I don't have any problem with a independent, dedicated Linguistics.SE if it's going to survive, but I'd hate to see all that hard work and loyal followership (such as it is, anyway) be threatened by insisting on independence. I agree that there's a lot of overlap with cogsci.SE on one side and ELU on the other, so I'd suggest that those who want an independent Linguistics site should ask themselves, "What questions couldn't I possibly ask on either of the other (bigger) sites?"
If you've got some ideas in mind already, more power to you. If not, think twice about promoting Linguistics. I for one would welcome more linguistic questions over on cogsci.SE (where I'm coming from, hence admittedly biased toward), especially since our biggest need at the moment seems to be more questions. That being said, the same seems to be the problem for Linguistics, so it's a tough call. We could both continue to struggle as long as it takes (or as long as they'll let us), or we could merge. As for questions that don't belong on cogsci.SE, if they can fit into ELU, might those questions be better off there on an established site with better usage stats anyway?
If not, again, more power to you. I've got nothing against Linguistics. If I weren't a psychologist (with interests outside of linguistics), and if Cognitive Sciences weren't the broader topic with (marginally) better stats so far, I'd probably be advocating the opposite merger...if that even makes sense. I feel some need to disclaim and apologize for advocating the merger because I assume cogsci.SE would be the one to keep its name and not necessarily change much, but I'm open to improvements if there are any to be made in this regard, and I think everyone else should be too. Including Linguistics.
Consider what you gain from independence, and also what it costs you. Your questions are focused on linguistics, so you gain focus, but lose breadth. Some of you want to move in opposite directions along this dimension, so it's not a simple choice. Which is best? Would you lose so much focus by having to sort through non-linguistic, cognitive questions? Would merging necessarily mean losing any of your tag system's organization? Granted, some of it would probably go with the questions that are more appropriate on ELU, but if this is an option worth exploring, maybe someone should be talking to ELU about merging in that bit of your system somewhat faithfully. It's conceivable that your tag structures would remain intact individually, though in this case, some would split off to one site and some to the other. This also implies that off-topic flagging and migration issues would arise for those questions that really straddle the divide right now and would have a hard time finding the right home in the case of a schism. Hopefully this would just mean each site's moderators would get to know each other a little better, but it seems likely that some frustration would come along with this too.
On the upside, cogsci.SE might not need any further boost than this to launch out of beta (depending on what else might still be holding us back behind the scenes, I suppose), ELU would get a little boost, Lingustics' questions would get a boost from the extra readership, and Linguistics' users would be compelled to migrate and broaden their horizons in one direction, the other, or both, and both destination sites would be enriched to have them join. However, some pressure might be exerted implicitly on questions to be more oriented toward English and the humanities or more oriented toward cognitive science (for better or worse); some dedicated linguists would find their attention much more divided across two separate sites, each with their own broader pools of content; and the Linguistics SE would of course be no more. Hopefully everyone's reputation scores over there could be preserved somehow—maybe transferred to one of the two target sites of their choosing, if Area51 doesn't have a more established policy in place already.
If it's going to have a hope of happening, some of these complications should be resolved in advance, even if only tentatively, and Linguistics should really judge what kind of users it has and how they'd feel about this. The status quo is going to be popular I'm sure, but it's still something of a gamble as long as the status quo = "beta", so don't let the users forget that either if you care to raise the question over there. I don't know what the SE bigshots might do to promote cogsci.SE when it launches, or what additional perks ELU might already have for being out of beta, but any such promotions/perks are additional incentives for people to consider, so those should be identified and marketed along with the idea itself, if it is to be marketed further at all.