I spent a formative part of my first career doing dialect fieldwork in an effort to document conservative forms of Chinese that would soon disappear. In the Spring of 2010, I was asked how to pick a location that would be useful and find a really good illiterate informant. (For many Western fieldworkers studying unwritten languages, illiterate speakers are preferred.) Below was my response.
As for "useful", it depends on what you want to use it for. I went to the places I did in order to answer a specific set of questions about issues in classification. But my view is that any location is fine, provided you have superb informants there and know how to make full use of them. The informant is always the limiting factor. If I felt myself free to continue doing fieldwork, I would work with any appropriate person I could find, regardless of what they spoke. If you have a good informant, you can get good data, and with good data you can always find interesting things. So never mind the place — any will do.
I have not had the best luck with illiterate people in China. Basically, I think that literacy is such a central part of Chinese culture that anyone who has linguistic skills will learn how to read. It's quite a different matter with some of the minority languages.
There are definite character traits associated with being a good informant, and having a "verbal personality" is one of them. The ability to explain the same word consistently on different days, to enunciate tones clearly and be able to say whether two syllables have the same tones or not, to explain the functioning of grammar words and idioms — things like these are what we are interested in, and knowing Mandarin and how to read does not lessen the informant's ability to do them. But in my experience, someone Chinese who has these skills and has nonetheless not learned how to read — well, such people are few.
The personality and ability of the informant are the key to everything. I wouldn't insist on illiteracy.