“Should my discussion be ‘theoretical’ or not?”
Theory will always be abstract and use abstract concepts. What is important is only that they be clearly and precisely defined, and that their usefulness in explaining phenomena in some domain is exhibited or can be. The interdependence of theory construction and close observation of empirical or textual data will be realized partly in the use of both abstract nouns and concrete descriptive adjectives. In the humanities in particular today, these are the two poles of much discourse and knowing how to work with them is one of the keys to good writing. With theoretical discourse, there is a danger often of jargon, which is using concepts that are either clichés or used casually, without sufficient attention to the actual meaning or range of meanings of the term, so that the discussion becomes obfuscatory rather than illuminating. But a purely concrete discourse would not be a solution in most cases. It would run the risk of turning a careful scholarly study of a work into an evaluative review fit for a newspaper. The writer's complaint that theoretical terms are disconnected from concrete things is simply and wholly mistaken. Take, for example, the word "infinity." It has been given more than one definition, but it is a metaphysical and mathematical concept that cannot be reduced to some example, just as it does not derive from the effort to some conceptualize some particular thing. The same would apply to, say, "power" or "ideology," both concepts in political theory, the former also one in metaphysics and physics (it is related to the concept of force).
I suggest that this is true of most abstract concepts. Take that of number. You may think that we understanding things being numbered and countable because we count things like this stone and that one and then generalize from stones to anything that can be countable, and from the marks of the counted (1, 2,...) the idea of number. I think it can shown that is false, and that the concepts of number and counting are abstract and, however they arose historically (maybe a God hit a budding mathematician on the head with stones in succession, and then he had the idea of number), logically the concepts cannot be derived from what is in fact their use. If this is true, and if philosophical metaphysics is as close to pure mathematics as it was at their origin (in ancient Greece), this may shed some light on the necessity of theory.
Theoretical abstractions are not only shorthand; they are a feature of language that make it possible for us to conceptualize what we otherwise could only observe and describe. For example, if I understand by "infinity" the limitlessly countable, than I must imagine something I cannot observe or experience: a constructible infinite series of countable things. I suggest this is true of all concepts. "Blue" does not merely name a particular visual sensation, nor even a field of sensations that can be grouped together: it names a rule for identifying such sensations, and without concepts serving as such rules, we would never get there. We could not reason, because reasoning requires concepts. And so we could not write or read argumentative essays and texts, or understand anything that requires a demonstration. Demonstrated in arguments is not just what is supported by evidence or a reason and so is true; it is also what we mean by the words we use, and these are related.
I don't see a difference in kind but only one in degree between everyday concepts and abstract theories.
Why are abstract concepts so useful in science and scholarship, in some fields particularly? Because we want to interpret phenomena and explain them. Empirical science does this with testable hypotheses, the humanities do it in a way that is freer and arguably makes use of concepts almost as if they were no different from adjectives used to describe apparent qualities, but with no semblance of scientific rigor. Philosophy develops such concepts independently of any field of derivation or application; what we call "theory" in the humanities draws on philosophy but is "impure" and uses observable facts to test theories and theories to interpret and explain them. Often the best work in studies of literature and the arts aims to develop a new theory based on the data of the matter being studied. This is also true in the social sciences, at least at their more interpretive, as in most anthropology and much sociology.
People who manage to write pure theory in the humanities are essentially doing philosophy, and it's not a bad thing that some people do this, especially in this country, where what is called philosophy in Europe is otherwise ghettoized.
There are humanities scholars who eschew theory (or claim to). This must be weighed as an option for its pros and cons. If the art work or text being discussed is specific to a place and time as is always more or less the case, scholarship that makes no broad theoretical claims risks its relevance being limited to scholars of that place and time or even that work.
The prejudice against theory (and philosophy) that many Americans have must be considered as part of the general reflex of anti-intellectualism in this country, perhaps with a dash of defending humility against arrogant folly thrown in.
The claims against theory can be disproven simply by exhibiting illuminating discussions of texts and works that involve the deduction from observation and evidence of abstract claims that can at least potentially be given wider application by future scholars, as sometimes happens. But that also means one may want to import concepts.
Take, again, the concept of "power," let's say in Foucault's understanding of it. You are a sociologist describing some interaction in some institutional context. You find it useful for your thinking to make reference to Foucault's concept. Of course you footnote a working definition. You show how it helps to explain what you are studying, you wonder whether Foucault's understanding of how power works can be made more precise, particularly if you find some ambiguities or lacunae in it, to which you may devote some paragraphs analyzing. You think you can go a step further and add to our understanding of how power works, let's say in public hospitals in psychiatric interviews with low-income Hispanic female patients. You will have other factors to account for, obviously. Your discussion of the work of previous theorists is just properly scientific.
Now, let's say you are discussing the paintings of Jean-Louis David. Among the many things you might discuss are his representations of social power relations. Foucault's concept may or may not be relevant; a close reading of the paintings will determine that; you should, as always, look very closely at the work and aim to do so without prejudice. Don't use any theorists' concepts just because they are fashionable, or because you are lazy; use them if and only if they help you clarify and analyze what is at stake in the paintings for your readers, including the community of scholars who know these works and the literature on them.
Most people agree that in humanities and social science scholarship there is a principle of the primacy of the object (the text, artwork, situation, etc. that you are trying to understand). A correct understanding of this also sheds light on how the interpretive social sciences and the humanities are closer than one might think, and therefore also on our ideas of what is a science and how scientific and artistic thought are related.