> The explanation can't be that neat.

It's not an absolute, I was telling OP that, in fact, there was some research on what he mentioned. Basically, that memory is linguistic-context dependent, but as a subset of cognitive-context dependent (as well as physiological-, affective-, and several types of context). This doesn't mean that memories are ONLY linked to language, they have lots of different associations, things that work as a cue of the memory, this (language) is only one of them.

> You can have a stroke and lose language and still retain and form memories.

What does it mean to "lose language"? Are you unable to speak, to express yourself, to comprehend others, all together? What memories do you retain and form? Are you talking about semantic memories, or episodic memories? How do you measure that those memories are retained and formed, if you cannot test the subject due to the impossibility of communication?

It's a lot more difficult than absolutist statements.

Edit: typo (thinks->things)