According to WALS, most languages using SOV as basic order of subject, object and verb have some kind of personal agreement markers. As far as I know, these affixes rise by grammaticalization of clitic personal pronouns, when the clitics are adjacent to verb. However, that seems problematic to me, because in case of developing personal preffixes, object prevents affixation of subject pronoun to verb, and in case of suffixes, postion of suffix doesn't match position of its ancestral pronoun in SOV languages. I know two possible solutions of this problem:
earlier OSV word order - as far as I know, that is word order attested (or reconstructed) in eralier historical stages of Turkic and Mongolian languages.
SOV transformed to VSO, with clitic personal pronouns for both subject, and object, like in case of modern spoken French. Order of clitics and verb reflects former SOV word order.
Are these two explanations the only possible options? Or perhaps some other possibilites are attested, or at least theorized?