Repairing Reference Identification Failures by Relaxation

REPAIRING REFERENCE IDENTIFICATION FAILURES BY RELAXATION

B r a d l e y A. G o o d m a n B B N Laboratories I0 M o u l t o n Street C a m b r i d g e . Mass. 0 2 2 3 8

ABSTRACT

T h e goal of thls w o r k is the e n r i c h m e n t of h u m a n - m a c h l n e m t e r a c t I o n s in a natural l a n g u a g e envlronment. 1 W e w a n t to provide a f r a m e w o r k less restrictive t h a n earlier o n e s b y allowing a s p e a k e r leeway tn forming a n u t t e r a n c e a b o u t a task a n d in determining the conversational vehicle to deliver it, A s p e a k e r a n d listener c a n n o t be a s s u r e d to h a v e the s a m e beliefs, contexts, b a c k g r o u n d s or goals at e a c h point in a conversation. As a result, dlfflcultles a n d mistakes arise w h e n a listener interprets a s p e a k e r s utterance. T h e s e mistakes c a n lead to various kinds of m l s u n d e r s t a n d l n g s b e t w e e n s p e a k e r a n d hstener. including reference failures or failure to u n d e r s t a n d the speaker's mtentlon. W e c a l l these m t s u n d e r s t a n d l n g s m l s c o m m u n m a t l o n S u c h m~stakes constitute a klnd of "ill-formed" input that c a n slow d o w n e n d possibly b r e a k d o w n c o m m u n i c a t i o n . O u r goal ~s to recognize a n d Isolate s u c h m l s c o m m u n l c a t t o n s a n d c i r c u m v e n t them. Thls p a p e r will h l g h h g h t a particular class of m l s c o m m u n l c a t l o n - reference p r o b l e m s - by d e s c r l b m g a case study, includlng t e c h n i q u e s for avoldlng failures of r e f e r e n c e

I Introduction

Cohen, Perrault a n d Allen s h o w e d in thelr p a p e r " B e y o n d Question A n s w e r i n g " [8~ that ",.. users of cluestlon-answerzng s y s t e m s expect t h e m to do m o r e t h a n just a n s w e r isolated questions - - they expect s y s t e m s to e n g a g e tn conversation. In doing ~o, the s y s t e m ts e x p e c t e d to allow users to be less than meticulously hteral ~n c o n v e y i n g their zntentlons, a n d tt is e x p e c t e d to m a k e hnguxstlc a n d p r a g m a t i c use of the previous discourse." Following in thelr footsteps, w e w a n t to build robust natural l a n g u a g e processing s y s t e m s that c a n detect a n d recover from m l s c ~ m m u n l c a t t o n . T h e d e v e l o p m e n t of s u c h systems requires s study o n h o w people c o m m u n i c a t e a n d h o w they recover from p r o b l e m s In c o m m u n i c a t i o n . This p a p e r s u m m a r i z e s the results of a dissertation [13] that tnvestlgates the kinds of m l s c o m m u n l c a t l o n that o c c u r in h u m a n c o m m u n i c a t i o n with a special e m p h a s i s o n reference prooiems, i.e.. p r o b l e m s a listener has determining w h o m or w h a t a s p e a k e r ts talking about. W e h a v e written c o m p u t e r p r o g r a m s a n d algorithms that d e m o n s t r a t e h ~ w o n e could h a n d l e s u c h p r o b l e m s m

I T h i s reseorcn was suDDorted in p o r t by the Oefenee

Advonce4 Reseorch P r o ~ e c t Aqency under ¢ o n t r = c t Neee14--77--

C - ~ 3 7 8 .

the c o n t e x t of a natural l a n g u a g e u n d e r s t a n d 2 n g system. T h e s t u d y of m z s c o m m u n l c a t l o n is a n e c e s s a r y task w l t h m s u c h a c o n t e x t since a n y c o m p u t e r capable of c o m m u n l c a t ~ n g with h u m a n s tn natural l a n g u a g e m u s t be tolerant of the tmprecIse, lll-devlsed or c o m p l e x u t t e r a n c e s that people often use.

O u r c u r r e n t r e s e a r c h [25, 26] views m o s t dialogues as being cooperatlve a n d goal directed, l,e.. a s p e a k e r a n d h s t e n e r w o r k together to achieve a c o m m o n goal. T h e interpretation of a n u t t e r a n c e involves Identifying the underlying plan or goal that the u t t e r a n c e reflects [5. I, 2 3 ] . Thls plan, however, is rarely, d ever, obvious at the surface s e n t e n c e level. A central issue In the interpretation of u t t e r a n c e s ts the t r a n s f o r m a t i o n of s e q u e n c e s of imprecise, zll- devised or c o m p l e x u t t e r a n c e s into w e l l - s p e c l h e d plans that might b e carried out b y dialogue participants. Within thls context, m l s c o m m u n l c a t l o n c a n occur.

W e ere particularly c o n c e r n e d with cases of m x s c o m m u n l c a t l o n from the heater's viewpoint, s u c h as w h e n the h e a r e r is m a t t e n t l v e to. c o n f u s e d about, or misled a b o u t the zntentlons of the speaker. In ordinary e x c h a n g e s s p e a k e r s usually m a k e a s s u m p t i o n s regarding w h a t thelr listeners k n o w a b o u t a topic of discussion. T h e y w111 leave out details t h o u g h t to be superfluous [2. 19]. Since the s p e a k e r really does not k n o w exactly w h a t a listener k n o w s a b o u t a topic, tt ts easy to m a k e s t a t e m e n t s that c a n b e misinterpreted or not u n d e r s t o o d b y the listener b e c a u s e not e n o u g h details w e r e presented. O n e principal s o u r c e of trouble Is the description c o n s t r u c t e d b y the s p e a k e r to refer to a n actual object in the world. T h e d e s c m p t l o n c a n be tmpreclse, confused, a m b i g u o u s or over!v speclflc. It might be interpreted u n d e r the w r o n g context. This leads to dlfflculty for the h s t e n e r w h e n figuring out w h a t oblect ~s being described, that Is. ref.erence identification errors. S u c h descriptions are "all- f o r m e d " input, the b l a m e for ill-formedness m a y lie partly with the s p e a k e r a n d partly with the listener T h e s p e a k e r m a y h a v e b e e n sloppy or not t a k e n the h e a r e r into consideration, the listener m a y be either remiss or unwilling to admit h e can't u n d e r s t a n d the s p e a k e r a n d to ask the s p e a k e r for clarification, or m a y slmply feel that he h a s u n d e r s t o o d w h e n he zn fact has not.

Thls w o r k ts part of a n o n - g o i n g e f f o r t to develop a reference Identlfzcatmn a n d plan recognition m e c h a n i s m that c a n exhibit m o r e " h u m a n - h k e ' tolerance of s u c h utterances. O u r goal zs to build a m o r e robust s y s t e m that c a n handle errorful utterances, a n d ~hat c a n be i n c o r p o r a t e d in exlstlng systems. As a start, w e h a v e c o n c e n t r a t e d o n

r e f e r e n c e tdentlflcatzon. In c o n v e r s a t i o n people use

imperfect descriptions to c o m m u n i c a t e a b o u t objects; s o m e t i m e s their p a r t n e r s s u c c e e d zn u n d e r s t a n d i n g a n d occasionally they fail. A n y c o m p u t e r h o p i n g to play the part of a listener m u s t b e c a p a b l e of taking w h a t the

s p e a k e r says a n d either deleting, adapting or clarifying it. W e are developing a t h e o r y of the use of extensional descrlptlons that will help e x p l a m h o w people successfully use s u c h imperfect descriptions. W e call thls the theory of reference m l s c o m m u n l c a t i o n

Section 2 of this p a p e r highlights s o m e aspects of n o r m a l c o m m u n i c a t i o n a n d t h e n provides a general discussion o n the types of m i s c o m m u n l c a t l o n that o c c u r In conversation, concentrating primarily o n reference p r o b l e m s a n d motivating m a n y of t h e m with Illustrative protocols. Section 3 presents possible w a y s a r o u n d s o m e of the p r o b l e m s of m i s c o m m u n x c a t i o n in reference. Motivated there is a partial Implementation of a reference m e c h a n i s m that attempts to o v e r c o m e m a n y reference problems.

[image:2.612.315.552.43.717.2]

W e are following the t a s k - o m e n t e d p a r a d i g m of G r o s z [ 1 4 ] s i n c e i t Is e a s y t o s t u d y ( t h r o u g h v i d e o t a p e s ) . It p l a c e s t h e w o r l d In f r o n t o f y o u ( a p r i m a r i l y e x t e n s i o n a l w o r l d ) , a n d It l i m i t s t h e d l s c u s s l o n w h l l e s t i l l p r o v i d i n g a r l c h e n v i r o n m e n t f o r c o m p l e x d e s c r i p t i o n s . T h e t a s k c h o s e n a s t h e t a r g e t f o r t h e s y s t e m Is t h e a s s e m b l y o f a t o y w a t e r p u m p . T h e w a t e r p u m p Is r e a s o n a b l y c o m p l e x , c o n t a i n i n g f o u r s u b a s s e m b l i e s t h a t a r e b u i l t f r o m p l a s t i c t u b e s , n o z z l e s , v a l v e s , p l u n g e r s , a n d c a p s t h a t c a n b e s c r e w e d o r p u s h e d t o g e t h e r . A l a r g e c o r p u s o f d i a l o g u e s c o n c e r n i n g t h l s t a s k w a s c o l l e c t e d b y C o h e n ( s e e

[7. 8. 9 ] ) . T h e s e d i a l o g u e s c o n t a i n e d i n s t r u c t i o n s f r o m a n " e x p e r t " t o a n " a p p r e n t i c e " t h a t e x p l a i n t h e assembly of the toy water p u m p , Both participants w e r e w o r k i n g t o a c h i e v e a c o m m o n g o a l - t h e s u c c e s s f u l a s s e m b l y of t h e p u m p T h l s d o m a i n Is r l c h m p e r c e p t u a l i n f o r m a t i o n , a l l o w i n g f o r c o m p l e x d e s c r i p t i o n s of e l e m e n t s i n it. T h e d a t a p r o v i d e e x a m p l e s of imprecision, confusion, a n d ambiguity as we!l as attempts to correct these p r o b l e m s

T h e following e x c h a n g e exemplifies o n e s u c h situation. Here A Is instructing J to a s s e m b l e part o f t h e w a t e r pump. R e f e r t o F i g u r e l ( a ) f o r a p i c t u r e of t h e p u m p . A a n d J a r e c o m m u n i c a t i n g v e r b a l l y b u t n e i t h e r c a n s e e t h e o t h e r . ( T h e b r a c k e t e d t e x t In t h e e x c e r p t t e l l s w h a t w a s a c t u a l l y o c c u r r i n g w h i l e e a c h u t t e r a n c e w a s s p o k e n . ) N o t l c e t h e c o m p l e x i t y o f t h e s p e a k e r ' s d e s c r i p t i o n s a n d t h e r e s u l t a n t p r o c e s s i n g r e q u i r e d b y t h e l i s t e n e r , T h l s d i a l o g u e i l l u s t r a t e s w h e n l i s t e n e r s r e p a i r t h e s p e a k e r s d e s c r i p t i o n in o r d e r t o f l n d a r e f e r e n t , w h e n t h e y r e p a i r t h e i r m z t l a l r e f e r e n c e c h o i c e o n c e t h e y a r e g i v e n m o r e i n f o r m a t i o n , a n d w h e n t h e y f a l l t ~. c h o o s e a p r o p e r r e f e r e n t In Linp 7, A :[,=scribes the two holes on the BAjEVALVE as "the h t t l e h o i e " J m u s t r e p a i r t h e d e s c r l p t l o n , r e a h z m g that A d o e s n t r e a l l y m e a n " o n e h o l e b u t is r e f e r r i n g t o t,~e ' t w o ' h o l e s . J a p p a r e n t l y d o e s t h i s s i n c e h e d o e s n t c o m p l a i n a b o u t a s d e s c r i p t i o n a n d c o r r e c t l y a t t a c h e s the B A S E V A L V E to the T U B E B A S E F i g u r e l i b ) s h o w s the c o n f i g u r a t i o n of the p u m p a f t e r the T U B E B A S E is a t t a c h e d t o t h e M A I N T U B E "n L m e I 0 , [n L m e 13. J i n t e r p r e t s " a r e d p l a s t i c p i e c e " t o r e f e r t o t h e . V O Z Z L E W h e n A a d d s t h e r e l a t i v e c l a u s e " t h a t h a s f o u r g i = m o s on it." J is forced to drop the N O Z Z L E as the r e f e r e n t a n d to se{ect the S L L D E V ~ L V E In L m e s i7 a n d 18, A'S description "the o t h e r - - t h e o p e n part o f the maln tube. the lower valve" is ambiguous, a n d J selects the w r o n g s l t e , n a m e l y t h e TUBEBAEE, in w h i c h t o i n s e r t

the SLIDEVALVE. Since the SL/DEVALVE flts, J doesn't

d e t e c t a n y trouble. L~nes 20 a n d 21 k e e p I f r o m t h i n k i n g t h a t s o m e t h i n g is w r o n g b e c a u s e t h e p a r t f i t s l o o s e l y , In L~nes 27 a n d 2 8 , J i n d i c a t e s t h a t A d l d n o t g l v e him e n o u g h z n f o r m a t l o n to perform the requested a c t i o n . In L m e 3 0 . J f u r t h e r c o m p o u n d s t h e e r r o r i n Line 18 b y putting t h e S P O U T on t h e TUBEBASE.

E x c e r p t 1 ( T e l e p h o n e )

A. I. Now t h e r e ' s a b l u e c a p [J g r a b s t h e T U B E B A S E ]

2. t h a t h a s t w o l i t t l e t e e t h s t i c k i n g 3. o u t o f t h e b o t t o m o f it.

J: 4. Y e a h .

A. 5. O k a y O n that take the

6. brlght shocking pink piece of plastic [J takes B A S E V A L V E ]

7. a n d stick the little hole over the t e e t h .

[J s t a r t s to i n s t a l l t h e BASEVALVE. b a c k s off, l o o k s at it a g a i n a n d t h e n g o e s a h e a d a n d i n s t a l l s it]

J. 8 O k a y

A: 9 N o w s c r e w t h a t b l u e c a p o n t o I 0 . t h e b o t t o m o f t h e m a l n t u b e . [J s c r e w s TUBEBASE onto M A I N T U B E ]

J. 11. O k a y

A. 12 Now. t h e r e ' s a - - 13. a r e d p l a s t i c p i e c e [J starts for N O Z Z L E ]

1 4 t h a t h a s f o u r g i z m o s o n It. [J switches to SLIDEVALVE]

J. 1 5 Yes.

A 16 O k a y Put the u n g t z m o e d e n d In the u h

1 7 the o t h e r - - t h e open

1 8 p a r t o f t h e m a l n t u b e , t h e l o w e r v a l v e

[3 p u t s SLIDEVALVE i n t o h o l e i n TUBEBASE, b u t A m e a n t OUTLET2 of MAINTUBE]

I 19 All r i g h t

A 20 !t ;ust hts loosely It .doesnt '~I have to f'.t right. Okay. then take .~2 the clear plastic elbow ]omt [J takes S P O U T ]

J 23 All r i g h t

A $4 A n d put tt over the bottom opening, too.

[J trees installing S P O U T o n TI/BEBASE] l -~ O k a y

a. 28. O k a y Now. t a k e t h e - -

2 7 W h i c h e n d a m I s u p p o s e d to p u t It over')

28 Do y o u k n o w °

A. -:'9 Put t h e - - p u t t h e - - t h e big e n d - - 3 0 the blg e n d over it.

[J p u s h e s big e n d of S P O U T o n T U B E B A S E . twlstlng zt to f o r c e it on]

[image:3.612.59.285.77.297.2]

NO:zZe

Figure I:

I,~.d)

I'

(a) (b)

T h e T o y W a t e r P u m p

C

2 M i s c o m m u n i c a t i o n

People m u s t a n d d o m a n a g e to resolve lots of (potentaal) m a s c o m m u m c a t a o n In e v e r y d a y conversataon. M u c h of it as resolved s u b c o n s c a o u s l y wlth the h s t e n e r u n a w a r e that a n y t h i n g is wrong, O t h e r m l s c o m m u m c a t l o n is resolved wath the listener actively deleting or replacang m f o r m a t a o n m the s p e a k e r s utterance until It flts the c u r r e n t context. S o m e t i m e s thls resolutlon Is p o s t p o n e d until the questlonable part of the utterance is actually needed. S h l l . w h e n all these fail. the h s t e n e r c a n ask the s p e a k e r to clarlfy w h a t w a s said. 2

T h e r e are m a n y aspects of a n utterance that the h s t e n e r c a n b e c o m e c o n f u s e d a b o u t a n d that c a n lead to m a s c o m m u n a c a t t o n . T h e h s t e n e r c a n b e c o m e c o n f u s e d a b o u t w h a t the s p e a k e r intends for the referents, the actaons, a n d the goals described b y the utterance, Confuslons often a p p e a r to result from confhct b e t w e e n the c u r r e n t state of the conversation. the overall goal of the speaker, or the m a n n e r In w h i c h the s p e a k e r p r e s e n t e d the anformatlon. H o w e v e r , w h e n the h s t e n e r steps b a c k a n d is able to discover w h a t k~nd of confuslon ~s occurring, t h e n the confusion c a n qulte possibly be resolved.

2.1 C a u s e s of m l s c o m m u n i c a t i o n

Thls sectaon attempts to motlvate a p a r a d l g m for the kinds of conversation that w e studled a n d traes to point out places m the p a r a d l g m that leave r o o m for m l s c o m m u m c a t l o n .

~'.1.1 E f f e c t s o f t h e s t r u c t u r e of t a s k - o r i e n t e d d i a l o g u e s

T a s k - o r i e n t e d c o n v e r s a t l o n s h a v e a speclfic goal to b e achleved: the p e r f o r m a n c e of a task (e.g.. [14]). T h e partlclpants in the dlalogue c a n h a v e the s a m e skill level a n d they c a n slmply w o r k t o g e t h e r to a c c o m p l i s h the task; or o n e of them, the expert, could k n o w m o r e a n d could direct the other, the apprentlce. to p e r f o r m the task. W e h a v e c o n c e n t r a t e d prlmarlly

o n the latter case - d u e to the protocols that w e

e x a m l n e d - but m a n y of o u r observations c a n b e g e n e r a h z e d to the f o r m e r case, too. W e will refer to thls as the a p p r e n t l c e - e x p e r t domaln.

T h e v l e w p o m t s of the expert a n d a p p r e n t l c e differ greatly In a p p r e n t l c e - e x p e r t e x c h a n g e s . T h e expert, having a n u n d e r s t a n d l n g of the f u n c t l o n a h t y of the e l e m e n t s in the task. h a s m o r e of a feel for h o w the e l e m e n t s w o r k together, h o w they go together, a n d h o w the indlvldual e l e m e n t s c a n b e used. T h e a p p r e n t l c e n o r m a l l y h a s n o s u c h k n o w l e d g e a n d m u s t b a s e hls declslons o n p e r c e p t u a l features s u c h as s h a p e [15].

T h e structure of the task affects the structure of the dlalogue [14}. partlcularly t h r o u g h the c e n t e r of attentlon of the e x p e r t a n d apprentlce. Thls is the p h e n o m e n o n called focus [14. 20. 24]. whlch, in t a s k - orlented dlalogues Is a very real a n d operational thlng (e.g., focus is u s e d In resolving a n a p h o r a c references). Shafts ~n focus c o r r e s p o n d dlrectly to the task, ats subtasks, the oblects an a task a n d the s u b p l e c e s of e a c h object F o c u s a n d focus shifts are g o v e r n e d b y m a n y rules [14. :~0, 24] C o n f u s a o n m a y result w h e n e x p e c t e d shafts d o not take place. For example. If the expert c h a n g e s focus to a n object but n e v e r discusses Its s u b p a e c e s ~such as a n obvaous a t t a c h m e n t surface) or n e v e r b o t h e r s to talk a b o u t the object r e a s o n a b l y s o o n after its antroductlon (Le., b e t w e e n the tame of ~ts m t r o d u c t l o n a n d its use. without digressing in a well- s t r u c t u r e d w a y In b e t w e e n (see [20])), t h e n the a p p r e n t l c e m a y b e c o m e confused, leavang hlm r~pe for m l s c o m m u n l c a t l o n . T h e reverse anfluence b e t w e e n focus a n d oblects c a n lead to trouble, too. A shzft In focus b y the expert that d o e s not h a v e a m a n H e s t a t l o n In the apprentlce's world wall also perplex the a p p r e n t i c e

F o c u s also influences h o w descr:ptlons are f o r m e d [15, 2]. T h e level of detail requlred in a description d e p e n d s directly o n the e l e m e n t s currently highlighted b y the f o c u s If the oblect to be descrabed Is samflar to other e l e m e n t ~ m focus, the expert m u s t be m o r e speclhc m the formulation of the descraptlon or m a y conslder shlftmg focus a w a y from the posslbly a m b i g u o u s objects to o n e w h e r e the amblgulty w o n t occur.

2.2 C o n s e q u e n c e s of m i s c o m m u n i c a t l o n

In thls section we will make It clear t h a t people d o m : s c o m m u n l c a t e a n d yet they often m a n a g e to flx thlngs. W e will look at speclfic forms of m l s c o m m u n l c a t l o n a n d descrlbe w a y s to detect them. We will h z g h h g h t relatlonsh;ps between different m l s c o m m u n z c a t ; o n p r o b l e m s but won't necessarzly d e m o n s t r a t e w a y s to resolve e a c h of them.

2An analysis of c l a r i f i c a t i o n suodialogues can be found ;n [ 1 7 ) .

2.2.1 I n s t a n c e s of m t s c o m m u n / c a t i o n

T h e r e a r e m a n y w a y s h e a r e r s c a n g e t c o n f u s e d d u r i n g a c o n v e r s a t i o n . F i g u r e 2 o u t l i n e s s o m e of t h e m t h a t w e r e d e r i v e d f r o m a n a l y z i n g t h e w a t e r p u m p p r o t o c o l s . T h i s s e c t i o n d e f i n e s a n d i l l u s t r a t e s m a n y of t h e m t h r o u g h n u m e r o u s e x c e r p t s . E a c h e x c e r p t i s m a r k e d i n p a r e n t h e s e s t o s h o w w h a t m o d a l i t y of c o m m u n i c a t i o n w a s u s e d ( s e e [9] f o r a d e s c r i p t i o n a b o u t t h e c o l l e c t i o n of t h e s e e x c e r p t s ) . E a c h b r a c k e t e d p o r t i o n of t h e e x c e r p t e x p l a i n s w h a t w a s o c c u r r i n g a t t h a t p o i n t i n t h e d i a l o g u e . T h e c o n f u s i o n s t h e m s e l v e s , c o u p l e d w i t h t h e d e s c r i p t i o n a t t h e e n d of t h i s s e c t i o n o n h o w t o r e c o g n i z e w h e n o n e of t h e m i s o c c u r r i n g , p r o v i d e s m o t i v a t i o n f o r t h e u s e of t h e a l g o r i t h m o u t l i n e d i n S e c t i o n 3 a s a m e a n s f o r r e p a i r i n g c o m m u n i c a t i o n p r o b l e m s . We w i l l o n l y d i s c u s s r e f e r e n t c o n f u s i o n tn t h i s p a p e r . The o t h e r f o r m s of c o n f u s i o n - A c t i o n . Goal, a n d C o g m t i v e L o a d - a r e d e s c r i b e d i n [11. 13]. A n o t h e r c a t e g o r i z a t i o n of c o n f u s m n s t h a t l e a d t o c o n v e r s a t i o n f a i l u r e c a n b e f o u n d in [22].

• F i g u r e 2: A t a x o n o m y of c o n f u s m n s

R e f e r e n t ~ o n f u s l o n o c c u r s w h e n t h e l i s t e n e r is u n a b l e to c o r r e c t l y d e t e r m i n e w h a t t h e s p e a k e r i s r e f e r r i n g t o w i t h a p a r t i c u l a r d e s c r l p t m n . [t o c c u r s w h e n t h e d e s c r i p t i o n s In t h e u t t e r a n c e a r e a m b i g u o u s o r i m p r e c i s e , w h e n t h e r e IS c o n f u s i o n b e t w e e n t h e s p e a k e r a n d l i s t e n e r a b o u t w h a t t h e c u r r e n t f o c u s o r c o n t e x t Is, o r w h e n t h e d e s c r i p t i o n s in t h e u t t e r a n c e a r e e i t h e r i n c o r r e c t o r i n c o m p a t i b l e w i t h t h e c u r r e n t o r g l o b a l c o n t e x t .

E r r o n e o u s S p e c i f i c i t y

A m b i g u o u s ( a n d . t h u s , i m p r e c i s e ) d e s c n p t x o n s c a n c a u s e c o n f u s i o n a b o u t t h e r e f e r e n t . E x c e r p t 2 b e l o w i l l u s t r a t e s a c a s e w h e r e t h e s p e a k e r ' s d e s c r i p t i o n is u n d e r s p e c x f l e d - i t d o e s n o t p r o v i d e e n o u g h d a t e d t o p r u n e t h e s e t of p o s s i b l e r e f e r e n t s d o w n t o o n e .

E x c e r p t 2 ( P a c e - t o - F a c e )

S 1. And n o w t a k e t h e l i t t l e r e d 3. p e g ,

[P t a k e s PLUG] 3. Yes,

4. a n d p l a c e i t xn t h e h o l e a t t h e 5. g r e e n e n d .

[P s t a r t s t o p u t PLUG i n t o OUTLETR of MAINTUBE] 6. no

7. t h e - - i n t h e g r e e n t h i n g [P p u t s PLUG i n t o g r e e n p a r t of PLUNGER]

P: 8. Okay.

In L i n e 4 a n d 5, S d e s c r i b e s t h e l o c a t i o n t o p l a c e a p e g i n t o a h o l e b y g i v i n g s p a t i a l i n f o r m a t i o n . S i n c e t h e l o c a t i o n i s g i v e n r e l a t i v e t o a n o t h e r l o c a t i o n b y " i n t h e h o l e a t t h e g r e e n e n d " , i t d e f i n e s a r e g i o n w h e r e t h e p e g m i g h t go i n s t e a d of a s p e c i f i c l o c a t i o n . In t h i s p a r t i c u l a r c a s e , t h e r e a r e t h r e e p o s s i b l e h o l e s to c h o o s e f r o m t h a t a r e n e a r t h e g r e e n e n d . T h e l i s t e n e r c h o o s e s o n e - t h e w r o n g o n e - a n d i n s e r t s t h e p e g i n t o i t . B e c a u s e t h i s d i a l o g u e t o o k p l a c e f a c e t o f a c e , S i s a b l e t o c o r r e c t t h e a m b i g u i t y i n L i n e s 6 a n d 7.

A s p e a k e r ' s d e s c r i p t i o n c a n b e i m p r e c i s e i n s e v e r a l p o s s i b l e w a y s . (1) It m a y c o n t a i n f e a t u r e s t h a t d o n o t r e a d i l y a p p l y i n t h e d o m a i n . In f i n e 3, E x c e r p t 3, t h e f e a t u r e " f u n n y " h a s n o r e l e v a n c e t o t h e l i s t e n e r . It i s n o t u n t i l A p r o v i d e s a f u l l e r d e s c r i p t i o n i n L i n e s 5 t o 8 t h a t E i s a b l e t o s e l e c t t h e p r o p e r p i e c e . (2) It m a y u s e a v a g u e h e a d n o u n c o u p l e d w i t h f e w o r no f e a t u r e v a l u e s ( a n d c o n t e x t a l o n e d o e s n o t n e c e s s a r i l y s u f f i c e t o d i s t i n g u i s h t h e o b j e c t ) . In E x c e r p t 4, Line 9, " a t t a c h m e n t " i s v a g u e b e c a u s e a l l o b j e c t s in t h e d o m a i n a r e a t t a c h a b l e p a r t s . T h e e x p e r t ' s u s e of " a t t a c h m e n t " w a s m o s t l i k e l y t o s i g n a l t h e a c t i o n t h e a p p r e n t i c e c a n e x p e c t t o t a k e n e x t . The u s e of t h e f e a t u r e v a l u e " c l e a r ' * p r o v i d e s l i t t l e b e n e f i t e i t h e r b e c a u s e t h r e e c l e a r , u n u s e d p a r t s e x i s t . The s i z e d e s c r i p t o r " l i t t l e " p r u n e s t h i s s e t of p o s s i b l e r e f e r e n t s d o w n t o t w o c o n t e n d e r s . (3) E n o u g h f e a t u r e v a l u e s a r e p r o v i d e d b u t a t l e a s t o n e v a l u e is t o o v a g u e l e a d i n g t o t r o u b l e . In E x c e r p t 5, Line 3, t h e u s e of t h e a t t r i b u t e v a l u e " r o u n d e d " t o d e s c r i b e t h e s h a p e d o e s n o t s u f f i c i e n t l y r e d u c e t h e s e t of f o u r p o s s i b l e r e f e r e n t s ( t h o u g h , i n t h i s p a r t i c u l a r i n s t a n c e , A c o r r e c t l y i d e n t i f i e s i t ) b e c a u s e t h e t e r m is a p p l i c a b l e t o n u m e r o u s p a r t s In t h e d o m m n . A m o r e p r e c i s e s h a p e d e s c r i p t o r s u c h a s " b e l l - s h a p e d " o r " c y l i n d r i c a l " w o u l d h a v e b e e n m o r e b e n e f i c i a l t o t h e l i s t e n e r ,

E x c e r p t 3 ( T e l e p h o n e )

E: I. All r i g h t .

2. Now.

3. T h e r e ' s a n o t h e r f u n n y l i t t l e 4. r e d t h i n g , a

[A i s c o n f u s e d , e x a m i n e s b o t h NOZZLE SX.,mr-VALVE ]

5. l i t t l e t e e n y r e d t h i n g t h a t ' s 6. s o m e - - s h o u l d b e s o m e w h e r e on 7. t h e d e s k , t h a t h a s u m - - t h e r e ' s 8. l i k e t e e t h on o n e e n d . [E t a k e s SLIDEVALVE]

a n d

A: 9. Okay.

E: 10. I t ' s a f u n n y - l o o - - h o l l o w , 11. h o l l o w p r o j e c t i o n on o n e e n d 12. a n d t h e n t e e t h on t h e o t h e r .

E x c e r p t 4 ( T e l e t y p e )

A: I. t a k e t h e r e d t h i n g w i t h t h e 2. p r o n g s o n i t

3. a n d f i t i t o n t o t h e o t h e r h o l e 4. of t h e c y l i n d e r

5. s o t h a t t h e p r o n g s a r e 6. s t i c k i n g o u t

[image:4.612.62.556.89.739.2] [image:4.612.56.297.97.403.2] [image:4.612.318.553.103.719.2]

R: 7. o k

A: 8. n o w t a k e t h e c l e a r l i t t l e 9. a t t a c h m e n t

10. a n d p u t o n t h e h o l e w h e r e y o u 11. j u s t p u t t h e r e d c a p o n

12. m a k e s u r e it p o i n t s 13. u p w a r d

R: 14. o k

F, x e e r p t 5 ( T e l e t y p e )

S; I. Ok,

2. p u t t h e r e d n o z z l e o n t h e o u t l e t 3. of t h e r o u n d e d c l e a r c h a m b e r

4. o k ?

A: 5. got it.

I m p r o p e r F o c u s

F o c u s c o n f u s i o n c a n o c c u r w h e n t h e s p e a k e r s e t s u p o n e f o c u s a n d t h e n p r o c e e d s w i t h a n o t h e r o n e w i t h o u t l e t t i n g t h e l i s t e n e r k n o w of t h e s w i t c h (i.e., a f o c u s s h i f t o c c u r s w i t h o u t a n y i n d i c a t i o n ) . An o p p o s i t e p h e n o m e n o n c a n a l s o h a p p e n - t h e l i s t e n e r may feel t h a t a f o c u s s h i f t h a s t a k e n p l a c e w h e n t h e s p e a k e r a c t u a l l y n e v e r i n t e n d e d o n e . T h e s e r e a l l y a r e v e r y s i m i l a r - o n e Is viewed m o r e s t r o n g l y f r o m t h e p e r s p e c t i v e of t h e s p e a k e r a n d t h e o t h e r f r o m t h e l i s t e n e r .

E x c e r p t 6 b e l o w l U u s t r a t e s a n m s t a n c e of t h e f i r s t t y p e of f o c u s c o n f u s i o n . In t h e e x c e r p t , t h e s p e a k e r (S) s h i f t s f o c u s w i t h o u t n o t i f y i n g t h e l i s t e n e r (P) of t h e s w i t c h . As t h e e x c e r p t b e g i n s , P ,s h o l d i n g

t h e TUBEBASE. S p r o v i d e s in Lines 1 to 16

m s t r u c t z o n s f o r P to a t t a c h t h e CAP a n d t h e SPOUT to o u t l e t s OUTLETI a n d OUTLET2, r e s p e c t i v e l y , o n t h e MAINTUSE. U p o n P ' s s u c c e s s f u l c o m p l e t i o n of t h e s e a t t a c h m e n t s . S s w i t c h e s f o c u s m Lines 17 to 20 to t h e TUSESASE a s s e m b l y a n d r e q u e s t s P to s c r e w tt on to t h e b o t t o m of t h e M,e/NTUSE. White P c o m p l e t e s t h e t a s k . S r e a l i z e s s h e left o u t a s t e p in t h e a s s e m b l y - t h e p l a c e m e n t of t h e SLIDEVALVE i n t o OUTLET2 of t h e M,eJNTUSE b e f o r e t h e SPOUT ts p l a c e d o v e r t h e s a m e o u t l e t . S a t t e m p t s to c o r r e c t h e r m i s t a k e by r e q u e s t i n g P to r e m o v e " t h e p i e s "~ p i e c e in ~ n e s 22 a n d 23. S i n c e S n e v e r i n d i c a t e d a s h i f t in f o c u s f r o m t h e TUSESASE b a c k to t h e IPOUT, P m t e r p r e t s " t h e p i e s " to r e f e r to t h e TUSESASE.

E x c e r p t 6 ( F a c e - t o - F a c e )

S 1. And p l a c e

2. t h e b l u e c a p t h a t ' s l e f t [P t a k e s CAP]

3. o n t h e s i d e h o l e s t h a t a r e

3The whole ward here i s " p l e e t i c . " People in general tend to be good ot p r o c e e d i n q b e f o r e h e o r i n 9 the whole u t t e r o n c e or even the whole word.

4. o n t h e c y l i n d e r , [P l a y s d o w n TUBEBASE]

5. t h e s i d e h o l e t h a t is f a r t h e s t 6. f r o m t h e g r e e n e n d .

[P p u t s CAP o n OUTLET! of MAINTUBE]

P: 7. Okay.

S; 8. And t a k e t h e n o z z l e - l o o k i n g 9. p i e c e ,

[P g r a b s NOZZLE]

10. n o

11. I m e a n t h e c l e a r p l a s t i c o n e , [P t a k e s SPOUT]

12. a n d p l a c e it o n t h e o t h e r h o l e [P i d e n t i f i e s O ~ of MA1NTUBE]

13. t h a t ' s left,

14. s o t h a t n o z z l e p o i n t s a w a y 15. f r o m t h e

[P i n s t a l l s SPOUT o n OUTLET2 of MAINTUBE]

16. r i g h t .

P: 17. Okay.

S: 18. Now

19. t a k e t h e

20. c a p b a s e t h i n g [P t a k e s TUBEBASE]

21. a n d s c r e w it o n t o t h e b o t t o m , [P s o r e w s T U B E B A S E on)L~3NTUBE]

22, ooops,

[S r e a l i z e s s h e h a s f o r g o t t e n t o h a v e P p u t S L I D L ~ A L V E i n t o OUTLET2 of MAINTUBE] 23. u n - u n d o t h e p i e s

[P s t a r t s t o t a k e TUBEBASE off MAINTUBE]

24. n o

25. t h e c l e a r p l a s t i c t h i n g t h a t I 26. t o l d y o u t o p u t o n

[P r e m o v e s SPOUT]

27. sorry.

28. And p l a c e t h e l i t t l e r e d t h i n g [P t a k e s $LID~ALVZ]

29. tn t h e r e f i r s t ,

[P m s e r t s S L X D ~ A L V Z i n t o OUTLET~ of M[AINT~E] 30. it f i t s l o o s e l y in t h e r e .

E x c e r p t 7 b e l o w d e m o n s t r a t e s t h e l a t t e r t y p e of f o c u s c o n f u s z o n t h a t o c c u r s w h e n t h e s p e a k e r (S) s e t s u p o n e f o c u s - t h e M,4]NTUBE, w h i c h is t h e c o r r e c t f o c u s In t h i s c a s e - b u t t h e n p r o c e e d s in s u c h a m a n n e r t h a t t h e l i s t e n e r (J) t h i n k s a f o c u s s h i f t to a n o t h e r p i e c e , t h e TUBESASE, h a s o c c u r r e d . T h u s , Line 15 r e f e r s t o " t h e l o w e r s i d e h o l e in t h e M,41NTUBE" f o r S a n d " t h e h o l e in t h e TUBEBASE" f o r J. J h a s no way of r e a l i z i n g t h a t h e h a s f o c u s e d i n c o r r e c t l y u n l e s s t h e d e s c r i p t i o n a s h e i n t e r p r e t s it d o e s n ' t h a v e a r e a l w o r l d c o r r e l a t e ( h e r e s o m e t h i n g d o e s s a t i s f y t h e d e s c r i p t i o n s o J d o e s n ' t s e n s e a n y p r o b l e m ) o r if, l a t e r in t h e e x c h a n g e , a c o n f l i c t a r i s e s

d u e t o t h e m i s t a k e (e.g,, a r e q u e s t e d a c t i o n c a n n o t b e p e r f o r m e d ) . In Line 31, J i n s e r t s a p i e c e i n t o t h e w r o n g h o l e b e c a u s e of t h e m i s u n d e r s t a n d i n g in Line 15. Line 31 h i n t s t h a t J may h a v e b e c o m e s u s p i c i o u s t h a t a n a m b i g u i t y e x i s t e d b u t s i n c e t h e t a s k w a s s u c c e s s f u l l y c o m p l e t e d (i.e., t h e r e d p i e c e fit i n t o t h e h o l e in t h e b a s e ) , a n d s i n c e S did n o t p r o v i d e a n y c l a r i f i c a t i o n , h e a s s u m e d h e w a s c o r r e c t .

h c e r p t 7 ( T e l e p h o n e )

S: 1. U m now.

2. Now w e ' r e g e t t i n g a l i t t l e 3. m o r e difficult.

J: 4. ( l a u g h s )

S: 5. Pick o u t t h e l a r g e a i r t u b e [ l p i c k s u p S A N D ]

6. t h a t h a s t h e p l u n g e r in it.

[J p u t s d o w n STAND. t a k e s PLUNGER/MAINTUB~.

a s s e m b l y ]

J: 7. Okay.

S: 8. And s e t it o n ~ts b a s e ,

[J p u t s d o w n idAINTUBE, s t a n d i n g v e r t i c a l l y , o n t h e TABLE]

9. w h i c h is b l u e now, 10. r z g h t ?

[J h a s s h i f t e d f o c u s t o t h e TUBEBASE]

J: 11. Yeah.

$, 12. Base is blue. 13. Okay. 14. Now

15. Y o u ' v e g o t a b o t t o m h o l e still 16. to be filled,

17. c o r r e c t ?

J: 18. Yeah.

[J a n s w e r s t h i s w i t h MAINTUBE s t i l l s i t t i n t o n t h e TABLE; h e s h o w s n o i n d i c a t i o n of w h a t h o l e h e t h i n k s i8 m e a n t - t h e o n e o n t h e MAINTUBE. OUTLET2, o r t h e o n e in t h e TUBEBASE]

[J

S.

p i c k s

19. Okay.

20. You h a v e o n e r e d p i e c e 21. r e m a m m 8 ?

u p ldA/NTUBE a s s e m b l y a n d l o o k s a t TUBEBASE, r o t a t i n e t h e MAINTUBE s o t h a t TUBP-BASE is p o i n t e d u p , a n d s e e s t h e h o l e in i t ; h e t h e n l o o k s a t t h e SLIDEVALVE]

J: 22. Yeah. 3. 23. Okay.

24. Take t h a t r e d piece. [j t a k e s SIJDEVALVE]

25. I t ' s g o t f o u r l i t t l e f e e t o n 26. it?

J: 27. Yeah.

S; 28. And p u t t h e small e n d i n t o 29. t h a t h o l e on t h e a i r t u b e - -

30. o n t h e big t u b e .

[J

s t a r t s J; 31. On t h e v e r y b o t t o m ? t o p u t it i n t o t h e b o t t o m h o l e of TUBEBASE - t h o u g h h e i n d i c a t e s h e is u n s u r e of h i m s e l f ]

S: 32. On t h e b o t t o m , 33. Yes.

M i s f o c u s c a n a l s o o c c u r w h e n t h e s p e a k e r i n a d v e r t e n t l y l e f t s t o d i s t i n g u i s h t h e p r o p e r f o c u s b e c a u s e h e did n o t n o t i c e a p o s s i b l e a m b i g u i t y ; o r w h e n , t h r o u g h n o f a u l t of t h e s p e a k e r , t h e l i s t e n e r j u s t fails t o r e c o g n i z e a s w i t c h in f o c u s i n d i c a t e d by t h e s p e a k e r . ~ x c e r p t 7 a b o v e is a n e x a m p l e of t h e f i r s t t y p e b e c a u s e S f a i l e d t o n o t i c e t h a t a n a m b l g u z t y e x i s t e d s i n c e h e n e v e r e x p l i c i t l y b r o u g h t t h e TUBEBASE e i t h e r i n t o o r o u t of f o c u s . He j u s t a s s u m e d t h a t J h a d t h e s a m e p e r s p e c t i v e a s hzm - a p e r s p e c t i v e in w h i c h u o a m b i g u i t y o c c u r r e d .

Wrong Context

C o n t e x t d i f f e r s f r o m f o c u s . The c o n t e x t of a p o r t i o n of a c o n v e r s a t i o n is c o n c e r n e d with t h e p o : n t of t h e d i s c u s s i o n in t h a t f r a g m e n t a n d with t h e s e t of o b j e c t s r e l e v a n t t o t h a t d i s c u s s i o n , t h o u g h n o t a t t e n d e d t o c u r r e n t l y . F o c u s p e r t a i n s to t h e e l e m e n t s w h i c h a r e c u r r e n t l y b e i n g a t t e n d e d to in t h e c o n t e x t . F o r e x a m p l e , two p e o p l e c a n s h a r e t h e s a m e c o n t e x t b u t h a v e d i f f e r e n t f o c u s a s s i g n m e n t s wt~hm it - w e ' r e b o t h t a l k i n g a b o u t t h e w a t e r p u m p b u t y o u ' r e d e s c r i b i n g t h e MA/NTUB£ a n d I'm d e s c r l b m g t h e AIRCH,4MB£,q. A l t e r n a t i v e l y , we c o u l d JUst be u s l n g d i f f e r e n t c o n t e x t s - I t h i n k y o u ' r e t a l k i n g a b o u t t a k i n g t h e p u m p a p a r t b u t y o u ' r e t a l k i n g a b o u t r e p l h ^ l n g t h e p u m p w i t h n e w p a r t s - m b o t h c a s e s we m~v be s h a r i n g t h e s a m e f o c u s - t h e p u m p - b u t o u r c o n t e ~ , s a r e t o t a l l y off f r o m o n e a n o t h e r . ~ The k i n d s of m i s u n d e r s t a n d i n g s t h a t c a n o c c u r b e c a u s e of c o n t e x t p r o b l e m s a r e s i m i l a r to t h o s e f o r f o c u s p r o b l e m s : (1) t h e s p e a k e r might s e t u p o r be xn o n e c o n t e x t for a d i s c u s s i o n a n d t h e n p r o c e e d in a n o t h e r o n e w i t h o u t e f f e c t i v e l y l e t t i n g t h e l i s t e n e r k n o w of t h e c h a n g e , (2) t h e l i s t e n e r may feel a c h a n g e in c o n t e x t h a s t a k e n p l a c e w h e n in f a c t t h e s p e a k e r n e v e r I n t e n d e d one, o r (:3) t h e L i s t e n e r fails t o r e c o g n i z e a n i n d i c a t e d c o n t e x t s w i t c h by t h e s p e a k e r . C o n t e x t a f f e c t s r e f e r e n c e b e c a u s e it h e l p s d e f i n e t h e s e t of a v a i l a b l e o b l e c t s t h a t a r e p o s s i b l e c o n t e n d e r s f o r t h e r e f e r e n t of t h e s p e a k e r ' s d e s c r i p t i o n s . If t h e c o n t e x t s of t h e s p e a k e r a n d l i s t e n e r differ, t h e n m | s r e f e r e n c e might r e s u l t .

B a d AnaloEy

An analogy (see [I0] for • discusslon on a n a l o g i e s ) is a u s e f u l way to h e l p d e s c r l b e an o b j e c t by a t t e m p t l n g to be m o r e p r e c i s e by u s i n g s h a r e d p a s t e x p e m e n c e a n d k n o w l e d g e - e s p e c : a l l y s h a p e a n d f u n c t i o n a l r e f o r m a t i o n . If t h a t p a s t e x p e r x e n c e o r k n o w l e d g e d o e s n ' t c o n t a i n t h e r e f o r m a t i o n t h e s p e a k e r a s s u m e s it d o e s o r i s n ' t t h e r e , t h e n t r o u b l e o c c u r s . T h u s . o n e m o r e way r e f e r e n t c o n f u s i o n can o c c u r Is by d e s c r i b i n g a n oh}act u s i n g • p o o r a n a l o g y . An a n a l o g y u s e d t o d e s c r i b e a n o b j e c t m i g h t n o t be s p e c : f i c

4Groez [14, lS] would dem~ril~ t h i s as o d i f f e r e n c e in " t a s k DIane J ~ i l e Rai¢ltlNnt [2e, 21] m~uld say that the "c0mlmmjcativa gCNlie" dJffare¢l.

e n o u g h - c o n f u s i n g t h e l i s t e n e r b e c a u s e s e v e r a l p i e c e s m i g h t c o n f o r m t o t h e a n a l o g y o r , tn f a c t , n o n e a t a l l a p p e a r t o f i t b e c a u s e d i s c o v e r i n g a m a p p i n g b e t w e e n t h e a n a l o g o u s o b j e c t a n d s o m e p i e c e i n t h e e n v i r o n m e n t Is t o o d i f f i c u l t . In E x c e r p t 8, J a t f i r s t h a s t r o u b l e c o r r e c t l y s a t i s f y i n g A's f u n c t i o n a l a n a l o g y " s t o p p e r " i n " t h e bag b l u e s t o p p e r " , b u t f i n a l l y s e l e c t s w h a t h e c o n s i d e r s t o b e t h e c l o s e s t m a t c h t o " s t o p p e r " .

E x c e r p t 8 ( T e l e p h o n e )

A: I. Okay. Now.

2. take the big blue

3. stopper that's laying a r o u n d [J g r a b s ~diCI4AMBER]

4. .. a n d take the black 5, r i n g - -

J: 6. The b i g b l u e s t o p p e r ?

[J i s c o n f u s e d a n d t r i e s t o c o m m u n i c a t e i t t o A; h e is h o l d i n g t h e AIRCHAMBER h e r e ]

A. 7 Yeah.

8. the blg blue s t o p p e r 9. a n d t h e b l a c k r i n g

[J d r o p s AIRCHAMBER a n d t a k e s t h e O - R I N G a n d t h e TUBEBASE]

In o t h e r c a s e s tt m i g h t be t o o s p e c i f i c - c o n f u s i n g t h e l i s t e n e r b e c a u s e n o n e of t h e a v a i l a b l e r e f e r e n t s a p p e a r to f i t it. In Line 8 of E x c e r p t 6, " n o z z l e - l o o k i n g " f o r m s a p o o r s h a p e a n a l o g y b e c a u s e t h e o b j e c t b e i n g r e f e r r e d t o a c t u a l l y Is a n e l b o w - s h a p e d s p o u t . The " n o z z l e - l o o k l n g " p a r t of t h e d e s c r i p t i o n c o n v i n c e d t h e l i s t e n e r t h a t w h a t h e w a s l o o k i n g f o r w a s s o m e t h i n g s p e c i f i c l i k e a n o z z l e ( w h i c h xs a s m a l l s p o u t ) . S o m e t i m e s , w h e n a n o b l e c t xs a c l e a r r e p r e s e n t a t i v e of a s p e c i f i e d a n a l o g y c l a s s , t h e a p p r e n t 2 c e m a y b e c o m e c o n f u s e d , w o n d e r i n g w h y t h e e x p e r t b o t h e r e d to f o r m a n a n a l o g y m s t e a d of j u s t d i r e c t l y d e s c r i b i n g t h e o b j e c t a s a m e m b e r of t h e c l a s s . H e n c e , tt w o u l d n o t be s u r p r i s i n g d t h e a p p r e n t i c e t g n o r e u t h e b e s t r e p r e s e n t a t n v e of t h e c l a s s f o r s o m e l e s s o b v i o u s e x e m p l a r . T h u s , f o r e x a m p l e , It ts b e t t e r t o s a y " n o z z l e " i n s t e a d of " n o z z l e - l o o k i n g . " In E x c e r p t 9, t h e d e s c r i p t i o n " h i p p o p o t a m u s f a c e s h a p e " ( a s h a p e analogy) tn Lines 2 a n d 3, a n d " c h a m p a g n e top" (a s h a p e a n a l o g y ) in Line 9. e r e t o o s p e c l h c a n d t h e h s t e n e r ts u n a b l e t o e a s i l y f i n d s o m e t h i n g c l o s e e n o u g h to m a t c h e i t h e r of t h e m . He c a n ' t d i s c o v e r a m a p p i n g b e t w e e n t h e o b l e c t in t h e a n a l o g y a n d o n e in t h e r e a l w o r l d .

E x c e r p t 9 ( A u d i o t a p e )

M; I. t a k e t h e b r i g h t p l n k f l a t 2. p i e c e of h i p p o p o t a m u s f a c e 3. s h a p e p i e c e of p l a s t i c 4. a n d y o u n o t i c e t h a t t h e t w o 5. h o l e s o n xt

[M i s t r ~ t n g t o r e f e r t o BASEVALVE] 6. m a t c h

7. a l o n g w i t h t h e two 8. p e g h o l e s o n t h e 9. c h a m p a g n e t o p s o r t of 10. l o o k i n g b o t t o m t h a t h a d

I I . t h r e a d s o n It [M i s t r y i n E t o r e f e r t o TUBEBASE]

D e s c r i p t i o n i n c o m p a t i b i l i t y

I n c o m p a t i b l e d e s c r i p t i o n s c a n l e a d t o c o n f u s i o n a l s o . A d e s c r i p t i o n i s i n c o m p a t i b l e w h e n (1) o n e o r m o r e of t h e s p e c i f i e d c o n d i t i o n s , i.e., t h e f e a t u r e v a l u e s , do n o t s a t i s f y a n y of t h e p i e c e s ; (2) w h e n o n e o r m o r e s p e c i f i e d c o n s t r a i n t s do n o t h o l d (e.g.. s a y i n g " t h e l o o s e o n e " w h e n a l l o b j e c t s a r e t i g h t l y a t t a c h e d ) .

o r (3) if n o o n e o b j e c t s a t i s f i e s al_~l of t h e f e a t u r e s s p e c i f i e d i n t h e d e s c r i p t i o n . In L i n e s 7 a n d 8 of E x c e r p t 9 a b o v e , M's u s e of " t h e t w o p e g h o l e s " l e a d s

t o b e w i l d e r m e n t f o r t h e l i s t e n e r b e c a u s e t h e d e s c r i b e d o b j e c t h a s n o h o l e s i n it. M a c t u a l l y m e a n t " t w o p e g s " .

2.2.2 D e t e c t i n g m i s c o m m u n i c a t l o n

P a r t of o u r r e s e a r c h h a s b e e n t o e x a m i n e h o w a l i s t e n e r d i s c o v e r s t h e n e e d f o r a r e p a i r of a n u t t e r a n c e o r a d e s c r i p t i o n d u r i n g c o m m u n i c a t i o n . The i n c o m p a t i b i l i t y of a r e f e r e n t o r a c t i o n is o n e s i g n a l of p o s s i b l e t r o u b l e . The a p p e a r a n c e of a n o b s t a c l e t h a t b l o c k s o n e f r o m a c h i e v i n g a g o a l is a n o t h e r i n d i c a t i o n of a p r o b l e m .

I n c o m p a t i b i l l t y

Two k i n d s of i n c o m p a t ~ b t l t t y , a c t i o n o r r e f e r e n t . a p p e a r In t h e t a x o n o m y of c o n f u s i o n s . T h e s t r o n g e s t h i n t t h a t t h e r e is a r e f e r e n c e p r o b l e m o c c u r s w h e n t h e l i s t e n e r f i n d s n o r e a l w o r l d o b j e c t to c o r r e s p o n d to t h e s p e a k e r ' s d e s c r i p t i o n . T h i s c a n o c c u r w h e n (1) o n e o r m o r e of t h e s p e c i f i e d f e a t u r e v a l u e s xn t h e d e s c r i p t i o n a r e n o t s a t i s f i e d by a n y of t h e p i e c e s ( e . g . s a y i n g " t h e o r a n g e c a p " w h e n n o n e of t h e o b j e c t s a r e o r a n g e ~ . {2) w h e n o n e o r m o r e s p e c i f i e d c o n s t r a i n t s do n o t h o l d (e.g., s a y i n g " t h e r e d p l u g t h a t f i t s l o o s e l y " w h e n a l l t h e r e d p l u g s a t t a c h t i g h t l y ) , o r (3) If no o n e o b j e c t s a t i s f i e s a l l of t h e f e a t u r e s s p e c i f i e d m t h e d e s c r i p t i o n (I.e., t h e r ' e - t s , f o r e a c h f e a t u r e , a n o b j e c t t h a t e x h i b i t s t h e s p e c i f i e d f e a t u r e v a l u e , b u t n o o n e o b j e c t e x h i b i t s a l l of t h e v a l u e s ) . An a c t i o n p r o b l e m xs l i k e l y ~f I l) t h e l i s t e n e r c a n n o t p e r f o r m t h e a c t i o n s p e c i f i e d by t h e s p e a k e r b e c a u s e of s o m e o b s t a c l e ; (2) t h e h s t e n e r p e r f o r m s t h e a c t i o n b u t d o e s n o t a r r l v e a t i t s i n t e n d e d e f f e c t (I.e., a s p e c i f i e d o r d e f a u l t c o n s t r a m t l s n t s a t i s f i e d ) ; o r (3) t h e c u r r e n t a c t i o n a f f e c t s a p r e v i o u s a c t i o n tn a n a d v e r s e way, y e t t h e s p e a k e r h a s g i v e n no s i g n of a n y i m p o r t a n c e t o t h i s s i d e - e f f e c t .

G o a l o b s t a c l e

A g o a l o b s t a c l e o c c u r s w h e n a g o a l ( o r s u b g o a [ ) o n e is t r y i n g to a c h i e v e ts b l o c k e d T h i s b l o c k a g e c a n r e s u l t m c o n f u s i o n f o r t h e h s t e n e r b e c a u s e he d i d n o t e x p e c t t h e s p e a k e r t o g i v e him t a s k s t h a t c o u l d n o t b e a c h i e v e d . O f t e n . t h o u g h , i t p o i n t s o u t f o r t h e h s t e n e r t h a t s o m e m l s c o m m u n i c a t i o n ( s u c h a s m l s r e f e r e n c e ) h a s o c c u r r e d .

G o a l r e d u n d a n c y

G o a l r e d u n d a n c y o c c u r s w h e n t h e r e q u e s t e d g o a l ( o r s u b g o a l ) is a l r e a d y s a t i s f i e d . In s o m e s e n s e , xt xs a s p e c i a l k l n d of g o a l o b s t a c l e w h e r e t h e g o a l t o be f u l f i l l e d i s b l o c k e d b e c a u s e i t i s a l r e a d y s a t i s f i e d . It is a s i m p l e g o a l o b s t a c l e b e c a u s e n o t h m g h a s t o b e d o n e t o g e t a r o u n d it. H o w e v e r , i t c a n l e a d t o c o n f u s i o n on

t h e p a r t of l i s t e n e r s b e c a u s e t h e y m a y s u s p e c t t h e y m i s u n d e r s t o o d w h a t t h e s p e a k e r h a s r e q u e s t e d s i n c e t h e y w o u l d n ' t e x p e c t a r e a s o n a b l e s p e a k e r Lo r e q u e s t t h e p e r f o r m a n c e of a n a l r e a d y c o m p l e t e d a c t i o n . I t p r o v i d e s a h i n t t h a t m i s c o m m u m c a t i o n h a s o c c u r r e d .

3 R e p a i r i n g R e f e r e n c e F a i l u r e s

3. I I n t r o d u c t i o n

T h e previous section dlustrated h o w t a s k - oriented natural l a n g u a g e m t e r a c t l o n s in the real world c a n i n d u c e contextually p o o r utterances. Given all the possibilities for confusion, w h e n confusions d o occur, they m u s t b e resolved If the task is to b e p e r f o r m e d . T h i s s e c t i o n e x p l o r e s t h e p r o b l e m o f f i x i n g r e f e r e n c e f a i l u r e s .

R e f e r e n c e I d e n t i f i c a t i o n i s a s e a r c h p r o c e s s w h e r e a l i s t e n e r l o o k s f o r s o m e t h i n g i n t h e w o r l d t h a t satisfies a speaker's uttered description. A c o m p u t a t l o n a l s c h e m e for performing reference h a s evolved from w o r k b y other artificial intelligence researchers (e.g., see [14]). That tradltlonal a p p r o a c h s u c c e e d s if a referent ~s found, or falls d n o referent ts f o u n d {see Figure 3 ( a ) ) . H o w e v e r , a reference identlficatlon c o m p o n e n t m u s t be m o r e versatile t h a n those c o n s t r u c t e d m the traditional m a n n e r . T h e excerpts provided m the prevlous section s h o w that t h e t r a d i t i o n a l a p p r o a c h i s w r o n g b e c a u s e people's r e a l behavlor zs m u c h m o r e elaborate. In particular. hsteners often find the correct referent e v e n w h e n the speaker's descrlpt)on does not

describe

a n y object In the world. For example, a s p e a k e r could descrlbe a blue block as the "turquoise block." M o s t listeners w o u l d go a h e a d a n d a s s u m e that the blue block w a s the o n e the s p e a k e r meant.

A k e y feature to reference identlficatlon is "negotlatlon." Negotlatlon in

reference

ldentlhcatlon c o m e s in two forms. First. It c a n o c c u r b e t w e e n the listener a n d the speaker. T h e listener c a n step back, e x p a n d greatly o n the speaker's descrlptlon of a plausible referent, a n d a s k for c o n h r m a t l o n that h e h a s indeed f o u n d the correct referent. For example, a h s t e n e r could mltlate negotiation wlth 'Tin confused. Are y o u talking a b o u t the thlng that is klnd of flared at the top? C o u p l e inches long. It's kind of blue." Second. negotiation c a n be wlth oneself. Thls type of negotiation, called self-negotlatlon. Ls the o n e that w e are most c o n c e r n e d wlth in thls research. T h e listener conslders aspects of the speaker's descrzptlon, the context of the c o m m u m c a t l o n , a n d the listener's o w n abdltles. H e t h e n a p p h e s that d e h b e r a t l o n to d e t e r m i n e w h e t h e r o n e referent candldate :s better t h a n a n o t h e r or. if n o candidate Is found, w h a t are the m o s t likely places for error or confuslon. S u c h negotlatlon c a n result in the listener testing w h e t h e r or not a partlcular referent works. For example, linguistic descrlptlons c a n influence a listener's perception of the world. T h e listener m u s t ask himself w h e t h e r h e c a n percelve o n e of the oblects in the world the w a y the s p e a k e r described it. in s o m e cases, the listener's perceptlon m a y overrule the descrlptlon b e c a u s e the listener can't percelve ~t the w a y the s p e a k e r described it.

To r e p a i r t h e t r a d i t i o n a l a p p r o a c h w e h a v e developed a n a l g o r i t h m t h a t c a p t u r e s f o r c e r t a i n c a s e s t h e l i s t e n e r ' s a b d i t y t o n e g o t i a t e w i t h h i m s e l f f o r a r e f e r e n t It c a n l o o k f o r a r e f e r e n t a n d . If I t d o e s n ' t

f i n d o n e , i t c a n t r y t o f i n d p o s s i b l e r e f e r e n t c a n d i d a t e s t h a t m i g h t w o r k , a n d t h e n l o o s e n t h e s p e a k e r ' s d e s c r i p t i o n u s i n g k n o w l e d g e a b o u t t h e s p e a k e r , t h e c o n v e r s a t i o n , a n d t h e l i s t e n e r h i m s e l f . T h u s . t h e r e f e r e n c e p r o c e s s b e c o m e s m u l t i - s t e p a n d r e s u m a b l e T h i s c o m p u t a t i o n a l m o d e l , w h i c h

I

c a l l "FWIM" f o r " F i n d W h a t I M e a n " , i s m o r e f a i t h f u l t o t h e d a t a t h a n t h e t r a d i t i o n a l m o d e l ( s e e F i g u r e 3 ( b ) ) .

C u r r e n t

I_ ~

RefePence

~u....=

Component

~ m i ~ = t

C u r r e n t

Reference -~ ~,,=¢..=

Component

~ ~J~milure

R e l a x a t i o n 1

Component

T¢.-,,- u

(a) Traditional

(b) FWIM

F i g u r e 3: A p p r o a c h e s t o reference ] d e n t d l c a t l o n

O n e m e a n s of m a k i n g s e n s e of a n a p p r o x l m a t e d e s c r i p t i o n i s t o d e l e t e o r r e p l a c e p o r t l o n s of i t t h a t d o n ' t m a t c h o b j e c t s In t h e h e a t e r ' s w o r l d . [n o u r p r o g r a m we a r e u s l n g " r e l a x a t i o n " t e c h n i q u e s t o c a p t u r e t h i s b e h a v i o r . O u r r e f e r e n c e i d e n t l h c a t l o n m o d u l e t r e a t s d e s c r i p t i o n s a s a p p r o x i m a t e It r e l a x e s a d e s c r i p t i o n i n o r d e r t o f i n d a r e f e r e n t w h e n t h e h t e r a l c o n t e n t of t h e d e s c r i p t i o n f a l l s t o p r o v i d e t h e n e e d e d I n f o r m a t i o n . R e l a x a t i o n . h o w e v e r , i s n o t p e r f o r m e d b l i n d l y o n t h e d e s c r i p t i o n We t r y t o m o d e l a p e r s o n ' s b e h a v i o r b y d r a w l n g o n s o u r c e s of k n o w l e d g e u s e d b y p e o p l e . We h a v e d e v e l o p e d a c o m p u t a t i o n a l m o d e l t h a t c a n r e l a x a s p e c t s o f a d e s c r l p t l o n u s i n g m a n y of t h e s e s o u r c e s of k n o w l e d g e . R e l a x a t i o n t h e n b e c o m e s a f o r m of c o m m u m c a t l o n r e p a i r [ 4 ] t h a t h e a r e r s c a n u s e .

3 . 2 T h e r e l a x a t i o n c o m p o n e n t

W h e n a d e s c r i p t i o n f a i l s t o d e n o t e a r e f e r e n t In t h e r e a l w o r l d p r o p e r l y , I t Is p o s s i b l e t o r e p a i r tt b y a r e l a x a t l o n p r o c e s s t h a t i g n o r e s o r m o d i f i e s p a r t s of t h e d e s c r l p t l o n . S i n c e a d e s c r i p t i o n c a n s p e c l f y m a n y f e a t u r e s of a n o b j e c t , t h e o r d e r In w h i c h p a r t s of It a r e r e l a x e d Is c r u c i a l (i.e.. r e l a x i n g Ln d i f f e r e n t o r d e r s c o u l d y i e l d m a t c h e s t o d i f f e r e n t o b j e c t s ) T h e r e a r e s e v e r a l k i n d s of r e l a x a t i o n p o s s i b l e One c a n i g n o r e a c o n s t i t u e n t , r e p l a c e It w i t h s o m e t h i n g c l o s e , r e p l a c e i t w i t h a r e l a t e d v a l u e , o r c h a n g e f o c u s (i.e.. c o n s i d e r a d i f f e r e n t g r o u p of o b j e c t s . ) . T h i s s e c t i o n d e s c r l b e s t h e o v e r a l l r e l a x a t l o n c o m p o n e n t t h a t d r a w s o n k n o w l e d g e s o u r c e s a b o u t d e s c r i p t i o n s a n d t h e r e a l w o r l d a s i t t r i e s t o r e l a x a n e r r o r f u l d e s c r i p t i o n t o o n e f o r w h i c h a r e f e r e n t c a n b e s d e n t l f i e d .

3 . 2 . 1 F i n d a r e f e r e n t u s i n g a r e f e r e n c e m e c h a m s m

[image:8.612.315.554.160.283.2]

r e f e r e n c e m e c h a n i s m Is t o d e t e r m i n e w h e t h e r o r n o t a s e a r c h of t h e ( t a x o n o m i c ) k n o w l e d g e b a s e t h a t we u s e t o m o d e l t h e w o r l d Is n e c e s s a r y . F o r e x a m p l e , t h e r e f e r e n c e c o m p o n e n t s h o u l d n o t b o t h e r s e a r c h i n g - u n l e s s s p e c i f i c a l l y r e q u e s t e d t o do so - f o r a r e f e r e n t f o r i n d e f i n i t e n o u n p h r a s e s ( w h i c h u s u a l l y d e s c r i b e n e w o r h y p o t h e t i c a l o b j e c t s ) o r e x t r e m e l y v a g u e d e s c r i p t i o n s ( w h i c h do n o t c l e a r l y d e s c r i b e a n o b l e c t b e c a u s e t h e y a r e c o m p o s e d of i m p r e c i s e f e a t u r e values). A n u m b e r of aspects of discourse pragmattcs c a n be u s e d in that determination (eg., the use of a

delctlc In a definite n o u n phrase, s u c h as "thls X" or "the last X", hints that the object w a s either m e n t i o n e d p r e v i o u s l y o r t h a t i t p r o b a b l y w a s e v o k e d by s o m e p r e v i o u s r e f e r e n c e , a n d t h a t i t is s e a r c h a b l e ) b u t we will n o t e x a m i n e t h e m h e r e .

T h e k n o w l e d g e base contains linguistic descriptions a n d a descrlptton of the listener's vlsual s c e n e itself. In o u r Implementation a n d algorithms, we a s s u m e It is r e p r e s e n t e d In K L - O n e [3], a s y s t e m for describing t a x o n o m i c knowledge. K L - O n e is c o m p o s e d of C O N C E P T s , R O L E s o n concepts, e n d links b e t w e e n them. A C O N C E P T Is like a set. representing those elements described b y it. A S U P E R C link ( ' = = > " ) is u s e d b e t w e e n c o n c e p t s to s h o w set Inclusion. For example, consider Figure 3. T h e S u p e r C from C o n c e p t B to C o n c e p t A is like stating B C A for two sets A a n d B A n INDIVIDUAL C O N C E P T ts u s e d to g u a r a n t e e that the subset speclhed by a c o n c e p t Is u n i q u e T h e [ndlvldual C o n c e p t D s h o w n m the figure Is d e h n e d to be a u n i q u e m e m b e r of the subset specified by C o n c e p t C R O L E s o n c o n c e p t s are like n o r m a l attributes a n d slot hllers m o t h e r k n o w l e d g e representation languages. T h e y define a functlonal relatlonshlp b e t w e e n the c o n c e p t a n d other c o n c e p t s

Concept

C I n d i v i d u a l

Concept

Figure 4: A K L - O n e T a x o n o m y

A s s u m i n g t h a t a s e a r c h of t h e k n o w l e d g e b a s e Is c o n s i d e r e d n e c e s s a r y , t h e n a r e f e r e n c e s e a r c h m e c h a n i s m ts r e v o k e d . The s e a r c h m e c h a n i s m u s e s t h e K L - O n e C l a s s t h e r [ 1 6 ] t o s e a r c h t h e k n o w l e d g e b a s e t a x o n o m y . T h l s s e a r c h Is c o n s t r a i n e d b y a f o c u s m e c h a n l s m b a s e d on t h e o n e d e v e l o p e d by G r o s z [14]. The C l a s s a f l e r ' s p u r p o s e Is t o d i s c o v e r all a p p r o p r m t e ~ u b s u m p t l o n r e l a t i o n s h i p s b e t w e e n a n e w l y f o r m e d

d e s c r l p t t o n and all o t h e r

descriptions

In a g w e n t a x o n o m y . With respect to reference, t h i s m e a n s that all possible (descriptions of) referents of the descrlptlon will be s u b s u m e d by tt after It has b e e n classLhed rote the k n o w l e d g e base taxonomy. If m o r e than o n e candidate referent Is below ( w h e n a descrlptlon A Is s u b s u m e d by B. w e say A ts "below" B) the classified description, then, unless a quantifier in the description specified m o r e t h a n o n e element, the speaker's description is ambiguous. If exactly o n e descr~ptlon Is below it, t h e n the intended referent is a s s u m e d t o h a v e b e e n f o u n d . F i n a l l y , if n o r e f e r e n t is

f o u n d b e l o w t h e c l a s s i f i e d d e s c r x p t i o n , t h e r e l a x a t i o n c o m p o n e n t i s i n v o k e d . We will o n l y c o n s i d e r t h e l a s t c a s e i n t h e r e s t of t h e p a p e r .

3.2.2 C o l l e c t v o t e s f o r o r a g a i n s t r e l a x i n g t h e d e s c r i p t i o n

It is n e c e s s a r y t o d e t e r m i n e w h e t h e r o r n o t t h e l a c k of a r e f e r e n t f o r a d e s c r i p t i o n h a s t o do w i t h t h e d e s c r i p t i o n i t s e l f (i.e.. r e f e r e n c e f a i l u r e ) o r o u t s i d e f o r c e s t h a t a r e c a u s i n g r e f e r e n c e c o n f u s i o n . F o r e x a m p l e , t h e p r o b l e m m a y b e w i t h t h e flow of t h e c o n v e r s a t i o n a n d t h e s p e a k e r ' s a n d h s t e n e r s p e r s p e c t i v e s o n i t ; i t m a y b e d u e t o m c o r r e c t a t t a c h m e n t of a m o d i f i e r ; i t m a y b e d u e t o t h e a c t i o n r e q u e s t e d ; a n d so on. P r a g m a t i c r u l e s a r e I n v o k e d to d e c i d e w h e t h e r o r n o t t h e d e s c r x p t l o n s h o u l d be r e l a x e d . T h e s e r u l e s will n o t be d i s c u s s e d h e r e so we will a s s u m e t h a t t h e p r o b l e m l i e s in t h e s p e a k e r s d e s c r i p t i o n .

3.2.3 P e r f o r m t h e r e l a x a t i o n of t h e d e s c r i p t i o n

If r e l a x a t i o n Is d e m a n d e d , t h e n t h e s y s t e m m u s t (1) f i n d p o t e n t i a l r e f e r e n t c a n d i d a t e s , ( 2 l d e t e r m i n e w h i c h f e a t u r e s in t h e s p e a k e r ' s d e s c r i p t i o n to r e l a x a n d in w h a t o r d e r , a n d u s e t h o s e o r d e r e d f e a t u r e s to o r d e r t h e p o t e n t i a l c a n d i d a t e s w i t h r e s p e c t to t h e p r e f e r r e d o r d e r i n g of f e a t u r e s , a n d (3~ d e t e r m i n e t h e p r o p e r r e l a x a t i o n t e c h n i q u e s to u s e a n d a p p l y t h e m to t h e d e s c r i p t i o n .

F i n d p o t e n t i a l r e f e r e n t c a n d i d a t e s

B e f o r e r e l a x a t i o n c a n t a k e p l a c e , p o t e n t i a l c a n d i d a t e s f o r r e f e r e n t s ( w h i c h d e n o t e e l e m e n t s in t h e l i s t e n e r ' s v i s u a l s c e n e ) m u s t f i r s t b e f o u n d T h e s e c a n d i d a t e s a r e d i s c o v e r e d by p e r f o r m i n g a " w a l k " tn t h e k n o w l e d g e b a s e t a x o n o m y in t h e g e n e r a l v l c l m t y of t h e s p e a k e r s c l a s s i f i e d d e s c r i p t i o n . A K L - O n e p a r t i a l m a r c h e r is u s e d to d e t e r m m e h o w c l o s e t h e candidate d e s c r i p t i o n s f o u n d d u r i n g t h e w a l k a r e to t h e s p e a k e r s d e s c r i p t i o n , The p a r t i a l m e t c h e r g e n e r a t e s a n u m e r i c a l s c o r e t o r e p r e s e n t how well the d e s c r l p t l o n s m a t c h ( a f t e r f i r s t g e n e r a t i n g s c o r e s a t t h e f e a t u r e l e v e l to h e l p d e t e r m m e how t h e f e a t u r e s a r e t o be a l i g n e d e n d h o w w e l l t h e y m a t c h ) . T h i s s c o r e is b a s e d on i n f o r m a t i o n a b o u t K L - O n e a n d d o e s n o t t a k e m t o a c c o u n t a n y i n f o r m a t i o n a b o u t t h e t a s k d o m a i n . The o r d e r i n g of f e a t u r e s a n d c a n d i d a t e s f o r r e l a x a t i o n d e s c r i b e d b e l o w t a k e s I n t o a c c o u n t t h e t a s k d o m a i n . The s e t of b e s t d e s c r i p t i o n s r e t u r n e d by t h e m a r c h e r ( a s d e t e r m i n e d by s o m e c u t o f f s c o r e ) a r e s e l e c t e d a s r e f e r e n t c a n d i d a t e s .

O r d e r t h e f e a t u r e s a n d c a n d i d a t e s f o r r e l a x a t i o n

At t h i s p e a t t h e r e f e r e n c e s y s t e m i n s p e c t s t h e s p e a k e r ' s d e s c r i p t i o n a n d t h e c a n d i d a t e s , d e c i d e s w t l t c h f e a t u r e s to r e l a x a n d in w h a t o r d e r . 5 a n d g e n e r a t e s a m a s t e r o r d e r i n g of f e a t u r e s f o r r e l a x a t i o n . Once t h e f e a t u r e o r d e r Is c r e a t e d , t h e r e f e r e n c e s y s t e m u s e s

50f course, om=a one ~ r t i c u l a r candidate is selected. then deciding which f e a t u r e s to relax is r e l a t i v e l y t r ( v i a l - one simply c(mporee f e a t u r e by f e a t u r e between the candidate d e s c r i p t i o n ( t h e t a r g e t ) and the speaker's d e s c r i p t i o n ( t h e ~ t t e r n ) and notes any d i s c r e p a n c i e s .

[image:9.612.57.289.393.521.2]