governance






Governance and Agreements · FNEF


FNEF thunderbird, artwork by Art Thompson

First Nations Education Foundation

LanguageCloud

Governance and Agreements

How FNEF works with communities, in writing. Everything on this page can be read, downloaded, filled in and signed from here: the agreement itself, the consent form for speakers, the agreement we use when a research partner or company is involved, how we work day to day, and the principles behind all of it. Nothing here is a funding contract. Your community keeps its language, its data and its tools, always.

What we commit to

These are the terms every community gets, written into the MOU below. They do not change from one community to the next.

You own it. All language content, recordings and data remain the property of the community they come from. FNEF never owns it and uses it only at your direction. (MOU Section 3)
You are never locked in. You hold a complete copy of your collection on your own computers, and you can export everything, with the source code of your installation, at no charge within thirty days of asking. (Section 4)
If FNEF ever stopped. Twelve months’ notice and a named fallback steward, so the work does not stop with us. (Section 4)
No AI without your written yes. No model is built from your language data without an agreement your community signs, and speakers consent individually. (Section 5)
Confidential by default. What we learn in your community stays in your community unless you say otherwise. (Section 7)
A charity, not a vendor. FNEF is a federally registered charity working on a cost-recovery basis. It sells no product and holds no rights in community language data.

How it fits together

1. The MOUSets out how we work together and what the community keeps. Non-binding on money; binding on ownership, continuity and data. Sign once.
Community or organization signs
2. Project schedulesEach project (a dictionary, an app, a language plan) gets its own short statement of work under the MOU, with its own budget and dates.
Added as projects come
3. Speaker consentEvery person who records says, in plain language, what their voice may be used for, and can change their mind later.
Each speaker signs
4. Data collaborationWhen recordings are to be used to build a speech, text or translation model, a separate agreement names the purpose, the builder and the benefits.
Community signs, per project
5. How we workShort, practical guides for the language team: bringing an idea forward, who to call, what happens when access or permissions change.
Not signed; used every day

1. The Memorandum of Understanding fill in and sign

Our standard MOU, Template V2 (October 2026). It establishes how we would work together, with specific projects added later as schedules. The fillable PDF has every field on the Quick Fill page, the signature page and the Schedule of Communities ready to type into, so it can be completed without printing.

To sign: open the fillable PDF in Adobe Acrobat Reader (free) or in your browser on a computer, complete the Quick Fill page, tick the boxes that describe your community or organization, type your name in the signature box as /s/ Your Name or sign it with Acrobat’s Fill & Sign tool, save, and email the file to scott.jeary@fnef.ca. We countersign and return the completed copy. Prefer paper? Print it, sign, and send a photo or scan. If you would rather talk it through first, or adjust anything, say so; we can send a Word copy for mark-up.

2. Speaker consent each speaker

A two-page, plain-language form for anyone who records their voice for the project. It says what the speaker agrees to, keeps speech technology as a separate yes, and explains what stays theirs and how to withdraw. It can be completed on a phone and signed with a finger, typed into as a fillable PDF, or printed and signed on paper. The language office keeps the record.

Tip for language offices: add your details to the phone link so speakers see them filled in, for example consent.html?community=Your%20Community&language=Your%20Language&office=you@example.ca.

3. Language Data Collaboration Agreement per project

Used only when language recordings or texts are to be used to build a speech-to-text, text-to-speech or translation model. It has three roles: the Data Holder (the Nation, or individual contributors), the Builder (the lab, agency or company building the model, or FNEF itself), and FNEF as Coordinator. The community keeps its data and decides where the model goes; the builder keeps its methods; speakers consent individually; reuse needs a yes. It puts the principles in section 5 into contract form.

Template Version 1 (September 2026). Because each project has its own purpose, builder, funding and benefits, the fillable version is meant to be completed together by the community, the builder and FNEF, and each Party should have it reviewed by its own advisor before signing. The italic guidance notes are for the drafter and come out of the signed copy.

4. How we work: practical guides for the language team

These four pieces explain in everyday terms how the MOU works in practice. They are written for a language coordinator, not a lawyer, and they are updated as we learn.

Supporting community-led language work version 2, October 2026

The guide itself: how people bring ideas forward, how permission follows the purpose, what an NDA does and does not do, how consent and withdrawal are handled, and how the community keeps its collection and tools. Read the guide (PDF)

Three one-page companions

  • Sharing an idea for the language project: a conversation starter for a new resource or project. Fill in (PDF) · Print (PDF)
  • People supporting the language project: a shared contact sheet so everyone knows who to ask. Fill in (PDF) · Print (PDF)
  • When access or permissions need to change: what happens when someone withdraws a recording, joins, or steps away. Read (PDF)

5. The principles behind it

Community Language Data and AI: Principles for a Fair Agreement (Version 7). Three things, three owners: the community owns the data, the builder owns its methods, and the trained model follows the community. Speakers consent. Reuse is expected, with a yes. Written so a Chief and Council, a language committee, a lab director or a funder can read it in five minutes. These principles are adopted in Section 5 of the MOU and govern any speech or AI work FNEF takes part in.

Questions

Scott Jeary, CEO · scott.jeary@fnef.ca · 604.340.5170 · www.FNEF.ca · www.LanguageCloud.ca

First Nations Education Foundation · Suite 1460, 701 West Georgia Street, PO Box 10156 LCD Pacific Centre, Vancouver, BC V7Y 1E4 · Thunderbird by Art Thompson (Nuu-chah-nulth, Ditidaht) · LanguageCloud™ is a trademark of the First Nations Educational Foundation.


Scroll To Top