I've given it some more thought, and the only thing that comes to mind is that I'd really like to let either a character or a narrator in a story reveal that they are biased or even lying without actually letting anyone know that they are. You would have to figure it out by their words or wording alone.* So I'm wondering if there are any branches of linguistics that deal with that sort of thing? How or why people reveal things without meaning to?
Ah, I see. There are many possible approaches to this question. You could go the cognitive linguistics route. "Framing" would be a useful concept (Lakoff, via fame semantics), as would be "Profiling" (Langacker).
Framing means that when you use a term, you're also invoking a frame of meanings that gives rise to the term. Thus, depending on what terms you use, certain things are harder to argue than others. For example, I'm an atheist. But by calling myself an "atheist" I'm already ceding ground to theists, by invoking what I'm not - a theist, rather than exploring what I
actually believe and giving a non-theist-coloured term to it (such as "naturalist" or "materialist", each of which again conjurs up different contexts and may or may not be to my advantage.)
Profiling means something similar: A word/construction/... usually profiles something on a base. This is rather complex, and I'm not an expert. But, for example, Langacker says that a verb profiles a relation on the base of its participents. I think this means that the "I receive a present," "I am given a present," and "Joe gives me a present," all refer to the same state of affairs, but with different profiles:
"Receive" profiles the relation between recepient (me) and object (the present), leaving out the giver (Joe). "Give" profiles the three-way relation between the recepient (me), the object (the present) and the giver (Joe). Passive voice (as in "I was given a present,") shifts the "landmark" and reverses the "trajectory", but still retains the the profiling of the giver (Joe). (This is why amateur critics often talk about how the passive voice obscures the agent, while a verb like "receive", which does the same thing, gets away with it: The profile draws attention to what's not expressed.)
***
You could also approach the question via pragmatics.
The Gricean maxims of conversation come to mind. Basically, it's about what about a situation you think should be said. The maxims are Quality (Say the truth), Quantity (Give neither too little nor too much information), Relation (Be relevant), and Manner (be clear). If you think one of these maxims is violated, you may have detected a difference in perspective. An obvious example: "I've fired my homosexual gardener." You know he only employed one gardener. There is no question he's a bigot, really. Why else would that adjective be in there, than to indicate the reason for firing the gardener. Most statements are not as obvious.
***
There are pitfalls: Take
this article about rape and responsiblity attribution, as concerns the use of the passive voice. There's an interesting discussion of ambiguity, and then you get to the methodology. This includes this little gem:
Bohner said:
Nominal forms (e.g. ‘and then the rape occurred’) were coded as passive because of their impersonal nature.
Ah. If nominal forms were coded as passives because of their impersonal nature, wouldn't it have been more straightforward to code both passives and nominal forms as "constructions of an impersonal nature"? Why, then, does this run under the heading of "passive voice". Also, why do they not separate agented and unagented passives (if it is the impersonal nature they are after)? A sentence like "And then she was raped by that castration-deserving arsehole" is very different in responsibility attribution from something like "Yeah, well, she got herself raped." (Notice how I'm exaggerating extremes to make a point here? What does that say about
my position? Go and google my post history, and you will find lots of rants of mine targeting the silly [<--aha!] rule against passive voice....)
This is getting convoluted now, isn't it? An article that seeks to find out bias, gets targeted by me for it's own bias - in biased language... So what's my point? Be vigilant. Don't even trust studies. Interesting new concepts are
especially seductive.
Just a little taste of what a difficult subject you're interested in. (But it's soooo interesting, isn't it?)