Download presentation
Presentation is loading. Please wait.
Published byBlake McCarthy Modified over 9 years ago
1
Comparing approaches Part XVIII: Approaches to dialogue Peter Kühnlein
2
Comparing approaches
3
Parameters for theories Levin & Moore (1977) Mann (1988) Mann (2002) Garrod & Pickering (2003) Grosz & Sidner (1986) Bratman (1999) Clark (1996) A number of theories have been discussed Comparing approaches
4
Dialogue games - intuition Established patterns represented as sets of knowledge structures are dialogue-games called dialogue-games (DGs). They capture shared conventional knowledge about communication and the use of communication to achieve goals Comparing approaches
5
Parameters for theories Levin & Moore (1977)Pattern Mann (1988) Mann (2002) Garrod & Pickering (2003) Grosz & Sidner (1986) Bratman (1999) Clark (1996) A number of theories have been discussed Comparing approaches
6
Dialogue and Alignment Alignment and routinization A „routine“ is an expression that is „fixed“ to some extent it has a higher frequency than the frequency of its component words would lead to expect it has a particular analysis at each level of linguistic representation highly frequent in dialogue „Routinization“ establishes routines on the fly: if an interlocutor uses an expression in a particular way, it may become a routine for the purpose of the conversation Comparing approaches
7
Parameters for theories Levin & Moore (1977)Pattern Mann (1988) Mann (2002) Garrod & Pickering (2003)Routine Grosz & Sidner (1986) Bratman (1999) Clark (1996) A number of theories have been discussed Comparing approaches
8
( OCP) ( f ) ( c): (a) ICP initiated DS intending 1. OCP to think DS possesses f 2. OCP to think f correlated in way c with ψ-ing that p 3. OCP to think, on the basis of the fulfillment of 1. and [2.] that ICP intends OCP to think that ICP ψs that p 4. OCP, on the basis of the fulfillment of 3. to think that ICP ψs that p and (for some cases) 5. OCP, on the basis of the fulfillment of 4., himself to ψ that p By initiating DS ICP meant that *ψp is true iff ψ in the above formulae names an appropriate propositional attitude, e.g., belief. * is a „mood marker“. Comparing approaches
9
Parameters for theories Levin & Moore (1977)Pattern Mann (1988) Mann (2002) Garrod & Pickering (2003)Routine Grosz & Sidner (1986)Properties of utterances Bratman (1999) Clark (1996) A number of theories have been discussed Comparing approaches
10
Conventional language coordination problem C 13 Signals are deliberate actions. Some are performed as parts of conventional languages like English, Dakota, Japanese, or American Sign Language… C 70,71 A convention, according to Lewis (1969), is a community's solution to a recurrent coordination problem. … What makes something a convention? According to Lewis, it has these five properties: A convention is: 1.a regularity r in behaviour 2.partly arbitrary 3.that is common ground in a given community C 4.as a coordination device 5.for a recurrent coordination problem s. Comparing approaches
11
Parameters for theories Levin & Moore (1977)Pattern Mann (1988) Mann (2002) Garrod & Pickering (2003)Routine Grosz & Sidner (1986)Properties of utterances Bratman (1999) Clark (1996)Convention A number of theories have been discussed Comparing approaches
12
Core Intuitions Aims & State-of-the-Art Dialogue is the most basic and natural form of language use Hence, psycholinguistics should provide an account of the basic language processing mechanisms in dialogue Thesis: Dialogue is coordinated behaviour in that the representations that underly discourse become aligned. Comparing approaches
13
Parameters for theories Levin & Moore (1977)Pattern Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentation Grosz & Sidner (1986)Properties of utterances Bratman (1999) Clark (1996)Convention A number of theories have been discussed Comparing approaches
14
The formal notion of a dialogue game basic assumptions - natural dialogue has a recognizable „episodic“ structure and coherence which conform to a set of conventions - virtually all communication is characterizable as goal pursuit activity (broader theoretical perspective) - a dialogue reflects intentions and goal pursuit activity on the part of each participant - goals are consequential over the period of pursuit - goals are regarded as having scopes: intervals over which they prevail - speaker exibits goal pursuit and provides suitable information - hearer has tacit knowledge of possible goal sets - hearer´s knowledge includes relationships between goals (subordination) Comparing approaches
15
Parameters for theories Levin & Moore (1977)PatternBelief Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentation Grosz & Sidner (1986)Properties of utterances Bratman (1999) Clark (1996)Convention A number of theories have been discussed Comparing approaches
16
Conventional language common ground shared basis C 93 The technical notion of common ground was introduced by Robert Stalnaker (1978; cf. Karttunen and Peters, 1975) … Two people's common ground is, in effect, the sum of their mutual, common, or joint knowledge, beliefs, and suppositions. C 94 Common ground (shared basis) p is common ground for members of community C if and only if: 1.every member of C has information that basis b holds; 2.b indicates to every member of C that every member of C has information that b holds; 3.b indicates to members of C that p. Comparing approaches
17
Parameters for theories Levin & Moore (1977)PatternBelief Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentation Grosz & Sidner (1986)Properties of utterances Bratman (1999) Clark (1996)ConventionBel. A number of theories have been discussed Comparing approaches
18
Discourse structure a structure of intentions G&S 178 Typically, an ICP will have a number of different kinds of intentions that lead to initiating the discourse. … The kinds of intentions that can serve as discourse purposes or discourse segment purposes are distinguished from other intentions by the fact that they are intended to be recognized…. Discourse purposes and discourse segment purposes share this property with certain utterance-level intentions that Grice (1969) uses in defining utterance meaning. Comparing approaches
19
Parameters for theories Levin & Moore (1977)PatternBelief Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentation Grosz & Sidner (1986)PropertiesIntention of utterances Bratman (1999) Clark (1996)ConventionBel. A number of theories have been discussed Comparing approaches
20
A major goal of [DMT] is to provide a descriptive account for the coherence of a wide diversity of natural dialogues. A dialogue is said to be coherent if a person who has good access to the dialogue is left with the impression that every part of the dialogue contributed to the remainder, or equivalently that there are no parts whose presence is not easily explained. M02, 3-4: Dialogue coherence arises from the intentions (also called goals) of the dialogue participants. Comparing approaches
21
Parameters for theories Levin & Moore (1977)PatternBeliefIntention Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentation Grosz & Sidner (1986)PropertiesIntention of utterances Bratman (1999) Clark (1996)ConventionBel. A number of theories have been discussed Comparing approaches
22
Core Intuitions Aims & State-of-the-Art Dialogue is the most basic and natural form of language use Hence, psycholinguistics should provide an account of the basic language processing mechanisms in dialogue Thesis: Dialogue is coordinated behaviour in that the representations that underly discourse become aligned The linguistic representations employed by the interlocutors become aligned at many levels of representation Alignment is the result of a largely automatic process greatly simplifies production and comprehension Comparing approaches
23
Parameters for theories Levin & Moore (1977)PatternBeliefIntention Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentationProcess Grosz & Sidner (1986)PropertiesIntention of utterances Bratman (1999) Clark (1996)ConventionBel. A number of theories have been discussed Comparing approaches
24
The notion of Intention Commitment When you intend to perform a certain action φ, you are committed to perform φ. This commitment is at work even in cases where you intend to φ at a considerably later time, say after one year. The fact to note is to intend something means that your choices have to be compatible with that intention. You can’t consistently intend to φ and perform actions that will prevent you from φing. So your intentions will be interdependent in a larger framework. Call the larger frameworks plans. Comparing approaches
25
Parameters for theories Levin & Moore (1977)PatternBeliefIntention Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentationProcess Grosz & Sidner (1986)PropertiesIntention of utterances Bratman (1999)Intention as part of plan Clark (1996)ConventionBel. A number of theories have been discussed Comparing approaches
26
Joint activities joint actions joint construal C 192 Uptake […] presupposes understanding. Although [participant A might reach] a construal of [participant B's] utterance, is it the one [B] intended – is it the one [B] will accept? I will call this the joint construal problem. … There is a tight link between the way two people settle on a joint construal of a signal (level 3) and the way they propose and take up joint projects (level 4). Comparing approaches
27
Parameters for theories Levin & Moore (1977)PatternBeliefIntention Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentationProcess Grosz & Sidner (1986)PropertiesIntention of utterances Bratman (1999)Intention as part of plan Clark (1996)ConventionBel.Construal A number of theories have been discussed Comparing approaches
28
Conventional language C 13 Signals are deliberate actions. Some are performed as parts of conventional languages like English, Dakota, Japanese, or American Sign Language… C 75 Languages like English are conventional signaling systems par excellence. Most English speakers, for example, have contingency plans that include this pairing of conditionals, which I will call a signaling doublet: Speaker: If you intend to denote the cipher naught, you can utter the word zero. Addressee: If a speaker utters the word zero, he or she can be denoting the cipher naught. This doublet happens to be conventional. … It is a coordination device for a recurrent coordination problem – speakers wanting to denote naught and their addressees wanting to recognize this. Comparing approaches
29
Parameters for theories Levin & Moore (1977)PatternBeliefIntention Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentationProcess Grosz & Sidner (1986)PropertiesIntention of utterances Bratman (1999)Intention as part of plan Clark (1996)ConventionBel.ConstrualCoordination A number of theories have been discussed Comparing approaches
30
A major goal of [DMT] is to provide a descriptive account for the coherence of a wide diversity of natural dialogues. A dialogue is said to be coherent if a person who has good access to the dialogue is left with the impression that every part of the dialogue contributed to the remainder, or equivalently that there are no parts whose presence is not easily explained. M02, 3-4: Dialogue coherence arises from the intentions (also called goals) of the dialogue participants. It arises especially from the way that the conventions of dialogue cause the participants to adopt and dismiss groups of intentions. Grouping of intentions is the foundation for coordination of the activities of dialogue participants. Comparing approaches
31
Parameters for theories Levin & Moore (1977)PatternBeliefIntentionPlaying Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentationProcess Grosz & Sidner (1986)PropertiesIntention of utterances Bratman (1999)Intention as part of plan Clark (1996)ConventionBel.ConstrualCoordination A number of theories have been discussed Comparing approaches
32
Core Intuitions Aims & State-of-the-Art Dialogue is the most basic and natural form of language use Hence, psycholinguistics should provide an account of the basic language processing mechanisms in dialogue Thesis: Dialogue is coordinated behaviour in that the representations that underly discourse become aligned Alignment differs from the classical (Lewis, Clark) kind of coordination in that it is a psychological mechanism, not a strategy in behaviour Comparing approaches
33
Parameters for theories Levin & Moore (1977)PatternBeliefIntentionPlaying Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentationProcessAlignment Grosz & Sidner (1986)PropertiesIntention of utterances Bratman (1999)Intention as part of plan Clark (1996)ConventionBel.ConstrualCoordination A number of theories have been discussed Comparing approaches
34
G&S 199 Discourse structure a structure of intentions The intentions that serve as DP/DSPs are natural extensions of the intentions that Grice (1969) considers essential to developing a theory of utterer's meaning. There is a crucial difference, however, between our use of discourse- level intentions in this paper (and the theory, as developed so far) and Grice's use of utterance-level intentions. We are not yet addressing the issue of discourse meaning, but are concerned with the role of DP/DSPs in determining discourse structure and in specifying how these intentions can be recognized by an OCP. Comparing approaches
35
Parameters for theories Levin & Moore (1977)PatternBeliefIntentionPlaying Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentationProcessAlignment Grosz & Sidner (1986)PropertiesIntentionRecognition of utterances Bratman (1999)Intention as part of plan Clark (1996)ConventionBel.ConstrualCoordination A number of theories have been discussed Comparing approaches
36
Plans are hierarchically organized. And sometimes plans comprise joint actions by a number of people. Examples are: Two people dancing together. (This is a good example with all the features.) These two should be said to jointly perform some action. Given that their joint action is constituted by individual actions, it is nevertheless an issue to explain the notion of intending a joint action. V4: We intend to φ iff 1.I intend that we φ and you intend that we φ, 2. each of us intends to φ because of 1 and meshing subplans of 1, and 3.1 is common knowledge among us. This suggestion doesn’t settle the question whether groups can jointly intend something: There are two readings of “we intend” – one distributive, and one collective reading. The notion of Intention Jointly intending to act Comparing approaches
37
Parameters for theories Levin & Moore (1977)PatternBeliefIntentionPlaying Mann (1988) Mann (2002) Garrod & Pickering (2003)RoutineRepresentationProcessAlignment Grosz & Sidner (1986)PropertiesIntentionRecognition of utterances Bratman (1999)Intention asShared part of planIntentions Clark (1996)ConventionBel.ConstrualCoordination A number of theories have been discussed Comparing approaches
38
Coverage wrt the benchmark dialogue Comparing approaches
39
(A) Inst:So, jetzt nimmst du Well, now you take Cnst:eine Schraube a screw. Inst:eine orangene mit einem Schlitz. an orange one with a slit Cnst:Ja. Yes Coverage wrt the benchmark dialogue Comparing approaches
40
(A) Inst:So, jetzt nimmst du Well, now you take Cnst:eine Schraube a screw. Inst:eine orangene mit einem Schlitz. an orange one with a slit Cnst:Ja. Yes Available Bolts Coverage wrt the benchmark dialogue Comparing approaches
41
(B) Inst: Und steckst sie dadurch, also And you put it through there, let’s see Cnst: Von oben. From the top. Inst: Von oben, daß also die drei festgeschraubt werden dann. From the top, so that the three bars get fixed. Cnst: Ja. Yes. Intended Junction Intended Result Coverage wrt the benchmark dialogue Comparing approaches
42
(A) Inst: Well, now you take Cnst: a screw. Inst:an orange one with a slit Cnst: Yes. (B) Inst: And you put it through there, let’s see Cnst: From the top. Inst: From the top, so that the three bars get fixed. Cnst:Yes. Coverage wrt the benchmark dialogue Comparing approaches
43
Coverage wrt the benchmark dialogue Levin & Moore (1977) Mann (1988) Mann (2002) Garrod & Pickering (2003) Grosz & Sidner (1986) Bratman (1999) Clark (1996) And here are the scores Comparing approaches
44
Coverage wrt the benchmark dialogue Levin & Moore (1977)Idealizations required; Mann (1988) Mann (2002) Garrod & Pickering (2003) Grosz & Sidner (1986) Bratman (1999) Clark (1996) And here are the scores Comparing approaches
45
Coverage wrt the benchmark dialogue Levin & Moore (1977)Idealizations required; Mann (1988)relation between segments unclear; Mann (2002) Garrod & Pickering (2003) Grosz & Sidner (1986) Bratman (1999) Clark (1996) And here are the scores Comparing approaches
46
Coverage wrt the benchmark dialogue Levin & Moore (1977)Idealizations required; Mann (1988)relation between segments unclear; result not plausible Mann (2002) Garrod & Pickering (2003) Grosz & Sidner (1986) Bratman (1999) Clark (1996) And here are the scores Comparing approaches
47
Coverage wrt the benchmark dialogue Levin & Moore (1977)Idealizations required; (directive/action pair not covered;) Mann (1988)relation between segments unclear; result not plausible Mann (2002)(covers directive/action pair) Garrod & Pickering (2003) Grosz & Sidner (1986) Bratman (1999) Clark (1996) And here are the scores Comparing approaches
48
Coverage wrt the benchmark dialogue Levin & Moore (1977)Idealizations required; (directive/action pair not covered;) Mann (1988)relation between segments unclear; result not plausible Mann (2002)(covers directive/action pair) Garrod & Pickering (2003)(too early to decide) Grosz & Sidner (1986) Bratman (1999) Clark (1996) And here are the scores Comparing approaches
49
Coverage wrt the benchmark dialogue Levin & Moore (1977)Idealizations required; (directive/action pair not covered;) Mann (1988)relation between segments unclear; result not plausible Mann (2002)(covers directive/action pair) Garrod & Pickering (2003)(too early to decide) Grosz & Sidner (1986)Idealizations required; stack regime must be changed Bratman (1999) Clark (1996) And here are the scores Comparing approaches
50
Coverage wrt the benchmark dialogue Levin & Moore (1977)Idealizations required; (directive/action pair not covered;) Mann (1988)relation between segments unclear; result not plausible Mann (2002)(covers directive/action pair) Garrod & Pickering (2003)(too early to decide) Grosz & Sidner (1986)Idealizations required; stack regime must be changed Bratman (1999) Clark (1996) And here are the scores Comparing approaches
51
Coverage wrt the benchmark dialogue Levin & Moore (1977)Idealizations required; (directive/action pair not covered;) Mann (1988)relation between segments unclear; result not plausible Mann (2002)(covers directive/action pair) Garrod & Pickering (2003)(too early to decide) Grosz & Sidner (1986)Idealizations required; stack regime must be changed Bratman (1999) Clark (1996)Idealizations required And here are the scores Comparing approaches
Similar presentations
© 2025 SlidePlayer.com Inc.
All rights reserved.