1. Att ventilera en (pragmatik-) uppsats. Pragmatik VT06. 2. Ventilering (20-30
minuter). 1. Respondenten (författaren) berättar om sistaminuten-ändringar. 2.
Ventilering (20-30 minuter)
Att ventilera en (pragmatik-) uppsats
1. Respondenten (författaren) berättar om sistaminuten-ändringar. 2. Opponenten presenterar uppsatsen. 3. Respondenten kompletterar presentationen. 4. Opponenten ställer kritiska frågor, som respondenten besvarar efter bästa förmåga. 5. Auditoriet (åhörarna) ställer frågor.
Pragmatik VT06
1
Presentation av uppsatsen ! !
! ! !
2
Att läsa en uppsats kritiskt
Lyft fram huvuddragen Presentera så att någon som inte läst uppsatsen kan förstå vad den handlar om Om lämpligt, visa exempel för att öka åhörarnas förståelse Använd oh, ppt, skriv på tavlan.. Ha inte med någon värdering här, dvs säg inte vad du tycker är bra eller dåligt
!
Exempel på frågor att titta på: "
"
"
"
Knyter avslutningen an till inledningen? Dvs, gör författaren det hon säger sig göra? Vilken metod använder författaren? Är den lämplig? Ger uppsatsen de resultat som författaren påstår? Saknas något? Är något onödigt?
3
Att ställa kritiska frågor till respondenten !
! ! !
Praktiskt
Koncentrera dig på innehållet, inte formen "
4
!
Dvs ta inte upp stavfel och liknande
Utgå ifrån de saker du fann under din kritiska läsning av uppsatsen Ställ gärna frågor om sådant du tyckte var oklart Ha gärna med positiva saker med uppsatsen
!
5
Före 23 mars kommer opponenter och respondenter att paras ihop, och ett schema kommer läggas ut Opponenten behöver få uppsatsen några dagar före ventileringen
6
A pragmatic issue
Centering Theory !
Pragmatik VT06
Just how are pronouns and descriptive NPs interpreted (resolved) in a discourse?
7
8
Centering theory
An example (1) Jane likes Mary. (2) She often brings her flowers. (3) She chats with the young woman for ages
!
Centering a key element of local discourse coherence "
A system of rules and constraints that govern: !
! !
the relationship between what the discourse is about and some of the linguistic choices made by discourse participants choice of syntactic structure type of referring expression (proper noun, definite or indefinite description, reflexive or personal pronoun, etc)
(Brennan, Friedman & Pollard 1987) 9
Centering theory !
!
10
Main themes (1) !
Attempts to characterise the texts that can be considered coherent on the basis of the way discourse entities are introduced and discussed Attempts to predict which entities will be most salient at any given time
Discourse is viewed dynamically "
!
A sentence/utterance is a transition from an input state to an output state
The state " " "
determines which entities are under discussion: the centers of attention represents the utterance’s anaphoric potential captures the relative salience of various discourse entities
(Poesio et al 2000)
11
12
The Centering model
Main themes (2) !
The transitions (between states) are classified according to amount of change involved " "
Transitions involving only little change: coherent discourse Transitions involving much change: incoherent discourse
13
14
Backward-looking center (CB) !
Forward-looking center (CF)
A state comprises a backward-looking center and a forward-looking center list
A forward-looking center list (CF)
! "
!
A backward-looking center (CB)
A list of discourse referents
CF(Un): a list of all the referents in utterance n
! "
A discourse referent A link to the previous sentence ! The most significant entity under discussion in both the current and previous sentence ! CB(Un): the bw-looking center of the nth utterance !
Ordered according to salience (= grammatical obliqueness) 1.
!
2. 3. 4.
The subject: first element, most salient (least oblique) • Is also the preferred center (PF) Direct object Indirect object Adjuncts: last elements, least salient (most oblique)
15
The algorithm (2)
The algorithm (1) 1. CONSTRUCT "
"
"
2. FILTER
The alternative possibilities for anaphoric resolution are identified Each possibility maps pronouns in the sentence to discoure entities, respecting agreement features For each possibility: ! CF(Un) consists of all the referents of NPs in the sentence !
16
Possibilities are discarded unless all of the following criteria are met: 1.
2.
CB(Un) is chosen from CF(Un-1) or NIL 3. 17
If there are pronouns in the current sentence, then one of them is the bw-looking center of the current sentence (the “Rule 1”) The bw-loooking center is mapped onto the entity mentioned in the current sentence which is highest ranked in the previous sentence’s fwdlooking center Syntactic coreference constraints are upheld 18
The algorithm (4)
The algorithm (3)
3. CLASSIFY
4. SELECT
Classify each possibility as one of the transition types: " " " "
Chose the best possibility, using the ranking over transition types: continue > retain > smooth shift > rough shift
continue retain smooth shift rough shift
“>” = “is preferred over” 19
20
Transition types (1) !
Transition types (2)
State transitions are classified according to amount of change involved:
"
Retain:
Bw-looking center is unchanged, i.e CB(Un-1) = CB(Un) ! Bw-looking center is no longer in preferred position, i.e CB(Un) ! CP(Un) !
"
Continue: ! Bw-looking center is unchanged, !
i.e CB(Un-1) = CB(Un) Bw-looking center is also preferred center in the new utterance, i.e CB(Un) = CP(Un)
21
Transition types (3)
"
22
Transition types (3)
Smooth shift:
New bw-looking center is different from old, i.e CB(Un-1) ! CB(Un) ! Bw-looking center is the same as the preferred center, i.e CB(Un) = CP(Un)
"
!
Rough shift: ! New bw-looking center is different from old, !
23
i.e CB(Un-1) ! CB(Un) Bw-looking center is no longer in preferred position, i.e CB(Un) ! CP(Un)
24
Sentence (1): Jane likes Mary.
Consider the following discourse again: (1) Jane likes Mary. (2) She often brings her flowers. (3) She chats with the young woman for ages Question: What do the pronouns and the definite description (underlined) refer to?
1. Construct: CF(U1): CB(U1): NIL CP(U1): Jane 2. Filter: x 3. Classify: 4. Select: -
25
26
Sentence (2): She often brings her flowers.
Sentence (2): She often brings her flowers.
1. Construct: CF(U2): or or or CB(U2): Jane or Mary CP(U2): Jane or Mary
2. Filter: By (1): “she” or “her” refers to CB By (2): CB(U2) is Jane By (3): & are ruled out
27
Sentence (2): She often brings her flowers.
28
Sentence (2): She often brings her flowers.
3. Classify: (a) CF(U2): CB(U2): Jane CP(U2): Jane So, CB(U2) ! CB(U1), CB(U2) = CP(U2) i.e smooth shift (b) CF(U2): CB(U2): Jane CP(U2): Mary So, CB(U2) ! CB(U1), CB(U2) ! CP(U2) i.e rough shift
4. Select: smooth shift
29
30
Sentence (3): She chats with the young woman for ages
So CT predicts: (1) Jane likes Mary (2) She often brings her flowers = Jane often brings Mary flowers
1. Construct: CF(U3): or or or CB(U3): Jane or Mary or flowers or NIL CP(U3): Jane or Mary
31
Sentence (3):
She chats with the young woman for ages
2. Filter: By (1): CB(U3) is Jane or Mary By (2): CB(U3) is Jane By (3): & are ruled out
33
Sentence (3):
She chats with the young woman for ages
32
Sentence (3):
She chats with the young woman for ages
3. Classify: (a) CF(U3): CB(U3): Jane CP(U3): Jane So, CB(U3) = CB(U2), CB(U3) = CP(U3) i.e continue (b) CF(U3): CB(U3): Jane CP(U3): Mary So, CB(U3) = CB(U2), CB(U3) ! CP(U3) i.e retain
34
So: (1) Jane likes Mary (2) She often brings her flowers = Jane often brings Mary flowers (3) She chats with the young woman for ages = Jane chats with Mary for ages
4. Select: continue
35
36
Poesio et al: Specifying the Parameters of Centering Theory: a Corpus-Based Evaluation using Text from Application-Oriented Domains
Poesio et al ! !
CT is a parametric theory There are different versions of CT "
!
Examples of parameters & values: " "
!
They give different values to different parameters Should “utterance” be a full sentence, or a clause, or..? What counts as the realisation of a reference? Only direct (Jane.. She..) or also via bridging (The painting.. The frame..)?
And how does it work for different languages?
37
Poesio et al !
!
Poesio et al
First goal: "
38
!
Which values for the parameters would make the claims of CT most accurate?
"
Second goal: "
First goal involves automatic check of two central CT claims:
"
Evaluate CT predictions in domains of interest for real applications: corpora of instructional texts and descriptions of museum objects
!
Constraint 1: All utterances of a segment except for the first have exactly one CB Rule 1: If any discourse referent is pronominalised, the CB is
Another claim that can be checked: "
Rule 2: continue > retain > shift
39
Literature !
"
!
First 10 pages are a good introduction to Centering Theory http://www.stanford.edu/~dib/
! !
Se schemat Inlämning onsdag 8 mars
Massimo Poesio et al. 2000. Specifying the Parameters of Centering Theory: a Corpus-Based Evaluation using Text from Application-Oriented Domains. Proceedings of the 38th ACL. "
!
Övning för Centering Theory
David Beaver. 2004. The Optimization of Discourse Anaphora. Linguistics and Philosophy 27(1), pp. 3-56. "
40
http://cswww.essex.ac.uk/staff/poesio/
S E Brennan, M W Friedman & C J Pollard. 1987. A Centering Approach to Pronouns . Proceedings of the 25th ACL. "
http://acl.ldc.upenn.edu/P/P87/P87-1022.pdf 41
42