Well that is a million dollar question. There are as many definitions of good interpreting as there are interpreters, researchers, institutions and clients.
First of all as one of my students wrote (quoting Patrick Kermit I think), the interpreter is the common language of the staff and the client. And that common language has to be understandable of course.
There are some statements you can make about good interpreting:
1. Interpreting is good when it works for the participants.
2. Interpreting is good when it conveys the meaning of an utterance from speaker to listener (you will have to define “meaning” though :-)).
3. Interpreting is good when it serves its purpose.
But these three statements does not allow my client to put on his or her headphones and immediately say: “That was good interpreting”. It does not enable the patient to judge whether I am interpreting the whole meaning – “everything” or not.
When interpreters take exams there are different ways to try to ensure objectivity in the light of good interpreting. The first thing is to hire a jury, several people judging the same thing gives objectivity, or at least inter-subjectivity. Another idea may be do decide that the interpreters MUST render e.g. 80% of all meaning-bearing units, and then you count…
None of these methods is of course water tight, and does it really ensure good interpreting? In Grenada in Spain there is a group of researchers, ECIS, who has found out that interpreting clients claim that one particular thing is important for quality (let’s say choice of correct word), but when the speeches are tweaked and one issue does not work in the interpreting (e.g.) intonation, then the client scores the interpreting lower and may even argue that the word choice was bad.
You may also have every meaning-bearing unit correct in your interpreting without producing any intelligible utterance.
Another group of researchers concluded that exactness in the interpreting and neutrality were less important features than trust for community interpreters. The clients were most pleased with interpreters they felt they could trust, interpreters who took an active role. Very little was mentioned on exactness of the interpreting.
So, to sum up; the definition of good interpreting is not something that all parties necessary agree on, but good, or the lack thereof, interpreting is something that affects all parties.

This post is part of a list, 30 days of interpreting. You can see the whole list here.