@article{oai:nagoya.repo.nii.ac.jp:00027994,
 author = {Niwa, Satomi},
 journal = {IVY},
 month = {Oct},
 note = {There have been various approaches ro explain how temporal interpretation is established, especially for complex sentences. In English, there is a phenomenon called Sequence of Tense (SOT). (1) a. John said that Bill was sick. b. John said that Bill is sick. c. Mary said that he told a lie. (1a) is an example where SOT is applied. There are two possible interpretations for (1a): Bill's being sick occurs at the same time of John's reporting it and Bill's being sick occurs before John's reporting it. (1b) is an exception of SOT. (1b) means that Bill has been sick since John's report or since sometime before John's report. To give account for how these readings are derived, Hornstein (1977, 1981, 1990) has introduced three entities that denote time: E(vent time), R(eference time) and S(peech time). Stowell (1993a, b) considers the E and R to be a thematic role given to arguments of T^0, and, introducing a new projection ZP, temporal interpretation mechanism is accounted for by binding. In this paper, inspired by the suggestion in Stowell (1993a, b) that tense has something similar to pronominals, and with further analysis of semantic properties, a suggestion for a possible temporal interpretation mechanisam is made. Although (1a) and (1c) have the same structure, the reading that the embedded event occurs at the same time of the matrix event is not available for (1c). The difference lies in the semantic type of the predicates. Being an individual-level predicate functions as mapping the temporal location of the event as a duration on the axis of time, while a stage-level predicate functions to map the event as a "point". There are other factors that affect temporal meaning of a phrase. Temporal adverbs make the temporal location of an event more precise or detailed. In English, discourse imposes restrictions on V for its selection of proposition. By reflecting on the identity of each tense, it seems that tense is similar to pronominals in some ways. It has its own identity, bur it can be partially specified. Temporal identity can be gained through comparison with other temporal items. Pronominals have their own identity, but not fully specified, compared with R-expressions. The identity of pronominals can be obtained through binding. Thus, for temporal identification, I proposed a mechanism with binding. Following Chomsky (1999), I assume that syntactic, semantic, and phonological computation is completed phase by phase. Semantic features add up within a projection to establish the meaning of the projection as a whole. Temporal semantic features gather within a projection and percolate up along the same line. Then only V of a complement clause can enter a binding relation. This mechanism can provide a proper account of how each reading is derived for each construction.},
 pages = {83--108},
 title = {Temporal Identification through Binding},
 volume = {36},
 year = {2003}
}