Syntactic Analyses for Parallel Grammars: Auxiliaries and Genitive NPs

Syntactic Analyses for Parallel Grammars: Auxiliaries and Genitive NPs Miriam Butt and Christian Fortmann and Christian Rohrer institut fiir Masc.hin(:ll(: Sprachv(:rarbeitung Universil, St Stutl;gart Azenbergstr. 12 70174 Stutl;gart, Germany { mul, t I for(,man n ]rohrer } (c~i ms. uni-stu (;l;gar t.de Abstract This paper tbeuses on two disparate as- I)e(:ts of German syntax from the per- speetive of paral]eJ grammar develop- menl;. As part of a eOOl)erative project, we present an innovative approach to auxiliaries and multiph: genitive NPs ill German. The LFG-based imph:men- ration t)resented here avoids unnessary structural eonq)lexity in the representa- tion of auxiliaries by challenging the tra- ditional analysis of auxiliaries as raising verbs. The approach developed for mul- tiple genitive NPs provides a mor(: at)- straet, language independent representa- ti(m of genitives associated with nomi- nalized verl)s. Taken together, the two ai)proa(:hes rei)resent a step (;owards pro- viding uniformly al)plical)le treatments for differing languages, thus lightening the burden for machine translation. 1 Introduction Within the cooperative parallel grmmnar project PARGRAM (IMS-Stuttgart, Xerox-Pale Alto, Xerox-Grenoble), the analysis and representation of stru(:tures in the grammars must bc viewed from a more gh)bal perspective than l;hat of tile individual languages (German, English, French). One major goal of PAll.GRAM is the development of broad coverage grammars which are also mod- ular and easy to nm.intain. Another major goal is the construction of parallel analyses for sen- tences of the same type in German, English, and French. If this can be achieved, the prol)h'm faced by machine translation (MT) could tm greatly re- duced. Due to the recent developnmnt of a faster and more powerful version of the LFG (Lexical- Functional-Grammar) based Grammar Writer's Workbench (Kaplan and Maxwell 1993) at Xerox, tile implementation of a linguistically adequate, broad coverage grammar aI)pears via.tile. Given the flexible t)rojection-based architecture of LFG (l)alrymple el; al. 1995) and the MT approach pre- sented in Kaplan et al. (1989), 1 a rolmst MT sys- tem is already in place. In this paper, we (:oncentrate on two issues within the broader perspective of I>AI{GII.AM: the treatment of auxiliaries and the transi)arent rep- resentation of multilfle genitive NPs in German. 'these phenoinena I'eI)resent two are.as for which generally accepted proposals exist, but, whose iln- plelnentation in the context of par'aiM gram- mar development throws up quest;ions as to their wider, crosslinguistic, feasibility. With restleet to auxiliaries, the standard raising at)preach that is usually adopted yields undesirable structural com- plexity and results in idiosyncratic, language par- ticular analyses of the role of auxiliaries. With regard to genitive NiPs, the standard analysis tbr German yields strtlctures which are too alnbigu- ous for a succesflfl at)plication of madfine transla- tion. The fbllowing sections present a solution in that morphological wellformedness conditions are. stated at a separate eonlponent, the 'morTfl~ology projection. Furthermore, a representation of argu- inent structure is implenmnted that is related to, 1)ut not idcnl;ical to the ret)resentation of grain- matical flmel,ions. Language particular idiosyn- cratic requirements are thus separated out from the language universal information l(:quired for …

Content truncated to 500 words