File Information
File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/metho/96/c96-1092_metho.xml
Size: 20,828 bytes
Last Modified: 2025-10-06 14:14:11
<?xml version="1.0" standalone="yes"?> <Paper uid="C96-1092"> <Title>Applying Lexical Rules Under Subsumption</Title> <Section position="4" start_page="543" end_page="546" type="metho"> <SectionTitle> 2 Passive by Lexical Rule </SectionTitle> <Paragraph position="0"> It has been assumed in a variety of syntactic fi'ameworks that the active/passive alternation should be treated as a lexical process: Bresnan (1982) in LFG, Dowty (1982) in Categorial Grammar, and Pollard and Sag (1987) for IIPSG. German exhibits two types of passNes: personal passives, as in (lb), and impersonal passives, as in b. Der 'Poten wurde gedaeht.</Paragraph> <Paragraph position="1"> 'The dead were commemorated.' For personal passives the accusative object NP of a transitive verb, e.g. den Mann in (la), in its active form corresponds to an NP with nominative case, e.g. der Mann in (11)). In impersonal passives a datiw; or genitive NP complement of a transitive verb, e.g. the dative NP dem Mann in (2), exhibits the same case assignment in the active and passive forms. 'IS formMly capture the relationship between the l, Es for active and passive forms of German verbs in IlPSG, Kiss (1992) formulates the LRs in Fig. 1 (henceforth referred to as the PPLt{ and the IPLR, respectively), u The LRs in b'ig. I employ the HPSG feature geometry of Pollard and Sag (1994). a '\]'he specillcation of a syntactic category (CNI') in tlPSO >However, see Kathol (1994) and Pollard (1994) for an account of German passive without LRs.</Paragraph> <Paragraph position="2"> aIn order to reduce the size of the feature structures, prefixes of paths that begin with the SYNSEM attribute have been omitted as much as possible in Fig. 1 and all other feature structures that are shown in this paper.</Paragraph> <Paragraph position="3"> According to Kiss, specifying two different case wtlues under one reentrancy (cf. tags \[~ in the PPLR) is a shorthand notation for identity of the two categories in all respects, except for the case value. '\]'he original includes the feature VALENCE (abbreviated as VAL in lqg. 1), which in turn specifies for verbs under the features SUILI and COMPS the sub-ject and non-subject complements. COMPS takes a list of categories, (:ailed synsem objects, as its vahn;. If the list is nomempty, the h:ftmost cate gory in the list represents the direct object. The intended effect of the PPI, I{ is to promote the direct object of a. transitive verb in the l,l,\] for the actiw~' form to the subject of the passive form. l,'ol: impersonal passives the COMPS list of any transitive verb whose leftmost complement is marked by genitive or dative case remains unchanged, while the singleton list of the subject value of the active form becomes the empty list in the I,E for the passive form. Note that the case specifications on the left-hand side of the rules m:e crucial since they condition which classes of transitive verbs appear in personal and impersonal passives.</Paragraph> <Paragraph position="4"> Viewed procedurally, the PPI,I{, is meant to ap-ply to \[,Es for transitiw; verbs such as kau,\[en as shown in |+'ig. 2. 4 For transitive verbs it makes no ditthrence whether we use unitication or sub sumption as the test of ;,pplicability tbr the PP l, lC '\['he LE for any transitive verb is more specific than the feature stru<-ture of the input description of the PPI,I{, since the PHON wdue and the COMPS and SOl+;1 wJues will be further instandated than in the input description of the PPM~. In particular, the COMPS list of the PPLR is entirely schematic R)r any non-empty list of categories whose tlrst element is an accusative N P, while the COMPS vahle for kaufi'.n is a list with exactly one element that has the same category and case spc'cification as the NP in the PIq, R.</Paragraph> <Paragraph position="5"> Therefore, {;he LE will both unify with and be subsumed by the input Datnre structure of the IJPLI{. However, as we will see in the next section, the choice of subsumption or unification makes a crucial dilDrence when \[,Es are themselves highly schematic and underspecitled.</Paragraph> <Paragraph position="6"> tbrmulation of the I,Rs by Kiss (1992) differs from the version presented here in minor definitionM details.</Paragraph> <Paragraph position="7"> ttowever, these differences are entirely orthogonal to the theoretical issues discussed in this paper.</Paragraph> <Paragraph position="8"> 4l&quot;ollowing abbreviatory conventions in HPSG, the subscripted tags in Pig. 2 stand for the index va.lues contained in the CONq?I'\]NT specifications of the NPs. /,roduccd inl;o tJm I/I)SG lil.crt~tm:c l)y Ilinrichs nnd Nttka,znwa (1989) to account for the topica.lization of w, rhM c(mstituents aml ff)r the auxiliary ()CI'lll UI. flip ('onsl;ruction in a \]11 (~ornHtll VE\]I, BAI, (?()MI'I,EXLS~ i,c. lllai,l vcA'bs t.ogethcr with non-linitc auxiliaries, such as /e.~e~ k(';m~en c~m bc toldca.lize.d, ;ts show. iu (4).</Paragraph> <Paragraph position="9"> (4) l:'s('n kiinncn wird <'r cs.</Paragraph> <Paragraph position="10"> rca.d (:;m will hcit 'lie will be able rca.d il;.' In addition, the verbM complex serves as the, domain over which auxilim:ics c;u, t><', I'ronl,ed. This so-called AUXII,IAII.Y I,'1,11' c(ms(.rucl.ion i)osil.k)ns finite auxilia, rics such as &quot;,:ird in (5) to the M't in {,he vcrl)a.l COml>h',x , insl.~nd of 1,h<~ customa.ry scntencc-lina.I posit,ion for suhordim:d;c cla.uses. (5) lch ghmlm m(:ht, da.13 cr ~'s wird lose., kSnncl,. l bdiew, \[,ot tim*holt will read (:a.n '1 don'l; 1)clicvc th{:d; hc will 1)c a.I)lc 1;o I'e~M it. 'tbpica.lization {rod auxilia.ry-tlip, I, hus, t)rovidc cru(:ia\] evidence for i;re;~l;ing Ina, in verbs a.ud auxiliaries as constil,ucnLs. The prol)oscd consl;itueiH; sLrllcl;urc l'cquircs that, sul)ca.tcgorizal.io,l in for,w> lion M>out non-wM)al COml)h;m0,nts is l>rOl/agal;ed from the main verb to the l,op of the vcrlml comf>lcx. In IIPS(~ l,his can he achieved by structure shnring the complements of the ina.in vcrt) with the subcal;egorization in form;d;ion of each ~mxiliary in I;hc scnl;ence. '\['his lends to I,Es for auxiliaries such :'II:PS(~ pra.cticioucrs h~tve a,doptcd the notion of arg*tmcnt compositio*t to ~tcc:ou nt for a, wu:icty of syn-. L~t(:ti<: coimtrncl, ions in <lift>rent \]~ulgua,gcs, including citric-climbing in /t~di~m (Mona.chest :199;I), the syn-t~(:l;i(: properl, ies of ~tuxilia, rics in /g'cnch (Alto*lid a, nd (',oda, rd 1994), nnd (:ross-scriM dcpc'ndcncics in Dutch (ll.entier 1994). as kSnncn '<:~m' show. in Fig. 3.</Paragraph> <Paragraph position="11"> l l/) 0 i'd PIION (k6n.cn } \[IIEA I)1VI&quot;O I{.M hsc \] ' \[VAI,I(:OMPS (\[~+ ( V \[( oml's \[~ ) where \[\] - list(-, \[>'YNSI:M I ...IIIEAI) .... &quot;b\]) I:igurc 3: A l)brcvintcd I,E I'or auxiliary kSnue~l k&m, en requires a lmsc infinilJw~' coml)h;t~mnt~ a.s imli<'atcd in tl,&quot; (;OMITS vMuc in Fig. 3. The (JOMI>S value of kSnnc.n also conta, ins ~ (possibly cinl)t,y ) list o\[' non-w~rl)a.l categories (idcnl, ilicd by tim ta.g 1) that I, he gow~'rned w'xt) sul>ca.tc.goriz~s for. c; In other words, the COMICS vMuc of the governed vcrh is merged with the (7OMI'S list of t'gnncn itsclr. Formally this merging is nchic;vcd t)y al)l)en<ling the COMI'S list of' the governed v(;rh with l.h(~ olm-ch~mcnt list 1;hal. collsisl;s o\[' l.h(; govern('d v(;rh its('lF. Sin(:(' auxilia.rics Im.vc I.() bc able Ix) c(md)inc with dill'crcnl~ types of vcrl)s (%g. intra.nsitives, I, ra.nsitivcs, ditra.nsil.ivcs, etc.), the (',()M I'S lis~ oF I, hc governed w',rh has l,o r<,: main uudcrslmcilicd. It. is this undcrspccilication such as kgnncn and the ims,~ivc: I,I{ for (;erman, it malC/cs ~ <'.ruciM difforcncc whc(,hm' the I,l{ appli<'~s under unifica.|;ion or mMcr subsumption. Firs/, consi<h;r mliIical;ion as the tcsl, for nl, lJicability of' l,l{s. Since the M.\] for k5n'nc~z does not; c,o,tain ~my infornm.tion that is inconsistent with the in.</Paragraph> <Paragraph position="12"> t>ul; s/~ccilication o\[&quot; the I>PI,I{ in Fig. l, bhc rule is a.I)plic, al)h~' a.nd will produce the derived LI:, for c; For lurthcr dc(,a.ils a.s to wily Lh(&quot; list of ra.isc'd clc. ,mints h~m to be restricted to (xmta.in only non-verba\] (:~m~gorics sec \]linrichs and N~tkaza.wa. (1994).</Paragraph> <Paragraph position="13"> When the LE for k6nnen is unified with the left-hand side of the PPLR, the COMPS list of kb'nnen - and, via structure sharing, the COMPS value of the governed verb - becomes further instantiate& This COMPS list now contains as its left-most element a category with accusative case. In accordance with the PPLR, this leftmost element is promoted to the SUBJ value of kdnnen, while the remainder of the COMPS list of the verb governed by k6nnen, identified by tag 3, is retained. Ilowever, this derived LE would have the undesirable consequence that it admits ill-formed sentences such as (6).</Paragraph> <Paragraph position="14"> (6) * Das Auto wurde kaufen gekonnt.</Paragraph> <Paragraph position="15"> the car was bought can In (6) the auxiliary k6nnen has been passivized and the direct object of the transitive verb has been promoted as the subject of k6nnen. |lowever, since in German only main verbs can be passivized, the sentence is ungrammatical. 7 Ungrammatical sentences such as (6) can be successfiflly ruled out if the PPLR, is applied to an I,E only if the input specification of the LR subsumes the LE (Hypothesis B). The subsumption requirement for LR application is based on the insight that LRs should apply only to LEs that are instantiated at least to the extent that the input description of the LR minimally requires. In the case at hand, the list of raised arguments in the LE for kdnnen in Fig. 3 is totally unspecified - it can be any list of non-verbal synsem objects, inchiding the empty list. The COMPS list of the left-hand side of the PPLR, on the other hand} requires the leftmost element to carry accusative >rhere are some cases of &quot;long distance&quot; passives, i.e. passives which involve the complelnent of an embedded verb, that at least some German speakers accept, e.g. Der Hund wurde vergesscn zu fPSttcrn ('It was forgotten to feed the dog'): However, we are not aware of any German speakers that would allow passives with raising verbs such as k6nnen.</Paragraph> <Paragraph position="16"> case. Therefore the COMPS list of kSnnen does not subsume the COMPS list on the left-hand side of the PPLR, or vice versa. Accordingly, no subsutnption relationship exists between the input specification of the PPM{ as a whole and the LE for kb'nnen, tIence if tIypothesis B is assumed, the Lt{ can be successfiflly blocked, s However, even under subsumption nothing blocks the PPLR from applying to the transitive verb kaufen, as discussed in section 2. Therefore the grammatical sentence in (7) can be derived successfully.</Paragraph> <Paragraph position="17"> This section will consider the interaction of highly schematic LEs with another LR commonly used in IIPSG: the Complement Extraction LR, formulated for English by Pollard and Sag (1994). Fig. 5 shows the version of the Complement Extraction LexicM Rule for German (henceforth: CELRG) that has been proposed by Hinrichs and Nakazawa (1994).</Paragraph> <Paragraph position="19"> for German - Hinrichs and Nakazawa 1994 The CELRG moves an element from the COMPS list of a verb to its SI,ASH set. The value of the feature SLASH contains those items that are realized in left dislocated position, e.g. as a topicalized constituent in sentenceqnitiM position. Analogous to the applicability of tile PPLR, the CELt{G is applicable to LEs of transitive verbs such as kaufcn shown in Fig. 2, under both unification and subsumption. \[File result of such an application is the same as shown in Fig. 2, except that the direct object is placed on the SLASH set instead of the SUB3 list. For the analysis of sen8The reader may wonder whether one couldn't get around the requirement that the LR for passive apply under subsmnption by restricting the rule to apply only to main verbs. At first glance this seems like a possible way-out, since the cases that we havc considered problematic for applying the LR under unification involve LEs of auxiliaries. IIowever, the (:lass of argument raising verbs is not restricted to auxiliaries. Verbs such as verstehen ('know how'), as in Er versteht Parser zu implementiercn ('He knows how to impleInent parsers.'), fall into the same class. But verstehen is a main verb, not an auxiliary. Thus, even if the LR would be reformulated to apply 'only to auxiliaries, the following ungrammaticM sentence could not be excluded: * Parser werden verstanden zu implementieren.</Paragraph> <Paragraph position="20"> lation of the relevant SLASH value that is introduced via the CEIA{G in the LE for kaufcn. 9 (8) l)as Buch wird Peter gekauft haben kgnnen.</Paragraph> <Paragraph position="21"> the book will Peter bought have can 'Peter will have been able to buy the book.' The SLASII wdue in Fig. 6 is percolated h'om the non-terminal node for the verb ka'ufl:n, by the Nonlocal I,'eaturc Principle to the sister node of the topicalized constitueut des Buck 'Fhe top local tree is licensed by the llead-Filler II) Schema which binds oil' the SLASII wdue so that the sentence node has an empty SI,ASH set;./deg The CEI, I{G restricts topicalized constituents to phrasal categories. This restriction is necessary to rule out sentences such as (9) in which a single lexicM item, i.e. a word in terms of the type hierarchy of HPSG, is t, opiealized.</Paragraph> <Paragraph position="22"> (9) * K6nnen wird Peter des Auto gekauft haben.</Paragraph> <Paragraph position="23"> can will 1)eter the car bought have As in the ease of the PPLR, the difference between Hypotheses A and B comes into play when we consider the interaction of the (21!'LtLG with highly schematic entries such as the ones Ilinrichs and Nakazawa (1994) assume for attxiliaries in German. If Hypothesis A is assumed, then the CEI,I{G will be applicable to the type of I,E shown for kSnnen in Fig. 3, since such an I,l'\] will unify with the input descrit)tion of the LI{,. One of the possible outputs of the I3{ to such an auxiliary entry wouht look identical to l;he putative, output of the PPLIt shown in Fig. 4, expect that one of the elements from the COMPS list of the auxiliary is assigned to the SLASH set instead of the SUBJ list;. However, this would have the undesirable consequence that the SLASII wdne in 9The tree in Fig. 6 ~Lssumcs the tlat constituent structure for Germ~tn cbutse structure proposed by tIinrichs and Nakazttwa (1994). However, the issue raised with respc'ct to the CEI,RG in this p~tper is orthogonal to overall assumptions of German obtuse structure.</Paragraph> <Paragraph position="24"> Nakazawa (1.994) for further details on the Nonlocal Feature Principle and the Head-Filler ID Schema.</Paragraph> <Paragraph position="25"> the analysis of topicalized sentences that contain auxiliaries could originate in the LE for the main verb, but also in the LE for each auxiliary present in the sentence. Hence, in addition to the. tree in Fig. 6, three additionM trees are admitted for sentence (81) in which the SLASII value originates in one of the pre-terminal nodes for the anxiliaries.</Paragraph> <Paragraph position="26"> These nodes are marked lbr emphasis by an asterisk in Fig. 6. This ambiguity is, of course, totally spurious since it does not correlate with a difference in semantics or any other relevant linguistic property of the sentence. I!'rom a computational perspective, such spurious ambiguities are highly undesirable since they force the parser into considering multiple analyses where a single analysis suflbes. The spurious ambiguity that we have just identified is particularly pernicious, since it; would aiD.el a wide range of sentences in any grammar of German that employs argument composition and the (JEI~I{G: all assertion main clauses that contMn auxiliaries would be affected since in assertion clauses the initiM constituent is the result of topicalization. null Once again undesirable consequences of overapplying an Lit under unification can be avoided if applicability of LRs instead requires subsumption (llypothesis B). Since the CEM{G limits extracted constituents to phr~ses and since the COMPS list; of an auxiliary does not restrict, its elements in the same w~ty, the LI,\] for auxiliaries ~md the input description of the CEI,I{G do not stand in a subsumption relation, llenee under llypothesis B the CI,3,RG applies only to main verbs, and the tree in Fig. 6 is the only tree admitted.</Paragraph> </Section> <Section position="5" start_page="546" end_page="547" type="metho"> <SectionTitle> 6 Computational Consequences </SectionTitle> <Paragraph position="0"> Finally, we will consider the computational implications that the adoption of lfypothesis B has for the processing of LRs in a computational system.</Paragraph> <Paragraph position="1"> Since consensus on how to provide an adequate denote|tonal semantics for I,Rs has not yet been reached, it would go well beyond the scope of this paper to develop a fully worked-out proposal on how to process lJ{s. n A very promising approach USee CMcagno and Pollard (1995) ~tt(l Menrers (11995) for further discussion.</Paragraph> <Paragraph position="2"> on how to integrate LRs into processing systems for HPSG has been developed independently by wm Noord and Bouma (1994) and by Meurers and Minnen (1995). It turns out that the conclusions reached in this paper can be easily integrated in the general framework that these authors provide. Common to these two proposals is the idea of treating LRs as (ltorn clause) constraints on LF, s of the kind shown schematically in Fig. 7.</Paragraph> <Paragraph position="3"> on relations between lexical entries l)efinite clause schemata as in Fig. 7 (for n 0) define relations between base LEs listed in the lexicon and derived l, Es that are obtained via the application of a sequence of LRs. Each I,R is then viewed as a two-place definite relation as illns\[rated in Fi R. 8 for the l'Pl,t{ of Fig. l. It is important to distinguish two tasks that need to be performed in computing with lexical rules: 12 1. the algorithm that decides for a given lexicaI entry whether a lexical rule is applicable to it, and 2. the algorithm that; computes for a given lex- null ical entry the output specification of the lex~ ical rule, i.e. the derived lexical entry.</Paragraph> <Paragraph position="4"> The subsumption test for lexical rule application that we have argued for in this paper pertains to the first task. The execution of the definite clauses of the kind shown schematically in Fig. 7, which encode the possible relations between base mid derived lexical entries, pertains to the second task.</Paragraph> <Paragraph position="5"> Regarding the first task, in van Noord and Bouma's approach the sequences of lexical rules that are applicable to a given base lexical entry have to be specified by the grammar developer along with delay statements, which allow goal freezing at run riffle of not sufficiently ins\[an\[fated relations. In Meurers and Minnen's approach such sequences are automatically generated from 12This distinctkm is rightly emphasized by Cab cagno and Pollard (1995).</Paragraph> <Paragraph position="6"> the set of base lexical entries and the set of lexical rules specified by the grammar. The allowable sequences of lexi('al rules are compiled into finite state automata which are in turn encoded as de\[: inite clause attachments to base lexical entries. Note thai; both approaches are general enough to accommodate different assumptions about the applicability of lexicat rules to lexical entries, i.e. they are compatible with both Itypotheses A and B. Whether or not at given lexical rule applies to a. lexical entry in van Noord and l{ouma's al)proach needs to be stipulated by the grarnmar writer who is in theory fl:ee to use either a unification or subsumption test. In Meurers and Minhen's approach the deduction rules for automatically generating a finite-state encoding of lexical rules can likewise be based on a subsumption check or a unification clieck. In this paper we have argued on empirical grollnds that subsumption should be the relevant operatiw-, criterion. The theoretical results of tit(', present paper c~m therefore be straightforwardly integrated into a lexical rule compiler of the sort described by Meurers and Minneu in which applicability of lexieal rules is checked automatically under subsumption.</Paragraph> </Section> class="xml-element"></Paper>