Abstract
This study proposes an extended FrameNet approach for the description of connectives. The meanings of connectives are described with respect to the two frames evoked by each of the conjoined clauses, whose combinational patterns are termed “frame valences”. Taking the English polysemous connective while as an example, features of each meaning were statistically analyzed based on the frame valences using correspondence analysis. The correspondence analysis has revealed that in the contrastive use the same frame tends to be evoked in the conjoined clauses. To test this result, this study has further examined the contrastive connective whereas, which has firmly supported the results of the correspondence analysis and shown that frames that are closely related via ‘frame-to-frame relations’ can be evoked in the contrastive uses. These findings, in turn, corroborate the validity of the frame-based approach for the description of connectives.
