The role of emotions in ethics is often taken by philosophers and others as antithetical to rationality. On the most basic level (in undergraduate philosophy exams and elsewhere), stating an opinion in the form "I feel that p" can be a way of sidestepping the demand for reasons. But emotions can sometimes also be seen as supplying reasons for moral judgment to the extent that they involve evaluations--and a way of communicating them across different moral perspectives.
I think that typical human emotions are best viewed as having two connected evaluative aspects, affective and conceptual. First, to the extent that emotions are built on positive and negative feeling, feeling of a sort that we naturally find rewarding or aversive, they can be understood just at that level as "saying" something evaluative about their objects, representing their motivational significance for the agent.
Sadness evaluates something negatively, for example, in the first instance just by registering the thought of that object of attention in unpleasant feeling. It amounts to feeling bad or uncomfortable about that object--perhaps just something indefinite; the surrounding situation. We might call this sort of negative "gut feeling" an affective evaluation.
This first aspect of emotional evaluation is something we can attribute unproblematically to infants and to nonhuman animals that seem to be capable of emotion. It's also something humans can pick up from each other without much prior understanding of the content of an emotion or the grounds for it (if any)--just by imitating the bodily responses of agents in one's surroundings, for the less reflective form of empathetic transfer known as "emotional contagion."
Typical adult human emotions also involve evaluation in a further sense, however, with explicit evaluative thought content. Feeling can have a proposition as its object: that something has turned out badly, say. This second, more complex form of emotional evaluation allows for distinctively moral emotions and for more sophisticated forms of empathy, mediated by moral imagination: putting oneself in someone else's shoes, in the standard image, reacting from another cognitive standpoint.
Imaginative awareness of harm to others can thus communicate negative feelings like sorrow by way of evaluative thought: a thought about how bad things are or would be for someone else. Ultimately, this process of identification with others gives rise to more complex feelings, on the order of emotional guilt, where we see ourselves as responsible for some harm done or contemplated and "simulate" others' responses.
We can see such emotions as involving evaluation on two levels: a negative feeling applied to a negative thought--the thought that one made things bad for someone, say. But of course it's important that emotions don't have to start out at that level of complexity, if they're to figure in a developmental account of moral judgment.
Moral emotions seem to be prior to moral thoughts; or perhaps one should say they're the original form of moral thinking. Consider how we teach a child the judgment of moral wrong. We react with various forms of disapproval to this or that lapse: "Bad boy!" said in an angry or disappointed tone, and the like. We're appealing here to the child's felt sense of ease with himself and others, as something communicable wordlessly from early on, via interpersonal acceptance or rejection, the allocation of positive or negative personal attention, or even just pleasant or unpleasant forms of holding and skin contact.
What ultimately gives us moral emotions like guilt is the fine-tuning and sophisticated social influence--the criticism and modeling of emotions as responses to various more specific objects--that language makes possible. But preanalytic emotional evaluation also survives into adult life--as the vague sense of uneasiness with a proposed alternative, say.
For purposes of analysis, this can be expressed as an indefinite propositional attitude--discomfort at the thought that something isn't quite right--in order to show that it's part of the evidence we appeal to in formulating or justifying a judgment. It's more than just a vestige of infantile response but can also serve as the bearer of important evaluative information--sometimes more accurate information than gets encoded in our developed moral thought.
Of course one shouldn't rest content with "I feel that p" as an expression of moral opinion. It may take some serious work to figure out what a feeling really amounts to--both philosophic analysis, or the attempt to specify an object that would give reasons for the reaction, and the kind of psychological inquiry needed to uncover hidden sources of feeling. My suggestion is just that feeling is often the proper starting point for rational ethical judgment--and sometimes even an appropriate substitute for it, under time-pressure, or with limited knowledge, in the heat of moral decision.