While I agree we do a lot of our problem solving with symbolic languages (streams of images), even if we define "thought" as symbolic language processing, I believe many great experts in philosophy and law do internalize the relationships between concepts and operate on it on a more subconscious level to get there faster, going back to the symbolic language to validate their reasoning processes.
I wouldn't call those underlying processes "thinking", but it is a matter of definition.
This is also why those who just use LLMs to write those court submissions we've read about fail: there was no non-thinking reasoning happening, but just a stream of words coming out, and then you need to validate everything, which is hard, time-consuming and... boring.
Another issue is that a lot of tasks in the modern world are rooted in language, law or philosophy is in large part just word games, you won't be able to get far thinking about them without language, as those concept don't have any direct correlate that you could experience by other means.
Overall I do agree that there are plenty of problems we can solve without language, but the type of problems that can and can't be solve without language would need some further delineation.