Persian (Iranian, Indo-European) may be a counterexample (Keenan, p.c. Givón, see Givón 1975: 122)There are so many counterexamples that I doubt that such a claim has really been made. See quotes below for possible clarification.
1. Non-accusative object in the terminology of Givón means “indirect object”.2. “One striking feature of direct object is that it shares the case marking properties of the accusative, but only when the accusative is UNMARKED.”3. “It seems, thus, that it is the unmarkedness of the accusative which makes it such a natural promotional case, and this is only natural: by promoting other objects into the same case of the unmarked accusative, they lose their original semantic case but do not gain another SEMANTIC case. But this is precisely what happens in promotion to subject …”
1. Non-accusative object in the terminology of Givón means “indirect object”.2. “One striking feature of direct object is that it shares the case marking properties of the accusative, but only when the accusative is UNMARKED.”3. “It seems, thus, that it is the unmarkedness of the accusative which makes it such a natural promotional case, and this is only natural: by promoting other objects into the same case of the unmarked accusative, they lose their original semantic case but do not gain another SEMANTIC case. But this is precisely what happens in promotion to subject …”