The problem with this approach is that anecdotal evidence isn’t much use at proving anything.
Let’s look at your orginal proposition:
I assume by “fitness indicators” you mean something akin to the peacock’s tail. The tail confers no survival benefit on the peacock, but it does advertise to peahens that its owner is healthy enough to support such a useless extravagence.
I don’t see how emotions fit this framework. Some emotions (fear, disgust) have an obvious survival value. Others like love and anger may be harder to explain, but they still don’t seem to work the same way the peacock’s tail does. Is a towering rage sexier than a fit of pique? Where is the selection pressure?
Furthermore, while people can hide their emotions, they do it more through mental self-control than through language. (Keeping a “poker face”, for example.) In fact, talking can often make it harder to hide your emotions – think of the edge of anger creeps into your voice when you talk to someone you dislike, or the way a mourner may hold his feelings inside for hours, only to break down when he tries to speak.