wish to annotate, then provides them with phrases of that length. Annotators can type their annotation and then simply press Enter to move on.
9. Honor our colleague’s request that we share his interview far and wide [community request]
   • By allowing members of the public to annotate publicly-shared audio files in any Unicode script and further share their annotations with the public, Dito facilitates wide, cross-lingual dissemination of specialized spoken language.

Some sample annotations created and displayed using Dito can be found at Yakaloco Dito, a Dito portal that we host. Click on any storybook card, click “Add Console” to view an alternative interpretation of the storybook, then click the Play button to immerse yourself in annotated natural speech.

We developed Dito to save users time with transcribing, translating, collaborating on, and studying from recordings of spoken audio in all languages and dialects. Now that the technology is developed, we are invested in tailoring it to fit the specific workflows of existing language communities and the projects they are engaging in.

We hope that developing these products kickstarts the generation of data and tools for language learners to make their journeys easier, more fun, and more fulfilling. Did we mention that spoken languages and dialects hold age-old knowledge about maintaining sustainable ecosystems? Our support for spoken languages is tied, hand-in-hand, with our support for human-animal-land relationships that keep societies synchronized with, adaptive to, and sustainable within our ever-changing environment.