Ethics of Personal Politics

I have been thinking lately about the relationship between ethics and (personal) politics in the documentation of endangered languages (or the study of any language, really). Specifically, what are our responsibilities as researchers to either disseminate or withhold particular types of information about a language (putting aside for the moment issues of information gathered incidentally about the speakers) that may arise in the course of such activities, and may be very interesting scientifically, but may also be sensitive for the community of speakers? I am thinking especially of cases where there may be a small-to-medium-sized group of speakers, many or all of whom know each other, but who may not all agree on (a) how best (or whether) to document/preserve/revitalize their language, or (b) on matters of the language itself.

Here are a few situations that I can think of off the top of my head; I have not come across any of these specific situations myself, so please comment if you have better or other examples.

(1) A consultant is seen to have more authority on the language than his/her fellow speakers. Possible issues:
-Other speakers do not volunteer contrary opinions, even though they may use the language differently; data may differ greatly between natural conversation and response to elicitation;
-Other speakers do volunteer contrary opinions, but the primary consultant disagrees with them (and may or may not express wishes about what to do with these contrary opinions).

(2) A consultant is seen to have (or sees her/himself to have) less authority on the language than other speakers. Possible issues:
-Data from this speaker are inconsistent between naturalistic production and elicited production;
-This speaker feels uncomfortable either with naturalistic or elicited data collection.

(3) Community members who speak the language as a second language have more authority in the community than those who speak it as a first language; the researcher finds that the data are inconsistent between the two groups but the community defers to the second language speakers’ opinions when asked.

(4) Speakers identify certain data points as belonging to different dialect groups and the researcher does not find that the data fall into such groups; OR, speakers do not identify into separate dialect groups, but the researcher finds such patterns in the data.

And so on. Has anyone dealt directly with any situations similar to these, either successfully or unsuccessfully? What resources have people found useful for approaching such situations? I’m interested here not necessarily in how to deal with potentially difficult social situations in fieldwork, but in what our ethical obligations or responsibilities are as language scientists, in terms of publishing and drawing conclusions from data collected from such situations (assuming that we have permission to do so in the first place).