The verb is attested in two different phonogramic forms, 沙音 and 三, both of which are uncontroversially reconstructed as *sam-.
沙音 (*sam-)
When expressing the sense of "to take A as B", Old Korean appears to have natively used the expression "A乙 B沙音", with A taking the accusative particle 乙 (*-ol) and B combining directly with the verb *sam- without an intervening particle.
Gugyeol texts are all close translations of a Chinese original, and the Literary Chinese text often used 以 (yǐ, “to use”) in the construction "以 (yǐ)A為／为B" for the sense of "to take A as B". These constructions were translated into Old Korean as "A乙 以厼 B沙音", with A continuing to take the accusative particle but now becoming the object of the verb 以 (*PSU-) instead, with the literal meaning "to take as B while using A".
In Middle Korean, however, "to take A as B" is generally expressed as "A로 (-lwo) Bᄅᆞᆯ (-lol) 삼다 (samta)", with B taking the accusative particle ᄅᆞᆯ (Yale: -lol) and A now taking the instrumental particle 로 (Yale: -lwo). An Ye-ri theorizes that this was the result of syntactic influence from the Chinese "以 (yǐ)A為／为B" construction, as Chinese 以 (yǐ) was usually perceived as being equivalent to the Korean instrumental particle 로 (Yale: -lwo). Thus speakers may have initially imitated the Chinese syntax by using the instrumental particle for A rather than the accusative, after which B was reanalyzed as being the direct object of 삼〯다〮 (Yale: sǎm-tá).