With reference to classification, it is an important skill in learning as well as in language retrieval. The reason that it tends to be suggested with reference animals/transport etc in a generic SALT clinic is that tends to appeal to the broadest spectrum of kids that come to SALT and resources for new equipment etc tend to be quite limited. Planning time for therapy targets is probably about 10 minutes a target when you are dealing with a clinic caseload so the opportunities for individualising it so that it makes sense in the context of the child are very limited, unfortunately. So out come the handouts.
In terms of isses with the idea of classificaton being non-functional, it depends on the view you take of how language intervention should operate, I guess. There's a 100% functional approach and the tackle-the-underlying-impairment-for-functional-gain approach.
Teaching classification based on your own specific interests is going to be more functional but in terms of school/life etc, the fact that you're not interested in farms or jungles or household furniture is a bit of a problem when you come across everyday vocabulary and don't understand it because you paid no attention to it because it doesn't interest you.
The way I look at this is that I (as an NT person) decided very early on that I had no use for Maths. None. All the options that I made in my entire academic career avoided anything computational. I found it difficult so it didn't interest me and I avoided it. Academically, this didn't affect me one whit. I have two first class honours degrees - one in English Language and Lit and one in Speech Therapy.
Unfortunately, however, I am now doing an MSc and seriously struggle with the stats. I find budgeting for bills difficult etc. I make stupid financial decisions.
The fact that I don't like something doesn't mean I will never need those skills.
In the same way, learning to categorise vocabulary so that you can retrieve it may not be a huge barrel of laughs but it is designed to have long term functional validity. It has bee a feature of every ABA programme I have ever encountered - object by function, categories, receptive-by-feature-function-class.
In terms of the long-term picture, what I have seen with students with autism at secondary is that they can have huge technical vocabularies related to their specific spheres of interest but massive gaping holes in semantic knowledge elsewhere. Functionally, this translates into them feeling overwhelmed by the content of lessons that bore them and "tuning out" and also contribute to misunderstandings in conversation with adults and peers.
In terms of teaching students with autism, when basic underlying conceptual relationships are missing it means that a huge amount of time is wasted revisiting and plugging in these gaps when you have a lesson, say, on ecology and have to learn information about different climates but you don't even realise that polar bears are animals. Given that students with autism have difficulty rapidly categorising based on incidental learning and experience, this is an issue worth addressing.
I have seen quite bright students stumble over seemingly "obvious" information.
I know that public SALT provision is terribly watered down and may seem nonsensical as a result as everything is so generic. However, it doesn't mean it's all crap y'know 