Lexical Rules: What are they?
Andrew Bredenkamp, Stella Markantonatou and Louisa Sadler Department of Language and Linguistics University of Essex, UK {andrewb,m~trks,louisa}~essex.~c.uk
Abstract ive Alternation, ctc) and word formation phenom- ena (inflectional and derivational morphology). In Horizontal redundancy is inherent to lex- fact, Pollard and Sag also refer to declension class ica consisting of descriptions of fnlly membership and similar facts as horizontM rela- formed objects. This causes an unwel- tions, and as we shall see, the boundary between come expansion of the lexical database vertical and horizontal relations is not immutably and increases parsing time. To eliminate fixed once and for all. it, direct relations between descriptions The notion of lexical rule is often given some of fnlly formed objects are often defined. status at the level of linguistic or psychological These are additional to the (Typed Mul- theory. (Pollard & Sag 1987) make reference to tiple) Inheritance Network which already a generative or procedural interpretation of lexicM structures the lexicon. Many implement- rules as a deductive mechanism which can be de- ations of horizontal relations, however, ployed on a needs only basis, for example, to gen- fail to generate lexieal entries on a needs- erate words from a single base form. The concep- driven basis, so eliminate neither the tion of lexical rules as essential generative devices problem of lexicon expansion nor that of (rather than static statements expressing (sub-) inefficient parsing. Alternatively, we pro- regularities), is shared in much in:llnential work pose that lexical entries are descriptions (e.g. (Bresnan 1982), (Pinker 1989)), although of objects open to contextnal specifica- it is by no means universM, even within tIPSG. tion of their properties on the basis of Viewed from an implcmentational perspective, on- constraints defined within the type sys- the-fly application of lexicM rules brings with it a tem. This guarantees that only those number of distinct advantages which follow from grammatical lexical entries are infered the drastic reduction in the size of the lexicM data- that are needed for efficient parsing. The base (lexical construction is less time consmning proposal is extremely modest, making and parsing time should be reduced as lexical look use of only basic inference power and ex- up is less ambiguous, etc). At first sight then it ap- pressivity. pears that the benefit of adding an external Lexical l(ule component outweighs the disadvantages (ex- ternal powerNl mechanisms). We will first show 1 Lexical Rules: what are they? that their role is less clear than this suggests and Within the strongly lexical framework of HPSG, certainly more problematic, before suggesting in lexi('al rules are used to express relations among Section 2 an alternative which eschews any extra descriptions ..... a kind of indirect "horizontal re- mechanisms. latedness" (Pollard & Sag 1987, 209) which can be 1.1 Horizontal and Vertical Redundancy contrasted with the vertical relations between the type(s) of lexical elements. Type relations are, of The parallel drawn above between vertical related- course, captured directly as the monotonic (typed) ness (expressed with the type system) and ho- multiple inheritance network itself, which struc- rizontM relatedness among descriptions of fully tures the lexicon. formed objects is however rather misleading. Typical examples of horizontal redundancy in Monotonic multiple inheritance networks are most the hierarchical lexicon thus conceived arc the Al- naturally used to represent generalisations over ternation phenomena (e.g. Dative Shift, the Locat- the properties that (groups of) linguistic objects
163 share - inspection of any network will eonfirin constrained to account for 'exceptional' behaviour, that they are usually deployed to express what that is, for those words which do not participate is essentially a componential analysis of objects to a given horizontal relation despite the fact that and of the relationship between them (defined on their description makes them appropriate candid- the basis of this analysis). On the other hand, ares for the relation (verb alternations offer several horizontal relations among descriptions (very of- examples of these situtation, for instance, 'giving' ten lnodelled by means of lexical rules) are es- verbs which do not exhibit the so-called 'dative sentially relations holding directly between objects shift' phenomenon). themselves. While this intuition is clear, this is Modelling of'exceptional' behaviour leads either much less adequate an approach for morpholo- to an extreme complexity of the type system or gical relatedness, where a componential approach to non-lnonotonic solutions (Flickinger 1987) b(> may often appear just as natural as an object re- cause it turns out that certain horizontM relations, latedness view, especially if the formalism includes usually defined over types, nmst be blocked for in- fimctionally dependent values, permitting the ex- dividuM objects. pression of allomorphic wtriation and the like. In fact, many putatively horizontal relations may be 1.3 hnplementing Horizontal Relations simply re-expressed within a type hierarchy by Several different implementations of horizontal re- viewing them Dom a compouential perspective, lations exist. All of them add extra machinery obviating the need for expressing them on the %o- and some add extra expressive power to the core rizontal" dimension which may lead 6o the use of mechanism. lexical rules. 13ut this is only possible once one Most frequently, horizontM relations are imple- frees oneself from a view of lexical relatedness as mented as unary rules operating at parsing time something which holds essentially between words within a dcrivationM component. Such a compon- (objects which correspond to maximal types, that ent is tided to the inheritance machinery fbr in- is types at the bottom of the type hierarchy). dependent reasons, mainly because of' the limited llorizontal relations are perhaps most naturally expressivity of the type systeul. With Ll/.s, some captm:ed by an extra device (Ll{s) external to the lexical entry is considered as 'basic' and all other lexical network and associated inference mechan- lexical entries are derived fl:om it introducing oth- ism-- see (Krieger gc Nerbonne 1993) and (Cal- erwise unjustified directionality to the grammar. cagno 1995) for recent ItPSG proposals. Some re- In addition, the derivational implementation of ho- cent work ((Mem'ers 1995) and (l{iehemann 1994:)) rizontal relations fails to produce lexical entries a.s partly departs fl:orn this view by expressing rela- needed, instead, it produces lexica.l entries accord- tions between objects using the vertical axis (that ing to the system's internM algorithm of searching is, using the type system), but again tile starting the rule space. Considerable ambiguity is intro- point is 'complete' lexical objects. duced with unpleasant results for parsing time. 1.2 Why avoid Horizontal RelationsY Extra machinery for blocking these rules in or- der to account for exceptional behaviour is also Horizontal relations have a number of undesirable necessary. features as well as requiring an external meclmn- Alternatively, Lt{s may be compiled out bat, ism. HorizontM relations (between objects) are nnder this approach too, problems like direction- in principle pretty much unconstrained. Vertical ality and the blocking of Ll{s as well as expensive relations are more constrained becmlse they are ambiguity at parsing time remain unsolved. based on componential analysis, starting out Dora the set of properties that objects have. On the 2 An alternative proposal other hand, any object can be related to any other object by stipulation in an external mechanism. In In this paper we explore an alternative to hori- architectural terms, it is simply accidental (if for- zontal relatedness which exploits the idea that it is tuitous) that lexical rules are often used to relate often possible to conceive of the linguistic objects minimMly different objects -- they are capable of in such a way as to eliminate potential sources much more promiscious behaviour. of ambiguity and additional external mechanisms. This state of,affairs is amply demonstrated in To illustrate our approach we will propose an ac- the literature, which abounds with attempts to count of a subset of Verb Alternation phenom- constrain horizontal relations by appeal to subsi- ena which rely on what are essentially underspe- diary principles (predicate locality in LFG, con- cilled lexicM entries. The lexicon will then con- straints of a psycholinguistic nature in the work of tain one (verbal) entry and the system will rely (Pinker 1989), etc). Horizontal relations must be only on the existing resources (the type hierarchy)
164 to provide the different interpretations of the pre- pointer is included as an extra feature of the dicate which license the distinct eoml)lementation value of SYNSF,IVI[LOC[CONT[NUCLEUS. This fea- patterns. Analysis is incremental and determin- t/ire we ttalile SFM(ANTIC) CONS('FII.AINq'S) a,tl([ istic and the procedure relies mainly on what we we make it apl)rol)riate tbr the same values that will call 'trivial type inference'. In the sections the prepositional SYNSEMII,OC[CONT]NUCI,EUS is that follow [irst we discuss the linguistic at)l)roaeh assigned. The lexica] entry for to load would look underlying our proposal, second we eOmlmre our as in ((~). proposal to existing underspecification tq)proaches and finally, we give some details of the implementa- tion which relies on no special fc'+~tures or external devices. 2.1 -Underspeeificatlon +L ...... {++,+I ...... <++7> We will exemplify ore- approach by treating a ltl¢I, load [~]1 subset of verb alternati(ms which conform to |^,.,2 [L~ the following general schema (]). These include f;[ )N 'l' It* NT IAtl+i[~ [] the so-called spray/load (locative) Mtornation, the | ...... +[ ..... , wipe~clear a.lternation, the b~vak/hit alternation [ (o,,.,~t[ ..... etc (l,evin 19!)a). (~) v Nl'j [l'~ Np~] -+ v NP~ [P~ Nt'.j] The lexieal entry for the pre4)osition with is given I)el()w: We adopt the view that verb l)redicatcs are open to contextual information (which ntttst I)e contras- ted to the approaches whereby verb predicates are ...... (+:~) treated am fully formed objects which dictate tit+; exact nature of their dependents). Consider the + ...... + ...... ,::::, predicate load: with_contact Alt{+2 ~[~] J (2) The peasant loaded the horses. (3) The peasant loaded the horses on the boat. (6) is an mtdcrspcciJicd entry which gets filr- (4) The peasant loaded tit(: horses with hay. ther specified at parsing time when art appropri- ate PP is a tta+ehed, l"or instance, if a with-[)l ~ is (2) is ambiguous I)ctwc,m (3)and (4:) ,inch one encountered, then tm interpretation according Lo of which is not ambiguous. 'l'he contextual factor which the location sm'faccs as the direct object of that resolves the ambiguity is the semantics of the the verb is injured. head of the prepositional complement which here There are some theoretical reasons why we have is tt~ken to specii~y whether the direct ob.ieet of the chosen to include a "pointer" to prepositional se- verb is understood as the location and the oblique mantics rather than making it compatible (uui[i- complement as the locatum or v[ee versa. The cru- M)le) to verbal semantics as Weehsler (Wechsler tim a.ssumption here is that prepositions have their 1994) has proposed, l,'irstly, if verl)al and pre- own semantics, an idea first exploited in ((lawron positional semantics were uniliable then we would 1!)86). not bc able to explicitly state, in the semantics the We use/IPSG to model our approach. (5) gives relation which each feature structure encodes as the fragment of the type system constraining the there would be a clash of constants (relation nanms wdues of the path in SYNSEMILOC]CONTINUCLEUS are constants). Secondly, identifying the semantics the (word) description of prepositions which par- of verbs with ttmt of prepositions does not allow ticipate to the locative alternation l)henomenon. for expressing certain types of diverse behaviour within the class of a.lternating verbs. For instauce, IH!'I, ± both and show locative, alternation, but At{GI co~+tc'nt load stuff A II.G2 co;it e~tt only the former admits optional PP complements. With to stuff the interpretation under which lees- ItEI, with ItEI, on V i+t ... lion is a direct object admits an optional P I) eotn- AI/.G 1 (loe.atio~t) AI{GI (locat~tm) plement (8) while the interpretation under u, hich AI{G2 (locat~m) AI{G2 (location) [,++...... 1[ ...... ] Iocatum is a direct argument admits an obligat- We %rthermore assume that the semantics of ory one (9). Similarly, while both versions of to the predicates include a pointer to the semantics of load are related to passive adjectives (loaded carl, the prepositional complements they license. This loaded hay), only the "location" version is related
165 cont~tct optional obligatory inference power' is independently needed to deal, for instance, with (10): if NP1 is a subtype of NP then rule (10) will work only if trivial inference Oll-GO~ power is available when the sequence NP1, VP is on_con.opt on_con-obl with-con_opt with_con~bl encountered. (10) S --+ NP, VP Figure 1: Type system fragment encoding prepos- There are proposals in the literature which itional alternation build on the idea of using underspecified entries. However, several of them use additional, external to such an adjective in the case of to stuff (stuffed powerful mechanisms to simulate type inference. pillow, *stuffed feathers). The exact treatment of (van Noord ~: Bouma 1994) use underspecified such phenomena, however, goes beyond the scope verb entries and FP~OLOG delayed evaluation tech- of our discussion here which concentrates on the niques to insert adjuncts in Dutch VPs without use of underspecification. using lexical rules which would cater for the ne- cessary variations of the subcategorisation list of (8) Mary stuffed the pillow with feathers. verbs. In another proposal using underspeciflc- (9) Mary stuffed the feathers into the pillow. ation (Sanfilippo 1995) type inference (feature Optionality of PP complements can also be cap- structure grounding) is simulated by relying on tured easily with this proposal. With to load (2), an external mechanism as powerful as PItOLOG. (3) and (4), the PP complement is optional. The In dealing with different complementation pat- grammar must have access to three different ver- tern phenomena, Sanfilippo constructs type sys- sions of to load, one with zero PP complements and tem fragments where the meet of the alternative two with a PP complement participating in the al- complements is defined and subtypes verbs accord- ternation discussed above. One approach would ing to complement types. Therefore, the informa- involve defining two lexieal rules; an alternative tion about the alternation is duplicated in the type would be to express all three possibilities directly. system as it is encoded both on tile complement Both are problematic, of course. Consider the situ- types and the verb types. The same information ation when the grammar has two PS rules for VPs, is encoded again on a table of clauses which relate one for discharging a [NP,NP] SUBCAT list and one a verbal "meet" type with a maximal complement for discharging a [NP,NP, PP] list. Without harm- type and a maximal verb type. Such type resolving ing generality, assume that the bivalent version of clauses are provided for each alternation pattern. to load is in the lexicon and two lexical rules gen- PS rules are annotated with procedures which pick erate the trivalent versions. To process a trivalent up the correct verb type resolving clause when the version, the parser will backtrack on the bivalent appropriate complement is encountered. Both the version, will use a lexical rule and then, it will clauses and the searching procedures are mechan- either succeed or it will backtrack again and use a isms external to the inferencing mechanism that is second lexical rule. directly related with the type system. Sanfilippo's To avoid this, the following solution may be ad- approach, though powerful and flexible, seems ex- opted. First, the type system is augmented to al- travagant for phenomena like verb alternations of low for declaring the property of being an optional the kind discussed here as well as inflection phem- or an obligatory prepositional complement, as in omena of the kind discussed in (Krieger & Ner- figure 1. bonne 1993). In such cases the system can take Second, a PS structure rule is introduced of the advantage of the fact that type inference can be following sort: driven by the combination of the information that VP --+ V[SUBCAT[NP,NP,P(optional)P]], NP is related to two separate strings (preposition and Only one trivalent, underspecified version of to verb, verb ending and verb stem) as is exemplified load is necessary. The parsing of a trivalent ver- in our proposal. sion as before would involve backtracking on the Furthermore, in our approach no horizontal re- rule dealing with optional complements but then lations exist as the lexicon contains only one entry the rule dealing with obligatory ones would be and no other entry is ever generated. Instead, the chosen and it would succeed anyway. single lexical entry is interpreted on the fly, each Only limited inference power is necessary for time according to well-specified constraints. Con- this set up to work: the system must be able to sequently, no ambiguity problems result with a infer that the unification of a subtype with its su- nice effect, on parsing time. In this sense, using un- pertype is of the type of the subtype. This 'trivial derspecification defined in the type system is more
166 econolnic than using lexical rules or a "static" ver- distinction. sion of underspecification which is defined in the For the construction of the VP, a simple rule lexicon. For instance, (Krieger &: Nerbonne :1993) was used (VP -+ V NP PP), of the following form have used a specialised macro, the so-called dis- (14): tributive (or named) disjunction, in a treatment 04) of German verb inflectional morphology: Id :{a ign=>phrasal :{ While it is true that distributive disjunction aynsom=>synsom:{ Ioel=>lool:{ does not add any expressive power to the sys- cat=>cat :{ tem (though a piece of machinery, the specialised head=>HEAD, subj:>[SUBJ] , macro, must be supported), if the macro is ever comps->[]}, cont~nt=>Cont ent }}}} called all the legal combinations are thereby gem < [ crated and added to the lexicon. In this, the situ- ld : {s ign=>l~ X[col :{ aynsom->syns~m: { ation is precisely the same us with lexieal rules, for loel->locl:{ ca%=>cat :< in each case, what is provided is simply a compact hoad=>llEAD=>vorb :{}, sub j=> [SUB J} , representation of an ambiguity. comps~> [OBJ I ,OBJ2] } This can be also exemplified from the domain contont=>Content}}}}. Id :{si~n=>phPasal :{ of Verb Alternation phenomena. (11) will generate synsem~>OBJ l~>synsom:{ locl=>loel:{ two lexical entries with at1 identical I'IION string. taro>cat ;{ h~ad=>noun :~}}}}}}, .... < ...... > ld :{sil~n=>phrasal :{ (]*) synsom=>OBJ2=>myns om: { locl:>locl :{ cat=>cat :{ h~ad=>prop :{} }}}}}] , [ ......
167 som_constr=> tion approach. loc_altornation:{ arg2_stato=>A2S, argS_sta~o:>hSS}}}}}}}}. References The prepositional entries now simply provide the "missing" part of the semantics, nanMy the Joan Bresnan (ed). 1982. The menial representa- Ioeum/locatum distinction: tion of grammatical relations. MIT Press (16) Mike Calcagno. 1995. Interpreting lexical rules. ¢1 with " In AQUILEX Workshop on lexical rules. Cam- ld:< sign=>loxical:{ bridge, UK, August 9-11, 1995 m PHON_LEX[with], synsem=>synsem:{ Daniel Paul Flickinger. 1987. Lcxical l~ulcs in the locl=>locl:< cat=>cat:{ Hierarchical Lexicon. PhD. Stanford University. head=>prep:{ pform=>with}, subj=>[], Jean Mark Gawron. 1986. Situations and Preposi- comps=>[synsom:{ tions. Linguistics and Philosophy 9, pp.327-382 locl=>locl:< cat:>cat:{ hoad=>noun:{caso=>acc}, Hans-Ulrich Krieger & John Nerbonne. 1993. subj=>[], comps:>[]}, Feature-BAsed Inheritance Networks for Com- content=>inst_paoa:{ putational Lexicons. In (ads) Ted lbiscoe, Va- rel=> Rol}}}]}, con~ent=>ins~_psoa:{ leria di Paiva and Ann Copstake Inheritance, rel => Rel, Defaults and the Lexicon. Cambridge University sem~constr => with_variant:{ arg2_stato=>loca~um, Press, 90--136 arg3_stato=>locum}}}}}}. Beth Levin. 11993. English Verb Classes and Al- ternations. A Preliminary Investigation. The 4 Conclusion University of Chicago Press. We have shown that horizontal redundancy is in- Walt Detnmr Meurers. 19!)5. Towards a semantics herent to a lexicon consisting of descriptions of for lexical rules as used in HPSG. In AQUILEX fully formed objects. To eliminate horizontal re- Workshop on lexical rules. Cambridge, UK, Au- dundancy, direct relations between descriptions of gust 9-11, 71995 fully formed objects must be defined externally to Steven Pinker. 1989. Learnability and Cognition: the Typed Mulitple Inheritance Network or unin- .The acquisition of argument structure. MIT tuitive solutions must be pursued. Available ira- Press plementations of horizontal relations fail to satisfy the reasons that dictate their implementation: the Carl Pollard & Ivan A. Sag. 1987. Information- on-need generation of lexical entries and efficient based Syntax and Semantics. Volume 1. Funda- parsing. Alternatively, we proposed that lexical mentals. Center for the Study of Language and Information entries are descriptions of objects which allow for further contextual specification of their properties Susanne Riehemann, Morphology and the Hier- on the basis of clearly defined constraints. We archical Lexicon (ms.) Stanford University, have shown that this is an easily implementable Stanibrd. proposal even in environments with lean inference Antonio Sanfilippo. 1.995. Lexical Polymorphism power and expressivity because it relies on very and Word Usage Extensibility. In AQUU,EX basic machinery which is available for independ- Workshop on lexieal rules, Cambridge, UK, Au- ent reasons. gust 9--11, 1995 This approach can be adopted whenever inforru- a.tion can be distributed among independent sur- Gertjan wm Noord and Gosse Bomna. 1994. Ad- face strings. Under the light of this proposal, many juncts and the processing of lexical rules. In of the phenomena which have been argued in (Pof Proceedings of the 15th International Cm@r- ence on Computational Linguistics lard & Sag 1987) to justify the horizontal related- (COLING), Kyoto, 1994 ness approach can be viewed as different 'inter- pretations' of a 'core' lexical entry according to Stephen Mark Wechsler. 1994. Preposition Selec- well-specified types of 'context'. tlowever, it must tion Outside the Lexicon. To appear in Procee& be noted here that this is not always a simple task. ings of the Thirteenth West Coast Conference Roughly speaking, the less specific the contextual on Formal Linguistics information is the more inference power and ex- pressivity is needed to retain the underspecifica-
168