University of Southern California Dissertations and Theses

Definiteness types in Spanish: A study of natural discourse

(USC Thesis Other)

Definiteness types in Spanish: A study of natural discourse

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these wiil be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. Bell & Howell Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA 800-521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. D EFIN ITEN ESS TYPES IN S P A N IS H : A STU D Y OF N A T U R A L DISCOURSE by M a que la B rizu e la A D isse rta tio n Presented to the F A C U LT Y OF TH E G R A D U A T E SC H O O L U N IV E R S IT Y OF S O U TH E R N C A L IF O R N IA In P a rtia l F u lfillm e n t o f the R equirem ent fo r the Degree DOCTOR OF P H ILO SO PH Y (G eneral L in g u is tic s ) M a y 1999 C o p y rig h t 1999 M aquela B rizu e la Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 9933743 Copyright 1999 by Brizuela, Maquela All rights reserved. UMI Microform 9933743 Copyright 1999, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Road Ann Arbor, MI 48103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 90007 This dissertation, written by ...................ifc a & S iS . B.Tizuela^........................................... under the direction of h. . ?£..... Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY D ean tte Studies Date .. 1999 DISSERTATIO N C O M M IT T E E ~ \ . ~ ^ T ............................... Chairperson ] .... Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. To m y parents and sisters (with gratitude) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iii A c k n o w le d g m e n ts M a n y in d iv id u a ls have m ade s ig n ific a n t c o n trib u tio n s to m y d isse rta tio n . F irs t, I w o u ld lik e to th a n k C ecilia B rizu e la w h o helped m e to co lle ct the data. I w o u ld also lik e to acknow ledge the vo lu n te e rs w h o p a rtic ip a te d in the d iffe re n t e xp erim e nts and the in te rv ie w s th a t p ro v id e d the data fo r m y stu d y. A special thanks to the m em bers o f m y d isse rta tio n com m ittee fo r sh a rin g th e ir expertise and w is d o m fro m th e ir respective su b d iscip lin e s. Each o f them c o n trib u te d in a ve ry special w a y, and a ll o f them devote d tim e and energy to the c ritic a l re a d in g process, m a kin g e d ito ria l suggestions a t each stage o f m y research. To Jack and E laine I ow e a great deal o f a p p re cia tio n ; as m y m entors, th e y helped m e to overcom e a ll the obstacles th a t I faced d u rin g the w ritin g o f th is d is s e rta tio n . W hen I w as at UCSB (d u rin g the W in te r and S p rin g o f 1995), I m et several pe ople fro m w h o m I have benefited e n o rm o u sly in discussions o f m y data and fin d in g s : Pat C lancy, W a lly C hafe, Jack D uB ois, C. E m m o tt, Lo retta O 'C o n n o r and Sandy Thom pson. Thanks to the CogLab a t USC fo r o ffe rin g me n o t o n ly a s tim u la tin g e n viro n m e n t to w o rk in b u t also the o p p o rtu n ity to share a p h ysica l space w ith m y office-m ates. Thanks to La ura G onnerm an and K aren M arblestone, fo r b e in g u n d e rsta n d in g and fo r b e in g good frie n d s . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I also th a n k P rof. R ich a rd Joh n w h o helped m e w ith the sta tistica l analysis. I th a n k a ll th e people w h o helped m e w ith the E n g lish language: G ayle V ie rm a r C y n th ia H a g stro m and M a ry Beth Tegan. T h is d isse rta tio n is d e d ica te d to m y parents and sisters w h o su p p o rte d m e u n c o n d itio n a lly th ro u g h o u t the w h o le process. La st b u t d e fin ite ly n o t least, a v e ry sp e cia l thanks to G a b rie l K a p la n , fo r h is in va lu a b le h e lp w ith the s ta tis tic s , fo r lo n g ta lks a b o u t th e im p lic a tio n s o f th is w o rk , and fo r his ca rin g s u p p o rt and encouragem ent w h ic h helped m e to enjoy th is la rg e endea vor. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table of Contents D e d ica tio n P age.......................................................................................................i i A ckn o w le d g m e n ts................................................................................................. i i i L is t o f Tables and F ig u re s ...................................................................................ix A b stra ct......................................................................................................................x i C h a p te r 1: In tro d u c tio n 1.1. G o a l................................................................................................................1 1.2. The phenom ena.......................................................................................... 1 1.3. P re vio u s A p p ro a ch e s...............................................................................2 1.3.1. The c o n trib u tio n o f the sem antic a p p ro a ch : th e n o tio n o f u n iq u e n e s s ................................................................................................3 1.3.2. The c o n trib u tio n o f the d is c o u rs e /p ra g m a tic approach: the n o tio n o f a P-set........................................................................................ 4 1.3.3. The c o n trib u tio n o f the c o g n itiv e appro ach: p ro c e d u ra l m e a n in g and the in te ra ctio n o f fa c to rs .............................................4 1.5. The s tru c tu re o f the d is s e rta tio n ..........................................................7 C h a p te r 2: L ite ra tu re R e vie w 2.0. In tro d u c tio n ............................................................................................... 8 2.1. S panish D e te rm in e r.................................................................................8 2.1.1. S panish D e fin ite A rtic le ..........................................................................8 2.1.2. The D em o n stra tive s in S panish...........................................................10 2.1.3. P ronouns and D e fin ite N u ll Subjects in S p a n is h ........................11 2.1.4. Ina de qua cy o f these approaches..........................................................11 2.2. U niqueness................................................................................................ 12 2.2.1. C ritic is m s o f the uniqueness re q u ire m e n t.......................................13 2.2.2. The n o tio n o f uniqueness is n o t a b s o lu te .......................................14 2.2.3. U niqueness is a s u ffic ie n t b u t n o t a necessary c o n d itio n 15 2.3. The n o tio n o f a P -se t............................................................................. 16 2.4. D iscourse S tru c tu re ................................................................................19 2.4.1. The D iscourse U n it..................................................................................20 2.4.2. L in g u is tic In d ica to rs o f D iscourse U n its ......................................... 20 2.4.3. The P roperties o f D iscourse U n its ...................................................... 21 2.5. P ro ce d u ra l M e anin g o f D efiniteness T y p e s :...................................22 2.5.1. P ro ce d u ra l M e a n in g .............................................................................. 22 2.5.2. D efin itene ss Types Signal Levels o f A c c e s s ib ility ........................ 23 2.5.2.1. The A c c e s s ib ility T h e o ry ...................................................................... 23 2.5.2.2. The A c tiv a tio n T h e o ry ......................................................................... 25 2.5.2.3. N ecessary and S u fficie n t C o n d itio n s ................................................. 26 2.5.3. E vidence and M o tiv a tio n .................................................................... 29 2.5.4. S u m m a ry ..................................................................................................31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. v i 2.5.5. Factors C o n trib u tin g to A c c e s s ib ility ..............................................32 2.5.5.1. D ista n ce ................................................................................................... 32 2.5.5.2. T o p ic a lity and Frequency o f M e n tio n s........................................... 32 2.5.5.3. D iscourse S tru c tu re ............................................................................. 34 2.5.5.4. L in g u is tic Size.........................................................................................34 2.5.5.5 S ynta ctic P o s itio n ................................................................................. 36 2.5.5.6. T h e m a tic R o le .......................................................................................37 2.5.5.7. D e ixis and A c c e s s ib ility....................................................................... 38 2.5.5.8. In te ra c tio n a l A p p ro a ch e s...................................................................41 2.5.6. A cc e s s ib ility o f E n titie s vs. A c c e s s ib ility o f S ets........................ 42 2.5.6.1. A cc e s s ib ility o f co n te xts ......................................................................42 2.5.7. In te ra c tio n o f Factors............................................................................ 44 2.6. P ro m in e n c e ............................................................................................ 45 2.7. The Psycho lin g u is tic A p p ro a c h .......................................................46 2.8. C o n c lu s io n ............................................................................................. 49 C h a p te r 3: M e th o d o lo g y 3.1. N a tu re o f D ata: N a tu ra l O ra l D iscourse.........................................50 3.2. The S ubjects...........................................................................................50 3.3. The S tru ctu re o f the In te rv ie w ........................................................52 3.4. The T ra n s c rip tio n s .............................................................................. 52 3.4.1. In to n a tio n a l U n its ............................................................................... 52 3.4.2 T ra n s c rip tio n c o n v e n tio n s ............................................................... 55 3.5. The V a ria b le s.........................................................................................56 3.5.1 V a ria b le 1: T y p e .................................................................................... 56 3.5.2. V a ria b le 2: C on text (P -se t)................................................................. 58 3.5.3. V a ria b le 3: S yntactic P o s itio n ........................................................... 63 3.5.4. V a ria b le 4: Frequency o f M e n tio n ...................................................64 3.5.5. V a ria b le 5: N u m b e r o f W o rd s........................................................... 64 3.5.6. V a ria b le 6: T hem atic R ole...................................................................64 3.5.7. V a ria b le 7: D ista n ce ............................................................................. 65 3.6. The analyses...........................................................................................65 3.6.1. T e xtu a l U n its .........................................................................................65 3.6.2. A p p ro p ria te n e ss C o n d itio n s ............................................................ 66 3.6.3. A p p lie d lo g is tic regression................................................................. 66 3.7. C onclusions.............................................................................................67 C hapter 4: D e fin in g D e fin ite n e ss Types 4.0. In tro d u c tio n .............................. 68 4.1. The d e fin itio n o f de fin ite n e ss........................................................... 68 4.1.1. The d e lim ita tio n o f a P-set..................................................................69 4.1.2. The uniqueness re q u ire m e n t.............................................................70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vii 4.2. P-sets........................................................................................................ 72 4.2.1. D efiniteness types re fe r to d iffe re n t levels o f a cce ssib ility 73 4.2.1.1. H ig h accessibility P-sets.......................................................................74 4.2.1.2. M id d le accessibility P-sets...................................................................74 4.2.1.3. L o w accessibility P-sets........................................................................75 4.2.2. T e xtu a l P -sets........................................................................................ 76 4.2.2.1. D iscourse M a rk e rs ............................................................................... 78 4.2.2.2. U n it o p e n e rs......................................................................................... 80 4.2.2.3. C o m b in a tio n o f discourse m arkers and u n it openers................80 4.2.2.4. Q uestions o r com m ents b y the in te rv ie w e r................................82 4.2.2.5. Q uotes...................................................................................................... 82 4.2.2.6. C o-occurrence o f factors......................................................................83 4.3. A p p ropriaten ess co n d itio n s o f de finiten ess ty p e s ..................... 84 4.3.1 The d e fin ite a rtic le ................................................................................ 86 4.3.2. Preposed de m o n stra tive d e te rm in e r...............................................87 4.3.3. Postposed d e m o n stra tive d e te rm in e r........................................... 88 4.3.4. D e m o n stra tive p ro n o u n ................................................................... 89 4.3.5. P ersonal P ro n o u n ................................................................................92 4.3.6. C litic .........................................................................................................92 4.3.7. D e fin ite N u ll S u b je ct......................................................................... 93 4.4. M o tiv a tin g the appropriateness c o n d itio n s ..................................93 4.4.1. L o w accessible vs. h ig h accessible P-sets........................................ 95 4.4.2. P hon olog ical size o f d iffe re n t definiteness ty p e s.......................96 4.4.3. A m o u n t o f m a te ria l........................................................................... 97 4.4.4. D istance and F requency..................................................................... 99 4.4.4.1. D ista n ce.................................................................................................. 99 4.4.4.2. Frequency o f m entions o f the e n tity re fe rre d to .......................100 4.4.4.3. Frequency o f definiteness ty p e .......................................................101 4.5. C o n c lu s io n .......................................................................................... 102 C h a p te r 5: P rom inence 5.0. In tro d u c tio n ........................................................................................103 5 .1 . The d e fin itio n o f p ro m in e n c e .......................................................104 5. 2. Factors considered to be re le va n t fo r p ro m in e n ce ..................105 5.2.1. D ista n ce................................................................................................ 106 5.2.2. Frequency..............................................................................................107 5.2.3. S yntactic p o s itio n .............................................................................. 107 5.2.4. Them atic ro le ....................................................................................... 108 5.2.5. E xam ple o f c o d ific a tio n fo r the fo u r fa cto rs.............................. 108 5.3 In te ra ctio n o f factors.......................................................................... 109 5.3.1. F irs t q u a n tita tiv e p a irw is e a n a ly s is ............................................. I l l 5.3.2. The degree o f success o f each factor: re la tive stre n g th............118 5.3.3. Second q u a n tita tiv e analysis: h o w the fo u r factors fu n c tio n sim u lta n e o u sly..................................................................................... 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V l l l 5.3.4. T h ird q u a n tita tiv e a n alysis: L o g is tic Regression A n a ly s is .... 126 5.3.5 Sum m ary fo r 5.3................................................................................... 129 5.4. E xp la in in g the ra n k in g ..................................................................... 129 5.4.1. M e ntal re p re se n ta tio n b y G e m sb a ch e r........................................130 5.4.2. M y discourse re p re se n ta tio n th e o ry .............................................133 5.4.3. M o tiv a tin g the h ie ra rc h y o f fa c to rs ..............................................134 5.4.3.1. G ram m atical and th e m a tic ro le .......................................................136 5.4.3.2. D istance and fre q u e n c y ..................................................................... 137 5.4.4. The in te ra c tio n o f fa cto rs in th e p re vio u s lite ra tu re ...............139 5.5. C o n c lu s io n ........................................................................................... 140 Chapter 6: Conclusions 6.0. In tro d u c tio n ............................................................................................... 141 6.1. The appropriateness c o n d itio n s o f definiteness typ e s....................142 6.2. M o tiv a tin g the ap pro pria tene ss c o n d itio n s ...................................... 143 6.3. T e xtua l P -sets............................................................................................. 144 6.4. The in te ra ctio n o f fa cto rs and th e d e fin itio n o f p ro m in e n c e 145 6.5. M o tiv a tin g the p ro m in e n ce h ie ra rc h y ............................................. 146 6.6. F u ture d ire ctio n s........................................................................................ 149 A p p e n d ix A ...........................................................................................................152 References.............................................................................................................. 156 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ix L is t o f Tables Table 1: In fo rm a tio n a b o u t the subjects..................................................... 51 Table 2: L in g u is tic in d ic a to rs and consensus fo r s h iftin g discourse u n its ....................................................................................................... 79 Table 3: Percentage o f definiteness types and a cce ssib ility o f P -sets... 85 Table 4: Prep. D em . and D em . P ronoun: the ro le o f distance to w a rd s th e antecedents.................................................................. 91 Table 5: A note o n E S T E /E S E ........................................................................ 91 Table 6: A m o u n t o f lin g u is tic m a te ria l and ty p e o f N P ........................98 Table 7 F actor D istance: m ean and m e d ia n o f n u m b e r o f IU s fro m the a n a p h o r to th e an tece den t.....................................................100 Table 8: F actor Frequency: m ean o f n u m b e r o f m e n tio n s o f the e n tity re fe rre d to b y the an aphor in the p re v io u s 25 IU s.... 101 Table 9: F requency o f definiteness ty p e s ...................................................102 Table 10: A ntecedents a n d th e ir co m p e tito rs in re la tio n to fo u r factors ................................................................................................................ 109 Table 11: The in te ra c tio n o f D istance and F requency o f M e ntion s ....116 Table 12: The in te ra c tio n o f D istance and F requency o f M e ntion s and th e ir re la tiv e s tre n g th .....................................................................120 Table 13: R atio o f success o f D istance b y degree w ith respect to Frequency o f M e n tio n s .................................................................. 121 Table 14: R atio o f success o f Frequency o f M e n tio n s b y degree w ith respect to D istance............................................................................ 121 Table 15: C o d in g fo r the second q u a n tita tiv e a n a lysis........................... 123 Table 16: C o m p a riso n o f m odels w ith d iffe re n t fa cto rs..........................128 Table 17: The R a n kin g o f factors fo r p ro m in e n ce .....................................129 Table 18: The in te ra c tio n o f D istance an d S ynta ctic P o s itio n ...............152 Table 19: The in te ra c tio n o f D istance and S ynta ctic P o s itio n and th e ir re la tiv e s tre n g th .................................................................... 152 Table 20: The in te ra c tio n o f D istance and T h e m a tic R o le ....................152 Table 21: The in te ra c tio n o f D istance a n d T h em a tic R ole and th e ir re la tiv e s tre n g th ...............................................................................153 Table 22: The in te ra c tio n o f Frequency o f M e n tio n s and S yntactic P o s itio n ............................................................................................... 153 Table 23: The in te ra c tio n o f Frequency o f M e n tio n s b y S yntactic P o sitio n a n d th e ir re la tive s tre n g th ............................................ 154 Table 24: The in te ra c tio n o f Frequency o f M e n tio n s and Them atic R o le .......................................................................................................154 Table 25: The in te ra c tio n o f F requency o f M e n tio n s b y Them atic R ole and th e ir re la tiv e s tre n g th ..................................................155 Table 26: The in te ra c tio n o f S yntactic P o s itio n and T hem atic R o le ...................................................................................................... 155 Table 27: In te ra c tio n S ynta ctic p o s itio n b y T h e m a tic R ole and th e ir re la tiv e s tre n g th ............................................................................... 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X F igures F ig ure 1. C o m b in a tio n o f factors. Ties in c lu d e d ...................................... 124 F ig u re 2. C o m b in a tio n o f factors. Ties re -cod ed.......................................125 F ig u re 3. M e n ta l R epresentation o f a T e x t................................................ 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A bstract D e fin ite n e ss T ypes in S panish: A S tu d y o f N a tu ra l D iscourse W hy do speakers select one typ e o f d e fin ite N o u n Phrase (NP) ra th e r tha n another? T h is d isse rta tio n explores th e sem antic, pragm atic and te x tu a l differences betw een the fo llo w in g definiteness types in Spanish u sin g n a tu ra lly o ccu rrin g discourse: (1) la idea 'th e idea' (2) esta idea 'th is id e a ’ (3) la idea esta 'th is idea he re' [lite ra lly ] 'th e idea th is' (4) esta 'th is ' (5) ella 'she ' (6) pro piensa. 'p ro th in k s ' (7) Ella Z o piensa 'She th in k s it ' In this d isse rta tio n , I p ro v id e the appropriateness co n d itio n s fo r each definiteness typ e , based on the uniqueness re q u ire m e n t and a m odel o f discourse re pre senta tion. R efe rring expressions p la y a fun da m en tal ro le in fo rm in g a coherent discourse representation th ro u g h the lin k in g a n d h ig h lig h tin g o f in fo rm a tio n , m a kin g it m ore o r less accessible (G em sbacher, 1990; G ivon, 1992; G rosz, Joshi, and W einstein, 1995). O ne c o n trib u tio n o f this d isse rta tio n is to p ro vid e an op era tiona l m o del o f discourse representation based on an analysis o f lin g u is tic in d ic a to rs th a t d e lim it discourse u n its. Several researchers, m o st n o ta b ly A rie l 1990 and G ivo n 1992, have argued th a t d iffe re n t re fe re n tia l expressions m a rk acce ssibility on the basis o f the fo llo w in g factors (am ong others): 1) distance (closer is m ore accessible); 2) fre q u e n cy o f m e n tio n (m ore fre q u e n t is m ore accessible); 3) synta ctic p o s itio n (subjects are m ore accessible than objects); 4) the m atic ro le (a g e n ts/e xp e rie n ce s are m ore accessible Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. xii th a n them es). N o -o n e , h o w e ve r, has e xp la in e d the re la tiv e c o n trib u tio n o f each o f these factors to a cce ssibility n o r h o w th e y in te ra ct. T his d is s e rta tio n p ro v id e s a m ore com plete account o f a cce ssib ility b y a n a ly z in g th e ir in te ra c tio n in the in te rp re ta tio n o f d iffe re n t re fe re n tia l expressions. Three d iffe re n t q u a n tita tiv e analyses have been con d u cte d : a p a irw is e com pariso n o f factors, a d e s c rip tiv e analysis o f h o w th e fo u r factors fu n c tio n sim u lta n e o u s ly , a n d an a p p lie d m u ltip le regression. A ll three in d ica te th a t distance a n d fre q u e n cy o f m e n tio n are h ig h ly s ig n ific a n t factors fo r antecedent sele ction w h ile syn ta ctic and th e m a tic roles are n o t s ig n ific a n t. T h is d is s e rta tio n challenges th e v ie w th a t the m e n ta l re p re se n ta tio n o f discourse in clu d e s syn ta ctic and th e m a tic in fo rm a tio n (G o rdo n e t al, 1993; M cK o o n e t al., 1993b). Instead , these fin d in g s are consistent w ith the orie s o f discourse re p re se n ta tio n th a t m o d e l d iffe re n t levels o f a c tiv a tio n fo r en titie s. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 C ha pter 1 In tro d u c tio n 1.1. G o a l The g o a l o f th is d is s e rta tio n is to propose a m o del o f d e fin ite n e ss types in S panish th a t achieves th e fo llo w in g : 1) it is based on an analysis o f n a tu ra l data, n o t y e t u n d e rta ke n in Spanish (w ith one e xce ption , cf. B e n tiv o g lio , 1983); 2) it in te gra tes c e rta in key in s ig h ts fro m d iffe re n t research tra d itio n s and goes b e yo n d the p re v io u s discrete approaches; 3) it m akes an e m p iric a l c o n trib u tio n , g iv in g a m o re th o ro u g h account o f c e rta in fu n d a m e n ta l n o tio n s th a t have p re v io u s ly been discussed (uniqueness, a cce ssib ility, prom inence); 4) it m akes a th e o re tic a l c o n trib u tio n b y e x p la in in g h o w d iffe re n t factors in te ra c t in the in te rp re ta tio n and selection o f d e fin ite NPs. 1.2. T h e ph eno m en a I p ro v id e ap pro pria tene ss c o n d itio n s fo r definiteness typ e s1 (1-7 b e lo w ) in S panish N a tu ra l D iscourse. D e fin ite A rtic le (1) la idea 'th e id e a ' Preposed D e m o n stra tive (2) esta idea 'th is id e a ' Postposed D e m o n stra tive (3) la idea esta 'th is id e a h e re ', lit. 'T he idea th is ' D e m o n stra tive P ro n o u n (4) esta 'th is ' P ersonal P ro n o u n (5) ella 'she' D e fin ite N u ll S ubject (6) pro piensa. 'p ro th in k s ' C litic o f D O and IO (7) Ella lo piensa 'She th in k s it ' 1 Possessives are not considered in this study. For instance his book, involves reference to the entity 'book' and to the possessor o f the book; this NP is bivalent, referring to two entities at the sam e time, which is w h y I did not take them into account. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 M y d isse rta tio n addresses (l)-(7 )/ based o n analyses o f te xtu a l data. B y e xa m in in g n a tu ra l discourse, the fo llo w in g question can be answ ered: w h y does the speaker select one ty p e o f N P ra ther than another? In o rd e r to answ er th is q u estion I take in sig h ts fro m d iffe re n t theories and p ro v id e a u n ifie d account th a t in clu d e s k e y n o tions (such as uniqueness, p ro m in e n ce and se t-accessib ility) fro m d iffe re n t perspectives: the p ra g m a tic account in H a w k in s , 1991; Sperber & W ilso n , 1986; a n d th e fu n c tio n a l account in A rie l, 1990; Chafe, 1994; E pstein, 1994. I re -d e fin e these n o tions and I in co rp o ra te them in to a n e w m odel. 1.3. P re vio u s approaches The issue o f definiteness has been approached fro m d iffe re n t perspectives. I t is possible to classify the m v e ry lo o se ly in the fo llo w in g categories: • The sem antic approach th a t has c o n trib u te d the n o tio n o f uniqueness (cf. 2.2) • The d isco u rse /p ra g m a tic approach th a t has co n trib u te d the n o tio n o f a P-set o r discourse u n it (cf. 2.3) • The c o g n itiv e approach th a t firs t, has in q u ire d a b out the d iffe re n t c o g n itiv e factors th a t in flu e n ce reference selection and th a t second, has con sidered the p ro ce d u ra l m e a n in g o f re fe re n tia l expressions. In th is m o d e l, d e fin ite expressions are in stru ctio n s to the liste n e r to p ic k u p the rig h t reference fo r the in te rp re ta tio n o f the d e fin ite N P . Each o f these approaches presents us w ith d iffe re n t problem s. F o r instance, the sem antic approach is based e x c lu s iv e ly on in ve n te d Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 exam ples and on in tu itio n s . I t does n o t deal w ith language in use. In a d d itio n , the scholars in th is fra m e w o rk are n o t concerned a b o u t h o w th e lin g u is tic theory th e y p ro v id e m ig h t be represented in the m in d s o f speakers and listeners. I p ro v id e an a lte rn a tive account in cha pte r 4 based on n a tu ra l discourse and m o tiv a te d b y a th e o ry o f m e n ta l re p re se n ta tio n . The d isco u rse /p ra g m a tic approach is som etim es vague and n o t v e ry ca re fu l in the d e fin itio n o f basic term s. M o re o ve r, the vagueness and la ck o f precision, m akes it som etim es d iffic u lt to re p lic a te the analyses in th is approach and th e re fo re , th e y la ck re lia b ility (cf. C hapter 2). F in a lly , the c o g n itiv e ap pro ach presents d iffe re n t factors th a t in flu e n ce reference selection b u t it does n o t te ll us h o w these d iffe re n t factors in te ra ct, an issue th a t I w ill address in chapter 5. 1.3.1. The c o n trib u tio n o f th e sem antic approach: th e n o tio n o f uniqueness The n o tio n o f uniqueness is firs t presented b y R ussell, 1905 (cf. section 2.2) and then re in te rp re te d b y H a w kin s, 1978, 1991. Three d iffe re n t re strictio n s o f the uniqueness re q u ire m e n t have been presented. F irst, H a w kin s (1991) re stricts the re q u ire m e n t o f uniqueness sta tin g th a t it does n o t h o ld in absolute term s b u t w ith in p ra g m a tic a lly d e lim ite d sets (P-sets). Second, B im e r & W a rd (1994) have sho w n th a t the re q u ire m e n t o f uniqueness does n o t h o ld w h e n th e e n tity re fe rre d to is n o t re le v a n tly d iffe re n tia b le fro m an y o th e r e n tity o f the sam e k in d . In effect, the en titie s w ith in the re le v a n t P-set are n o t in d iv id u a te d in any Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 s ig n ific a n t w a y fo r the in te rp re ta tio n o f the N P . T h is issue is also discussed in ch a p te r 4. T h ird , th e uniqueness re q u ire m e n t seems to be co n tra d icte d w h e n w e observe h o w the d e fin ite n u ll subject, th e c litic and the p e rson al p ro n o u n fu n c tio n . A ll o f these re fe re n tia l expressions m ay re fe r to n o n -u n iq u e e n titie s , because there is m o re th a n one re fere nt s e m a n tica lly co m p a tib le w ith the d e fin ite N P . H o w e v e r, there is so m e th in g u n iq u e a b o u t those e n titie s th a t are selected fo r reference in te rp re ta tio n : th e m o st p ro m in e n t e n tity is p ic k e d fo r reference selection a n d th e re fo re , in these cases, uniqueness is achieved th ro u g h p ro m in e n ce (cf. ch a p te r 5). 1.3.2. T h e c o n trib u tio n o f th e d isco u rse /p ra g m a tic approach: the n o tio n o f a P-set A s m e n tio n e d above, the uniqueness re q u ire m e n t h o ld s w ith in a re le va n t P -set. These sets in clu d e p o rtio n s o f the com m on kn o w le d g e , the p h y s ic a l situ a tio n , and the te xt. I extend and a p p ly these n o tio n s to discourse re pre senta tion in ch a p te r 4. 1.3.3. T h e c o n trib u tio n o f the c o g n itiv e appro ach: p ro ce d u ra l m e a n in g an d th e in te ra c tio n o f factors F u n c tio n a l the orie s (A rie l, 1990; C hafe, 1994; G iv o n , 1992) have described d e fin ite N Ps u s in g n o tio n s such as a cce ssib ility, a ctiva tio n and to p ic a lity . E ven th o u g h these concepts are d iffe re n t, there is a basic in s ig h t th a t in te g ra te s them : d iffe re n t d e fin ite N P s s ig n a l d iffe re n t leve ls o f e n tity a c tiv a tio n . In oth er w o rd s, d e fin ite expressions have p ro ce d u ra l m e a n in g , because the y are in s tru c tio n s to th e liste n e r to p ic k the rig h t re fe re n t fo r the in te rp re ta tio n o f th e d e fin ite expression. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 This has been tested in d iffe re n t w ays in d iffe re n t languages. These theories fa il, h o w e ve r, to g iv e a u n ifie d account o f th e d iffe ren ces am ong these NPs (cf. ch a p te r 2). A n d w h ile the in s ig h t I e xp lo re m ay be im p lic it in e a rlie r analyses, no th e o ry addresses i t e x p lic itly . I m ake the n o tio n o f a cce ssib ility o p e ra tio n a l, c la im in g th a t th e d e fin ite a rtic le , the preposed d e m o n stra tive d e te rm in e r, the p o stp o se d d e m o n stra tive d e te rm in e r and the d e m o n s tra tiv e p ro n o u n re fe r to u n iq u e e n titie s w ith in d iffe re n t P-sets w h ic h are ra n ke d a cco rd in g to d iffe re n t levels o f a cce ssib ility. F or instance, no one has re m arke d o n th e im p o rta n ce o f co n sid e rin g tw o d iffe re n t k in d s o f a cce ssib ility- the a c c e s s ib ility o f d iffe re n t contexts (te x t, co n te xtu a l kn o w le d g e , p h y s ic a l s itu a tio n ), w h ic h I c a ll set-accessib ility, and the a cce ssib ility o f e n titie s , w h ic h I ca ll prom inence. It is clea r th a t the le v e l o f a cce ssib ility in th e e n tity 's con text w ill have an e ffe ct o n the le ve l o f a cce ssib ility o f the e n tity its e lf. H o w e ve r, th e d is tin c tio n betw een p ro m in e n ce a n d set- a cce ssib ility a llo w s us to m ake a g e n e ra liza tio n a b o u t a ll d e fin ite expressions as is sh o w n in chapters 4 and 5. M o re o ve r, A rie l (1990) in h e r a cce ssib ility th e o ry , p o in ts o u t th a t h e r analysis is re s tric te d to o n ly one fa c to r w h ic h in flu e n ce s a cce ssib ility (distance) a lth o u g h she acknow ledges th a t o th e r factors need to be taken in to account. I extend the th e o ry o f a cce ssib ility b y lo o k in g a t o th e r factors. Several factors have been sh o w n to have an e ffe c t o n the prom ine nce o f e n titie s. F o r instance it has been s h o w n th a t s h o rt size is re la te d to pro m in e n ce (to p ic a lity , G iv o n , 1992; a cce ssib ility, A rie l, 1990). I argue th a t fo u r d iffe re n t factors need to be ta ke n in to account Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 fo r a d e fin itio n , o f p ro m in e n ce : distance, fre q u e n cy, syn ta ctic p o sitio n , and them atic p o s itio n . I assum e th a t these fa cto rs are re la te d , and show th a t th e y have d iffe re n t levels o f im p o rta n ce . In cha pte r 5 ,1 p ro v id e three d iffe re n t q u a n tita tiv e analyses (a p a ir-w is e com parison, the analysis o f h o w th e fo u r factors fu n c tio n s im u lta n e o u s ly , and a lo g is tic regression a n a lysis) w h ic h reveal the in te ra c tio n o f these fo u r factors and th e ir re la tiv e strengths. S um m ing u p , th is d isse rta tio n w ill account fo r a m o d e l o f definiteness types in S panish u sin g the fo llo w in g ke y n o tio n s: uniqueness, a cce ssib ility o f P-sets and a cce ssib ility o f e n titie s (i.e. p ro m in e n ce ). The c o n trib u tio n o f th is d isse rta tio n is b o th th e o re tic a l and practical. I co n trib u te to the general lite ra tu re o f S panish b y p ro v id in g the firs t com prehensive account o f d e finiten ess types, in c lu d in g the postposed d e m o n stra tive d e te rm in e r, w h ic h has n o t y e t been extensively stu d ie d . T h e o re tica lly, I in te g ra te c e rta in k e y in sig h ts fro m d iffe re n t research tra d itio n s in a m ore com plete u n ifie d account. M y approach is eclectic in a p o s itiv e sense, ta k in g th e uniqueness re q u ire m e n t fro m th e lo g ic a l and sem antic tra d itio n s an d re in te rp re tin g it: firs t, b y d e lim itin g its scope to P-sets (H a w k in s , 1991), e la borating th is la tte r n o tio n , and second b y e x p la in in g a p p a re n t shortcom ings (irre le v a n t uniqueness and uniqueness achieved th ro u g h prom ine nce ). M o re o ve r, I describe h o w these d e fin ite NPs d iffe r and w h y som etim es th e y are n o t possible in ce rta in contexts. I use and extend the a cce ssib ility th e o ry and I p ro v id e a th e o re tica l and data m o tiva te d acco u n t o f prom inence. P ra c tic a lly , th is acco u n t gives a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 b e tte r d e s c rip tio n o f the d iffe re n t usages attested in the data I analyzed. T herefore, I p ro v id e a general m o d e l th a t im p ro ve s o n the th e o ry o f d e fin ite expressions o v e ra ll b y re v e a lin g the in te ra c tio n o f d iffe re n t ke y concepts (such as uniqueness, a cce ssib ility and prom ine nce ) w ith in a u n ifie d th e o ry . 1.5. The structure of the dissertation T h is d isse rta tio n is org a n ize d as fo llo w s . C hapter 2 presents the p re vio u s approaches to d e fin ite expressions. C hapter 3 presents the m e th o d o lo g y o f th is stu d y. C ha pter 4 and 5 present an answ er to the p rim a ry q u e stio n posed: w h y does th e speaker select one ty p e o f N P ra th e r th a n another? C ha pter 4 analyses the appropriateness c o n d itio n s fo r d iffe re n t definiteness types in term s o f a cce ssib ility o f the P-sets in w h ic h th e ir referents are u n iq u e . C ha pter 5 sp e lls o u t the factors th a t m ake e n tities m ore o r less accessible w ith in a P-set, w ith the re s u lt th a t uniqueness can be achieved in the re le va n t sets. C hapter 6 presents the conclusions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 C h a p te r 2 L ite ra tu re R e vie w 2.0. In tro d u c tio n In th is chapter, I p re se n t the m o st re le va n t stu d ie s o f d e finiten ess types h ith e rto . F irs t, th e lite ra tu re on S panish definiteness types is sum m arized. The re le v a n t stud ies are in co m p le te because they do n o t take in to account su b tle p ra g m a tic phenom ena. T hen, I tu rn to the general lin g u is tic lite ra tu re and I organ ize the p re s e n ta tio n aroun d three ke y n o tions: the n o tio n o f uniqueness, the n o tio n o f P-set, and the n o tio n o f a cce ssib ility o r p ro c e d u ra l m eaning. F irs t, I discuss the re vise d re q u ire m e n t o f uniqueness presented in H a w k in s (1991) and ta ke n fro m R ussell (1905). F o r instance, the re q u ire m e n t o f uniqueness is n o t absolute: it h o ld s w ith in the lim its o f a P-set (p ra g m a tica lly d e lim ite d set o f e n titie s, H a w k in s , 1991). Second, I present d iffe re n t the orie s o f d isco urse re p re se n ta tio n , and the n o tio n o f a discourse u n it, w h ic h I in te g ra te in to H a w k in s ' (1991) th e o ry o f P-sets (H a w k in s , 1991). P-sets can be te x tu a l o r n o n te x tu a l and the y are ra n ke d a cco rd in g to a cce ssib ility (A rie l, 1990). A c c e s s ib ility is the last ke y n o tio n to be sum m arize d. D iffe re n t scholars have ta lk e d about these issues fro m v e ry d iffe re n t pe rsp e ctive s and fra m e w o rks. This cha pte r is a com prehensive and c ritic a l su m m a ry o f th e m . 2.1 S panish D e te rm in e rs 2.1.1. S panish D e fin ite A rtic le The lite ra tu re o n th e d e fin ite a rtic le (el/la/los/las) in S panish is sparse. The lite ra tu re m ig h t be s u b -d iv id e d in to tw o g roup s. The firs t Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 g ro u p m a in ta in s th a t the d e fin ite a rtic le re fers to fa m ilia r/g iv e n e n titie s , w h ile th e in d e fin ite a rticle refers to n e w e n titie s (Pease- G o rrise n , 1980; K le in , 1976; B e ll,1982). T h is ap p ro a ch has been show n to be in a deq uate in the lite ra tu re on E n g lish (cf. H a w k in s , 1978). The second g ro u p (B u ll, 1965, K in g , 1992, G arcia Fajardo 1989, 1991) in co rp o ra te s the uniqueness in s ig h t (R ussell, 1905; H a w kin s, 1978). These a u th o rs assert th a t the d e fin ite a rtic le refers u n iq u e ly , w h ile the in d e fin ite a rtic le refers to one e n tity o f m any. A s B u ll (1965: 215) p u ts it: 'th e d e fin ite a rtic le indicates to ta lity (one u n iq u e ) and the in d e fin ite a rtic le m a rks p a rtitive n e ss (one o f m a n y )'. B u ll (1965) m e n tio n s the fa ct the 'th e use o f the D e fin ite A rtic le /In d e fin ite A rtic le d e te rm in e r re fle cts the p o in t o f v ie w o f the spe aker'. T his k in d o f in s ig h t is w h a t leads H a w k in s to re fo rm u la te R usse ll's uniqueness th e o ry to one w h ic h is tie d to the purposes o f th e p a rtic ip a n ts as is sh o w n b e lo w . A m o re m o d e m a tte m p t to account fo r th e d e fin ite a rtic le in S panish is th a t o f L e o n e tti (1993) w h o in clu d e s S perber & W ilso n 's (1986) p ra g m a tic p rin c ip le s . C onsider, fo r instance exam ple (1) taken fro m W ils o n (1992): (1) I switched from linguistics to geography. The lectures were less boring. T he in te rp re ta tio n o f the lectures as geography lectures and n o t linguistic lectures is based o n the p rin c ip le o f relevance th a t speakers and liste n e rs share (W ilso n , 1992). T his p rin c ip le m a in ta in s: 'try to f u lf ill y o u r co m m u n ica tive purposes w ith the le a st possible e ffo rt'. W hen re fe rrin g w ith a d e fin ite a rticle the lis te n e r is in s tru c te d to lo o k Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 fo r th e re fe re n t in the d iffe re n t possible contexts (the m in im u m nu m b e r o f assum ptions o r b e lie fs th a t h e lp the listen er to in te rp re t the d e fin ite N P ). The liste n e r stops searching w hen an in te rp re ta tio n is coherent w ith the relevance p rin c ip le . In (1), the liste n e r stops w h e n the second utte ra n ce (The lectures were less boring) is in te rp re te d as a ju s tific a tio n o f the firs t u ttera nce (I switched from linguistics to geography), and the refore 'th e le ctu re s' are in te rp re te d as 'th e ge ography lectures.' The questions fo r th is approach are: h o w are contexts d e fin e d precisely? h o w do th e y w o rk in n a tu ra l data? h o w does the p ro ce d u re to in te rp re t the d e fin ite a rticle d iffe r fro m the procedures to in te rp re t o th e r definiteness types (fo r instance personal p ro n o u n s )? 2.1.2. T he D e m o n stra tive s in S pan ish D e m o n stra tive s in S panish have been stu d ie d as d e ic tic elem ents fo llo w in g the tra d itio n a l analysis in general lin g u is tic s (Lyons, 1977). K in g (1992) and C ia p u sd o (1988) are exam ples. D e ix is , as K in g (1992) p u ts it, is the sem antic n o tio n th a t allow s the speakers to use them selves as the p o in t o f reference fo r the id e n tific a tio n o f real w o rld e n titie s . The p ro x im a l d e m o n stra tive in Spanish (este/a) refers to e n titie s close to the speaker, the n o n -p ro x im a l d e m o n stra tive (ese/a) refers to e n titie s close to the liste n e r, and fin a lly the rem ote d e m o n s tra tiv e (aquel/aquella) refers to e n titie s th a t are fa r fro m b o th p a rtic ip a n ts . The co n s tru c tio n w ith the postposed d e m o n stra tive (la idea esta [lit] 'the ide a th is ,' 'th is idea here') has o n ly been stu d ie d s u p e rfic ia lly . Sole and Sole (1977) a ffirm th a t it expresses d isd a in and B ergen (1977) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 also claim s th a t it in v o lv e s a p e jo ra tiv e m eaning. Fernandez R am irez (1987) asserts th a t the p o s tp o s itio n o f de m onstratives is used w ith an evo cative o r expressive fu n c tio n . T he expressive fu n c tio n is re la te d to the p e jo ra tiv e m e anin g m e n tio n e d above. The e vo ca tive fu n c tio n is d e fin e d as 'a re trie v a l in m e m o ry' (u n lla m a m ie n to a la m e m o ria ) and it refers to the fa ct th a t the postpo sed d e m on strative m ig h t re fe r back to an e n tity th a t has been m e n tio n e d before, and is n o t p a rt o f the co n ve rsa tio n th a t is g o in g o n a t th e m om ent o f speech. I w ill re la te th is usage to the n o tio n o f a c tiv a tio n (Chafe, 1976, 1994; A rie l, 1988, 1990) and to the fa ct th a t one o f th e m a in fu n ctio n s o f the postposed d e m o n stra tive is to re trie ve e n titie s th a t are n o t fu lly a ctiva te d (cf. cha pte r 4). 2.1.3. Pronouns and D efinite N ull Subjects in Spanish F o r the syn ta ctic subject p o s itio n , the o p p o s itio n o f an expressed subject (use o f the p ro n o u n ) vs. zero has been in v e s tig a te d in several studies. For instance B e n tiv o g lio , (1983) fo u n d th a t in spoken Spanish, m a x im u m c o n tin u ity (closer, fre q u e n tly m e n tio n e d , u n a m b ig u o u s antecedents) is lin g u is tic a lly encoded b y verb agreem ent and m in im u m c o n tin u ity is encoded b y e xiste n tia l c o n stru ctio n s w ith in d e fin ite NPs. A s B la ckw e ll (1998) sum m arizes it, th re e d iffe re n t exp la n a tio n s have been g iv e n fo r th is . The p ro n o u n is used to a vo id a m b ig u ity , to m a n ife st co n tra st an d to m a rk em phasis. 2.1.4. Inadequacy of these approaches The p ro b le m w ith a ll these accounts is th a t, firs t, th e y do n o t p ro v id e accurate ap propriateness c o n d itio n s fo r these d e fin ite n e ss types. T hey use vague term s, such as p e jo ra tive m e anin g (fo r the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 postposed d e m o n stra tive ), em phasis (fo r the p ro n u n c ia tio n o f the subject), expressive fu n c tio n (fo r the postposed d e m o n stra tive ) none o f w h ic h are cle a rly d e fin e d . M o re o ve r, th e y de fine the d iffe re n t dem on strative s as closer to o r fu rth e r fro m the speaker. T h is ch a ra cte riza tio n is in s u ffic ie n t as H a n ks (1992) and Strauss (1993) have sh o w n fo r E nglish. F o r instance, in Strauss' data there are cases o f self re p a ir w here the speaker is n o t sure w h ic h de m on strative to use1. The tra d itio n a l n o tio n o f d e ixis sh o u ld be extended (cf. K irsn e r, 1979; G arcia, 1975; and P iw e k e t al., 1996). F in a lly , a ll these sh o rtco m in g s are re la te d to the fa ct th a t m o st o f these stu d ie s are based o n n a tiv e speaker in tu itio n s , and, in general, the da ta are v e ry lim ite d (C ia p u scio , 1988 and B la ckw e ll, 1998 are exceptions). 2.2. U niqueness The re q u ire m e n t o f uniqueness fo r d e fin ite re fe re n tia l expressions fo llo w s a lo n g tra d itio n . A c c o rd in g to R ussell's (1905) an alysis, the fo llo w in g d e fin ite d e s c rip tio n : (2a) The professor is sober. m akes three claim s: (a) existence: there is a p rofesso r (b) uniqueness: the re is o n ly one professor (c) p re d ica tio n : th is in d iv id u a l is sober. 1 Consider the following example of self-repair (Strauss, 1993: 406) which cannot be explained following the traditional deictic explanation: "people" is referred to with both the proximal and distal demonstrative. If the German judges chose tuh try:: thee (these), those people, hh yuh see, W- you gotta remember something- There is a funny thing about legality Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 I f the sentence is true, the n (a)-(c) w ill be tru e , and hence (2a) e n ta ils the co n ju n ctio n o f (a)-(c). The im p o rta n t d is tin c tio n betw een (2a) and the corresp ond ing in d e fin ite d e scrip tio n (2b) lies in the uniqueness claim . (2b) A professor is sober. The tru th o f sentence (2b) re quires th a t there sh o u ld be a t least one professor w h o is sober. It is lo g ic a lly co m p a tib le w ith there being m ore th a n one professor, o r w ith o n ly one professo r, and so (2b) is n e u tra l to uniqueness and does n o t c o n tra d ic t it. The n o tio n o f uniqueness is developed in H a w k in s (1991) w he re he explains n u m e ro u s exam ples o f d e fin ite a rtic le usage in te rm o f it. In exam ples such as, A president has resigned vs. The president has resigned., the firs t exam ple m ig h t re fe r to a u n iq u e e n tity , w h ile the la tte r im p lie s u n iq u e n e ss. 2.2.1. C ritic is m s o f the u n iq u e n e ss re q u ire m e n t The re q u ire m e n t o f uniqueness has been c ritic iz e d . F irst, the uniqueness re q u ire m e n t does n o t h o ld in absolute term s (section 2.2.2.). Second, in some cases, the n o tio n o f uniqueness is a s u ffic ie n t b u t n o t a necessary c o n d itio n fo r definiteness (section 2.2.3). F in a lly , som etim es, uniqueness seems to be n e ith e r a necessary n o r a s u ffic ie n t c o n d itio n because the definiteness typ e refers to a n o n -u n iq u e e n tity (i.e. the re are com peting referents). It w ill be argued th a t these la tte r cases can be reconciled w ith a th e o ry o f uniqueness since the crite ria n o w in c lu d e the n o tio n o f pro m in e n ce (cf. ch a p te r 5). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 2.2.2 T h e n o tio n o f uniqueness is n o t ab solute The re q u ire m e n t o f uniqueness is n o t absolute: it h o ld s w ith in the d o m a in o f a pra g m a tica lly d e lim ite d set o f en tities (P-set). As H a w k in s (1991: 407) asserts, w h e n re fe rrin g to the professor, w e are n o t re fe rrin g to the o n ly professor in the w o rld , i.e. the unique p ro fe sso r in absolute term s. K adm on (1990), s im ila rly , su p p o rts a less absolute n o tio n o f uniqueness w hereby d e fin ite d e scrip tio n s are 'u n iq u e in some w a y ' (K adm on, 1990: 282). B ut h e r p ro p o sa l is too vague and as H a w kin s (1991: 407) puts it: "m e re ly encourages the a lte rn a tive v ie w th a t uniqueness m ay not, a fte r a ll, be an in v a ria n t p ro p e rty o f d e fin ite d e s c rip tio n s ". K adm on (1990) inco rpo rate s the n o tio n o f uniqueness in to the discourse representation th e o ry o f K am p (1981) in w h ic h p ra g m a tic a lly su p p lie d m a te ria l achieves the uniqueness effect. In the process, h e r analysis accounts fo r a lo t o f sub tle data in v o lv in g b o u n d anaphors and donkey sentences fo r w h ic h H e im (1982) gave a n o n uniqueness account. But it is n o t clear e xa ctly w h a t the p ragm atic p rin c ip le s are th a t achieve the uniqueness e ffe ct and how th e y operate in K adm on's (1990) fra m ew o rk. H a w k in s (1991) makes K adm on's in tu itio n e x p lic it. H e states th a t the uniqueness requirem ent does n o t h o ld in absolute term s. Instead, g iv e n the d e fin ite expression the professor, w e are re fe rrin g to the u n iq u e p rofesso r w ith in a set o f e n titie s d e lim ite d b y the speaker w h ic h is, in tu rn , m u tu a lly m a n ife st to the lis te n e r (Sperber & W ils o n , 1986). The n o tio n o f a P-set is s till n o t as p re cise ly defined as one m ig h t lik e . E pste in (1994) argues th a t it can lead to m u ltip le ad-hoc Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 in te rp re ta tio n s. A n d i f the n o tio n o f a P-set is n o t re s tric te d in some w a y, then the n o tio n o f uniqueness lacks s u p p o rt because i t is dependent on the d e lim ita tio n o f a P-set. 2.2.3. U niqu en ess is a s u ffic ie n t b u t n o t a necessary c o n d itio n fo r d e fin ite n e ss types. B im e r and W a rd (1994) have claim ed th a t th e uniqueness re q u ire m e n t is a s u ffic ie n t, b u t n o t a necessary, c o n d itio n fo r d e fin ite N Ps in E nglish. F o r instance, an exam ple such as Pass me the salt shaker m ay be u tte re d in th e presence o f m ore th a n one s a lt shaker, i.e. w h e n the e n tity re fe rre d to is n o t unique. A n o th e r e xa m ple is I took the bus, w here b o th the speaker and the lis te n e r k n o w th a t there are num erous buses th a t c o u ld s a tis fy the d e fin ite d e s c rip tio n . H o w e ve r, in b o th exam ples, it is im p o rta n t to note th a t it does n o t m a tte r w h ic h sa lt shaker the lis te n e r p icks o r w h ic h bus the speaker to o k. In such cases, B im e r and W a rd (1994) assert, the e n tity re fe rre d to is n o t re le v a n tly d iffe re n tia b le fro m a n y o th e r e n tity o f the som e k in d , i.e. the e n titie s w ith in the re le va n t P-set are n o t in d iv id u a te d in any s ig n ific a n t w a y fo r e ith e r th e speaker o r the lis te n e r. T h is k in d o f e xp la n a tio n also accounts fo r L e o n e tti s exam ples: (3) Puso la mano sobre la mesa. 'H e /s h e p u t h is /h e r hand on the ta b le .' (4) Cuando llegamos, ella estaba en la ventana. 'W hen w e a rriv e d , she was a t the w in d o w '. (5) Beso a su madre en la mejilla. 'H e /s h e kissed h is /h e r m o th e r on h e r cheek.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 In each case, th e re is m ore th a n one e n tity th a t m ig h t be re fe rre d to b y the d e fin ite a rtic le (tw o hands, m ore than one w in d o w , and tw o cheeks). H o w e v e r, w h ic h h a n d h e /s h e p u t on the ta b le , w h ic h w in d o w she w as at, and w h ic h cheek sh e /h e kissed is irre le v a n t to the purposes o f co m m u n ic a tio n . 2.3. T h e n o tio n o f a P-set H a w k in s ' (1978) m a in c o n trib u tio n to the th e o ry o f d e fin ite n e ss is to p ro v id e p ra g m a tic c o n d itio n s fo r the in te rp re ta tio n o f the uniqueness re q u ire m e n t: it h o ld s o n ly w ith in the lim its o f a p ra g m a tic a lly d e lim ite d set o f e n titie s (P-sets), and the n o tio n o f a P-set is the n fu rth e r ela borated in h is 1991 article . In 1978, he c a re fu lly describes a ll the m ajor usages fo r the d e fin ite a rtic le in E n g lish and he dem onstrates h o w th e uniqueness re q u ire m e n t h o ld s w ith in the lim its o f a determ ined P-set. T h ere fore , fo r each usage o f the D e fin ite A rtic le , one type o f P-set is d e lim ite d . The speaker activates a P-set th a t is m u tu a lly m a nifest to the lis te n e r b y re fe rrin g u n iq u e ly to the re le va n t e n tity in th a t set. The P-sets can be co n stitu te d by: a) the p re v io u s discourse; b) the im m e d ia te s itu a tio n o f utterance; c) a la rg e r o r sp e cific s itu a tio n th a t p ro p e rly co n tains the s itu a tio n o f u ttera nce; d ) associative re la tio n s h ip s , and e) e s ta b lis h in g re la tive s o r genitives. The p re v io u s d isco u rse set is con stitute d b y th e e n titie s th a t w ere m e ntion ed in th e p re v io u s text. F o r instance in (6), an exam ple fro m Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 m y corpus, la miijer 'th e w o m a n ' refers u n iq u e ly to th e w o m a n {una mujer, 'a w o m a n ,’ lin e 1) w h o w as m entioned in th e p re vio u s discourse set. (6) nosotros teniamos una m ujer cjm e trabajaba con nosotros que ahora no trabaja mas (2 IU omitted) y la m ujer era m uy conocida we had a w o m a n w h o used to w o rk w ith us b u t she d o e sn 't w o rk anym ore (2 IUs o m itte d ) and the w o m a n w as v e ry w e ll kn o w n The im m e d ia te s itu a tio n o f uttera nce set is co n stitu te d b y those e n titie s tha t can a ctu a lly be seen b y b o th the speaker an d liste n e r. For instance in Pass me the bucket, the b u cke t is present and can be seen b y b o th the speaker and the hearer. The la rg e r o r s p e c ific s itu a tio n set is co n stitu te d b y those e n tities th a t can be id e n tifie d b y the speaker and hearer b y v irtu e o f th e ir com m on m em bership in a la rg e r c o m m u n ity (la rg e r s itu a tio n : The president resigned, w he n w e are ta lk in g about the o n ly p re s id e n t o f the c o u n try the speaker and the lis te n e r share) o r b y v irtu e o f th e ir com m on m em bership in a s m a ll c o m m u n ity (fo r instance, in a fa m ily w e can say the dog if there is o n ly one dog in the house). The associative re la tio n s h ip set is co n stitu te d b y e n titie s th a t are re la ted by associations d e fin e d b y encyclopedic kno w le d g e o r fram es (M in s k y , 1975). For instance, in the fo llo w in g exam ple, I bought a book. The author is very fam ous, w e re fe r u n iq u e ly to the author Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 because w e k n o w th a t in general, stereo ty p ic a lly , a b o o k has a u n iq u e a u th o r. The e s ta b lis h in g re la tiv e o r g e n itiv e P-set is co n stitu te d b y the e n titie s in fe rre d fro m the in fo rm a tio n p ro v id e d w ith in the N P its e lf. For instance, it is possible to sta rt an utterance w ith , the front page of the newspaper w here the a p p ro p ria te P-set d e lim ite d fo r the uniqueness re q u ire m e n t is established w ith in the g e n itiv e of the newspaper g ive n th a t in general, people o w n one house. R elative clauses can also fu n c tio n in th is w ay. F o r instance in Pass me the bucket which is over there the re la tiv e which is over there gives us the fo llo w in g in fo rm a tio n : pass m e the o n ly b u cket th a t is located o ve r there, regardless o f those located elsewhere. T his la s t usage dem onstrates th a t fa m ilia rity o f the e n tity (H e im , 1982) is n o t a necessary c o n d itio n fo r the usage o f the d e fin ite a rtic le . P-sets are d e lim ite d sub jectively b y speakers, a n d E pstein (1994) p o in ts o u t th a t th is vague d e fin itio n can lead to in n u m e ra b le ad-hoc pra g m a tic d e lim ita tio n s th a t su b ve rt the v e ry con cept o f u n iq u e reference. I t is, thus, necessary to specify the param eters b y w h ic h P-sets are d e fin e d . For instance, it is necessary to c la rify a n d re fine the n o tio n o f p re vio u s discourse set. W h ich un ique e n tity a g iv e n definiteness typ e refers to in the p re vio u s discourse set w ill de pen d on the discourse stru ctu re , i.e. on the w a y the m ental repre senta tion o f the te xt is constructed in the m in d o f b o th the speaker and he are r. The fo llo w in g section deals w ith p re vio u s approaches to disco urse stru ctu re . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 2.4. D isco u rse S tru ctu re There is agreem ent in th e lite ra tu re on discourse s tru c tu re on the fo llo w in g p o in ts: firs t, texts are in te rp re te d b y b u ild in g a m e ntal re p re se n ta tio n o f them (G em sbacher, 1990; G em sbacher e t al., 1995; G ivo n , 1995). Second, th is m e n ta l re pre senta tion consists o f discourse u n its w h ic h are h ie ra rc h ic a lly o rg a n ize d (M ann & T h om p son, 1988; G ivo n , 1992; Fox, 1987). D iscourse u n its are d e fin a b le p o rtio n s o f a te xt, and th e y are som etim es d e lim ite d b y lin g u is tic s in d ic a to rs . These discourse u n its have in trin s ic p ro p e rtie s , fo r instance v e ry o fte n th e y have the sam e subject, tense an d aspect form s. F in a lly , definiteness typ e s re fle c t the h ie ra rc h y o r 'n a tu re ' o f the m e n ta l re p re se n ta tio n (Fox, 1987; M a n n & T hom pson, 1988: 270; S idner, 1983: 273). F or instance, spe cific definiteness types are used at the b e g in n in g o f the discourse u n it (V o n k et al., 1992). A s Fox (1987) states: a n y tre a tm e n t o f anaphora m u st seek its u n d e rs ta n d in g in the h ie ra rc h ic a l s tru ctu re o f the te xt-typ e b e in g used as a source o f data. Texts m a y be produ ced and h e a rd /re a d in a lin e a r fa sh io n , b u t th e y are designed and u n d e rsto o d h ie ra rc h ic a lly , and th is fa ct has d ra m a tic consequences fo r the lin g u is tic co d in g e m p lo ye d . G em sbacher (1990) presents experim ents in w h ic h it is clear th a t a change in th e scenery (p a rtic ip a n ts and events) has consequences fo r the discourse re p re se n ta tio n in the m in d s o f the p a rtic ip a n ts and, th e re fo re , fo r th e lin g u is tic fo rm s used to represent these discourse e n titie s . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 2.4.1. T he D isco u rse U n it D iffe re n t term s th a t have been assigned to u n its o f discourse in clu d e : 'n o d e s' (G ivo n , 1992), 'fie ld o f reference' (G rim es, 1975), 'e p iso d e ' (V an D ijk , 1982), 'p a ra g ra p h ' (H in d s, 1979), 'c o n te x t space' (R eichm an, 1978), 'd o m a in o f reference' (S anford a n d G a rro d , 1981:118; C o rn ish , 1986), 'd o m a in o f in te rp re ta tio n ' (S ten ning , 1978), 'focus space' (G rosz, 1981), scenery (V o n k at al., 1992). This m u ltip lic ity o f term s show s us h o w p ro lific and, a t the same tim e , h o w c o n fu s in g the fie ld o f discourse pragm atics is The n o tio n o f a discourse u n it is n o t a lw a ys c le a rly d e fin e d in these approaches. Q uestions th a t need to be addressed are: w h a t co n stitu te s a discourse u n it? W hat m akes one d isco u rse u n it d iffe re n t fro m a n oth er and h o w can w e recognize w he n one ends and w hen a n oth er starts? 2.4.2. L in g u is tic In d ic a to rs o f D iscourse U n its It has been sh o w n th a t discourse u n its are g e n e ra lly fram ed b y lin g u is tic in d ic a to rs . F o r instance, som e discourse m a rke rs (S ch iffrin , 1987) are used to in itia te discourse u n its (see, fo r insta nce B rizuela, 1992 and B rizu e la e t al. 1999). H esita tions (C hafe, 1994: 138) are claim ed to be used b y the speaker to g a in tim e w h ile processing the change o f discourse u n its. P o la n yi e t al. (1983: 265) assert th a t a n e w discourse u n it is m a rke d , in general, p h o n o lo g ic a lly , s y n ta c tic a lly and se m a n tica lly. F o r instance, well is a m a rke r o f in itia liz a tio n o f a u n it, ok is a m a rke r o f c o m p le tio n and so is a m a rke r o f re s u m p tio n . M eta com m ents, such as Where was I?, Sorry to interrupt but..., are also sh ifte rs. H o w e ve r, note th a t, as P olan yi (1994) and C o rn is h (1986) state, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 these in d ica to rs are n o t alw ays re q u ire d ; n o r do th e y occur in e ve ry change o f discourse u n it. W h ich in d ic a to rs are used to d e lim it discourse units? H o w do th e y achieve th e ir effect? A m ore d e ta ile d analysis on th is to p ic, based on n a tu ra l data, is needed. 2.4.3. T he P rop ertie s o f D isco u rse U n its O ne p ro p e rty o f discourse u n its is th a t the y are se m a n tica lly coherent, i.e. the y, are 'about' the same them e. V an D ijk (1982) fo r instance, defines 'e p iso d e ' as a coherent sequence o f p ro p o s itio n s th a t denotes facts th a t are interconnected in a possible w o rld . C o rn ish (1986) claim s th a t discourse u n its s h o u ld have th e m a tic u n ity ; the same p a rticip a n ts, tense, lo ca tio n , and actions. H in d s (1979: 136) states th a t a 'p a ra g ra p h ' (discourse u n it) is 'a u n it o f speech o r w ritin g th a t m a in ta in s a u n ifo rm o rie n ta tio n (s p a tia lly , te m p o ra lly , th e m a tic a lly o r in term s o f p a rtic ip a n ts ).' R eichm an (1978) asserts th a t 'a g ro u p o f utterances re fe rrin g to a single episode o r issue form s the basis fo r a con text space.' I t is n o t clear w h a t is it to be ’about' the same them e. Is it to ta lk ab out one e n tity o r is it to ta lk about one aspect o f an en tity? W e can th in k o f a fifte e n m in u te co n ve rsa tio n a l te x t b e in g a b o u t one them e, b u t it is clear th a t th a t te x t w ill consist o f m ore th a n one discourse u n it. A n o th e r p ro p e rty o f discourse u n its is sh a rin g the sam e subject, tense, and aspect form s. Paredes S ilva (1993: 43) states th a t discourse u n its w ith the same subject, tense, and aspect form s are v e ry w e ll connected. The connectedness o f a discourse u n it is re flected lin g u is tic a lly in P ortuguese, b y the presence o r absence o f p ro n o u n s. The m ore connected the discourse u n it, the m ore lik e ly th a t e n titie s Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 are re fe rre d to w ith o u t lin g u is tic m a te ria l (zero) ra th e r th a n u s in g a p ro n o u n . A change in an y o f these fo rm a l devices: subject, tense o r aspect fo rm s m ay re s u lt in a change o f discourse u n it (see also B rizuela, 1997). B u t e xa ctly w h ic h factors are re le van t (subject, tense, o r aspect) in d e fin in g a discourse u n it? W h ich o f these are m o re im p o rta n t? A n d , w h ic h ones co n stitu te necessary o r s u ffic ie n t c o n d itio n s fo r the connectedness o f a u n it? 2.5. T he P roce dural M e a n in g o f D e fin ite n e s s T ypes In the fo llo w in g section, I present approaches in w h ic h definiteness types s u p p ly p ro ce d u ra l m eaning. T h e y are the speaker's in s tru c tio n s to the lis te n e r to fin d the re fe re n t fo r th e in te rp re ta tio n o f the N P . D iffe re n t theories h o ld th is vie w : fo r instance, a cce ssib ility th e o ry (A rie l, 1990), a c tiv a tio n th e o ry (Chafe, 1976, 1994) and oth e r pra g m a tic approaches as w e ll such as G u n d e l et a l., 1993. 2.5.1. P rocedural M e a n in g A rie l claim s th a t w h e n speakers use a re fe rrin g expression they are g iv in g p ro c e d u ra l in fo rm a tio n about w he re to fin d in fo rm a tio n (o f any typ e , a n a p h o ric m a tch in g , ph ysical re fe re n t m a tch in g , b rid g e inferences) in o rd e r to fu lly understand the N P . T he n o tio n o f p ro ce d u ra l m eaning is also used b y Sperber & W ils o n (1986) w h o d is tin g u is h re p re se n ta tio n a l fro m p ro ce d u ra l m e anin g. R epresentational m e anin g is th e m eaning o f co n te n t w o rd s and th e ir le xica l and g ra m m a tica l m eaning. D eterm iners, p ro n o u n s and discourse p a rticle s th a t in s tru c t the hearer to m ake an inference in re la tio n to dive rse th in g s have p ro c e d u ra l m e anin g. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 A n o th e r scholar w h o considers the p ro c e d u ra l m eaning o f d e fin ite expressions is G iv o n (1992). G iv o n asserts th a t the g ra m m a r o f re fe re n tia l coherence is n o t p rim a rily a b o u t reference. R ather it is ab o u t id e n tify in g and accessing th e lo ca tio n s ('m e n ta l file s ’, 'storage nodes') in episodic m e m o ry. N o m in a l re fere nts serve as file lab els, and th e y activa te the storage nodes in w h ic h in c o m in g in fo rm a tio n is stored. A rie l (1990), G iv o n (1992) and S perber & W ils o n (1986) agree in asserting th a t definiteness types are in s tru c tio n s to the hearer a b o u t w he re to lo o k fo r the re fe re n t, and th e re fo re th e y have p ro c e d u ra l m e a n in g . 2.5.2. D e fin ite n e ss Types s ig n a l le ve ls o f a c c e s s ib ility The p ro ce d u ra l m e a n in g o f re fe re n tia l expressions in v o lv e s in s tru c tin g th e hearer to lo o k fo r the in fo rm a tio n needed in th e rig h t place, i.e. w ith in m ore o r less accessible contexts. Theories o f a cce ssib ility and a ctiva tio n are presented b e lo w . Several scholars (A rie l, 1990; Chafe, 1976, 1984; G u n d e l e t al., 1993 in te r alia) have cla im e d th a t the selection o f d e fin ite n e ss types depends on the le v e l o f a cce ssib ility o r a ctiva tio n o f the e n tity /c o n te x ts re fe rre d to. The d e fin itio n o f acce ssibility o r a c tiv a tio n va rie s fro m one scholar to the next, b u t the basic in s ig h t is the same: d e finiten ess types are m a rke rs o f accessibility. 2.5.2.I. The a c c e s s ib ility th e o ry A rie l's (1988, 1990) h yp o th e sis is th a t n a tu ra l languages p rim a rily p ro v id e speakers w ith the m eans to code the a cce ssib ility o f the re fe re n t to the addressee. F o llo w in g A rie l, a cce ssib ility is th e Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 d e te rm in in g p rin c ip le th a t accounts fo r the choice o f d iffe re n t re fe re n tia l form s. She proposes a h ie ra rc h y o f accessibility (presented fro m lo w to h ig h accessibility) (a) F u ll nam e + m o d ifie r (b) F u ll nam e (c) L o n g d e fin ite d e s c rip tio n (d) S hort d e fin ite d e s c rip tio n (e) Last nam e (f) F irs t nam e (g) D is ta l d e m o n stra tive + NP + m o d ifie r (h) P ro xim a l d e m o n stra tive + N P + m o d ifie r (i) D is ta l d e m o n stra tive + NP O ') P ro x im a l d e m o n s tra tiv e + N P (k) Stressed p ro n o u n + gesture (1) Stressed p ro n o u n (m ) unstressed p ro n o u n (n) c litic iz e d p ro n o u n (o) agreem ent, re fle xive s, gaps, p ro E very category lis te d above is considered a m a rke r o f som e leve l o f accessibility. L o w m a rkers are those u ttere d on the basis o f general kn o w le d g e in u n m a rke d in itia l re trie va ls. M id d le a cce ssib ility m arkers p o in t to the ph ysica l s itu a tio n . F in a lly , h ig h a cce ssib ility m a rkers refer to sh o rt distance and s a lie n t referents. The evidence she presents is based on the distance betw een the re fe rrin g expression and th e antecedent needed fo r its in te rp re ta tio n (p ron oun s: s h o rt distance; dem onstratives: m e d ia l distance; d e fin ite a rticle : lo n g distance). T h is h ie ra rc h y is u n ive rsa l, even th o u g h d iffe re n t languages have d iffe re n t num bers o f a cce ssib ility m arkers, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 and th e re fo re the accessibility range and set o f usages fo r each m a rke r depends o n the num ber o f o th e r co m p e tin g fo rm s (A rie l, 1990: 83). 2.5.2.2. T h e a c tiv a tio n th e o ry C hafe (1976,1994) states th a t language is dependent on the speaker’s b e lie fs about a ctiva tio n states in the hearer's m in d . H e considers three a ctiva tio n states o f in fo rm a tio n : a ctive , se m i-a ctive , and in a c tiv e . These states are re la te d to s h o rt and lon g term m em ory. A c tiv e in fo rm a tio n is located in s h o rt m e m o ry, sem i-active in fo rm a tio n in lo n g te rm m e m o ry an d in a c tiv e in fo rm a tio n is co m p le te ly n e w fo r the hearer. There m a y be m ore a ctiva tio n states tha n these, and the boundaries betw een the m m ay be less categorical tha n th is d iv is io n suggests (69). In th is respect h is the ory can be com pared to A rie l’s, w h ic h also presents the dim ension o f a cce ssib ility as a c o n tin u u m w ith o u t discrete states. Speakers' beliefs ab out the hearer's m in d have various sources: p re v io u s lin g u is tic in te ra ctio n - th in g s said w ith in the same discourse, o r rem em bered fro m p re vio u s con versa tions. O thers are d e rive d fro m no n lin g u is tic in te ra ctio n (fo r insta nce the speech situ a tio n ), fro m shared experiences, and fro m shared cu ltu re s. The n u m b e r o f d iffe re n t re fere nts th a t can be active at the same tim e is v e ry sm a ll, and any re fe re n t, unless it is repeated, w ill q u ic k ly cease b e in g active. F o llo w in g C hafe, (1994: 39) a t any given m om ent the m in d can focus on no m ore th a n a sm a ll segm ent o f e ve ryth in g it kno w s. Consciousness is a v e ry lim ite d a c tiv a tio n process. H e com pares i t to visio n , w h ic h can o n ly focus o n a lim ite d am ount o f in fo rm a tio n a t one tim e and is in co n sta n t m o tio n . This lim ite d Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 capacity o f consciousness is re fle cte d lin g u is tic a lly in the b rie f spurts o f language th a t he calls in to n a tio n u n its (cf. cha pte r 3). R egarding de finiten ess types, C hafe claim s th a t accented p ro n o u n s express givenness (ideas a lre a d y active in the consciousness o f the addressee) and the d e fin ite a rtic le refers to id e n tifia b le e n titie s. The n o tio n o f id e n tifia b ility (C hafe, 1994: 93) is in d e p e n d e n t o f a c tiv a tio n cost. I t has three com ponents: a) it is assum ed to be a lre a d y shared d ire c tly o r in d ire c tly b y the lis te n e r b) it is ve rb a lize d in a s u ffic ie n tly id e n tify in g w a y c) it is c o n te x tu a lly sa lie n t. A lth o u g h the q u e stio n o f w h e th e r a re fe re n t is k n o w n o r u n k n o w n to the lis te n e r is irre le v a n t to a c tiv a tio n cost it is cru cia l to id e n tifia b ility . 2 Chafe's approach can be re la te d to A rie l's general c la im th a t definiteness types are m a rkers o f accessibility. In C hafe's v ie w , language reflects the fact th a t e n titie s are m ore o r less a ctiva te d in o u r consciousness. 2.5.2.3. Necessary and sufficient conditions for definiteness types F o llo w in g the same lin e o f reasoning, G u n d e l e t al. (1993: 275) assume th a t the fo rm o f re fe rrin g expressions depends o n the assumed c o g n itiv e statuses o f the re fe re n t, i.e. on assum ptions th a t a cooperative 2Hawkins (1978) has show n that identifiability (or familiarity in Heims' terms) is neither a necessary nor a sufficient condition for the usage of the definite article. For instance in the follow ing example: W hat's wrong with Bill? Oh the woman he went out with last night was nasty to him, the underlined NP could be a perfectly appropriate first-mention definite description (cf. also section 2.3). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 speaker can reasonably m ake re g a rd in g th e addressee's kno w le dg e and a tte n tio n state in the p a rtic u la r co n te xt in w h ic h the expression is used: b o th the speaker and hearer k n o w th a t d iffe re n t d e te rm in e rs and p ro n o m in a l fo rm s c o n v e n tio n a lly sig n a l d iffe re n t c o g n itiv e statuses (in fo rm a tio n about lo c a tio n in m e m o ry and a tte n tio n state) th e re b y e n a b lin g the addressee to re s tric t the set o f possible referents. (G u nde l e t a l., 1993: 275) T h e y propose a h ie ra rc h y o f assum ed c o g n itive statuses w ith th e ir re spective lin g u is tic s fo rm s. O n ly the re le v a n t lin g u is tic form s are presented fo r E n g lish and Spanish. In focus A ctivated F am iliar U n iq u ely Id e n tifia b le R eferential Type Id en tifiab le English i t HE, this, that, this N that N the N indefinite this N a N Spanish zero el 'he' EL este, ese, aquel este N ese N aquel N el N zero N un N Each status on the h ie ra rc h y is a necessary and s u ffic ie n t c o n d itio n fo r the a p p ro p ria te use o f a d iffe re n t fo rm . They assert th a t in u s in g a p a rtic u la r fo rm speakers assum e th a t the associated co g n itiv e status is m e t and since each status e n ta ils a ll lo w e r statuses, th e y also sig n a l th a t a ll lo w e r statuses (on the rig h t) have been m et (G u n d e l e t al., 1993: 275). T herefore, fo r instance, a lth o u g h the c o n v e n tio n a l m e anin g o f the in d e fin ite a rtic le in E nglish is th a t the re fe re n t is a t least typ e id e n tifia b le , it is possible to fin d usages in w h ic h the in d e fin ite a rtic le m eans u n iq u e ly id e n tifia b ility . H ow eve r, th e ir data an a lysis show s th a t th is scenario is v e ry in fre q u e n t. They e x p la in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 this phenom enon b y c la im in g th a t, fo r instance, the use o f the in d e fin ite a rtic le c o n v e rs a tio n a lly im p lica te s the n e g a tio n o f the stronger c o g n itiv e a ssu m p tio n (b y Q -im p lic a tu re l 3 ), in th is case, if the e n tity re fe rre d to is in te rp re te d as typ e id e n tifia b le , then it is n o t u n iq u e ly id e n tifia b le . A n d th a t is w h y reference to u n iq u e ly id e n tifia b le e n titie s is n o t u s u a lly m ade u sin g the in d e fin ite a rtic le . B ut, th is im p lic a tio n o n ly e xp la in s preferences fo r usages. A n d since a conversational im p lic a tu re is n o t a necessary inference, it is possible th a t in som e contexts th e in d e fin ite a rticle does re fe r to a u n iq u e ly id e n tifia b le e n tity . 4 The p ro x im a l d e te rm in e r d e m o n stra tive {this N , este N ) signals an activa ted e n tity , i.e. loca ted in the cu rre n t sh o rt-te rm m e m o ry. The d is ta l d e te rm in e r d e m o n stra tive (th a t N , ese N ) signals a fa m ilia r e n tity. In other w o rd s, th e addressee is able to u n iq u e ly id e n tify the intended re fe re n t because he a lre a d y has a representation o f it in m em ory (e ith e r lo n g o r s h o rt te rm m em ory). F in a lly , the d e fin ite a rticle (the N , e l N ) sign als a u n iq u e ly id e n tifia b le e n tity (the addressee can id e n tify the speaker's in te n d e d referent on the basis o f the n o m in a l alone). The p ro b le m w ith th is th e o ry is th a t the co g n itive statuses th a t G undel et al. propose to d e fin e re fe re n tia l expressions are s tip u la te d . 3 Q-implicature 1 involves the negation of a stronger proposition (Levinson, 1987a). 4 For instance they present the follow ing example: The senate on August 10 voted 50 to 48 to spend $736,400 fo r a third Senate Gymnasium due to be built in T H E /A 137.7 million-dollar host senate office building opening in January. Since the descriptive information necessary to identify the referent of the underlined NP is encoded in the nominal itself, an explicit marker of identifiability is unnecessary and an indefinite article is equally appropriate there to refer to a uniquely identifiable entity. (Gundel et al. 1993: 296) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 The o n ly evidence the y g ive fo r the existence and re a lity o f the separate co g n itive statuses is the m eaning o f the co rre sp o n d in g lin g u is tic fo rm . The c irc u la rity o f th e ir argum ent is clear: th e y have n o t g ive n e xte rn a l evidence fo r these no tio n s. A lso , the u n iv e rs a lity o f these statuses is questionable, esp ecially since the y are s tip u la te d . M o re o ve r, the assum ption th a t E n g lish is the 'perfect' language, m e anin g th a t it is the o n ly language w h ic h discrim inates e ve ry sing le status, is ra th e r susp icio us. T h at the re is a co rre la tio n betw een c o g n itive status and lin g u is tic fo rm is v a lid and is related to A rie l's and C hafe's insights. The co n tro ve rsy resides in d e te rm in in g a) the re le v a n t co g n itive statuses and th e ir na ture , and b) th e ir re la tio n s h ip to each other. 2.5.3. E vidence and m o tiv a tio n fo r the th e o rie s o f p ro ce d u ra l m e a n in g The evidence fo r these theories varie s. F o r instance, A rie l (1990) presents evidence fo r her accessibility h ie ra rch y based on a te xtual data analysis. She asserts th a t e n titie s closer to the an aph or are m ore accessible th a n e n titie s fa rth e r fro m the anaphor. It is clear th a t th is evidence is also u se fu l fo r anaphoric usages. B u t, w h a t happens w ith d e ictic usages? G un de l e t al. (1993) also present a data analysis in 5 d iffe re n t languages (C hinese, Russian, E nglish, S panish an d Japanese). The p ro b le m w ith th is analysis, as show n above, is th e n a tu re o f the statuses. F in a lly , C hafe’s analysis is based on q u a lita tiv e data analysis o f discourses. The vagueness o f the d e fin itio n o f th e term s a ctiva te d , Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 se m i-a c tiv a te d and in a c tiv a te d m ake it d iffic u lt to v e rify his statem ents. M o tiv a tio n also varies w ith in these theories. G u n d e l e t al. (1993) and C hafe (1976, 1994) assum e a co m p a rtm e n ta l v ie w o f m em ory. R eferents are located in d iffe re n t com partm ents (i.e. s h o rt o r lo n g te rm m e m o ry ). A rie l (1990) asserts th a t he r th e o ry o f acce ssibility re quires a n o n co m p a rtm e n ta lize d s tru ctu re o f the m e m o ry. C ontexts are ra n ke d according to degree o f accessibility, a n d language reflects th a t w ith m arks o f h ig h o r lo w acce ssibility. M e m o ry theories are s h iftin g in th is d ire ctio n . A s she claim s: A fu rth e r c ru c ia l d e ve lo p m e n t in th e v ie w o f m e m ory s tru c tu re seems ra th e r s im ila r to the changed concept o f co n te xt w e have argued fo r. Ju st as con text cannot p ro b a b ly be d iv id e d g e o g ra p h ica lly if it to account fo r lin g u is tic b e h a vio r, so it seems th a t a t least som e researchers are n o w casting d o u b t on the 'geographic' m a p p in g o f va rio u s m em ory stores. (13) She asserts th a t one o f the th e o rie s th a t sup ports th is v ie w is th a t o f the P a ra lle l D is trib u te d Processing (M cC le lla n d and R u m e lh a rt, 1986a). T hey argue th a t m em ory consists o f m o d u la r u n its re ce ivin g and se n d in g o u t sign als to other su ch m odules. A n d she goes on: The advantage o f P arallel D is trib u te d Processing m odels fo r A c c e s s ib ility T heory is th a t an a b so lu te ly rig id m e m o ry d iv is io n is n o t re q u ire d an y m o re . I believe th a t degrees o f A c c e s s ib ility correspond to d iffe re n t statuses w ith in m e m o ry, w h e re m a te ria l m a y be e xtre m e ly sa lie n t, i.e. h ig h ly a ctiva te d , or, vice versa u n a ctiva te d o r fa d in g aw a y (16) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 The v ie w o f M c C le lla n d and R um elhart, m ay be re la te d to d iffe re n t approaches (fo r instance C ow an, 1993 and M e D o n a ld e t al, in press) th a t su p p o rt the sam e lin e o f reasoning: m e m ory is n o t co m p a rtm e n ta l (d iv id e d in s h o rt vs. lo n g te rm m e m ory). R ather, it consists o f nodes w ith m o re o r less a ctiva tio n . So-called s h o rt o r w o rk in g m em ory consists o f those nodes th a t surpass a ce rta in th re sh o ld in a c tiv a tio n le v e l. Therefore, the stru ctu re o f m e m o ry as p o stu la te d b y P ara lle l D is trib u te d Processing explains w h y re fe re n tia l devices are m arkers o f a cce ssib ility. 2.5.4. S um m ary o f 2.5.1,2.5.2., 2.5.3. E ven th o u g h a ll th e the orie s sum m arized above (2.5.1, 2.5.2., 2.5.3.) cla im th a t the se le ctio n o f d e fin ite expressions is re la te d to the state o f a c tiv a tio n o f the re fe re n t, th e y d iffe r in some aspects. F irs t, the taxonom ies used b y th e d iffe re n t theories are d iffe re n t. Second, w h ile G u n d e l et al., (1993) propo se discrete statuses in an im p lic a tio n a l re la tio n sh ip , A rie l (1990) and C hafe (1994), consider the states to be con tinuo us. R egarding A rie l (1990) and Chafe (1994), C hafe m akes the p o in t th a t the speaker m akes assum ptions ab out the lis te n e r state o f m in d . A rie l's a cce ssib ility, in con trast, seems to be m o re in d e p e n d e n t o f b o th the speaker and liste n e r. H e r levels o f a cce ssib ility are m u tu a lly m a nifest to b o th the speaker and liste n e r (cf. S perber & W ilso n , 1986) and hence, th e y seem to be m ore an 'o b je ctive ' p ro p e rty o f the e n titie s to w h ic h th e speaker and liste n e r have access. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 2.5.5. Factors c o n trib u tin g to a cce ssib ility D iffe re n t factors have been taken in to account in the p re v io u s lite ra tu re fo r the d e fin itio n o f acce ssibility. In th is section, I presen t the m ost im p o rta n t ones: distance, to p ic a lity , frequency, e p iso d ic stru ctu re , lin g u is tic size, syn ta ctic p o s itio n , the m atic ro le , d e ixis, and in te ra c tio n purposes. 2.5.5.1. D istance The firs t, and m o st e xte n sive ly stu d ie d fa cto r, is d ista n ce fro m the anaphor. I t is w id e ly accepted th a t anaphors w ith little lin g u is tic m aterial re fe r to s h o rt distance antecedents (cf. G ivo n , 1983; B e n tivo g lio , 1983). G iv o n (1992), fo r instance takes in to account so- called re fe re n tia l distance, i.e. the num ber o f clauses fro m th e la st occurrence in the p re ce d in g discourse fro m 1-20 clauses back. S anford and G arrod (1981) state th a t the am o u n t o f in te rv e n in g te x t since the m ost recent m e n tio n co n trib u te s to acce ssibility o f an e n tity . Gem sbacher (1990) claim s th a t distance its e lf is n o t the im p o rta n t fa cto r fo r accessibility. W h a t is im p o rta n t is the fact th a t the greater the distance betw een anaphors and antecedents, the greater the p ro b a b ility o f e n co u n te rin g in te rv e n in g co m p e tito rs. 2.5.5.2. T o p ic a lity and fre q u e n cy o f m e n tio n s The second fa c to r c o n trib u tin g to a cce ssibility is to p ic a lity . The n o tio n o f to p ic has lo n g been a m a tte r o f controversy. H o w e ve r, Gem sbacher et al., (1995) o p e ra tio n a lize d it b y co n sid e rin g tw o m a in characteristics. F irs t, to p ic a l e n titie s are fre q u e n tly m e n tio n e d and, second, to p ic a l e n titie s appear before n o n -to p ica l e n titie s. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 T his la st issue raises som e p ro b le m s. A num ber o f experim ents have sho w n th a t 'firs t m e n tio n e n titie s ’ are m ore accessible th a n 'second m e n tio n e n titie s ’. B ut, i t is n o t clear h o w the firs t m e n tio n e ffe ct is re la te d to the p re v io u s research o n given-new in fo rm a tio n (fo r instance C hafe 1976, 1994), w he re it has been claim ed th a t g iv e n in fo rm a tio n is presented firs t. I f it is the case th a t to p ic a l/im p o rta n t in fo rm a tio n u su a lly appears firs t, a n d th a t g ive n in fo rm a tio n appears firs t, the n n e w in fo rm a tio n w o u ld a lw a ys re fe r to n o n -im p o rta n t in fo rm a tio n . Y et th is co n clu sio n appears to be c o u n te r-in tu itiv e (fo r a lo n g e r discussion o f th is issue, cf. G em sbacher, 1990: 47 and C hafe, 1994: 73). M o re o ve r, w h ile there m a y be a firs t-m e n tio n effect fo r sentences o r sh o rt discourses, it is n o t clea r h o w th is effect co u ld be im p le m e n te d fo r lo n g e r discourses. Since the w o rk presented in th is thesis deals w ith lo n g conversation, I w ill th e re fo re n o t take th is la st aspect o f to p ic a lity in to account. A n o th e r factor re la te d to to p ic a lity is re fe re n tia lity . N o n - re fe re n tia l n o u n phrases are n o t v e ry im p o rta n t/to p ic a l in discourse. G iv o n (1992) states th a t im p o rta n t to p ics are m ore attended to, m e m orize d and re trie ve d m ore q u ic k ly (A nderson e t a l 1983; G em sbacher and Shroyer 1988; G em sbacher and H argreaves, 1988). A lso , topics are m ore fre q u e n tly m e n tio n e d . G ivon (1992) m easures the nu m b e r o f m entions o f the e n tity ca ta p h o rica lly, i.e. sub sequ ently, in the fo llo w in g 10 clauses. F re q u e n t e n titie s are o b v io u s ly m ore a ctiva te d , because each m e n tio n o f a n e n tity enhances its a c tiv a tio n . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 2.5.5.3. D iscourse stru ctu re A n o th e r fa cto r th a t has an effect o n the a cce ssib ility o f e n titie s is the e p is o d ic s tru c tu re . F o r instance, Fox (1987) has sh o w n h o w rh e to ric a l h ie ra rc h ic a l s tru c tu re plays an im p o rta n t ro le in the in te rp re ta tio n o f anaphors (see section 2.4. above and B rizu e la , 1997). I w ill discuss som e o f these issues in cha pte r 4, w h e re I present a data based an alysis o f discourse structure. 2.5.5.4. L in g u is tic size The lin g u is tic size (a tte n u a tio n in A rie l’s (1990) term s) o f d iffe re n t N Ps is also re la ted to accessibility. The fa ct th a t sh o rte r form s (fo r instance p ro n o u n s) re fe r to h ig h ly accessible e n titie s has been sh o w n b y C la n cy (1980), and b y Chafe (1976,1994), am ong others. From a p s y c h o lin g u is tic p o in t o f v ie w , G a rro d (1994:334) claim s th a t the m ore atte n u a te d the expression is, the m ore its in te rp re ta tio n re lie s d ire c tly on the im m e d ia te lin g u is tic context. T h is is a n o th e r w a y o f sa yin g th a t the s h o rte r the fo rm , the m ore accessible it is. G iv o n (1992) also presents the g ra m m a tica l co d e -q u a n tity p rin c ip le w h ic h states th a t there is an in ve rse re la tio n s h ip betw een a cce ssib ility o f the e n tity and the re la tiv e size o r gra m m a tica l c o m p le x ity o f a lin g u is tic fo rm . In the P ragm atic tra d itio n , L e vin so n (1987a) p ro vid e s the m in im iz a tio n p rin c ip le w h ic h says th a t tw o o p p o s in g forces are in v o lv e d in re fe re n tia l fo rm preferences. Speakers are pressured to co n fo rm to the M in im iz a tio n p rin c ip le (say as little as necessary), w h ic h pushes them to select shorter fo rm s, b u t th e y also need to consider lis te n e r needs, i.e. th e y need to m ake sure th a t the liste n e r can recognize the in te n d e d re fere nt. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 R elevance th e o ry sets o u t a s im ila r c o n flic t betw een processing cost ('le a st e ffo rt’) and co n te xtu a l effects (in th is case, re c o g n itio n o f the re fe re n t). A c c o rd in g to Sperber a n d W ils o n (1986), a liste n e r is guaranteed b y th e speaker th a t th e fo rm she chooses strike s th e rig h t balance betw een these tw o pressures. A lth o u g h a ll these approaches can account fo r a larg e n u m b e r o f cases, there are at least tw o c o u n te r exam ples. F irst, o ve r-sp e cifie d NPs m a y serve a discourse s tru c tu rin g fu n c tio n . V o n k et al., 1992 show ed th a t o ve r-sp e cifie d NPs are used to m a rk discourse s tru ctu re , i.e. a change o f scenery. A n d second, som etim es frequency can overcom e the p rin c ip le o f size-accessibility. A rie l (1990: 84) asserts th a t n e ith e r the L e vinson (1987a) n o r the S perber & W ils o n (1986) p ro p o sa ls can e x p la in w h y the d e m o n stra tive d e te rm in e r is lo n g e r than the d e fin ite a rtic le , even th o u g h the fo rm e r refers to m ore accessible e n titie s th a n the la tte r. She asserts: in fact, note th a t in E n g lis h a n d H e b re w , as w e ll as in m a ny o th e r languages, it is th e d e fin ite a rticle w h ic h is s h o rte r/m o re attenuated, a n d th e re fo re a ctu a lly p re d ic te d to be the h ig h e r, ra th e r th a n th e lo w e r A cce ssib ility M a rk e r (th e m ore in fo rm a tiv e m a rk e r, in L e vinson 's te rm in o lo g y ) (84) A rie l fu rth e r a ffirm s th a t th e p ro b le m o f these th e o rie s is th a t th e y o n ly co n sid e r m in im iz a tio n o r a tte n u a tio n as a fu n c tio n o f ease o f in te rp re ta tio n . T hey do n o t acco u n t fo r the fa c t th a t degrees o f a tte n u a tio n are also related to fre q u e n cy, i.e. m ore fre q u e n t fo rm s (so ca lle d u n m a rke d form s) ten d to g e t sh o rte r across tim e . Since the d e fin ite a rtic le is m uch m ore com m on in use, it tends to ge t sh o rte r Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 th a n the d e m o n stra tive , and in S em itic languages it even tends to be c litic iz e d . 2.5.5.5. Syntactic position A n o th e r fa c to r considered fo r the a cce ssib ility o f d iffe re n t definiteness types is the syntactic p o s itio n o f th e antecedents o f the anaphors. The ty p o lo g ic a l approach po stu la te s th a t synta ctic po sitions are ra nked a cco rd in g to s tru c tu ra l c o m p le x ity . F o r instance, Prim us (1993) proposes a syn ta ctic hie rarch y th a t ranks p o s itio n s according to th e ir a sym m e trica l c-com m and rela tion s. T h e re fo re , the NP im m e d ia te ly d o m in a te d b y speaker (u s u a lly ca lle d Subject) outranks an N P im m e d ia te ly d o m in a te d b y VP (u s u a lly ca lle d O bject) because the fo rm e r a s y m m e tric a lly c-com m ands the la tte r. T h ere fore , the subject p o s itio n is m ore accessible than the object p o s itio n . T h a t is to say, the sim p le r the syn ta ctic p o s itio n s tru c tu ra lly , th e m o re accessible it is. S im p lic ity in H a w k in s ' (1994: 42) term s in v o lv e s the am oun t o f stru ctu re in a tree d ia g ra m . The syntactic h ie ra rc h y in term s o f a sym m e trica l c-com m and relations has a v e ry precise co rre la tio n w ith co m p le xity: i f A a sym m e trica lly c-com m ands B, th e n the stru ctu ra l d o m a in o f B con tains A b u t n o t vice versa; hence s tru c tu ra l dom ain sizes increase, as w e proceed dow n the syn ta ctic h ie ra rch y. The a cce ssibility h ie ra rc h y fo r syntactic p o s itio n (also presented in G ivon , 1992) is, fro m h ig h to lo w accessibility: Subject < O bject < O thers G iv o n (1992) corroborates th is discrete o rg a n iz a tio n : the subject is consistendy m ore to p ic a l (accessible) th a n th e d ire c t object, and the d ire c t object m o re to p ic a l th a n the in d ire c t o b je ct (G ivo n , 1983a, 1984a.). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 H e asserts th a t hum an languages encode, g ra m m a tic a lly , o n ly three discrete levels o f to p ic a lity o f the p a rtic ip a n ts in e ve n ts/sta te s: (a) m a in to p ic = subject (b) secondary to p ic = d ire c t object n o n to p ic = a ll other roles. A lm o r (1996) and C arp en ter and Just (1977) also sh o w the im po rta nce o f syntactic co n strain ts in the a cce ssib ility o f referents. 2.S.5.6. T h e m a tic ro le Them atic ro le is an a d d itio n a l fa cto r assum ed to c o n trib u te to acce ssibility. R egarding the m atic hie rarch ies, P rim u s (1995) claim s th a t there are tw o types o f th e m a tic in fo rm a tio n conveyed b y a ve rb . One typ e o f in fo rm a tio n in vo lve s the n u m b e r and co n te n t o f basic them atic argum ents. The second typ e is the in fo rm a tio n a b o u t th e m a tic dependencies represented b y the re la tiv e s tru c tu ra l p o s itio n o f the p a rtic ip a n ts in the the m atic s tru c tu re o f a verb o r sentence. B y the c rite rio n o f the them atic dependency re la tio n s h ip be tw e e n the p a rtic ip a n ts o f a verb o r sentence, P atients and E xperienciers are th e m a tic a lly dependent o n A gents. P atients are d e p e n d e n t on E xperienciers. The fo llo w in g h ie ra rch y has been p ropo sed: A g e n ts /E x p e rie n c e s < P atients < O th ers M o re o ve r, G em sbacher (1990: 12) asserts th a t, in general, A gents are m ore a ctiva ted th a n N on-agents: Sem antic agents ten d to be m ore anim ate (C la rk, 1965; Johnson, 1967), m ore active (O sgood, 1971), m o re p o s itiv e ly evaluated (Johnson, 1967), and m ore im ageable (James, 1972; James, T hom pson & B a ld w in , 1973). Because o f these characteristics, several th e o rists have suggested th a t agents are m ore lik e ly to a ttra c t a tte n tio n (Z u b in , 1979), stim u la te em path y (K u n o , K abaraki, 1977), and m atch the speaker o r liste n e r's perspe ctive (M a cW h in n e y, 1977). Sem antic agents are also ty p ic a lly th e ir sentences' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 synta ctic subjects (G reenberg, 1963), to p ics (G iv o n , ed. 1983), and them es (T o m lin , 1983) I t is clear th a t A gen ts are the m ost accessible th e m a tic ro le , w ith the rest o f the h ie ra rc h y n o t c le a rly defined. 2.5.5.7. Deixis and Accessibility The te rm d e ix is is used to re fer to the fu n c tio n o f personal and d e m o n stra tive p ro n o u n s w h ic h relate utterances to th e sp a tio - te m p o ra l co-ordinates o f the act o f utterance. D e ixis refers to the lo ca tio n and id e n tific a tio n o f persons, objects, events, processes and a c tiv itie s b e in g re fe rre d to , in re la tio n to the s p a tio -te m p o ra l context o f the speech e ve nt (L yons, 1977: 637). Lyons (1977: 638) states th a t the canonical s itu a tio n o f a speech e ve nt is egocentric because speakers relate e v e ry th in g to th e ir ow n p o in t o f v ie w . T hey are a t the ze ro -p o in t o f the sp a tio -te m p o ra l co ordina tes o f the d e ic tic con text. T his e g o ce n tricity is te m p o ra l because the ro le o f speakers changes fro m one p a rtic ip a n t to th e o th e r as the con versa tion proceeds, an d th e p a rticip a n ts m a y m o ve a ro u n d as they are ta lk in g : the sp a tio -te m p o ra l ze ro -p o in t (the h e re -a n d -n o w ) is d e te rm in e d b y the place o f the speaker at the m o m e n t o f th e speech e v e n t. Lyons adds: D e m o n stra tive p ro n o u n s and d e m o n stra tive a d jectives, lik e the E n g lish 'th is ' and 'th a t', as w e ll as d e m o n stra tive adverbs, such as 'he re' and 'there', are p rim a rily d e ictic; and, w h e n th e y have th is fu n ctio n , th e y are to be in te rp re te d w ith respect to the lo ca tio n o f th e p a rtic ip a n ts in the d e ic tic co n te xt. R o u g h ly speaking, th e d is tin c tio n betw een 'th is ' an d 'th a t' and betw een 'here' a n d 'th e re ', depends u p o n p ro x im ity to the zero p o in t o f th e d e ictic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 con text: 'th is b o o k' means 'th e b o o k w h ic h is he re’ o r the b o o k w h ic h is near to the speaker (Lyons, 1977: 646)5 K irs n e r (1979) claim s th a t de m on strative s are used to d ire c t the a tte n tio n o f the liste n e r ra th e r th a n s im p ly in d ica te the distance betw een the e n tity re fe rre d to and the speaker. The essence o f d e m o n stra tive s, fo llo w in g K irs n e r (1979), is in s tru c tio n a l (cf. p ro c e d u ra l m e a n in g 2.5.1) ra th e r th a n d e s c rip tiv e . K irs n e r p ro vid e s three strategies fo r choosing betw een h ig h o r lo w d e ixis fo r D u tch d e m o n stra tive s: 1) N o te w o rth in e ss (s h o w in g the speaker’s in te re st in the e n tity re fe rre d to ), 2) G ivenness (cf. C hafe, 1976, 1994), and 3) F o re g ro u n d . S trategy 2 is related to th e assu m ptio n th a t the m ore d iffic u lt it is fo r the addressee to fin d th e re fere nt, the m ore s tro n g ly the speaker w ill urge the liste n e r to fin d the re ferent. Therefore, fo r a lo w accessible e n tity (n o t given) a h ig h d e ixis device w ill be used. D e ixis has also been d e fin e d as "th e force w ith w h ic h the liste n e r is in s tru c te d to fin d the re fere nt" (G arcia, 1975: 65). T h a t is, h ig h de ixis refers to antecedents th a t are n o t o b vio u s, th a t are re la tiv e ly h a rd to fin d , o r th a t sh o u ld u n d e r no circum stances be m issed. L o w de ixis re fe r to antecedents th a t are ob vious, th a t are easy to fin d , o r w here it does n o t m u ch m a tte r if the re fe re n t is id e n tifie d o r n o t. The n o tio n o f d e ixis can be re la te d to the acce ssibility discussed above. D e ic tic elem ents are d e fin e d as 'in s tru c tio n s to the liste n e r' s im ila r to a cce ssib ility m arkers, i.e. w ith p ro ce d u ra l m eaning. In general, h ig h d e ixis devices have been re la te d to m arkers o f h ig h e r a cce ssib ility (such as the p ro x im a l de m on strative s) and lo w de ixis 5cf. King's (1992) implementation of this description to Spanish in section 2.1.2. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 devices are m arkers o f lo w e r accessibility (such as d is ta l dem onstratives; A rie l, 1990; G un de l et al, 1993). H ow eve r, G arcia's (1975) and K irsn e r's (1979) in te rp re ta tio n o f de ixis does n o t re la te d ire c tly to the th e o ry o f acce ssibility. N ote th a t h ig h d e ixis devices re fe r to b o th less obvious antecedents (less accessible), and to v e ry sa lie n t antecedents (those w h ic h u n d e r no circum stances are m issed). These tw o kin d s o f antecedents cannot relate to h ig h a cce ssib ility devices, a t least d ire c tly . M oreover, P iw e k e t al.'s (1996: 847) re sults sho w th a t proxim ates are used p re fe ra b ly to re fe r to less accessible e n titie s. K irs n e r (1979) and P iw e k et al. (1996) sh o w th a t th is is the case fo r D u tc h dem onstratives. N o t-g iv e n antecedents (less accessible) are re fe rre d to b y a h ig h a cce ssibility device: the p ro x im a l de m on strative . H o w e ve r, th is claim needs fu rth e r research. O ne reason is th a t the exp erim e nts presented b y P iw e k et al. (1996) use a p ro b le m a tic d e fin itio n o f accessibility. They have designed an e x p e rim e n t in w h ic h a researcher gives in stru ctio n s to the listen er to re arra ng e a g ive n b lo ck b u ild in g . The factors d e te rm in in g a cce ssib ility w ere: the shape o f an object, and the object loca tion. I f an object has exceptional shape, it is m ore accessible. I f the object is in the area to w h ic h the speaker e x p lic itly d ire cte d the a tte n tio n o f the addressee, it is m ore accessible. F in a lly , i f the object is adjacent to an object w h ic h was m a n ip u la te d o r ta lk e d ab out im m e d ia te ly before the re fe rrin g act too k place, it is also m ore accessible. The d e fin itio n o f 'exceptional shape' is n o t a guarantee o f accessibility. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 2.5.5.8. In te ra c tio n a l A pproaches Tao (1994) states th a t d e ixis sh o u ld be re la te d to the in te ra c tiv e nature o f language, and therefore, the in d e x ic a l g ro u n d s are n o t fix e d (fo r instance, close o r fa r from , the speaker). O n the c o n tra ry , th e y s h ift d yn a m ica lly as the in te ra c tio n proceeds. D e m o n stra tive s, th u s, re fle ct the speaker's a ssu m p tio n ab out the liste n e r's state o f m in d . Tao gives an account o f d e m o n stra tive s in M a n d a rin C hinese, co n s id e rin g d iffe re n t factors. O ne o f these factors is socia l distance. H e states th a t dem onstratives are n o t o n ly used fo r reference m a k in g b u t also to show social a ttitu d e to w a rd s some e n titie s o r facts th a t the in te rlo cu to rs are ta lk in g about. Language, in th is v ie w , is a to o l fo r social o rg a n iz a tio n in general. Strauss (1993: 409) states th a t the p ro x im a l this in E n g lis h m arks h ig h focus and separateness and independence fro m the lis te n e r (409), w h ile the d is ta l that/it sh o w s o lid a rity am o n g in te rlo c u to rs . A n o th e r in te ra ctio n a l ap pro ach is offe re d b y L a u ry (1997). These in te ra c tio n a l approaches extend a cce ssib ility th e o ry b y asserting th a t a cce ssib ility is n o t an in trin s ic p ro p e rty o f e n titie s , b u t th a t it is an in te rp re ta tio n o f the speakers. I con cur w ith th is la st con ten tion because i t is tru e th a t a h ig h a cce ssib ility device m ig h t be used to re fe r to a lo w a cce ssib ility e n tity to m ake it becom e m ore accessible. A s Fox (1987) asserts, re fe re n tia l devices have a co n ve n tio n a l m e a n in g th a t can be v io la te d o n p u rp o se . In o th e r w ord s, th e y are g e n e ra lly used in ce rta in contexts (fo r instance to re fer to h ig h a cce ssib ility e n titie s), b u t som etim es th e y are used to create the re q u ire d co n te xt (to tu rn a non-accessible e n tity in to a s a lie n t one). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 2.5.6. A c c e s s ib ility o f e n titie s vs. A c c e s s ib ility o f sets. W h a t is i t th a t is re a lly accessible to the in te rlo cu to rs? I t seem s, a t firs t, th a t acce ssibility is a p ro p e rty o f the m e n ta l representation o f e n titie s w h ic h in tu rn are s tru c tu re d in a m e n ta l re p re se n ta tio n o f discourse. The o rg a n iz a tio n o f disco urse is a co n tro ve rsia l and co n fu sin g to p ic , as show n above (cf. 2.4). W ith in acce ssibility th e o ry , e n titie s b e lo n g to d iffe re n t types o f so-called contexts, and these contexts have d iffe re n t leve ls o f a cce ssib ility. A lth o u g h it is n o t e x p lic itly stated, it is im p lie d th a t e n titie s in h e rit the le ve l o f a cce ssib ility o f the contexts th a t th e y b e lo n g to . The d is tin c tio n betw een the a cce ssib ility o f e n tities and the a cce ssib ility o f contexts is n o t c le a rly d e lim ite d in the lite ra tu re su rve ye d . 2.5.6.I. A c c e s s ib ility o f contexts (P-sets) In th is section, I p re se n t d iffe re n t h ie rarch ies o f a cce ssib ility o f contexts. F or instance, A rie l (1990: 68) assumes th a t context types are h ie ra rc h ic a lly ordered a cco rd in g to degree o f acce ssibility to the liste n e r. The p re vio u s te x t is the m o st accessible con text, fo llo w e d b y the p h y s ic a l s itu a tio n , and th e n e n cyclo p e d ic kn o w le d g e . W ith in relevance th e o ry (Sperber & W ils o n , 1986), th e ra n k in g is as fo llo w s : a) p h ysica l s itu a tio n (e n titie s w h ic h are v is u a lly present) b) p re vio u s te x t c) the set o f associative e n titie s d) the set o f e n titie s th a t b e lo n g to encyclopedic kn o w le d g e e) the lo g ica l fo rm o f the sentence associated w ith the utterance b e in g processed Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 S perber & W ilso n (1986: 142) state: N o tice a p o in t o f c ru c ia l im p o rta n ce fo r relevance th e o ry: ju s t as processing an ite m o f in fo rm a tio n in a con text in vo lve s som e e ffo rt, so accessing a context in v o lv e s some e ffo rt. The less accessible a context, the greater the e ffo rt in v o lv e d in accessing it, and conversely. C hafe (1994: 101) m e n tio n s th a t referents th a t are presen t in the im m e d ia te e n viro n m e n t are m o re s a lie n t than m o re d is ta n t referents. W ith in every social g ro u p th e re are ce rta in referents th a t are salien t fo r the m em bers o f th a t g ro u p . G ivo n (1992: 12) claim s th a t d e te rm in in g the source (antecedence) o f definiteness type s can be done fro m three sources6 : a) the m e n ta l m o d e l o f the speech s itu a tio n (p h ysica l s itu a tio n set) b) the pe rm a n e n t se m a n tic-cu ltu ra l m e m o ry (encyclopedic kn o w le d g e ) c) the e p isodic m e m o ry o f the cu rre n t te x t (p re vio u s text) G ivo n notes th a t the sp e cific d e ta ils o f the o rg a n iz a tio n o f these u n its in m e ntal m odels have y e t to be w o rke d o u t. The m a in d iffic u lty rests w ith the fact th a t th is s tru c tu rin g is in te rn a l an d subjective. G iv o n (1992) does n o t see o v e rw h e lm in g grounds fo r d e te rm in in g the re la tiv e m arkedness o f the th re e dom ains. The im m e d ia te speech s itu a tio n is accessible a u to m a tic a lly at a n y given co m m u n ica tive m o m e n t. The e xtre m e ly h ig h to p ic a lity o f "I", "y o u ", "he re", "th is" an d "n o w " suggests th a t th e ir referents are a u to m a tic a lly activa te d 6 I added the text between parentheses to facilitate the connection with the approaches presented above. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 d u rin g the s itu a tio n . P erm anent s e m a n tic -c u ltu ra l m em ory is a u to m a tic a lly activated b y vo ca b u la ry. C o-reference searches th ro u g h e p isodic m e m o ry are trig g e re d b y the sp e cific gram m atical devices subsum ed u n d e r the gram m ar o f re fe re n tia l coherence. Thus, a ll o f these scholars recognize three d iffe re n t sources o f in fo rm a tio n : 1) the s itu a tio n o f speech 2) the te xt 3) c u ltu ra l based kn o w le d g e (g lo b a l, based on fram es o r sem antic re la tio n sh ip s such as w h o le -p a rt) A ll agree th a t these three contexts are ra nked according to acce ssib ility, alth oug h there is disagreem ent on the specific o rd e r. 2.5.7. T he in te ra ctio n o f factors K ib rik (1996) offers the o n ly w o rk th a t attem pts to account fo r the in te ra c tio n o f factors c o n trib u tin g to acce ssibility (see also B rizu e la , in press). H e firs t lists a ll the factors th a t have been considered fo r the selection o f the anaphora (lin e a r distance, rh e to ric a l distance, p a ra g ra p h boundaries, sem antic and syn ta ctic ro le o f the antecedent and s lo p p y id e n tity o f referents). Second, he enters these factors in to a sta tis tic a l p ro g ra m (SHOEBOX) th a t m akes a co rre la tio n w ith the d iffe re n t types o f Russian NPs. H e is inte reste d, p rim a rily , in the d iffe re n ce betw een reduced vs. no n -re d u ce d NPs. T h ird , if th e c o rre la tio n betw een the values o f the factors and the specific ty p e o f N P is .66 o r greater, he includes th is fa c to r in h is m odel. F o urth , he takes the factors selected and assigns an a rb itra ry w e ig h t to the d iffe re n t values o f the factors. For instance, he assigns to the subject p o s itio n an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 a c tiv a tio n v a lu e o f 0.4 and one o f 0.3 fo r n o n-su bject po sitio n s. F in a lly , g iv e n these a rb itra ry values, he m easures the le ve l o f a c tiv a tio n o f each e n tity in a w ritte n p rose co rp o ra b y a d d in g the d iffe re n t scores and p re d icts the typ e o f N P th a t is g o in g to be used a cco rd in g to th is a c tiv a tio n score. I agree w ith K ib rik th a t the c u rre n t challenge to the th e o ry o f anaphora is to account fo r the in te ra c tio n o f factors th a t in flu e n ce antecedent selection. Several factors ha ve been s tu d ie d in d e p e n d e n tly, b u t no one, except fo r K ib rik , addresses th e q u estion o f the in te ra c tio n o f factors. T herefore, h is in s ig h t is v a lu a b le in th is respect and needs to be taken as a p o in t o f de parture . N onetheless, K ib rik 's approach fa ils to a rtic u la te th is in s ig h t w ith a clea r and re lia b le m e th o d o lo g y (see cha pte r 5). 2.6. P ro m in e n ce E p ste in (1994) focuses h is d is s e rta tio n on the d e fin ite a rticle . H e rejects th e uniqueness re q u ire m e n t fo r th e d e fin ite a rticle . H e asserts th a t th e conveys the prom ine nce o f a re fe re n t, and th a t th is fu n c tio n is m ore im p o rta n t th a n the uniqueness re q u ire m e n t. E pstein (1994) asserts th a t th e speaker em ploys a rticle s to com m unicate personal a ttitu d e s to w a rd s a re fe re n t. T herefore, th e re fe re n tia l v ie w , represents o n ly a p a rtia l v e rsio n o f the o v e ra ll p ic tu re o f w h a t is h a p p e n in g w he n one a rtic le is selected. E pstein argues th a t the d e fin ite a rticle possesses an expressive fu n c tio n , in a d d itio n to its re fe re n tia l fu n c tio n . Speakers co m m u n ica te th e ir o w n sub jective e v a lu a tio n o r assessm ent o f the re fe re n t (fo r exam ple, th a t the re fe re n t is p ro m in e n t in the discourse Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 con text). E pste in does n o t, h o w e ve r, p ro v id e a d e fin itio n o f pro m in e n ce , a lth o u g h it fo rm s a m a jo r p a rt o f h is a rg u m e n ta tio n . 2.7. T he P s y c h o lin g u is tic approach In th is section I present tw o p s y c h o lin g u is tic approaches. The firs t is the cen terin g th e o ry (G rosz, 1981; G ordon et al., 1993, G rosz e t al., 1995) th a t o rig in a te d w ith in the c o m p u ta tio n a l lin g u is tic fra m e w o rk . I t is a th e o ry o f reference th a t p ro v id e s ru le s o f reference in the fo rm o f p rin c ip le s th a t can be in co rp o ra te d in a lg o rith m ic and co m p u ta tio n a l p rogram s. F o r instance, based m e re ly on the d is trib u tio n o f a n a p h o ric references in sh o rt dialogues, G rosz e t al. (1995) postulate th a t a p ro n o u n s h o u ld be used w h e n re fe rrin g to the re fe re n t in focus if any p ro n o u n is to be used at a ll. There are tw o m a in p roble m s w ith th is the ory: firs t, a m a jo r n o tio n , 'focus o f a tte n tio n ’, is n o t cle a rly d e fin e d . Second, as A lm o r (to appear) dem onstrates, th e ir p rin c ip le s la ck psych olog ical s u p p o rt. The second p s y c h o lin g u is tic approach is th a t o f G em sbacher (1990,1997). She p ro vid e s a th e o ry o f m ental re presentation o f the te x t th a t, cou pled w ith experim ents, a llo w s us to understand w h y som e o f the factors c o n trib u tin g to a cce ssib ility m entioned above c o n trib u te w ith greater o r less stre n g th to reference id e n tific a tio n . R e fe re n tia l expressions are cues fo r the b u ild in g o f re fe re n tia l coherence. A te x t is coherent i f tw o co n d itio n s are fu lfille d . F irst, the te xt sh o u ld be co n n e cte d , th a t is, it sh o u ld establish re la tio n sh ip s betw een the d iffe re n t p a rts o f the text. Second the te x t sh o u ld be g ro u n d e d , in o th e r w o rd s , the in fo rm a tio n in the te x t sh o u ld be h ie ra rc h ic a lly o rg a n ize d . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 I f the m e n ta l re p re se n ta tio n (g iv e n these tw o c o n d itio n s ) b u ilt in the m in d o f the liste n e r, is s im ila r to the one in the m in d o f the speaker, then, the te x t is coherent (G em sbacher and G ivo n , 1995). G em sbacher (1997) presents fiv e types o f coherence: 1) re fe re n tia l coherence, w h ic h is re la ted to w h o o r w h a t is be ing discussed, 2) te m p o ra l coherence w h ic h is re la ted to w hen the events th a t are be ing discussed happen, 3) lo c a tio n a l coherence, related to w h e re these events occur, 4) causal coherence, w h ic h is related to w h y these events occur an d fin a lly 5) s tru c tu ra l coherence w h ich is re la ted to the fo rm in w h ic h events are described in the te x t o r discourse. G em sbacher (1997) claim s th a t each o f these types o f coherence is cued b y e ith e r im p lic it o r e x p lic it signals. R e fe re n tia l expressions co n trib u te to the c o n n e c tiv ity o f the te xt (by re fe rrin g back and fo rw a rd ), and re fe r to the h ie ra rc h ic a lly s tru c tu re d in fo rm a tio n (in P-sets). A cco rd in g to G em sbacher (1997), in te rlo c u to rs b u ild a m e n ta l representation o f the te x t w h ile th e y are ta lk in g . She claim s th a t w e b u ild m e n ta l representation u s in g m e m ory nodes: The b u ild in g blocks o f the m e n ta l structures are m e m o ry nodes. M e m o ry nodes are activa ted b y in co m in g s tim u li. In itia l a c tiv a tio n fo rm s the fo u n d a tio n o f m e n ta l stru ctu re s. Once m e m o ry nodes are activated, th e y tra n s m it processing signals to enhance (increase) o r suppress (decrease o r dam pen) other nodes' a c tiv a tio n . T hus, once m e m o ry nodes are activated, tw o m echanism s c o n tro l th e ir le v e l o f a c tiv a tio n : suppression and e n h a n ce m e n t. S uppression and enhancem ent are the tw o m echanism s th a t c o n tro l the a c tiv a tio n o f the nodes in m em ory. S uppression is re la ted Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 to the distance fa cto r: the closer the antecedent, the less the p o s s ib ility o f e n co u n te rin g a co m p e tito r fo r the reference, w h ic h con sequ ently, w o u ld suppress the a ctiva tio n o f the antecedent. T his is consistent w ith p sych o lo g ica l experim ents (C la rk & S engul, 1979; E h rlic h & Rayner, 1983; F rederiksen, 1981) th a t sh o w the im p o rta n ce o f p ro x im ity fo r the in te rp re ta tio n o f anaphors. F or instance, C la rk & S engul (1979) c la im th a t sentences c o n ta in in g references to re ce n tly m entioned antecedents g e n e ra lly take less tim e to read th a n sentences re fe rrin g to d is ta n t antecedents. E h rlic h & R ayner (1983) state th a t readers spend lo n g e r fix a tin g the te xt im m e d ia te ly a fte r e n co u n te rin g a p ro n o u n th a t refers to a d is ta n t antecedent as com pared to a less d is ta n t one. In general, lin g u is tic processes tend to reduce re le va n t dom ains fo r any lin g u is tic p u rp o se . F or instance, H a w k in s (1998) states th a t there is a perform ance preference to m in im iz e the dom ains in w h ic h syntactic and sem antic re la tio n s, such as c o n s titu e n t g ro u p in g , are processed w ith in a sentence. Enhancem ent, is related to the fa cto r 'fre q u e n cy o f m e n tio n s': each tim e an e n tity is repeated, it gains a c tiv a tio n ; the refore , the m ore fre q u e n t, the m ore activated. In chapter 5 , 1 show th a t these tw o factors are, ind eed , the m ore im p o rta n t ones fo r reference in te rp re ta tio n o f definiteness types. In o th e r experim ents, n o u n phrases have been d is tin g u is h e d fro m pronouns p sych o lin g u is tic a lly . For instance, G arrod , e t al. 1994 (41) sho w th a t n o u n anaphors o n ly a ctiva te surface in fo rm a tio n a b o u t th e ir antecedents, whereas pronouns a ctiva te deeper conceptual in fo rm a tio n about them , w h ich , in general, takes m ore tim e . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 G arrod & S anford (1990) state th a t p ro n o u n s seem to operate b y d ire c tly id e n tify in g antecedent discourse e n titie s. F u lle r d e fin ite n e ss types operate in a m ore general w a y to recover discourse roles. These is a diffe ren ce betw een focu ssin g an antecedent as an e n tity as opposed to focussing an antecedent as a ro le . P ronouns are devices fo r m a in ta in in g reference to h ig h ly focused antecedent e n titie s (K a rm ilo ff S m ith 80, M a rsle n -W ilso n e t a l 82). 2.8. Conclusion In th is chapter, I su m m a rize d th e m a jo r c o n trib u tio n s in the lite ra tu re to the in te rp re ta tio n o f de finiten ess types. I show ed th a t, in the lite ra tu re on S panish d e fin ite n e ss type s, the accounts are v e ry lim ite d , and th e y o v e rlo o k su b tle p ra g m a tic phenom ena. F ro m the lite ra tu re o n E n glish and o th e r languages, I described the th re e m a jo r key term s fo r th is d isse rta tio n ; these are: uniqueness; P-set; and accessibility. The sem antic n o tio n o f uniqueness p o stu la te d b y R ussell (1905) and rescued by H a w kin s (1991) is a v e ry im p o rta n t p a rt o f m y d e fin itio n o f definiteness type s (cf. cha pte r 4). The re q u ire m e n t o f uniqueness is n o t an absolute, b u t h o ld s w ith in the lim its o f a set d e lim ite d p ra g m a tica lly b y th e speaker (P-set). These P-sets, in tu rn , are ra nked according to degrees o f acce ssibility. Factors c o n trib u tin g to acce ssibility have been presented, b u t the lite ra tu re to date does n o t fu lly account fo r the in te ra c tio n and re la tiv e stre n g th o f these factors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 C h a p te r 3 M e th o d o lo g y 3.1. T he N a tu re o f th e D ata: N a tu ra l O ra l D isco u rse The s tu d y presented in th is thesis is based o n n a tu ra lis tic , co n ve rsa tio n a l discourse, in an a tte m p t to e xp lo re w h a t is h a p p e n in g in the m in d o f the lis te n e r and o f the speaker d u rin g the o n -lin e processing o f language. The data consist o f three tria d ic conversations (in c lu d in g the researcher and tw o subjects) o f a p p ro x im a te ly one h o u r each, recorded in in fo rm a l settings. To assure as n a tu ra l as possible a discourse context, I recorded the conversations o f p e ople I kne w fa irly w e ll. The in te rlo c u to rs are a ll y o u n g people in th e ir m id to late tw e n tie s (a ll w ere 23-29 years o f age). M o re d e ta ils o n the subjects, the m e th o d and the analyses are in c lu d e d in th is chapter. 3.2. The S ubjects The subjects w hose discourse w as analyzed in th is s tu d y w ere a ll m o n o lin g u a l n a tiv e speakers o f A rg e n tin e S panish, except fo r the b ilin g u a l researcher w hose firs t language is A rg e n tin e Spanish. I w o u ld have p re fe rre d to in clu d e o n ly the speech o f the three p a irs o f speakers described b e lo w ; ho w ever, because o f the constant exchange o f co n ve rsa tio n a l tu rn s th a t take place in frie n d ly , in fo rm a l conversations, it w as necessary to in c lu d e the speech o f the researcher in th e data analysis in o rd e r to trace the o n -g o in g reference to any e n tity . Table 1 p ro v id e s in fo rm a tio n a b o u t the subjects w hose speech w as stu d ie d . I recorded the firs t p a ir, K a ly and G ustavo, a t th e ir hom e a fte r h a v in g a v e ry in fo rm a l lu n c h in a re la xed e n viro n m e n t. K a ly Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 and G ustavo are husband and w ife . K a ly is a h ig h school graduate and G ustavo is c u rre n tly fin is h in g a u n iv e rs ity degree in accounting. The second p a ir, F lora and V iv ia n a , are sisters. The re co rd in g to o k place on a Sunday m o rn in g a t th e apartm ent th e y share. F lora re ce n tly received a degree in S ocial W o rk , and V iv ia n a is w o rk in g on h e r degree in accounting. I m e t w ith the th ird p a ir, M a rin a and M a rio (b o y frie n d and g irlfrie n d ), in a sm all cafe. M a rin a re ce n tly received h e r degree in S ocial W o rk , and M a rio has co m p le te d a h ig h school ed u ca tio n . Thus, because o f the in tim a te n a tu re o f the re la tio n sh ip s, a ll the conversations w ere v e ry a n im a te d and in fo rm a l.1 Table 1: Information about the subjects 1) In te rv ie w 1: K aly-G ustavo 2) In te rv ie w 2: F lo ra -V ivia n a 3) In te rv ie w 3: M a rin a - M a rio Interview N am e Gender R elationship Education Age 1st K a ly F w ife H igh school graduate 23 p a ir G ustavo M h u sb a n d Finishing College- Accounting 24 2 n d F lo ra F siste r College graduate: Social Work 24 p a ir V iv ia n a F siste r Finishing College: Accounting 23 3rd M a rin a F g irlfrie n d College graduate: Social Work 24 p a ir M a rio M b o y frie n d High school graduate 25 M aq F researcher Graduate student 29 1Notice that example (9) in chapter 4 comes from an interview that was not coded for the quantitative analyses in chapter 4 and 5. This specific group conversation included two young males (M and E) and two young females (C and Maq), all of them native speakers of Argentine Spanish between 23-29 years old. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 3.3. T h e s tru c tu re o f th e in te rv ie w A ll th e in te rv ie w s to o k place in Buenos A ire s , A rg e n tin a in J u ly 1994. F or each in te rv ie w , I ha d p re p a re d som e questions to use to e lic it conversation, i f necessary, b u t in general I s im p ly encouraged the speakers to ta lk ab out top ics th a t in te reste d the m . I alw ays yie ld e d the flo o r i f the co n ve rsa tio n w as flo w in g flu e n tly and n a tu ra lly , w h ic h was u su a lly the case. To o p e n the conversations, I w o u ld ask th e subjects to te ll m e about them selves: th e ir occupations, th e ir inte rests, etc. If, a t some p o in t the co n ve rsa tio n seemed to s ta ll, I w o u ld ask Labov (1972) w e ll- kn o w n 'close to death experience' q u e stio n o r o th e r types o f questions th a t im p ly h ig h in v o lv e m e n t such as: w h a t do y o u th in k about in fid e lity b e tw een husband and w ife? The p u rp o se o f in tro d u c in g h ig h in vo lve m e n t to p ics w as to encourage discourse in w h ic h the speakers concentrated o n the con ten t o f w h a t th e y w ere sa yin g ra th e r than m o n ito rin g th e fo rm th e y used o r the s tru c tu re o f th e ir discourse. 3. 4. T h e tra n s c rip tio n s 3.4.1 In to n a tio n a l U n its The tra n s c rip tio n s w ere done fo llo w in g the conventions in D u Bois et al. (1992). The data w ere segm ented in to in to n a tio n a l u n its (IU s). A s suggested b y Chafe (1994), D u Bois e t al. (1992) and G ivo n (1992: 7), th e n o tio n o f in to n a tio n a l u n it is m o re satisfa ctory th a n the n o tio n o f sentence fo r several reasons. F irs t, it b e tte r reflects h o w speakers a c tu a lly produce language. T h a t is, speakers do n o t alw ays ta lk in com plete sentences, b u t inste ad p ro d u ce discourse in spu rts, som etim es e q u iv a le n t to in co m p le te sentences, som etim es to ru n -o n Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 sentences. M o re o ve r, C hafe (1994) claim s th a t th e segm entation in IU s fo llo w s the p h y s io lo g ic a l re q u ire m e n t o f b re a th in g , w h ic h a t the same tim e operates in s yn ch ro n y w ith som e basic fu n c tio n a l segm entations o f discourse (Chafe, 1994: 57). The features th a t cha racte rize IU s m ay in v o lv e a n y o r a ll o f the fo llo w in g : changes in fu n d a m e n ta l fre que ncy (p e rce ive d as p itc h ) changes in d u ra tio n (perceived as th e sh o rte n in g o r le n g th e n in g o f sylla b le s o f w o rd s) a cce lera tion- d e ce le ra tio n . changes in in te n s ity (p erce ive d as loudness) a lte rn a tio n o f v o c a liz a tio n w ith silence (p erce ive d as p a u sin g ) changes in voice q u a lity o f va rio u s k in d s , and som etim es, changes o f tu rn s Each IU ve rb a lize s the in fo rm a tio n a ctive in th e speaker's m in d at its onset, th a t is, th e S's focus o f consciousness a t th a t m om ent (C hafe, 1994: 63). C hafe (1994) has sho w n th a t th is n o tio n is u n ive rsa l and a p plicab le to co m p le te ly d iffe re n t languages. A s C hafe (1994: 63) pu ts it: I t is th ro u g h th is d yn a m ic process o f successive a ctiva tio n s, firs t fo r the speaker and th e n , th ro u g h the utterance o f an IU fo r the L [listener] th a t language is able to p ro v id e an im p e rfe ct b rid g e betw een one m in d and a n o th e r. There are d iffe re n t types o f IU s: re g u la to ry , sub stan tive and fra g m e n ta ry. R e g u la to ry IU s are g e n e ra lly discourse m a rkers (D M s, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. s' 4 O N U l £ « . W N J H - ‘ 0 0 v j a N U i r f a > o j | v j l_i 54 S c h iffrin , 1987) and serve d iffe re n t fu n ctio n s. F o r instance, as I w ill show in chapter 4, D M s m ay be used to s tru c tu re discourse, i.e. to m a rk lim its o f discourse u n its . W hen D M s alone co n stitu te an IU , th e y are considered re g u la to ry . A t o th e r tim es, w h e n th e y are accom panied b y o th e r lin g u is tic m a te ria l, D M s co n stitu te su b sta n tive IU s. S ubstantive IU s h e lp the discourse progress b y a d d in g in fo rm a tio n , m a k in g com m ents, etc. F ragm entary IU s are those IU s th a t have n o t been fin is h e d b y the speaker because she has been in te rru p te d o r because she is re -th in k in g w h a t she is g o in g to say. In (1) b e lo w , lin e 2 is a fra g m e n ta ry IU : the speaker (F) w a n te d to say 'he started to get in te g ra te d ' b u t th e n she d id n o t fin is h h e r statem ent u n til the n e xt IU w h e n she says th a t 'he g o t in te g ra te d m u ch m o re'. In con trast, lin e 6 is a re g u la to ry IU , since the D M aparte 'besides' is s tru c tu rin g discourse b y o p e n in g a new discourse u n it (cf. cha pte r 4 fo r an extensive account). The o th e r IU s (1, 3, 5, 7) in th is exam ple are a ll sub stan tive IU s, as they co n ta in co n te n t and co n trib u te to m o v in g th e discourse fo rw a rd . (1) F: y ahi en quinto eh como que em— ya se integro muchisimo mas (2 IU s) V : pero a ella no le gustaba en ese momento aparte bueno le gustaron un monton F: an d the re in fifth grade e h lik e he— he g o t in te g ra te d m u ch m ore (2 IU s) V : b u t she d id n 't lik e h im at th a t tim e besides w e ll she lik e d a lo t (o f guys) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 C hafe (1994: 66) proposes th a t the fact th a t the n u m b e r o f w o rd s in an IU rem ains w ith in a n a rro w range fo r any one language re flects a co n stra in t o n the capacity o f active consciousness. 3.4.2. Transcription conventions Some o f the m o st com m on tra n s c rip tio n co n ve n tio n s used in th is d is s e rta tio n are: • carriage re tu rn : IU • @: la u g h • - tru n ca te d IU • -: tru n ca te d w o rd • (...): o m itte d te xt • XX: u n in te llig ib le te xt • <Q Q >: quote im ita tin g the voice q u a lity o f the speaker w h o u tte re d the quote N o tic e th a t s p e llin g conventions -and n o t p h o n e tic co n ve n tio n s- have been fo llo w e d th ro u g h o u t the exam ples. H o w e ve r, the tra n s c rip tio n o f the shortened versio n o f the D M bueno 'w e ll' has been represented as bueh. The 'h ' represents the a s p ira tio n a t the end o f the w o rd . T h is is used because in A rg e n tin a , th is co n ve n tio n is q u ite com m on in new spapers, cartoons o r o th e r contexts w h e re o ra l language is represented. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 3.5. T h e va ria b le s O nce the con versa tions w ere tra n scrib e d and segm ented in to IU s, a ll the d e fin ite expressions w ere id e n tifie d and coded acco rdin g to a set o f e ig h t varia bles re le va n t to the analyses I w ill p re se n t in C hapter 4; a subset o f fo u r o f these va ria bles are used in th e analyses in C hapter 5 (i.e., distance, fre q u e n cy o f m e ntion , syn ta ctic p o s itio n an d them atic ro le ). F o r each a n a p h o ric d e fin ite expression, I also coded b o th its an tece den t and its p o s s ib le co m p e tito r. Each o f the e ig h t varia bles has va ria n ts o r values th a t are e xp laine d b e lo w . 3.5.1. V a ria b le 1: T yp e In th is va ria b le , I coded the type o f N P . F o r th e analyses th a t w ill fo llo w in chapter 4, th is w o u ld be the dependent va ria b le . The fo llo w in g seven types w e re in clu d e d : 1) D e fin ite a rtic le : la fs idea /s 'th e id e a /s ' o r elflos libro/s 'the b o o k /s ' I t is im p o rta n t to n o tice th a t the generic d e fin ite a rtic le has been e xcluded because I o n ly in c lu d e d re fe re n tia l NPs (fo r a d e fin itio n o f generics, cf. D eclerck, 1986) 2) P reposed d e m o n stra tive : form s o f este, ese, an d aquel fro m the th re e -w a y d e ictic system in Spanish. B e lo w I p re se n t Sole and Sole's (1977: 309-310) d e fin itio n o f the three p o s s ib ilitie s ; C ha pter 2 p ro v id e d a m o re extensive account. Este: w h a t is near and closely connected w ith th e speaker, Ese: an object n o t to o fa r rem oved fro m the speaker, o r near the liste n e r an d close ly connected w ith th is person, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 aquel: w h a t is fu rth e r re m ove d fro m the speaker in tim e , space, o r th o u g h t, o r w h a t is rem ote fo r b o th the speaker and the person addressed. esta idea - estas ideas th is ide a - those ideas (fe m in in e ) este libro - estos libros th is b o o k - th a t b o o k (m asculine) esa idea - esas ideas th a t idea - those ideas (fe m in in e ) ese libro - esos libros th a t b o o k - those books (m asculine) aquella idea - aquellas ideas th a t idea o ve r th e re - those ideas o ve r there aquel libro - aquellos libros th a t bo ok o ve r there - those books o ve r there 3) P ostposed de m o n stra tive : la idea esta "this idea here", lit. "The idea th is '. A g a in , th e postposed d e m o n stra tive m ay appear in a ll genders an d n u m b e rs and in the th re e possible d e ictic fo rm s (este-ese- aquel) 4) D e m o n stra tive P ronoun: esta 'th is '. I have exclu ded the n e u te r d e m o n s tra tiv e p ro n o u n esto, because its reference is v e ry vague: in general, it refers to ideas o r to a g roup o f sentences, and therefore its re fe re n tia lity is d u b io u s.2 2For instance in (2) the first esto refers to all the problems that K had w ith a sofa she bought and she did not get at the right time, etc. It is not clear, how ever, what part of the conflict is the one she is referring to w hen she states a todo esto ’ w ith all this’. ( 2) K: la cosa es que fu i un dia para c- a todo esto yo me pelee con mi papa porque mi papa me dice vos estas muy alterada dice te veo muy nerviosa Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 5) P ersonal P ronoun: ella 'sh e ', el ’he’, ellas ’th e y fe m in in e ’, ellos ’th e y m a scu lin e ’ 6) D e fin ite N u ll Subject (DFNLSJ): pro piensa 'p ro th in k s ' The agreem ent o f the ve rb licenses the zero subject, o r so-called p ro d ro p . T herefore, it is the agreem ent th a t re quire s the re p re se n ta tio n o f an N P, and it is the agreem ent th a t is d ire c tly related to the antecedent. 7) C litic o f D ire ct and In d ire c t O bject: a) D ire c t O bject: Ella Z o piensa 'She th in k s it ' b) In d ire c t O bject: Le dijo "h e /sh e to ld h im /h e r’. 3.5.2. V a ria b le 2: C o n te xt (P-set) T h is va ria b le codifies the source o f in te rp re ta tio n fo r d e fin ite expressions. D efiniteness, as I w ill de fine it in chapter 4, re quire s the d e lim ita tio n o f a P-set and uniqueness ho lds w ith in the lim its o f th a t P-set. The fo llo w in g varia nts capture the levels o f a cce ssib ility th a t I propose in C ha pter 4. 1) A n a p h o ric same IU , same discourse u n it no se que te esta pasando desde que claro yo con esto parece una estupidez K: the thing is that I went one day to c- with all this (at the same time all this was happening) I got into a fight with my father because my dad says you are very distressed I don't know what is going on zvith you since of course I with this (neuter demonstrative pronoun) it seems stupid Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 2) A n a p h o ric p re v io u s IU , same disco urse u n it 3) A n a p h o ric p re vio u s 5 IU s, same discourse u n it 4) A n a p h o ric p rio r to 5 IU s, same discourse u n it 5) R elative clause 6) P re p o sitio n a l Phrase 7) A d je ctive 8) D eictic 9) A n a p h o ric, d iffe re n t discourse u n it 10) A ssociative A n a p h o ra 11) La rge r s itu a tio n 12) U niverse o f D iscourse V ariants 1) to 4) correspond to a c u rre n t discourse u n it (see chapter 4 fo r an e xp la n a tio n ), o r those P-sets d e lim ite d in the p re vio u s text. I d iv id e d the p re vio u s te xt in to sub-sets acco rdin g to distance. V a ria n t 1) refers to those cases w here the antecedent is located in the same IU . For instance in exam ple (3) the antecedent o f the n u ll subject me miraba 'he w as lo o k in g a t m e' (in lin e 2), is the n u ll subject o f me huia 'he was a v o id in g m e' w h ic h is located in the same IU (lin e 2). (3) 1 El pibe me miraba me huia me miraba con una cara me miraba asi como diciendo the g u y w as lo o k in g a t me he was a v o id in g m e lo o k in g a t m e w ith th a t w e ird lo o k he was lo o k in g at m e as he was sa yin g V a ria n t 2) refers to those cases w he re th e antecedent is located in the previous IU . For instance, in exam ple (3), the antecedent o f the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 n u ll subject o f the verb hirfa '(h e ) w as a v o id in g ' is 'the b o y ’, lo ca te d in the p re v io u s IU . V a ria n t 3) refers to those cases w h e re the antecedent is lo ca te d in the p rio r 5 IU s (except fo r th e p re v io u s IU ). F or instance, in e xa m ple (14) b e lo w (p . 53) the antecedent o f esa plata 'th a t m oney' is lo ca te d 2 IU s p rio r to the anaphor. V a ria n t 4) refers to those cases w h e re the antecedent is loca ted p rio r to 5 IU s. F or instance in exam ple (4) the n u ll subject o f th e ve rb iba in lin e 7 refers back to a re fe re n t lo ca te d 6 IU s p rio r to th e a n a p h o r, i.e ., mi Ho 'm y un cle' in lin e 1. E ven th o u g h the antecedent is d is ta n t, it is v e ry accessible since it belongs to th e same discourse u n it. (4) G: segun una charla que tuve despues con mi Ho era para que yo vaya hable y concr- digamos un nuevo plazo para pagar M q : claro G: me pro iba a dar un nuevo plazo G: a cco rdin g to a ta lk I ha d w ith m y u n cle la te r it w as so th a t I go ta lk an d set u p - I mean a n e w in s ta llm e n t p la n to p a y M q : o f course G: fh e l w as g o in g to g ive m e a n e w de a d lin e V a ria n ts (5) to (7) re fe r to those cases w here the d e fin ite N P is in te rp re te d in a P-set d e lim ite d in the fo llo w in g text. There are three p ossible types: 5), the re la tive clause; 6), the p re p o sitio n a l phrase; and 7) th e a d je ctive . For instance, in exa m ple (5) que da el curso 'w h o is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T — I N C O 61 teaching the class' is d e lim itin g th e P-set w he re la mina esta 'the w om a n th is ' s h o u ld be in te rp re te d u n iq u e ly . (5) la mina esta que da el curso (5') 'the w o m a n th is w h o is tea ching the class' V a ria n t 6) refers to those cases w here the P-set is d e lim ite d in a p re p o s itio n a l phrase. F or instance in (6) lin e 2, th e p re p o s itio n a l p h rase del gamexane 'o f gam exane' is d e lim itin g th e P-set w here the d e fin ite N P la vez esa 'th e tim e th a t' sh o u ld be in te rp re te d u n iq u e ly . (6) F: la vez la vez esa del gamexane esa vez era fin de ano F: the tim e the tim e th a t o f the gam exane th a t tim e w as N e w Year D ay I t is im p o rta n t to n o tice here th a t la vez 'th e tim e ' in lin e (1) is n o t considered in the c o d ific a tio n because it is a false sta rt. The speaker (F) realizes th a t the d e fin ite a rtic le is n o t enough to in te rp re t the re fe re n t o f vez 'tim e '. F needs to use a postposed d e m o n stra tive w h ich , as I w ill show in chapter 4, refers to non-accessible P-sets. V a ria n t 7) refers to those cases w here the P-set is d e lim ite d b y an adjective. In the d e fin ite N P el trabajo intelectual, the a d jective intelectual is d e lim itin g th e P-set w h e re trabajo 'w o rk ' is u n ique . V a ria n t 8) refers to those re fe re n tia l expressions w hose re fe re n t is present in the speech s itu a tio n (d e ictic usage). In th is case, references to p e riod s o f tim e such as un dia, el ano pasado, etc. have been also c o d ifie d as d e ic tic usages. In (7), fo r instance, this table and that piece of furniture are present in the s itu a tio n o f speech. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 (7) K : el total eran dos mil quinientos M q: que que que son? esta esta mesa y ese modular? (7 ’) K : the to ta l w as tw o thousand and fiv e h u n d re d M q : w h a t w h a t w h a t is it th is th is table and th a t piece o f fu rn itu re ? V a ria n t (9) refers to those cases w h e re , even th o u g h the P-set is d e lim ite d in the p re vio u s discourse, the antecedent in located in a n o n activa ted discourse u n it. T h is is e x e m p lifie d and explained in greater d e ta il in cha pte r 4. V a ria n t 10, associative anaphora, is re la te d to those e n titie s th a t are associated b y v irtu e o f fram es o r scrip ts (M in s k y , 1975). For instance, in the fo llo w in g exam ple, I bought a book. The author is very famous., the P-set consists o f those e n titie s re la te d to 'a b o o k', in th is case 'a b o o k has an a u th o r'. I f the speaker and liste n e r can id e n tify the re fe re n t o f the N P b y v irtu e o f th e ir com m on m em bership in a c o m m u n ity (fo r instance The president resigned), the P-set consists o f the e n titie s in c lu d e d in a la rg e r s itu a tio n — V a ria n t 11). F in a lly , th e least re stricte d P-set (V a ria n t 12) is the u n ive rse o f discourse P-set th a t includ es a ll the possible en tities th a t m a y be re fe rre d to. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 3.5.3. Variable 3: Syntactic Position The va ria b le 'syn ta ctic p o s itio n ' codes the s yn ta ctic p o s itio n fo r the anaphor, the antecedent a n d the co m p e tito r. The fo llo w in g va ria n ts w ere in c lu d e d in th is va ria b le : 1) M a trix /e m b e d d e d subject. See exam ple (8) b e lo w fo r m a trix (esa vez ’th a t tim e ’) and (9) b e lo w fo r em bedded (el ’h e ’)- 2) M a trix /e m b e d d e d d ire c t object. See exam ple (10) b e lo w fo r m a trix (la plata del alquiler ’the m o n e y o f the re n t’) a n d (11) b e lo w fo r em bedded (el curso 'th e class') 3) M a trix /e m b e d d e d in d ire c t object. See exam ple (12) b e lo w fo r m a trix (a los profesores 'to the teachers') and (13) b e lo w fo r em bedded. 4) M a trix /e m b e d d e d P re p o sitio n a l Phrase. See exam ple (14) b e low fo r m a trix (de esa plata 'o f th a t m oney') and (15) b e lo w fo r em bedded (de los salones 'o f th e classroom s' an d en la mesa 'o n the ta b le '). (8) esa vez era fin de ano 'th a t tim e was N e w Y ear's Eve' (9) porque el no nos dijo pagame 'because he d id n o t say to us p a y m e' (10) yo tengo la p la ta del alquiler 'I have the m o n e y o f the re n t' (11) que da el curso 'w h o teaches the class' (12) le tiraban a los pro fesores 'th e y th ro u g h to the teachers' (13) porque le compraban a los profesores (in v e n te d exam ple) (14) yo puedo disponer de esa pla ta ' I can dispose o f th a t m oney' (15) las puertas de los salones 'the doors o f th e classroom s' se paraba en la m esa 'he w o u ld sta n d u p o n th e ta b le ' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 3.5.4. V a ria b le 4: F requency o f M e n tio n The v a ria b le frequency o f m e n tio n codes the n u m b e r o f occurrences o f the e n tity re fe rre d to b y th e an aphor w ith in th e 25 p rio r IU s. F or those cases w here there are o th e r p o te n tia l antecedents, the n u m b e r o f m e n tio n s o f the c o m p e tito r w ith in the 25 p rio r IU s are also coded. F o r th is fa cto r, G ivo n (1983a: 13) considers 20 clauses to the le ft o f the anaphor. I added 5 m ore to the n u m b e r 20 because IU s are s h o rte r th a n clauses. In general, sentences are com prised o f m o re th a n one IU . 3.5.5. V a ria b le 5: N u m b e r o f W o rd s T h is v a ria b le codes th e n u m b e r o f w o rd s o f the re le v a n t N P . A s A rie l (1990) claim s, lon ger NPs w ill re fe r to less accessible e n titie s . Since e n titie s are n o t re a d ily accessible, th e speaker needs to p ro v id e m ore in fo rm a tio n to the liste n e r w ith respect to the e n tity b e in g re fe rre d to . F o r instance the N P los mitos que hay en la sociedad 'th e m y th th a t w e can fin d in society' consists o f 7 w o rd s , w h ile el patriarcado 'th e p a tria rc h y ' consists o f 2 w o rd s. 3.5.6. V a ria b le 6: T h e m a tic R ole The v a ria b le 'th e m a tic ro le ' codes th e th e m a tic ro le fo r the an aph or, th e antecedent and th e c o m p e tito r. The fo llo w in g v a ria n ts w ere in c lu d e d in th is varia ble: 1) A gent-E xperiencer: those cases w h e re the e n tity is the in s tig a to r o f th e a ctio n (I wrote the letter) o r th e experiencer o f a fe e lin g o r state (JJike Spanish) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 2) Them es: those e n titie s affected b y an oth er p a rtic ip a n t (e.g., I wrote the letter) 3) O th e r: the re m a in in g them atic roles, (e.g., in s tru m e n t: I wrote the letter with m y pen) 3.5.7. V a ria b le 7: D ista n ce T h is va ria b le codes the distance fro m the a n a p h o r to the antecedent and the c o m p e tito r, m easured in term s o f IU s. 3.6. T h e A n a lyse s A ll these va ria b le s an d v a ria n ts w ere entered in to SPSS (S ta tistica l Package fo r S ocial Sciences), to m a n ip u la te th e data m ore e a sily. F o r chapter 5 , 1 ca rrie d o u t an A p p lie d L o g is tic R egression. In the fo llo w in g sections I describe the d iffe re n t analyses th a t I conducted in th is d isse rta tio n . 3.6.1 T e xtu a l u n its F or the analysis o f te x tu a l P-sets, I asked three d iffe re n t n a tive speakers o f A rg e n tin e S panish to d iv id e d iffe re n t texts in to discourse u n its . The jud ges are tw o v e ry h ig h ly educated y o u n g w o m e n and m yse lf. I p ro v id e d them w ith a ll the tra n scrip tio n s o f m y in te rv ie w s . The in s tru c tio n s w ere in te n tio n a lly vague: 'please d iv id e the fo llo w in g texts in to d iffe re n t pieces o f an y so rt.' If the y w ere n o t sure, I encouraged them to set u p th e ir o w n crite ria . A ll texts w e re d iv id e d by a ll three jud ges. A fte r th e y re tu rn e d the d iv id e d te xts to m e, I selected o u t those u n its th a t w ere d e lim ite d b y lin g u is tic in d ic a to rs , such as D M s. The re sults are presented in chapter 4 (4.2.2). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 3.6.2. A p p ro p ria te n e ss C o n d itio n s F or the q u a n tita tiv e an alysis presented in section 4.3 (the appropriateness co n d itio n s o f d e fin ite n e ss types) I recoded the va ria b le 2 (context) in the fo llo w in g w a y. V ariants 1-4 w ere g ro u p e d to g e th e r and re-coded as: anaphoric, sam e discourse u n it. V ariants 5-6 w ere g ro u p e d to g e th e r and w ere re-coded as: fo llo w in g text. V a ria n t 7 was n o t con sidered because it was ve ry in fre q u e n t. V ariants 8-12 w ere n o t changed. Each one w as in d e p e n d e n tly coded as: 8) D eictic 9) A n a p h o ric, d iffe re n t disco urse u n it 10) A ssociative A n a p h o ra 11) Larger s itu a tio n 12) U niverse o f D iscourse 3.6.3 A p p lie d lo g is tic re g re ssio n F or chapter 5 ,1 conducted a a p p lie d regression analysis because it a llo w s us to see the in flu e n ce o f each va ria b le in the selection o f the d ependent variable. In lin e a r re g re ssio n , the least-squares m e thod is used to estim ate the param eters o f th e m odel. In other w o rd s, w e select regression coefficients th a t re s u lt in th e sm allest sum s o f squared distances betw een the observed a n d the p re d icte d values o f the dependent variable fo r the set o f data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 In lo g is tic regression, the param eters o f the m o d e l are estim ated u sin g the m a x im u m -lik e lih o o d m e th o d . T h a t is, the co e fficie n ts th a t m ake o u r observed results m ost 'lik e ly ' are selected (N o ru sis, 1994:3). To in te rp re t the results o f a lo g is tic regression analysis, it is im p o rta n t to u n derstan d tha t: 1) L o g is tic Regression a n ticipate s sig m o id e a l o r S cu rve re la tio n sh ip s ra th e r than lin e a r re la tio n s h ip s . In a sig m o id e a l re la tio n s h ip th e p ro b a b ilitie s o f an e v e n t o c c u rrin g change a t sm a lle r increm ents as the values o f the in d e p e n d e n t v a ria b le reach th e ir h ig h and lo w values (Sweet, 1999). 2) The c o n s ta n t valu e (Beta zero) w ill te ll us the p o in t at w h ic h the regression lin e is p re d icte d to to u c h the Y axis (in o th e r w o rd s the p re d icte d va lu e o f Y w hen X=0). 3) The C o e ffic ie n ts values (Betas 1 to n ) w ill p ro v id e us w ith the in fo rm a tio n a b o u t the slope o f the re gre ssio n lin e . The slope ind ica tes the am o u n t o f change in the d e pen den t v a ria b le (Y) w h ic h is associated w ith a one u n it change in the in d e p e n d e n t va ria b le X. M o re d e ta ils and the results o f th is analysis are presented in chapter 5 (5.3.4) 3.7 C o n clu sio n s In th is chapter the m e th o d o lo g y fo r the q u a n tita tiv e analyses conducted in th is d isse rta tio n w ere presented. The n a tu re o f the data, d e tails a b out the subjects, the s tru c tu re o f the in te rv ie w as w e ll as tra n s c rip tio n conventions w ere de scribe d. F in a lly , the va ria b le s and va ria n ts used in the d iffe re n t analyses w e re discussed w ith exam ples fro m m y corpus. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 Chapter 4 D efining D efiniteness Types 4.0. Introduction In th is ch a p te r I p ro v id e a d e fin itio n o f d e finiten ess based on the n o tio n o f P-sets a n d uniqueness. The n o tio n o f a P-set w h ic h o rig in a te d in the p ra g m a tic tra d itio n is exte nd ed and a p p lie d to n a tu ra l discourse, c o n trib u tin g to the n o tio n o f disco urse re pre senta tion. The n o tio n o f uniqueness is take n fro m the lo g ic a l tra d itio n (R ussell, 1905; H a w kin s, 1991) an d in te rp re te d in the lig h t o f an e m p iric a l account. The ch a p te r w ill be organized as fo llo w s : firs t, in section 4.1., I w ill describe th e n o tio n o f de finiteness, s u p p le m e n tin g the lo g ic a l/p ra g m a tic approach w ith an an alysis o f n a tu ra l data. M y d e fin itio n o f d e fin ite n e ss types in vo lve s a disco urse re p re se n ta tio n th a t is co n stitu te d b y p ra g m a tic a lly s tru c tu re d sets o f e n titie s w ith d iffe re n t levels o f a cce ssib ility (cf. section 4.2.2.). A n im p o rta n t p a rt o f this cha pte r is a com prehensive account o f th e n o tio n o f discourse re pre senta tion focused o n the s h iftin g o f disco urse u n its (cf. section 4.2.1.). In sectio n 4.3., I present the a p p ro pria tene ss co n d itio n s o f the definiteness types based o n the analysis o f n a tu ra l data. In section 4.4., I p ro v id e c o g n itiv e m o tiv a tio n s fo r the a p p ro p ria te n e ss co n d itio n s o f these d e finiten ess types. 4.1. The definition of definiteness In th is ch a p te r I w ill p ro v id e an acco unt o f the fo llo w in g definiteness types: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 D e fin ite A rtic le (D e fA rt+ N ) Preposed D em o nstrative (D e m + N ) Postposed D em onstrative (D e fA rt+ N + D e m ) D e m o n stra tive P ro n o u n (D e m ) P ersonal P ronoun (Pers P ro n o u n ) (1) la idea 'th e id e a ' (2) esta idea 'th is id e a ' (3) la idea esta 'th is idea here' lit. 'T h e idea th is ' (4) esta 'th is ' (5) ella 'she ' D e fin ite N u ll Subject (D e f N u ll S u b j) (6) pro pie nsa. 'p ro th in k s ' C litic o f D O and IO (C litic ) (7) Ella h piensa ' She th in k s i t ' F o llo w in g H a w k in s (1991) d e finiten ess is defined as fo llo w s : each tim e speakers use a de finiten ess typ e to re fer to an e n tity th e y d e lim it a p ra g m a tica lly s tru c tu re d set o f e n titie s (P-set) w h e re th e e n tity is u n iq u e . F o r instance in (8) el pibe 'th e b o y ' and la mujer 'th e w o m a n ' are re fe rrin g to the u n iq u e e n tity w ith in a p re vio u s text P-set. In o th e r w o rd s, in the p reviou s te xt P-set the re is o n ly one bo y and one w om a n. the b o y studies g raph ic design and the w om an was v e ry w e ll k n o w n (h e /sh e ) had a lo t o f experience 4.1.1. The d e lim ita tio n o f a P-set A P-set (as show n in cha pte r 2) is a p ra g m a tica lly s tru c tu re d set o f e n titie s w ith in the o v e ra ll u n ive rse o f discourse, w h ic h de fines the param eters fo r the uniqueness re q u ire m e n t. H a w kin s (1991) asserts th a t the uniqueness re q u ire m e n t fo r the d e fin ite a rticle is re s tric te d to a p ra g m a tic set. In oth er w o rd s, speakers d e lim it a P-set and th e n th e y re fe r to a u n ique e n tity w ith in th e lim its o f th a t P-set. el pibe estudia disefio grafico y la mujer era m uy conocida D e f N u ll Subj tenia mucha experiencia Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 The p roble m w ith the n o tio n o f P-set is th a t it is v e ry loosely d e fin e d . For instance, w h a t co u ld , in p rin c ip le , co n stitu te a pragm atic d e lim ita tio n o f e n titie s w h e re the re q u ire m e n t o f uniqueness can hold? Even th o u g h a ll the d iffe re n t types o f P-sets th a t have been discussed b y H a w kin s (1991) are clear I w ill try , in th is dissertation, to d e fin e some a d d itio n a l types o f s tru c tu rin g , s p e c ific a lly those w h ic h are discourse sensitive. I, the refore , set u p the c rite ria to d e fin e te xtual boundaries and I describe the in te rn a l p ro p e rtie s o f these te xtu a l P-sets (cf. section 4.2.2). 4.1.2. The u n ique ness re q u ire m e n t The uniqueness re q u ire m e n t is a s u ffic ie n t c o n d itio n fo r a ll d e fin ite NPs. H ow ever,’ uniqueness is n o t a necessary co n d itio n : there are exam ples o f w h a t B im e r and W a rd (1994) c a ll irre le v a n t u n iq u e n e ss, and fo r som e definiteness types, uniqueness is achieved th ro u g h prom inence. I w ill discuss each o f these cases in w h ich d e fin ite references are n o t te c h n ic a lly un ique , w h e n I present the appropriateness co n d itio n s fo r the d iffe re n t definiteness types. In the fo llo w in g exam ple (9) it is in te re stin g to observe h o w the speakers are aw are o f the im p lic a tio n o f uniqueness w h e n the D e fA rt+ N is used. W hen re fe rrin g to 'the school' th e y are im p ly in g th a t it is the o n ly one, a statem ent th a t is o b vio u sly false (there is m ore th a n one school in Buenos A ire s!). I t is true, h o w e ve r, th a t this specific school is the best in the c ity , and therefore it is u n iq u e in a sense. The speaker, C, in (9) in a w a y, challenges the im p lic a tio n th a t th is specific school is the o n ly one (o r the best one) b y asking an a p p a re n tly genuine w h -q u e stio n : Perdon, que colegio? 'Excuse m e, w h a t school?' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in lin e 12 ( it is obvious fo r a ll p a rtic ip a n ts w h ic h school th e y are ta lk in g about). M o reo ver, speaker E is aw are o f the uniqueness re q u ire m e n t w hen u sin g the D e fA rt+ N , w h e n he e x p lic itly states si fuera el unico 'as it (the school) w as th e o n ly one' in lin e 20. (9) 1 M : Yo estudio b io lo g ia 2 basicam ente 3 digam os a e l lo conozco d e l colegio (...') 4 E: Yo soy Esteban 5 lo conozco a M a rtin d e l colegio 6 hacemos cosas parecidas 7 M q: ,;Oue hace la gente que se recibe en el colegio 8 aparte de Uds? 9 Yo tenia la idea de que el colegio era mas b ie n 10 tira n d o a las h u m a n id a d e s. 11 M : Esa es la idea que tie n e n m uchos. 12 C: Perdon. £que colegio? 13 M : El Buenos A ire s 14 (...) 15 M : ademas vos ju n ta s a dos personas del colegio 16 y se pregu nta n 17 £ vos estudiabas en e l colegio? 18 A p a rte en el colegio 19 porque el colegio es cu a lq u ie ra 20 E: com o si fue ra e l u n ic o 21 M : me encuentro con com paheros de la fa cu lta d 22 p o r a h i m e ven con e l y dice n 23 ah ahi esta e l o tro que es d e l colegio. 24 (9 ') del un ico colegio que se vo. 1 M : I stu d y B io lo g y 2 ba sically le t's say th a t I k n o w h im fro m the school 3 E: I am Esteban. I k n o w M a rtin fro m the school, w e do 4 s im ila r th in g s 5 M q: W hat do people w h o gra d u a te d fro m the school do. 6 besides yo u guys? 7 I had the idea th a t th e school w as specialized 8 in H u m a n itie s . 9 M : That is the idea th a t a lo t o f people have 10 C: Excuse m e, w h a t school? 11 M : The 'Buenos A ire s ' 12 (...) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 13 M : besides y o u p u t tw o people fro m th e school to g e th e r 14 and th e y ask to each o th e r 15 d id y o u go to th e sch o o l? 16 Besides, to the school, 17 because the school is a n y school 18 E: as it was the o n ly one 19 M : I ru n in to u n iv e rs ity classm ates 20 and m aybe th e y see us tog eth er and they say 21 th e re it is the o th e r one fro m th e school 22 the o n ly school w h a t do I know ? 4.2. P-sets A P-set is a p ra g m a tic a lly s tru c tu re d set o f e n titie s w ith in the o v e ra ll u n ive rse o f discourse w h ic h defines the param eters fo r the re q u ire m e n t o f uniqueness (H a w k in s , 1991). I w ill extend the n o tio n o f a P-set p a rtic u la rly to te x tu a l P-sets w ith in th is e m p iric a l s tu d y . The fo llo w in g P-sets are considered fo r the d e fin itio n o f de finiten ess types, fo llo w in g H a w k in s (1978, 1991) (cf. C hapter 2.3): 1) the p re vio u s te x t (d iv id e d in to discourse u n its as e xp la in e d in section 4.2.1. be low ) 2) the P-set d e lim ite d b y a p re p o s itio n a l phrase o r re la tiv e clause 3) th e im m e d ia te s itu a tio n o f uttera nce set 4) th e la rg e r o r spe cific s itu a tio n set 5) th e set o f associated e n titie s 6) the w h o le u n iverse o f discourse, co m p ris in g a ll possible entities th a t anyone m ay re fe r to. To the q u estion w h y do speakers select one typ e o f N P ra th e r th a n a n oth er, I w ill answ er: in o rd e r to re fer to P-sets w ith d iffe re n t le v e l o f acce ssibility. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 4.2.1. D efiniteness types refer to P-sets of different levels of accessibility G ive n the m e n ta l re presentation o f texts as discussed in chapter 2 , 1 assum e th a t as speakers are ta lk in g , d iffe re n t e n titie s (o r m e ntal representations o f the e n titie s) have m ore o r less a c tiv a tio n /a c c e s s ib ility . The m e n ta l re pre senta tion o f a te x t is h ie ra rc h ic a lly organized, because e n titie s are m ore o r less accessible d e p e n d in g on the con textu al sources (P-sets) they b e long to . F o r instance, e n titie s b e lo n g in g to the p re v io u s te x t have one le v e l o f a c tiv a tio n and en titie s b e lo n g in g to the p h ysica l s itu a tio n have a d iffe re n t le ve l o f a ctiva tio n . E n titie s w ith in these P-sets are also ra nked according to accessibility. I t is im p o rta n t to d is tin g u is h the a cce ssib ility o f P-sets fro m the acce ssibility o f en tities. D e fA rt+ N , D em + N , D e fA rt+ N + D e m and D em re fe r to a u n iq u e e n tity w ith in a P-set, and the refore the re la tiv e a cce ssib ility o f e n titie s is n o t v e ry im p o rta n t. Som etim es, Pers P ron oun , C litic and D e f N u ll Subj re fe r te ch n ica lly to a n o n -u n iq u e e n tity , i.e. th e y re fe r to the m ost p ro m in e n t e n tity w ith in the d e lim ite d P-set. Therefore, a com pariso n betw een the d iffe re n t levels o f a cce ssib ility o f e n titie s is re q u ire d . Three m ajor degrees o f a cce ssib ility can be d istin g u ish e d fo r P- sets: H ig h accessibility P-sets (the p re vio u s te x t set [c u rre n t discourse u n it]1 and the p h ysica l s itu a tio n set), m id accessibility P-sets (the p re vio u s te x t [the antecedent is located w ith in a d iffe re n t discourse u n it], the set d e lim ite d b y the fo llo w in g te xt), and lo w accessible P-sets (the set o f associated e n titie s, the la rg e r s itu a tio n set, the u n ive rse o f 1 The notion of a discourse unit is further eUaborated in section 4.2.2. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 discourse set). In the fo llo w in g sections, I ju s tify th is ra n k in g according to degrees o f accessibility. 4.2.1.1. H ig h a cce ssib ility P-sets A s speakers are ta lk in g , w h a t they hear (the e n titie s th a t are m e n tio n e d in the p re vio u s te xt, the discourse u n it in w h ic h the d e fin ite N P is located) and w h a t they see (the e n titie s th a t are located in the p h ysica l s itu a tio n P-set) are the m ost accessible P-sets. There is some co n tro ve rsy a b o u t w h ic h o f these P-sets has a h ig h e r le ve l o f acce ssibility (cf. G ivo n , 1992; A rie l, 1990). I cla im th a t th e spe cific degree o f a cce ssib ility o f these tw o kin d s o f P-sets depends o n the s itu a tio n and on the genre o f the text. F or instance, if p a rtic ip a n ts are ta lk in g about the pieces o f fu rn itu re in a house in w h ic h th e y are b o th p h y s ic a lly located, the p h ysica l s itu a tio n w ill be m ore accessible th a n the te xt, b u t if the y are ta lk in g on the phone, the ph ysical s itu a tio n w ill be less a ctiva ted . 4.2.1.2. M id a cce ssib ility P-sets The p re vio u s te xt set th a t constitutes a d iffe re n t discourse u n it tha n the one w here the re fe re n t o f a d e fin ite N P is lo ca te d is less accessible th a n the p re vio u s te xt set th a t constitutes th e same discourse u n it in w h ic h the d e fin ite N P is located (i.e. the c u rre n t discourse u n it). The p o rtio n o f the p re vio u s te x t th a t constitutes a d iffe re n t discourse u n it is n o t as a ctiva te d as the discourse u n it th a t is b e in g processed at the same tim e th a t th e definiteness type is p rodu ced. T he p re vio u s te x t set th a t belongs to a d iffe re n t u n it is less accessible th a n th e p h ysica l s itu a tio n set, the set o f en titie s th a t the p a rticip a n ts can a c tu a lly see. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 The P-set d e lim ite d b y a fo llo w in g p re p o s itio n a l phrase o r a re la tiv e clause is processed w ith a de la y in com pariso n to th e h ig h accessible P-sets. F or instance, in the fo llo w in g exam ple (m e n tio n e d in 2.3): Pass me the bucket which is over there, th e re la tiv e clause w hich is over there is processed after the processing o f the N P the bucket, and th e re fo re w ith a d e la y in com parison to h ig h accessible P-sets. In co n tra st, e n titie s m e n tio n e d in the p re vio u s te x t are a lre a d y processed and a va ila b le fo r reference in te rp re ta tio n . A g a in , even th o u g h it is n o t clear w h ic h o f these tw o P-sets is the m o re accessible one (p re vio u s te xt, d iffe re n t discourse u n it; fo llo w in g te xt), i t is clea r th a t th e y are less accessible th a n th e h ig h accessible P-sets. 4.2.I.3. L o w a cce ssib ility P-sets F in a lly , there are lo w a cce ssib ility P-sets. These in c lu d e the associated e n titie s set, the la rg e r s itu a tio n set and the u n ive rse o f discourse set. F irst, the P-set o f associated e n titie s is a lo w a cce ssib ility P-set. In o rd e r to relate tw o e n titie s b y fram es o r scripts (M in s k y , 1975) it is necessary to m ake an inference. For instance, in the exa m ple I bought a book. The author is great, w e re fe r u n iq u e ly to the a u th o r o f the b o o k, because w e in fe r th a t s te re o ty p ic a lly a book has a u n iq u e a u th o r. Second, there are some P-sets d e lim ite d b y v irtu e o f the com m on m em bership o f the p a rtic ip a n ts in a la rg e r c o m m u n ity , fo r instance tw o citize n s o f the same c o u n try can say the president, re fe rrin g to th e o n ly p re sid e n t in th a t c o u n try . W h ile h ig h a n d m id accessible P-sets can be d e lim ite d b y any tw o p a rtic ip a n ts , th is lo w e r Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 a cce ssib ility P-set re q u ire s ce rta in com m on kn o w le d g e betw een p a rtic ip a n ts . F in a lly , the to ta l u n iverse o f discourse is the least accessible P-set because it in clu d e s a ll the p o ssible e n titie s th a t can be re fe rre d to , i.e. w h e n an e n tity is u n iq u e in ab solute term s. S um m ing In the n e xt section, I w ill define ce rta in a d d itio n a l types o f P-sets th a t are discourse sen sitive. I w ill p ro v id e the c rite ria to recognize te xtu a l b o und arie s and I w ill describe th e ir in te rn a l p ro p e rtie s. 4.2.2. T e xtu a l P-sets In th is section, I w ill d e fin e te xtu a l P-sets. A s m e n tio n e d in chapter 2, discourse s tru c tu re p la ys an im p o rta n t ro le in d e lim itin g the reference o f ce rta in d e fin ite n e ss types. Q uestions th a t need to be addressed are: w h a t co n stitu te s a discourse u n it (o r te x tu a l P-set)? W hat m akes a discourse u n it d iffe re n t fro m a n o th e r and h o w it is recognized w h e n one ends a n d another up: A c c e s s ib ility o f P-sets • h ig h a cce ssib ility P-sets: • lo w acce ssibility P-sets: • m id a cce ssib ility P-sets: Types o f P-set the p re v io u s te x t (antecedent in the sam e P-set as anaphor) the p h y s ic a l s itu a tio n set the p re v io u s te x t (antecedent w ith in a d if f . discourse u n it) the set d e lim ite d b y the fo llo w in g te x t the set o f associated e n titie s the la rg e r s itu a tio n set the u n ive rse o f discourse set Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 starts? Several scholars ta lk ab out 'a ch u n k o f discourse', o r 'a piece o f te x t', 'episodes', etc. (cf. 2.4.1). There are tw o p o in ts th a t I need to m ake here a b o u t these chunks or p o rtio n o f texts. F irst, discourse u n its occu py a ce rta in p o rtio n o f a te x t th a t are som etim es d e lim ite d b y lin g u is tic in d ic a to rs . F in d in g lin g u is tic in d ic a to rs th a t alw ays signal a change o f discourse u n it con tribu tes a m eans to develop an a p p ro p ria te m e th o d fo r d e fin in g discourse s tru ctu re . In m y data fo u r factors have been fo u n d re le va n t fo r m a rk in g the s h iftin g o f discourse u n its : th e occurrence o f discourse m a rkers, u n it openers, questions b y the in te rv ie w e r, and quotes. The occurrence o f a discourse m a rker o r a u n it opener, and a qu estion o r a com m en t b y the in te rv ie w e r are s u ffic ie n t c o n d itio n s to s ta rt a new discourse u n it. The occurrence o f a quote is n e ith e r a necessary n o r a s u ffic ie n t c o n d itio n fo r a discourse u n it s h ift in m y corpus. This fa cto r is presented anyw ay, because th is d is trib u tio n m ig h t be an a rtifa c t o f m y data and m ay n o t be true fo r o th e r corpora. Second, these discourse u n its have in trin s ic p ro p e rtie s, fo r instance, v e ry ofte n th e y have the same subject, tense, and aspect fo rm s. H o w e ve r, none o f these factors c o n s titu te s u ffic ie n t co n d itio n s to re co g n izin g a discourse u n it, even th o u g h th e y are h ig h ly correlated, i.e. a change o f the subject, tense o r aspect fo rm s co-occurs, in general, w ith a s h iftin g o f a discourse u n it. F o r m y analysis o f the d e lim ita tio n o f discourse u n its , three d iffe re n t judges (n a tive speakers o f S panish2) in c lu d in g m yse lf in d e p e n d e n tly d iv id e d d iffe re n t pieces o f te x t in to discourse u n its. 2 Cf. chapter 3 for details on the experiment. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 Then I exam ined how o fte n th e ir ju d g m e n ts o f sh ifts agreed w ith m ine, i.e. w h a t percentage o f the b o u n d a rie s I no ted w ere also noted b y them . The id e n tific a tio n o f the lin g u is tic in d ic a to rs th a t lim it discourse u n its coincided 85% o f th e tim e . Some o f the judges m ade a d d itio n a l m a rkin g s w h ic h w ere n o t sig n a le d lin g u is tic a lly , b u t I o n ly considered those w ith lin g u is tic in d ic a to rs . In the fo llo w in g section, I present the lin g u is tic in d ica to rs th a t th e judges agreed to be discourse u n it bo und arie s. Table 2 (below ) show s the n u m b e r o f judges w h o agreed fo r each s h ift (second colum n: consensus). T h e table also show s the lin g u is tic in d ica to rs th a t co-occur w ith a s h ift o f u n it o f discourse: n u m b e r o f discourse m arkers o r co m b in a tio n w ith u n it openers, questions b y the in te rv ie w e r, quotes, change o f subject a n d change o f tense and aspect. 4.2.2.I. L in g u is tic In d ica to rs o f D isco u rse U n its: D iscourse M a rkers S c h iffrin (1987) defines disco urse m a rkers as lin g u is tic elem ents th a t bracket u n its o f ta lk. The k in d s o f u n its d e lim ite d b y discourse m arkers in clu d e turns at ta lk , speech acts, and te xtu a l discourse u n its. For instance, B rizuela (1992) show s h o w discourse m arkers d e lim it the stru ctu re o f a n a rra tive , i.e. a te xtu a l d isco urse u n its, and A nde rsen et al. (1995) and B rizuela e t al. (1999) sh o w h o w discourse m arkers m a rk d iffe re n t registers o f speech. The fo llo w in g com binations o f disco urse m arkers appear o n ly if there is a s h ift o f discourse u n it: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 y bueh de ahi 'a n d w e ll fro m there' y bueno despues 'a n d w e ll la te r’ por ejemplo 'fo r in s ta n c e ’ entonces bueno ’th e n w e ll’ bueno entonces (aparte)'well th e n ’ y bueno despues' a n d w e ll la te r’ Table 2: Linguistic Indicators and Consensus for Shifting Discourse Units (Y = Yes; N = No)___________________________ N Discourse Unit Shift Cnsnsusof Judges Number of DM or comb, w ith unit opener Qstions or comments by the Intrvwer Quotes Change of Subject Change of tense and aspect 1 3 2 N N Y N 2 2 3 N N Y Y 3 3 3 N N Y Y 4 3 2 N N Y Y 5 3 0 Y N Y Y 6 3 2 N N Y Y 7 3 2 Y N Y Y 8 3 1 N N Y Y 9 3 2 N N Y Y 10 3 1 N Y Y Y 11 2 2 N N Y Y 12 2 2 N N Y Y 13 3 1 N N Y N 14 3 1 N N Y Y 15 3 2 N N Y N 16 1 3 N N Y Y 17 3 1 Y N Y Y 18 3 1 Y N N Y 19 3 3 Y N Y Y 20 3 3 N N N Y 21 3 1 N N Y Y 22 3 3 N N Y Y 23 2 0 Y N Y Y 24 2 2 N N Y N 25 2 1 Y N Y Y 26 3 1 N N Y Y 27 3 1 N N Y Y Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 4.2.2.2. L in g u is tic In d ic a to rs o f D isco u rse U n its : U n it O peners In m y data, I fo u n d the fo llo w in g lin g u is tic co n stru ctio n s th a t I c a ll 'u n it openers', appearing o n ly a t the b e g in n in g o f a discourse u n it: lo que pasa es que 'w h a t happens is' una vez 'once' un dia ’one d a y' el asunto es que ' the issue is th a t' la cosa es que 'th e th in g is th a t' de ahi ' 'fro m the re' A ll these co n structio ns are e ith e r c lo s in g a digression, i.e. the co n ve rsa tio n w e n t on to a n o th e r to p ic {el asunto es que, la cosa es que), o r fu rth e r e la b o ra tin g a to p ic {de ahi), o r o p e n in g a ju s tific a tio n {lo que pasa es que), o r s ta rtin g a s to ry {un dia, una vez). 4.2.2.3. L in g u is tic In d ic a to rs o f D isco u rse U n its: D iscourse m a rkers and u n it openers In m y data, I also fo u n d a c o m b in a tio n o f discourse m a rkers, and u n it openers: bueno el asunto es que 'w e ll th e issue is th a t' (bueno) la cosa es que (w e ll) th e th in g is th a t y bueh de ahi 'a n d w e ll fro m there' In tab le 2, (th ird colu m n) it is po ssible to observe the n u m b e r o f discourse m a rkers and u n it openers th a t co-occur w ith a change o f discourse u n it (1-2-3 or N i f there is none). In 93% o f the cases, a discourse m a rke r o r a u n it opener accom panies a discourse u n it s h ift (o n ly 2 cases do n o t have a discourse m a rk e r o r u n it opener). Therefore, the appearance o f a discourse m a rke r o r u n it opener, o r a co m b in a tio n o f b o th is a s u ffic ie n t in d ic a to r o f a discourse u n it s h ift. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 In the fo llo w in g exam ple (10) there are tw o discourse u n it sh ifts, the firs t one is m a rke d b y the co m b in a tio n o f discourse m arkers bueno entonces aparte in lin e s 7-8. The second one starts w ith the co m b in a tio n o f th e D iscourse M a rk e r bueno and the u n it opener el asunto es que (lin e 14). 1 (10) G: te das cuenta 2 cuando vos 3 M q: si s i te das cuenta 4 G: estas en u n a re u n io n 5 vie n e n y estan con cara de o rto 6 M q: s i te das cuenta 7 G: b u e n o entonces 8 K : a p arte se arm o u n despelote trem endo 9 esta to d a la fa m ilia peleada 10 G: @ @ p o r ese tem a 11 bueh p e ro eso es aparte 12 eso es u n tem a aparte 13 D e f N u ll Subj no tien e nada que ve r 14 b u eno e l asunto es que com o p u d im o s juntam os la plata... 15 e l asunto que 16 M q: le pagaste tod o de una vez 1 (10') G: yo u can see it 2 w h e n y o u 3 M q : yes yes y o u can see it 4 G: y o u are in a get together 5 the y com e and th e y are pissed 6 M q: yes y o u can see it 7 G: w e ll th e n 8 K : besides w e have a te rrib le mess 9 a ll the fa m ily is a n g ry to each o th e r 10 G: because o f th a t 11 w e ll b u t th a t it's a d iffe re n t th in g 12 D e f N u ll Subj has n o th in g to do w ith th a t 13 w e ll th e issue is th a t w e gather the m oney as w e cou ld... 14 the issue is th a t 15 M q : y o u p a id h im e v e ry th in g a t once Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 4.2.2.4. L in g u is tic In d ic a to rs o f D iscourse U n its : Q uestions o r com m ents b y th e in te rv ie w e r Even th o u g h the in te rv ie w e r (m yse lf) in m y data does n o t ask m a n y questions in o rd e r to le t the speakers ta lk , som etim es she feels the need to in te rru p t the speaker, to m ake sure th a t she understands fu lly w h a t is g o in g on. In these cases, h e r in te rv e n tio n s re s u lt in a change o f discourse u n it, o r in a p a re n th e tica l discourse u n it th a t is used to answ er th a t qu estion . E xam ples o f qu estion s/state m en ts b y the in te rv ie w e r w ith th is fu n c tio n are: Pero que pasd h u t w h a t h a p p e n e d ’ podes contarme un poquito 'can y o u te ll m e a little b it m ore' pero entonces contame 'b u t th e n te ll m e' pero cdmo termino 'b u t h o w d id th is e n d ’ cdmo empezo h o w d id i t s ta rt’ asi que cdmo fue que ’so h o w w as th a t' E ve ry tim e there is an in te rv e n tio n b y the in te rv ie w e r o f th is sort, there is also a change o f discourse u n it. O u t o f 27 shifts, th is happens 26% o f the tim e (cf. Table 2: 7 o u t o f 27). N otice th a t in the tw o cases w h e re there is no discourse m a rke r o r u n it opener (#5 and #23), there is a question or com m ent b y the in te rv ie w e r. Therefore, a q u e stio n o r a com m ent b y the in te rv ie w e r is also a s u ffic ie n t in d ic a to r o f a discourse u n it s h ift. 4.2.2.5. L in g u is tic In d ic a to rs o f D iscourse U n its : Q uotes Q uotes b y characters o f a p re v io u s ly to ld s to ry are used to sta rt a ne w discourse u n it th a t is re la te d to the p re vio u s discourse u n it. For instance in the fo llo w in g exam ple (11), speaker K is te llin g a s to ry ab out a sofa th a t she b o u g h t and th a t caused her a lo t o f tro uble . In lin e s 3-9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. \ O O O S C h U l ^ W M i - * O v O C C \ 1 0 \ O l ^ U N ) 83 she explains w h y she w a n te d to go b y h e rse lf to ta lk to the m anager o f the shop w here she b o u g h t it. T hen, in lin e 10, she refers to the p re v io u s discourse u n it a b o u t the sofa b y u sin g a q u o te b y one o f the p e o p le fro m th e shop (bueno nosotros se lo subimos 'w e ll w e w ill take it [the sofa] u p sta irs'). (ID 1 el silldn me lo tienen que subir ustedes (3 IUs omitted) fu i sola el no estaba te digo mird pa il tenia que ir a la escribania andate a la escribania quiero ir sola porque si estan ellos no me dejan hablar yo me tengo que pelear con el tipo X X X yo sola <Q bueno nosotros se lo subim os Q > y o u have to take the sofa u p (to m y apartm ent) (3 IU s o m itte d ) I w e n t by m yse lf he w as n o t here I'm te llin g y o u , lo o k fo r— he had to go to the re a l estate o ffice go to the real estate o ffice I w a n t to go b y m y s e lf because if th e y com e w ith m e, th e y d o n 't a llo w m e to speak I have to argue w ith the g u y X X X b y m yse lf 10 <Q w e ll w e w ill take it u p sta irs Q > 4.2.2.6. C o-occurrences o f factors. W e have seen th a t discourse u n its ty p ic a lly share the same p a rticip a n ts, the same tense an d aspect form s (cf. sectio n 2.4.3). T herefore, it is com m on th a t w h e n a change o f disco urse u n it occurs, a change in one o f these elem ents w ill also occur. L e t us see h o w o u r discourse u n its w ere used. Table 2 shows th a t o u t o f the 27 u n it sh ifts Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 considered in th is analysis, there is also a change in subject in 93%, and there is a change o f e ith e r tense o r aspect in 85%. A n d i f th e subject does n o t change (#18 and #20), there is alw ays a change o f tense and aspect o r vice versa, i.e. i f the ten se/a spect are the same, there is a change o f subject (#1,13,15,24). A change o f subject, tense o r aspect w o u ld be a s u ffic ie n t c o n d itio n to a u n it s h ift, i f in e ve ry sin g le occurrence o f such a change, there is also a u n it s h ift. H o w e ve r, th is co rre la tio n w as n o t attested in m y data, i.e. it is n o t the case th a t in e ve ry single case o f subject, aspect o r tense change in m y data, there was a discourse u n it s h ift. A t this p o in t, I o n ly presented the co rre la tio n s fo r the p o rtio n s o f texts th a t co n stitu te discourse u n its acco rd in g to the judges co n su lte d (o n ly those w ith lin g u is tic in d ic a to rs as e xp laine d above, in section 4.2.1.1). In exam ple (10), p . 81 there is a change o f subject (fro m 'yo n ' , lin e 6 to 'a m ess', lin e 8) and a change o f tense (fro m present das, lin e 6 to p re te rit armo, lin e 8) and aspect (h a b itu a l present das, lin e 6, vs. p u n c tu a l p re te rit, armo, lin e 8). In the second sw itch , there is a change o f tense and a change o f aspect (u n ive rsa l present tiene, lin e 13, p u n c tu a l p re te rit pudimos, lin e 13) and a change o f subject (D e f N u ll Subj 3 rd pe rson s in g u la r, lin e 13 to 'w e ', lin e 14). 4.3. A p p ro p ria te n e ss c o n d itio n s o f d e fin ite n e ss type s In the n e xt sections, I w ill p ro v id e the appropriateness co n d itio n s o f these definiteness types based on the analysis o f m y data. These sections o f the chapter are based o n the analysis o f a corpus o f data th a t consists o f a ro u n d 3 h o u rs o f in te rv ie w s and n a tu ra l conversations in S panish betw een y o u n g ad ults and m y s e lf (cf. chapter Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 3 fo r an extensive d e scrip tio n o f the data analyzed). I analyzed a ll o f the d e fin ite expressions, a to ta l o f 930 (376 D e fA rt+ N s, 62 D em +N s, 10 D e fA rt+ N + D e m s, 33 D em s, 64 Pers P ronouns, 186 C litic s and 199 D ef. N u ll Subjs). Table 3 sum m arizes the usages o f d iffe re n t definiteness types b y the acce ssibility o f P-sets. M o st o f the definiteness types re fe r to h ig h accessible P-sets (cf. D em + N , D em , Pers P ronoun, C litic and D e f N u ll Subj). H o w e ve r, D e fA rt+ N + D e m refers m o stly to m id accessible P-sets (70%) and D e fA rt+ N refers alm ost e xclu sive ly to lo w accessible P-sets. It is im p o rta n t to n o tice th a t D e fA rt+ N m ay refer to any k in d o f P-set. In w h a t fo llo w s I w ill analyze these d iffe re n t definiteness types in tu rn . Table 3: Percentage o f definiteness types and accessibility o f P-sets A ccof P -set DefArt +N DefArt+ N+Dem Derrn-N Ctem Pers Pronoun Clitic Def Null Subj n= % % % % o/ /o % % H IG H 46 30 81 94 87 96 100 687 M ID 20 70 17 6 13 3 0 117 LO W 34 0 2 0 0 1 0 126 n= 376 10 62 33 64 186 199 930 W hen a n a ly z in g these data I m ade the assum ption th a t the statistics o f m y sam ple o f data are re lia b le , typ ica l and n o t id io s y n c ra tic , based on the fa ct th a t m y data are q u ite extensive and n a tu ra lis tic (cf. chapter 3). Therefore, if a usage p o s s ib ility is n o t attested in m y data, it is considered to be im possible. I recognize tha t there m ay be ce rta in usages th a t d o n o t occur in m y data because they are ra th e r in fre q u e n t, and th a t th e y m ig h t occur, in p rin c ip le , in oth er sam ples. I f th is is so, it w o u ld g ive m e a sm a ll n u m b e r o f a d d itio n a l usage p o s s ib ilitie s an d it could add ce rta in item s to the data d is trib u tio n in table 3, so th a t w h e re Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 the re is c u rre n tly a zero there m ig h t be som e s m a ll n u m b e r o f occurrences fin d . I also m ade the assu m ptio n th a t th e s ta tis tic a l d is trib u tio n o f the preferences and dispreferences fo r these uses o f the d iffe re n t definiteness types is ty p ic a l fo r S panish in general. M e th o d o lo g ic a lly , th is is the o n ly assum ption th a t I can reason ably m ake: I am d e scrib in g accurately w h a t is in m y data and I am d e riv in g statem ents about usage re g u la ritie s fro m it. A g a in , h o w e ve r, the re m ig h t, in p rin c ip le , be a d iffe re n t d is trib u tio n in o th e r te x t sam ples, o r in o th e r dia le cts o f S panish. 4.3.1 T he d e fin ite a rtic le W hen speakers use a d e fin ite a rtic le in S panish (la idea 'the idea') to re fe r to an e n tity , they: 1) d e lim it a P-set w ith a n y degree o f a cce ssib ility 2) re q u ire uniqueness w ith in the lim its o f th e d e lim ite d P-set. As sho w n in table 2, D e fA rt+ N is the o n ly d e fin ite typ e th a t refers p ro d u c tiv e ly to lo w accessible P-sets (D e m + N and C litic m ay also re fe r to lo w accessibility P-sets, b u t o n ly v e ry in fre q u e n tly : 2% fo r D em + N and 1% fo r C litic ). The D e fA rt+ N w as a p p ro x im a te ly e ve n ly d is trib u te d am ong the d iffe re n t a cce ssib ility leve ls o f P-sets: 34% o f the usages re fe r to lo w e r acce ssibility P-sets, 46% o f the usages re fe r to h ig h acce ssibility P-sets, and 20% o f the uses re fe r to m id acce ssibility P-sets. I t is also im p o rta n t to no tice, th a t in these 3 h o u rs o f con versa tion, o u t o f 930 definiteness types, 376 (40%) are D e fA rt+ N . The D e fA rt+ N is the m o st fre q u e n t definiteness type . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 I also have som e data on the fre q u e n cy o f th e n o n -u n iq u e ly re fe rrin g uses o f th e D e fA rt+ N (the irre le v a n t uniqueness types o f B im e r and W a rd , 1994). I counted o n ly 2.7% o f usages o f D e fA rt+ N w h ic h are n o t re le v a n tly d iffe re n tia b le . F o r insta nce, Fueron a comprar algo al supermercado 'T h e y w e n t to b u y so m e th in g a t th e s u p e rm a rke t' (see cha pte r 2.2.3.) 4.3.2. P reposed d e m o n stra tive d e te rm in e r W hen speakers use a preposed d e m o n stra tive d e te rm in e r in S panish to re fe r to an e n tity , they: 1) d e lim it: p re fe ra b ly: a P-set o f h ig h acce ssibility less p re fe ra b ly: a P-set o f m id acce ssibility least p re fe ra b ly: a P-set o f lo w acce ssibility 2) re q u ire uniqueness w ith in the d e lim ite d P-set. A s sh o w n in tab le 3, 81% o f the references b y D em +N re fe r to the h ig h a cce ssib ility P-sets (p reviou s te x t set and p h y s ic a l s itu a tio n set). F o r instance in (12), ’esa p la ta ' refers back u n iq u e ly to the p re v io u s te x t P-set, c u rre n t discourse u n it (la plata del alquiler 'the m o n e y fro m th e re n ta l'). (12) K : yo tengo la plata del alquiler de mi papa me ent— que yo puedo disponer de esa plata (12’) K : I have th e m o n e y fro m m y fa th e r’s re n ta l y o u see— th a t I can use th a t m oney Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 W hen any d e m o n stra tive (e ith e r d e te rm in e r o r p ro n o u n ) is used fo r m a k in g a d e ictic contrast, the uniqueness re q u ire m e n t h o ld s w ith in the P-set d e lim ite d b y the d e ic tic sem antics o f the d e m o n stra tive . In o th e r w o rd s, g iv e n a p h y s ic a l s itu a tio n P-set, este w ill re fe r to th e u n iq u e e n tity close to the speaker, ese w ill re fe r to the u n iq u e e n tity close to the liste n e r and aquel w ill re fe r to the u n iq u e e n tity in the p h y s ic a l s itu a tio n fa r fro m b o th the speaker and the lis te n e r. 4.3.3. P ostposed d e m o n stra tive d e te rm in e r W hen speakers use a postposed d e m o n stra tive d e te rm in e r in S panish to re fe r to an e n tity , they: 1) d e lim it: p re fe ra b ly: a P-set o f m id acce ssibility less pre fe ra b ly: a P-set o f h ig h acce ssibility 2) re q u ire uniqueness w ith in the lim its o f the d e lim ite d P-set. The D e fA rt+ N + D e m does n o t re fe r to a lo w a cce sib ility P-set in m y data. Table 2 show s th a t 70% o f the usages o f D e fA rt+ N + D e m re fe r to m id a cce ssib ility P-sets (fo llo w in g te x t set and p re vio u s te xt, d iffe re n t discourse u n it set). For instance, in (13), D e fA rt+ N + D e m refers u n iq u e ly w ith in the P-set d e lim ite d b y the fo llo w in g p re p o s itio n a l phrase del gamexane. In o th e r w o rd s, there is o n ly one tim e th a t th e y had a b u g bom b set up d u rin g h ig h school. (13) F: ... la vez esa del gamexane esa vez era fin de ano Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 (13') F: ... the tim e th a t o f the b u g bom b th a t tim e was N e w Y ear D ay M y data do n o t s u p p o rt p re vio u s approaches (fo r instance, Sole & Sole, 1977 and Bergen, 1977) w h ic h state th a t the postposed d e m o n stra tive d e te rm in e r necessarily expresses d is d a in o r a p e jo ra tiv e m eaning (cf. chapter 2). F o r instance, in the fo llo w in g exa m ple (14) (taken fro m e l H abla C u lta de Buenos A ires: 498, cf. references) the p a rticip a n ts are re fe rrin g to a great dancer th a t the speaker is g o in g to see p e rfo rm and it is clear th a t a p e jo ra tive m eaning is n o t im p lie d . E: Ah, ila bailarina? jEstnpenda! Yo ahora la voy a ver X X X 3A h la mujer esa es fantdsticaXXX E: O h, the dancer? She is great! I am going to see h e r n o w X X X oh th a t w om an is fantasticX X X In the data analysed in th is d isse rta tio n , none o f the exam ples carries the p e yo ra tive m eaning. In the to ta l sam ple co lle cte d , o n ly 11% o f the occurrences o f the postposed dem onstrative co n ve y a p e jo ra tive m eaning (e.g. el tarado ese 'th e s tu p id g u y over there') 4.3.4. D e m o n stra tive p ro n o u n W hen speakers use a d e m o n stra tive p ro n o u n in S panish to re fer to an e n tity , they: 1) d e lim it: (14) preferably: m uch less p re fe ra b ly: a P-set o f h ig h a cce ssib ility a P-set o f m id a cce ssib ility 3 XXX = unintelligible text Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 2) re q u ire uniqueness w ith in the lim its o f the d e lim ite d P-set. I t is im p o rta n t to n o tice th a t D em cannot refer to a lo w acce ssibility P-set. A s sh o w n in tab le 3, 94% o f the usages o f D em re fe r to h ig h a cce ssib ility P-sets (p h y s ic a l s itu a tio n P-set and p re v io u s te x t set). F or instance in (14), M q is re fe rrin g u n iq u e ly to the d iv a n in the p re vio u s te xt, c u rre n t u n it o f discourse. (15) K : trdigame el divan bueno M q: me imagino qne seria mncho mas caro ese (15') K : b rin g m e the d iv a n w e ll M q: I im a g in e th a t th a t one w o u ld be m uch m o re exp ensive B oth D e m + N an d D em , re fe r to a u n iq u e e n tity in a h ig h acce ssibility P-set. The usage o f these tw o definiteness types d iffe rs in the fo llo w in g respects, h o w e ve r: firs t, D em refers m ore fre q u e n tly to the p h ysica l s itu a tio n P-set; and second D em refers to a m o re accessible e n tity w ith in the h ig h a cce ssib ility P-set. F irs t, D em refers m ore fre q u e n tly to the ph ysical s itu a tio n P-set: o u t o f its to ta l usages, 55% o f the tim e D em is used to re fe r to the ph ysical s itu a tio n . B y con trast, o n ly 44% o f the to ta l usages o f D e m + N are deictic. Second, there is a d iffe re n ce betw een the acce ssibility o f e n titie s w ith in the h ig h accessible P-sets. A closer e n tity w ill be m o re accessible than a fu rth e r e n tity . Table 4 show s th a t 71% o f the a n a p h o ric usages o f D em + N re fe r back to an antecedent located in the p re v io u s 2-5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 international u n its . In con trast, m ore th a n 62% o f the tim e , D em re fe rs back to the im m e d ia te ly p re ce d in g in to n a tio n a l u n it. T h is suggests th a t Dem refers to th e closer e n titie s and m ore accessible e n titie s w ith in the h ig h accessible P-sets. Table 4: Prep. Dem and Dem Pronoun: The role of the distance towards the antecedents. N P Anaphoric Previous IU Anaphoric previous 2-5 IUs T o ta l D e m + N 7 - 29% 17 - 71% 24 D e m 8 - 62% 5 - 38% 13 In m y data there are no exam ples o f the rem ote d e m o n stra tive in Spanish {aquel / aquella). Tables 5 (A ,B ) show the d iffe re n t usages fo r este (table 5,A ) and ese (ta ble 5, B) fo r h ig h accessible P-sets, fo r each d e m o n stra tive typ e (D e fA rt+ N + D e m , D em + N , and D em ). Este is used to re fer to the p h ysica l s itu a tio n in 81% o f the cases. O n the o th e r h a n d , ese is used to re fe r to the p re vio u s te x t (same discourse u n it) in 67% o f usages. T h is ske w in g in fre que ncy sup ports the c la im th a t th e p h ysica l s itu a tio n P-set is m ore accessible th a n the p re v io u s te xt because este, the p ro x im a l d e m o n stra tive , is used m a in ly to re fe r to the p h ysica l s itu a tio n P-set. Table 5: a note on ESTE/ESE (Raw numbers) A) ESTE T yp e o f D em A n a p h o ric Usages (same discourse u n it) D e ictic Usages (p h ysica l s itu a tio n set) D e m + N 3 16 D e fA rt+ N + D e m 2 0 D e m 2 14 T O T A L 7 30 (81%) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 B)ESE A n a p h o ric Usages (same discourse u n it) D eictic Usages (physical s itu a tio n set) D e m + N 20 11 D e fA rt+ N + D e m 0 0 D em 11 4 T O T A L 31 (67%) 15 4.3.5 Pers P ron oun W hen speakers use a Pers P ro n o u n in Spanish to re fe r to an e n tity , they: 1) d e lim it: preferably: a P-set o f h ig h a cce ssib ility m u ch less p re fe ra b ly: a P-set o f m id a cce ssib ility 2) re q u ire uniqueness w ith in the lim its o f the d e lim ite d P-set. U niqueness can p o ssib ly be achieved th ro u g h prom inence b y e lim in a tin g less p ro m in e n t e n titie s w ith in the same P-sets (cf. ch a p te r 5). 4.3.6 C litic W hen speakers use a c litic in S panish to re fer to an e n tity , they: 1) d e lim it: preferably: a P-set o f h ig h a cce ssib ility m u ch less p re fe ra b ly: a P-set o f m id o r lo w accessibility 2) re q u ire uniqueness w ith in the lim its o f the d e lim ite d P-set. U niqueness can p o ssib ly be achieved th ro u g h prom inence b y e lim in a tin g less p ro m in e n t e n titie s w ith in th e same P-sets (cf. cha pte r 5). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 4.3.7. D e f N u ll S u b j W hen speakers use a D e f N u ll Subj in S panish to re fe r to an e n tity , they: 1) d e lim it a P-set o f h ig h accessibility. T hey d o n o t d e lim it any o th e r P-set. 2) re q u ire uniqueness w ith in the lim its o f the d e lim ite d P-set. U niqueness can p o s s ib ly be achieved th ro u g h p ro m in e n ce (cf. chapter 5). 4.4. M o tiv a tin g th e appropriateness c o n d itio n s In section 4 .3 .1 described the appropriateness co n d itio n s o f definiteness types and n o w I w ill address the m o re e xp la n a to ry question: w h y d o these definiteness types have the appropriateness co n d itio n s th e y do? D e fA rt+ N , D em +N , D e fA rt+ N + D e m and D em have a s tric t uniqueness re q u ire m e n t w ith in th e P-set th a t is activa ted . The reference o f D e fA rt+ N can occur in a n y k in d o f P-set, and th is is the p re fe rre d d e finiten ess typ e fo r e n titie s in the lo w acce ssibility P-sets. D e fA rt+ N + D e m refers to a unique e n tity w ith in a m id acce ssibility P- sets. D em + N and D em b o th refer to a u n iq u e e n tity w ith in a h ig h accessibility P-sets. T he difference betw een D e m + N and D em has been related to the k in d o f P-sets they refer to (D em to p h y s ic a l situ a tio n sets and D em + N to p re v io u s text) and to the le v e l o f a cce ssib ility o f e n titie s w ith in the P-sets (D em refers to closer e n titie s a n d D e m + N to fu rth e r en tities). G ive n th a t b o th e n titie s located in the p h y s ic a l s itu a tio n and closer e n titie s are m ore accessible, I have co n clu d e d th a t D em has greater a cce ssib ility th a n D em +N . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 Pers P ronoun, C litic and D e f N u ll Subj re fe r to h ig h a cce ssib ility P-sets an d th e y often achieve uniqueness th ro u g h p ro m in e n ce (fo r a d e fin itio n o f prom inence cf. ch a p te r 5). The appropriateness c o n d itio n s o f the d iffe re n t d e fin ite n e ss types can be g ive n a fu n c tio n a l m o tiv a tio n a lo ng th e fo llo w in g lines. F irst, i t is to be expected th a t h ig h a cce ssib ility P-sets can be re fe rre d to w ith a w id e r v a rie ty o f d e finiten ess types, fo r the reasons I enum erate b e lo w in 4.4.1. Second, the p h o n o lo g ic a l size o f these fo rm s corresponds to the accessibility o f th e P-sets th e y re fe r to because i f the e n tity is v e ry accessible, v e ry little lin g u is tic m a te ria l is needed to re fe r to it (cf. 4.4.2 an d G ivon , 1992; A rie l, 1990). T h ird , d e fin ite n e ss types re fe rrin g to lo w e r acce ssibility P-sets w ill need to c o n ta in m ore in fo rm a tio n in o rd e r to m ake sure th a t the liste n e r w ill in te rp re t the N P c o rre c tly . Therefore, NPs w ith definiteness types th a t re fe r to lo w e r a cce ssib ility P-sets are expected to be lo n g e r and m ore co m p le x th a n those re fe rrin g to h ig h a cce ssib ility P-sets (cf. 4.4.3). F in a lly , a lth o u g h the distance and frequency fa cto rs are g o in g to be discussed exte nsively in the fo llo w in g chapter (5), I w ill presen t som e p re d ic tio n s here. I have sh o w n above th a t de fin ite n e ss types re fe r to P-sets o f d iffe re n t levels o f acce ssibility: D e f N u ll Subj, C litic , Pers P ro n o u n , D em and D em + N re fe r to h ig h a cce ssib ility P-sets. D e fA rt+ N + D e m refers to m id d le a cce ssib ility P-sets and D e fA rt m a y re fe r to any k in d o f P-set. T herefore it is expected th a t, w ith in a g ive n P-set (fo r instance, p re vio u s te xt, c u rre n t discourse u n it), D e f N u ll Subj, C litic , Pers P ronoun, D em , and D e m + N w ill re fe r to m o re accessible e n titie s (closer and m ore fre q u e n t e n titie s ). I t is also lik e ly th a t Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 D e m + N w o u ld re fe r to e n titie s m ore accessible th a n D e fA rt+ N + D e m . F in a lly , it is also expected th a t D e fA rt+ N + D e m w o u ld re fe r to m ore accessible e n titie s th a n D e fA rt+ N . A n d th is is ind eed the case as sh o w n b e lo w . M o reo ver, d e fin ite n e ss types w h ic h are n o t re stricte d to a ce rta in ty p e o f P-set (such as D e fA rt+ N ) are expected to be used m ore fre q u e n tly th a n definiteness typ e s w h ic h re fe r to a sp e cific P-set (fo r instance, D e fA rt+ N + D e m w h ic h refers m a in ly to m id a cce ssib ility P- sets). 4.4.1. L o w accessible vs. h ig h accessible P-sets I t is n o t accidental th a t languages have g ra m m a tica l d istin ctio n s to d is tin g u is h am ong v e ry accessible P-sets, w hereas th e y do n o t m ake co rre sp o n d in g d is tin c tio n s am ong d iffe re n t types o f less accessible P- sets. E n titie s become accessible w h e n w e ta lk a b o u t them . People u s u a lly in c lu d e m ore th a n one p a rtic ip a n t and e ve n t w h e n ta lk in g and th e re fo re it is necessary to d is tin g u is h am ong these p a rtic ip a n ts and events u sin g m arkers o f d iffe re n t degrees o f acce ssibility. L o w accessible en titie s are less fre q u e n tly re fe rre d to , and th a t is w h y languages re q u ire m ore lin g u is tic devices to re fe r to m o re accessible e n titie s th a n to less accessible e n titie s. In the data presented in this s tu d y , fo r instance, the re are fiv e d iffe re n t de finiten ess types th a t re fe r to h ig h acce ssibility P-sets (D e m + N , D em , Pers P ron oun , C litic , D ef N u ll Subj). H o w e ve r, o n ly D e fA rt+ N refers p ro d u c tiv e ly to lo w a cce ssib ility P-sets (associated e n titie s P-set, la rg e r s itu a tio n P-set and th e u n ive rse o f discourse P -set). M o re o ve r, w h e n re fe rrin g to lo w accessible P-sets (such as fra m es), a v e ry general definiteness type is used th a t is n o t reserved e x c lu s iv e ly fo r lo w a cce ssib ility P-sets: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 D e fA rt+ N can re fe r to a ll k in d o f P-sets, in c lu d in g those o f lo w accessibility. 4.4.2. P h o n olog ical size o f th e d iffe re n t de fin ite n e ss types: In chapter 2 ,1 described h o w , in general, sh o rte r form s re fe r to h ig h ly accessible e n titie s (cf. se ctio n 2.5.4.4., and C lancy, 1980; C hafe, 1976, 1994; G arrod, 1994; L e vin so n , 1987a; Sperber and W ilso n , 1986). T h a t exp lains w h y D e fA rt+ N + D e m an d D e fA rt+ N re fe r to e n titie s w ith in less accessible P-sets. B o th definiteness types are lo n g e r th a n the re st (D e fA rt+ N consists o f tw o w o rd s , and D e fA rt+ N + D e m consists o f 3 w o rd s ). This statem ent also e xp la in s w h y D em + N refers to less accessible entities th a n D em (D e m + N consists o f 2 w o rd s and D em consists o f ju s t 1). H o w e ve r, it does n o t e xp la in w h y D e fA rt+ N refers to e n titie s th a t belong to less accessible P-sets than those re fe rre d to b y D e m + N . D em +N is lo n g e r th a n D e fA rt+ N (este vs. el, 2 sylla b le s vs. 1 sylla b le ). Therefore, b y th is lo g ic , D e m + N sho uld re fe r to less accessible e n titie s, and th is is n o t the case. A rie l (1990) p ro vid e s an e xp la n a tio n fo r th is apparent counterexam ple w h ic h I presented in chapter 2 (section 2.5.4.4.). B asically, the re are tw o related reasons. F irst, there is a h is to ric a l reason, and second the re is the frequency factor. D e fA rt+ N d e rive s h is to ric a lly fro m the d e m o n stra tive and it has became m ore general, m ore fre q u e n t and less sp e cific. A s A rie l (1990) p o in ts o u t D e fA rt+ N is sho rter and is an u n m a rke d fo rm . T hat is w h y it is n o t s u rp ris in g th a t it refers to any k in d o f P-set, and th a t it is the o n ly d e fin ite typ e th a t refers p ro d u c tiv e ly to lo w accessible P-sets. O n the o th e r h a nd, D em +N is m ore sp e cific, re fe rrin g to m ore accessible Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 e n titie s th a t are present in a p re vio u s te xt, o r in the p h ysica l s itu a tio n . Its m eaning has becom e spe cialized. It is in te re stin g to no te th a t fo r reference in te rp re ta tio n , tw o processes take place fo r those sh o rte r definiteness types, w h ile o n ly one process takes place fo r the lo n g e r definiteness types. F or lo n g e r definiteness types (D e fA rt+ N , D em + N , D e fA rt+ N + D e m and D em ) the d e lim ita tio n o f P-sets is e n o u g h fo r u n iq u e reference. R e fe rrin g w ith sh o rte r definiteness types (Pers P ronoun, C litic and D e f N u ll Subj) h o w e ve r, m ay in v o lv e tw o d iffe re n t processes. F irs t, the P-set is d e lim ite d . T his is the sam e process th a t takes place fo r th e sho rter definiteness types. Second, e n titie s th a t are se m a n tica lly and m o rp h o lo g ic a lly co m p a tib le w ith the d e fin ite N P are com pared to each o th e r in term s o f p rom ine nce (cf. chapter 5). The apparent c o m p le x ity fo r reference assignm ent o f the Pers P ron oun , C litic and D e f N u ll Subj (tw o processes vs. one process fo r the D e fA rt+ N , D em + N and D e fA rt+ N + D e m ) can be exp laine d b y the fa ct th a t these NPs are sh o rte r in term s o f p h o n o lo g ica l size: the co m p le xity is balanced: easier to produ ce (sh o rt in p h o n o lo g ica l size), h a rd e r to in te rp re t (tw o processes vs. one). 4.4.3. A m o u n t o f m a te ria l A lo n g these lines, d e finiten ess types re fe rrin g to lo w e r a cce ssib ility P-sets o r e n titie s w ill need to co n ta in m ore sem antic m o re in fo rm a tio n in o rd e r to m ake sure th a t the liste n e r can in te rp re t the N P accurately, and th e re fo re , th e y are expected to be lo n g e r than those re fe rrin g to h ig h a cce ssib ility P-sets and e n titie s. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 T able 6 show s the m ean n u m b e r o f w o rd s in the N P a cco rd in g to d e fin ite n e ss types. F o r insta nce in (13) la vez esa del gamexane 'th e tim e th a t o f the b u g b o m b ', has been coded as 5 w o rd s, and in (15) el divan 'th e d iv a n ' has been coded as 2. A s expected, a t one extrem e, D e fA rt+ N + D e m has the greate st m ean n u m b e r o f w o rd s (4.3) w h ile Pers P ro n o u n (1) and D em (1) have th e sm allest, a t the o th e r extrem e. (D e f N u ll Subj and C litic have been o m itte d fo r ob vious reasons, the firs t has n o p h o n o lo g ica l exp ressio n, a n d the second does n o t a llo w fo r a n y m o d ifie r). Table 6: Am ount o f linguistic Material and type of NP D e fA rt+ N D e m + N D e fA rt+ N + D e m D em Pers P ro n o u n M e a n 2.786 2.339 4.3 1 1 I t is expected th a t D e fA rt+ N + D e m m a y a llo w fo r the occurrence o f m ore lin g u is tic m a te ria l to s p e c ify m e anin g. The e n tity re fe rre d to is n o t v e ry accessible and th e re fo re m o re in fo rm a tio n is needed. The m ean o f w o rd s fo r th e D e fA rt+ N is s lig h tly greater th a n the m ean o f w o rd s fo r D em + N (2.786 fo r D e fA rt+ N and 2.339 fo r D e m + N ). T h is is to be expected: D e m + N refers to m ore accessible P-sets and th e re fo re , it does n o t need m u ch e xtra lin g u is tic m a te ria l fo r th e rig h t in te rp re ta tio n o f the anaphor. I t is re le v a n t here to consider the sta n d a rd de via tio n s o f these m ean va lu e s (i.e. the e stim a tio n o f th e distance o f the va ria b le fro m its m ean va lu e ). The standard d e v ia tio n o f D e fA rt+ N (1.436) is greater th a n th e sta n d a rd d e v ia tio n o f D e m + N (.904). T h is is also as expected: the D e fA rt+ N has a w id e r range o f Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 usages and th e re fo re it occurs w ith a w id e r range o f p h o n o lo g ica l m a te ria l. 4.4.4. D istan ce and fre q u e n cy 4.4.4.I. D istance Table 7 presents the m ean and m edian fo r in to n a tio n a l u n its th a t in te rve n e betw een a definiteness type and its te x tu a l antecedent (a n a p h o ric usages, in c lu d in g those re fe rrin g to a d iffe re n t u n it o f discourse). A s expected, D e fA rt+ N refers to antecedents located several in to n a tio n a l u n its aw ay fro m the N P (6.61 in to n a tio n a l u n its ), and the D e f N u ll Subj refers to closer antecedents (located in th e p re vio u s tw o in to n a tio n a l u n its : 1.96). There is n o t m uch to say a b o u t D e fA rt+ N + D e m , since there are o n ly tw o occurrences o f anaphoric usages. In co m p a rin g D e fA rt+ N and D em +N in re la tio n to distance, it is u s e fu l to lo o k a t the m edian. The m ean (6.6. fo r D e fA rt+ N and 8.2. fo r D em + N ) is con tam in ate d b y the fact th a t o v e ra ll the re are m ore occurrences o f D e fA rt+ N , and therefore the sm a lle r nu m be rs m ake a sm a lle r average. I a vo id th is p ro b le m b y co n sid e rin g the m edian. The m e d ia n fo r D e fA rt+ N is 3 and the m edian fo r D e m + N is 2. This is expected because D em + N refers to m ore accessible P-sets and therefore to m ore accessible en titie s. O n the con trary, D e fA rt+ N s till has a w id e range o f p o s s ib ilitie s and it is used to re fer back to n e a rb y and fu rth e r antecedents (th e y co u ld re fe r to the p re vio u s in to n a tio n a l u n it o r to as m u ch as 50 in to n a tio n a l u n its p re vio u s to the N P ). H o w e ve r, the m e d ia n is 3, one in to n a tio n a l u n it fu rth e r back th a n D em + N . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 Table 7: Factor Distance: Mean and Median of number of IUs from the anaphor to the antecedent DefArt+ N DerrH-N DefArt+ N+Detn D tem Pers Pron. Clitic Def Null Subj M e a n 6.61 8.2 4 1.5 2.6 2.3 1.96 M e d ia n 3 2 4 1 2 1 1 4A.4.2. Frequency o f m e n tio n s o f the e n tity re fe rre d to Table 8 presents the m ean fre que ncy o f m e n tio n o f the e n tity referred to in the p re vio u s 25 in to n a tio n a l u n its acco rdin g to d iffe re n t definiteness type s. As expected, D e fA rt+ N + D e m refers to the least m entioned e n tity (2 tim es) fo llo w e d b y D em + N (2.67 tim es), D e fA rt+ N (2.9 tim es) and D em (3.8 tim es). A g a in , it is im p o rta n t to consider th e v e rs a tility o f D e fA rt+ N to re fe r to any k in d o f P-set and th a t e xp lains the greater m ean nu m be r o f m entions (2.9) com pared w ith D em + N . A lth o u g h it is im p o rta n t to n o tice th a t the diffe ren ce betw een the tw o is s till m in im a l. D em , Pers P ronoun, C litic and D e f N u ll Subj re fe r to v e ry fre q u e n tly m e n tio n e d e n titie s (th e y re fe r to en titie s th a t have been m entioned a t least 5 tim es: C litic : 5 tim es, Pers P ronoun: 6.8 tim es and D e f N u ll Subj 6.2 tim es). The fa c t th a t C litic refers to less fre q u e n tly m entioned e n titie s (5 tim es) tha n Pers P ronoun (6.8 tim es) and th a t D e f N u ll Subj refers to less fre q u e n tly m e ntion ed e n titie s th a n Pers P ron oun (6.2. tim es vs. 6.8 tim es) is e xp la in e d in chapter 5: frequency is n o t the o n ly fa cto r to p la y a ro le in the prom inence o f an e n tity and m oreover, it is n o t the m o st im p o rta n t fa c to r. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 Table 8: Factor Frequency: Mean of number o f mentions o f the entity referred to by the anaphor in the previous 25 IUs. De£Art+ N DerrH-N De£Art+ N+Dem Dem Pers Pron Clitic Def .Null Subj M e a n 2.9 2.67 2 3.8 6.8 5 6.2 4.4.4.3. F requ ency o f d e fin ite n e ss types Table 9 presents the frequency o f each de fin ite n e ss typ e . Tw o forces are p la y in g a ro le here. F irs t, m ore u n re s tric te d fo rm s (those th a t can re fe r to m ore d iffe re n t kin d s o f e n titie s) w ill be m o re fre q u e n t. Therefore, it is n o t s u rp ris in g th a t D e fA rt+ N is the m o st fre q u e n t definiteness typ e (40%: 376 usages o u t o f a to ta l o f 930 analyzed). Second, those d e finiten ess types th a t re fe r to h ig ly accessible e n titie s and th a t are v e ry easy to produce (because th e y are v e ry sh o rt) w ill also be v e ry fre q u e n t (cf. D e f N u ll Subj 21%, 199 o u t o f 930 and C litic 20%, 186 o u t o f 930). A t the o th e r extrem e, lo n g e r fo rm s and those m o st spe cified in m e anin g w ill be the least frequent: D e fA rt+ N + D e m is the longest fo rm ( it consists o f a t least 3 w ord s) and it is re s tric te d to m id a cce ssib ility P-sets, the refore it is v e ry in fre q u e n t (2% ). D em is n o t v e ry fre q u e n t (4% ). T w o c o n tra d ic to ry forces are a t w o rk here. O n the one ha nd, D em is s h o rt an d therefore u n sp e cifie d w h ic h s h o u ld re s u lt in fre q u e n t usage. O n the oth er ha nd, it does n o t re fe r to v e ry accessible e n titie s, w h ic h conspires against its fre que ncy o f usage. The re s u lt th a t seems to be w in n in g is the reference to less accessible e n titie s. F in a lly , D e m + N and Pers P ro n o u n are m o re o r less e q u a lly fre q u e n t (D e m + N = 7% and Pers P ro n o u n = 7% ). Pers P ro n o u n is less Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 fre q u e n t th a n D e f N u ll Subj and C litic , because i t is lo n g e r, even th o u g h i t refers to h ig h accessible e n titie s . D em + N does n o t re fe r to h ig h accessible e n titie s b u t, it is lo n g enough to p ro v id e s u ffic ie n t in fo rm a tio n to balance th a t la ck o f acce ssibility, are th e re fo re it is o f m e d ia n fre que ncy. Table 9: Frequency o f definiteness types DefArt+ N Dem+N DefArt+ N+Dem Dem Pers Pron Clitic Def.Null Subj Of fa 40 7 2 4 7 20 21 n = 376 62 10 33 64 186 199 4.5. C o n clu sio n T h is cha pte r d e a lt w ith the d e fin itio n o f d e fin ite n e ss (section 4.1), w h ic h in clu d e s a d e lim ita tio n o f a P-set and the re q u ire m e n t o f uniqueness. A n im p o rta n t p a rt o f the chapter was d e vo te d to p ro v id in g c rite ria to d e fin e lin g u is tic in d ica to rs fo r th e d e lim ita tio n o f discourse u n its based on the analysis o f data (section 4.2.2). S ection 4.3. p ro v id e d the appropriateness c o n d itio n s o f definiteness types based on levels o f acce ssibility fo r P-sets. F in a lly , section 4.4. p ro v id e d a c o g n itiv e m o tiv a tio n fo r the a p p ro pria tene ss c o n d itio n s o f d e fin ite n e ss types. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 C hapter 5 The in te ra c tio n o f factors O r T he n o tio n o f p ro m in e n ce 5.0. In tro d u c tio n In chapter 4 , 1 fo rm u la te d the appropriateness co n d itio n s fo r the definiteness types in S panish based on the re q u ire m e n t o f uniqueness and the accessibility o f P-sets. This chapter w ill be devoted to d e fin in g prom inence (i.e. the a cce ssib ility o f e n titie s w ith in P-sets) in o rd e r to accom m odate som e a p p a re n t counter-exam ples to uniqueness w ith in the general th e o ry presented here (cf. 4.1.2). The question to be addressed here is: w h a t are the factors re le va n t to a d e fin itio n o f prom inence and w h a t is th e ir re la tive strength? Four factors w ill be analyzed in m y data: distance, frequency o f m e n tio n , synta ctic p o s itio n and the m atic ro le . I w ill show th a t closer en titie s and m o re fre q u e n t e n titie s are m o re p ro m in e n t regardless o f th e ir syn ta ctic p o s itio n and th e m a tic ro le . I w ill also p ro v id e a co g n itive m o tiv a tio n fo r th is g e n e ra liza tio n . M y evidence comes fro m three d iffe re n t q u a n tita tiv e analyses: a p a ir-w is e com parison o f the d iffe re n t factors, a co n sid e ra tio n o f the fo u r factors fu n c tio n in g sim u lta n e o u s ly and an a p p lie d m u ltip le regression. To access a set o f en titie s (a P-set) is d iffe re n t fro m a ctiva tin g in d iv id u a l en titie s. W hen u s in g a definiteness typ e , the uniqueness re q u ire m e n t ho lds w ith in the boundaries o f the P-set d e lim ite d b y the speaker on lin e . In cha pte r 4 , 1 show ed h o w th is w o rk s fo r D e fA rt+ N , D em + N , D e fA rt + N + D e m , and Dem . H o w e ve r, fo r Pers P ronoun, D e f N u ll Subj, and C litic th e e n tity re fe rre d to m ay som etim es be te c h n ic a lly n o n -u n iq u e w ith in the lim its o f the d e lim ite d P-set (in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 oth e r w o rd s, there m ay be another e n tity co m p a tib le in gender and nu m be r w ith the re fe re n tia l expression, i.e . a c o m p e tito r or no n antecedent). In these cases, the speaker w ill re fe r to the m ost p ro m in e n t e n tity w ith in the lim its o f th e d e lim ite d P-set, i.e. the e n tity w h ic h stands o u t u n iq u e ly in term s o f its prom ine nce . 5.1. The definition of prom inence The n o tio n o f prom inence is c ru c ia l fo r re fe re n t id e n tific a tio n w ith the fo llo w in g definiteness types: 1) P ersonal P ronouns El come. 'H e eats' 2) D ef. N u ll Subject p ro Come, 'p ro eats' 3) C litic s Lo p ro come, 'p ro eats it' A ll D ef. N u ll Subjects and a lm o st a ll C litic s (97%) are anaphoric. F u lly 76% o f the Personal P ronouns are a n a p h o ric. A m on g these anaphoric definiteness types, o n ly 41% o f Per. P ronoun, 27% o f D ef. N u ll Subject and 16% o f C litic re fe r u n iq u e ly w ith in the d e lim ite d p ra g m a tic a lly stru cture d set o f e n titie s. The rest o f these NPs are te ch n ica lly am biguous in reference an d m a y re fe r to another e n tity w ith in the P-set. In these cases, there is a c o m p e tito r th a t is com patib le w ith the antecedent o f the anaphor in g e nde r, nu m be r, and sem antic p ro p e rtie s. B asically, the anaphor c o u ld have been in te rp re te d w ith the co m p e tito r's reference w ith o u t re s u ltin g in an 'unacceptable' sentence a lth o u g h , o f course, the re s u ltin g sentence w o u ld n o t be coherent w ith the rest o f the text. For instance, in lin e 3 o f exam ple (1) below , b o th p re vio u s sentences p ro vid e p e rfe c tly acceptable antecedents fo r the D e f N u ll Subj, a lth o u g h o n ly la mujer 'th e w o m a n ' is coherent w ith the p re vio u s te x t (fo r a d e fin itio n o f coherence, cf. G em sbacher et al., 1995). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 I t is a fact th a t speakers a n d listen ers m anage to re fe r u n a m b ig u o u s ly d espite non-uniqueness. M y h yp o th e sis is th a t w ith in a d e lim ite d P- set, en titie s are ra n ke d a cco rd in g to degree o f p ro m in e n ce and m y goal is to co n trib u te to a co h e re n t n o tio n o f prom inence. (1) 1 el pibe estndia disefio grafico 2 y la muier era m uy conocida 3 (Def. N idi Subject) tenia mucha experiencia d ’) 1 'the b o y studies g ra p h ic d e sig n 2 and the w om a n w as v e ry w e ll k n o w n 3 (h e /s h e ) had a lo t o f experience' 5. 2. Factors co n sid e re d to be re le v a n t fo r a th e o ry o f p ro m in e n ce The factors th a t are considered to be re le v a n t fo r a th e o ry o f prom inence (distance, fre q u e n cy o f m e n tio n , syn ta ctic p o s itio n , and th e m a tic ro le [cf. C ha pter 2 ]) have a ll been e xte n sive ly discussed in the lite ra tu re . I t is n o t clear, h o w e ve r, h o w they a ll in te ra c t. In o th e r w o rd s, w h a t is th e ir re la tiv e stre n g th , and are any o f these factors m ore re le va n t o r stronger? In o rd e r to answ er these questions I ra n d o m ly selected 109 occurrences1 o f Per. P ro n o u n (22), D ef. N u ll Subject (43), and C litic (44) th a t re fe r to an antecedent, b u t w h ic h w ere also co m p a tib le w ith a second possible antecedent in the te xt. In o th e r w o rd s , th e antecedent selected fo r the in te rp re ta tio n o f the d e fin ite N P had a c o m p e tito r, and since the d e fin ite re fe re n t w as no lo n g e r u n iq u e , p ro m in e n c e has to stand in fo r u n iq u e n e ss. 1 All the personal pronouns that refer to a non unique entity within a P-set were considered. In m y coding, the first 43 and 44 occurrences of DFNLSJ and CL which do not refer to a unique entity within a P-set were considered. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 The fo llo w in g factors w e re take n in to acco unt in these data: distance fro m the antecedent an d fro m the c o m p e tito r to the anaphor; fre que ncy o f m e ntion , i.e. h o w m a ny tim es the e n tity re fe rre d to b y the antecedent and the c o m p e tito r w as m e n tio n e d in the p re vio u s 25 in to n a tio n a l u n its (IU s, fo r a d e fin itio n o f IU , cf. ch a p te r 3); the syn ta ctic p o s itio n o f the la s t m e n tio n o f the antecedent and the co m p e tito r; and the s e m a n tic /th e m a tic ro le o f th e la s t m e n tio n o f the antecedent and the c o m p e tito r. 5.2.1. D istance D istance was coded in th e fo llo w in g w a y: I cou nte d the nu m b e r o f in to n a tio n a l u n its , fro m th e a n a p h o ric N P to th e antecedent and the nu m b e r o f IU s fro m the a n a p h o ric N P to the c o m p e tito r. Then, I calculated the difference b e tw e e n the tw o . In exam ple (1) above, fo r instance, the d iffe re n ce betw een the n u m b e r o f IU s fro m the D e f. N u ll Subject in lin e 3 to the co m p e tito r el pibe in lin e 1 is 2; and the n u m b e r o f IU s fro m the D ef. N u ll Subject in lin e 3 to the selected antecedent (in lin e 2) is 1. The diffe ren ce, the refore is 1. N ote th a t th is n u m b e r 1 says n o th in g a b o u t the absolute distance betw een the a n a p h o r and the selected antecedent, since th is n u m b e r m ay be the re s u lt o f 2 m in u s 1 w h e re b o th antecedent and c o m p e tito r are re a lly close (in the p re vio u s lin e ) o r it c o u ld be the re s u lt o f 10 m inu s 9, w here b o th are n o t re a lly close. I t is im p o rta n t to n o te th a t m ost o f the antecedents o f o u r a n a p h o ric NPs are located v e ry close to the anaphor. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 5.2.2. Frequency Frequency o f m entions w as coded in the same m anner b y c a lc u la tin g the difference in m e n tio n s betw een the antecedent and the c o m p e tito r. Thus, if the antecedent is m e n tio n e d 10 tim es and the co m p e tito r 6, the difference is 4. S u m m in g u p , distance is equal to the n u m b e r o f IU s fro m the c o m p e tito r to the anaphor m in u s th e n u m b e r o f IU s fro m the antecedent to the anaphor. F requency o f m e ntion s is equal to the n u m b e r o f m entions fo r the antecedent m in u s the num ber o f m e ntion s fo r the com petito r. T h is fo llo w s the assum ption th a t closer antecedents (sm aller nu m b e r o f IU s fro m antecedent to anaphor) and m ore fre q u e n t antecedents (grea ter n u m b e r o f m entions) are m ore p ro m in e n t. 5.2.3. S yntactic p o s itio n F o r synta ctic p o sitio n , subjects (b o th em bedded and m a in subjects) w ere considered m ore p ro m in e n t th a n objects (b o th em bedded and m ain objects), and objects, m ore p ro m in e n t th a n any other syn ta ctic p o sitio n (cf. 2.5.4.5). I assigned an a rb itra ry nu m be r to the syn ta ctic p o sitio n o f the la st m e n tio n o f the antecedent and to the synta ctic p o s itio n o f the last m e n tio n o f the co m p e tito r. Then, I calculated the difference betw een the tw o . The h ie ra rch y is the fo llo w in g : subjects > objects > o th e r 3 > 2 > 1 T herefore, if the c o m p e tito r is in an object p o sitio n (2), and the antecedent is in a subject p o s itio n (3), the d iffe re n ce is 1. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 5.2.4. T h e m a tic ro le A lo n g the sam e lines, agents and experiencers w ere considered m ore p ro m in e n t th a n them es, and them es m o re p ro m in e n t th a n a ll the rest o f the sem antic roles (cf. 2.5.4.6). A g a in , an a rb itra ry num ber w as assigned to th e the m atic ro le o f the la s t m e n tio n o f the antecedent and to the the m atic ro le o f the la st m e n tio n o f th e co m p e tito r. Then, I calculated the d iffe re n ce betw een the tw o . agents-experiencers > th e m e > o th e r 3 > 2 > 1 For instance, if the co m p e tito r is an agent (3) and the antecedent is them e (2), the diffe ren ce is 1. 5.2.5. E xam ple o f co d in g fo r the fo u r fa cto rs In exam ple (1) above, fo r instance, re g a rd in g distance, the antecedent (la mujer, 'the w o m a n ') o f the D e f. N u ll Subject is closer th a n the c o m p e tito r (el pibe ’the b o y'): la mujer is in the p re vio u s IU to the anaphor an d el pibe is tw o IU s p rio r to the anaphor. W ith respect to frequency, the antecedent (la mujer 'th e w o m a n ') is m ore fre q u e n tly m e n tio n e d th a n the c o m p e tito r (el pibe 'the b o y): la m ujer has been m e n tio n e d fo u r tim es p rio r to th e an aph or: (una mujer 'a w o m a n ', D ef. N u ll Subject trabaja, 'h e /s h e w o rk s ', D ef. N u ll Subject tenia 'she had', la mujer 'the w om a n'). W ith respect to syn ta ctic p o s itio n , b o th the la st m e n tio n o f the antecedent (la mujer 'the w om a n') and the la s t m e n tio n o f the c o m p e tito r (el pibe 'th e b o y') are in the subject p o s itio n . W ith respect to th e m a tic ro le , the antecedent (la mujer) is an experiencer (cf. chapter 3 fo r c o d in g con ventio ns) and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 the c o m p e tito r (el pibe) is an agent. B o th experiencer and agent are considered to have the same le ve l o f p ro m in e n ce (cf. cha pte r 3). 5.3. In te ra c tio n o f factors Table 10 presents the fo u r fa cto rs th a t have been considered to e x p la in w h y an antecedent p re v a ils (w in s ) o ve r its co m p e tito rs: distance, fre q u e n cy o f m e ntion , s y n ta c tic p o s itio n and th e m a tic ro le in the 109 occurrences analyzed. The ta b le show s the n u m b e r o f usages in w h ic h th e successful antecedent has a h ig h e r va lu e fo r the fa cto r in q u e stio n , com pared w ith the n u m b e r in w h ic h the unsuccessful c o m p e tito r has a h ig h e r valu e. Table 10: Antecedents and their competitors in relation to four factors Distance Freq. o f Mntns Syntactic pos. Them. Role ANTCDENT W INS 87% (n=95) 67% (n=73) 36% (n=39) 32% (n=35) CMFTTOR W IN S 7% (n=8) 24% (n = 26) 16% (n = l7) 16% (n=17) TIE 6 % (n=6) 9% (n=10) 49% (n=53) 52% (n=57) In the second ro w (antecedent w in s ) o f Table 10, w e can see th a t in 87% o f the cases, the antecedent is closer th a n the co m p e tito r, in 67% the antecedent is m ore fre q u e n tly m e n tio n e d tha n the c o m p e tito r, and in 36% an d 32%, the last m e n tio n o f th e antecedent has a m ore accessible syn ta ctic p o s itio n o r th e m a tic ro le th a n the la st m e n tio n o f the co m p e tito r. The th ird ro w , lis ts those occurrences in w h ic h th e c o m p e tito r is closer (7%), m ore fre q u e n t th a n the antecedent (24%) and has a m ore accessible syntactic p o s itio n (16%) and th e m a tic ro le (16%) th a n th e antecedent. In these cases, th e antecedent selection m ay be accounted fo r b y o th e r factors th a n th e one b e in g q u a n tifie d . The la s t Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 ro w (Tie) presents those cases in w h ic h b o th the antecedent and c o m p e tito r are located in th e sam e international u n it (6% ), o r th e y are m e n tio n e d the same n u m b e r o f tim e s (9% ), o r th e y have th e sam e syn ta ctic p o s itio n (49% ), o r th e sam e the m atic ro le (52%). Table 10 show s th a t th e antecedent is g e n e ra lly closer to the anaphor than the c o m p e tito r, an d it is ge n e ra lly m ore fre q u e n t. O n the o th e r h a n d , the syn ta ctic p o s itio n and the m atic ro le o f the antecedent v a ry i.e. it is n o t the case th a t th e syn ta ctic p o s itio n and th e m a tic ro le o f the antecedent is in g e n e ra l m o re p ro m in e n t tha n the s y n ta c tic p o s itio n and the m atic ro le o f the co m p e tito r. W e can co n clude , th e re fo re , th a t distance is th e m o st im p o rta n t fa cto r fo r th e sele ction o f an antecedent, fo llo w e d b y fre q u e n cy o f m ention, syn ta ctic p o s itio n , and sem antic ro le in th a t o rd e r. The num bers in T able 10 are lim ite d in the fo llo w in g respect. T h ey do n o t show h o w to d e te rm in e antecedent selection in each exam ple o f reference assig nm e nt to an anaphor, n o r do th e y show w h a t the re la tive stre n g th o f each fa c to r is fo r each occasion o f use. Table 10 gives us a g lo b a l o v e rv ie w in w h ic h the factors are considered sim u lta n e o u sly. T his m eans th a t, w e cannot answ er q u estion s such as the fo llo w in g on the basis o f these data. Is it the case th a t closer antecedents are also m o re fre q u e n t? Is it the case th a t w h e n the antecedent is a subject (the m o re p ro m in e n t synta ctic p o s itio n ) it is also an agent? W hat happens w ith distance w hen the c o m p e tito r is m ore fre q u e n tly m entioned th a n th e antecedent o f the anaphor? A m ore d e ta ile d q u a n tific a tio n fo r each piece o f data is re q u ire d in w h ic h the d iffe re n t factors are com pared p a irw is e . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I l l 5.3.1. First quantitative pairwise analysis In o rd e r to answ er the questions posed in the p reviou s section, I firs t com pared the antecedent and the c o m p e tito r fo r these 109 exam ples in re la tio n to each o f the fo u r factors considered p a irw ise . R egarding distance a n d frequency o f m e n tio n , I coded those occurrences in w h ic h th e antecedent w as m ore p ro m in e n t in distance (closer) than the c o m p e tito r (i.e. the antecedent 'w in s ' w ith respect to distance) and those in w h ic h the antecedent w as less p ro m in e n t in distance (it was fu rth e r fro m the an aph or, i.e. the antecedent loses re g a rd in g distance). In th is analysis, I disre g a rd e d those cases in w h ic h the antecedent w as pla ce d a t the same distance (in the same IU ) as the co m p e tito r because th is fa cto r does n o t th e n fa v o r any o f the possible antecedents fo r reference in te rp re ta tio n . A s a re su lt, the to ta l o f coded cases was a sm a lle r n u m b e r than 109. I, then, to o k th is subset o f occurrences (a ll those cases w here the c o m p e tito r and antecedent d iffe r w ith respect to the distance factor, i.e. th e y are n o t located in the same IU ) and coded them in re la tio n to a n o th e r fa cto r, fo r instance frequency o f m ention. A g a in , I coded those cases in w h ic h the antecedent 'w in s ’ in re la tio n to fre q u e n cy o f m e n tio n o v e r th e co m p e tito r, i.e. the antecedent is m ore fre q u e n t th a n the c o m p e tito r, and those cases in w h ic h the antecedent 'loses', i.e. is less fre q u e n t th a n the co m p e tito r. A g a in , I disregarded those cases in w h ic h b o th the antecedent and the co m p e tito r w ere m e n tio n e d the same n u m b e r o f tim es. I p a id special a tte n tio n to the cases o f c o n flic t, i.e. those cases in w h ic h one o f the factors is m ore p ro m in e n t fo r the antecedent, i.e. 'w in s ' (and therefore, less p ro m in e n t fo r th e co m p e tito r) and the o th e r fa cto r was less Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. U l ^ W N J ' M 0 \ ( J l ^ « N ) 112 p ro m in e n t fo r the antecedent, i.e. 'loses' (and the c o m p e tito r w as th e re fo re m ore p ro m in e n t fo r the c o m p e tito r). Then, I repeated th is k in d o f analysis fo r each fa c to r in re la tio n to the others (distance- fre q u e n cy o f m e n tio n , d ista n ce -syn ta ctic p o s itio n , d ista n ce -th e m a tic ro le , fre q u e n cy o f m e n tio n -s y n ta c tic p o s itio n , frequency o f m e n tio n - th e m a tic ro le , synta ctic p o s itio n -th e m a tic ro le ). L e t us exam ine som e exam ples. In exam ple (2) b e lo w the D ef. N u ll S ubject in lin e 7 m a y re fe r ba ck to ellaj (in lin e 1) o r to Gracielai (in lin e 4)- The in te n d e d antecedent Graciela is located 3 IU s p rio r to th e an aphor w h ile the c o m p e tito r (ella) is located 7 IU s p rio r to th e anaphor. The difference is th e re fo re 4. R egarding, fre que ncy h o w e ve r, th e antecedent (Graciela) has been m e n tio n e d o n ly 3 tim e s, w h ile the c o m p e tito r (ella) has been m e n tio n e d 10 tim es in the p re v io u s 25 IU s. The d iffe re n ce is 7. T h is is an exam ple o f c o n flic t, in w h ic h the antecedent (Gracielai, , in lin e 4) is close r b u t less fre q u e n t th a n the c o m p e tito r (ella), in lin e 1). (2) (FL8 2-220) 1 V : yo no se la version de e lla jque le habra contado (2 IU o m itte d ) y yo le conte a G ra c ie la i porque G ra ciela i qne es la hermana del jefe bueno pero no no D ef. N u ll S u b je c ts fa como la hermana digamos no D e f. N u ll S u b je ct!jjes re-macanuda que se yo (2') V : I d o n 't kn o w w h a t shej m ig h t h a ve said to h im (2 IU s o m itte d ) an d I to ld G racielai because G racielai w h o is the boss' sister w e ll b u t Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 0 0 N ] O N U l t ^ . W r v J M O O M O \ U l ^ W N ) H 113 6 shei does n o t act as i f she w as I m ean 7 s h e i/j is v e ry nice w h a t do I k n o w E xam ple (3) b e lo w presents the o p p o s ite s itu a tio n , in w h ic h distance fa vo rs the c o m p e tito r and fre q u e n cy fa vo rs the antecedent. In (3) the Per. P ro n o u n e /i (in lin e 7) m a y re fe r back to un chico del contingentej (in lin e 5) o r to Juan Pabloi (lin e 1)- The in te n d e d antecedent Juan Pablo is located 15 IU s p r io r to the an aph or w h ile the c o m p e tito r (un chico del contingente) is lo ca te d 2 IU s p rio r to the anaphor. The d iffe ren ce is the refore 13. R e g a rd in g , frequency h o w e ve r, th e antecedent (Juan Pablo) has been m e n tio n e d 5 tim es, w h ile the c o m p e tito r (un chico del contingente) has been m e n tio n e d o n ly once in the p re vio u s 25 IU s. The d iffe re n ce is 4. T h is is another exam ple o f c o n flic t, in w h ic h the antecedent is m ore fre q u e n t th a n the co m p e tito r b u t the c o m p e tito r is closer th a n the antecedent. (3) V : y de Juan Pabloi no le habia dicho nada a la amiga (10 IU s o m itte d , dig ression ) F: y bueno despues nos fuim os de viaje de egresados en setiembre eh ahi @@@@ a m i me gustaba un chico del contingentej que iba con nosotros de otro colegio y entonces un dia e li/j se acerca y dice <Q che a vos quien te gusta Q> (3 ') V : and she h a sn 't to ld h e r g irl frie n d a n y th in g ab out Juan Pablo (10 IU s o m itte d , dig ression ) F: and w e ll th e n w e w e n t to a g ra d u a tin g trip in Septem ber u h th e re @ @ @ @ I lik e d a g u y fro m the g ro u p th a t w as co m in g w ith us fro m a n o th e r school and th e n one da y he approaches m e and he says <Q hey, do y o u lik e som eone fro m here? Q> Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 S om etim es, how ever, b o th factors fa v o r the same antecedent. T h is is in d e e d the m ore fre que nt scenario. L e t us exam ine exam ple (4) b e lo w . The antecedent o f the D ef. N u ll S ubject in lin e 4 {la mujeri 'th e w o m a n ', lin e 3) is closer than the c o m p e tito r {el 'hej', lin e 1) b y 7 IU s. The antecedent is also m ore fre q u e n t th a n the co m p e tito r b y 1 m e n tio n . N o te th a t the antecedent is m o re p ro m in e n t th a n the c o m p e tito r fo r b o th factors, the refore , the c o m p e tito r is n o t th re a te n in g the se le ctio n o f the antecedent v e ry m uch. (4) 1 elj quedo co— 2 (7 IU s o m ite d ) 3 la mujeriera m uy conflictiva viste 4 D e f. N u ll S u b je c ti/j estaba siempre pensando eh que ... (4 ') 1 he re m a in e d as— 2 (7 IU s o m ite d ) 3 the w o m a n w as v e ry contentious y o u k n o w 4 sh e /h e w as alw ays th in k in g u h th a t... I t is also possible tha t b o th factors fa v o r the co m p e tito r, as in (5) b e lo w . In th is exam ple, the C litic (in lin e 12) m a y re fe r back to el ejercito 'the arm y' (in lin e 6) o r to the D e f. N u ll Subject, w h ic h refers to Juan Pablo in lin e 1. The rig h t antecedent (the D ef. N u ll Subject in lin e 1) is located 22 IU s p rio r to the an aphor w h ile the co m p e tito r {el ejercito) is located 6 IU s p rio r to the anaphor. The diffe ren ce is th e re fo re 15. R egarding frequency, the antecedent (Juan Pablo) has been m e n tio n e d once, w h ile the c o m p e tito r {the army) has been m e n tio n e d three tim es in the p re vio u s 25 IU s. D espite the fa ct th a t b o th factors con spire against the antecedent, the liste n e r d id n o t have Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 any tro u b le in te rp re tin g the referent. In th is exam ple, the p ra g m a tic co n te xt v e ry h e a v ily fa vo rs the selection o f Juan Pablo as the re fe re n t o f the definiteness typ e because the a cce pta bility o f th e 'in the second year I kn e w w h o /w h a t th e a rm y w as’ is ve ry d o u b tfu l. (5) 1 F: Def. N u ll S ubject detia que era muy 2 [m uy denigrante 3 M q: [me imagino 4 F: (D ef. N u ll S ubject: e llo s, e l ejercito) denigraban a la familia 5 (12 IU s o m itte d ) 6 si te (D ef. N u ll Subject: el eje rcito) queria mandar a la Quiaca 7 bueno te ibas a la Quiaca 8 (1 IU o m itte d ) 9 M q : asi que como fu e que 10 F: entonces no si 11 M q: segundo ano? 12 F: segundo ano mas o menos lo, ubicaba (5’) 1 F: (Def. N u ll Subject: he) to ld me th a t it was v e ry 2 ve ry d e n ig ra tin g 3 M q: I can im ag ine 4 F: (Def. N u ll Subject: th e y , the arm y) the y w o u ld d e n ig ra te the 5 fa m ily 6 (12 IU s o m itte d ) 7 if (D ef. N u ll S ubject= the arm y) the a rm y w a n te d to send yo u to 8 la Q uiaca 9 w e ll yo u w o u ld go to La Quiaca 10 (2 IU s o m itte d ) 11 M q: so, h o w w as th a t yo u ... (m et Juan Pablo) 12 F: then, no, yes 13 M q: second year in h ig h school 14 F: in the second yea r I kn e w w ho he was Table 11 show s the in te ra ctio n o f the fre q u e n cy o f m e n tio n and distance factors. I t com pares the num ber o f occurrences o f a co m p e tito r th a t are closer to the an a p h o r (w ith an antecedent e ith e r m ore o r less fre q u e n t th a n the c o m p e tito r) w ith the nu m be r o f occurrences o f an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 antecedent closer to the anaphor) a n d in w h ic h the antecedent is e ith e r m o re o r less fre q u e n t th a n the c o m p e tito r). Table 11: The interaction o f Distance and Frequency o f M entions D istance F requ ency o f M n . C om p. C loser A n t. C lo s e r T o ta l A n t. m ore fre q n t R g i'Ig g a H & S ig 64 70 C om p , m ore frq n t jssm sm ssm a m m 24 T o ta l 8 86 94 There are, the refore , fo u r c o m b in a to ria l p o s s ib ilitie s . In th e firs t and m o st fre q u e n t case, the antecedent is selected b y b o th factors fa v o rin g prom inence (cf. 64 in table 11): in these cases it is b o th closer to th e anaphor and m ore fre q u e n t th a n its co m p e tito r. In the second (a nd the least fre q u e n t case) the c o m p e tito r is b o th closer and m o re fre q u e n t (ju s t 2 exam ples). M ore in te re s tin g , h o w e ve r, are the cases in w h ic h the antecedent and the c o m p e tito r have d iffe re n t values fo r these tw o factors (the shaded cells in ta b le 11). In o th e r w o rd s , the antecedent is closer to the anaphor b u t less fre q u e n t th a n the c o m p e tito r (22 in table 11), o r else th e antecedent is fu rth e r b u t m o re fre q u e n t th a n the co m p e tito r (6 in ta b le 11). These tw o la st p o s s ib ilitie s are s ig n ific a n t because if I com pare th e n u m b e r o f occurrences o f these so ca lle d c o n flic t cases I can see w h ic h fa c to r has m ore s tre n g th w h e n it is opposed b y the o th e r factor. F or insta nce, in 22 occurrences, even th o u g h the c o m p e tito r w as m ore fre q u e n t th a n the antecedent, the c lo s e r antecedent was selected; and in o n ly 6 occurrences d id the o p p o site occur. E ven th o u g h the antecedent w as fu rth e r fro m the a n a p h o r, the m ore fre q u e n t antecedent w as p icke d fo r reference Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 assignm ent. O f course one w o u ld expect to fin d m a n y m o re cases in w h ic h b o th factors are p o s itiv e , i.e. in w h ic h b o th co n sp ire to m ake the antecedent w in and th is is ind eed w h a t happens (cf. 64 in tab le 11). A n d , con verse ly, one w o u ld expect there to be v e ry fe w cases in w h ic h the antecedent w in s despite the fa ct th a t b o th fa cto rs fa v o r the c o m p e tito r (cf. 2 in table 11). In 22 (79%) o u t o f these 28 c o n flic t cases (th e shaded cells), the antecedent is closer to the an aph or even th o u g h the c o m p e tito r is m o re fre q u e n tly m e ntion ed. In 6 cases (21%) th e antecedent is m ore fre q u e n t even th o u g h the c o m p e tito r is located closer to the anaphor. These cases o f c o n flic t are p ro b le m a tic fo r liste n e rs since i t is h a rd e r fo r the m to decide w h ic h N P sh o u ld be selected fo r the in te rp re ta tio n o f the an a p h o r. A n d it appears in these cases th a t distance w in s o ve r fre q u e n cy o f m e n tio n m ost o f the tim e , i.e. the lis te n e r w ill p ic k the closer antecedent ra th e r th a n the m ore fre q u e n t one. H ence, the dista nce fa c to r is stro nge r th a n the frequency o f m e n tio n fa cto r. In th is p a irw is e com parison, I disre ga rde d those cases in w h ic h the antecedent and the c o m p e tito r have e q ual n u m b e rs o f m e n tio n s and are p o s itio n e d a t the same distance fro m th e an a p h o r, because I am in te re ste d in cases o f c o n flic t in w h ic h the in te rp la y o f d iffe re n t factors is a p p a re n t. M o reo ver, I expect th a t as the d iffe re n ce in distance increases, the n u m b e r o f cases in w h ic h the antecedent is closer to the a n a p h o r th a n th e c o m p e tito r w ill be greater. I f the distance diffe re n ce decreases, the n u m b e r o f cases in w h ich the antecedent is closer to the a n a p h o r th a n the c o m p e tito r w ill be sm aller. T h is is, in d e e d , w h a t I have fo u n d . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 5.3.2. T he degree o f success o f each fa c to r: re la tiv e stre ngth I have sh o w n th a t distance fro m th e a n a p h o r is m ore im p o rta n t than fre que ncy o f m e n tio n fo r the selection o f the rig h t antecedent w hen there is a co m p e tito r. The next q u e stio n to ask is: w h y is it the case th a t distance does n o t w in 100% o f the tim e , if, as I have show n above, distance is a v e ry im p o rta n t factor? T he in te ra c tio n o f these factors is v e ry com plex and each factor, can be m o re o r less strong. For instance, consider distance, ft is n o t the case th a t distance is alw ays the m ost im p o rta n t fa cto r, irre sp e ctive o f the o th e r factors. I f an antecedent is v e ry fre q u e n tly m entioned, it can be selected fo r the in te rp re ta tio n o f the reference, even i f it is fa r aw ay. In oth er w ords, being located v e ry close to the anaphor is n e ith e r a necessary n o r a s u ffic ie n t c o n d itio n fo r an antecedent to be selected. There are other factors w h ic h p la y a ro le as w e ll (consider, fo r instance, the ro le o f the pragm atic co n te xt in exam ple (5) above). T h e re fo re , the strength o f each fa cto r is also im p o rta n t. In o rd e r to q u a n tify the degree to w h ic h each fa cto r assisted its e lf, I assigned a n u m e rica l ra n k in g as fo llo w s . F o r instance, degree 1 fo r distance w as assigned to those cases in w h ic h th e difference in distance betw een the antecedent and the co m p e tito r w as 1-5 IU s, i.e. the distance factor w as n o t v e ry stro n g and degree 2 distance w as assigned to those cases in w h ic h the diffe ren ce was 6 o r m ore IU s, i.e. the distance factor was v e ry strong. S im ila rly , fo r frequency o f m e ntion , degree 1 w as assigned to those cases in w h ic h the diffe ren ce in fre q u e n cy o f m e n tio n betw een the co m p e tito r and the antecedent was 1-5, i.e. th e fre que ncy o f Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 m entions fa cto r was n o t v e ry strong, and degree 2 w as assigned to those cases in w h ic h the d iffe re n ce was greater th a n 6, i.e. th e frequency o f m e n tio n fa cto r w as v e ry strong. hr exam ple (4) above, the d iffe re n ce in distance betw een the antecedent and the c o m p e tito r was 7 (degree 2), and th e diffe ren ce in frequency betw een the antecedent and th e co m p e tito r w as 1 (degree 1). In th is case, the distance fa cto r was stro n g e r than the fre q u e n cy factor. In exam ple (5) above, the c o m p e tito r was closer (d iffe re n ce in distance = 15) and m o re fre q u e n t (d iffe re n ce in fre que ncy = 2) tha n the antecedent. Frequency co n trib u te s less s tro n g ly tha n distance to the prom inence o f the c o m p e tito r in th is case. Table 12 show s th a t the greatest diffe ren ce in distance (degree 2: the antecedent is e ith e r closer o r fu rth e r fro m the c o m p e tito r b y 6 or m ore IU s) is e x e m p lifie d b y 47 occurrences (cf. to ta l o f second and fifth colum n in table 12, cells fram ed w ith b la ck th ic k b o rd e r). The antecedent is closer to the anaphor tha n the co m p e tito r in 94% (44 occurrences, fifth co lu m n ), and o n ly in 3 cases (6%, second colu m n) the co m p e tito r is closer to the anaphor tha n the antecedent. Degree 1 is also e xe m p lifie d b y 47 occurrences (fo u rth and th ird c o lu m n fram ed w ith double lin e b o rd e r). For th is degree o f distance, in 42 cases (89%: fo u rth colu m n in tab le 12) the antecedent is closer th a n the co m p e tito r, and in o n ly 5 cases (11% , th ird colu m n) th e co m p e tito r is closer to the anaphor th a n the antecedent. For the fre q u e n cy o f m e n tio n fa cto r, o u t o f 28 occurrences o f degree 2 (cf. to ta l o f th ird and s ix th ro w in table 12, fra m e d w ith black th ic k bo rd e r), in 82% (23: cf. th ird ro w in table 12) the antecedent is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 m o re fre q u e n t th a n the co m p e tito r. H o w e v e r, fo r degree 1, the antecedent is m o re fre q u e n t th a n the c o m p e tito r in o n ly 71% o f the tim e s (47, cf. fo u rth ro w in table 12, o u t o f 66 (a d d itio n o f fo u rth and fifth ro w in ta b le 12, fra m e d w ith d o u b le lin e b o rd e r). Table 12: The interaction of Distance and Frequency o f M entions and their relative strength F requ ency o f M n ts . I Ant. more frq by 6 or more mntns degree 2________ D istance ■ Comp, clsr by 6 or more IUs degree 2 Comp, clsr by 1-5 IUs degree 1 Ant. clser by 1-5 IUs degree 1 Ant more frqnt by 1-5 mnts degree 1 Comp more frqnt by 1-5 mnts degree 1 Comp more frqnt by 6 or more mnts degree 2 T otal Ant clser by To tal 6 or more IUs degree 2 23.(82%) 47.(71%) 19 (29%) 5 (18%) 3 (6%) 5 (11%) 42 (89%) 44 (94%) Table 13 show s th a t the e ffe ct o f distance on antecedent selection is s lig h tly greater in degree 2 th a n in degree 1 (94% vs. 89% ), b u t it is a lre a d y h ig h ly s ig n ific a n t even fo r degree 1. In o th e r w o rd s , if the antecedent is m u ch closer to the a n a p h o r th a n the c o m p e tito r, it w ill h a ve m ore p o s s ib ilitie s to w in o ve r th e c o m p e tito r. H o w e ve r, if the d iffe re n ce in distance is n o t as b ig (degree 1), th e n there is a greater chance th a t it w ill be overcom e b y o th e r factors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 Table 13: Ratio of success o f Distance by degree w ith respect to Frequency of M entions Distance A n te ce d e n t closer to a n a p h o r C o m p e tito r close r to a n a p h o r Degree 2 (stronger) 94% (44 o u t o f 47) 11% (3 o u t o f 47) Degree 1 (w eaker) 89% (42 o u t o f 47) 6% (5 o u t o f 47) D ifference 5% 5% Table 14 show s th a t the effect o f fre q u e n cy o n antecedent selection is greater in degree 2 th a n in degree 1 (82% vs. 71%). I f the antecedent is m u ch m o re fre q u e n tly m e n tio n e d th a n th e c o m p e tito r, it w ill have m o re p o s s ib ilitie s to w in o ve r the c o m p e tito r. H o w e ve r, i f the diffe ren ce betw een m e ntion s is n o t as b ig (degree 1), th e n there is a greater chance o f b e in g overcom e b y o th e r factors. Table 14: Ratio o f success of Frequencv of Mentions by degree with respect to Distance Freauencv o f M e n tio n s Antecedent more frequent Competitor more frequent Degree 2 (stronger) 82% (23 o u t o f 28) 18% (5 o u t 28) Degree 1 (w eaker) 71% (47 o u t o f 66) 29% (19 o u t o f 66) D ifference 11% 11% I f w e observe tables 13 and 14 w e can see th a t as lo n g as the degree o f d iffe re n ce increases fo r b o th distance and fre que ncy o f m e ntion , the re le v a n t fa c to r is m ore successful in antecedent selection. In other w o rd s, the stro nge r the fa cto r (degree 2), the less lik e ly it is to be overcom e (o r defeated) b y o th e r factors th a t m a y conspire fo r the selection o f th e c o m p e tito r fo r the reference o f th e anaphor. I f w e com pare the d iffe ren ce betw een degree 1 and degree 2 in table 13 fo r distance (la st ro w : 5%) and in tab le 14 fo r fre que ncy (last Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 ro w : 11%), w e note th a t the d iffe re n ce betw een degree 1 and 2 is greater fo r fre que ncy (11%) than it is fo r distance (5%). This is as expected. If degree 1 is already v e ry stro ng fo r a g iv e n fa cto r, as it is in distance, then degree 2 cannot be s ig n ific a n tly h ig h e r. I f degree 1 is w eaker, as it is in fre que ncy o f m ention, the n degree 2 can be s ig n ific a n tly h ig h e r. T herefore, distance is m ore im p o rta n t o r it is stro n g e r th a n frequency o f m e n tio n . M o reo ver, i f w e com pare h o w these tw o factors in te ra ct, it is in te re stin g to see th a t o u t o f a to ta l o f 24 occurrences (cf. table 11), w hen the fa cto r frequency favors the c o m p e tito r, 95% o f the tim es (22 occurrences) the antecedent is closer to the anaphor. O n the con trary, o u t o f a to ta l o f 8 occurrences w h e n th e fa cto r distance fa vo rs the co m p e tito r, o n ly 75% o f the tim es (6 occurrences) the antecedent is m ore fre q u e n tly m entioned th a n the c o m p e tito r. In o th e r w o rd s, distance fa vo rs the antecedent m ost o f the tim e , even if the co m p e tito r is m ore fre q u e n t. O n the o th e r h a n d , even w h e n the c o m p e tito r is closer, frequency o f m e ntion does n o t p la y an im p o rta n t role. I p e rfo rm e d the same c a lc u la tio n fo r the re m a in in g factors. I then com pared every factor to e ve ry o th e r, p a irw is e in the same w ay th a t I e xp la in e d above fo r distance an d fre que ncy o f m e n tio n . In p a rtic u la r, I considered the c o n flic t cases (w here the factors are p u llin g in d iffe re n t dire ctio n s: one to m ake th e antecedent m ore p ro m in e n t, and the o th e r to m ake the c o m p e tito r m o re p ro m in e n t) and I ranked them acco rd in g to degrees o f stre n g th . The data are presented in ap pen dix A , tables 18-27. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 5.3.3. Second Q u a n tita tiv e A n a ly s is : h o w th e 4 factors fu n c tio n s im u lta n e o u s ly E ven th o u g h the p re vio u s m e th o d presented is v e ry in fo rm a tiv e and de tailed about the in te ra c tio n o f each fa cto r w ith the oth er, it does n o t a llo w us to see h o w th e fo u r factors fu n c tio n at the same tim e . Therefore, I n o w presen t a second d e scrip tive q u a n tita tiv e m e thod th a t supplies m ore in fo rm a tio n a b o u t the in te ra ctio n o f these factors. I recoded the fo u r d iffe re n t fa cto rs in such a w ay th a t w he n com bined together each fa cto r w ill be th e firs t, second, th ird o r fo u rth d ig it in the strin g . Table 15 show s th e d e ta ils fo r the coding. Table 15: Coding for the second quantitative analysis D istance: 2000 = A ntecedent closer 1000 = C o m p e tito r closer 0000 = Tie F requency o f M e n tio n s: 200 = A n t m ore frequent 100 = C om p, m o re fre q u e n t 000= Tie S ynta ctic P o sitio n : 20 = A n t m ore accessible 10 = C om p, m o re accessible. 00 = Tie T h e m a tic R ole: 2 = A n t m ore accessible 1 = C om p, m o re accessible. 0 = Tie F or instance, a co d in g o f: 2222 m eans th a t the antecedent is closer, m ore fre q u e n t and m ore accessible b o th in synta ctic p o s itio n and th e m a tic ro le than the co m p e tito r. F ig u re 1 shows th a t the m ost fre q u e n t co m b in a tio n o f factors is 2200 (18.3%) fo llo w e d b y the co m b in a tio n 2222 (11.9% ), 2100 (8.3%) and 2220 (7.3%) in th a t order. A ll o f them p riv ile g e the p o sitio n o f distance. In a ll the analysis I have done so fa r, d ista nce is the stronger factor. The firs t co m bina tion (2200) show s th a t m o s t o f th e tim e, b o th distance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 and frequency fa v o r th e antecedent, w h e n there is a c o m p e tito r a va ila ble fo r reference id e n tific a tio n . In th is c o m b in a tio n , synta x and th e m a tic ro le do n o t p la y a ro le . A ll fo u r factors are im p o rta n t in the second co m b in a tio n . The antecedent w as p ic k e d fo r reference id e n tific a tio n and th e re fo re , it is expected th a t it is v e ry accessible fo r a ll the possible factors. The th ird co m b in a tio n (2100), a llo w s us to conclude th a t dista nce is m ore im p o rta n t th a n fre q u e n cy, even though, frequency is a v e ry im p o rta n t fa cto r as sh o w n in co m b in a tio n s 1 and 2. F in a lly , the last co m b in a tio n (2220), m a y lea d us to conclude th a t syntax is m ore im p o rta n t th a n th e m a tic ro le , a lth o u g h , in th is corpus, there is n o t s u ffic ie n t evidence to s u p p o rt th is. Figure 1: Combination o f Factors: Ties included 2 (2200) Percentage 20 T 18 -- 16 -- 14 -- (2222) 3 (2100) 4 (2220) 10 C N o C N Co-occurrence of Factors In a second analysis, the so-called ties (0 values) w ere recoded as 1000, 100, 10 and 1 re sp e ctive ly (fo r distance, fre q u e n cy o f m ention, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 syn ta ctic p o s itio n and the m atic ro le ). I w as interested to k n o w w h a t m akes th e lis te n e r select the antecedent and n o t the co m p e tito r, in o th e r w o rd s w h ic h fa cto r m akes the d iffe re n ce . I f a fa cto r is coded as 0 i t is n o t in fact, fa v o rin g the antecedent and th a t is w h y I m erged the m w ith those cases in w h ic h the c o m p e tito r w as favored. F ig u re 2 show s even m o re c le a rly th a t the c o m b in a tio n 2211 (antecedent closer a n d m ore fre q u e n t, syn ta ctic p o s itio n an d th e m a tic ro le are n o t fa v o rin g the antecedent) is the m ore fre q u e n t (34% ), fo llo w e d b y the co m b in a tio n 2111 (21%) and 2222 (13% ). A g a in , distance is o v e rw h e lm in g ly m ore p re d ic ta b le th a n th e rest o f the factors, fo llo w e d b y fre que ncy. S yntactic p o s itio n an d th e m a tic role do n o t p la y a ro le . These lim ite d and d e scrip tive re su lts are g o in g to be su p p o rte d b y a m ore so p h istica te d sta tistica l a n a lysis presented in the n e xt section. Figure 2: Combination o f factors: Ties re-coded Percentage 35 -r (2211) 30 2 (2111) 25 (2222) 20 15 10 CN CN CN CN CN CN CN CN CN CN CN CN CN Co-occurrence o f Factors Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 5.3.4. T h ird q u a n tita tiv e a n alysis: lo g is tic regression a n alysis A m u ltip le re gre ssio n analysis is the id e a l s ta tis tic a l m e thod to deal w ith m y research q u e stio n because it a llo w s us to see the in flu ence o f each va ria b le in the selection o f the dependent va ria b le . I a p p lie d a L o g is tic R egression because the dependent va ria b le is cate go rical and dichotom ous (fo r a d e s c rip tio n o f the dependent va ria b le see chapter 3). The L o g istic R egression pro ce d u re estim ates the outcom e o f a d ich otom ous depen den t v a ria b le acco rd in g to m u ltip le in d e p e n d e n t va ria b le s. O ne o f the m o st im p o rta n t questions is w h ic h o f these factors is m ore im p o rta n t fo r th e antecedent selection. P u t a d iffe re n t w a y: w h a t is the c o n trib u tio n o f each va ria b le fo r o u r m odel? To e va luate the variables, the R s ta tis tic is used to lo o k a t the p a rtia l c o rre la tio n betw een the dependent va ria b le and each o f the in d e p e n d e n t variables. R can range fro m -1 to 1. A p o s itiv e valu e indicates th a t as the va ria b le increases in valu e so does the lik e lih o o d o f the event o ccu rrin g . I f R is negative the o p posite is tru e . S m all values fo r R in d ica te th a t the v a ria b le has a sm a ll, p a rtia l c o n trib u tio n to the m o del (N o ru sis, 1994:5). P u t another w a y , R indicates the degree to w h ic h the sta tistica l m o d e l explains v a ria tio n in the dependent va ria b le , i.e. R ind ica tes h o w m u ch o f the flu c tu a tio n in the dependent va ria b le is e xp la in e d b y the ind epe nde nt va ria b le s (Sweet, 1999). To evaluate the m o d e l, the n o tio n o f lik e lih o o d (L) w as used, i.e. the p ro b a b ility o f the observed re sults, g ive n the param eters estim ated. Since the lik e lih o o d is a s m a ll nu m be r, less tha n 1, it is cu sto m a ry to use -2 tim es the lo g o f th e lik e lih o o d (-2LL) as a m easure o f h o w w e ll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 the estim ated m o d e l fits the data. A good m o d e l is one th a t re sults in a h ig h lik e lih o o d o f the observed re su lts. T h is translates to a sm a ll value fo r -2 L L ( if a m o del fits p e rfe c tly , the lik e lih o o d is 1, and -2 tim e s the lo g lik e lih o o d is 0) (N o ru sis, 1994:10). T herefore, if one im p o rta n t fa cto r is added to the m odel, the -2 L L w ill be sm a lle r because th a t fa c to r im p ro v e d th e m o d e l. Table 16 presents 6 possible m o dels, the firs t one co n sid e rin g the fo u r factors (distance, frequency o f m e n tio n , syn ta ctic p o s itio n and them atic ro le , ro w 2), the rest ta k in g o u t one fa cto r, o r tw o re sp e ctive ly. The firs t c o lu m n presents the -2 L L used to evaluate each m o d e l in com parison to the o th e r. The th ird an d fo u rth co lu m n presen t a ll th e s ig n ific a n t fa cto rs w ith its co rre sp o n d e n t R statistics. T w o param eters are g o in g to be considered to com pare the re la tiv e stre n g th o f the factors: 1) the sig n ifica n ce o f each fa cto r (co lu m n 3, table 16) in each m o d e l and its corresp ond ent R statistics (co lu m n 4, table 16). 2) the -2 L L th a t evaluates the m o d e l as a w h o le (co lu m n 2, in table 16). The re su lts in table 16 show the fo llo w in g : • D istance is the m ost im p o rta n t fa c to r because firs t, distance is a lw ays s ig n ific a n t i f present (cf. th ird co lu m n ) and second, if distance is n o t in clu d e d (ro w 3, table 16), the lik e lih o o d o f the m o del decreases as exp laine d above: cf. 45.664 vs. 117.152 • Syntax is the least im p o rta n t fa c to r because firs t, syntax is n o t s ig n ific a n t in an y m o del (cf. th ird co lu m n , table 16), and second, the m o del w ith o u t syntax (ro w 5, ta b le 16) does n o t increase its Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 lik e lih o o d v e ry m u ch (cf. w /o u t syntax: 46.088 w ith a ll fo u r factors: 45.664). • F requency is m ore im p o rta n t than syn ta x and th e m a tic ro le , b u t less im p o rta n t th a n distance. F irst, the m o d e l w ith o u t distance (ro w 3, table 16) has o n ly one s ig n ific a n t fa cto r: frequency. Second, in o rd e r to com pare th e fre q u e n cy o f m e n tio n fa c to r a n d the them atic ro le , a m o d e l w ith o n ly these tw o factors w as ru n (ro w 7, table 16), an d fre q u e n cy w as the o n ly s ig n ific a n t fa cto r. F in a lly , in a m o del w ith o u t fre que ncy (ro w 4, tab le 16) o n ly distance p re v a ils . • T h em a tic ro le is o n ly s ig n ific a n t w hen syn ta x is n o t p re se n t (ro w 4, ta b le 16). H o w e ve r, distance in th is m o d e l is o v e rw h e lm in g ly m ore p re d icta b le (cf. R = .3677). Table 16: Comparison o f models w ith different factors. Factors considered in the m odel -2 Log L ikelihood Significant factor R statistics 4 factors 45.664 distance (p c.O l) .3589 W /o u t distance 117.152 frq n cy (p c.O l) .2743 W /o u t fre que ncy 51.989 distance (p c.O l) .3945 W /o u t syntax 46.088 distance (p c.O l) th r l (pc.05) D ist:.3677 T h R l:.1337 W /o u t the m r l 52.677 distance (p c.O l) .4072 Freq and the m r l 120.687 freqncy (pc.O l) .2823 T herefore, it is clea r th a t distance is m ore im p o rta n t th a n the rest o f the factors and th a t syn ta x is the least im p o rta n t fa cto r. The rest o f th e data sho w th a t fre q u e n cy is second in im p o rta n ce . Some facts th a t need fu rth e r research are: firs t, w h y is th e m a tic ro le s ig n ific a n t in th e m o d e l w ith o u t syn ta x, w h ile frequency o f m e n tio n in th is m o del is n o t sig n ifica n t? A n d the second question is, w h y the -2 L L o f the m odel Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 w ith o u t them atic ro le (52.677) is w orse th a n the m o d e l w ith o u t fre que ncy (51.989). These issues need fu rth e r research, fo r n o w , I have sh o w n s u ffic ie n t evidence fo r the fo llo w in g ra n k in g o f factors fo r antecedent selection: d ista n ce , frequency o f m e n tio n , th e m a tic ro le and syn ta ctic p o s itio n (in th a t o rd e r). 5.3.5. S um m ary o f 5.3 S um m ing u p , I have considered fo u r factors o f relevance fo r a d e fin itio n o f prom inence (distance, frequency o f m e n tio n , syn ta ctic p o s itio n and them atic ro le ). I have conducted three d iffe re n t q u a n tita tiv e analyses (a p a irw is e com parison, a analysis o f h o w the fo u r factors fu n c tio n s im u lta n e o u s ly , and a lo g is tic regression analysis) w h ic h have show n th a t distance is the m ost im p o rta n t fa cto r th a t can m ake an e n tity p ro m in e n t, fo llo w e d b y fre q u e n cy o f m e n tio n . S yntactic p o s itio n and th e m a tic ro le do n o t p la y a ro le in antecedent se le ction in discourse. S u m m in g u p : Table 17: The Ranking of Factors for Prominence Syntactic position Distance > Frequency > Thematic role In the fo llo w in g se ctio n I w ill p ro v id e a c o g n itiv e m o tiv a tio n fo r the ra n k in g in tab le 17. 5.4. E x p la in in g th e ra n k in g Each o f m y q u a n tita tiv e analysis has co n firm e d the ra n k in g o f factors g ive n in table 17. In th is section, m y goal w ill be to p ro v id e a ra tio n a le o r m o tiv a tio n fo r the ra n kin g . I w ill b e g in b y re v ie w in g Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 som e re le va n t ideas o f G em sbacher (1990) on h o w in fo rm a tio n is m e n ta lly represented o n -lin e . 5.4.1 T he m e n ta l re p re se n ta tio n o f a te x t in G em sbacher's th e o ry (1990) In chapter 2 , 1 presented G em sbacher's (1990) th e o ry o f a m ental re pre senta tion o f a text. G em sbacher (1990: 2) cla im s th a t w e b u ild the m e n ta l re p re se n ta tio n u sin g m em ory nodes. S uppression and enhancem ent c o n tro l the a c tiv a tio n o f nodes in m e m o ry. Suppression is re la te d to the distance factor: the closer the antecedent, the less the p o s s ib ility o f e n coun terin g a co m p e tito r fo r th e reference, w h ic h consequently, w o u ld suppress the a c tiv a tio n o f th e antecedent. T h is agrees w ith p sych olog ical exp erim e nts (C la rk & Sengul, 1979; E h rlic h & Rayner, 1983; Frederiksen, 1981) th a t reveal the im p o rta n ce o f closeness in the in te rp re ta tio n o f anaphors. In general, it appears th a t lin g u is tic processes are p re fe rre d th a t reduce relevant dom ains fo r any lin g u is tic operation. For instance, H a w kin s (1998, 1999 and to appear) argues th a t there is a perform an ce preference to m in im iz e the dom ains in w h ic h synta ctic and sem antic re la tion s, such as co n stitu e n t g ro u p in g , are processed w ith in a sentence. E nhancem ent is related to frequency o f m e n tio n : each tim e an e n tity is repeated, it gains in a ctiva tio n . T h ere fore , the m ore frequent and e n tity is, the m ore activated. G em sbacher (1990) is n o t ve ry clear a b o u t h o w syntactic and th e m a tic in fo rm a tio n is represented m e n ta lly , n o r about w h a t in fo rm a tio n is re le va n t fo r the representation o f th e te x t and fo r an aph oric in te rp re ta tio n . W hen p resen ting th e fir s t m e n tio n effect Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 (the fa ct th a t e n titie s th a t are firs t m e n tio n e d are m o re accessible tha n the rest, see chapter 2 fo r a m ore extensive e x p la n a tio n ), she disregards the im p o rta n ce o f syn ta ctic and sem antic in fo rm a tio n . R egarding sem antic in fo rm a tio n she asserts: com prehenders m u st represent sentences in such a w a y th a t firs t-m e n tio n e d p a rtic ip a n ts are m o re accessible. B u t sem antic ro le is n o t the fa cto r u n d e rly in g th is greater acce ssibility. (G em sbacher, 1990: 14) In th is qu ote she is re fe rrin g to th e m a tic ro le s such as agents o r them es. R e g a rd in g synta ctic p o s itio n she adds: the advantage o f firs t m e n tio n is n o t lo s t w h e n b o th the firs t and second-m entioned p a rtic ip a n ts are syn ta ctic subjects. The advantage o f firs t m e n tio n is n o t lo s t even w h e n firs t-m e n tio n e d p a rtic ip a n ts are n o t syn ta ctic subjects. (G em sbacher, 1990: 24) H o w e ve r, some synta ctic in fo rm a tio n is represented in the m e n ta l re p re se n ta tio n o f the te xt: exp erim e nts w ith clauses th a t are m o re d e p e n d e n t than those w e m a n ip u la te d suggest th a t com pre hen ders b u ild ric h m e n ta l representations th a t c a p tu re cla u sa l dependencies. (42) The q u e stio n w e m u st ask here is: h o w d o com prehenders m e n ta lly repre sent synta ctic in fo rm a tio n associated w ith entities? G em sbacher (1990) claim s th a t syn ta ctic in fo rm a tio n is n o t re taine d fo r a lo n g tim e because i t is p a rt o f the surface fo rm o f each in d iv id u a l utterance. B u t tw o d iffe re n t syn ta ctic fo rm s can co n ve y the same m eaning, and the syn ta ctic co d in g o f a re fe rrin g expression can v a ry w ith each d iffe re n t m e n tio n in a te xt. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 C onsider a passage o f te xt. I f it is w e ll-co m po sed, th e n a ll sentences s h o u ld co n ve y the same them e. B u t a ll sentences do n o t have th e same synta ctic fo rm . O ve r the span o f the passage the sentences' syn ta ctic fo rm s changes faster th a n th e ir the m e; therefore, syn ta ctic fo rm is considered surface in fo rm a tio n , w h ile them e is n o t. (G em sbacher, 1990: 162) H o w e ve r, sem antic, syn ta ctic and p ra g m a tic in fo rm a tio n is encoded a t som e le ve l: in fo rm a tio n fro m o th e r sources (such as sem antic, syntactic, and p ra g m a tic context) sho uld also trig g e r suppression and enhancem ent, b u t m ore s lo w ly and less p o w e rfu lly . (112) Therefore, it is n o t clea r e xa ctly h o w m u ch syn ta ctic and the m atic in fo rm a tio n is a v a ila b le in the m e ntal re p re se n ta tio n o f a text in he r v ie w , and if th is in fo rm a tio n is indeed encoded in the m ental re p re se n ta tio n . S um m ing u p , it seems th a t G em sbacher som etim es considers synta ctic and sem antic in fo rm a tio n to be p a rt o f the m e n ta l representation o f the te x t an d som etim es she does n o t. In a n y event it is n o t clear, if the in fo rm a tio n is present, w h a t its fo rm is. A t one p o in t she states th a t sem antic a n d syn ta ctic context is re presented in m em ory cells (G em sbacher, 1990: 104), b u t she stops there an d does n o t c la rify fu rth e r. A lth o u g h G em sbacher (1990) m entions the fa c t th a t re fe re n tia l distance and fre que ncy (w h ic h she includes in to p ic a lity ) p la y a ro le in anaphor in te rp re ta tio n , th e precise m anner in w h ic h th is in fo rm a tio n Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 is stored in the m e n ta l re pre senta tion o f a te x t is n o t clear. R elevant questions to ask o f th is m odel are: w h a t kin d s o f in fo rm a tio n are w e accessing w h e n re p e a tin g a g iv e n e n tity ? W h a t is the in fo rm a tio n th a t is recovered? H o w is syntactic and sem antic p o s itio n represented? H o w is distance represented in the m e n ta l re pre senta tion o f the text? 5.4.2. M y discourse representation theory M y discourse repre senta tion th e o ry is based on G em sbacher's (1990) m o del. I assume th a t th e m e n ta l re pre senta tion o f a te x t consists o f a h ie ra rc h ic a l arrangem ent o f d isco u rse and n o n discourse u n its (te x tu a l and n o n te xtu a l P-sets, cf. C h a p te r 2 and 4 fo r m ore details) as fo llo w s : 1) T e xtu a l P-sets: each discourse is h ie ra rc h ic a lly organized: discourse u n its are arranged w ith d iffe re n t levels o f a c tiv a tio n (cf. cha pte r 4). T h is m o del agrees w ith th e Jo h nson -La ird (1983) m ental m o d e l re p re se n ta tio n and the va n D ijk and K in tsch (1983) s itu a tio n a l m o d e l re p re se n ta tio n . 2) N o n -te x tu a l P-sets: no n disco urse u n its in clu d e the im m e diate s itu a tio n set (p hysical s itu a tio n ), th e set co n stitu te d b y scripts o r en cyclopedic kno w le dg e, and the la rg e r s itu a tio n te x t th a t are also ra nked acco rdin g to accessibility (i.e . th e y have d iffe re n t levels o f a c tiv a tio n , cf. chapter 2). 3) E n titie s w ith in P-sets are also ra n ke d according to acce ssibility (i.e. th e y have d iffe re n t levels o f a c tiv a tio n ). O ne o f the e n titie s w ith in the h ig h a cce ssib ility P-sets is the m o st p ro m in e n t. F ig u re 3 b e lo w presents 3 te x tu a l P-sets (discourse u n its: ovals) and tw o n o n -te xtu a l P-sets (circles). The d iffe re n t thicknesses o f the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 lines sig n a l the d iffe re n t levels o f a c tiv a tio n . E n titie s w ith in P-sets (K is the m o st p ro m in e n t e n tity and the re m a in in g x's are other e n titie s w ith in th e m ore activated te xtu a l u n it) are also ra nked according to acce ssibility. The accessibility o f e n titie s is o n ly gram m aticalized fo r h ig h a cce ssib ility P-sets (cu rre n t d isco urse u n it and p h ysical s itu a tio n set). T h is m eans th a t speakers can o n ly m ake a prom inence d is tin c tio n am ong the e n titie s o f h ig h acce ssibility P-sets. T h is is com m on in m ost languages as n o te d in chapter 4 (4.4.1). Figure 3: Mental representation o f a text N o n -te xtu a l P-set N o n -te xtu a l P-set T e xtu a l P-sets 5.4.3. M o tiv a tin g the h ie ra rch y o f factors F ig u re 3 represents the h ie ra rc h ic a l n a tu re o f the m ental re pre senta tion o f the text. G iven the fa c t th a t discourse is sequential, and th a t its m e n ta l representation is h ie ra rc h ic a l, the m ental o p e ra tio n to tra n s fo rm sequential strin gs to h ie ra rc h ic a l stru cture s should take Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 tim e . H o w th is m e n ta l o p e ra tio n takes place has been a m a tte r o f controversy. F irs t, the re is th e th e o ry o f co m p a rtm e n ta l m e m o ry w h ic h states th a t the re is a b u ffe r o r w o rk in g m e m o ry com ponent th a t represents e n titie s in a d iffe re n t w a y than e n titie s w ith in the lo n g term m e m ory o r pe rm a n e n t re p re s e n ta tio n o f the te xt. T here have been d iffe re n t theories and v ie w s o n th is p o in t and no ne o f the m has resolved the issue s a tis fa c to rily . F o r instance, i t is n o t clea r h o w large w o rk in g m e m o ry is, i.e. h o w m a n y w o rd s can be sto re d w ith in th a t te m p o ra ry com ponent. N onetheless, if w e accept a th e o ry lik e th is, it w o u ld m ake sense th a t the o rg a n iz a tio n o f e n titie s in so-called w o rk in g m e m o ry s h o u ld be d iffe re n t fro m the h ie ra rc h ic a l o rg a n iza tio n o f the m o re p e rm a n e n t m e ntal re p re se n ta tio n o f the text. Since discourse p ro d u c tio n is sequential, g ive n th a t s trin g s o f w o rd s are produ ced in re a l tim e (one a fte r the other), it w o u ld n o t be s u rp ris in g th a t the o rg a n iza tio n o f th is te m p o ra ry com ponent s h o u ld also be sequential. I f so, (and fu rth e r research is needed o n th is ), e n titie s processed closer in tim e s h o u ld be m ore accessible th a n m ore d ista n t e n titie s in tim e . A n d the clo se r e n titie s w o u ld be m o re accessible than a n yth in g th a t w as a lre a d y processed h ie ra rch ica lly, i.e. assigned to a P- set w ith a le ve l o f a c tiv a tio n . A d iffe re n t th e o ry o f m e m o ry (C ow an, 1993; M e D o n a ld et al. in press, M cC le lla n d and R u m e lh a rt, 1986b) rejects th is co m p a rtm e n ta l v ie w o f m e m ory and cla im s th a t th e o n ly d iffe re n ce betw een en tities w ith in w o rk in g m e m o ry a n d e n titie s w ith in a m o re p e rm a n e n t re pre senta tion is the le v e l o f a c tiv a tio n . T h e re fo re , w o rk in g m em ory e n titie s have a h ig h e r le v e l o f a c tiv a tio n (above a g iv e n th re sh o ld ) and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 e n titie s alre ad y processed a n d loca ted in the m ore perm anent re pre senta tion o f the te x t are less a ctiva te d . I t is possible to conclude th a t the discourse th a t has ju s t been p ro d u ce d has a v e ry h ig h le v e l o f a c tiv a tio n , even h ig h e r th a n fo r a n y o f the e n titie s represented in the p e rm a n e n t com ponent o f th e m e m o ry. I can n o w address the c ru c ia l issue o f w h y the factors are ra n ke d in the o rd e r th e y are (cf. ta b le 17). I am g o in g to argue, firs t, th a t the g ra m m a tica l p o s itio n an d th e m a tic ro le are n o t pe rm a n e n t p ro p e rtie s o f e n titie s and tha t th is e x p la in s th e ir ra n k in g . Second, I w ill also try to m o tiv a te the fact th a t dista n ce is h ig h e r o n the ra n k in g th a n frequency. 5.4.3.I. Grammatical and them atic role S yntactic and th e m a tic in fo rm a tio n do n o t p ro v id e co n siste n t id e n tific a tio n o f an e n tity . W e can have m a n y d iffe re n t g ra m m a tica l and the m atic roles associated w ith one and the sam e e n tity , because these gram m a tical p ro je c tio n s are a fu n c tio n o f in d iv id u a l sentences. The fa c t th a t one e n tity appears as a subject does n o t m ean th a t it w ill be re trie ve d late r in discourse in th e sam e syn ta ctic p o sitio n . G ra m m a tica l p o s itio n an d th e m a tic ro le are im p o rta n t fo r sentence in te rn a l representation (see fo r instance R e in h a rt 1983) b u t th e y are n o t s ig n ific a n t fo r a discourse re p re se n ta tio n . S yntactic and th e m a tic in fo rm a tio n are n o t re pre sente d in th e m e n ta l re p re se n ta tio n o f the te x t because th e ir in fo rm a tio n is te m p o ra ry (i.e. th is in fo rm a tio n is n o t stored in any k in d o f p e rm a n e n t m e m o ry storage). I t is clear th a t distance and frequency p la y a m o re im p o rta n t ro le th a n g ra m m a tica l and th e m a tic ro le fo r antecedent selection. A n d the fa ct th a t I coded Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 the gra m m a tica l and the m atic ro le o f the m o st recent m e n tio n o f an antecedent does n o t m ean th a t th a t sp e cific ro le has any p rio rity , p re cise ly because the m atic and g ra m m a tica l roles do n o t p ro v id e co n siste n t id e n tific a tio n . S.4.3.2. D istance and fre q u e n cy The next question to ask is w h y distance sho uld be m o re im p o rta n t th a n frequency. F o r b o th fa cto rs m echanism s o f su p p re ssio n and enhancem ent are at w o rk (G em sbacher, 1990). F or distance, the closer the antecedent, the m o re enhanced, and therefore the m o re the co m p e tito r is suppressed. The closer the antecedent is, the less o p p o rtu n ity there is fo r o th e r co m p e tito rs to intervene and suppress the antecedent’s a ctiva tio n . S im ila rly , fo r frequency, each tim e an e n tity is m entioned, its a c tiv a tio n is enhanced and con sequ ently the a c tiv a tio n o f the co m p e tito r is suppressed. There are d iffe re n t po ssible exp la n a tio n s fo r the fact th a t distance is m ore im p o rta n t tha n fre que ncy. F irs t, th e re is a general preference in b o th perform ance and g ra m m a r to m in im iz e the re le va n t d o m ain s fo r lin g u is tic operations (H a w k in s , 1998). Therefore, the in te rp re ta tio n o f an an aphor sho uld be re so lve d w ith in th e shortest distance possible. The p rin c ip le 'fin d the closest antecedent, regardless o f its fre q u e n cy o f m e n tio n ' is an e ffic ie n t and econom ic processing o p e ra tio n . A n d m a kin g a frequency assessment (b y co m p a rin g the nu m be r o f m e ntion s fo r d iffe re n t entities) is p la u s ib ly a m ore com plex processing o p e ra tio n . O n the other ha nd, the fre q u e n cy va lu e o f an e n tity c o u ld be a u to m a tic a lly calculated as soon as the an aph or is processed, and therefore, co u ld be an in te g ra l p a rt o f the a ctiva tio n o f e n titie s. In Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 o rd e r to e x p la in the ra n k in g in table 17 w ith in th is a lte rn a tive v ie w (cf. C ow an, 1993 and M e D o n a ld e t a l in press), w e w o u ld need to assume th a t recency o f m e n tio n re sults in a greater a c tiv a tio n th a n num ber o f m entions and the precise m o tiv a tio n fo r th is w o u ld need to be ju s tifie d . F in a lly , g iv e n m y discourse re p re se n ta tio n th e o ry , the m ental representation o f a te x t is n o t a u to m a tic o r instantaneous. O n the co n tra ry, the m e n ta l re p re se n ta tio n is b u ilt w h ile speakers and listeners are ta lk in g . The sequential n a tu re o f speech is d iffe re n t fro m the h ie ra rch ica l s tru c tu re o f the m e n ta l re p re se n ta tio n . Therefore, there sh o u ld be a tra n s itio n fro m th e se q u e n tia l s trin g o f the speech to the h ie ra rch ica l re p re se n ta tio n in w h ic h the e n titie s are n o t yet assigned to a P-set o r to a le ve l o f fre que ncy o f m e n tio n . This is m ore im p o rta n t fo r Pers. P ron oun , C litic , D ef. N u ll S ubject because the p h o n o lo g ica l m a te ria l does n o t s u p p ly m u ch in fo rm a tio n and th e ir in te rp re ta tio n relies m ore h e a v ily in the in fo rm a tio n p ro v id e d b y the antecedent. Therefore, w hen lo o k in g fo r an antecedent, the firs t place to search is the se q u e n tia lly organ ized, v e ry te m p o ra ry com ponent in w h ic h e n titie s are s till in the process o f b e in g assigned a le ve l o f a ctiva tio n and a co rre sp o n d in g P-set. I f th is te m p o ra ry com ponent is sequential, the closer e n tity w o u ld be m ore accessible th a t the one n o t present o r fu rth e r aw a y in th is te m p o ra ry com ponent. The issue th a t th is raises is, h o w lo n g the sequential, u n s tru c tu re d s trin g can be. Tw o fu rth e r questions m u st be asked: w h y is i t the case th a t n o t a ll the antecedents are located in a closer p o s itio n th a n th e ir Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 com petitors? A n d , if distance is so im p o rta n t, w h y does fre q u e n cy p la y any ro le a t all? F or the fe w cases in w h ic h the closer antecedent w as n o t selected, som e o th e r factors (fo r instance, fre q u e n cy o f m e n tio n , syn ta ctic p o s itio n o r the m atic ro le ) w ere fa v o rin g it. T herefore, distance is v e ry im p o rta n t b u t it is n o t th e o n ly fa c to r th a t is considered in antecedent se le ctio n . Second, the s ta tis tic a l c a lc u la tio n sho w ed th a t fre q u e n cy p la ys a ro le , even w h e n distance is n o t p re se n t in th e m o del. Language has a lo t o f re du n d a n cy and the closer antecedent is also, in general, the m ore fre q u e n t one. 5.4.4. The interaction of factors in the previous literature K ib rik (1996) states th a t the m a jo r challenge fo r c u rre n t the orie s o f anaphora is to account fo r th e in te ra c tio n o f factors th a t in flu e n ce antecedent selection. Several fa cto rs have been s tu d ie d in d e p e n d e n tly , b u t no-one except fo r K ib rik has y e t addressed the q u e stio n o f th e ir in te ra c tio n . Therefore, K ib rik ’s p o in t is v e ry va lu a b le in th is respect and needs to be taken as a p o in t o f d e p a rtu re . H o w e ve r, K ib rik 's o w n a p pro ach to th is p ro b le m fa ils fo r la ck o f a clear and re lia b le m e th o d o lo g y. F irs t, it is un clear h o w the co rre la tio n s are m ade in h is s ta tis tic a l p ro g ra m . H e states th a t factors w ere selected 'th ro u g h a special SH O EBO X filte rin g p ro ce d u re ' (265), b u t he does n o t te ll w h a t th is is. Second, i t is un clear h o w the a rb itra ry num bers fo r the values o f the factors have been assigned. T h ird , the exte rn a l evidence p ro v id e d fo r the p re d ic tio n o f d iffe re n t types o f N Ps, is based o n co m p le te ly a rb itra ry n u m be rs, obscure co rre la tio n s and an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 a rith m e tic c o m p u ta tio n . In o th e r w o rd s , h is m e th o d o lo g y lacks m a th e m a tica l s o p h is tic a tio n . 5.5. C o n c lu s io n In th is ch a p te r I have p ro v id e d to a d e fin itio n o f prom inence based on the in te ra c tio n o f fo u r d iffe re n t factors: distance, fre que ncy o f m e n tio n , syn ta ctic p o s itio n and th e m a tic ro le . A fte r co n d u ctin g three q u a n tita tiv e analyses (a p a ir-w is e com pariso n, a d e s c rip tio n o f h o w the fo u r factors fu n c tio n s im u lta n e o u s ly , an d an a p p lie d m u ltip le regression), I have co n clu d e d th a t distance is the m o st im p o rta n t fa c to r fo r m a kin g an e n tity p ro m in e n t, fo llo w e d b y fre q u e n cy o f m e n tio n . The synta ctic p o s itio n an d them atic ro le factors do n o t p la y a ro le in the prom ine nce o f e n titie s in discourse. M y ap pro ach is m o re com plete th a n p re vio u s approaches because it addresses a q u e stio n th a t m o st o f the research on anaphora has ig n o re d h ith e rto , n a m e ly: w h a t is the in te ra c tio n o f factors in the selection and in te rp re ta tio n o f d e fin ite NPs? In d o in g so, it com bines a tra d itio n a l approach, in w h ic h each fa c to r is com pared w ith e ve ry o th e r, w ith m ore so p h istica te d sta tis tic a l techniques. F or exam ple an a p p lie d lo g is tic regression is used in o rd e r to c o n tro l fo r a ll the possible com bina tions an d c o m p u ta tio n s th a t are n o t re a d ily ava ila ble w h e n ra w num bers are m a n ip u la te d . T his a n alysis gives greater re lia b ility to m y e m p iric a l cla im s an d p ro vid e s a n e w s ta tis tic a l m e thod fo r anaphora research. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 C h a p te r 6 C o n clu sio n s 6.0. In tro d u c tio n T h is d isse rta tio n has p ropo sed a m o d e l o f definiteness types in Spanish th a t achieves th e fo llo w in g : 1) it is based on an analysis o f n a tu ra l data, n o t ye t u n d e rta ke n in S panish (w ith one exception, cf. B e n tiv o g lio , 1983); 2) it integrates ce rta in ke y in s ig h ts fro m d iffe re n t research tra d itio n s and goes beyond p re v io u s approaches w ith in each o f these tra d itio n s (Sem antics [R ussell, 1905], D iscourse/P ra gm a tics [H a w k in s , 1991; A rie l, 1990], P sycholin guistics [G em sbacher, 1990]); 3) it m akes an e m p iric a l c o n trib u tio n , g iv in g a m ore th o ro u g h account o f ce rta in fu n d a m e n ta l n o tio n s in th is area (uniqueness, a cce ssib ility, prom inence); 4) i t m akes a th e o re tica l c o n trib u tio n b y e xp la in in g h o w d iffe re n t factors in te ra c t in the in te rp re ta tio n and selection o f d e fin ite NPs. In the process, I have o ffe re d an answ er to the q u e stio n th a t was raised in chapter 4: w h y do speakers select one type o f d e fin ite N P ra th e r th a n another? Speakers select d iffe re n t definiteness type s in o rd e r to re fe r to a P-set (i.e. p ra g m a tic a lly stru ctu re d set o f e n titie s) w ith an a p p ro p ria te le ve l o f a cce ssib ility. A ppropriateness c o n d itio n s have been p ro v id e d fo r the fo llo w in g de finiten ess types in Spanish: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 D e fin ite A rtic le (D e fA rt+ N ) Preposed D e m o n s tra tiv e (D em +N ) Postposed D e m o n stra tive (1) la idea 'th e id e a ' (2) esta idea 'th is id e a ' (3) la idea esta D e fA rt+ N + D e m 'th is idea h e re ' lit. 'T he idea th is ' D e m o n stra tive P ro n o u n (D em ) Personal P ro n o u n (Pers P ronoun) D e fin ite N u ll S ubject (D e f N u ll Subj) (4) esta 'th is ' (5) ella 'she ' (6) pro p ie n sa . C litic o f D O and IO (C litic ) 'p ro th in k s ' (7) Ella lo piensa ' She th in k s it ' I have also so lve d a p ro b le m fo r th e uniqueness th e o ry o f d e fin ite d e scrip tio n s. W hen the uniqueness c o n d itio n appears n o t to be satisfied, it is su p p le m e n te d b y the n o tio n o f prom inence. T h is d isse rta tio n has p ro v id e d a d e fin itio n o f p ro m in e n ce based o n the in te ra ctio n o f factors (such as distance a n d fre que ncy) th a t c o n trib u te to m a kin g the reference u n iq u e (cf. chapter 5). 6.1. T he a p p ro p ria te n e ss c o n d itio n s o f d e fin ite n e ss types In chapter 4, I fo rm u la te d the a p pro pria tene ss c o n d itio n s fo r the definiteness types in S panish based on the a n alysis o f a corpus o f data th a t consists o f a p p ro x im a te ly 3 ho urs o f in te rv ie w s and n a tu ra l conversations in S panish betw een y o u n g a d u lts and m y s e lf (cf. chapter 3 fo r an extensive d e s c rip tio n o f the data an alyzed ). I analyzed a ll o f the d e fin ite expressions, a to ta l o f 930 (376 D e fA rt+ N s , 62 D em +N s, 10 D efA rt+ N + D e m s, 33 D em s, 64 Pers P ron oun s, 186 C litic s and 199 D ef. N u ll Subjs). M o st o f the de finiten ess types re fe r to h ig h a cce ssib ility P-sets (D em +N , D em , Pers P ro n o u n , C litic an d D e f N u ll Subj). H o w e ve r, D e fA rt+ N + D e m refers m o s tly to m id a cce ssib ility P-sets and D e fA rt+ N Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 refers a lm o st e xclu sive ly to lo w a c c e s s ib ility P-sets. I t is im p o rta n t to note th a t D e fA rt+ N m ay re fe r to a n y k in d o f P-set. In section 4.3., I described th e appropriateness c o n d itio n s o f d e fin ite n e ss types, and in section 4.4. I e xp la in e d w h y these d e finiten ess types have the appropriateness c o n d itio n s th a t they do. D e fA rt+ N , D e m + N , D e fA rt+ N + D e m and D e m ha ve a s tric t uniqueness re q u ire m e n t w ith in the P-set th a t is a ctiva te d . The re fe re n t o f D e fA rt+ N can occur in any k in d o f P-set, an d th is is the p re fe rre d d e fin ite n e ss ty p e fo r e n titie s in th e lo w a cce ssib ility P-sets. D e fA rt+ N + D e m refers to a u n iq u e e n tity w ith in m id a cc e s s ib ility P-sets and D e m + N an d D em b o th re fe r to a u n iq u e e n tity w ith in h ig h a cce ssib ility P-sets. The d iffe re n ce b e tw e e n D em +N and D e m is re la te d to the k in d o f P-sets th e y re fe r to (D e m to p h ysica l s itu a tio n sets and D e m + N to p re vio u s te xt) and to th e le v e l o f acce ssibility o f e n titie s w ith in th e P-sets (Dem refers to clo se r e n titie s and D e m + N to fa rth e r e n titie s ). Pers P ronoun, C litic and D e f N u ll Subj refer to h ig h a cce ssib ility P-sets a n d th e y often achieve u n ique ness th ro u g h p ro m in e n ce , as discussed in chapter 5. 6.2. M o tiv a tin g the a p p ro p ria te n e ss c o n d itio n s T he appropriateness c o n d itio n s o f th e d iffe re n t d e fin ite n e ss types can be m o tiva te d alo n g the fo llo w in g lines. F irst, i t is to be expected th a t h ig h acce ssibility P-sets can be referred to w ith a w id e r v a rie ty o f definiteness types, fo r th e reasons I enum erated in 4.4.1. Second, th e p h o n o lo g ica l size o f these fo rm s corresponds to the a cce ssib ility o f the P-sets th e y re fe rre d to . I f the e n tity is v e ry accessible, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 v e ry little lin g u is tic m a te ria l is needed to re fe r to it (cf. 4.4.2 and G ivon, 1992; A rie l, 1990). T h ird , definiteness types re fe rrin g to lo w e r a cce ssib ility P-sets w ill need to co n ta in m o re in fo rm a tio n in o rd e r to ensure th a t the liste n e r can in te rp re t the N P co rre ctly. Therefore, NPs w ith definiteness types th a t re fe r to lo w e r a cce ssib ility P-sets are expected to be lo n g e r and m ore co m p le x th a n those re fe rrin g to h ig h acce ssibility P-sets (cf. 4.4.3). 6.3. T e xtu a l P-sets A P-set (as sho w n in chapter 2) is a p ra g m a tic a lly stru ctu re d set o f e n titie s w ith in the o v e ra ll u n ive rse o f disco urse , w h ic h defines the param eters fo r the uniqueness re q u ire m e n t. H a w k in s (1991) asserts th a t the uniqueness re quire m en t fo r the d e fin ite a rtic le is re stricte d to a p ra g m a tic set. In o th e r w o rd s, speakers d e lim it a P-set and the n they re fe r to a u n iq u e e n tity w ith in the lim its o f th a t P-set. I have p ro v id e d a re -in te rp re ta tio n o f the n o tio n o f a P-set th a t in vo lve s a m o d e l o f discourse re p re se n ta tio n . E ven th o u g h a ll the d iffe re n t types o f P-sets th a t have been discussed b y H a w kin s (1991) are clear, I have d e fin e d som e a d d itio n a l types o f s tru c tu rin g , sp e cifica lly fo r P-sets w h ic h are discourse sen sitive. The lin g u is tic in d ica to rs o f these te x tu a l P-sets and th e ir in te rn a l p ro p e rtie s are discussed in section 4.2.2. D iscourse u n its are ofte n d e lim ite d b y lin g u is tic in d ica to rs. F in d in g the lin g u is tic in d ica to rs th a t s ig n a l a change o f discourse u n it gives us a w a y o f d e fin in g discourse s tru ctu re . In m y data, fo u r factors w ere fo u n d re le va n t fo r m a rkin g th e s h iftin g o f discourse u n its: the occurrence o f discourse m arkers; u n it openers; questions b y the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 in te rv ie w e r; and quotes. The occurrence o f a discourse m a rke r o r a u n it opener, and a question o r a com m en t b y the in te rv ie w e r are s u ffic ie n t in d ic a to rs o f the onset o f n e w discourse u n it. The occurrence o f a qu ote is n e ith e r a necessary n o r a s u ffic ie n t co n d itio n fo r a discourse u n it s h ift in m y corpus. T h is fa cto r is here presented anyw ay, because th is d is trib u tio n m ig h t be an a rtifa c t o f m y data and m a y n o t be tru e fo r o th e r corpora. These discourse u n its have in trin s ic p ro p e rtie s; fo r instance, v e ry o fte n th e y have the same subject, tense, and aspect fo rm s. H o w e ve r, none o f these factors p ro v id e s a necessary c o n d itio n fo r re co g n izin g a discourse u n it, even th o u g h they are h ig h ly co rre la te d : e.g. a change o f the subject, tense o r aspect form s, in general co-occurs w ith a s h ift o f discourse u n it. 6.4. T he in te ra c tio n o f factors a n d th e d e fin itio n o f p ro m in e n ce In chapter 5 , 1 proposed a d e fin itio n o f prom inence based on the in te ra c tio n o f fo u r d iffe re n t factors: distance, frequency o f m e n tio n , syn ta ctic p o s itio n , and them atic ro le . M y approach goes be yo n d p re vio u s approaches b y addressing a qu estion th a t m ost o f th e research on anaphora ignores: h o w do these factors in te ra ct w hen se le ctin g and in te rp re tin g a d e fin ite NP? I conducted three q u a n tita tiv e analyses (a p a ir-w is e com parison o f the d iffe re n t factors, a d e scrip tio n o f h o w the fo u r factors fu n c tio n sim u lta n e o u sly, and an a p p lie d m u ltip le regre ssio n). In the firs t q u a n tita tive analysis, I com pared each fa c to r w ith e ve ry o th e r, in o rd e r to assess th e ir re la tiv e strength, esp ecially in cases Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 o f c o n flic t w h e re one fa c to r p re fe rre d one antecedent and another fa cto r p re fe rre d a n o th e r. M y second q u a n tita tiv e analysis goes a step fu rth e r, b y lo o k in g v e ry d e s c rip tiv e ly a t h o w the factors stu d ie d are d is trib u te d sim u lta n e o u sly in m y set o f data. This is possible because the corpus is lim ite d and m anageable. A p ro b le m w ith th is a n a lysis is th a t the results m a y be an a rtifa c t o f m y p a rtic u la r data. The th ird q u a n tita tiv e analysis is a so p h istica te d s ta tis tic a l m e thod (an a p p lie d lo g is tic regression) th a t a llo w s fo r a ll the possible co m b in a tio n s and c o m p u ta tio n s w h ic h are n o t re a d ily a va ila b le w hen ra w num bers are m a n ip u la te d . T his analysis s u p p o rts m y q u a lita tiv e analysis and the tw o p rio r q u a n tita tiv e analyses, a n d it gives the analysis greate r re lia b ility . The co n clu sio n th a t em erges fro m these analyses is th a t distance is the m o st im p o rta n t fa c to r to m ake an e n tity p ro m in e n t, fo llo w e d by fre que ncy o f m e n tio n . S yntactic p o s itio n and th e m a tic ro le do n o t p la y a ro le in the p ro m in e n ce o f e n titie s in discourse. These facts w ere sum m arize d in T able 17 (cf. 5.3.4) w h ich is repeated here: Table 17: The Ranking of Factors for Prominence Syntactic position D istance > Frequency > Thematic role 6.5. M otivating the prominence hierarchy In sections 5 .4 .3 ,1 e xp la in e d w h y syn ta ctic p o s itio n and them atic ro le do n o t p la y a ro le in reference in te rp re ta tio n a n d w h y the distance fa cto r is m o re im p o rta n t th a n th e frequency fa cto r. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 S ynta ctic an d them atic in fo rm a tio n do n o t p ro v id e co n siste n t id e n tific a tio n o f an e n tity . W e can have m a n y d iffe re n t g ra m m a tica l and th e m a tic ro le s fo r the same e n tity because th e y are a fu n c tio n o f in d iv id u a l sentences. The fa c t th a t one e n tity appears as subject does n o t m ean th a t it w ill be re trie v e d la te r in discourse in the sam e synta ctic p o s itio n . G ra m m a tica l p o s itio n an d the m atic role are im p o rta n t fo r sentence in te rn a l re p re se n ta tio n (see fo r instance R e inh art 1983) b u t the y are n o t s ig n ific a n t fo r a discourse re pre senta tion. S yntactic and the m atic p o s itio n are n o t represented in the m e n ta l re p re se n ta tio n o f the te x t because th e ir in fo rm a tio n is te m p o ra ry, i.e. th is in fo rm a tio n is n o t sto re d in a n y k in d o f p e rm a n e n t m e m ory storage. I t is clear th a t distance and frequency p la y a m ore im p o rta n t ro le th a n g ra m m a tica l and th e m a tic ro le fo r antecedent selection. A n d , ju s t because I have coded the gram m atical and the m atic ro le o f the m ost recent m e n tio n o f the antecedent, does n o t m ean th a t th a t sp e cific ro le has any p rio rity , pre cise ly because th e m a tic and g ra m m a tica l ro le s do n o t p ro v id e con sistent id e n tific a tio n . There are d iffe re n t possible exp la n a tio n s fo r the fact th a t distance is m ore im p o rta n t th a n frequency. F irs t, the re is a general preference in language to m in im iz e the re le va n t do m ain s fo r any lin g u is tic o p e ra tio n (H a w k in s , 1998). T herefore, th e in te rp re ta tio n o f an an aph or sh o u ld be re so lve d w ith in the shortest distance possible. The p rin c ip le fin d the closest antecedent, regardless o f its frequency of mention in vo lve s a m o re econom ic processing o p e ra tio n . B y contrast, m a k in g a frequency assessm ent (by co m p a rin g th e n u m b e r o f m entions o f d iffe re n t e n titie s ) is a m uch m ore com plex processing op era tion. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 T his is so, unless w e assume th a t the fre q u e n cy va lu e o f an e n tity is a u to m a tica lly ca lcu la te d as soon as th e anaphor is processed, and the refore th a t it is p a rt o f the a ctiva tio n o f e n titie s. In o rd e r to e xp la in the ra n k in g in ta b le 17 u n d e r th is a ssu m p tio n (cf. C ow an, 1993 and M e D on ald e t a l in press), w e w o u ld need to assume th a t the recency o f an e n tity s till re su lts in greater a c tiv a tio n th a n m ore fre q u e n t m entions (an a ssu m p tio n th a t re q u ire s fu rth e r research). The a c tiv a tio n o f e n titie s re c e n tly pro d u ce d co u ld s till be m o re accessible th a n access to e n titie s a lre a d y processed, th e re fo re , and distance co u ld again be the m ore im p o rta n t factor. T hat is w h y distance is the m ore im p o rta n t factor. F in a lly , g ive n the discourse re p re se n ta tio n th e o ry presented in sectio n 5.4.2, the m e n ta l re p re se n ta tio n o f a te x t is n o t au tom a tic o r instantaneous. O n the c o n tra ry , th e m ental re p re se n ta tio n is b u ilt w h ile speakers and listen ers are ta lk in g . The se q u e n tia l n a tu re o f speech is d iffe re n t fro m the h ie ra rch ica l s tru c tu re o f th e m e ntal representation. T herefore, there sh o u ld be a tra n s itio n fro m the se q u e n tia l s trin g o f the speech to the h ie ra rch ica l re pre senta tion in w h ic h the en tities are n o t y e t assigned to a P-set o r to a le ve l o f fre q u e n cy o f m e ntion . T h is is m ore im p o rta n t fo r Pers P ronoun, C litic , and D e f N u ll Subj because th e p h o n o lo g ica l m a te ria l does n o t s u p p ly m u ch in fo rm a tio n , and th e ir in te rp re ta tio n re lie s m ore h e a v ily on the in fo rm a tio n p ro v id e d b y th e antecedent. T h ere fore , w h e n lo o k in g fo r an antecedent, the firs t place to search is the se q u e n tia lly organized, v e ry te m p o ra ry com pon ent in w h ic h e n titie s are s till in the process o f b e in g assigned a le ve l o f a c tiv a tio n and a co rre sp o n d in g P-set. I f th is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 te m p o ra ry com ponent is seq uen tial, th e closer e n tity w o u ld be m ore accessible tha n the one n o t pre se n t o r fu rth e r aw a y in th is te m p o ra ry co m p o n e n t. T w o questions are raised a t th is p o in t: w h y is it the case th a t n o t a ll antecedents are located in a closer p o s itio n th a n th e ir com petitors? A n d , if distance is so im p o rta n t, w h y does frequency p la y an y ro le at a ll? For the fe w cases in w h ic h the closer antecedent w as n o t selected, som e other factors (frequency o f m e n tio n , syn ta ctic p o s itio n and th e m a tic role) w ere fa v o rin g it. T h ere fore , distance is v e ry im p o rta n t b u t i t is n o t the o n ly fa cto r considered in antecedent selection. Second, the sta tistica l c a lc u la tio n show ed th a t fre q u e n cy pla ys a ro le , even w he n distance is n o t p re se n t in the m odel. Language is re d u n d a n t and the closer antecedent is, in general, the m ore fre q u e n t one. The sta tistica l analysis a llo w e d us to evaluate the d iffe re n t factors w ith o u t co n sid e rin g the te m p o ra ry com p o n e n t o f the m o d e l (the b u ffe r). The m o del behaves as i f the e n titie s w ere assigned to a le ve l o f a c tiv a tio n and P-set in sta nta neo usly. T h ere fore , the a c tiv a tio n o f e n titie s (w hatever the source o f a c tiv a tio n is, closeness o r frequency) is im p o rta n t fo r antecedent selection, a n d there is n o th in g in h e re n t in th e a ctiva tio n re s u ltin g fro m distance o r frequency. 6.6. F u ture d ire ctio n s In chapter 4 ,1 discussed th e necessary and s u ffic ie n t co n d itio n s fo r d is tin g u is h in g one discourse u n it fro m another. T h is is o n ly a b e g in n in g . F u rth e r research needs to be done on th is to p ic, in c lu d in g Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 m ore da ta analysis. F o r instance, it is possible th a t w e w ill fin d m ore lin g u is tic in d ic a to rs th a t d e lim it discourse u n its . In chapter 4 , 1 also presented a h ie ra rch y o f P-sets arranged according to degree o f a cce ssib ility. As sho w n in chapter 2 (section 2.5.5.I.) n o t e v e ry b o d y agrees o n th e arrangem ent o f th is h ie ra rch y; fu rth e r research is needed in o rd e r to p ro v id e the e xp e rim e n ta l data th a t gives evidence fo r the re la tiv e acce ssibility o f each context. T h is s tu d y has im p lic a tio n s fo r theories o f the m e ntal re pre senta tion o f a te xt, since synta ctic and th e m a tic roles w ere fo u n d in chapter 5 to be less im p o rta n t tha n distance and fre que ncy o f m e n tio n . A n e w lin g u is tic pe rspe ctive em erges fro m th is in te g ra tio n o f p ra g m a tic approaches w ith e m p iric a l discourse studies and p s y c h o lin g u is tic /e x p e rim e n ta l approaches. T h is n e w fra m e w o rk p ro vid e s a d e ta ile d m e n ta l re pre senta tion o f texts, w h ic h can be used fo r re la te d stud ies, such as n a tu ra l language processing. The lite ra tu re su rve ye d in th is d isse rta tio n has show n th a t m e ntal re pre senta tions o f discourse v a ry acco rd in g to d iffe re n t m odels. Those o rig in a tin g in p h s y c h o lin g u is tic s are based o n e xp e rim e n ta l data, b u t th e y do n o t te ll us m u ch about spe cific lin g u is tic facts. For instance, G em sbacher's (1990) m o d e l is n o t clear a b o u t the exact n a tu re o f syn ta ctic p o s itio n and th e m a tic ro le in discourse re pre senta tion (cf. chapter 5). D iscourse theories (A rie l, 1990, C hafe, 1994, G undel e t al., 1993) have based th e ir research o n n a tu ra l data analysis, b u t th e y do n o t go the e xtra step o f p ro p o s in g h o w these data are represented in the m in d o f the speaker. F or instance, Chafe's (1994) consciousness th e o ry is ra th e r vague, an d is n o t based o n a re lia b le m e th o d o lo g y. A rie l Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 (1990) m e n tio n s th a t the P a ra lle l D is trib u te d Processing m o d e l presents a p p ro p ria te d e scrip tio n s o f m e n ta l repre senta tions o f language b u t she does n o t c la rify h o w exa ctly it does so. C hafe (1994) and G u n d e l e t al. (1993) m e n tio n a co m p a rtm e n ta l th e o ry o f m em ory, b u t th e y d o n o t p ro v id e any evidence s u p p o rtin g th is statem ent. A m ore com prehensive acco unt o f the w a y language is represented d u rin g p ro d u c tio n is needed, and th is d is s e rta tio n has p ro v id e d o n ly th e firs t step in a ch ie vin g a m o re in te g ra te d an d in te rd is c ip lin a ry m o d e l o f discourse re p re se n ta tio n , as it re la te d to d e fin ite expressions in S pan ish. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 A p p e n d ix A T h e re m a in in g factors (C h a p te r 5) Table 18: The interaction of Distance and Syntactic Position D istance S vnta ctic P o s itio n C om p. C loser A n t. C lo se r T o ta l Ant. more accessible 31 34 Comp, more accessible 2 £H S (82gM $(S 3 16 5 45 50 Table 19: The interaction of Distance and Syntactic Position and their relative strength D istance — > S vntactic P o s itio n 1 Comp, clsr by 6 or more IUs degree 2 Comp, clsr by 1-5 IUs degree 1 Ant. clser by 1-5 IUs degree 1 Ant clser by 6 or more IUs degree 2 Total Ant. more accss. degree 2 1 1 6 5 13 (87%) Ant. more accss. degree 1 1 0 8 12 21 (60%) Cmpt more accss. degree 1 0 1 6 7 14 (40%) Cmpt more accss. degree 2 1 0 1 0 2 (13%) T o ta l 3 (11%) 2 (9%) 21 (91%) 24.(89%) 50 Table 20: The interaction of Distance and Thematic Role D istance T h e m a tic R ole C om p. C loser A n t. C lo se r T o ta l Ant. more accessible 26 29 Comp, more accessible l 16 4 41 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 Table 21: The interaction of Distance and Thematic Role and their relative strength D ista n ce — > T h e m a tic R ole 1 Comp, clsr b v 6 o r more IUs degree 2 Comp, clsr by 1-5 IUs degree 1 Ant. clser by 1-5 IUs degree 1 Ant clser by 6 or more IUs degree 2 T otal A n t more accss. degree 2 1 1 2 1 5 (100%) A n t more accss. degree 1 1 0 11 12 24 (60%) Comp more accss. degree 1 0 1 8 7 16 (40%) Comp more accss. degree 2 0 0 0 0 0 (0%) Total 2 (9%) 2 (9%) 21 (91%) 20 (91%) 45 Table 22: The interaction of Frequency of M entions and Syntactic Position F requencv o f M e n tio n s S vntactic P o s itio n C om p , m ore fre q u e n t A n t. m o re fre q u e n t T o ta l A n t more accessible 28 36 Comp, more accessible 5 3 K g ( 6 0 % )|lilli 17 13 40 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 23: The interaction of Frequency of Mentions by Syntactic Position and their relative strength Freauenci/ o f M e n tio n s — > S ynta ctic P o s itio n Comp more frqnt by 6 or more mnts degree 2 Comp more frqnt by 1-5 mnts degree 1 Ant more frqnt by 1- 5 mnts degree 1 Ant. more frqnt by 6 or more mntns degree 2 Total A n t more accss. degree 2 1 4 7 2 14 (87%) A n t more accss. degree 1 0 3 12 7 22.(59%) Comp more accss. degree 1 0 4 7 4 15 (41%) Comp more accss. degree 2 1 0 1 0 2 (13%) T otal 2 (13%) 11 (29%) 27 (71%) 13 (87%) 53 Table 24: The interaction o f Frequency of M entions and Thematic Role F requ encv o f M e n tio n s T h e m a tic R ole C om p, m o re fre q u e n t A n t. m o re fre q u e n t T o ta l A n t more accessible mmmmm 26 32 Comp, more accessible 3 15 9 38 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 Table 25: The interaction of Frequency of Mentions by Thematic Role and their relative strength Frequence/ o f M e n tio n s — > T h e m a tic R o le Comp more frqnt by 6 or more mnts degree 2 Comp more frqnt by 1-5 mnts degree 1 Ant more frqnt by 1- 5 mnts degree 1 Ant. more frqnt by 6 or more mntns degree 2 T otal A n t more accss. degree 2 1 1 4 0 6 (100%) Ant. more accss. degree 1 0 4 14 8 26 (63%) Comp more accss. degree 1 1 2 11 1 15 (37%) Comp more accss. degree 2 0 0 0 0 0 (0%) T otal 2 (18%) 7 (20%) 29 (80%) 9 82% 47 Table 26: The interaction of Syntactic Position and Thematic Role S yntactic P o s itio n T h e m a tic R ole C om p, m ore accessible A n t. m ore accessible T o ta l A n t more accessible 25 27 Comp, more accessible 6 8 8 35 Table 27: Interaction Syntactic position by Thematic Role and their relative strength S yntactic P o s itio n — > Th«?nMtic.RQte degree 2 degree 1 degree 1 degree 2 T o ta l A n t more accss. degree 2 1 0 0 5 6 (100%) A n t more accss. degree 1 0 1 16 4 21 (72%) Comp more accss. degree 1 0 6 1 1 8 (28%) Compmore accss. degree 2 0 0 0 0 0 (0%) T otal 1 (10%) 7 (30%) i z (70%) 10 (90%) 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 R eferences A lm o r, A m it. 1996. N P A n a p h o ra and Focus - the In fo rm a tio n a l Lo ad H ypothesis. Ph. D . D isse rta tio n , B ro w n U n iv e rs ity , P rovidence. R I. _________ . to appear. C o n stra in ts an d M echanism s in T heories o f A n a p h o r Processing. A rc h ite c tu re s and M echanism s in Sentence C om prehension, ed. b y. Ed. P ickering , M . C lifto n , and M . C rocker. C am b ridg e U n iv e rs ity Press: C am bridge, E ngland A nde rsen , Elaine, M aquela B rizu e la , Beatrice D u p u y , and Laura G onnerm an. 1995. The a c q u is itio n o f D iscourse M arkers as S o cio lin g u istic V a ria b le s: A C ro s s -lin g u is tic C om parison. Proceedings o f the T w e n ty-se ve n th A n n u a l C h ild Language Research Forum ., ed. b y Eve C la rk, 61-70. S tanford: C enter fo r the S tudy o f Language and In fo rm a tio n . A n d e rso n , A , SC G arrod and AJ S anford. 1983. The A cce ssib ility o f P ro n o m in a l A ntecedents as a F u n ctio n o f E pisode S hifts in N a rra tiv e Text. Q u a te rly Jo u rn a l o f E xperim en tal P sychology . 35A. 427-440. A rie l, M ira . 1988. R e fe rrin g an d A cce ssib ility. Journal o f L in g u is tic s .24. 67-87. ____________. 1990. A ccessing N ou n-P hrase A ntecedents. N e w Y o rk: R outledge. ____________. 1996. R e fe rrin g E xpressions and the + /- C oreference D is tin c tio n . Reference and R eferent A cce ssib ility, ed. b y Fretheim , T h o rste in and Jeanette K . G undel, 13-35. A m sterdam : John B enjam ins P u b lish in g C om pany. B e ll, T o ny. 1982. The sem antic sig n ifica n ce o f the in d e fin ite a rticle in Spanish. H isp a n ia . 619-624. B e n tiv o g lio , Paola. 1983. T o p ic c o n tin u ity and d is c o n tin u ity in discourse: a s tu d y o f spoken L a tin -A m e rica n Spanish. T o p ic c o n tin u ity in discourse: a q u a n tita tiv e cross-language stu d y, ed. b y T a lm y G ivo n , 255-311. P h ila d e lp h ia : John B enjam ins. Bergen, John J. 1977. The Sem antics o f Spanish C o u n t and M easure E n tity N ouns. Language Sciences 44.1-9. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 B im e r, B etty & G re g o ry W a rd . 1994. U niqueness, F a m ilia rity , and the D e fin ite A rtic le in E n g lish . P roceeding o f th e T w e n tie th A n n u a l M e e tin g o f the B erkeley L in g u is tic s S ociety, ed. b y Susanne G ahl, A n d y D o lb e y and C h ris to p h e r Johnson, 93- 102. Berkeley: B erkeley L in g u is tic S ociety, In c . B la ckw e ll, Sarah E. 1998. C o n stra in ts on S panish N P A n a p h o ra : the S yntactic versus the P rag m atic D o m a in . H is p a n ia 81,3. 606- 618. B rizu e la , M aquela. 1992. M arcadores d iscu rsivo s e n n a rra tiv a s de ninos b ilin g iie s . B ilin g iiis m o y a d q u is ic io n d e l espanol. Estudios en Espana y Estados U n id o s., ed. b y H e m a n U rru tia y C arm en S ilva -C o rva la n , 333-52. B ilbao: H o riz o n te . ______________1997. La E s tru c tu ra D is c u rs iv a y la S eleccion de D em onstratives en E spanol: Usos A n a fo rico s. A ctas d e l I C oloq uio L a tin o a m e rica n o de A n a lista s d e l D iscu rso , ed. b y B o liva r, A d ria n a y P aola B e n tiv o g lio , 71-81. C aracas:U niversidad C e n tra l de V enezuela. ____________. In press. T he se le ctio n o f d e fin ite exp re ssio n s in Spanish. P roceedings o f the In te rn a tio n a l C o g n itiv e L in g u istics C onference. A m sterdam . 1997. B rizu e la , M aquela, E laine A n d e rse n and L yn n e S ta llin g s . 1999. D iscourse M arkers as In d ica to rs o f R egister. H is p a n ia 82.118- 131. B u ll, W illia m . 1965. S panish fo r Teachers: A p p lie d lin g u is tic s . N e w Y ork: R onald Press. C arpenter, P. A . and M .A . Just. 1977. In te g ra tiv e Processes in com prehension. Basic Processes in R eading, P e rce p tio n and C om prehension, ed. b y D . La Berge and J. Sam uels, 217-241. H ills d a le , N.J: E rlb a u m . C hafe, W allace 1976. G ivenness, C ontrastiveness, D e fin ite n e ss, Subjects, Topics and P o in t o f V ie w . Subject an d T o p ic., ed. b y Charles N . L i, 25-55. N e w Y ork: A cadem ic Press. ____________. 1994. D isco u rse , C onsciousness a n d T im e . C hicago: The U n iv e rs ity o f C hicago Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 C iapuscio, G u im a r. 1988. La d e ixis y e l fu n cio n a m ie n to te x tu a l de los p ro n o m b re s. R evista argen tina de lin g u is tic a 4. 25-66. C lancy, Pat. 1980. R e fe re n tia l C hoice in E n g lis h and Japanese N a rra tiv e D iscourse. The Pear Stories, v o l. 3, ed. b y Chafe, W ., 127-202N orw ood; A b le x. C la rk , H .H . & C.J S engul. 1979. In Search o f R eferents fo r N ouns and P ronouns. M e m o ry and C o g n itio n 7. 35-41. C hristo ph erse n, P. 1939. The A rticle s: a S tu d y o f th e ir T h eory and Use in E n g lish . M u n ksg a a rd , C openhagen. C o rn ish , Francis. 1986. A n a p h o ric R elations in E n g lish and French. A D iscourse P erspective. N e w H a m p sh ire : C room H e lm . C ow an , N elson. 1993. A c tiv a tio n , a tte n tio n , a n d s h o rt-te rm m em ory. M e m o ry & C o g n itio n 21 (2). 162-167. D eclerck, Renaat. 1986. The m a n ifo ld in te rp re ta tio n s o f generic sentences. Lingua 68. 149-188. va n D ijk , Teun. 1982. Episodes as u n its o f discourse analysis. A n a ly s in g D iscourse: T ext and T a lk, ed. b y D eborah Tannen, 177-195. W a sh in g to n D C : G eorgetow n U n iv e rs ity Press. va n D ijk , T. and K in tsch . 1983. Strategies o f discourse com pre hen sion. N e w Y o rk: A cadem ic Press. D u B ois, John; Susanna C u m m in , Stephan S chuetze-C obum , Danae P aolino (eds.). 1992. D iscourse T ra n s c rip tio n . Santa Barbara Papers in L in g u is tic s V o lu m e 4. E h rlic h , K . & K . R ayner. 1983. P ronouns A ssig n m e n t and Sem antic In te g ra tio n d u rin g R eading: E ye-m ovem ents and Im m edia cy o f P rocessing. Jo u rn a l o f V e rb a l L e a rn in g and V erb al B ehavior. 22. 75-87. E l habla c u lta de Buenos A ires. M a teria le s pa ra su estudio. Tom o 1 y 2. 1987, U n iv e rs id a d N a cio n a l de Buenos A ire s. F acultad de F ilo so fia y Letras. Buenos A ire s. E pstein, R ichard. 1994. D iscourse and D efin itene ss: S ynchronic and D ia ch ro n ic P erspectives. Ph. D . D isse rta tio n . U n iv e rs ity o f C a lifo rn ia , San D iego. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 Fernandez R am irez, S alvador. 1987. G ra m a tica E spanola. Segunda e d icio n . 3.2. E l pronom bre. M a d rid : A rc o . Fox, B arbara. 1987. D iscourse S tructure and A n a p h o ra in W ritte n a n d C o n ve rsa tio n a l E nglish. C a m b rid g e : C am b ridg e U n iv e rs ity Press. F rederiksen, J. R. 1981. U n d e rsta n d in g anaphora: Rules used b y readers in a ssig n in g p ro n o m in a l re fe re n ts. (R ep ort N o 3- 4462) C a m b rid g e , M A : B o lt B eranek a n d N e w m a n Inc. G arcia, E rica. 1975. The R ole o f T h eory in L in g u is tic A n a lysis: The S panish P ro n o u n System . A m ste rd a m : N o rth H o lla n d . G arcia F ajardo, Josefina. 1989. E l se n tid o en la fu n c io n referencial de frases su sta ntiva s d e l espanol. N u e va R evista de F ilo lo g ia H isp a n ica 1.19-26. _________________ . 1991. E l s ig n ific a d o de lo s d e te rm in a n te s espanoles. N u e va R evista de F ilo lo g ia H isp a n ica 2. 737-752. G arrod , S im on. 1994. R esolving P ronouns a n d O th e r A n a p h o ric D evices: The Case fo r D iv e rs ity in D iscourse Processing. P erspectives o n Sentence Processing, ed. b y C lifto n , C harles, Jr., L y n n F ra zie r & K e ith R ayner. H ills d a le : Law rence E rlb a u m Associates. G arrod , S im on and A.J. S anford. 1990. R e fe re n tia l P rocessing in R eading: F o cusing on roles and in d iv id u a ls . C om prehension Processes in R eading, ed. b y D .A . B a llo ta , G.B. Flores d 'A rca is, and K . R ayner, 465-486. H ills d a le , N .J.: E rlb a u m G arrod , S im on, D a n ie l F reudenthal and E liz a b e th B oyle. 1994. The R ole o f D iffe re n t Types o f A n a p h o r in th e O n -lin e R e so lu tio n o f Sentences in D iscourse. Jo u rn a l o f M e m o ry and Language 33, 39-68. G em sbacher, M o rto n A n n . 1990. Language C om p reh en sio n as S tru ctu re B u ild in g . H ills d a le , N e w Jersey: Law rence E rlba um ____________________ . 1997. C oherence C ues M a p p in g D u rin g C o m p re h e n sio n . Processing In te rc la u s a l R e la tio n sh ip s: S tudies in th e p ro d u c tio n and C om p reh en sio n o f T ext., ed. b y J C osterm ans and M . Fayol. M a h w a h , NJ: LE A . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 G em sbacher, M o rto n A n n and H argreaves D . 1988. A ccessing sentence p a rtic ip a n ts : The advantage o f firs t m e n tio n . Journal o f M e m o ry and Language 27. 699-717. G em sbacher, M o rto n A n n and Shroyer, S. 1989. The cata ph oric use o f the in d e fin ite th is in spoken n a rra tive s. M e m o ry and C o g n itio n 17. 536-540. G em sbacher, M o rto n A n n & T a lm y G ivo n . 1995. C oherence in S pontaneous T ext. P h ila d e lp h ia : John B enja m ins. G ivo n , T a lm y. 1983a. T o p ic C o n tin u ity in D iscourse: a Q u a n tita tiv e C ross-language S tudy. P h ila d e lp h ia : John B enja m ins PC. 5-41. __________ . 1984a. D ire c t O bject and D a tiv e S h iftin g : Sem antic and P ragm atic Case. Objects, ed. b y F. P lank. M e w Y o rk: A cadem ic Press. ___________. 1992. The G ra m m a r o f R e fe re n tia l C oherence as M e n ta l Processing In stru ctio n s. L in g u is tic s 30. 5-55. __________ . 1995. C oherence in Text vs. C oherence in M in d . Coherence in Spontaneous Text., ed. b y G em sbacher, M o rto n and T G ivo n , 59-115. A m sterdam : John B enjam ins P u b lis h in g C om pany. G ordon, P.C., B.J. G rosz and L. G illio m . 1993. P ron oun s, Nam es, and the C e n te rin g o f A tte n tio n in D iscourse. C o g n itiv e Science, 17 (3), 311-348. G un de l, Jeanette, N ancy H edberg & Ron Zacharsky. 1993. C og nitive Status an d the F o rm o f R eferring E xpressions in D iscourse. Language 69.274-307. G rim es, J.E. 1975. The Thread o f Discourse. The H ag ue: M o u to n . G rosz. B. J. 1977. The R epresentation and Use o f Focus in D ia lo g U n d e rsta n d in g . Ph. D . dissertation. U n iv e rs ity o f C a lifo rn ia at B erkeley. _______ . 1981. F ocusing and D e scrip tio n in N a tu ra l Language D ialogues. Elem ents o f Discourse U n d e rsta n d in g , ed. b y A .K . Joshi, B.L. W ebber and I.A . Sag. C am bridge: C a m b rid g e U n iv e rs ity Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 G ro sz, B arbara and Candace S id n e r. 1986. A tte n tio n , in te n tio n s , and the s tru c tu re o f discourse. C o m p u ta tio n a l L in g u is tic s . 44-50. G rosz, B arbara, A ra v in d Joshi a n d S cott W einstein. 1995. C e n te rin g : A F ra m e w o rk fo r M o d e llin g th e Local Coherence o f D iscourse. C o m p u ta tio n a l L in g u is tic s 21-2. H a n ks, W illia m F. 1992. The In d e x ic a l G ro u n d o f D e ictic Reference. R e th in k in g C ontext: Language as an In te ra ctive P henom enon, ed. b y A . D u ra n ti and C. G o o d w in g . C am b ridg e: C am bridge U n iv e rs ity Press. H a w k in s , John A . 1978. D e fin ite n e ss an d Indefin itene ss. A S tu d y in Reference an d G ra m m a tic a lity P re d ictio n . L o n d o n : C ro o m H e lm . _______________ .1991. O n (In )d e fin ite A rtic le s : Im p lic a tu re s a n d (U n )g ra m m a tic a lity P re d ic tio n . Jou rna l o f L in g u is tic s 27. 405- 42. _______________ . 1994. A P e rfo rm a n ce T h e o ry o f O rd e r a n d C o n stitu e n cy. C am bridge: C a m b rid g e U n iv e rs ity Press. _______________ . 1998. Some Issues in a P erform ance T h e o ry o f W o rd O rd e r. C o n stitu e n t O rd e r in the Languages o f E uro pe, ed. b y A . S iew ierska, 729-780. B e rlin : de G ru yte r. 1999. P rocessing C o m p le x ity and F ille r-g a p dependencies across G ra m m a r. Language 75.2. to appear. T he R ole o f P rocessing P rin c ip le s in E x p la in in g Language U n ive rsa ls. Language T y p o lo g y and Language U niversals: A n In te rn a tio n a l H an dbo ok, ed. b y M . H a sp e lm a th , E. K oenig, W . O esterreicher and W . R aible. B e rlin : de G ru yte r. H e im , Irene. 1982. The Sem antics o f D e fin ite and In d e fin ite N P s. P hD d isse rta tio n . U n iv e rs ity o f M assachusetts, A m h e rst. H in d s , J. 1979. O rg a n iza tio n a l P atte rns in D iscourse. S yntax and Sem antics V o l 12: D iscourse an d Syntax, ed. b y G iv o n , T a lm y, 135-158. N e w Y ork: A cade m ic Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 H in tik k a , Jaakko & Jack K ulas. 1985. A n a p h o ra and D e fin ite D e scrip tio n s. T w o A p p lic a tio n s o f G am e-Theoretical Sem antics. D ord re ch t: D . R eide l PC. Joh n so n -L a ird , P .N . 1983. M e n ta l M o dels. C am bridge, M A : H a rv a rd U n iv e rs ity Press. K a d m o n , N . 1990. U niqueness. L in g u is tic s and P hilosophy 13. 273- 324. K a m p , H . 1981. A T h eory o f T ru th and Sem antic R epresentation. F o rm a l M ethods in the S tu d y o f Language: Proceedings o f the T h ird A m sterdam C o llo q u iu m , P a rt I, ed. b y J. G ro e n e n d ijk, M . S to kh o f and F. V e ltm a n . A m ste rd a m : M a th e m a tica l C entre. K a rm ilo ff S m ith, A . 1980. O sych o lo g ica l Processes U n d e rly in g P ro n o m in a liz a tio n a n d N o n p ro n o m in a liz a tio n . in C h ild re n 's C onnected D iscourse. Papers fro m the Parasession o n P ronouns and A n a p h o rs, ed. b y J. K re im a n and A . E. O jeda, 231-250. C hicago: C hicago L in g u is tic s Sodety. K le in -(A n d re u ), F lora. 1976. 'Sam e' vs. D iffe re n t" C ro s s lin g u is tic a lly : the a rtic le in E n g lish and Spanish. Papers fro m the T w e lfth R eg io na l M e e tin g o f the Chicago L in g u is tic Society, ed. b y S alikoko S. M u fw e n e e t al. 413-424. C hicago: U n iv e rs ity o f C hicago Press. K ib rik , A n d re j. 1996. A n a p h o ra in R ussian N a rra tiv e Prose: A C o g n itiv e C a lcu la tive A cco u n t. Studies in A naphora, ed. b y Fox, Barbara, 255-303. A m ste rd a m : John Benjam ins. K in g , L a rry D . 1992. The S em antic S tru ctu re o f Spanish. M e a n in g and G ra m m a tica l F o rm . P h ila d e lp h ia : John Benjam ins. K irs n e r, R obert. 1979. D e ixis in D iscourse: A n E xp lo ra to ry Q u a n tita tiv e S tud y o f the M o d e m D u tc h D em o nstrative A d je ctive s. In S yntax and Sem antics: D iscourse and S yntax. G iv o n , T a lm y (Ed.) V o l 12. N e w Y o rk: Academ ic Press. 355- 375. L a b o v, W illia m . 1972. Language in the in n e r c ity : Studies in the B lack E n g lish ve rn a cu la r. P h ila d e lp h ia : U n iv e rs ity o f P e nn sylvania Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 Lapesa, R. 1961. D e l d e m ostrativo a l a rtfc u lo . N u e va R evista de F ilo lo g fa H isp a n ica XV. 23-44. L a u ry, R itva . 1997. D em onstratives in In te ra c tio n . The em ergence o f a d e fin ite a rtic le in F in n is h . P h ila d e lp h ia : John B enjam ins. Leonetti, M a n u e l. 1993. Im p lic a tu ra s G eneralizadas y R elevancia. R evista E spahola de L in g iifs tic a , 23,107-139. ____________ . 1993. In press. E l a rtfc u lo d e fin id o y la co n stru ccio n d e l contexto. Levinson, Stephen. 1987a. P ragm atics and the gram m a r o f anaphora: a p a rtia l p ra g m a tic re d u c tio n o f b in d in g and c o n tro l phenom ena. Jo u rn a l o f L in g u is tic s 23. 379-434. _______________ .1989. A R e vie w o f R elevance. J o u rn a l o f L in g u is tic s 25. 455-472. Lyons, John. 1977. Sem antics. V o lu m e II. C am bridge: C am bridge U n iv e rs ity Press. M acD onald, M a ry e lle n and M o rte n H . C h ristia n se n . In press. In d iv id u a l D ifferences w ith o u t W o rk in g M e m o ry: a R e p ly to Just & C a rp e n te r and W aters & C aplan. Maes, A lfo n s & Leo N oo rdm a n. D e m o n stra tive N o m in a l A n a p h o rs: a Case o f N o n id e n tific a tio n a l M arkedness. L in g u is tic s 33. 255-282. M ann, W illia m C. and Sandra A . T hom pson. 1988. R h e to rica l S tru ctu re T h eory: T o w ard a F u n ctio n a l T h e o ry o f T ext O rg a n iza tio n . Text 8 (3). 243-281. M a rsle n -W ilso n , W ., E. Le vy and L .K . T y le r. '982. P ro d u cin g In te rp re ta b le D iscourse: the E stab lish m en t and M aintanece o f Reference. Speech, Place and A c tio n , ed. b y R.J. Jarve;;a and W . K le in , 339-378. C hichester: W ile y . M cC lellan d and R um elhart. 1986a (eds.) P a ra lle l D is trib u te d Processing. C am bridge: M IT Press. M cK oon, G ., R. R a tc liff and G. W a rd . 1993. S yntactic P rom inence E ffects o n D iscourse Processes. Jo u rn a l o f M e m o ry and Language. 32,593-607. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 M in s k y , M a rv in . 1975. A F ra m e w o rk fo r R epresenting K n o w le d g e . The Psychology o f C o m p u te r V isio n , ed. b y P. W in sto n , 211- 277. N ew Y o rk: M c G ra w -H ill. N o ru sis, M a rija J. 1994. SPSS A d va n ce d S tatistics 6.1. C hicago: SPSS Inc. 1-30. Paredes S ilva, Vera. 1993. S ubject O m ission and F u n c tio n a l C om pensation: E vidence fro m W ritte n B ra z ilia n P ortuguese. Language V a ria tio n and C hange V olum e 5, N u m b e r 1. 35-49. Pease-G orrisen, M a rg a rita . 1980. The Use o f the A rtic le in Spanish H a b itu a l and G eneric Sentences. L in g u a 51. 311-316. P iw e k, P .L.A . and A .H .M . C rem ers. 1996. D u tch an d E n g lish D em onstratives: a C o m p a riso n . Language Sciences. 18,3-4. 835-851. P o la n yi, L iv ia and R.J.H. Scha. 1983. The syntax o f discourse. T e xt 3 (3). 261-270. P o la n yi, L iv ia . To appear. The L in g u is tic S tructure o f D iscourse. P rim us, Beatrice. 1993. S yn ta ctic R elations. S yntax: an In te rn a tio n a l H andbook o f C o n te m p o ra ry Research , ed. b y J. Jacobs, A . vo n Stechow, W . S tem efeld & T. Vennem ann, 686-705. B e rlin : de G ru yte r. ______________. 1995. Cases a n d T hem atic R oles. U n iv e rs ity o f M u n ich . C h. 2-3. 34-80. R eichm an, Rachel. 1978. C o n ve rsa tio n a l coherency. C o g n itiv e Science 2. 283-327. R e in h a rt, Tanya. 1983. A n a p h o ra and Sem antic In te rp re ta tio n . London: C room H e lm . R ussell. J.R. 1905. O n D e n o tin g . M in d 14. 479-93. S anford A . J. & G arrod. S. 1981. U n d e rsta n d in g w ritte n language: E xp lo ra tio n s in com pre hen sion beyond the sentence. C hichester: W ile y . S c h iffrin , Deborah. 1987. D iscourse M arkers. C am bridge: C am bridge U n iv e rs ity Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 S idner, Candace. 1983. Focusing a n d D iscourse. D iscourse Processes 6.107-130. Sperber D an & D e ird re W ils o n . 1986. Relevance. C o m m u n ica tio n and C o g n itio n . O xfo rd : B a s il B la ckw e ll. (A n d Second E d itio n , 1995). Sole, Y olanda & C arlos A Sole. 1977. M o d e m S panish Syntax: A s tu d y in C ontrast. L e x in g to n , M assachusetts: D .C . H eath. S tenning, K . 1978. A nap ho ra as an ap pro ach to P ragm atics. L in g u is tic T heory and P sych o lo g ica l R e a lity, ed. b y J. Bresnan, M . H a lle & G. M ille r, 162-200. C am bridge: M IT Press. Strauss, Susan. 1993. W h y 'T h is ' an d 'T h a t' are n o t C om plete w ith o u t 'It'. In Beals K a th a rin e , G in a C ooke, D a vid K ath m an , Sotaro K ita , K a rl-E rik M c C u llo u g h & D a vid Testen. Papers fro m the 2 9 th R egional M e e tin g o f the C hicago L in g u is tic Society. C hicago: The C hicago L in g u is tic Society. Sweet, Stephen A . 1999. D ata A n a ly s is w ith SPSS. M assachusetts: A lly n and Bacon. 107-139. Tao, H o n g y in . 1994. D em o nstratives and the speaker's p o in t o f v ie w in M a n d a rin con versa tiona l discourse. Paper g iv e n a t the 6th N o rth A m e rica n C onference on C hinese L in g u is tic s . USC. M a y, 13-15. T erken and N ooteboom , SG. 1988. O p p o site effects o f accentuation on v e rific a tio n latencies fo r G ive n an d N e w in fo rm a tio n . Language and C o g n itive Processes 2:3 /4 .1 4 5 -6 3 . V o n k, W ietske, L e ttica G .M .M H u s tin x and W im H .G . Sim ons. 1992. The Use o f R e fe re n tia l E xpressions in S tru c tu rin g D iscourse. Language and C o g n itiv e Processes. 7 (3 /4 ). 301-333. W ilso n , D e ird re . 1992. Reference an d R elevance. U C L W o rk in g Papers in L in g u istics. V o l. 4, 167-92. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IMAGE EVALUATION TEST TARGET (Q A -3 ) 1 . 0 l.l 1.25 “ IS yo ISA H 3.2 9 m IB— £ i « 14 I 2-5 2.2 [2.0 1 . 8 150mm APPLIED A IIV U 4 G E . Inc 1 1653 East Main Street Rochester. NY 14609 USA Phone: 716/482-0300 - = r - = Fax: 716/288-5989 0 1993. Applied Image. Inc.. All Rights Reserved V Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Evaluating heaviness: Relative weight in the spoken production of heavy-NP shift

PDF

Syntactic structures in nominals: A comparative study of Spanish and Southern Quechua

PDF

On the nature of particles in Japanese and its theoretical implications

PDF

Preview of the vanishing hero: a study of the protagonists in Jacobean drama

PDF

Correspondence and faithfulness constraints in optimality theory: A study of Korean consonantal phonology

PDF

The syntax of clitic doubling in modern Greek

PDF

Representative twentieth-century short story cycles: studies in a literary genre

PDF

Frequency in sentence comprehension

PDF

Intra-lexical noun-verb dissociations: Evidence from Chinese aphasia

PDF

Morphology and the lexicon: Exploring the semantics-phonology interface

PDF

On the status of possessives

PDF

The erudite romanticism of Juan Eugenio Hartzenbusch in 'Los Amantes de Teruel'

PDF

On the syntactic structure of Spanish noun phrases

PDF

The phonemics and morphology of Hokkaido Ainu

PDF

The syntax of the Chinese BA-constructions and verb compounds: A morphosyntactic analysis

PDF

In the event of focus

PDF

The distribution of lexical and null subjects in the acquisition of L2 Spanish by adult speakers of English

PDF

Identity in crisis: Modernism and the texts of adolescence

PDF

Rhetorical abstraction as a facet of expected response: A structural equation modeling analysis

PDF

The act and the word: A study of abstraction versus the concrete in the work of Albert Camus

Asset Metadata

Creator Brizuela, Maquela (author)

Core Title Definiteness types in Spanish: A study of natural discourse

Degree Doctor of Philosophy

Degree Program General Linguistics

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag language, linguistics,Language, Modern,OAI-PMH Harvest,psychology, cognitive

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-440673

Unique identifier UC11350293

Identifier 9933743.pdf (filename),usctheses-c17-440673 (legacy record id)

Legacy Identifier 9933743.pdf

Dmrecord 440673

Document Type Dissertation

Rights Brizuela, Maquela

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA