File Information
File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/intro/98/p98-2247_intro.xml
Size: 2,124 bytes
Last Modified: 2025-10-06 14:06:40
<?xml version="1.0" standalone="yes"?> <Paper uid="P98-2247"> <Title>Detecting Verbal Participation in Diathesis Alternations</Title> <Section position="2" start_page="0" end_page="0" type="intro"> <SectionTitle> 1 Introduction </SectionTitle> <Paragraph position="0"> Diathesis alternations are regular variations in the syntactic expressions of verbal arguments, for example The boy broke the window ~-. The window broke. Levin's (1993) investigation of alternations summarises the research done and demonstrates the utility of alternation information for classifying verbs. Some studies have recently recognised the potential for using diathesis alternations within automatic lexical acquisition (Ribas, 1995; Korhonen, 1997; Briscoe and Carroll, 1997).</Paragraph> <Paragraph position="1"> This paper shows how corpus data can be used to automatically detect which verbs undergo these alternations. Automatic acquisition avoids the costly overheads of a manual approach and allows for the fact that predicate behaviour varies between sublanguages, domains and across time. Subcategorization frames (SCFs) are acquired for each verb and a hand-crafted classification of diathesis alternations filters potential candidates with the correct SCFs. Models representing the selectional preferences of each verb for the argument slots under consideration are then used to indicate cases where the underlying arguments have switched position in alternating SCFs. The selectional preferences models are produced from argument head data stored specific to SCF and slot.</Paragraph> <Paragraph position="2"> The preference models are obtained using the minimum description length (MDL) principle.</Paragraph> <Paragraph position="3"> MDL selects an appropriate model by comparing potential candidates in terms of the cost of storing the model and the data stored using that model for each set of argument head data. We compare the cost of representing the data at alternating argument slots separately with that when the data is combined to indicate evidence for participation in an alternation.</Paragraph> </Section> class="xml-element"></Paper>