Abstract
LFG has been widely used to analyze English language as well as other languages from linguistic point of view [Joan Bresnan 2001; Louisa Sadler 1996], including Chinese [Lian-Cheng Chief 1996; One-Soon Her. 1997]. A new direction in LFG research field is applying it to language computation, ranging from parsing to machine translation [Louisa Sadler, Josef van Genabith, and Andy Way 2000; Mark Johnson 2000; Miriam Butt, Stefanie Dipper, Anette Frank, and Tracy Holloway King 1999]. However, the LFG-based work in Chinese computing is rather rare [Lian-Cheng Chief, Chu-Ren Huang, Keh-Jiann Chen et al 1998].
The current framework of LFG shows two folds when being employed in Chinese computing tasks: it is quite powerful for linguistic representation, but seems not to be strong enough for Chinese computation—here exists some room for improving the formalism of LFG. This paper will center on these two issues.