You are almost entirely correct.
All nodes validate as much Bitcoin data as they possibly can. All nodes avoid needing to trust other nodes as much as they possibly can. If one piece of data reaches ten thousand full-nodes, that data will be validated ten thousand times. Once by each. No node ever assumes that data it obtains from another is valid, regardless of whether it is new or old, first or last, early stages or late stages, Monday or Tuesday, day or night, raining or sunny. All nodes are untrusted.
Miners create new blocks. A new block, if valid, counts as one confirmation of the transactions in it and an additional confirmation of all transactions in all earlier blocks since the beginning. Each node validates the new block for itself and counts confirmations for itself.
Is it correct to differentiate between a validated bitcoin transaction and a confirmed bitcoin transaction?
Yes, a transaction may be valid but unconfirmed.
Validation is, for example, checking that the sum of outputs does not exceed the sum of inputs.
Miners produce new blocks but other nodes don't confirm those blocks. Confirmation is only a count of how far from the end of the chain is the block that contains a transaction. It is pretty much only an abstract measure of how much money a criminal would have to spend in order to persuade the world your transaction never existed.
Those are two very different things.
Am I correct, or very confused?
It is probably possible to be both correct and very confused. So it may not be an either-or proposition :-)