Datasets:

WINGNUS
/

ACL-OCL

	{
	"paper_id": "J78-3018",
	"header": {
	"generated_with": "S2ORC 1.0.0",
	"date_generated": "2023-01-19T03:04:43.588576Z"
	},
	"title": "'The ~s e of Focus as a Tool for Disambiguation of Definite Noun Phrases",
	"authors": [
	{
	"first": "Ca",
	"middle": [
	"Boston"
	],
	"last": "Wrman",
	"suffix": "",
	"affiliation": {},
	"email": ""
	},
	{
	"first": "Dav~d",
	"middle": [
	"E"
	],
	"last": "Rumelhart",
	"suffix": "",
	"affiliation": {},
	"email": ""
	},
	{
	"first": "W",
	"middle": [],
	"last": "March",
	"suffix": "",
	"affiliation": {},
	"email": ""
	},
	{
	"first": "6~1",
	"middle": [],
	"last": "Allefi",
	"suffix": "",
	"affiliation": {},
	"email": ""
	},
	{
	"first": "",
	"middle": [],
	"last": "Unwinolondon",
	"suffix": "",
	"affiliation": {},
	"email": ""
	}
	],
	"year": "",
	"venue": null,
	"identifiers": {},
	"abstract": "Fahlman, Scott E. [I9771 A System for Reprasenting and U3ng Real-World K)rowIedp, M1.T. AI. Laboratory AI-XR-450, December. Grosz, Barbara [I9771 The Repbesq(rfafron and Use df Focus iq OIslogue ender+tand~n~. Stanbrd Research Institute Technical Note f 5 1, Menlp Park, Calthrnia Halli:day, MA. and Ruqarya Hasan 119761 Cohe~en in EMlish L~ndon, Longman press. krktunen, Lauri [I9681 What Makes Oefinrte Nwn mrases Definite? Paper p-3871, The RAND corpc)tstion, Santa Moncia, Callfornie. Kripke, Saul A. f 19723 Namrng and 1Vecersrty. in ,Sementica Natural Lannuene, Davidson and Harman (eds.",
	"pdf_parse": {
	"paper_id": "J78-3018",
	"_pdf_hash": "",
	"abstract": [
	{
	"text": "Fahlman, Scott E. [I9771 A System for Reprasenting and U3ng Real-World K)rowIedp, M1.T. AI. Laboratory AI-XR-450, December. Grosz, Barbara [I9771 The Repbesq(rfafron and Use df Focus iq OIslogue ender+tand~n~. Stanbrd Research Institute Technical Note f 5 1, Menlp Park, Calthrnia Halli:day, MA. and Ruqarya Hasan 119761 Cohe~en in EMlish L~ndon, Longman press. krktunen, Lauri [I9681 What Makes Oefinrte Nwn mrases Definite? Paper p-3871, The RAND corpc)tstion, Santa Moncia, Callfornie. Kripke, Saul A. f 19723 Namrng and 1Vecersrty. in ,Sementica Natural Lannuene, Davidson and Harman (eds.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Abstract",
	"sec_num": null
	}
	],
	"body_text": [
	{
	"text": "Candace L. Sidner MIT Artificial Intelligence Laboratory , Cambridge, Massachusetts When speakers1 utter or write sentences, they use certain words in the sentence to refer to people, places, object, times, events and ideas which exist in th-real world. When sentences are fdrmea into units of twb or more sentences, certain words refer back to othet referring Qxpresslons in the previous sentences. Among the words which can be used to refer to the real world as well as to refer back (which is called thb process of co-raterence) are _noun phrases containing a definite article, such as the (cglled defnps, hereafter).",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction to tho Problem",
	"sec_num": "1."
	},
	{
	"text": "Several aspects of com~rehension of defnps are open problems:",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction to tho Problem",
	"sec_num": "1."
	},
	{
	"text": "1.) What is the definition of the reference of a defnp?",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction to tho Problem",
	"sec_num": "1."
	},
	{
	"text": "That is, what to we mean by reference irr computational linguisitics~ 2.) How are defnpst which are used to co-refer into a discourse distinguished from those which refer to real world objects outside the discourse? 3.) Whaf methods of search will distinguish the referent of a defnp which refers to an entity outside the discourse context' 4.) What different ways can defnps be used to co-refer to other entitles in the discourse?",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction to tho Problem",
	"sec_num": "1."
	},
	{
	"text": "5.) How can co-referenq of'defnps be detected? 6.) What inferences and data structures will be needed, for the detection process? The work of Winograd 1197 1 I, Charniak 119721 and Rieger [I9731 suggest that inferencing is. crucial to the interpretation of co-reference.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction to tho Problem",
	"sec_num": "1."
	},
	{
	"text": "This paper presents a viewpoint from which to answer theseauestibns based on the concept of focus, as developed by Grosz [1977) and the author (Bullwinkle 119771). This paper extends' Grwz' work by qstablishing a framework for communication and a set of rples for use of focus in discourse. The approach taken here represents an alternative to the inference driven schemes of Charniak and Rieger.",
	"cite_spans": [
	{
	"start": 115,
	"end": 127,
	"text": "Grosz [1977)",
	"ref_id": null
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Introduction to tho Problem",
	"sec_num": "1."
	},
	{
	"text": "The description of the communication process given here contains fout simple assumptions which are generally true and will be taken as true in this work First, the speaker is assumed to be communicating about something.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "This assumption implies that the speaker is not speaking gibberish, that there are referring expressrons and either requests, questions, assertions or acknowledgements beinq made. The something which the communication is about will be called the focus of the discourse? Second, the hearer is assumed to be able to identify what the focus qf the discourse is. The speaker wants to communicat_e about something., and for the communication to occur, the hearer must be able to dist~nguish what the speaker is communicating about. Third, the speawr is not trylng to confuse or deceive the hearer. The-speaker uses referring expressions with the intention of referring to someone or something, or with the intention of describing something or some event. In Gricean3 terms, the b y w r d is \"Be perspicuous.\" The final assumption claims that the speaker assumes the hearer knows certain knowledge about the real-world which can be referred to during the communication process. Recent research (Cohen [1978] ), as well as the well known work of Searle [I9691 and Austin [I9621 describe models of the speaker's knowledge of what the hearer believes. I n this chapter, the weakest form of such a model 1 6 assumed:",
	"cite_spans": [
	{
	"start": 988,
	"end": 1001,
	"text": "(Cohen [1978]",
	"ref_id": null
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "the speaker assumes the hearer has enough real-world knowledge in common with the speaker to know the entities which the spebRer refers to, and that knowledge is what the speaker draws on in constructing a message 40r a hearer. These four assumptions will play an important part in the discussion of co-reference interpretatioh which follows.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "1. I wOn .use the term speaker to refw to the producer of a spoken or written drscourse and hearsr to refer to the receiver of the discourse.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "2. 1 don't want fo suggest that only one thing can be communicated in a dlscourse, for speakers do direct their attention from one thing to another. For the moment, I will speak 9f the focus as the first center of rttention in a speaker's dlscourse.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "3. Grice, HP. \"Logic and Conversatlbn\" etc. This paper makes the claim that the focus acts as m indax function for referring expressions. For those referring expressions which are anaphoric, the focus indicates where to look for an anfecendent, For those referring expressions which are names or descriptions of things in the world, the focus acts as a generation center for a process that chooses a r'epresentatian. of a real world entity vihich fits the name or description. However, the focus of a discourse alone is not sufficient to produce theWndexing behavior. The focus must be used hr cmiunction with a hierarchical semantic network of associations. The network will indicate what other concepts are related to the focus. It is a codification of some of the general knowledge speakers and hearers have about the real world. The network is a dynamic structure because the hearer adds t o hisfher general knowledge in the process of interpreting a piece of discourse, Focus must also act with a third piece of computational machinery, an inferencing mechanism. I t is used to infer from general knowledge and some suppositions that a certain proposition is true.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "An example will be helpful here. I n the discourse below, thb focus of discussion is the meeting of DO-1. 00-1 1 want to schedule a meeting with Ira.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "2 I t should be at 3 p. m.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "3",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "We can get toget her in his office. 4 Invite John to come, too.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "All four sentences give information about the focussed entity. Thus in DO both sentences 3 and 4 make no direct reference to the meeting of 00-1. As human hearers, we know that these sentences are related to the rest of 00 because-they give information about the focus meeting. I n 00-3 there are three clues which connect this sentence 2nd the rest of the discourse: the use of get together, the a-reference of we to the participants of the meetfng, and his office establishing a place fop a meeting. DO-4 intraduces an additional participant which can be surmised from the use of invite,%nd the fact that the ellipsis of the w e n t that John IS invited to is the focus.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "A piece of the hierarchical net needed hr DO is given below in figure 1. A prototypd meetings has associated places, times, participants, and purposes.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "The relation between meeting and place is one of occurrence while the relation between meeting and purposes is one of causality. When tXt-1 is encountered, the hierarchical net grows a new member: an instance sf meeting from 00-1. I t inhe* associated entities of meetings and some specific values tor the participant entdty. 00-2 induafes that something (callbd it) will occur at a particular time. The focus of 00-1 is meetihg, so the focus, meeting, is proposed as the antecedent o ! it. To The explanation about the use of focus is not really so simple because the focus of a discourse changes. The interpretation of focus requires a means of recognizing that the focus has changed to some other entity. In 00' the focus begins on meeting, but the it in W -3 has my office as its co-referent, not the meet~ng. Detecting this co-reference requires a means of noticing a shift of focus and using the inferencing mechanism to confirm the choice of to-referent, Focus shift detection will not be discubsed here; the reader is referred t o Bullwinkle [I9771 for a discussion of focus shift where the term \"sub-topic shift\" is used.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "W-1 I want to schedule a meeting with George, Jim, Steve and M~ke.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "2 We can meet in my office. 3 It's kind of small, but the *meeting won' t last very long enyway.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "T b Communication Pracass a d Focus",
	"sec_num": "2."
	},
	{
	"text": "The theory presented here distinguishes two kinds of rafat.ring The first is an idernal reference between a noun phrase and some pre-exlstlng database object. That database object represents a real world entity. 10 Figure 2 below internal reference links the noun phrase NPl wJimmy Carter\" to a representation of Jimmy Carter (who is described as presrdent of the US, etcll. How the noun phrase and the database object refer t q the real world is the classical semantic problem of reference (cf. Kripke [I9721 amng others) and is beyond the scope of this work Tha other kind of referring is co-reference. Co-reference links a nounghrase t o another noun phrsse. The two noun phrases are said to co-refer, aqd both intetnally refer to the sa' me database object, both refer to the real world object, I n Figure 1 , the dashed link from NP2 \"Jtmmy to N P~ is a EO-reference lin4 The dot-dash link from NP2 to the database object is a virtual internal reference link wh~ch results from the co-reference link from NP2 to W I andfrorn the Infenal reference link from FJP' 1 t o t h e d a t a b m Bject. Internal reference aitd co-reference links at distingyished because co-reference links csn be establishd more easily using discqurse contqxt, W,ch will'bs discussed i n detail,later in this paper. In the remainder of thfs paper when 1 speak of inteknal reference, I will drop the phrase \"internal\" and use only \"reference.\" (1) Ooes the expression refer to Someone or something?",
	"cite_spans": [],
	"ref_spans": [
	{
	"start": 215,
	"end": 223,
	"text": "Figure 2",
	"ref_id": null
	},
	{
	"start": 803,
	"end": 811,
	"text": "Figure 1",
	"ref_id": "FIGREF7"
	}
	],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "(2) What conceptual entity in the memory or the databese of the hearer's knawledge, If any, is denoted by f i e referring expression? (3) WhGn does a given ewprgssionn refer to the s a w entlly as another referring expression?",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "The expressibn Julius Caesar 1s used to ,refer, and can refer to the person represented in the hearer's knowledge as Julius These two decisions together with the initial assumptions appear to make &necessary and sufficient conditions for comprehension since b y deciding that Julius Caesar refers uniquely and choosing a conceptual entity, the hearer has decoded what entity the speaker was referring to.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "There are, however, situations where the hearer's choices t o the above decisions and the speaker's intended referent do not coincide. Suppose the hearer decides that Julius Caesar refers uniquely and refers to Julius Caesar, who W&E a Rodan emperor. The speaker may also have intended it t o refer uniquely, but to the author% deceased cat, whose name was Jul~us Caesar. Now there are three possibilities;",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "either the hearer knew about Julius Caesar the cat, but decided the expression referred to Julius &he emperor, or the hearer only knew about the emperor, or fhe hearer didn't know of either. I n the last case, the hearer \"found\" a referent by a chance from randomly linking up the name and some memory representirtion. The last possibility does not fit a description of reference comprehension of any kind. Randomly hooking up information from one's memory t@ what appears to be a referring expression may be e cognitive act, but intuitively no one would call it reference comprdhension.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "In the case w h~r e the hearer only knew about the emperor, it seems safe to conclude that the reference may C have been comprehended, but incompletely. As we shall see, there are many other clues In communicstion aboul the referent of terms than those given by r e f s r r i n~ expressions in isolation.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "Without these, reference comprehenslon is incomplete because the h&er has no merne of knowing whether s/hs may have the wrong referent. Even with the best set of clues, the hearer may still choose Julius the emperor, Hsre ws will say that comprshqnsion has tdkan place, completely but incorrectly, because the b a r e r hrs used all the relevant communication knowledge to decodt, the speaker's message. What can be concluded is that the speaker's rules for reference generation and/gr the spmaker's knowledge bf the hearer is faulty (thereby contradicting the speaker's assumption above).",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "In the ease where the hearer knows of both possibilities and; ch~uses the incorrect one, the hearer may have erred due t@ hilqrs to fotIow other cammunicatkn ctves or again because the epedket's rules and knowledge were lacking. In conclusion, a referring expression fs comprehend& US intended, if and bnfy H th0 same referent as that intended by the speaker Is chosen from the entittes in memory. fhe expression is otherwise just comprehended ehefi thg hearer chooses an entify from mmery which is denoted by the Suppose, for example, that the hearer believes the speaker hates t o even speak of cats. Then the hearer may conclude that Julius Caesar is most likely a reference to the emperor of Rome. I am not going to consider this possibility in the forthcoming discussion; instead I wili restrict the discussion to cues from the communication process. Hearer beliefs raise a separate set of philosophical as well as con)pi&ational problems and entends the scope of this study too broadly. However, the issues are significant In the total picture of reference and co-reference comprehension.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "In the remainder of this paper I will consider co-reference comprehensioh just from the hearer's point of view.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "Thus in discussing referential and co-referential exptessrons, I will be concerned with a model of how the hearer disambiguates these expressions used in drscdurse. By symmetry, ~h e might suppose that the generation of keferential expressions by a speabr could make use of a similar model. Such a supposition will remain untested in this paper and is to be verified by later work Furthermore, I will not be concerned with c~mprehensiori as intended s i~ this pr6CeSS requires the additional information of what ths hearer believes that the speak&r knows about Instead i will point .out at various times how the theory under discussion would need modification if hearer's beliefs were included",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Ratferame jn Computational Term",
	"sec_num": "3."
	},
	{
	"text": "Definite* noun phrases can be used to refer to entities in the real world.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Problam with Definite Noun Phrases",
	"sec_num": "4."
	},
	{
	"text": "Scott, and that when it is strictly used, rr defnp denotes uniquelyd4 Thus by using a definite article, a rperker 1s saying in aftact \"there is one abject In the *world denoted by the phr'ese that f o l l o~s and I mean that one.' Of courbe 4 dsfnp may be used to denotq someone without actud/y &noting anyone, as is the case with the woman who wrote jhfurl& This defnp is used to refer tp someone, but there is no conceptual representation in the 'hearer's (or for that matter, the speaker's) msrngry which corresponds to a real world -'4. By stictly used Russell h a n s used without ambiguity. entity assutnhrg the normal case5 Nothln(l in the s y n t r t k or semantic form of the expresskn itself suggest@ that the expressian has, na denotatiah Hsw can th# b8f4s determine whether the deftrp refer!? fq sdmeQ(1Q 08 flofqf Course, if there exists a memory entity the author 6f \\rylm6rk which Zs attributed as male, the hearer can decide thrb thrp wprsssion does not refer to anyone bn the ba~lb fat 1) mnfr~dfctlon, E h t if no memoty entity exists, the hearer cannot &el& whether the WOM8n who wrote Wsverly refeta fa anyom' This way of looking at defnps, however, fails to acwunfi for dl the phenomena of defnps because it involves an assump?lon which is not' true.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "The Russelli'an analysis has difficulties because defnps are not always used to refer! The problem is not only whether a particular defnp actually denote$ a real world object$ i t is alsb a question of whether tha defnps Is intended t o refer at all. Even more surprising, a defnp may be used to refer, but the speaner may not intend for the hearer to know the referent of the defnpr the defnp form Is used to indicafe that the referent is knowable, but possibly not significant for the communication at hand, OonneUan [19?7f points out that some defnps are used attributively. If we happen upon Smith who lies dead with foul wounds, one can say Fmith' s murderer is insane.* Used attributively, Smlth's murderer does not refer t o anyone, and the phrase does not dpscribb a pprticular person. I t is as if to say, Smrth was murdered and the murderer, whoever that may be, is insane. Thys the speaker using an attributive defnp does not assume that someone fits the description, whereas with a referential defnp the speaker expects the hearer to realize who Is being pointed to.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "The other distrnct~on a speaker can make is to use a defnp t o indicate that the referent is knowable. Thus if OW says:",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "(1) Larry read a ID^ of l~ngu~strcs in the hospUal, (2) Carry read a lot of linguistics in a hosp~tal, the (2) usage is not the same as the (1). While the hearer doesnot know which hospital the hospital refers to, it is clear it refers to some particular one. Comprehbnsion of the referentla1 term does natinvolvs finding a memory entity which represents the real world entity that the expression refers to.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "For reference comprehension, thfb ceneern is considerable the Chinese guv~rnment In (3) does not demand reference deambiguation, while 154) does; the disambiguation isaifficult because the expression cen refer to more than on% thing.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "(3) John got help from the Chlnese government in adopting an Orlantal child, (4) Get a visa for your trip from the Chfness govsrnmgmt.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "Another difficulty with defnps Is that sometimes they are used not to r e t q to or to describe spmific individuals or abjects, but to charecter~re o class of entities with the 6, Possible world semantrcs will not be discussed here. Iksues of transworld identity qnd designatiotl by defiMe asscriptions may require more machinery than Is considered m m properties of the head noqn phrase and any of its modifiers. THUS (el) used in this way does not refer to m individual. I t character~zes. a member of the class of Individuals who are men and boak writers. (el) is similar to attribution except that the description applies ta a class.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "(el) the man who writes books So far, then, the following clas%es of defnps can be stated. A defnp that is used to refer uniquely to one entity; whether onot such an entity exists in the real world, is a s~e c i f~c defnb.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "A defnp that character~zes a class of entities by meam of an nd~v~dual whose propetties are delineated by the properties rf the head noun phrase and its mod~f~ers is a generic defnp.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "A defnp is attributive ~f ~t describes an entity without eferring. A defnp can be amb~~uous in use (u-ambiguous rereafter) t t its use as a speclfi9 attributive or generic is not dent~fiable, whtle a defnp is ambiguous & reference r-ambiguous hereafter) if it is used specifically and there is nore than one ~n t~t y flf ting the description of the defnp.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Russell [I9051 says of the expression the suthor of Waverly that it denotes Sir Walter",
	"sec_num": null
	},
	{
	"text": "L~t t l e of language, ~f any at all, is sa~d without some surround~og contexts of information. For example, most conversat~ons hgppen In a location where there are other objects present. Mbst stor~es have at least the context of there being a story teller, a hearer and the story being told. There are contexts. wrth more presumed common knowledge, such as what the hearer knows of the speaker's own identity or some shared additional informatton between them (ag. they have children or parents in common).",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Gonorrl Role ef Conhxt in Disambiguation",
	"sec_num": "5."
	},
	{
	"text": "Contexts arb needed to determine what a detnp refers to. I f I say (5), when I am standing in my kitchen with a frtend, the defnp, if specrfrc, must refer to some unique object in the world.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Gonorrl Role ef Conhxt in Disambiguation",
	"sec_num": "5."
	},
	{
	"text": "(51 Get me the hot dtdh holder. There may be lots of hot drsh holders denoted by conceptual entities i n my fr~end's mind, but I an-referrrng to a specific one. Since nethlng in (5) distinguishes the one I mean from the whole collection, erther 1 have mtsused the language, or there is a contoxt whlcb contains only one such hot dish holder, and my friend ' l s aware of that context at the time of my saying (5). In this case, the necessary context is the kitchen, and the referent is probably an item in the kitchen. Reference made to 6n object external to the conversion in called extra-sentbntlal reference. ft is discussed here to exemplify the role of gne clasS of contexts used in reference deternrinatton I call contexts bf reference which q~i s t in addltianal to the one created by the discourse Implicit contexts.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Gonorrl Role ef Conhxt in Disambiguation",
	"sec_num": "5."
	},
	{
	"text": "In this pappr 1 will show how use of Implicit contexts can Wold the problem of searching a general database for the entities denoted by defnpr.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Gonorrl Role ef Conhxt in Disambiguation",
	"sec_num": "5."
	},
	{
	"text": "Other defnps make use of different implicit contexts. Instesd of an implicit context consisting a f objects near the speaker, the implicit context may be events that the speaker 'belle~es~are common to the bearer. The speaker who opens a dialogue with (6) bel~w is assurnbg some prevlous confkxt (a discussion with the hasrer or some other situation) where the reference of the A!. Lab Lunguap Oroup was first established.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Gonorrl Role ef Conhxt in Disambiguation",
	"sec_num": "5."
	},
	{
	"text": "In (71, the speaker i r r esain assurnirrg 8 prb-sstrblished referent, but since the hearer mav know of several different dogs, some specific cbntext must be chosen that will distinguish a single dog. Later in this paper some heurisltics for choosing a context will be drscussed.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Gonorrl Role ef Conhxt in Disambiguation",
	"sec_num": "5."
	},
	{
	"text": "(6) The A.I. Lab Language Group wants to meet next week.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Gonorrl Role ef Conhxt in Disambiguation",
	"sec_num": "5."
	},
	{
	"text": "The dog is sick again, Contextual informafi~n of yet another kind appears in story telling. At the beginning of a story, the hearer expects characters t o be introduced. Sometimes this is done with indefinite noun phroses, which are a way all -discourses introduce new items, butl often a story-teller uses names or defnps bs (8) below shows.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "(7)",
	"sec_num": null
	},
	{
	"text": "(8) The -heiress, lived the l~f e of a recluse. She died under mysterious circumstances, but the murderer was never found.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "(7)",
	"sec_num": null
	},
	{
	"text": "(8) is not a case of cataphor~c referencing (referring forward i n a text) stnce the phrase thaheiress can fully specify an object itself. However, heaters of (8) do not have to search their memories for a referent to the heiress in (8). They use the context of story begrnnrng to guide them in reference disambiguation.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "(7)",
	"sec_num": null
	},
	{
	"text": "Defnps must be dtsambiguated as generic or non-generic. As wi'll be shown later, generics in the midst of a discourse can be easily d~samb~guated, but in an initial sentence only implicit contexts may exist in which a co-referent can be found to use in d~sambtguation. Impl~cit cohtexts may be helpful in some cases, but in general they are not sufficient to indicate Qhe interpretat~on. However certain rules can be postulated based on observable 6entence\"ata. This data tndtcates that there are several levels of sententtal and phrasal infor mat ibn used for d~sambiguatilon. The rujes which will bm su,rrmari2ed6 here give preferences for generic and non-gener it readings. Some rules govern whether the defnp itself is preferred as a generic or nqn-generic reading. A small collection of phrases like the sun, the moon and the presrdent default to specific, well def~ned entifiles. Which entities are defaulted to depends on the presence or absence of an implicit context to which tlle phrase may co-refer. For other defnps, a",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "\"yes\" answer t o ((1) of the following questions indicates a specific reading preference, wh~le a \"yes\" 10 (2) and 3indicates a generic. I.) Is !here 9 specif~c indtvgdual so described? 2.) Is there 13 class use acceptabjp for this pair1 3 ) Are there many indivtduals described but none outstanding?",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "\"No\" answers to all causes a preferred specific reading. In addition to phrasal preferences, predicate argument relations for cert-ain verbs may indicate a preferred non-generic reading. I n these cases, a u-ambiguous defnp will b e taken as a non-generic, an& a generic defnp will cause an odd sentence.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "(9) The black man was movihg towards the window.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "(10) The woman who reads Total Woman is coming to dinner on Saturday.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "Some classes of speech acts are also dist~nctly generic. The is-s sentence below is always generic1 this reading may result from the use of is-a to indicate further characteristics of the subject. The announcement speech act in (12) is generic its tong as an implicit context does not e~i s t which contains en acceptable co-reference for the defnp.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "(1 1) The elephant 2 large mammal.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "(12) 1 went to tell 1r6u about the orangetang.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "Speaker-hearer assumptions about perspicuity can force a reading to be generic or nonrgenerk, The defnp in (13) is forced to be read generically because a specific reading would be r--ambiguous and therefore not perspicuous. On the other hand, (14) is odd since mv~fe requires a non-genetic object.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "Ibwever, because the speaker is assumed to be maintaining perspicu~ty, the hearer may attempt to read the defnp in (14) as a non-generic.",
	"cite_spans": [],
	"ref_spans": [
	{
	"start": 115,
	"end": 119,
	"text": "(14)",
	"ref_id": "FIGREF3"
	}
	],
	"eq_spans": [],
	"section": "Distinguishing Celerics",
	"sec_num": "6."
	},
	{
	"text": "Boston's soctal unrost. ",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "( I 3) 6111 considers the black man to be the source of",
	"sec_num": null
	},
	{
	"text": "Using the cancept of focus, rules governing the co-referential use of defnps in discourse can be stated. The rules for defnp co-reference which foitoy depend upon the ability of the hearer to identify focus. This process is a complex one and will not be d~scussed here. The reader is referred to Sidner [forthcoming] for full details. I n brief, the focus o f a sentence depends upon predicate eggurnent relations and in same cases, speclal syntactic farms, such as clefts and pseudoclef Is.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "-The Explicit Backwards Go-referone-Rule for h f n p Dirrambiglration",
	"sec_num": "7"
	},
	{
	"text": "In t h e simplest formulation, the rules for defnp co-reference states: the d~scourse focus provides s reference point for the co-references of defnps. As I will show%elow, the rule contains several subparts which must be stated separately. In this paper I will refer to cases of a defnp used anaphorlcally as explicit, backwards co-referencing (EBC). Tha E8C rule states that e defnp with the same noun phrase heed as the focus, and which appesrs in a sentence following the sentence with the focus, Is co-referential to the focus. 02-1 I ' m goihg to tell you about the eie hant e.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "-The Explicit Backwards Go-referone-Rule for h f n p Dirrambiglration",
	"sec_num": "7"
	},
	{
	"text": "The elephanti is the largest of he jungle mammals.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "---I",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "3 & weighs over 3000 pounds.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "4 At one point in i Q existence, the elephant8 had protect itself from the lion, 5 but now its;bherds are so large, that most lions won't even venture near.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "What the reader will nbtice about 02 is not only the co-referencislity of focus for the second and third uses of the elephant, but also the co-referenciality implies that these uses are generic. Where defnps in isolation are often ambiguous on the generic-specific ~Iassification, in discourse context, this rarely occurs since the focus provides the class type for the defnp. As stated, the EBC rule makes a true predtction about u-ambiguous defnps which occur in sentences fotlowing the focus: they are CP-referential with the focus, and hence disambiguated as non-generic. 03-2 below contains a defnp which is u-ambiguous in isblation, but in the discourse context, it refets to George's elephant, the reference of the focus.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "03-1 1 sent George an elephant last year for a biithday present.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "2 The elephant likes potatoes for breakfast.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "The EBC rule is inacqurate when applied to strictly generic defnps, and where it fails, the role of phrasal and sentential level processing in co-reference comprehension is indicated. 0 4 is an ind~cation of the problem. D4-2 is generlc in dsolation. Even in the context of D4, where the focus is Mary's ferret, hearers interpret the underlined defnp as generic.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "194-1 Mary got a ferret for Christmas last year. 2 The ierret is a very rare animal.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "The context cues of \"discourse are not strong enough to reverse a strongly generic reading of a defnp. In order for this to be so, sententral level processing must have occurred without consideration of the demands of the context. Sinc~ the EBC rule as stated predicts CQ-reference in cases like 04, it must be revised: specific and u-ambiguous defnps which contain the same noun phrase head as the focus, and which follow the focus in the dtscourse, co-refer with the focus.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "A further refinement on the EBC rule is needed. Consider the fairy ta/e book in 05-2. The E& rules predicts it FJHI be co-rehrential with the focus of book in 05-1. In It seems that dafnps which co-refer with the focus canno contain anymore informatiw than is known about the focus.",
	"cite_spans": [
	{
	"start": 165,
	"end": 167,
	"text": "In",
	"ref_id": null
	}
	],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "Thus one could say following 05-1. \"The book I bought is a fairy tale book by the Brothers Giimmm (slnce 05-1 states that ?he speaker bought the book), but one cannot say 05-2. Why can't a detnp that contains more information than the focus co-refer to the focus? Returning to the discussion of focus-shift earlier, a referring expresslbn following the focus is either co-referential to the focus or introduces an entity which is the potential new focus of the discourse. The difficulty with phrases like the fa;ry tde book is that one cannot tell if it is intended to co-refer, or because it is somewhat different from the focus, intended to be used as a potential new focus. The EBC rule mus! be revised to state: specific and u-ambiguous defnps which contain the same noun phrase head as the focus, which follow the focus in the discourse, and which do not contain more information than is known about the focus co-refer with the focus.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "The EBC rule explains why a negative qistential cannot be referred to using a defnp. A sample case, from Karttunen 119683, is given in 06. 06-2 is generally regarded as an unacceptable sentence followihg D6-1. The sehtence is certainly grammatical, so the assumption by Karttuwn is that the referential term the car is being used in sgme inappropriate manner.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "2",
	"sec_num": null
	},
	{
	"text": "2 + The car is black.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "06-1 I don't have a. car.",
	"sec_num": null
	},
	{
	"text": "Thq EBC kule predicts thqt the car co-refers with the focws in 06-1. But a car in that sarnar~c~does not have a referent (because the speaker has just said so). Thus the use of the defnp in 06-2 causes the hearer tB expect a reference wher in fact'there is no rsfer~ntiarentity. Thus *as long as there is a co-referan, &ntity specified b y the focus, a defnp may be W d , but the predication about the defnp must be semanMIly mcanirigful.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "06-1 I don't have a. car.",
	"sec_num": null
	},
	{
	"text": "Another form of expllcit fiackwards co-reierencing is slightly different fhdh the previous examples. II involves the use of lexical geneyaltrPltion af the f&us. Grosz /1977] first categorized the reKtion of focus and defnps with a more general noun in the ~rolin phrase head. Irr97, t5a pear old beast is a 1exM1 generalization of the dog, that is, its head noun is a Mrm which is a dasd generalization of the focus. 7 7. This term comes from the a'bservetion of Halliday end hasan 119761 that lexical cohesiaa includes the use of reiteration of four types: same word! synonym, superordinate, and general word.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "06-1 I don't have a. car.",
	"sec_num": null
	},
	{
	"text": "[letsrmitbing ths class generaltzarion ot tne tocus Ss possiole when the focus is represented J n the way that is assumed fn this paper: as an association network wlth an is-a hierarchical structure. Using that hierarchy, i t is pbssible to determine whether s phrase Jike bast I S hierarchically related to Salamut.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "06-1 I don't have a. car.",
	"sec_num": null
	},
	{
	"text": "Salamuh t6 the vet yesterday.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "07-1 Harold took",
	"sec_num": null
	},
	{
	"text": "2 The Door old beasti was quite fame.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "07-1 Harold took",
	"sec_num": null
	},
	{
	"text": "One might expect that some constraint on the amount of information in ,the lexical generdization, of the focus is peeded. This is the case, shice the wrderlinedRJefnp in 07-2'",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "07-1 Harold took",
	"sec_num": null
	},
	{
	"text": "is unacceptable following D7-f es a cb-referent wlth the focus.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "07-1 Harold took",
	"sec_num": null
	},
	{
	"text": "The beast who is old was qdte lam& -2\" The mangy, snarling, unfriendly beast was qulte lame. I t qppears from ail the cases f can find that any post-nominal modifiers or, a noun which is a lexic~l generaiizatton of the, focus force the defnp to be non-coreferentilal with the fbcus, while pre-namlnet dtadrfiers, no matter how complex, preserve co-referentiality. I t is unclear wh)r pre-norninal modifiers and post-nominal modrfiers have them different behevidrs.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "07-2'",
	"sec_num": null
	},
	{
	"text": "M a w deftnite Houri phrases which occur in discourse are not sKas of b X U W W d S co-refsrenoa to the focus. Grosz f 1977) st . . , , j est ed thdt the focus &inas other items implicitly into fogus as well, b y means of ass~aation. Such detnps are related to the focus in one af several ways. Slncb the focus is MI speclfred, these relationsh~ds can 'be easily determined. The focus rcts as an anchor pornt for findjng references for such defMps. In*, the defnp the time refers to the time of the ckfcourse focus, the meeting. Phis defnp use I will call impl~cib backyards co-reference. Such caws are to be distinguished from explic~f backwards s~o-referencing becduse the defnp is co-referential wjth an ent~ty that is closely associated with the focus rather than to the focus itself, The phenomenon of association between #wo noun phrases has h e n cited by Norman and Rumelhart ,119751",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "08-1 The pa group wants ro have a meeting.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "2 The time will be 3 p.m. on Tuesday.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "Implicit backwards co-rekrencing is c~nstratned by the association network surroundin& the focus. Any entity closely associated with the entity M C \" b represants the focus can be mentioned using a slmple aefnp, Thus in 09, sentences With Gccsptable detnps as wall as oms with macceptabls dsfnps ere glven.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "09-1 1 went t o a new restaurant with Sam.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "2 The waittr4ss wr$ nasty.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "3 The food was great.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "4 The soup was sa fy, brat the wine was good.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "5 * The rug was ugly. Such bn infe'rence Is possible given a Fahlman 139773 tygs net with i w b infererne points like heiress and murderer (and the information associated to heiress from the context thus far); from the net, Phe relation of the murderer to the heirmss &an be inferred. Such an inference does n~t produce a real-world enttty to which the murderer refgrs. Instead the Iderred reliitlon of murderer and heiress provides sufficient fnformal4on to produce the entity if It exists i n the database. When a denotation does not exist in fhe database, the inference between the mrepderer and the heiress sugeste that the speaker is attributing of come individual that s / h is e murderer.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "A O l l -1 I was driving on the freeway !ha o \\ b r d~.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "2 Suddenly tfm engine began to make a f a n v noise.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "8 1 stopped the car.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "4 When 1 opened the hood, t saw that the radiator was boiling over.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "With focus of freeway in Di 1-1, the relation of the engine cm be found since vekicles are driven on freeways a d vehicles have engines. 3 The last mesfing, which was at 9, was too tate, so schedule thfl bne earlier.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "F r m these examples, the nature of focus in discourse can be re-empnasized. I t is the focus which connects sentences of the drscoutse.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "In the process of determining the focus relat~on between s defnp and the focus, the link in the discourse is created. From these examples, one can predict that there ought to be cases of def nps which bear none of these relations to the focus end which result in unacceptable d~scourses, This prediction is qccurate as D l 2'",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "shows.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "012'-la I went downtown today, 1b and there were clowns pedorming in the square.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "2 * 1 saw the chak The focus rn 014 is the seffrinar of D14-1. The mah department cha~rman L ndt directty related to the seminar.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "Howeyer, tJm focus doe* provide an important plece ot .inforrnat~on; ~t is the sourea of the ellided event to which the cha~rrnan is invited. I rCence~ a14 is different from +-6 or 012' where there is no link befween the focus and the sentence in question. What can be: toncl~ded is that the focus is not a How can t ht$ denotation of the chairman of the math department be found? Slam the denotat~on of the chairman of the math department lies outside the discourse context) a more global context such as that of the speaker's situtation in time and spare must be used to determine a context o+ possible referen~s. Thts conte%fi &must be limited because 'thbre ar& potentially many mafh department chairmen in the speaker's an$ hearer's memories, I do not intend to describe just what such a ,ontext w~l l k o k Itbe, but 1 dawant to ind~cate that i t may be \"grown\" froM a search through the associative network to other ent~ties whlch are related to any of the objects in the d~scourse, incloding the speaker and hearer, The association net includes not only abstract represe$,ptions of genefal classes of ma1 wodd entities, it also contains representations of real world objects. The lassoc~atlons between real world objects can be gethered by a search method Wfch collects assodations close t 6 the W s and then extends for other assoc~ations until one Is found that matches the defnp in qwstton. Of course, it is possibfe that no entity will be found: such a circumstance is yet another mxarnple of the hearer knowing that a dafarr refers withobt being able to tell who the speaker had intended as the reference. The implications of ext ra-discourse contexts for computat ional models is clear: models are needed of what the speaker assumes hislher hearer knows about, so that the speaker does not produce referring expressions which the hearer cannot disambiguate, and models are needed of what the hearer assumes the speaker has beliefs about so that the hearer can decide what t o do with referring expressions which s/he cannot disambiguate.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "Eallrer in thls paper I mentioned the use of such aernps as the sun, the moon, and the Earth. These defnps have default referents in initial sentences of a discourse. Inside a discourse, they can act in one of two ways: related to the focus as pther a backwards co-reference, implicit co-reference, inferred reference, set-element reference or computed reference; or they may act as references to entities outside the context, The EBC rule predicts that such defnps will be taken as co-referring when the focus has the same defnp head. I hus a defnp like the sun or the moon will refer t o its default value only if the focus does not predict a referent based ow one of the five co-reference relations discussed here. These kinds of defnps are especially noteworthy because they are a clear example of e defnp that can be used in either role.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "The tour related co-reference relations specify ways in whrch a defnp can be oredtcted as co-reference to an entity associated wdh the focus.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "Other defnps either refer to objects outside the discourse context or the defnp fails to refer as intended by the speaker. The former case is marked by the presence of d~u o u r r e links elsewhere in the sentence t o the focus on by the use ol default reference. in those cases where the reference as intended fails, the hearer may attempt t o create a connection to the focus, as was suggested with 1310-7, and thereby maintain the perspicuity contract. Alternatively, the hearer msiy h r l to understand the referent.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "FOCUS can be used for dPsambiguati6n of generic defnps in a manner similar to the cases presented above. As discussed earlier, the E8C rule predicts that a defnp which is u-ambiguous will be generic or not based on the focus, andD that a strictly generic detnp is independent of the foms. Ths strictly generic defnp case, as shown in 015, p r e s a i s one means of shifting focus in a discourse with defnps.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "D15-1 I got a new ASR 33 thls week.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "2 The AW 33 15 an old but reliable output device, I f 015-3 were \"It was available long before the newer electronfic consoles,\" the focus would have shifted from the ASR33 which the speaker had gotten to the generic entity of ASR33 teletypes.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "The strictly generrc defnp used after a non-gemric focus is just one case of implrclt backwards co-retqrencp using associations. Other assoctatlpns occur as welt. nowevw, Implicitly related defnps are nat distinguishable as generic unless a full modifyiag nerm dwsz is attached, as 016 bhow.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "The hme as a simple defnp can be used only as en implicit co-reference to the focus of party. Only the complex noun phrase has the syntactic and semantic distinctions which reflect the generic usage. The simple defnp used implicitly takes its genericlspecific classification from the focus.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "D16-1 I want to have a party. l a The time of a party is hard to decide on, 2b The tide is hard to decide on.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "Set -element implicit co-reference exists for generic foci as well as specific focf. A significant difference is that the foci for the generic case can be a singular defnp, gr they can be a plural noun phrase with either 8 definite article or no article. The set mdmbersMp is indicated by a distinguishing modifier, just as with specific set-element Implicit oo-reference.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "An example of generic set-elerneit co-reference is given below with both a singular defnp focus and a plural unspecified focus. week. 2 I think thp owner of a motorcycle ought ta be required to take &wing lessons.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "The o w m of a nrblarcycle is generically related to the first sentence b y m inference of what happened to the agent. The same kind of machinery ttrd is used for specific inferred co-references can be used for making these inferences as well.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "HOW can generic inferred co-references be distinguished t o m specific inferred co-reference? A drictly generic defnp as in 018'2 remains generic. Those defnps which are u-ambiguous at the sententral levd, as wlth 010-2, can be dissmbiguated as specific because of the relation to the focus.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "The use of *a semantic network with a focus afid inference mechanism results in a computational theory of co-reference which makes use of representation properties such as prototype of ~rrtit~es, hierarchical connections and associative links between entities in the representation. The use of focus for co-reference rilles such as the EllC rules, implicit backwords co-reference rules and inferred co-reference relies on this net representation. An l~ference mechanism is necessary both to verify . co-reference predictions and to test suppositions used in inferential to-reference.",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	},
	{
	"text": "With the net arid a focua, predictlow about acceptabh co-reference for noun phrases has been shown and verlf ied with linguistic Bvldence. Psychofoglcal predict tons, such as implicit ca-reference requiring more time than expiitit co-reference, can also be tested although the related psychological literature has not been discussed in t h~s paper h e llimfts %f focus as a co-reference mechanism suggest that focus is central. for co-feference of noun phrases related to previous discotlure. For noun phrases that refer outside the discourse,-focys may also be used to generate a context of entities tram which a co-teferent may be chosen. Further research can extend the focus mechanism to rules. involving",
	"cite_spans": [],
	"ref_spans": [],
	"eq_spans": [],
	"section": "Implicit Backwards Co-rrf6renca",
	"sec_num": "8."
	}
	],
	"back_matter": [],
	"bib_entries": {},
	"ref_entries": {
	"FIGREF0": {
	"type_str": "figure",
	"uris": null,
	"num": null,
	"text": "-at-3 partnipants: Stanoczyk, Lewin confirm the proposal, the Inference mechanism~checks to see If meetings occur at times. Indeed they do, so the proposal of meeting as antecedent of it is accepted."
	},
	"FIGREF1": {
	"type_str": "figure",
	"uris": null,
	"num": null,
	"text": "f he following questJons about the Use of referential terms in natural language:"
	},
	"FIGREF2": {
	"type_str": "figure",
	"uris": null,
	"num": null,
	"text": "referring expression using all the available communication clues but does not choose the same entity as intended by the speaker. An expression wilt be. considered incompletely comprehended if the hearer fails to use all the communication cues available at the time the communication occurs. So far 1 have not considered the possibility of error o n the patt of the hearer because of the hearer's beliefs."
	},
	"FIGREF3": {
	"type_str": "figure",
	"uris": null,
	"num": null,
	"text": "Invite the man who reads The Grapes of Wrath to dinner, I warrt t o emphasize that the noun phrase, verb phrase and sentonce level rules are only preferences far readings. I n tHe w o r d +ase, as (15) shows, an initial sentence may contain a u-ambiguous phrase which, while phferred as getlePic, can be used either way. (15) The robot is r e~l a c i rig Ine car."
	},
	"FIGREF4": {
	"type_str": "figure",
	"uris": null,
	"num": null,
	"text": "The more c o m n , forms of explicit backwards co-jaferenclng are found in D l and 02 below: 01-1 1 want to have z M p a r t y j with Ipts of guests.2 The party; ought to be on Saturday so everyone can come."
	},
	"FIGREF5": {
	"type_str": "figure",
	"uris": null,
	"num": null,
	"text": "The d~f f~c u l t y with D l T is that a chair k not' associated with clowns, and the discourse does not suggest sny supposltlons that would link chairs and clowns iWemntially, There are, however, cases of aetnps which do not beat any relation to the focus and which are perfectlp acceptable in discourse. Cansfder the chairman of the math depurtment in Dl4 below, 014-1 George wants to have a sehinar to dicuss representation in frame-like languages.2 lb wants to invite the chairman of the math departmertf,"
	},
	"FIGREF6": {
	"type_str": "figure",
	"uris": null,
	"num": null,
	"text": "usefur reference point for the referent of the chairman of the math department. The santetrce b no? odd because of focus links. Thus something Cats& of the discdurse context must contam the needed dennfatl~n This conclusion points to the lrmitatton of the focus% 44 captures -oniy those reference relations which are infernat tt$ the digicourse. I n 8 sense, the focus i s a summary of the discourse conf-~t a d what ths hearer knows about it."
	},
	"FIGREF7": {
	"type_str": "figure",
	"uris": null,
	"num": null,
	"text": "The Aust railian aborigine represents an almost ext lnct hunter-scavenger social group. 1' Austra~llan aborigines represent an slmost extinct hunter-scavenger social group. 2 The abor~g~ne in the southern sections of Austrajlia sometimes gathers food, but the other aborigines do not. Inferred generic co-references also occur. -In 018, the owof a motorcycle is a generic defnp: CN8-1 Alronso was in an accident with a motorcycle last"
	},
	"TABREF0": {
	"html": null,
	"type_str": "table",
	"content": "<table><tr><td>n , eo-reference ---b Wl \"Jimmy Csrter\" Y</td><td>\\ NP2 \"Jimmy\"</td></tr><tr><td>/</td><td>\\</td></tr><tr><td colspan=\"2\">virtual</td><td>1</td></tr><tr><td colspan=\"3\">internal r e f e r e n d /</td></tr><tr><td/><td>K *</td></tr><tr><td colspan=\"3\">Oatabese Representation of Jimmy Carter</td></tr><tr><td colspan=\"2\">Name: Jimmy Carter</td></tr><tr><td colspan=\"2\">occupation: President of US</td></tr><tr><td colspan=\"2\">birthplace: Georgia</td></tr><tr><td colspan=\"3\">Caesar. T,o answer the first question above, the beaker must</td></tr><tr><td colspan=\"3\">decide that names are rea~rting expressions. To answer the</td></tr><tr><td colspan=\"3\">second question, the b a r e r must decide 1) whether Julius</td></tr><tr><td colspan=\"3\">Caesar refers uniquely und 2) what conceptual entity in the</td></tr><tr><td colspan=\"3\">heamr's memory represents the hearer's real-world referent.</td></tr></table>",
	"text": "Fig. 2. Rmferomo Link ~~~~~n Noun Phrcrn",
	"num": null
	},
	"TABREF2": {
	"html": null,
	"type_str": "table",
	"content": "<table><tr><td>fact,</td></tr><tr><td>English speakers find 05-2 B5-1 I bought a book today.</td></tr><tr><td>2</td></tr></table>",
	"text": "an odd sentence Jn the discourse because it is not clrar what the fakg tale book has to do with the rest of 05. The fairy tale book is by the Brothers Grimm.",
	"num": null
	},
	"TABREF4": {
	"html": null,
	"type_str": "table",
	"content": "<table><tr><td colspan=\"2\">I f I39 included 09-7 below, r a e h a r e m mfght attempt t i</td></tr><tr><td>-7</td><td>5he elephant with the green tutu danced an</td></tr><tr><td colspan=\"2\">impressive Jig.</td></tr><tr><td>speaker.</td><td/></tr><tr><td colspan=\"2\">010-1 The helritss ilved the life of a recluse.</td></tr><tr><td/><td>2 $he dled u~d e r mysterious c\\rcumstames, but</td></tr><tr><td/><td>the murderer Was never found.(</td></tr><tr><td>i n</td><td/></tr><tr><td>Non-ahnple defnps have inflh!tdy more varlbty because the</td><td/></tr><tr><td>m o d i k r s can specify the rslatton of the defnp to the focus at</td><td/></tr><tr><td>hand as in 09-6. Non-simple defnps which do not suggest</td><td/></tr><tr><td>some con~ectlon are less ecceptable, but hearers, i n reliance</td><td/></tr><tr><td>a~t the petsptcuity maxim, may attempt some connection. T h a</td><td/></tr></table>",
	"text": "Jike the band that plays there.Another use of focus is as an inference point for infert'ed co-references. Inferred to-references, ike UJB murd8rw in 48), presented here as 014 are not mentioned explicitly In the preyious discourse nor can they be @mlbred closely essocibt6d b the focus On general principles. Their use ratiects arr inference about the focus an the part of We D6, the murderer represents an inference that the hires$ death was due to a specific type of circumstance, a murder.",
	"num": null
	},
	"TABREF5": {
	"html": null,
	"type_str": "table",
	"content": "<table><tr><td>the ref6rring expresston. However. another hearer upon</td></tr></table>",
	"text": "concrete example will illustrate my point. Suppose the hearer knows t a t the herress r s s killed b y knss. then an hearing DlO, fb k&afibr npt only contludes that the murderer refers to the murderer d the heihss, but alsa based on that conclusion, the hearer decides that Jones is denoted by points out the lettributtonal use of dxprassions whlch tbhnetlen has observed, The argument presentad here Is nof only about the nature of focus; it is a statemen? of whati$orrrration Is sufficient to hake up II descriptibn which uan d w p t e ;a unlque entity. Viewing ,inferred co-reference defnps as atttlbtdions*has ata Implication for a compu~atlonrl nrbdel whleh disrrrnblguates sm)r defnps. Thiq model m b t be rbis to use an exprearrion without kneblng Its referent hnd be able to link iup the dr)nbtstion to fhe ieferring expression if sown knodeclga m&es that denotation available at r a t e r polnt.",
	"num": null
	},
	"TABREF6": {
	"html": null,
	"type_str": "table",
	"content": "<table><tr><td/><td/><td>The ass0cidion chain here suggests that the</td></tr><tr><td colspan=\"3\">connection between 011-3 and the e n g h Inv~lves a few</td></tr><tr><td colspan=\"3\">inferences. These inferences ere part of 6 hearstqs general</td></tr><tr><td colspan=\"3\">knowledge and true of the world. They are part of the</td></tr><tr><td colspan=\"3\">knowledge in the association network. With &%O, )lswwr, the</td></tr><tr><td colspan=\"3\">inference about the murderer fnvolves a surmodtiarf which C</td></tr><tr><td colspan=\"3\">not necessarily true, since &lng u d t mysterious</td></tr><tr><td colspan=\"3\">oirsumstances does not necessarit), Implp cawdea; The</td></tr><tr><td colspan=\"3\">disfinction between implicff and infetred co-referm~e csfi thus</td></tr><tr><td colspan=\"3\">be stated: implicit co-reference involves Infern-</td><td>w)rtcfr are</td></tr><tr><td colspan=\"3\">true about the world, while inferred co-refsmrrca Smtv8s; a</td></tr><tr><td colspan=\"3\">supposition which the speaRer has: made which is</td></tr><tr><td colspan=\"2\">necessarily true.</td></tr><tr><td/><td colspan=\"2\">Another kind of i m~l i t i t focus-defw relatian exists in</td></tr><tr><td>01-2.</td><td colspan=\"2\">call this relation the set-element relaffzn slnm the</td></tr><tr><td colspan=\"3\">d~w n with r unicycle is an dement of tha set of clowns whlth</td></tr><tr><td colspan=\"3\">ttm focus denotes.</td></tr><tr><td/><td colspan=\"2\">Of 2-la I went downtown today,</td></tr><tr><td/><td>l%</td><td>and there were clowns wrformks in the</td></tr><tr><td/><td colspan=\"2\">square.</td></tr><tr><td/><td>2</td><td>The clown with a unicvcfa-did thb r d v</td></tr><tr><td/><td colspan=\"2\">fantastic stunt,</td></tr><tr><td/><td/><td>Tfieee easc)~ wej ellstet fa</td></tr><tr><td colspan=\"3\">dfstinguish than those of inferred referem b#8trse &$ back</td></tr><tr><td colspan=\"3\">noun is the singdar of the noun phrase teg~c#efiW in the</td></tr><tr><td colspan=\"3\">fwus W i e defnps using; the E N r*</td><td>$d-akmr\\t</td></tr><tr><td colspan=\"3\">to-reference demands a nrodifkt tHaf dibfln@&m@ @ tkam ttre</td></tr><tr><td colspan=\"3\">focus, Without fhe modifier, there is na meats 6 l c&&mWrrg</td></tr><tr><td colspan=\"3\">which member of the set b b e i g discureed</td><td>[1$'?7]</td></tr><tr><td colspan=\"3\">says of cases like t?w c m be</td></tr><tr><td colspan=\"3\">established without t k heed for Intereke.</td></tr><tr><td/><td colspan=\"2\">Another kind of focus relsii6n, M c h f call complrbd</td></tr><tr><td colspan=\"3\">reference, can be seeti in meeflng</td></tr><tr><td/><td colspan=\"2\">this week</td></tr><tr><td/><td colspan=\"2\">2 The rneetiq shodd be on Tuesd6y.</td></tr></table>",
	"text": "As With inferred references, the focus does rrbt mb ft prrssible to identify a specifle denotation with the ref.8nhg exppssion Instead the focus is the sef o O whkh thrq mfaent of that phrase is a member. ttot refer to the meeting mentioned In the prevkus senPew, but that msetihg can be used aIf one is known in the database, eks fi is a description of the entity required, es with inferred refstenees and set -element referenoel, Severd moditbre -",
	"num": null
	}
	}
	}
	}