Recent research shows that sound change (pronunciation shift) in

Sound change is not gradual.
Classical account needs it to be gradual.
So classical account must be discarded.

Conclusion: Sound change theory must be discarded.

Note that there is a gap here - 'classical account' is discussed in premises but conclusion talks about 'sound change theory.' So the assumption must connect the two.

(A) The data on which the classical account of sound-change theory was based are now known to be inaccurate.

The premises already tell us that classical account must be discarded. This option doesn't link classical account with the theory.

(B) The emergence of new sounds appears to be random.

We know it is sudden. Whether it is random or not, is out of scope.

(C) The meeting of linguistically disparate cultures can affect the sound of their languages in unpredictable ways.

No discussion on cultures.

(D) All theories of sound change rely heavily on the classical theory.

Correct. It says that all sound change theories rely on the classical account. So it makes sense that if we are discarding classical account, we should discard sound change theories too.

(E) For most languages, historical records of their earlier stages are scarce or nonexistent.

Doesn't fill the gap between premises and conclusion.

Answer (D)