I think the key to understanding Aristotle’s view on most epistemological subjects is that he doesn’t yet distinguish metaphysics and epistemology, or at least he doesn’t realize the import of this distinction. That’s not where his focus is. In part the reason is that a certain amount of method (of logic) needs to be developed before one can really start thinking about what its status is.
For example, in The Parts of Animals, Aristotle is concerned with the fact that birds and fish don’t form a single kind, and that the species of fish ought to be defined by a certain method. But he is just not concerned with the issue of whether these classifications and definitions are metaphysical (intrinsic) or epistemological (objective) – whether they exist independently of us, or whether they are the ways we need to classify and define in order to achieve knowledge.
Sometimes he does talk as though they are epistemological, but other times, and more often, he speaks about them as though they were metaphysical. I don’t think it’s right to say that he had a position on this issue. I think he simply didn’t see it. He is a realist by a sort of default. He argues against Plato’s view and he takes it as obvious that concepts have a basis in reality. But the distinction between their being real and their having a real basis, simply doesn’t come up for him. So he’s not explicitly on one side of it or the other. But, I think, in the end, it is correct to classify him as a sort of qualified case of a moderate realist. I don’t have time to argue for this here, but I will cite some of the relevant text for people who want to pursue the issue further.
The best example of Aristotle glimpsing the objectivity of concepts is the first book of the Parts of Animals. Here he is concerned with biological methodology including issues of how to classify and study animals and their parts. He is talking about how we should classify and is on the premise that the human mind is active in grouping, that there is a basis in reality for grouping, and that groupings can therefore be evaluated. Chapter 4 is the essential chapter here. He says that people (properly) “bring together” species animals under “one kind” (e.g. “bird”) for a reason, viz. that they differ “in the more and the less” rather than being the same only by analogy (as birds and fish are similar). This is, BTW, the closest he ever comes to measurement-omission, and this doctrine of difference in the more and the less is important to his biology. But equally interesting to is that on his view: *people* seem to do the “bringing together” and Aristotle is here interested in the *norms* that they should follow in doing so. The point isn’t that you should name the real universals that you find, but rather that you should bring things together when they differ only in the more and the less. Other parts of the book strongly imply that the reason why we should classify and inquire in certain ways rather than others is that it will make it possible to study and learn about things economically – if, he thinks, we studied animals one by one, rather than studying the parts that are common to many animals (e.g. stomachs) we would needlessly reduplicate our efforts.
But even in the Parts of Animals, A. is talking about why trout and bass are classified as fish, not why a certain individual qualifies as a trout. The reason why a certain individual is a trout (or a member of whatever the most indivisible species is) seems to be that it shares a form with other trout. (Although re: forms and individuals, see 644a24 and the interesting comments that Balme and Lennox make on it in their respective translations of the PA).
The more traditional places to look for information on universals in Aristotle include Posterior Analytics II.19, and Metaphysics I.1. In both chapters he talks about the progression of knowledge from sense-perception to apprehension of universals. Also the chapters from De Anima III that Daniel cites in his paper. You might want to look at the beginning of A.Po. I.11, which could be taken as an endorsement of moderate realism. There are a number of other discussions of the issue in the A.Po. and the Topics, mostly in passing. Once can find them using a decent index.
Categories 5 is useful for the claim that no substance is *in* another substance. This means that man (which Aristotle calls a “secondary substance”) cannot be *in* an individual man, it is not a *part* or *attribute* of him as some moderate realists think. Rather there is a distinct relationship between the universal man and particular men (the “said of” relationship). But is this a relationship between two mind-independent things? I don’t think there is a clear answer there. That’s not the question he’s focused on.
Depending on how in-depth one wants to go, and how far one wants to get into the issue of definition, the A.Po. as a whole and the Topics are very relevant, but it is hard to narrow it down to individual chapters, especially with the A.Po. Much of Metaphysics Eta is also relevant, though, here the focus isn’t on universals vs. particulars per se.
It is worth noting that Aristotle doesn’t really have a word for “concept” and he doesn’t talk much about “universals” in the since in which later philosophers use that term. (Again, signs that he’s not focused on this issue, per se.) Most of the discussion in the passages cited is about “what things are” or what their forms or essences are. In one (maybe the) use of the concept “universal” it refers primarily to a sort of relationship. Certain things hold “universally” of others (see A.Po. I.4). And it is important to identify things that hold in this way for the purpose of demonstration. This context colors all the discussion of universals in the A.Po., and perhaps elsewhere as well.
All of these citations are pretty much as they occur to me, so I might be leaving out something important. But they should be a good starting-point.