next up previous contents
Next: Lexical Redistribution Rules (LRRs) Up: System Overview Previous: Subcategorization frames

Blocks

Blocks are used to represent the tree substructures that are reused in different trees, i.e. blocks subsume classes of trees. Each block includes a set of nodes, dominance relation, parent relation, precedence relation between nodes, and feature equations. This follows the definition of the tree descriptions specified in a logical language patterned after Rogers and Vijay-Shanker[#!rogers-vijay94!#]. Blocks are divided into two types according to their functions: subcategorization blocks and transformation blocks. The former describes structural configurations incorporating the various information in a subcategorization frame. For example, some of the subcategorization blocks used in the development of the English grammar are shown in Figure D.2.28.1 When the subcategorization frame for a verb is given by the grammar developer, the system will automatically create a new block (of code) by essentially selecting the appropriate primitive subcategorization blocks corresponding to the argument information specified in that verb frame. The transformation blocks are used for various transformations such as wh-movement. These transformation blocks do not encode rules for modifying trees, but rather describe the properties of a particular syntactic construction. Figure D.3 depicts our representation of phrasal extraction. This can be specialized to give the blocks for wh-movement, topicalization, relative clause formation, etc. For example, the wh-movement block is defined by further specifying that the ExtractionRoot is labeled S, the NewSite has a +wh feature, and so on.

ps/lexorg/demosub.eps.gif
{Some subcategorization blocks

 

ps/lexorg/extract.eps.gif
{Transformation blocks for extraction

 


next up previous contents
Next: Lexical Redistribution Rules (LRRs) Up: System Overview Previous: Subcategorization frames
XTAG Project
http://www.cis.upenn.edu/~xtag