At least, in my field, there are at least two ways in which one can be driven in research. The way I see it, in Linguistics, one can either be language-driven or theory-driven.
Language-driven research are mostly true when it comes to fieldworkers. These are the scholars that are so fascinated with one language or language family. They establish field sites and work with the speakers for basically the rest of their career. They become known as the experts of anything that has to do with this language, whether it is about the sound system of the language, or the structure of the language, or the social aspects of the language. The disadvantage for scholars like these is that one needs to cover a lot of theoretical background, since basically, one is responsible for the various different aspects of the language.
Theory-driven research on the other hand are done by scholars who are interested in a small set of phenomena or research questions. Then, in order to investigate the research question, they find what ever language it is that can help them solve that question. They know bits and pieces of various languages, but only the pieces that are relevant to what they are interested in.
So, who am I? I suppose I belong to the latter type. I am interested in a set of research questions, and I am investigating that using the language that can provide me with answers. I know a lot about the topic that I am interested in, and I know stuff in different languages that relate to this research topic, but I know nothing about other aspects of the languages that I look at that are not relevant to what I am interested in.
Of course, there are pros and cons to the approach. And with this, the issue also comes up in what type of dissertation one would write. There are two possibilities.
One can write an "aspects" dissertation. This is the dissertation whose title usually goes along the lines of "Aspects of [insert big topic here]". A lot of dissertations that are based on fieldwork targeting a specific language is like this. Perhaps this is not what I want.
On the other hand, one can write a dissertation, where there is one grand thesis, and the chapters are basically points that lend evidence to this grand thesis. Thus, this is the "big idea" dissertation. I've seen several dissertations like this, and yes, I have to say, they are very logically sound.
Now, I have come to realize that the current state of my dissertation research needs to be improved. It needs to be reined in, so that it has one grand theme running through. And I think that is what I need to do from now on, to have a grand theme, and every family of experiments that I am going to propose will basically lend evidence in support of my grand theme.
The good thing is that my adviser told me that the grand theme doesn't need to be earth-shattering: in fact, only a few people do earth-shattering theories, theories which result in paradigm shifts and a change in world view. Charles Darwin and Noam Chomsky comes to mind as examples.
So there. I think I do have a grand thesis for this dissertation, therefore, the things that I would be talking about aren't just random clutter but instead revolving around a common theme trying to prove a common thesis.