Sei sulla pagina 1di 25

a0 buDg6c [D0D-)caD0Dal aD0 D SlIDUl0D m

XSl6DlaS
DavidBeaver,hammFrancez DmiqLevinson
Stan]om |n/vers/q
ntroduction
Consuuctions ofthe typeexempliedby ( I ) havebeenextensive|y studiedin|in-
guisticsmdphi|osophy. Inthegenerative|iteraturetheymegenera||yrefenedtoas
existentials.
( I ) Thereisaprob|emwith thismobilephone.
Oneofthemostextensive|ystudiedaspectsoftheseconsuuctionsisthedistribution
ofH' typesthatoccurinthero|eorpositionofumb/em in( I ). Fo||owing
commonpract|ce,werefertotheNPoccurringinthisrolepositionas/vot. Desp|te
extensive study, thereisnoconsensusasto whatkindsofNPs occur aspivots,or
why. Inthispaper,weargue.
a. Thatthedistributionof typesinpivotro|eismorecomp|exthanhasbeen
acknow|edgedinthel|terature.
b. Thatdenitenesseectsshou|dbeaccountedfornotbyconsuuction-specic
consuaints,butbygenera|markednessconsuaints,inpmicu|mmmkedness
consuaintsongrammatica|sub]ects.
c. Thatdeninesseectsofthekindexhibitedbyexistentia|svary systemat-
ical|y across|anguages, andthatunderstanding menamreofthis vmiation
a||owsustoidenti[anatura|c|assofconsuuctionscross-|inguistica||y.
In section we discuss there|ationbetweenexistentia| andnon-existentia| con-
suuctionsandout|ineaninfoma|theoryofpivotsbasedonmmkednessconsuaints
on sub]ects. Wea|sopresentqua|itauvedatahomEg|ish, Russian mdHebrew
thatdemonstratessystematicvariationinpivotpropertiesacrosslanguages. Insec-
tion we present quantitative datahom a cross-linguistic coqus study, which
demonsuatesthatthedistnbutionofpivotsinvmious|anguagesconespondstothe
expectations mising hom ourtheory. Section contains discussion oftwo phe-
nomenainRussian,thegen|tiveofnegationandtheoptiona||tyofthecopu|a]est '
|npresenttense|ocat|vesandpossessives. Weshowthatneitheroftheminvo|ves
adistribution|ikethatstandmd|yexpectedfordenitenesseects,butthatonour
c|assicationtherstinvo|vesac|emdenitenesseect,whi|ethesecondshowsa
weakereect.
Z3 by L0VlOC8VC, l8m8 8HOCZ 8HO Lmly CVtH8CH
5. LCCg888HO J. MCWC [Cd8), SAL!XV V-+J, th8O8,X: LCHC\H VC8ly.
20
L8VO O8VO, l8m8t t8HCOZ 8HO Lmlty OVH8CH
Z. The Marked Status of ExistentiaI Constructions
u H3uy l3n_u3_cS, Ihc COnSUuCIIOu COIIcSQOu0iu_ IO Ihc Ouc cXcDQIDc0 Dy Ihc
Lu_ISh ScnIcuCc In ( I ) 0cVI3IcS HOH Ihc C3nOuIC3l SIIuCIuIc O 3 0cCl3I3IIVc Scu-
IcuCc In Ihc l3u_u3_c. yQOO_IC3 SIu0IcS [c. _. ccZc I 992, LlMK I 978)h3Vc
QOIuIc0OuI V3IIOuS W3yS Iu WhICh UIS 0cVI3IIOu C3u Dc Ic8IZc0, IuCu0Iu_.
Word order LXiSIcuII3l ScuIcuCcS OIcu ShOW 3 IcVcIS3l OIhc C3uOuIC3l OI0cIIu_
O3 IcHc 3n0 3 lOC3IiVc Qh13Sc. tOI cX3HQc, Iu HuSSI3u 3 C3uOuIC3l
lOC3IiVc QIc0iC3IiOu COnSISIS O 3 IhcHc OlOWc0 Dy 3 OC3IIVc Qm3Sc
(?). hIS CODSQOu0S IO Ihc_cucI3C3uOuIC3ODOQIc0IC3IIOu Iu HuS-
Si3u, 3S Sccn Iu (3). u Ihc cXISIuII3l Ou Uc OIhcID3u0 Uc C3uOuIC3 OI0cI
IS IcVcISc0, 3u0 Ic lOC3IIVc QhI3Sc QIcCc0cS Ihc SuD]cCI, 3S Sccu Iu P
SIDl3I SIIu3IIOu hO0S IutIuuISh 3u0D3uy OUcI3u_u3_cS.
(2)


Kui_3 n3 SIOc
DOOK Ou I3Dc
` cDOOK IS Ou Ihc I3Dlc.
PlHc0 KOHQOSIIOI
PHc0 COHQOScI
`PlIc0IS 3 CODQOScI.
D3 SIOc ]cSI KuI_3
Lu I3Dlc COF DOOK
`cIc iS 3DOOK Ou Ihc I3Dc.
[RUSSIAN]
[RUSSIAN]
[RUSSlAN]
SpeciaI copuIas uH3uy M_u3_cS, Ihc cXISIcuII3l COuSUuCIIOu IuVOVcS 3 COQu3
OIVcIDIh3IIS0ISIIuCI HOD Ihc COQu3uSc0Iu C3uOuIC3 COuSUuCIIOuS. tOI
cX3HQc, In Ic HuSSI3u cX3HQcS 3DOVc, Ihc COQu3I ccDcuI]est ` IS OC-
CuISIucXISIcuII3l COuSIIuCIIOuS DuIuOIIu _cucI3l Iu COQuMCOuSIIuCIIOuS.
bIDl3I SQcCI3 COQu3S d Ouu0c. _. Iu1cDIcW 3u0uIKISh.
Locative pro-forms u DMy l3u_u3_cS, 3 OC3IIVc 0cICIIC cXcDc IS _Im3II-
C3IZc0 IuIO Ihc cXISIcuII3l COuSUuCIIOu, WhcIc II IS SIIIQQc0 OIIS 0cICIIC
uuCIIOn. LX3HQlcS 3IctIcuCh `IhcIc , 3u0 Lu_lISh the.
Morphosyntax PCIOSS l3u_u3_cS II IS COHDOu OI cXISIuII3 COuSIIuCIIOuS IO
ShOW nOu-C3uOuIC3 3u0 uuSI3Dc HOIQhOSynI3CIIC Q3IIcDS O3I_uDcuIIc-
3liZ3IiOu 3u03_IccHcuI. IVOIDSOIcuShOWunSI3DcC3Sc HMKIu_IuIh3I
Ihcy DuCIu3Ic DcIWccu C3uOuIC3l SuD]cCIH3IKIu_ 3u0 H3IKIu_ DOIc IyQI~
C3l O OD]cCIS. tOI cX3HQc, Iu DOIh HO0cH tcDIcW 3u0 Lu_lISh, QIVOIS
RuCIu3Ic DcIWccu uOHIn3IIVc M0 3CCuS3IiVc C3Sc. PSO, cXISIcuII3 COu-
SIIuCIiOnS OIcn cXhIDII uCIu3IIOn DcIWccn 3_IccHcuI M0 uO 3_IccHcuI
[i. c. IHQcISOn33_IccDcuIOu Ihc cXISIcuIi3l VcIDlCOQu3).
Bo Suencf. |on )cnontcttfv ano Dtsteuf|on tn Exisfenftats
hc Ic3IIOu DIWccu cXISIcuII3 COuSUCIIOuS 3u0 COQulM COuSUuCIIOuS h3S c3-
IuIc0 cXIcuSIVcy Iu SyuI3CIIC IcScMCh, WhcIc Ihc IWO COuSUCIIOuS 3Ic OIcu Sccu
3S0cDV3IIOu3yIc3Ic0,OI 3S h3VIu_ 3COHDOuuu0cIyIu_SUuCIuIc. HcSc3IChOu
Ihc 0ISIIIDuIIOu OQIVOIS, hOWcVcI, h3S uOI Iu _cucI3l cXQOIIc0 IhI S Ic3IIOuShIQ.
LOQu3I3u0cXISIcuII8COuSUuCIIOuSMc OHcuSIDIM Iu |UuIh-COu0lIl Ou3) HcM-
lu_, 0IHcIIu_ Ouly IuIRc W3y Uc D M_uDcuI ISIc3IZc0. huS, WhIc C3uOuIC3
SuD]cCISQIcCc0IhcCOQu3,QIVOIDSIcu0IOOlOWIhc COQu3OIcXISIcuIl3 VcID.
hlc C3uOuIC3 SuD]cCISQIcCc0c lOC3IIVc Qm3ScS, QIVOI S Icu0 IOOOW IhcD.
P u3IuI8 IuIIQIcI3IIOu O IhIS SIIu3IIOu IS Ih3I SQc8cIS ChOOSc IO Ic3lIZc
DS 3S QIVOIS Whcu Ihc uuD3IKc0 OQIIOu O Ic3lIZIu_ IhcD 3S SuD]cCIS ID COQu-
3I COuSIIuCIIOuS IS OI SODcIc3SOu uu3V3Il3Dc. hc Ic3SOu OI IIS uu3V3I3DlIIy
H3y Icl3I IO QIOQIIIS O Ihc QIVOI D IISc, IIS CODQ3IIDIlIy WIIh 3 C3uOuIC3l
COuSUuCIIOu. hyQOIhcSIZc Ih3I C3uOuIC3 COQuM COuSIIuCIIOuS M0 cXISIcuII3
COuSUuCIIOuS3IIuCOHQcUUOu, 3u0 Ih3I UISCODQcIIIIOu IS0IIcCIy Icl3Ic0 IO Ihc
0ISUIDuIIOu OQIVOIS.
(5) Markedness hypothesis hc OQQOSIIIOuDcIWccu cXlSIcuIl3 3u0
C3uOuIC3 SUuCIuIcS IS Wh3I uu0cIIcS Ihc 0ISIIIDuIIOu O DS lu
QIVOIuCIIOuCIOSS-Iu_uISUC3llj.
cOW W Su__cSIIh3IQIVOIDS3IcDSIh3I3I uOI_OO0CM0I03IcSOIuuCIIOu-
Iu_ 3S SuD]cCIS,OI3CKOQIOIOIyQIC3 SuD]cCIQIOQcIIIcS.
Z. 1. Ihe De/n/teness Eect
I IS Wc KuOWu Ih3I uOI 3l P IyQS Mc cQu3y cICIIOuS 3S QVO|S. PDOu_ Ihc
Klu0S ODS Ih3I h3V Dccu S3I0 uOI IO OCCuI 3S QIVOIS Mc 0cHuI|c DS 3S Iu
3u0 (7), 3u0 SUOu_ly Qu3uIIDc0 DS 3S Iu (8). hcSc IcSIIICIIOuS 3Ic KuOWu 3S Ihc
`0cHuIIcucSS cCI.
11hcI ISIhQIODcD|3_3Iu)WIIhIhISDODIcQhOuc.
(7) 11hcIcIShIDOu IhcCOuCh.
(8) 11hI IScVcIy CI_3IIu L3VI0S 03WcI.
hc DO0y O IScMCh 0c3lIu_ WIIh Ihc 0cDuIIcucSS ccCI [Scc c. _. tccu3u (2003),
MWISc 3u0 LOOQcI ( I 9& I ), NIlS3IK ( I 977), NCD3y ( I 99&), ZuCChI ( I 995) h3S
DOSIy Dccu D3Sc0 Ou COuSUuCIc0 Lu_lISh cX3HQS. LOuScQucuIy, Ih CIOSS-
Iu_uISIIC V33IIOu Iu IhccXIcuIOIhcScIcSUCIIOuS Ou QIVOIS h3S uOIDccu SIu0Ic0
SySIcH3IIC3y, uOI h3S lI Dccu I8cu IuIO 3CCOuuI Iu ODu3IIn_ IhcOIIcS O Ihc
0cDuIIcucSScHcCI.
hIc SODc 3uIhOIS h3Vc uOI0 Ih3I Ihc 0cDuIIcucSS cHcCI 0OcS uOI hO0
cQu3ly OI 3 3u_u3_cS [Scc c._. ZIV ( I 9&2) OI 1cDIcW, NOIO ( I vv4) OI I3I3u
3u0cuIcy (2002)OI bM0IuIM), IhcSc3uIhOIS0OuOIQIcScuI3|hcOIjOIhc 0cD~
uIIcucSSccCIIh3IcXQ3IuSCIOSS-Iu_uISIIC V3D3IIOu. bID3Iy, cVcu IhOu_h IIh3S
2I
22 L8VIOO8VO,l8m88HOOZ 8HO LmlyOVlH8CH
DccuuOIc0Ih3IcVcu IuLu_ISh,HOSI IyQcS Ih3Ih3VcDccuC3Hc0uOIIO OCCuI
iu cXISIcuII3 COuSUuCIIOuS 0O Iu3CIOCCuI [WIIh VMIOuS IuIcIcSIIu_ COuScQucuCcS
OI IhciI IuIcIQIcI3UOu), 3V3I3Dc ScH3uIIC IhcOIIcS OIhc 0cDuIIcucSS ccCIUc3I
II3S 3C3Ic_ODC3lD3uOu3C3SS OS, 3u0OCuSHOSIy Ou IIyIu_IO ChM3CIcIIZc
IhIS C3SS. U Ihc uOI3Dc cXCcQIIOu OM0 3u0 IDcI ( I 995), IhcIc h3Vc uOI
Dccu cXIcuSIVc 3IIcDQlS IO cXQ3Iu Ihc OCCuDcuCc O 0cDuIIcS 8u0 OIhcI IyQcS
IuLu_ISh cXISIcuII3l COuSUuCIIOuS IuIcIHSO_cucI3 Iu_uISIIC QIIuCIQcS.
NIKKcScu(2002)0cVcOQS 3IhcOIy OL3uIShcXQlcIIVcCOuSUuCIIOuSIh3IIS
IOOIc0IucX3CIy IhISIuIuIIIOu. NIKKcScuS 3u3ySISUc3ISIhc 0cHuIIcucSScHcCI 3S
cQIQhcuOHcu3 IO Ihc IuIcI3CIIOu OHMKc0ucSSCOuSU3IuIS Ou SuD]cCIS. bhc ShOWS
Ih3I uO SQcCI3 COuSU3uIS ucc0 IO Dc QOSIIc0 OI cXQcIIVc COuSUuCIIOuS. uSIc30,
h3IHOuIC 8I_uHcuICOuSU3IuISDuIIOu Wcl ODScIVc0 IyQOO_IC3 Icu0cuCicS Su-
HCc IO 0cIIVc Ihc 0ISUIDuIIOu OQIVOIS in L3ni Sh d Wcl 3S Lu_ISh. hc QIOIOIyQ-
IC3 SuD]cCIQIOQcIIy U3I D_uIcS Iu hcI 3u3ySIS IS 0cDuIIcucSS. PCIOSS 3u_u3_cS,
0cHui Ic SuD]cCIS Mc QIccIIc0 OVcI iu0cHuiIc OucS [LIVOn I 97&). u NIKKclScu S
IhcOIy, Uc VMI3IIOuIuIcl3IIVcI3uKIu_ OCOuSII3IuIS 3_3IuSIcXQcIIVc SuD]cCIS M0
3_3iuSIIu0cDuIIc SuD]cCIS 0cIIVcS Ihc0ISIIIDuIIOu OS IuL3uISh3u0Lu_lISh.
LcDuIIc S hOWcVcI Mc Ony 3 SuDScI O Ihc IyQcS Ih3I h3Vc Dccu
S3i0 nOI IO OCCuI Iu cXISIcuII3 COuSIIuCIIOnS. hIS IS uOI SuIQIISIu_ Ou Ihc QIcScuI
VIcW, SIuCcOIhcIIhm 0cDuIIcucSS,IhcIcMc HMyOIhcIQIOQcIIIcS Ih3ICh3I3CIcIIZc
C3uOuIC3 SuD]cCIS, 3u0 UcII IHQOII3uCc V3IIcS HOH 3u_u3_c IO lM_u3_c. hIlc
SODc O IhcD Mc _I33IIC3IZc0 Iu SOHc 3u_u3_cS, OIhcIS QIO0uCc SI3IISIIC3
Icu0cuCIcS. c3IDu ShOW Ih3IIhc COHQcIIUOu 0VIu_ Ihc 0ISDDuIIOu O0cDuIIc
S 3u0cXQcIIVcS IuNIKKcScu` SIhcOIy ShOul0DccXIcu0c0 IO 3l IyQcS.
hcI0c3U3IDMKc0ucSSC3uDcC3QIuIc0Dy3SSOCI3IIu_ 0cIcuIhIcIMChI-
C3 SC3cS_OcSD3CKIO WOIKDy NICh3cl bIVcISIcIu, 3u0ISDyuOWWcllcSI3DIShc0
IuIhc QQOO_IC3uCIIOu8 IIcI3IuIc, cSQcCI3y IuLQIIH3IIy hcOQ. bIlVcISIcIu
( I 976) QIOQOSc0Ih3I QQOlO_IC3y, clcHcuISIh3IMchI_h Ou 3QcISOul3uIH3CyhIcI-
3IChy Mc Dy 0c3uI3SSOCI3Ic0WIIhHOIcQIOHIucuIUcH3IIC IOcS. PISScu ( I 999)
QIOVI0cSIhcOllOWIu_ VcISIOu ObIVcISIcIuShIcI3IChy.
(9) cISOnl PuIH8y hIcIMChy.
8. LOCd| U|8OD > ||OU| NOuD 3|d > MumdD 3|d > PD| mdlU 3|d
> | DdD| mdlU |d
U. PgUDl > |dl|UDl
PISScu IHQlcHcuIS bIlVcISIcIuS I0c3S OIH3lly Iu L, 3u0 ShOWS hOW hIS _cucI3
IyQOO_IC3 I0c3 3QQIcS Iu QMIICu3I _I3MS. 1u cXISIcuII3 COuSIIuCIIOuS, Wc
|Mc II Ih3I Ihc KIu0 O SC3c OHcIc0 Dy bIVcISIcIu 3u0 PISScu ScIVcS Iu QM3c
IO OIhcI SC3cS SuCh 3S Ihc 0cDuIIcucSS SC3!c 0ISCuSSc0 Dy NIKKcScu, d Wcl 3S
OIhcI SC3cS ycI IO Dc QIcCIScy 0cDuc0. tOI cXMQc, Iu Ihc IIcI3IuIc Ou cXIS
IcuII3l COuSUuCIIOuS II IS COHDOuy ODScIVc0 Ih3IuSc OIhcCOuSUuCIIOu IS Iuu
cuCc0 Dy 0ISCOuISc QI3_H3IICS, 3u0 Ih3IQIVOIS Icu0 IO 0IHcI QI3_H3IiC3y HOH
SuD]cCIS. 3I0 8u0 IDcI ( I 995) C3IH Ih3I cXISIcuII3 ScuIcuCcS HuCIIOu IO Iu
UO0uCc ucW 0ISCOuISc IccIcuIS, 3u0 hcnCc Ih3IQIVOIS 3Ic hcMcI ucW. Hc3Ic0y,
HcIuh3II (2005) 3HOu_ OIhcIS h3S Cl3IHc0 Ih3I QIVOIS C3unOI Dc ScuIcuCc IOQICS,
Bao Suencf. (Non- )canotcattf ano NPDisteufton tu Extsfenftats
whi|eZucchi( I 995)c|aimsthattheycannotbepresuppositiona| . Itiswe||known
that canonica| sub]ects are often discourse topics and hence both hemer old md
presuppositiona| . Onasca|ebasedapproach,theseobservationswou|dD capmred
byemp|oying apragmaticsca|esuchasmeonein( I 0).
( I 0) Topica|ityscale.
D| SCDU|S D| C > D| SCDU|S D| D > RB|| D| D > RB|| DW
Such a sca|e is in |ine with properties thathavebeen independent|y c|aimed to
cone|atewithsub]ecthoodinthe|iterature, e. g. byKeenan( I 976),andareknown
tobeinvo|vedinphenomenasensitivetosub]ecthoodsuchasvoicealtemauons.
In mispaperwedonotprovideasca|eofsub]ectpropertiescrse. Rather,
wearguefororderingsofNPtypesthatarep|ausib|yre|atedtounderlyingsca|esof
subectproperties. Whi|e the denitenesseectis standard|y seen ascategorica|,
onthe sca|ebasedviewpresentedbyMiHe|senandadoptedandexpmdedhere,
itis inherent|ymorecomp|ex. Thedenitenesseectisthe productofakindof
optimizationre|ativetosub]ectproperties.
( I I ) 5ubject propcrties hypothesis.
NPs that exhibitproperties associated with sub]ecthood me at-
tracted to consuuctions invo|ving a cmonica| sub]ect, whereas
thosethatdonotdisp|aysuchpropertiesmeatuactedtoconsmc-
tionsthatdonotinvolvecanonica|sub]ects.
Ifthepictureustout|inedisconect,thanweexpectthreethings.
Gradient distribution: Since there me usua||ymany factors aectingmelikeli-
hoodthatacertain NP type wi||be sub]ect, we expectdierent NPs mat
wou|duaditiona||y intoonetype,e. g. strongNPsorweNPs(Mi|smk
I 977),tovaryintheirpropensityforoccuningineitherofthetwore|evant
constructions. In other words, the deniteness eect shou|dbe gradent
ratherthancategorica| .
Anti-deniteness effects: IfNPs that do notoccur often in existentia| consmc-
tions(thedenitenesseect)meprototypica|sub]ecu,menweexpectNPs
thatdooccurofteninexistentia|constructions,i.e.mepivots,nottobepro-
totypicm sub]ects. Thedenitenesseecthasa ip side, inthatcertainNP
types shou|dberesuictedintheir capacity tooccuras cmonicm subects.
TheseNPs shou|dthen be attracted tothepivotposition. This fo||ows
homthe |ogic ofmarkedness. NPsthatcun berealizedas canonica| sub-
]ectswi||besorea|ized.
5ystematic variation: Wenotonlyexpectlanguagestovaryinthedegreetowhich
theyshowadehnitenesseect(duetodierencesinwhichpropertiescount
towardscanonica|ityofsub]ects),butwea|soexpectthatvariationtobein
aconsistentdirection. Forexamp|e, weexpect thattwoNPtypesthatare
orderedin re|ationtotheirpropensityforsub]ecthoodwi||notswitchtheir
?J
Z4 L8VO O8VO, l8m88HOOZ 8HO Lmty OVH8CH
orderinanother|anguage.Simi|mly,weexpectthattheremeno|anguages
inwhichsuong|yqaantihedNPsmemuchworsesub]ectsandmuchbetter
pivots than weak indenites. Ourtheory therefore suggests imp|icationa|
universal s.
Fromanempirica|pointofview, suchmaccountenab|esareneddiagnosticfor
decidingwhen aconstructionexhibitsadenitenesseect. aconsuuctioninvo|ves
suchaneectifthedistributionofNPsinitiscone|atedwiththeorderingofN
types, seen as animp|icationa| aniversa|. In section 4 we discussthe use ofsuch
criteriawithreferencetoRussianNPdisubution.
Asmentioned em|ier, the gradience ofthe disuibution ofN pesin ex-
istentia| sentences and the systematicity ofthis gradience have not been noted in
the|iterature. Theobservationthatindenitesmeinfe|icitousassub]ectsundercer-
taininterpretationsutnotothers)isontheotherhandrathero|d. Inpmticu|m,as
Mikke|sen(2002. 25)notes,ithas|ongbeenobseredthatsoca||ed' weak' inde-
nitesmebadsub]ectscross-|inguistica||y,mdinfactmeungramaticmassub]ects
insome|anguages on a non-generic interpretation. However, thi sfactaboutthe
disuibutionofindeniteshasnottoourknow|edgebeenre|atedinanywaytothe
denitenesseectinexistentia|s.
2.2. Ev|dence /or Cmss-//ngu/st/c \ar/at|on |n E|stent|a/s
Wenow present Eng|ish, Russian andHebrewdatathatprovides iniua| evidence
ofcross-|inguisticvariation, andwe showthatthetypeofvmiationweobserveis
consistent with the hypotheses suggestedabove. The data we considerinvo|ves
pronominale|ements,andcentersonthreefactors.the|oca|ityofpronouns,theref-
erentia|ityofpronounsandthe po|mty oftheconsuuction. Basedonmesefactors,
Russim,HebrewandEng|ishformac|inewithmspecttopronountypesa||owedin
existentia|constructions.
Considerrstsentences( ! ?) - ( ! 4) asanswerstoaquestionlikewhere |s m)
wa//et? A||threesentencesareverymarkedasanswerstothiskindofquestion.
( I 2) ?? There' sitonthetab|e.
( ! J) ?? ona]est'nasto|e.
it COF ontab|e
'Thereisitonthetab|e. '
( !4) ??yes oto a| ha-su|xan
COF himonthe-tab|e
'There'sitonthetab|e. '
[RUSSlAN]
[HE8REW]
Nowconsiderthe sentences ( I 5) - (! 7), eva|aated as answersto a question|ike
where can l ]nd th|s book TheEnglishsentenceissti||verymarked,whereasthe
RussianandHebrew sentencesarenatara| .
AU bu8JEC1. [PON-)CANONICALIT AND NP!STR!8U1!ON lN LXISTEN1lALS
( l 5) * hcIc SIIIDIhc IDIMy.
[ ) Ou3 ]cSI V DIDIOIcKc.
lI. NOM IS IDIDIMy.
` hcy h3Vc IIIDIhc IDI3Iy. [II. . HcIcS IIID Uc IDI3Iy.)
( I 7) ycS OIO D3-SIHIy3.
COF hID ID-Ihc-IDI3Iy.
`hcy h3Vc IIID Ihc lIDIMy. [II. . hcIcS II ID Uc IDI3Iy.)
[RUSSl AN]
[HE8REW]
hc ScH3u|IC IuIcIQIcI3IIDD O Ihc QIVOI ID ( I 6)-( I7)I S 0IHcIcDI Ih3I Ih3I Iu ( I 2)-
( I 4). YIc Ihc 3IIcI ScDIcDCcS 3Ic uSc0IO 3uSWcI 3 QucSIIOD 3DDuI 3 Q3IIICuM
W3cI, IhcOIHcI McuSc0IDMSWcI3 QucSIIOu 3DOuI 3Dy IOKcD OUcDOOK uuCcI
0ISCuSSIOu. LOOScy, Wc D3y S3y U3I Ihc QIVDI Iu ( I 6)-( I 7) IS uSc0 cSS IccIcu-
II3ly. hc cX3CI D3IuIc OIhIS 0IHcIcDCc IccIcuII3IIy IS uOI 0IIcCIy IccV3uI IO
OuI CU cu| QuIQOScS, DuI Wc DOIc Ih3I II IS Ic3Ic0 IO Ihc 0ISUuCIIOD D30c Dy c. _.
NC3lly ( I vv) DcIWccu IyQc 3u0 IOKcu 0cuDIIu_ DS. c IhuSIccI IO Ihc PSIu
( )-( ) 3SIjQc-CcuDIIu_, 3D0 IhOSc Iu( I 2)-(I 4)3SIOKcu-0cuOIIu_. hc IccV3DI
QOIuIcfc IS Ih3I1cDIcW 3u0HuSSIM McDOIh HuCh HOIcQcIDISSIVcIh3u LD_ISh
Iu 3OWIu_IyQc-0cDOIID_ QIODOuDS ID cXISIcuII3 COuSIIuCIIDDS.
hc SIIu3IIDD IS 0IHcIcDI WhcD QDl3IIIy IS Ch3D_c0. DIc IDMID3Ic QIO-
uOuuS 3Ic SII VcQ H3IKc0 ID LD_ISh Iu uc_3IIOu COuIcXIS, ID HuSSI3D M0 1c-
DIcWuc_3Ic0CODIcXISSccDIOIcDDVcUcIHII3IIDDOQIVDISIOIyQc-IDIcIQIcI3IIOD.
huS, ( ! v) 3u0 (20)3Ic_OO0d MSWcIS IO Ihc QucSIIOD where /s m wa//et ? .
( I ) "hcIc ISD I II OD Ihc I3Dc.
( I v) c O ucI D3 SIOc lD3 SIOc]c]O ucI.
I I. GN uOI-IS OD I3Dcl OD I3Dc II. OEN uOI-IS.
` ISuOIODIhc I3Dc. [II. . hcIc ISu IIIOuIhc I3c. )
(2U) cyu OIO 3 h3-buXM.
COF hlD Du Ihc-I3Dc.
`ISuOIDD Ihc I3Dc. [II. . cIcISu IIIOu Ihc I3Dc. )
[RUSSlAN]
[HE8REW]
OW COuS0cI Ihc ScDIcDCcS ID (2I )-(22), IDVOVIu_ 3 OC3l QIODOuD. hcSc Scu-
IcuCcS 3Ic ID 3l IhIcc 3D_u3_cS ]uSI 3S DMKc0 d ( I 2) - ( I 4) WDcD uSc0 d 3u
3uSWcI |O 3 QucSIIOu IKc where are you`.
(2 ) ` ` hcIcI S Hc 3I hOHc.
(22) `]3]cSI 0DD3.
IS hODc
`hcIc SHc 3I hOHc.
(2J) ``ycS OII D3-D3yII.
COF Dc Iu. DEl-hODc
`hcIc SHc 3I hOHc.
[RUSSIAN|
[HE8Rw]
25
Z L8VdO8VOt,tMMf8HCOZ 8Hd LmtOVlH8CH
However, whenthepo|arityischanged, thereisadierencebetweenRussimand
Hebrew. whi|esuch sentencesmenotacceptab|e inHebrew, they are acceptab|e
andinfactcomoninRussian.
(24) * Thereisn' tmeathome.
(25) Men]anet doma.
I. osN not-ishome.
' I' mnothome.' (lit. . Thereisn' tmehome. )
(26) ?? eyn oti ba-bayit.
cor. Nsomein. osr-house
' I' mnothome.' (|it. . Thereisn' tmeathome. )
[RUSSIAN]
[HE8REW]
Thedataare sU arizedintab|e I , whichshowsasystematicincreaseintheac-
ceptabi|ityofpronouns. Eng|ishis|eastpemissive,Hebrewislesspermissivethan
Russian, andRussianismostpemissive. Importantly,notethatpropertiesassoci-
atedwithsub]ecthoodnice|ypredicthowacceptableapronountypeis|ike|ytobe
cross-linguistical|y. Themoresub]ect-|ikemeN, the|esslikelyitistobeaccept-
ab|einexistenual s.
Iye o/ mnoan Eam/es ENoL!sH Hssksw Russ|AN
os neg os neg os neg
Loca| (2I ) - (26)
m
- - - -
Non-|oca|
,
token ( I 5) - (20) - -

-
,
type ( I 2) - (I4) - -
Table! . Acceptabi|ityofpronounsinexistentia|sinEng|ish,Russian,andHebrew.
Aanging the pronouns according tohowcommon they are inexistentialscross-
linguistica||y,wegetthattypedenotingnon-|oca|pronounsmemostcommon,then
tokendenotingnon-|oca|pronouns, andthen |oca| pronouns, which only occurin
Russian,mdon|yundernegation.Thisorderingisexact|y whatweexpectifpivots
are badsub]ects. Itconesponds(inreverse)tothefo|lowingtwosub]ect-property
sca|es.
LDCB| ||DDDUD > DDD| DCB| |DDDUD
N|S W| R DKD |D||B| DDS > N|S W| R y | D||B| DDS
Thatloca|ityi safactorinsub]ectse|ectioniswel|knownfromtypologica|literature,
andfo||owsom|oca|pronounsbeinghigheston the animacy/promi nencesca|e.
Furthermore,thisorderemergesinthequantitativestudiesreportedbe|ow,andthe
convergence ofresults is obviouslynotincidenta|. Thattoken interpretations me
more sub]ect-|ike then type interpretationsfo||ows naturallyfrom the assumption
thatsub]ectstendtobepresuppositiona|andspecic.
Bo Suencf. (Non-)cnontcxfv no NP Dtsfateufto tn Extsfeftt
The tab|e abovesupports ourclaimthat the ordering ofN types in exis-
tentia|sconstitutesanimplicationuniversal . ifa|anguageallows|ocpronouns,
ita|soa|lowsnon-localpronouns,andifitallowstokendenotingpronounsita|so
allowstypedenotingones. A|thoughadetai|edtheopis|acking,itisunsnsing
thatnegationshou|daectpivotdismbution,sincenegationa|tersthesetofproper-
tiesthesub]ectisentai|edtohave. We predictthatsuchasymmeuy,whenitexists,
shou|dconsistentlyho|dinthesmedirection.Independentpolmty|icensingfacts
apart,noNP typesshouldbemorenatura|aspivotsinpositivecontextsthaninneg-
ativeones. Inthequantitative sudiesmportedinthenextsection, a simi|eect
wi|lbeseen. downwmdmonotonequantiers,includingthenegativequantiersno
inEng|ish andgeen 'no' inDutch, wi|| D shown tobemore suongly existential
manupwardmonotonequantiersinasensetoDmadeprecise.
Quantitative Data from Dutch and EngIish
Mostdataondenitenesseectsintheliteratureisdawnom]udgmentsoncon-
smctedexamp|es.We donotsuggestdispensingwith]udgmentsonarticia|data,
butthinkitusefultosupp|ementsuchevidencewithquantitauvedatabasedonnat-
ural|yoccurringexamp|es. MuchofthedatareportedhereisdrawnomGoog|e
semches. However, mmymethodo|ogic issuescomeup intheuseofdataob-
tainedhomthe web usingstmdmdsemch engines, so itwasimportanttocheck
thattheresultscou|dbedup|icatedusingsemchtoolsdedicatedtol inguisticwork
onstructuredcoqora. We reportonsomepre|iminmyresu|tsomthese|atterin-
vesugationsbe|ow.
In thersttwosectionsofthispaperwe suggestedmatme disuibuuon of
Nsasexistentia|pivotsbe understoodinterms ofacompetition. anN canap-
peminpivotpositionwhenitlackscanonic sub]ectproperties. This suggests a
newwayof|ookingatdenitenesseectsinexistentialconsuuctions. Standmd|y,
the dismbution isconsidered |n vacao. agivenNP typeeitherdoes or does not
occurinexistentialpivotposition. Whatwewi|lconsiderinsteadistheprevence
ofN types in existentia| pivotpositionre|ative to theirpreva|ence incmonica|
sub]ectposition.
J. !. Method
Thequantitative smdyinvolvesthefollowingvmiables.
NF type Thecl assesofNP weconsiderme mostly eitherstandmd|exica|Ncat-
egories(e. g. pronouns)orelsec|assiedaccordingtotheheaddeterminer.
Fosition For the web-based studies reported here we considered two templates.
TheexistentialtemplateforEnglishistheres is are [NP pattem] in, and
the canonic fom templateis[NP pattem] is me in. Here [ pattem]
ZJ
Z L8VOO8VO, t8m88HCOZ 8HO LmtyOVH8CH
might,forexamp|e,beevery where isawi|dcmdmatchingmysing|e
word. NotethatfortheGoog|esemcheswe genera||yrestrictedourse|ves
to NPs wim a sing|e noun, though in our suuctured coqus searches we
dispensedwiththisconsuaint.
The web based semches usedpattems inc|uding a preposiuon in. The
reasonisthatsimp|esemchesbasedonatemp|ate[NPpattem]isproduce
ahighpercentageofmatcheswhicharenotofthedesiredform[S[N[VP
[Vbe] ...]] ] . Semcheswhichare more specic,duetotheinc|usion ofan
addiuona| e|ement, produce ahigherpercentage ofmatchesofthedesired
syntacticfom. Oursearcheson structuredcorpora(al|pre-parsed)didnot
requireanexp|icitprepositionsincewecou|dexuactappropnatesyntactic
matcheswithahighsuccessrate.
Language InthissectionwereportonresultsforEng|ishandDutch,andinsection
4 wi||reportonquantitativeworkonRussian.Othersinourresearchgroup
haveperfomedquantitauve ana|ysesforDanishandSpmish,andwehope
toreportontheseata|aterdate.
Genre ForEng|ishweperformedstudiesusingcorporaconstructedhomquitedif-
ferentsources.Thewebisheterogeneous,andhasa|owpercentageofuan-
scribed speech. Thethree suucturedcorporaweusedwere. Switchbomd,
a speech corpus dawn hom te|ephone conversauons, Brown, a ba|anced
corpusinc|udingamixtureoftranscribedspeechandothertexttypes, and
WSJ, aco||ectionofnewspaperartic|esfromtheW
a
||StreetJouma|.
+
Frequency Theon|ydependentvariab|e isfrequency,i. e. thenumberofmatches.
However, for webderiveddatatherawnumberofmatchesusingstandmd
websearchtoo|sisnotadequatebecauseofme high incidence ofinappro-
pnatematches.Toosetthisprob|em,foreachseparatesearchwesamp|ed
theresu|ts. Then,ratherthanusing therawnumberofsemchmatches, we
usedad]ustedfrequencies,thatis, therawnumberofsearchmatchesmu|-
tip|iedbymeproportionofsearchresu|tswhichhavetheintendedformas
measuredbyoursamp|e.
Thereareanumberofwaystostudythere|ativefrequencyofNPqpesincanonic
a
|
andexistenua|consuuctions. For examp|e, it wou|dbereasonab|e to considerfor
agivenNPtypetheproportionofNPsofthattype which occurinexistentia|con-
suuctionsoutofthetota|numberoftokensofthatNPtypeineitherconsuuction.
However,weoptforadierentmemc,name|ytheratioofthenumberofcanonica|
tokenstothenumberexistentia|tokens.Thedownsideofusingthismetricisthatit
isundenedincaseswherewefoundzerooccurrencesofanNPtypeinexistentia|
pivotposition, but the upside isthat despite the sma|| numberofNPsforwhich
thememcfai|s,itnonethe|essbringsoutthedierencesbetweenNPtypesmore
c|ear|ythanotherstatistics we considered. Forcases where the ad]usted number
ofexistentia|pivotoccurrenceswaszero,we(natura||yenough) ueatedtheratioas
tendingtoinnity.
Bao Suencf. (Nou-)coouicoti1 om Dis:eisu:iou iu Exis:e1ots
hc I3u_c OV3II3IIOu OIhc LMOuIC3LXISIcuII3 I3IIO IS cuDDOuS, cVcu
DI C3ScS WhcIc II IS Wc 0cDuc0. H DuI Lu_ISh 03I3, 3 DS Wc COuSI0cIc0 h30
uDu-ZcIO uuDDcIS O cXISIcuU3 QIVDI DCCuDcuCcS, DuI Ihc I3IID I3u_cS DcIWccu
0. 02 3u0 I 70,000. PI Ihc OW cu0 Mc S D Ihc DID at most N WIIh 3 I3IID
O0. 0247. hcSc 3Ic IhuS cXUcDcy cXISIcuII3, Dy WhICh Wc Dc3u U3I SQc3KcIS
h3Vc 3 SIIOu_ QIccIcuCc OIQuIIIu_ UcD Iu cXISIcuII3 QIVOI QOSIIIOu I3UcI Ih3u
CMOuIC3 SuD]cCI QOSIIIOu Whcu Ihc ChDICc IS 3V3I3Dc. PI Ihc hI_h cu0 3Ic D3
QIOuOuuS, WIIh 3I3IIO O I 67,000,WhICh3Ic IhuS hI_hy CMOuIC3 .
O cXcDQIy hOW Ihc I3IIOS Mc C3Cu3I0, CDuSI0cI 3 SIDQlc cX3DQc ID~
VDVIu_ Ihc IyQc/ew N. c DISUy I3u ScMChcS Du Ihc QuOIc0 SUu_S there
s |s are /ew |n OI cXISIcuII3 DCCuDcuCcS 3u0 ']ew are |n. hc H3IChcS
cX3CIy OucWOI0, 3u0 Ihc j -OQcI3IDI ISIuIcIQIcIc0 3S 0IS]uuCIIOu. tOIIhccXISIcu-
II3 COuSUuCIIOu, IhcIc WcIc I 05,000hIIS, 3u0DI IhcC3uOuIC3COuSUuCIIOu IhcIc
WcIc 34,000.cD3uu3y S3DQc0 Uc IcSulS Dc3Ch Sc3ICh. tOIc3Ch COuSIIuC~
IIOu, Wc COuuIc0 Ihc QcICcuI3_c D C3ScS WhcIc DuI 3u3ySIS O c3Ch S8Qc 0I0
uDI CDDcSQOu0 IO Ihc SUuCIuIc Wc WcIc SccKIu_. c COuuIc0 3S cDOIS 3uy C3Sc
WhcIcIhCSyuI3X W3SuOIWh3IWcWcIcSccKIu_, I.c. NP beo PP OI Ihc C3uDu~
IC3 COuSIIuCIIOu 3u0 /et|ve-there beom NP PP OI Uc cXISIcuII3. H SOHc
C3ScS, 3S Whcu 3 ScuIcuCcDOuu0My IuIIcuc0ID Uc D3IChc0 SIIIu_, Ihc C3SSID~
C3IIDu W3Sc3Sy,DuIIu OIhcIC3ScS,3SIu0cCI0Iu_ WhcIhcI 3 CODQcX cXQIcSSIOu IS
3 OI0cCI0Iu_ WhcIhcI3uDCC0 cuCc Dthen IScXQcIIVc OIOC3IIVc, Ihc]u0_c~
DcuI IS SuDIc. tOI ScMChcS IcRIu_ VcIy SD3 uuDDcIS OIcSuIS (_ I00) Wc
ChcCKc03IcSulIS.tOIDDSIScMChcSlhcDuHDcIDIcSuISW3SDuChhI_hcI|uQID
I 0,000,000).tOIIhcSc WcQcIDIDc0MIuIlI3S8QcD20DSuIS. IhccDOII3Ic
W3S 3DOVc 50%,WcIhcu IuCIc3Sc0lhc S8Qc SIZc, S8QIu_uQ IO I 00IcSuIS IuUc
cWC3ScS WcIc Ic ScMChcS IcRc0 CDSc ID 90%cDDucDuS IcSuIS (4 C3ScS DuI
D46). u Ic C3Sc O/ew, I 0%DIhc cXISIcuII3 IcSuIS WcIc cIlDucOuS, 3D0 75%
OIhc C3uOuIC3l IcSuIS WcIc cIIDucOuS. Luc SOuICc D Ihc hI_h cDOI I3Ic DI/ew
Iu C3uDuIC3 COuSIIuCUOuS ISIh3ID3uy DIhc IcSuIS WcIc Iu 3CIOIa /ew. ` 3Sc0
DuIhcSccIIOI I3IcS, Wc C3Cu3Ic0 30]uSIc0 HcQucuCIcS. tOI/ew Iu cXISIcuII3S Ihc
30]uSIc0 D_uIc IS I05,000X 0. 90 ~ 94500, M0 DI/ew Iu C3uOuIC3 COuSUuCIIDuS
II IS 34,000X 0.25 " 8500. c Ucu C3Cu3I0 Ihc I3IID OIhc 30]uSIc0 C3uDuIC3
HcQucuCy IO Ic 30]uSIc0cXISIcuII3 HcQucuCy, WhICh Iu IhIS C3Sc yIc0S 3 I3IID O
8500/94500=0.09.
PS Wc Scc Iu Ihc3DDVc cX3DQc, cXISIculI3 DCCuDcuCcS OSDUc DD
/ew N OuIuuDDcI COHQM3Dc C3uDuIC3 CDuSUuCIIOuS Dy 3 3CIOI ODOIc Ih3D I0,
3u0 Wc C3u S3y Ih3I/ew I S SIIOu_y cXISIcuII3. uI SuCh 3 C3ID 3SD hI_hI_hIS
3u IDQOlI3uI IHII3IIOu O Ihc IyQc O Qu3uIII3IIVc 03I3 Wc h3Vc ODI3Iuc0 SD u, 3
IDII3IIOu Ih3I 3QQlIcS cQu3y IO 3 D IyQcS Wc h3Vc OOKc0 3I. hc Qu3uIII3IIVc
03I3Wc CO lcCIc00OcSuOI0cIcIDIuc WhcIhcIIhc 9I %OcXISIcuII3 OCCuDcuCcS O
/ew h3Vc 3u IuUIuSIC3y 0IHcIcuI Dc3uIu_ Ih3u Ihc 9% OC3uOuIC3 OCCuIIcuCcS.
II SOul 0 3QQcu Ih3I IhIS SQII CODcSQOu0S QIcCIScy IO Ihc SI3u03I0 0ISIIuCIIOu
DcIWccuQIOQOIIIOu3 3u0C3I0Iu8 uScS O/ew, IhcD U3I WOul0 IHQy Ih3IC3I0Iu3
uScS O]ew Iu lOC3IIVc COQu3 COuSUuCIIOuS V3SIy DuIuuDDcIQIOQOIIIOu3 uScS.
1OWcVcI, SI3u03I0 ScH3uIIC 3CCOuuIS D Ihc 0cDuIIcucSScHcCI Su__cSIIh3I WhIc
29
J0 L8VdO8VOf, l8m8tf8HCO2 8Hd LmlQOVn8Cn
theproportiona|useshou|da|waysbesuong,thecmdina|useshou|da||owforboth
weakand strongreadings. Itwou|d fo||ow thatproportiona| uses of]ew me very
rareindeed. Thenagain,perhapsthosestmdmdtheonesareinconect.Themethod-
o|ogica|messagewewanttomakeexp|icitisthatthedatawepresentinthissection
simp|ydoesnotdetemnewhetherproportiona|usesof ]ew everoccurinexistentia|
pivotposition,mdsimi|ar|ydoesnotdetermineingenera|whetheroccunencesof
agivenNPformcm ydierentmeaningsincanonicm sub]ectandexistentialpivot
position.
J. Z. Resa/ts
Forthe studyofEng|ish NP disuibution using Goog|e, resu|ts are presentedfora
rangeofNPtypesingureI .
|
|OC| rO
o 3rd $|Dg
C
mO$L
|OC| O$
|
|JCvCQ
m|
LhC
bOL
LhISJLhCS
Oh| |OC |o
LhULhO C

ODC
$OmC
$OmC h|Dg W
I |C I N W
m
Y

l w W
|
COu|C l W
dOO
ICw

DumCr|$
|
DO
I mO$I N
0. 0! 0. ! !0 !00 !000 !0000 !00000 I000000
#canonlcal/#exstenUal [log acalej
FigureI . Eng|ishCanonica|xistentia|ratios
Resa/t I: No categor|ca/ de]n|teness e_ect
Therstthing tonote abouttheresu|ts we obtained isthatmeyprovide no di-
rectevidencefor acategorica| denitenesseect. a|| the P types we searched
forinEng|ishthatarethemse|vesreasonab|y frequentoccurwith some non-zero
frequencyinexistentia|pivotposition.
Bm Suencf. (No )caontcattfv am NPDtsfaufton tn Extsfeftats
Resu/t Z.A stmng non-categor|ca/ de]n|teness eect
Resu|tI doesnotimplythatthere isnodenitenesseect. Ontheconumy,there-
sultsshowaverystrongdenitenesseect,butitisaneectofadierentsonmm
thatusual|yobserved. Lookingatthetab|e,weseethatonmelenhandsidewith
canonica|/existentia|(C)ratiosbetween0andI, mearangeofmost|yindenite
and downwardmonotoneNPtypes. On me nghthandside, with ratios between
20and200,000areamixtureofdenitesandproportionmdetenners. Thatis,
weNPsappemontheleft,andsuongNPsontheright,withthenotionsofwe
andsuong fa|lingonstandard lines. Butmedenitenesseect, as itappemsin
ourdata, isnotthecategoricaleectnoma||ydescribed. Rather, thedeniteness
eectconsistsofamassivegu|finthespaceofL ratios. LetussaymatoneNP
i smon e|stent|a/ thananotherifitsLratioislower,andthatoneNPi sL times
asexistentia|as anotherifitsLrationis IJo asbig. ThenforEng|ish, me|east
existenua|weak is20timesasexistenua|asthemostexistentia|suongN.
Resu/t J: Ant|-de]n|teness eects
The weak NPs vary in existentia|ityby a factor ofover20. Severa| determners
M suongly existentia|,notablyat most, no, /emdnumera|s, a|lofwhich favor
existentia| pivotpositionby afactor ofover I0 (afactor of24for at most). we
descnbe such NPs as mmifesting m ant|-de]n|teness eect. Note mat three of
theseheavi|yexistentia|NPsmedownwmdmonotone.Wespecu|atethatmeirhigh
existentia|ityre|atestothefactthatunlikeupwardmonotonecmdina|determners
they |ack generic or specic readings and do not ordinmi|y introduce discourse
referents. Asregards numerals, we suggestthatthehighdegreeofexistentia|ity
wefoundmayre|ate totheirlackinggenericreadings andtendingnottobegiven
specicinteqretations.
s
Although simple indenites headedbya(n) are oftenconsideredprototyp-
ica| examp|es ofdeterminers occuning in exisnti
a
l pivot position, in tens of
L ratiosthey menotatallprototypical. Infactsimp|eindenites (wim generic,
specic andotherreadings|umped togemer) occur at approximately equa| rates
incanonica| sub]ectpositionandexistentia|pivotposition. Witha L ratio ap-
proaching I in the Google data, simp|e indenites cannotD said toexmbit any
anti-denitenesseect(butseebe|owforadiscussionofgemevariationfora(n) .
WedenedmostNPfomsusing meheaddetennerasamatchngcrite-
rion,butthebmchart a|so includes theNPa door, whichtums outtobesuong|y
existentia|, favoring pivotpositionby a factor of I 3. I . Twofactors me relevant.
First, a door oen refers to an ob]ectwhich is partofsome otherelementmen-
tionedinthe sentence, e. g. roomorawal|. Thus,a door refers to adependent
e|ement,suchthatifyoumovethethingitdependson,thedoormovestoo, while
the reversedoes not hold. Subjects, we suggest, me preferab|y independent, so
thatneithertheirpositionnorexistenceinuinsica|lyreliesonmatofotherob]ects
refend to inthe same clause. A secondrelevantfactor isanimacy. universal|y,
sub]ectstendtobeanimate,soitmaybethatinanimatesaredisfavomdinsub]ect
3 I
JZ L8VdO8VO, l8m8t8HCO2 8Hd LmlOVH8CH
positionandtendtobemoreexistentia|thansimilmanimates. However,wedonot
yethavequantitativedatatoindicatewhetherthi sspeculativegenera|izationho|ds.
Whateverthe factorsthat cause vmous NPs to shrinkfromcanonica| sub]ectpo-
sitionincopularconsuuctions,thedata showthatwhileproportional mddenite
NPssuong|yfavorcanonica|sub]ectpositionoverpivotposition,merearesoNPs
whichsuonglyfavorexistentia|pivotposition.
Resu/t 4. Sub-order|ngs among de]n|tes
ThesuongNPsvmyinexistentia|itybyafactorofover, mdthereissystem-
aticitytothedisuibution.Weueatedrstorsecondperson(i. e. |ocal)pronounsas
aseparateNtypefrommirdpersonpronouns,inordertoseewhetherthereisany
dierenti eectintheirexistentialiq, although bom types me ready known to
vastlyprefercanonic overexistential pivotposition. Whatwefoundisthatloca|
pronounsare|essexistentithannon-locmpronouns,theexpectedresu|tgiventhat
cross-linguistical|yloca|pronounsareprototypical sub]ects. Demonsuauveswere
foundtobemoreexistentia|thanregularpronouns,alsoin|inewiththeirbeingless
cmonical sub]ects, and dista| demonsuatives memore existentia| thm proximal
demonsuatives,whichisonceagaintheexpectedresu|t.
a
Resu/t J. Ec/us|ves contr|bute to ex|stent|a/|g
ItisknownthatNPsheadedby exc|us|ves suchason/y occur in existentia| pivot
position, but not known whether this is to beatuibuted to the eect ofon/y or
to the independent licensinginexistentia| pivot position ofthe mgumentofon/y
see Beaver(Z4). Tocheckwhetherde exc|usive on/y ects|icensing, we
includedanNPfomconsistingofon/y followedbya|ocpronoun.Whatwend
isthatthepresenceofon/y makesmeNPsubstantiallymoreexistentia|,byafactor
ofapproximate|y l ,. With the caveatthattheremeprobab|y sepmateeects
wherebyunsuessedNPsaremoreexistentia|thanunsuessedNPs,andcomp|exNPs
moreexistentia|thmsimp|eNPs,weconcludethaton/y hasaneectinmakingNPs
moreexistentia|.
Resu/t . Genre strong/y aects use o/e|stent|a/s
Our studies using suucturedcoqora me sti|| inprogress, butpre|imnary resu|ts
areshowningureZ. Duetothesmallersizeofthesuucturedcorporarelativeto
thewebcoqus,wewere|imitedinthetypeofexpressionswecou|dmemingfully
investigate. Nonethe less, ourdatacoversalmostthesamerangeofcategoriesas
thewebbasedstudy.
ia
Thebroad trends in the websurey merepeated for a|| three suuctured
corpora. pronounsmethe leastexistentia|,downwardmonotoneNPsmethemost
existentia| andshow aclem anti-denitenesseect, anddenites mdprototypical
indenitesfallinbetweenwithdeniteshavingconsiderablyhigherCratiosthan
indenites. Thebroadsimilaritybetweenthewebresultsandthestructuredcoqus
Bo Suencf. on-)cnontcttfv no Dtsteufto Extsnftm
resu|tsprovides evidence thatdespitethec|earmethodo|ogica|prob|emsinvo|ved
in usinganintemetsearchengineforgatheringquantitativedata, mewebresu|ts
areinfact|inguistica||ysignicant.
Thereisawea|thoffunherinformationthatcanbedrawnoutofgureZ,
butforreasonsofspacewe|imitourse|vesto]ustonemrtherobservation. Consider
whatisprobab|ythemostvisua||y sa|ientproperqofmegure. meWSJ graph is
unifomyhigherthanmeothertwo. Inotherwords,wters(mdeditors)ofnews-
papertextuniformyavoidexistentia|consuuctionsre|ativetowhati sfoundinthe
broadmixofgemesintheBrowncorpus,andre|ativetowhatisfoundinthetypeof
everydayspeechcontmnedinSwitchbomd.Forexmp|e, newspapertextmepro-
totypica|indenitea(n) isaboutmreetimesd |ike|ytooccurincanonica|position
asinexistenti
a
|pivotposition. Foramixedcorpus,a(n) iss|ighHymore|ike|yto
appearinexistentia|pivotposition(Cratio .), mdforeveryday speecha(n)
isZ.5 umesas|ike|ytoappearinexistentia|positionascanonica|sub] ect position.
Thisisasuikingexamp|eofgenrevmiationwhich,toourknow|edge,hasnotpre-
vious|ybeennoticed. We specu|atethatthedismbutionofindenitesmghtre|ate
inparttothedesireofnewspaperwriterstoappemauthoritativeabouttheinfona-
tiontheypresent,aneectwhichmaybeobtainedifindenitesaregivenaspecic
reading. Ifthisreadingismoreeasi|yobtainedincanonicmsub]ectposition,pivot
positionwi||bedisfavoredforindenites.However,theeectmnsacrosstheboard
ofNPtypes(thoughitispmticu|ar|ystrongforindenites),sotheremustbeother
factors, e. g. a desire tokeep|engthdown, md soavoidexp|etives thatom an
editor'spointofviewmightappearunnecessmy.
O 5w|uhbrd ~mn ~ WS
FigureZ. ComparisonofL ratiosbyGenre
JJ
34 L8VdO8VOf,l8m8ff8HCO28HOLmOVn8Cn
Resu/t /. Dutch NP d|str|but|on g/oba//y m|rmrs Eng/|sh
TheDutchexistentialconstructionhasasimilarformtotheEnglishexistential ,with
aniniti alexpletiveer ' there' fol|owedbyverbandthennon-canonic
a
lsub]ect.The
range ofverbs thatcan occurinthis consuucuon ismuch wider in Dutch thmin
Engl ish, mthough, asforEnglish,occurrencesofthe standardcopu|mzqn ' tobe'
areprototypi cal. WeperformedastudyofDutchexistentialsusingGoogle,exam-
iningonlycaseswiththe standmdcopulmd mainverb. Teresultsmepresented
inthechartinhgureJ. Atthetimeweconductedthestudy(mid-20)themount
ofDutchintheGoog|edatabasewasabout5%oftheamountofEng|ish. Thuswe
weremoreconstrmnedforDuuhmanforEng|ishinourabilitytoconductsearches
forinequentpattems,andwedidnotattempttomllyreplicatetherangeofEnglish
pattemswesemchedfor.
i i
OIh [bC|dC]
M M
|OC| prO [| ]CJ]]Jm|]]
M M
|OC| pOSS [m|]DJ]Ou . . . ]
M M
CCh [C| kC]
M M
Ih|SJIhe [d|tJdC2C]
N N
mOSI dC mCCSIC]
M M
nmCS (a phCCr ", . .]
M N
0 [ C]
IhOSC [d d| C]
IhC | [dC]
IhC N S|Dg [ h I]
C [dC]
SOmC | [SO Dm| gC]
QI |CSt N [m|DSICDS]
JODC [CCD]
Cw [CDkC|C]
Dy [vCC| ] 8
ICw CCD pr] @
t mOSt N [D xm| ] Q
ICw [w| D|g] Q
Dum pr8| S @
smCth|ng [| e
r[Qeen
0. 00I 0. 0I 0. I I0 I00 I000 I 0000 I00000
# canoncal/#exstental [log scale]
FigureJ. DutchCanonical/Existentia|ratios
The results for Dutch mebroadly similarto thosefor English. First, there is a
cleardennitenesseectconsistingofalargegapinhequency spacebetweenNPs
that would staadardlybeclassiedas suong, mdthosethatwou|dbec|assiedas
wem, andw|thinthe strongNPs. Second, withinthestrongNPs, denite NPsare
considerablymoreexistentialthanpronounsandNPsheadedbyproportionaldeter-
miners. Third,prototypic indennites such aseen 'a' (whichis not distinguished
inoardatahomthesuessednon-schwavarimtnone)arefarombeingpro-
totypical asexistentialNPs, andhaveaLratioclcsetopmty.Theprototypically
existential NPsinc|udethoseheadedby downwmdmonotonedetennersd well
asthoseheadedbynumerals,andal|oftheseexhibitaclemanti-denitenesseect.
AD bU8. [PON-)CANON!CA!1Y AND NllS1Rl8U1!ON !N LXlS1EN1IAL
Resu/t . Dutcb Sommige 'some ' behaves as q stong
ltis interestingtonote thattheDutch indenite somm/ge ' some' (hereconsidered
asadeteminer,thoughitcana|sooccurasafu||NP)shouldc|ear|ybec|assiedas
asuongNl.Thisismoma|ousfromthepointofviewofstandardsemantictheories
ofthesuong/weakdistinction.However,theresultsweobtainedmirrortheana|ysis
ofdeJong( I 983)anddeHoop( I 995)whoshowthatsomm/ge ' some'behavesasa
suongdeterminer.Wetakethisconuenceofconclusionsasprovidingsupportfor
ourmethodo|ogy.
J.J. D/scuss/on: De Order o/ NPs
Sincewe found bothdenitenesseectsandanti-denitenesseects, we maysay
thatourqumtitativestudiesofEng|ishandDutchshowthefo||owing. tbe de]n/te-
ness eect /s a two s/ded co/n. We cansummarizethemainresultsintensofthe
fo||owingsca|e.
(27) ||DDDUDS, |DD|| DDB| N|S > D| D| DSC|| | DDS > |OD~
y|CB| | DD| D|S > RDD N|S
Bymto/ca/ /nde]n/tes we mean whichever indenites me mostcom-
mon ina|anguage, thoughfor |anguagesotherthan thosereportedinthis section
thereisasepmaissueofwhetherindenitenessisstandmd|ymmkedinmat|an-
guageat
a
||. ThusDutch somm/ge ' some' isnotaprototypica|indenite since it
hasmuch|owerequencythaneen 'one' , andEng|ishnumera|smealsonotpro-
totypic, having much |ower frequency than a(n) and |ess uses(e. g. no generic
use).
Somemrthergeneralizationscana|sobeseenintermsofsca|es. Asregmds
pronouns, for exmp|e, the ordering we obtained is what wou|d be expected on
the basis ofananysisbasedonthe sub]ectprop accountwedescribed. | OCB|
|D > DDD-| DCB| |D. Simi|m|y,formostofthe smdiesweperfomed weobtain
thenatura|orderingamongdemonstratives|DX| 0B| > D|SB| . However, it is in-
teresungtonote thatforthe Switchboard corpusthisorderingwas reversed. this
ispresumab|ybecause ofaprominentuse ofEng|ishproxim
a
| demonstratives to
presentnewdiscoursereferents. there 's th|s guy kow . . . Fina||y,forthesmdy
basedon the lmgestcorpus (i . e. the Eng|ishweb study) wehave |D > |DD|~
| DDB| N|, almough forotherstudieswe|ackedsumcientdatatosubstantiate this
observation.Ignoringthesubordcnngofdemonstratives, wesuggestthefol|owing
morenegrainedsca|ethanthatin(27).
(28) LDCB| |D > DDD| DCB| |D > |DD||DDB| N|S > D| D| D~
SC||| DDS, D0DDS|B|VS > |DDy|CB| | DD| D|S > RDD
N|S
35
36 L0VdO8VOt, l0m0ff0HCO2 0HdLmlfyOVn8Cn
4. dentifying Deniteness Effects: To Case Studies from Russian
Wenow discusstwo consuuctions whichhavebeen ama] or concem in work on
Russian syntax. The rst, Sub]ectGenitive ofNegation, istypica||y not seen as
manifestingadenitenesseect. We showthatthisconsuuctiondoesexhibitdef-
initeness eects, a|though the permissiveness of the consuuction re|ative to e. g.
theEng|ishexistentia|meansthatthewecanobservetheeectsnotby|ookingat
whichNPsmeforbidden,butatwhichNPsareprefened.Thatis,whatweactum|y
seememti-denitenesseects. Thesecondconstructionistheopuona|copu|ain
presenttense|ocatives/possessives,whichhasbeenc|aimed e|sewhere to showme
deniteness eect. Using quantitative data, we show that the eectin this con-
suuctionissomewhatdierenthomthedenitenesseectsobseredinmeomer
consuuctionswehaveconsidered.
4. I. Sub]ect Gen/t/ve o/ Negat/on. An Ant/-de/ n/teness Eect
Thechoiceofthecanonicalortheexistentia|fomfornegativesentencesinRussim
hasbeendiscussedduetothecasemarking oftheNPsin suchsentences. Whi|e
thesub]ectofthecanonicalsentenceshastheNominative case, mesubectofme
negatedexistentia| hastheGenitivecase(thiscasemarking,knownasGenitiveof
Negation,isoptiona|fordirectob]ectsinnegatedsentences). Inthepasttense,me
copu|ainthe canonica| sentences isin agreement with the sub]ect( I ), while the
copu|aintheexistentia|sentences(2)isalwaysinneutersingu|m(exmp|esfrom
(ParteeandBorschev,2002)).
(I) Nikto tam ne by|.
Noone. Noutherenegwas.u. so.
' Noonewasthere.'
(2) Nikogo tam ne bylo.
noone. osNtherenegwas. n. so
' Noonewasthere.'
The factors responsib|e for thechoice ofthefom md the meaning dif-
ferences between the foms have been much debated. The factors pmposed to
distinguishbetween the forms me Topicocus structure (Babby, I 980), agentiv-
ity(Paduceva, I 992),andperspectiva|stmcture(Partee andBorschev200I , 2002,
2004).Itisclearfromthe|atterpublicationsthatapure|yTopicocusexp|anauon
makesinconectpredictions,andthatbothagentivityandperspectiva|structurep|ay
aro|eindeteminingthechoiceofconstruction.
Atrst, thechoiceofthecanonicalortheexistentia|constructioninthese
casesdoesnotseemconnectedto thedehnitenesseect,asdenitemdindenite
phasesalikeoccurfreelyintheexistentialconstruction. Wecheckedthefrequen-
cies ofthe variants for some nounphasesin past tense sentences
t
,

ts
. Tab|e2
containsthe hequency ofcanonicmandexistential consuuctions with thel ocative
phrasetam ' there'forthewordorderNPLocP asin (29) and(30).
Bao Suenm. (Non)canontcattf ano NlDisteuton tn Exts
(29) Lta kna tam nc bya.
Jhsbook. NOVthcrcnotwas. F
`Jhsbookwasn' tthcrc.'
(30) Ltoj kn tam nc byIo.
Jhsbook. GEN thcrcnotwas. NEUT.
'Jhcrcwasn' t thsbookthcrc.'
Nl uansIaton canon.
]a l I I ,700
mo]a * my (t.) 3
ta ths * (t.) 0
t/ * thcsc 2
taka]a such (I.) 0
tak/e * such * (pl.) 0
cxst. cxst1tot
a
I
6, I 00 34%
I 00 97%
I 00 I 00%
80 979
80 I 009
90 I 00%
JabIc2. CanoncaI andcxstcntaI consuuctonsnncauvcpast mnsc scnmnccsoI
thctomNP tam ne \.
As thc tabIc shows, most Nls rapIy occur n thc canoncaI torm. Jhs tom s
onIy uscd substantaIy wth anmatcs, whch s rcasonabIc, vcn that tms tom
snaIs acntvty. Jabc 2shows a stron ant-dchntcncss chcct, wth mostNs
occumnamostcxcIusvcyn thccxstcnuaIconsuuctions. Jhc subj cctpropcrtcs
rcgurcmcntstorthccanonca torm nthsL arc hh, andac onIy satshcdby
Nlshhon thcpromncncc scaIcs. Wc can sU arzc mc datatrom mctabIc n
tcrmsotmctoIownNlordcrn.
(3I ) | DCB| |D > | DCB| DSSSS|V > DRDDS|B|VS, | DD| D|S
Jhcordcrngn(3 I ) s asub-ordcrnotthatobscrvcdnourLnIshdata.
Jhcrctorc, wcsugcstthatdcsptcthccnormousdstrbutonadhcrcnccsbctwccn
pvotposton n !nIsh cxstcntaIs and thc Kussan Gcntvc olNcauon, both
dsubutonsbc sccn as nstanccs otadchntcncsschcct, andboth dsmbutons bc
sccn asnvovnanNloccunn n anon-canoncaconsuuctonbccausctIacks
caoncaIsubjcctpropcrtcs.
4.2. De Ot/ona/ Cou/a /n Present 1ense Possess/vesocat/ves. A De]n/teness
Eect
ln ths sccton wcconsdcraKussanconsuucton, namcy posscssvcs/Iocatvcsn
thcprcscnttcnsc, thathasbccncamcdtoshowadchntcncsschcct. Wcuscguant-
tatvcdatatoshowthatthcchcctdcpatssnhcantIyIromthccanonca/cxstcntaI
aItcmatons dscusscdabovc.

38 L8VdO8VCf, l8m8ff8HCC2 0nd LmOVn8Cu
n Kussan, postvcIocatvcandposscssvcscntcnccsnprcscnttcnsccan
appcawth or wthoutthccopuIa (Kondrasova, I 996, Chaptcr 5.2). Jhc varant
wm thc copuIa typcay has a non-cxhaustvc mcan, and thc varant wthout
has a cxhaustvc mcanng. lorcxpIc, thcLnIsh scntcncc (32) canhavctwo
mcanns. Some o/John 's teachers are good or(A//) John's teachers an good. Jhc
tomcrscnsc s cxprcsscd by thc Kussan scntcncc (33), wth thc copuIa, ad thc
Iattcrscnscscxprcsscdbyscntcncc (34),wthout thccopuIa.
(32) !ohnhasoodtcachcrs.
(33) Ozona jcst' xorosc uctcja.
at!ohn ]est ' ood tcachcrs.
` Somc ot!ohn'stcachcrsacgood.'
(34) UOzonaxoroscuctcIja.
at!ohn good tcachcrs.
` (AII)!ohn'stcachcrsarcood.'
Kondrasova ( I 996, 2. 4 4)camsthatthcconsuucuons wmthc copuIa]est ' show
thcdchntcncsschcct.lnourtcmnoogy,thcvaratwmthccopuIasthccxstcn-
taI, ad thc onc wthoutstscanonca aItcmatvc. HcrcxapIcs andj udmcnts
ac.
Stronguanthcrsvs. wcaguathcrs.
(35) *V Moskvc jcst' maomIconcrov lvscInvsty.
n Moscow]est ' Icw poIccmcn. GNlaII Inusts. NOM
`1hcrc arc Icw poccmcn/aII thc Inusts n Moscow.' ( hcr
(40)
(36) V Moskvc jcst' ncmaomIconcrov lncskoI ' koparkov.
nMoscow]est ' not. Icwpoccmcn. GNl scvcraI paks. GEN
`JhcrcarcqutcaIcwpoIccmcn/scvcraIpaksnMoscow.' (" hcr
(4I )
Ochntcvs. ndchntcdcscrptons.
(37) *V Moskvc jcst' tot ccIovck,kotoryj kupI katnu z Lrmtaza.
n Moscow]es| ' thatpcrson who bouht pantnhomHcrmtagc
` n Moscow thcrc s thc pcrson who bouht a pantn hom thc
Hcmtagc.' (=hcr(42a)
(38) V Moskvc jcst` takcIjud , kotorycpokupajutkartny z Lrmtaza.
nMoscow]es| ' such pcopIcwho buy pantngshomHcrmtac
` ln Moscow thcrc arc pcopIc who buy pantngs hom thc Hcr-
mtagc.' (=hcr(43a)
Bao Suencf. (Non)canontcttfx ano N Disfmufion Exisims
JhcrcactwoprobIcmswmKondasova' sdcscrpuon. rst, accordntothccIas-
shcaton ot MIsark (I977), ma/o s not a suon N. 5uon Ns nvoIvc gua-
thcatonovcra doman, whIcwcakoncs arc numbcr dctcmncrs or cadnaIty
words'. AccordntothscIasshcaton,ma/o ntscadnaIntcqrctauonsawcak
N,sotshouIdbctcctousnancxstcntaI. JhcsamcprcdctonwouIdbcmadc
byKcnan(2003),asma/o swhathcumsconscrvatvconmcsccondaumcnL'
NcvcrthcIcss,ma/o snotuscdwththccopuIacvcnnthccadnaIntcrprctaton.
5ccond,thcscarchcswcmadc showcdthataIthouhthccxstcntaIratoot
m/o ` Icw'vs. nemnogo ` somc' andmnogo ` much' sndccdsubstanuaI,mcdhcr-
cnccbctwccn tot koto] `thc that' (m.) andtako kto 'a that' (m. ) s not
asIagc,andthcrcacmanycxampIcswthandwthoutmccopuIatorcachotthcsc
consuuctons- scchurc4.JhssdhcrcnthommcchcctobscrvcdnLnlsh,
as thc dhcrcncc nthcratobctwccnthcdchntc and ndchntc Nls n LnIshs
muchhhcr. Whatsmorcsukn, thc wordma/o has amuch hhcrL rauo
than thc ndchntcs, whIc thc word `Icw' n LnIsh has a rauo smIa to o|hcr
ndchntcs. AIthouhmorccxtrcmc, thc uncxpcc0dbchavorotma/o srcmns-
ccntotmcOutch somm/ge `somc' (scc sccuon 3), whch aIsobchavcs as asuon
NldcsptcdcnotnacardnaIquanthcr.urthcrnvcstatonotmcscmantcsot
ma/o srcgurcdnordcrto undcrstandthsbchavor.

m `k "
m
m " `q {m.}

k m`I M{m)
Io kkJ6 `mp
@
_...,
!
nem m "
meNmm
m m`a 0at g.}
emn
@,. {f.
"!!
m _.. * mat
0. 1 1 10 1 00 1 00
maea|ca| twistaatia|j|eqs|aj
urc4.LsacotcopuIan KussanprcscnttcnscIocatvcs/posscssvcs.
JhcdhcrcnccbctwccnthcratosoIthcquanthcrsnthctabIc,cxccptma/o, snot
Iac, andsnotcnouhtocstabIshwhcthcrmc dchntcncsschcctcxstsasnthc
constructonsdscusscdnsccton3. JhcthcorctcaIrcasontor thsmaybcthat, n
mscasc,prcscnccor abscncc oI]est ' hasonIy amnormpacton thc subjccthood
oIthcNl. WhcnthccopuIasprcscnt,thcNlstamcrhomtscanoncaIIocaton
n mc bcnnn ot thc scntcncc than whcn mc copuIa sabscnt. Jhc torm and
arccmcntpropcrtcsotthcNlarcthcsamc. Jhsmay bcmcrcasontorthc smaIcr
dhcrcnccsnmccanonca/cxstcntaIrato.
39
4
L8Vd O8VO, l8m8f8HCO20HdLmlfyOVH8CH
. ConcIusion
Wc havc rcvcacd a compcx dsubuton otNl typcs n cxstcntaI constructons
nvoIvn both dchntcncss and at-dchntcncss chccts. Wc havc shown, usn
guanttatvc data, matthc dchntcncss chcct s not catconcaIy prcdctabIc hom
N lorm. Jhc tcndcncy otdhcrcnt kndsotNls to appcarncxstcntaI, as op-
poscdtocaoncaI,constructons,vmcswdcIy,butthcrc acmaypossbIcIcvcIs
bctwccnmccxucmcs. JhccxpIanatontorthc dsubutonaI propcrtcs otthc Nls
svcnntcmsotthcrsubjcctpropcrtcs. AnNlcanbcrcaIzcdasancxstcntaI
pvotwhcntIackspropcncsthatwouIdmakctaoodsubjcct,soNlsthatposscss
morc subjcct-Ikc propcrtcs tcndto occurmorchcgucntIy n canonca consuuc-
tons.
Wchavcaucdthatths anaIyss mmcs concctprcdctons tor a numbcr
oldsubutonaI ssucs n Ouuh, EnIsh, Hcbrcw and Kussan, and wchavc ds-
cusscdsomcchaIIcnncxampIcsoIcross-Inustcvaraton,cascsnwhchthc
racotcxstcntaIpvotNltypcssdhcrcntnoncanuachomthatnanomcr.
Howcvcr, whcnthcdstrbutondcpcndsonprcscncc/abscnccotsubjcctpropcncs,
tssystcmatcaIyInkcdtoundcrynordcrnsotNls, suchas.
LDCB| |D > DDD| DCB| |D > |DD||DDB| N|S > D| D| DSC|| | DDS,
DRDDS|B|VS > |DDy|CB| |DD|D|S > RDD N|S,
N|S W|R DKD | D||B|DDS > N|S W|R y | D||B| DDS.
ltwc acrht, thcn thc ordcrns wchavctound shouIdbcunvcrsa . naycon-
suuctonnvoIvnnon-canoncarcaIzatonotasubjcct,andwhch sncompct-
tonwthacanoncaI subjcctconstmcuon,thcdrcctonaItyotth sordcrnwIbc
prcscrvcd. Jhs hypothcss s sucstvc otmturccross-Inu stcrcscarch. morc
Ianuacs, andmorcconstructons. Wc bcIcvcthatourguanttatvc,corpusbascd
mcthodooy bascd on thc usc oIcanoncaI/cxstcntaI ratos can bc appIcdhut-
mIIy to othcr Ianuacs, andmay bccxtcndcdto othcrconstructonstorwhch a
compctton-bascdanayssspausbIc.
Endnotes
Jhs papcr sbascd on rcscarch conductcdby ajont StantordC BcrkcIcy rc-
scach roup run by Oavd Bcavcr and Lnc MkkcIscn, addtonaI mcmbcrs d
GcrIot Bouma, AIcx Bratkcvch, lvn Gaca ^Ivarcz, Ioran 1accr and Laura
Whtton. Wc bcnchtcd Irom cxtcnsvc dscussons wth Grcory Wad and trom
commcntsotaudcnccsatNcw`orklnvcrsty,NorthwcstcmlnvcrstyandStan-
tordLnvcrsty,thc KNAW Acadcmy Cooguum Cogn/t/ve Foundat/ons o/Inter-
retat/on, andSALJA. SpccaIthanksto1udthJonhauscrtorvaIuabIccomcnts
onthcpapcrtscII.
!
Jhrouhoutthspapcrwcusc torcIcrtoaII nomnaI phrascs,crcumvcntn
thcNl/Oldstncton, whch s ncIcvanttothc ssucs dscusscdhcrc.
Bao Sumcf. (Non-)canontcat|f am NlDtsteuton PExtsf
)
Lambrccht(2000) dscusscsthcobjcct-Ikcpropcrtcsotpvotscross-InustcaIy.
3
Scc c. . WooItord (2005), who shows that IocaI pronouns arc Icss markcd mcn
non-IocaIpronounstorcascmarknpurposcs.
4
1hc suucturcd corpus scarchcs uscd thcutIqtgre2 on lcnn 1rccbank corpora
prc-pascdusnthcJrccbank2 annotatonudcIncs. Godhcyct
a
I . ( I 992), Ma-
cusctaI . ( I 994) .
5
1hc scarchcs wcrc ca cdout ovcr apcrod oIatcw days nAuust 200. Uoth
thc wcb and scachcnnctcchnoIoycvoIvcconstanuy, so rcpcatn thc tcstsnow
wouIdycIdnumcrcaIIydstnctrcsuIts.Wc wouIdcxpcctthcrauoswccaIcuIatcto
rcmanrouhIyconstantprovdcdthcmakcupot thcwcbnvoIvcsrouhIyconstant
proporuonsotvcn cmcs, andprovdcdmctmcpcnod s smaIIcnouhtonorc
Iauac chanc. AprobIcmatc ssuc s that mc tunctonaIty otGooIc's scach
cnnchachancdscvcraItmcsn thcIastyca,attmcotwrtn, thcbchavorot
mc wIdcad opcrator * s notcIca. Omcrscarch cnncs (c. . Yahoo) aIso aIow
wIdcads,butthctunctonatyoIthcscwdcadss
a
IsouncIca.
5
ScarchcnncsordcrthcrrcsuItsnawaythatnuoduccsbasasrcardsthctypcs
ottcxtwhchtcndtobchhIyrankcdn ascarch.1hcrctorc twouIdbcadvsabIc
tosampIcarandomscIcct onotscarchrcsuItsrathcrthanc. . thchrst20. Howcvcr,
torthcLnIshGooIcdatawcsampIcdonIythchrsttcwpacs otscachrcsuIts.
7
Wc cxpcrmcntcd wth addtonaI scarchcs thatcombnc thc rcuIarscarch wth
a ncatvc cntcron usn GooIc' s tcaturc to bIock unwantcd rcsuIts Ikc u
]ew. Howcvcr, ths ntroduccs compIcatons, hrstbccausc GooIc' scountrcsuIts
may not bc consstcnt across scachcs wth and wthout ncauvc scach crtcna,
andsccondbccausc suchscarchcsbIockIctmatcscarchrcsuItsncascthcscach
pacs contan an nstancc oIthc bIockcdcxprcsson cIscwhcrc. Jhcrctorc wcdd
notuscncatvccrtcrananyoIourscachcs,andnsmadrcIcdsoIcIyonsampIn
topckuponcnoncous data.
3
NumcraIs M thc onccIass otwcak dctcrmncrs tor whch our wcb rcsuItscrc
notdupIcatcdnstructurcdcorpora,sothcrhhcxstcntaItynourwcbdatamay
rcsuIthoman anomaIy nourscarchproccdurcs.
9
Wc ddnot ornaIy makc a scparatc scachtor but a scarch at tmc otwrt-
n ndcatcs that it s cvcn Icss cxstcntaI than IocaI pronouns. 1hs chcct runs
aanst thc ran oIunvcrsaI hcrarchcs nwhch thrdpcrson s Icssprotogpcal
asasubj cctthanhrstand sccond, andnamatcNlsac IcssprototypcaIthanan-
matc Nls. Howcvcr, it s an nunscaIIy wcak pronoun (t' s suon countcrpart,
aruabIy, bcn th/s), so whcn wc comparc it wth othcrpronouns thcrc s a con-
tound. wc ac comparn wcak occuncnccs otonc pronoun wth both wcak and
suon occuncnccsoIothcrs.
J0
n hurc 2, Nlsarcordcrcd accordnto thcrCLratonthc WS! corpus. or
ths rcasont appcas n thc raph thatn somc scnscNls n thcWS!corpushavca
smoothmonotoncdstrbutonwhIcthc dsubuton otNlsnthcothcr corporas
morcjumpy. Jhssani IIuson crcatcdbythcchoccoINlordcrn.
J
!
JhcOutch scarchcs wcrcsampIcdnthcscway asthcLn shscachcs. How-
cvcr, thc pcrson whopcHomcd thc sampIn (Bcavcr) s not a natvc spcakcr ot
Outch. ldcaIythcscarchcs wouIdbcrcpcaudusnanatvcspcakcr. AnaddtonaI
4I
42 David Bcavcr, l8m8ff8HCOZ 8Hd Dmiuy Lcvinson
compcatonsthatnOutchthccrtcnaIordctcrmnnwhcthcra occurrcncc oI
cr `thcrc' s an cxpIctvc or a ocatvc s cvcn morc vcxcd than tor LnIsh. or
cxampc, occuncnccs ot cr `thcrc' n mmcdatcIy post-vcrbal poston ac morc
common n Outch than n Lnsh and somc olthcsc occuncnccs may nvovc an
cxpctvc. Sncc wctac non-canoncaI poston (or othcr makn) otmc subjcct
asthccrucaIssuc, wcddnotcountsuchtokcnswthcxstcntaIs.
( )
Somc dscusson otthchcgucnccsand thcmcanns otmcvaats stoundn
lartcc andBorschcv(2002). Jhc dchntcncsschccts notmcntoncdmcn.
( 3
Wc chcckcd thc scntcnccs nthc ncgatvc past tcnsc tomtortcchncaI rcasons.
nthslorm thc caoncaI/cxstcntaI varantsdhcr n a way thatmakcs schn
practca .
References
Aisscn, !udith. ! 999, "Markcdncss and Subjcct Choicc in OptimaIty Thcory", uluru
!unguug8 und !nguSltC 18Oj 17, l ~39.
Babby, Lconard H. . ! 9&O, LtSl8nltu cnlcnccsund 8gulOn n uSSun, Karoma Pub-
lishcrs, Ann Arbor.
Barwisc, !on and Robin Coopcr. ! 9& ! , `Gcncralizcd Quantcrs and Natural Languagc` ,
!nguSlCS und hOSOQh) 4, l 592l 9.
Bcavcr, Oavid. 2OO4, 'Fvc ' Only` Picccs` , Jh8Or8ltCu LnguulC$ ,4564.
Bcntlcy, Oclia. 2OO2, 'Ocnitcncss Eccts. Evidcncc hom Sadman' , Ms. , Univcnsty of
Manchcstcr.
Borschcv, Vladimir and Barbara Partcc. 2Ol , 'Thc Russan Gcntvc of Ncgaton in
Existcntials Scntcnccs. Thc RoIc of Thcmc-Rhcmc Structurc Rcconsdcrcd` , ln
Hajcov, Eva, Pctr Sgall, !in Hana, and Toms Hoskovcc (cds. ), 1DvuuX d8 Lr-
cle!nguSlgu8 d8 rugu8, V 4, l &5~25O, !ohn Bcnjans, Amstcrsdam.
Clark, Evc. ! 978, 'LoationaIs. Existcntia|, Locativc, and Posscssvc Constructons' , ln
Grccnbcrg, !oscph H., Charlcs A. Fcrguson, and Edith A. Moravcsk (cds. ), Ln
v8rSuS O]umun Lnmc, VO. 4, &5 l 26, Stanford Unvcrsty Prcss, Stanford,
CA.
Frcczc, Ray. ! 992, 'ExistcntiaIs and Othcr Locativcs` , Lnguug8 , 55395.
Gvon, TaImy. ! 97&, 'Ocnitcncss and RctcrcntiaIity` , ln Grccnbcrg, !oscph H. , Charlcs A.
Fcrguson, and Edith A. Moravcsik (cds. ), nv8rSuS O]umun Lnguug8, VO. 4,
29!33O, Stanford Unvcrsity Prcss, Stanford, CA.
Godhcy, !. , E. Holliman, and !. McDanicI . I 2, 'SWTCHBOARO. Tclcphonc Spcch
Corpus for Rcscarch and OcvcIopmcnt' , ln L>Z, 5 ! 752O, lEEE, San Fran-
cisco.
dc Hoop, Hclcn. ! 995, 'On thc Charactcrization of thc Wcak-Strong Dstnction' , ln
Bach, E. , E. !clinck, A. Kratzcr, and B.H. Partcc (cds. ), @uunlgCulOn tn uluru
Lunguug8S, 42! 45O, Kluwcr, Dordrccht.
dc !ong, Francisca. ! 9&3, `Ommtg8 Nict, Andcrc WcI, dc Vcrk!aring van ccn Raadsclachtig
Vcrschil ' , !L1 , 229246.
Kccnan, Edward. ! 976, 'Towards a UnivcrsaI Dchnition of 'Subjcct` , ln Li, Charlcs (cd. ),
Subjcct and Topic, Acadcmc Prcss, Ncw York.
AU bU8HC1. [PON- )CANON!CAL!T AND P L!STR!8U1!ON !N LX!S1EN1!ALS
Kcnan, Edward. 2OO3, 'Thc Ocntcncss Ehcct. Scmantics or Pragmatics?` , uluru
LngNug8 8munlCS II, l &72l 6.
Kndrasova, Natali a Yuricvna. l 996, 18 jnluXO]LSl8nlu @uunlQCulOn, Ph. O. thcsis,
Univcrsity of Wisconsin, Madison.
Lambrccht, Kud. 2OO0, `Whcn Subjccts Bchavc Lkc Objccts. An Analysis of thc Mcrg-
ing of S and in Scntcncc Focus Constructons Across Languagcs` , lNd8S n
Lnguug8 Z4, 6l l 6&2.
Marcus, MitchclI P. , Gracc Kim, Mary Ann Marcnkcwcz, Robcrt Maclntyrc, Ann Bics,
Mark Fcrguson, Karcn Katz, and Bntta Schasbcrgcr. ! 994, 'Thc Pcnn Trccbank.
Annotating Prcdicatc Argumcnt Structurc' , ) umun Lnguug8 18ChnOOg)
WOrkShOj, I I 4l I9, Morgan Kaufmann, Plansboro, N!.
McNally, Loui sc. l 99&, `Existcntial Scntcnccs wth Exstcntal Quantication` , LnguSlCS
und hOSOQh) Z1, 353392.
Mikkclscn, Linc. 2OO2, `Rcanalyzing thc Ocntcncss Ecct. Evidcncc from Oanish' , ln
Workng Papcrs in Scandinavan Syntax, volumc 69.
Milsark, Gary. l 9J7, `Toward an Explanaton of Ccrtan Pcculiaritics of mc Exstcntial
Constructon in EngIish` , LnguSlC nujSS J, l 29.
Moro, Andrca. l 994, 18 uSng O]r8dCul8S, Cambrdgc Univcrsity Prcss.
Paduccva, Elcna. l 992, ` LScmantccskom Podxodc k Sntaksisu i Gcnitivnom Sub' cktc
Glagola )l , uSSun LnguSlCS I4, 5363.
Partcc, Barbara H. and Vladimir Borschcv. 2002, 'Gcntvc of Ncgation and Scopc of Ncga-
tion in Russian Existcntial Scntcnccs' , ln POOu jjnuCh8S lO uvC LnguSlCS
J.
Partcc, Barbara H. and Vladimr Borschcv. 2O, 'Tc Scmantics of Russian Gcniuvc of
Ncgation. Thc Naturc and Rolc of PcrspcctvaI Structurc` , nC88dngS O]8mun-
lCS und LnguSlC 18OQ I4.
Rcinhart, Tanya. 2OO5, 'Topics and thc Conccptual lntcrfacc` , In Kamp, Hans and Barbara
Partcc (cds. ), LOnlCl, EIscvcr.
Slvcrstcin, Michacl. l 976, `Hicrarchy of Fcaturcs and Ergativity' , ln Oixon, Richard (cd.),
rummulCu Lul8gOr8S n uSlruun Lnguug8S, Australian lnstitutc of Aborigi-
nal Studics.
Ward, Grcgory and Bctty Bimcr. ! 995, 'Ocntcncss and thc EngIish Existcnti al ` , Lun-
guug8 I7, J22J42.
Woolford, Ellcn. 2OO5, `Oihcrcntal Subjcct Markng at Argumcnt Structurc, Syntax, and
PF` , Ms. , Univcrsity of Massachusctts. To appar n Hclcn dc Hoop and Pctcr dc
Swart cds. , Dicrcntial Subjcct Markng. SNLLT scrcs, Kluwcr Acadcmc Pub-
li shcrs.
Ziv, Yacl. l 9&2, `Anothcr Look at Ocntcs n Exstcntials ` , LnguSlCS 1, 73 &&.
Zucchi, Alcssandro. l 995, 'Thc lngrcdcnts of Ocntcncss and thc Ocnitcncss Ehcct` ,
ulNru Lunguug8 8munlCS J, 337&.
+

Potrebbero piacerti anche