When voice leading in the eighteenth-century style, why is the melodic augmented second (e.g., B♮ to A♭ in the same voice) forbidden?
Over the years I've heard reasons ranging from "it's awkward" to "it's a hard interval to sing" or "it sounds bad," but none of these seem all that convincing to me; plenty of other intervals are more awkward, harder to sing, and perhaps sound worse, but they aren't forbidden.
In looking through three common textbooks, I see no clear explanation. I couldn't find clear logic in either Laitz (The Complete Musician) or the Clendinning/Marvin (The Musician's Guide to Theory and Analysis), and Gauldin (Harmonic Practice in Tonal Music) only states to avoid singing any augmented intervals (diminished intervals are okay, provided they are usually descending).
Is there a better—ideally acoustic, much like the rules against parallel perfect octaves—reason for this limitation?