Dear list members,
I've wondered for a long time why sanskrit sandhi has any dependence on grammar and is not solely determined by preceding or following letters/sounds.

For example why should internal sandhi have any differences from external sandhi. As MacDonell says, "The most notable divergence from external sandhi is the unchangeableness of the final consonans of verbal and nominal stems before terminations beginning with vowels, semivowels and nasal

Or why should  ī ,ū or e when dual terminations (and only when dual terminations) remain unchanged before vowels.

In other words, in these cases why should grammar and not just adjacent sounds determine whether sandhi occurs.
Thanks,
Harry Spier