Some considerations you might be missing:

A language, among other things, is an ongoing, long-term, collective effort by a culture to categorize understanding: to divide up what is known, knowable, (or mistaken) into chunky abstractions that can then be played with lego-style to assemble new insights, hypotheses, or what-have-you.

Each language carves up reality a little differently.

When there are more languages in use, there are more versions of this carving at play. Some languages can easily express things that other languages cannot. Some languages make distinctions that others do not. And so forth.

When a language dies and merely exists as documentation in another language, something is lost. The very things in the dead language that are most exceptional are the ones that defy translation; they have their edges rubbed off during the documentation process and become e.g. "the closest English equivalent of" rather than their rich original meaning.

When we lose living languages, we lose more than alternative ways of expressing certain concepts, but alternative ways of conceptualizing.

I agree; keeping endangered languages alive is nowhere near the top 1000 possible Effective Altruist causes.

*

It definitely sucks to be one of the remaining monolingual speakers of a dying language. It is tempting to say that they made a very stupid choice, but maybe it wasn't really their choice. Perhaps they grew up in a small village isolated from the rest of the world. Maybe everyone around was hostile to them for ethnic or religious reasons, so there wasn't much opportunity and incentive to communicate. Or maybe they are just not smart enough to learn a second language. Life is not fair.

But these people are relatively few, otherwise the language wouldn't be getting extinct.

For the remaining bilingual speakers of a dying language, I think the loss is mostly cultural. Their situation could be compared to those expats whose children decide that learning the native language of their parents is too much work. The parents may feel very sad about not being able to share their culture with their children.

This kind of suffering is not comparable to a typical Effective Altruist cause.

For everyone else, there is a cultural loss of all the knowledge that existed in given language and hasn't been translated. When the language is gone, that knowledge is probably irreversibly gone, too. I think the lost knowledge will mostly be of historical and cultural nature; but sometimes there could be e.g. useful medical advice.

Again, there is real value lost, but not comparable to a typical EA cause. (Even the useful medical advice is probably about some rare local illness. Otherwise, I suppose, the useful information probably would have already spread beyond the small community.)

*

Some people will feel about this topic strongly, even if it is not an EA-level cause.

The ones who care about the cultural loss for humanity, they should document as much as possible. Take a camera, interview all the remaining speakers about all possible topics; upload to YouTube. Record people when they sing, and when they work. Record as much as possible, even if it is low quality; more data is better (you can still throw it away later, but you won't be able to record it later). Let the bilingual speakers tell the entire story in their native language first, and only afterwards ask them to translate.

I suggest recording over writing, because it is easier for the interviewed person (unless they have some tribal taboo against recording), so you can get much more data that way; and as a side effect you also document the accent, gesticulation, etc. Recording people while they work, as they comment what they are doing, helps to establish the meaning of their words.

If you record enough data, the future generations can sort it out, probably with the help of an AI.

Other people will be more concerned about the personal fate of the remaining speakers of the language.

I think the best thing you could do for the speakers personally would be to provide good language education for the young, and translation services for the old. (Yes, this does not oppose the extinction of the language; maybe even encourages it.) Be very sensitive about it! The goal should be to make the speakers fluently bilingual rather than to reeducate them. Let the language die (or maybe survive) naturally when its time comes.

And for some people (they will probably be over-represented online) this entire situation will be just a convenient metaphor for their favorite political topic. They will propose solutions that seem like a metaphorical way of hurting their political enemies, and oppose the solutions that seem like a metaphorical victory for the enemies. This is sad but probably inevitable; don't get involved in that, don't let your friends get involved in that.

Language is a border on culture, like a big wall.
Within a big language like English, people naturally invent new words when trying to reach for concepts they can not yet say, and this creates a tiny fence around a subculture. you can step over it, but the taller the fence the more the subculture diffs the broader culture.

I say this to say there is any value at all in having different communication protocols in the world at all. From an optimalist perspective you'd want everyone to have the same, because communication leads to truth right? but humans aren't immune to propaganda;  listening is not a free action. If the world spoke the same words tomorrow, people would immediately fight and get polarized, and diverge as they all tried to carve away a tiny little society that's safe and better by their values.

But I think such a society is much better than the one we currently have. Large borders seem worse than a polycentric world where people can move between subcultures and pick up the best parts. Freedom of movement allows people to leave cultures worse for them and enter cultures that are better.

Anyways, besides that small caveat, at this point in society we should if anything be actively trying to replace and assimilate the small languages rather than preserve them.