I've always wanted to do this with dolphins. Basically shift human language into the range of sounds they can produce, and have a reverse path so that sounds they make can be heard in a range a human could make. Then you raise a young dolphin in a 'bilingual' environment. With humans speaking to and playing with the calf while their speech is shifted, and any vocalizations by the adults or calf shifted into a range easy for humans to mimic back.
Basically my thesis is that understanding is possible and that it's facilitated by being able to produce as well as hear a given vocalization.
You may be interested in reading up on Margaret Howe Lovatt[1]:
> In the 1960s, she took part in a NASA-funded research project in which she attempted to teach a dolphin named Peter to understand and mimic human speech.
There are several articles and podcasts which can provide more detail.
Basically my thesis is that understanding is possible and that it's facilitated by being able to produce as well as hear a given vocalization.