File Information
File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/metho/84/p84-1104_metho.xml
Size: 16,496 bytes
Last Modified: 2025-10-06 14:11:43
<?xml version="1.0" standalone="yes"?> <Paper uid="P84-1104"> <Title>Conceptual Analysis of Garden-Path Sentences</Title> <Section position="3" start_page="487" end_page="488" type="metho"> <SectionTitle> llI GARDEN PATH SENTENCES .... </SectionTitle> <Paragraph position="0"> Several different types of local ambiguities cause GPs.</Paragraph> <Paragraph position="1"> Misunderstanding sentences I, 2 and 3 is a result of confusing a participle for the main verb of a sentence. Although there are other types of GPs (e.g., imperative and yes/no questions with an initial &quot;have'), we will only demonstrate how LAZY understands or misunderstands passive participle and main verb conflicts.</Paragraph> <Paragraph position="2"> Passive participles and past main verbs are indicated by a * ed&quot; suffix on the verb form. Therefore, the definition of &quot;ed&quot; must discriminate between these two cases. The definition of &quot;ed= is shown in Figure 3a. A simpler definition for &quot;ed deg is possible if the morphology routine reconstructs sentences so that the suffix of a verb is a separate &quot;word&quot; which precedes the verb. The definition of &quot;ed&quot; is shown in Figure 3a. Throughout this discussion, we will use the name Root for the verb immediately following =ed&quot; on the C-list.</Paragraph> <Paragraph position="3"> If Root appears to be passive Then mark Root as a passive participle.</Paragraph> <Paragraph position="4"> Otherwise if Root does not appear to be passive Then note the tense of Root.</Paragraph> <Paragraph position="5"> Figure 3a. Definition of &quot;ed'.</Paragraph> <Paragraph position="6"> It is safe to consider this request only at the end of the sentence or if a verb is seen following Root which could be the main verb. One test that is used to determine if Root could be passive is: 1. There is no known main verb seen preceding &quot;ed', and 2. The word which would be the subject of Root if Root were active agrees with the selectional restrictions for the word which would precede Root if Root were passive (i.e., the selectional restrictions of the direct object if there is no indirect object), and 3. There is a verb which could be the main verb following Root.</Paragraph> <Paragraph position="7"> Figure 3b.</Paragraph> <Paragraph position="8"> One test performed to determine if Root does not appear to be passive is: 1. The verb is not marked as passive, and 2. The word which would be the subject of Root if Root were active agrees with the selectional restrictions for the subject.</Paragraph> <Paragraph position="9"> Figure 3c.</Paragraph> <Paragraph position="10"> Note that these tests rely on the fact that one request can examine the semantic or syntactic information encoded in another request.</Paragraph> <Paragraph position="11"> As we have presented requests so far, four separate tests must be true to fire a request (i.e., to execute the request's action): a word must be found in a particular position in the sentence, the worif must have the proper part of speech, the word must meet the selectional restrictions, and the parse must be in a state in which it is safe to execute the positional predicate. We have relaxed the requirement that the selectional restrictions be met if all of the other tests are true. This avoids problems present in some previous conceptual analyzers which are unable to parse some sentences such as &quot;Do rocks talk? = . Additionally, we have experimented with not requiring that the Time test succeed if all other tests have passed unless we are reanalyzing a sentence that we have previously not been able to parse. We will demonstrate that this yields the performance that people exhibit when comprehending GPs.</Paragraph> <Paragraph position="12"> LAZY processes a sentence one word at a time from left to right. When processing a word, its representation is added to the C-list and its requests are activated. Next, all active requests are considered. When a request is fired, a syntactic structure is built by connecting two or more constituents on the C-list. At the end of a parse the C-list should contain one constituent as the root of a tree describing the structure of the sentence.</Paragraph> <Paragraph position="13"> Sentence ~6) is a GP which people normally have trouble reading: (6) The boat 8ailed across the river sank.</Paragraph> <Paragraph position="14"> When parsing this sentence, LAZY reads the word &quot;the&quot; and adds it to the C-list. Next, the word &quot;boat&quot; is added to the C-list. A request from &quot;the s looking for a noun to modify is considered and all tests pass. This request constructs a noun phrase with &quot;the&quot; modifying &quot;boat'. Next, &quot;ed s is added to the C-list. All of its requests look for a verb following, so they can not fire yet. The work &quot;sail&quot; is added to the C-list. The request of Sed&quot; which sets the tense of the immediately following verb is considered. It check the semantic features of &quot;boat s and finds that they match the selectional restrictions required of the subject of &quot;sail'. The action of this request is executed, in spite of the fact that its Time reports that it is not safe to do so. Next, a request from &quot;sail&quot; finds that that &quot;boat&quot; could serve as the subject since it precedes the verb in what is erroneously assumed to be an active clause. The structure built by this request notes that *boat&quot; is the subject of &quot;sail'. A request looking for the direct object of &quot;sail&quot; is then considered. It notices that the subject has been found and it is not animate, therefore &quot;sail&quot; is not being used transitively. This request is deactivated. The word &quot;across&quot; is added to the C-list and &quot;the river&quot; is then parsed analogously to &quot;tile boat'. Next, a request from &quot;across&quot; looking for the object of the preposition is considered... and finds the noun phrase, &quot;the river'. Another request is then activated and attaches this prepositional phrase to &quot;sail'. At this point in tile parse, we have built a structure describing an active sentence &quot;The boat sailed across the river.' and the C-list contains one constituent. After adding the verb suffix and &quot;sink&quot; to the C-list we find that &quot;sink&quot; cannot find a subject and there are two constituents left on the C-list. This is an error condition and the sentence must be reanalyzed more carefully.</Paragraph> <Paragraph position="15"> It is possible to recover from misreading some garden path sentences by reading more carefully. In LAZY, this corresponds to not letting a request fire until all the tests are true. Although other recovery schemes are possible, our current implementation starts * over from the beginning. When reanalyzing (6), the request from &quot;ed&quot; which sets the tense of the main verb is not fired because all facets of its test never become true. This request is deactivated when the word &quot;sank&quot; is read and another request from &quot;ed&quot; notes that &quot;sailed&quot; is a participle. At the end of the parse there is oae constituent left on the C-list, similar to that which would be produced when processing &quot;The boat which was sailed across the river sank'.</Paragraph> <Paragraph position="16"> It is possible to parse SDGPs without reanalysis. For example, most readers easily understand (7) which is simplified from \[Birnbaum 81\].</Paragraph> <Paragraph position="17"> (7) The plane stuffed with marijuana crashed.</Paragraph> <Paragraph position="18"> Sentence (7) is parsed analogously to (6) until the word &quot;stuff&quot; is encountered. A request from &quot;ed&quot; tries t,, determine the sentence type by testing if &quot;plane&quot; could be the subject of &quot;stuff* and fails because &quot;plane&quot; does not meet the selectional restrictions of &quot;stuff'. This request also checks to see if &quot;stuff&quot; could be passive, but fails at this time (see condition 3 of Figure 3b). A request from &quot;stuff&quot; then finds that &quot;plane&quot; is in the default position to be the subject, but its action is not executed because two of the four tests have not passed: the seleetional restrictions are violated and it is too early to consider the positional predicate because the sentence type is unknow. A request looking for the direct object of &quot;stuff&quot; does not succeed at this time because the default location of the direct object follows the verb. Next, the prepositional phrase &quot;with marijuana&quot; is pawed analogously to &quot;across the lake&quot; in (6). After the suffix of &quot;crash&quot; (i.e., &quot;ed') and &quot;crash&quot; are added to the C-list; the request fr.m the &quot;ed' of &quot;stuff&quot; is considered, and it finds that &quot;stuff&quot; could be a passive participle because &quot;plane&quot; can fulfill the selectional restrictions of the direct object of &quot;stuff'. A request from &quot;stuff&quot; then notes that &quot;plane&quot; is the direct object, and a request from the &quot;ed&quot; of &quot;crash&quot; marks the tense of &quot;er~h'. Finally, &quot;crash&quot; finds &quot;plane&quot; as its subject. The only constituent of the C-list is a tree similar to that which would be produced by &quot;The plane which wasstuffed with marijuana crashed'.</Paragraph> <Paragraph position="19"> There are some situations in which garden path sentences cannot be understood even with a careful reanalysis. For example, many people have problems understanding sentence (8).</Paragraph> <Paragraph position="20"> (8) The canoe floated down the river aank.</Paragraph> <Paragraph position="21"> To help some people understand this sentence, it is necessary to inform them that &quot;float&quot; can be a transitive verb by giving a simple example sentence such as &quot;The man floated the canoe'. Our parser would fail to reanalyze this sentence if it did not have a request associated with &quot;float&quot; which looks for a direct object. &quot;~e have been rather conservative in giving rules to determine when &quot;ed&quot; indicates a past participle instead of the past tense. In particular, condition 3 of Figure 3b may not be necessary. By removing it, as soon as &quot;the plane stuffed&quot; is processed we would assume that &quot;stuffed&quot; is a participle phrase. This would not change the parse of (7). However, there would be an impact when parsing (0).</Paragraph> <Paragraph position="22"> (9) The chicken cooked with broccoli.</Paragraph> <Paragraph position="23"> With condition 3 removed, this parses as a noun phrase. With it included, (9) would currently be recognized as a sentence. We have decided to include condition 3, because it delays the resolving of this ambiguity until both possibilities are clear. It is our belief that this ambiguity should be resolved by appealing to episodic and conceptual knowledge more powerful than sclectional restrictions.</Paragraph> </Section> <Section position="4" start_page="488" end_page="489" type="metho"> <SectionTitle> IV. PREVIOUS WORK </SectionTitle> <Paragraph position="0"> in PARSIFAL, Marcus' parser, the misunderstanding of GPs is caused by having grammar rules which can look ahead only three constituents. To deterministically parse a GP such as (1), it is necessary to have a look ahead buffer of at least four constituents.</Paragraph> <Paragraph position="1"> PARSIFAL's grammar rules make the same guess that readers make when presented with a true GP. For a participle/main verb conflict, readers prefer to choose a main verb. However, PARSIFAL will make the same guess when processing SDGPs. Therefore, PARSIFAL fails to parse some sentences (SDGPs) deterministically which people can parse without conscious backtracking. In LAZY, the C-list corresponds to the look ahead buffer. When parsing most sentences, the C-list will contain at most three constituents.</Paragraph> <Paragraph position="2"> }\]owever, when understanding a SDGP or reanalyzing a true garden path sentence, there are four constituents in the C-list. Instead of modeling the misunderstanding of GPs, by limiting the size of the look-ahead buffer and the look ahead in the grammar, LAZY models this phenomenon by deciding on a syntactic representation before waiting long enough to disamhiguate on a purely syntactic basis when semantic expectations are strong enough.</Paragraph> <Paragraph position="3"> Shieber models the misunderstanding of GPs in a LALR{I) parser \[Aho 77\] by the selection of an incorrect reduction in a reduce-reduce conflict. In a participle/main verb conflict, there is a state in his parser which requires choosing between a participle phrase and a verb phrase. Instead of guessing like PARSIFAL, Shieber's parser looks up the &quot;lexical preference&quot; of the verb. Some verbs are marked as preferring participle forms; others prefer being main verbs. While this lexicai preference can account for the understanding of SDGPs and the misunderstanding of GPs in any one particular example, it is not a very general mechanism. One implication of using lexical preference to select the correct form is that some verbs are only understood or misunderstood as main verbs and others only as participles. If this were true, then sentences (10a) and {10b) would both be either easily understood or GPs.</Paragraph> <Paragraph position="4"> (10n) No freshmen registered for Calculus failed.</Paragraph> <Paragraph position="5"> (lOb) No car registered in California should be driven in Mezico.</Paragraph> <Paragraph position="6"> We find that most people easily understand (10b), but require conscious backtracking to understand (10a). Instead of using a predetermined preference for one syntactic form, LAZY utilizes semantic clues to favor a particular parse.</Paragraph> </Section> <Section position="5" start_page="489" end_page="489" type="metho"> <SectionTitle> V. FUTURE WORK </SectionTitle> <Paragraph position="0"> We intend to extend LAZY by allowing it to consult and episodic memory during parsing. The format that we have chosen for requests can be augmented by adding an EPISODIC facet to the test. This will enable expectation to predict individual objects in addition to semantic features. We have seen examples of potential garden path sentences which we speculate are misunderstood or understood by consulting world knowledge {e.g., 11 and 12) (11) At MIT, ninety five percent of the freahmen registered for Calculus passed.</Paragraph> <Paragraph position="1"> (1~) At MIT, five percent of the freshmen registered foe Calculus failed.</Paragraph> <Paragraph position="2"> We have observed that more people mistake &quot;registered&quot; for the main verb in (11) than {12). This could be accounted forby the fact that the proposition that &quot;At MIT, ninety five percent of the freshmen registered for Calculus&quot; is more easily accepted than &quot;At MIT, five percent of the freshmen registered for Calculus'. Evidence such as this suggests that semantic and episodic processing are done at early stages of understanding.</Paragraph> <Paragraph position="3"> VI. CONCLUSION We have augmented the basic request consideration algorithm of a conceptual analyzer to include information to determine the time that an expectation should be considered and shown that by ignoring this information when syntactic and semantic expectations agree, we can model the performance of native English speakers understanding and misunderstanding garden path sentences.</Paragraph> </Section> <Section position="6" start_page="489" end_page="489" type="metho"> <SectionTitle> VII. ACKNOWLEDGMENTS </SectionTitle> <Paragraph position="0"> This work was supported by USAF Electronics System Division under Air Force contract F19628-84-C-0001 and monitored by the Rome Air Development Center.</Paragraph> </Section> class="xml-element"></Paper>