File Information
File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/metho/90/c90-3029_metho.xml
Size: 21,640 bytes
Last Modified: 2025-10-06 14:12:29
<?xml version="1.0" standalone="yes"?> <Paper uid="C90-3029"> <Title>Two Principles of Parse Preference</Title> <Section position="2" start_page="0" end_page="0" type="metho"> <SectionTitle> 2 Brief Review of the IAtera- </SectionTitle> <Paragraph position="0"> ture Most previous work on parse preferences has concerned itself with tile most notorious of the ambiguities--the attachment ambiguities of postmodifiers. Among the first linguists to address this problem was Kimball (1973). tie proposed several processing principles in an attempt to account for why certain readings of ambiguous sentences were more salient than others. Two of these principles were Right Association and Closure.</Paragraph> <Paragraph position="1"> In the late 1970s and early 1980s there was a great deal of work among linguists and psycholinguists (e.g. Frazier and Fodor, 1979; Wanner and Maratsos, 1978; Marcus, 1979; Church, 1980; Ford, Bresnan, and Ksplan, 1982) attempting to refine Kimbali's initial analysis of syntactic bias and proposing their own principles govering attachment. Frazier and Fodor pro~ posed the principles of Minimal Attachment and Local Association. Church proposed the A-over-A Early Closure Principle; and Ford, Bresnan and Kaplan im troduced the notions of Lexical Preference and Final Arguments.</Paragraph> <Paragraph position="2"> The two ideas that dominated their hypotheses and discussions were Right Association, which says roughly that postmodifiers prefer to be attached to the nearest previous possible head, and a stronger principle stipulating that argument interpretations are favored over adjunct interpretations. This latter principle is implied by Frazier and Fodor's Minimal Attachment and also by Ford, Bresnan and Kaplan's Lexical Preference.</Paragraph> <Paragraph position="3"> In recent computational linguistics, Shieber and Pereira (Shieber, 1983; Pereira, 1985) proposed a shift-reduce parser for parsing English, and showed that Right Association was equivalent to preferring shifts over reductions, and that Minimal Attachment was equivalent to favoring the longest possible reduction at each point.</Paragraph> <Paragraph position="4"> More recently, there have been debates, for example, between Schubert (1984, 1986) and Wilks et al. (1985), about the interaction of syntax with semantics and the role of semantics in disambiguating the classical ambiguities.</Paragraph> <Paragraph position="5"> We take it for granted that, psychologically, syntax, semantics, and pragmatics interact very tightly to achieve disambiguation. In fact, in other work (Hobbs et al., t988), we have proposed an integrated framework for natural language processing that provides for this tight interaction. However, in this paper, we are considering only syntactic factors. In the semantically and pragmatically unsophisticated systems of today, these are the most easily accessible factors, and even in more sophisticated systems, there will be examples that semantic and pragmatic factors alone will fail to disambiguate.</Paragraph> <Paragraph position="6"> The two principles we propose mawr be viewed as generalizations of Minimal Attachment and Right Association. null</Paragraph> </Section> <Section position="3" start_page="0" end_page="0" type="metho"> <SectionTitle> 3 Most Restrictive Context </SectionTitle> <Paragraph position="0"> The first principle might be called the Most Restrictive Context principle. It can be stated as follows: Where a constituent can be placed in two different structures, favor the structure that places greater constraints on allowable constituents. null For example, in John looked for Mary.</Paragraph> <Paragraph position="1"> &quot;for Mary&quot; can be interpreted as an adverbial signaling the beneficiary of the action or as a complement of the verb &quot;look&quot;. Since virtually any verb phrase can take an adverbial whereas only a very few verbs can take a &quot;for&quot; prepositional phrase as its complement, the latter interpretation has the most restrictive context and therefore is favored.</Paragraph> <Paragraph position="2"> A large number of preferences among ambiguities can be subsumed under this principle. They are enu- null merated below.</Paragraph> <Paragraph position="3"> 1. As in the above example, favor argument over adverbial intepretations for postmodifying prepositional phrases where possible. Thus, whereas in John cooked for Mary.</Paragraph> <Paragraph position="4"> &quot;for Mary&quot; is necessarily an adverbial, in &quot;John looked for Mary&quot; it is taken as a complement. Subsumable under this heuristic is the preference of &quot;by&quot; phrases after passives to indicate the agent rather than a location. This heuristic, together with the next type, constitutes tile traditional Minimal Attachment principle. This heuristic is very strong; of 47 occurrences examined, all were in accord with tile heuristic. 2. Favor arguments over mere modifiers. Thus, in John bought a book from Mary.</Paragraph> <Paragraph position="5"> the favored interpretation is &quot;bought from Mary&quot; rather than &quot;book from Mary&quot;. Where tile head noun is also subcategorized for the preposition, as in, John sold a ticket to the theater.</Paragraph> <Paragraph position="6"> this principle fails to decide among the readings, and tile second principle, described in the next section, becomes decisive.</Paragraph> <Paragraph position="7"> This principle was surprisingly strong, but perhaps for illegitimate reasons. Of 75 potential ambiguities, all but one were in accord with the heuristic. The one exception was IIDTV provides television images with finer detail than current systems.</Paragraph> <Paragraph position="8"> and even this is a close call. However, it is often very uncertain whether we should say verbs, nouns, and adjectives subcategorize for a certain preposition. For example, does &quot;discussion&quot; subcategorize for &quot;with&quot; and &quot;about&quot;? We are likely to say so when it yields the right parse and not to notice the possibility when it would yiehl the wrong parse. So our results here may not be completely unbiased.</Paragraph> <Paragraph position="9"> 3. Favor complement interpretations of infinitives over purpose adverbial interpretations. In John wants his driver to go to Los Angeles.</Paragraph> <Paragraph position="10"> the preferred interpretation has only the driver and not John going to Los Angeles.</Paragraph> <Paragraph position="11"> Of 44 examples of potential ambiguities of this sort that we found, 41 were complements and only 3 were purpose adverbials. Even these three could have been eliminated with the simplest seleetional restrictions. One example was the following tie pushed aside other business to devote all his time to this issue.</Paragraph> <Paragraph position="12"> 2 163 which could have been parsed analogously to He pushed strongly all the young researchers to publish papers on their work.</Paragraph> <Paragraph position="13"> A particularly intriguing example, remembering that &quot;provide&quot; can be ditransitive, is the following: That is weaker than what the Bush administration needs to provide the necessary tax revenues.</Paragraph> <Paragraph position="14"> 4. Favor the attachment of temporal prepositional phrases to verbs or event nouns. In the preferred reading of John saw the President during the campaign. null the seeing was during the campaign, since &quot;President&quot; is not an event noun. In the preferred reading of The historian described the demonstrations during Gorbachev's visit.</Paragraph> <Paragraph position="15"> the demonstrations are during the visit. This case can be considered an example of Minimal Attachment if we assume that all verbs and event nouns have potential temporal arguments. Of 74 examples examined, 66 were in accord with this heuristic. Two that did not involved the phrase &quot;business sin~e August 1&quot;. 5. Favor adverbial over object interpretations of temporal and measure noun phrases. Thus, in John won one day in Hawaii.</Paragraph> <Paragraph position="16"> &quot;one day in tIawaii&quot; is preferentially the time John won and not his prize. In John walked 10 miles.</Paragraph> <Paragraph position="17"> &quot;10 miles&quot; is a measure of how far he walked, not what he walked. This is an example of Most Restrictive Context because noun phrases, based on syntactic criteria alone, can always be the object of a transitive verb, whereas only temporal and measure noun phrases can function as adverbials. This case is interesting because it runs counter to Minimal Attachment. Here arguments are disfavored.</Paragraph> <Paragraph position="18"> Of fifteen examples we found of such ambiguities, eleven agreed with the heuristic. The reason for the large percentage of examples that did not is that sports articles were among tlmse examined, and they contained sentences like Smith gained 1240 yards last season.</Paragraph> <Paragraph position="19"> This illustrates the hidden dangers in genre selection. 6. Favor temporal nouns as adverbials over coinpound nominal heads. The latter interpretation is possible, as seen in Is this a CSLI Thursday? But the preferred reading is the temporal one that is most natural in I saw the man Thursday.</Paragraph> <Paragraph position="20"> 7. Favor &quot;that&quot; as a complementizer rather than as a determiner. Thus, in I know that sugar is expensive.</Paragraph> <Paragraph position="21"> we are probably not referring to &quot;that sugar&quot;. This is a case of Most Restrictive Context because the determiner &quot;that&quot; can appear in any noun phrase, whereas the complementizer &quot;that&quot; can occur only after a small number of verbs. This is a heuristic we suspect everyone who has built a moderately large grammar has implemented, because of the frequency of the ambiguity.</Paragraph> <Paragraph position="22"> 8. An initial &quot;there&quot; is interpreted as an existential, where possible, rather than as a locative. We interpret There is a man in the room.</Paragraph> <Paragraph position="23"> as an existential declarative sentence, rather than as an utterance with an initial locative. Locatives can occur virtually anyplace, whereas the existential &quot;there&quot; can occur in only a very small range of contexts. Of 30 occurrences examined, 29 were in accord with the heuristic. The one exception was There, in the midst of all those casinos, is 'Ptump's Taj Mahal.</Paragraph> <Paragraph position="24"> 9. Favor predeterminers over separate noun phrases. In Send all the money.</Paragraph> <Paragraph position="25"> the reading that treats &quot;all the&quot; as a complex determiner is favored over the one that treats &quot;all&quot; as a separate complete noun phrase in indirect object position. There are very many fewer loci for predeterminers than for noun phrases, and hence this is also an example of Most Restrictive Context.</Paragraph> <Paragraph position="26"> 10. Favor preprepositional lexical adverbs over separate adverbials. Thus, in John did the job precisely on time.</Paragraph> <Paragraph position="27"> we favor &quot;precisely&quot; modifying &quot;on time&quot; rather than &quot;did the job&quot;. Very many fewer adverbs can function as preprepositional modifiers than can function as verbal or sentential adverbs. Of 28 occurrences examined, all but one were in accord with the heuristic. The one was Who is going to type this all for you? _k 164 3 11. Group numbers with prenominal unit nouns but not with other prenominal nouns. For example, &quot;I0 mile runs&quot; are taken to be an indeterminate number of runs of l0 miles each rather than as exactly l0 runs of a mile each. Other nouns can firnction the same way as unit nouns, as in &quot;2 car garages&quot;, t)ut it is vastly more common to have the mlmber attached to the head noun instead, as in &quot;5 wine glasses&quot;. Virtually any noun can appear as a prcnominal noun, whereas only unit nouns can appear in the adjectival &quot;10-mile&quot; constrnction. Iience, for unit nouns this is the most restrictive context. While other nouns can ..~ometirnes occur in this context, it is only through a reinterpretation as a unit noun, as in ';2 car garages&quot;. 12. Disfavor headless structures. Headless strucl.nres impose no constraints, and are therefore never the most re,~trictive context, and thus are the least favored in cases of ambiguity. An example of this case i,~ the sentence John knows the best man wins.</Paragraph> <Paragraph position="28"> which we interpret as a concise form of John knows (that) the best man wins.</Paragraph> <Paragraph position="29"> rather than aq a concise form of John knows the best (thing that) man wins 0.</Paragraph> </Section> <Section position="4" start_page="0" end_page="0" type="metho"> <SectionTitle> 4 Attach Low and Parallel </SectionTitle> <Paragraph position="0"> The second principle might be called the Attach Low and Parallel principle. It may be stated as follows: Attach constituents as low as possible, and in parallel with other constituents ifpossible. null The cases subsumed by this principle are quite het- null erogeneou S.</Paragraph> <Paragraph position="1"> 1. Where not overridden by the Most Restrictive Context principle, favor attaching postmodifiers to the closest possible site, skipping over proper nouns. Thus, where neither the verb nor the noun is subcategorized for the preposition, as in John phoned a man in Chicago.</Paragraph> <Paragraph position="2"> or where both the verb and the noun are subcategorized for the preposition, as in John was given a book by a famous profes-</Paragraph> </Section> <Section position="5" start_page="0" end_page="0" type="metho"> <SectionTitle> SOIL </SectionTitle> <Paragraph position="0"> the noun is favored as the attachment point, since that is the lowest possible attachment point in tile parse tree. This case is just tile traditional Right Association. null The subcase of prepositional phrases with &quot;of&quot; is significant enough to be mentioned separately. We might say that every noun is subcategorized for &quot;of&quot; and that therefore &quot;of&quot; prepositional phrases are nearly always attached to the immediately preceding word. Of 250 occurrences examined, 248 satisfied this heuristic, and of the other two Since the first reports broke of the CIA's activities, ...</Paragraph> <Paragraph position="1"> He ordered the destruction two years ago of some records.</Paragraph> <Paragraph position="2"> tile second would not admit an incorrect attachment. in any case.</Paragraph> <Paragraph position="3"> We examined 148 instances of this case not involving &quot;of&quot;, temporal prepositional phrases, or prepositions that are subcategorized for by possible attachment points. Of these, 116 were in accord with the heuristic and 32 were not. An example where this heuristic failed was They abandoned hunting for food production. null For a significant number of examples (34), it did not matter where the attachment was made. For instance, in John made coffee for Mary.</Paragraph> <Paragraph position="4"> both the coffee and the making are for Mary. We counted these cases as being in accord with the heuristic, since the heuristic would yield a correct interpretation. null This is perhaps the place to present results on two very simple algorithms. The first is to attach prepositional phrases to the closest possible attachment point, regardless of other considerations. Of 251 occurrences examined, 125 attached to the nearest possibility, I09 to the second nearest, 14 to the third, and 3 to the fourth, fifth, or sixth. This algorithm is not especially recommended.</Paragraph> <Paragraph position="5"> The second algorithm is to attach to the near: est possible attachment point that subeategorizes for the preposition, if there is such, assuming verbs and event nouns to subcategorize for temporal prepositional phrases, and otherwise to attach to the nearest possible attachment point. This is essentially a summary of our heuristics for prepositional phrases. Of 297 occurrences examined, this yielded the right answer on 256 and the wrong one on 41.</Paragraph> <Paragraph position="6"> 4 165 2. Favor preprepositional readings of measure phrases over readings as separate adverbials. Thus, in John walked 10 miles into the forest..</Paragraph> <Paragraph position="7"> we preferentially take &quot;10 miles&quot; as modifying &quot;into the forest&quot; rather than &quot;walked&quot;, so that John is now 10 miles from the edge of the forest, rather than merely somewhere in tile forest but 10 miles from his starting point. Since the preposition occurs lower in the parse tree than the verb, this is an example of Attach Low and Parallel. Note that this is a kind of &quot;Left Association&quot;.</Paragraph> <Paragraph position="8"> 3. Coordinate &quot;both&quot; with &quot;an~&quot;, if possible, rather than treating it a~s a separate determiner. In John likes both intelligent and attractive women.</Paragraph> <Paragraph position="9"> the interpretation in which there are exactly two women who are intel!igmit and attractive is disfavored. Associating &quot;both&quot; with the coordinated adjectives rather than attaching it to the head noun is attaching it lower in the parse tree.</Paragraph> <Paragraph position="10"> 4. Distribute prenominal nouns over conjoined bead nouns. In &quot;oil sample and filter&quot;, we mean &quot;oil s,-~i,q>le and oil filter&quot;. A principle of Attach Low wo,.~i,:l not seem to be decisive in this case. Would ~{. ~,e~m that we attach &quot;oil&quot; low by attaching it to &quot;sample&quot; or that we attach &quot;and filter&quot; low by attaching it to &quot;sample&quot;. It is because of examples like this (and the next case) that we propose the principle Attach Low and Parallel. We favor the reading that captures the parallelism of the two head nouns.</Paragraph> <Paragraph position="11"> 5. Di~.t.ri}mte determiners and noun complements over conjoined head nouns. In &quot;the salt and pepper on the table&quot;, we treat &quot;salt&quot; and &quot;pepper&quot; as conjoined, rather than &quot;the salt&quot; and &quot;pepper on the table&quot;. As in the previous case, where we have a choice of what to attach low, we favor attaching parallel elements low. 6. Favor attaching adjectives to head nouns rather than prenominal nouns. We take &quot;red boat house&quot; to refer to a boat house that is red, rather than to a house for red boats. Like all of our principles, this preference can be overridden by semantics or convention, as in &quot;high stress job&quot;. IIere again we could interpret Attach Low ~ telling us to attach &quot;red&quot; to &quot;boat&quot; or to attach &quot;boat&quot; to &quot;house&quot;. Attach Low and Parallel tells us to favor the latter.</Paragraph> </Section> <Section position="6" start_page="0" end_page="0" type="metho"> <SectionTitle> 5 Interaction and Overriding </SectionTitle> <Paragraph position="0"> There will of course be many examples where both of our principles apply. In the cases that occur with some frequency, in particular, the prepositional phrase attachment ambiguities, it seems that the Most Restrictive Context principle domina.tes Attach Low and Parallel. It is unclear what the interacti6ns between these two principles should be, more generally. null These principles can be overridden by more than just semantics and pragmatics. Comma.s in written discourse and pauses in spoken discourse (see Bear and Price, 1990, on the latter) often function to override Attach Low and Parallel, as in John phoned the man, in Chicago.</Paragraph> <Paragraph position="1"> Specify the length, in bits, of a word.</Paragraph> <Paragraph position="2"> It is the phoning that is in Chicago, and the specification is in bits while the length is of a word. Similarly, commas and pauses can override the Most Restrictive Context principle, as in John wants his driver, to go to Los Angeles.</Paragraph> <Paragraph position="3"> Itere we prefer the purpose adverbial reading in which John and the driver both are going to Los Angeles.</Paragraph> </Section> <Section position="7" start_page="0" end_page="0" type="metho"> <SectionTitle> 6 Cognitive Significance; </SectionTitle> <Paragraph position="0"> The analysis of parse preferences in terms of these two very general principles is quite appealing, and more than simply because they subsume a. great many cases. They seem to relate somehow to deep principles of cognitive economy. The Most Restrictive Context principle is a matter of taking all of the available information into account in constructing interpretations. The &quot;Low&quot; of Attach bow and Parallel is an instance of a general cognitive heuristic to interpret features of the enviromnent ~ locally as possible. The &quot;Parallel&quot; exemplifies a general cognitive heuristic to see similarity wherever possible, a heuristic that promotes useful generalizations.</Paragraph> </Section> class="xml-element"></Paper>