This article discusses the structure of double object constructions in English. It argues that Larson's analysis of double object constructions has issues related to anaphor binding and semantic relations between the two objects. The author points out that the double object construction was common in Old English, while the dative construction began to appear in Middle English. To address these issues, the author proposes that the double object construction is a basic form, and the dative construction is derived from it through passivization within VP. The article begins by introducing the asymmetrical relations between NP1 and NP2 in double object constructions, as demonstrated by Barss and Lasnik (1986). Larson (1988) proposes radical structures for double object constructions, suggesting that the structure underlying both dative and double object constructions is (3a), where V2 and PP form a predicate, and V'2 is the subject of NP1. The S-Structure of double object constructions is derived from (3a) via a process similar to passivization. However, the author argues that Larson's analysis has problems, particularly in accounting for data observed by Pesetsky (1992). For example, Larson's analysis cannot explain the grammaticality of (4b), where the position of Theme c-commands Goal. The author proposes an alternative structure for double object constructions that accommodates data such as (4)-(5) as well as (1)-(2). The author reviews Larson's analysis of double object constructions, highlighting the asymmetric c-command relationships between NP1 and NP2. The author then presents an alternative structure for double object constructions, proposing that the structure underlying both dative and double object constructions is (6a), from which the structure of a dative construction is derived by applying passivization within VP. The author examines the consequences of the proposed analysis and offers some final remarks. The article concludes that the structure of double object constructions is the one underlying dative constructions, reflecting the historical development of these constructions. The author also suggests that in the structure of double object constructions [V NP1 NP2], NP1 and NP2 form VP, whose head is a null have, e, denoting a possession relation that holds between NP1 and NP2. The structure for dative constructions is derived from the structure of double objects by passivization within the VP. The analysis proposed here can account for the difference in anaphor binding as in (4)-(5) and presumably the contrast in scope as in (45a, b).This article discusses the structure of double object constructions in English. It argues that Larson's analysis of double object constructions has issues related to anaphor binding and semantic relations between the two objects. The author points out that the double object construction was common in Old English, while the dative construction began to appear in Middle English. To address these issues, the author proposes that the double object construction is a basic form, and the dative construction is derived from it through passivization within VP. The article begins by introducing the asymmetrical relations between NP1 and NP2 in double object constructions, as demonstrated by Barss and Lasnik (1986). Larson (1988) proposes radical structures for double object constructions, suggesting that the structure underlying both dative and double object constructions is (3a), where V2 and PP form a predicate, and V'2 is the subject of NP1. The S-Structure of double object constructions is derived from (3a) via a process similar to passivization. However, the author argues that Larson's analysis has problems, particularly in accounting for data observed by Pesetsky (1992). For example, Larson's analysis cannot explain the grammaticality of (4b), where the position of Theme c-commands Goal. The author proposes an alternative structure for double object constructions that accommodates data such as (4)-(5) as well as (1)-(2). The author reviews Larson's analysis of double object constructions, highlighting the asymmetric c-command relationships between NP1 and NP2. The author then presents an alternative structure for double object constructions, proposing that the structure underlying both dative and double object constructions is (6a), from which the structure of a dative construction is derived by applying passivization within VP. The author examines the consequences of the proposed analysis and offers some final remarks. The article concludes that the structure of double object constructions is the one underlying dative constructions, reflecting the historical development of these constructions. The author also suggests that in the structure of double object constructions [V NP1 NP2], NP1 and NP2 form VP, whose head is a null have, e, denoting a possession relation that holds between NP1 and NP2. The structure for dative constructions is derived from the structure of double objects by passivization within the VP. The analysis proposed here can account for the difference in anaphor binding as in (4)-(5) and presumably the contrast in scope as in (45a, b).