Tracking Jespersen’sCycle

Paul Kiparsky and Cleo CondoravdiStanfordUniversity, andPARC andStanfordUniversity

Wedescribefour successiveroundsof Jespersen’scyclein Greekandanalyzetheprocessastheiterationof asemanticallydrivenchainshift. Thecontrastbetweenplainandemphaticnegationis aneasilylostyetnecessarypartof language,hencesubjectto repeatedrenewalby morphosyntacticand/orlexicalmeans.

Keywords: negation,grammaticalization,Greekdialects,historicalsyn-tax,syntacticchange.

1 Trajectories of negation

1.1 Structuralinvarianceandlexicalvariation

Certainstructuralpropertiesof negationin Greekhavebeenstableoverthreemillen-nia. All dialectsatall stagesdistinguishtwo typesof negation,EMPHATIC andPLAIN.Emphaticnegationis alwaysa bipartitestructure(possiblydiscontinuous)that con-sistsof a negativeheadplus an additionalfocusedindefiniteNP or adverb. But intheir lexical form thenegativeexpressionsvarywidely, especiallytheir focusedindef-inite component.(1) illustratesthisparadoxicalcombinationof structuralstabilityandconstantlexical innovation.It displaystheplain andemphaticversionsof ‘nothing’,‘not any’ of themodernCretandialectandthreeof its antecedentstages.

(1) PLAIN EMPHATIC

(I) AncientGreek o>u . . . ti o>u-de . . . én

(II) EarlyMedievalGreek (oÎ)dèn . . . ti dèn . . . tÐ-pote

(III) Greekdialects dèn . . . tÐpote dèn . . . k�n tÐpotedèn . . . prama. . .

(IV) Cretan dèn . . . prama dèn . . . dros�dèn . . . �pantoq . . .

The negationsystemof otherstagesanddialectsof the languageis built the sameway. What accountsfor this ubiquitouspairing of negators?What causesthe high

rateof lexical replacementin this domain?And how canthetwo bereconciled?Theanswerto all thesequestionslies in the semanticgroundingof the processknownas JESPERSEN’ S CYCLE. A first clue to the answercomesfrom the natureof thesynchronicanddiachronicrelationsbetweenthetwo typesof negation.

1.2 Thetypologyof negativeexpressions

Emphaticnegationalwayscontainsa focusedindefiniteexpressionwhich is drawnfrom a relativelysmall stockof itemswith a characteristicrangeof meanings.It iseithera MINIMIZER (Horn 1989:452,Krifka 1995)or a GENERALIZER; eachcanbeeithernominalor adverbial.

A nominal minimizer denotesa negligible number,amount,or part of some-thing,e.g.ClassicalGreekoÎ dè én “not evenone”,ModernGreekdialectal(dè . . . )dros(i)� “(not even)a dewdrop”,gouli� “a sip”, triq�ri “a hair”, roujoÔni “a nos-tril”, klwnÐ “a twig”. It strengthensthe forceof the negationQUANTITATIVELY bymakingit stricter. In stating“I did not drink (even)a drop”, “I did not find (somuchas) a twig” a speakerextendsthe negationevento the most insignificantamounts,which on the ordinarylenient interpretationof a negationmight be exemptfrom it.Correspondingly, anadverbialminimizeris a degreeadverbmeaning“not evento thesmallestdegree”,e.g. the slightestbit. It likewise strengthensthe forceof negationquantitativelyby makingit stricter.

A nominalgeneralizerdenotesa maximallygeneraltypeor class,andstrengthensthe negationQUALITATIVELY , by extendingits scopeto includeeverythingin thatmaximalsortaldomain(“nothing of anykind”, “nobodywhatsoever”,“not in a mil-lion years”,“not ever”). Typical examplesareMedievalGreekdèn . . . tÐ-pote “noth-ing whatever”andModernGreekdialectaldè . . . prama “not a thing”. An adverbialgeneralizeris normallya manneradverbmeaning“in anywaywhatsoever”.

Quantitativeandqualitativestrengtheningcanevenbe combined,asin the Pon-tic/Cappadociantype(Neg). . .enase‘not onething’, i.e. ‘not evenoneitem[theleastnumber— quantitativestrengthening]of anysortwhatsoever[qualitativestrengthen-ing]’.

A nominalminimizercanbeextendedto awidersortaldomain;atthemaximalex-tensionit canbecomeadegreeadverb.Thesemanticdevelopmentis “minimal piece”> “minimal quantity”> “minimal degree”.This developmenthasmadeadverbsoutof Englishabit andtheirGreekcounterpartssuchasklwnÐ ‘twig’ andyÐqalo ‘crumb’.

(2) Nominalminimizergeneralized

a. dànnot

êqoumehave

klwnÈa twig

nerìwater

‘we don’thaveadropof water’(literally, ‘a twig of water’)(Kea,Salvanos1918)

2

b. dànnot

êqoumehave

klwnÈa twig

(ywmÐ)bread

‘we don’t havea crumbof bread’(literally, ‘a twig of bread’)(ibid.)

(3) Final stage:nominalminimizerturnedinto degreeadverb

a. dànnot

koimataisleeps

klwnÐtwig

‘he doesn’tsleepawink’ (literally, ‘a twig’) (Kerkyra,ibid.)

b. denot

pon°hurt

yÐqalocrumb

‘I don’t feelpainatall’ (literally, ‘a crumb’)(Macedonia,Hatzidakis1917)

While emphaticnegationmaybesynchronicallyformedby theadditionof anex-pressionsuchask�n ‘even’ or potè ‘ever’ to anindefiniteconstruedwith plain nega-tion, theconverserelationdoesnotoccur:plainnegationis neverbuilt from emphaticnegationby theadditionof somede-emphasizingelement.In thisprecisesense,plainnegationis formally UNMARKED andemphaticnegationis formally MARKED.

Diachronically, ontheotherhand,plainnegationis usuallyderivedfrom emphaticnegation.Inspectionof (1) showsthateachplainnegationin this particulartrajectoryis etymologicallyidenticalwith theemphaticnegationof theprecedingstage.Indeed,everyplain negationof Greekwasonceanemphaticnegation,at leastin sofar asitsorigin canbedetermined.1

Thegeneralizationsjust formulated— thatemphaticnegationis formedcomposi-tionally with aminimizeror generalizer,andneverconversely, andthatplainnegationis diachronicallyderivedfrom emphaticnegation— hold widely for otherlanguagesaswell. Therearenumerousexamplesof emphaticnegationschanging“by them-selves”into plain negations.Wheneverwe cantracetheorigin of plain negationsinIndo-European,theyturn out to beetymologicallyidenticalto earlieremphaticones.This is trueof Englishnot, no, andnothing, Frenchneandnon, Latin non andnihil.Thegeneralizationholdsnot only for clausalnegation,but for independentnegationaswell. Yesandno wereoriginally reservedfor emphaticassertionanddenial,andsupplantedtheir plain counterpartsyeaandnay in Middle English.Instancesof plainnegationsconverselydevelopingemphaticmeaningsdonotseemto beattested.

1.3 Thecycle

Observationof suchpatternsof changein GermanicandRomancenegationled Jes-persen(1917)to posit a historicalprocessof repeatedweakeningandreinforcementnowknownasJESPERSEN’ S CYCLE, whichhesummarizedasfollows:

1Thatwouldincludeo>u(k), if theidentificationof -ki in Homerico>uki with theIndo-Europeanindefinite-kwi- is correct.

3

. . . theoriginal negativeadverbis first weakened,thenfoundinsuffi-cientandthereforestrengthened,generallythroughsomeadditionalword,andthis in turn may be felt asthe negativeproperandmay thenin thecourseof time besubjectto thesamedevelopmentastheoriginal word.(Jespersen1917:4)

For Jespersen,then, the weakeningof the negationis a matterof phoneticre-duction,and its strengtheningby additionalwords is motivatedpartly by the needto maintainthedistinctionbetweennegationandaffirmation,andpartly to makethenegationmorevivid. He suggeststhatnegationtendsto beweaklystressed“becausesomeotherword in thesamesentencereceivesthestrongstressof contrast”andasaresultbecomesaclitic. Thecontrastbetweenaffirmativeandnegativesentencesbeingnotionallyimportant,whenthephoneticattritionof negationcausesit to befelt asin-sufficient,it is reinforcedby anaddedword in orderto restorethethreatenedcontrast.Suchreinforcementalsoserves“to increasethephoneticbulk” of thenegative(p. 14),and“to makethenegativemoreimpressiveasbeingmorevivid or picturesque,gener-ally throughanexaggeration,aswhensubstantivesmeaningsomethingverysmallareusedassubjuncts”(p. 15).

Theroleof phoneticweakeningin thishypotheticalscenario,howeverplausibleitmight seem,is not backedup by anydataasfar aswe know. Our analysisof Greekturnedupnosupportfor Jespersen’sassumptionthatphonologicalweakeningtriggersthe strengtheningof negation.Therearealsosomegeneralreasonsto doubtit. Foronething, phoneticweakeningis too generala phenomenonto explainthe specificpropertiesof this unusualpatternof change.It is a ubiquitoussoundchange,but itrarely triggersmorphosyntacticchangedirectly, let alonecyclic trajectories,which(asJespersen1917:4himselfnoted)arespeciallycharacteristicof negation.And onewould like moreconvincingparallelsof phonologicalweakeningprocessesdirectlytriggeringsyntacticreanalysis.In attestedchangesof negativeexpressions,thecausa-tion usuallygoesin theotherdirection:phonologicalreductionof plainnegativesmaybe morphosyntacticallyconditioned,and,in particular,contingenton their semanticweakening. Negationsare commonlyobservedto split on the basisof differencesin function. In English,the clausalnegativeheadnot andthe argumentnaughtareetymologicallythe same,andhavedivergedaccordingto their morphophonologicalfunction,no doubtasa resultof associateddifferencesin stress.ThesamegoesforFrenchne ‘not’ andnon‘no’, bothfrom non (< neunum). A similarcasefrom Greekis the phonologicalsplit of oÎdèn into de(n) ‘not’ andude ‘no’ (= SG îqi) in Bova(Calabria)(Taibbi & Caracausi478). Bovaalsoprovidesan illustrationof a phono-logical reductionof a negativepolarity item in its modifierfunction,leadingto a splitbetweenkanèna, Fem.kammÐa ‘someone,anyone’versusk�na, Fem.kamm� ‘some,any’ (Rohlfs1949:122).

Thereforewe will assumethat the reinforcementof negationby a postverbalin-definite(the“strengthening”)is notaresponseto thephoneticweakeningof thehead.Instead,we will follow morerecentanalysesof Jespersen’scyclein seekingthedriv-

4

ing forceof thecyclein pragmaticsandsemantics.

Emphaticnegationtendsto increasein frequencydueto pragmaticallymotivatedoverusewhich is characteristicof inherentlyboundedevaluativescales. This rise infrequencyat theexpenseof plain negationhasan “inflationary” effect,well attestedalsoin politenesssystems,hypocoristics,pejoratives,andscalaradjectivesof all kinds(Dahl 2001).2 Uncontroversially, an obligatoryelementcannotbe emphatic,for toemphasizingeverythingis to emphasizenothing. Therefore,when emphaticnega-tion risesin frequencyto thepoint whereit approachesobligatoriness,it necessarilyweakensto regularnegation.

Thevirtueof thisaccountis thatit explainstheobserveddirectionalityof change,for it allowsnomechanismby whichplainnegationscouldmutateinto emphaticnega-tionsthroughnormalusage.However,it is still insufficient,for thetypologicalobser-vationsof the precedingsectionimply that someof the changesmustbe intercon-nected:theymustconstitutea CHAIN SHIFT. This is indeedhowJespersendepictsthecycle.He imaginesit happeningin two phases.Thefirst, whichcanconstituteaniter-ablechainshift on its own,involvesaweakeningof thenegationplusacompensatorystrengtheningby meansof someaddedword. Thesecondconsistsof a reanalysisofthestrengthenerastheprimaryexponentof negation(thatis, asthenegativehead).

Let usthereforemarrytheJespersenianchainshift ideato thepragmatic/semanticmechanismproposedby Dahl andothers.We endup with the following view of itsnatureandmotivation.Thecontrastthatthechainshift maintainsis not thatbetweenaffirmationandnegation,asJespersenassumes,but the contrastbetweenplain andemphaticnegation. And theweakeningthatunderminesthecontrastis not phoneticweakeningof plainnegation,butsemanticweakeningof emphaticnegation.

The ideathat thefirst phaseof thecycle is a chainshift involving plain andem-phatic negationprovidesthe beginningof an answerto the diachronicpart of ourpuzzle. If weakeningandstrengtheningalwaysgo handin hand,thenit follows thatthecontrastbetweenplain andemphaticnegationwill bemaintainedat eachstageofthelanguage.

In the nextsectionwe examinethe mechanismbehindthe changemoreclosely,andaddressthequestionhow, unlike morefamiliar chainshiftsmechanisms,it gen-eratesa circular trajectory. Our answeris basedon an analysisof the pragmaticsandsemanticsof emphaticnegation,outlinedhereinformally andto beelaboratedinanotherpaper.

2As Dahl pointsout, not everyfrequentword (andnot eveneveryscalarpredicate)is proneto undergo“bleaching”, andnot all “bleaching” is dueto this kind of inflationaryeffect. We think thatsometypesofsemanticweakeningarereally automaticresultsof lossof lexicalor morphologicalitemsin asemanticfield(deblocking). For example,wheredid not acquireits new directionalmeaningthrough“bleaching” fromfrequentuse,but simply becauseit automaticallytook over the meaningof whither whenthat word waslost.

5

1.4 An interpretationof Jespersen’scycle

To modela chainshift we needat leasttwo things:a principlethatrequiresthemain-tenanceof somecontrast,anda processthat disruptsthe contrastby alteringoneoftheelementsthatexpressit. Therequirementthat thecontrastbemaintainedentailsthat any neutralizationin the relevantdomainwill be accompaniedby someotherchangethat preservesthe contrast,or immediatelyfollowed by someotherchangethatrestoresit. Sucha sequenceof changesconstitutesa chainshift.

Chainshifts areusually invokedin phonology, wheretheir statusis largely un-questioned(but in realityquiteproblematic,Gordon2002).But if chainshiftsexistatall, thenon generalgroundsit oughtto bepossibleto makea strongercasefor themin morphosyntax,especiallyin coremorphosyntacticcategoriessuchasnegation.Thereasonis thatmanysuchcategoriesareuniversal,andtheir formalexpressionis highlyconstrainedby principlesof grammar. Whensucha categoryis lost, it mustberegen-erated,andtherearealimited numberof possiblewaysin whichit canberegenerated.

Anotherway to put thispoint is asfollows. Theprincipleof contrastmaintenancecaneitherrequirethata particulargrammaticalor lexical distinctionbepreserved,orthataparticularphonologicalorgrammaticaldevice(say, agivenphonemicor featuralopposition)shouldbearsomefunctionalload.Thechangesdrivenby thesetwo typesof contrastmaintenanceareknownas“push-chains”and“drag-chains”,respectively.Jespersen’s cycle(at leastaswe understandit) is both,sincethecontrastit maintainsis bothfunctionallydeterminedandhighly constrainedin its formalexpression.

We haveseenthat emphaticnegationsarebuilt morphosyntacticallyfrom plainnegations,andweakenbackto plainnegations.This impliestwo processes.

(4) a. Morphological/syntacticstrengthening:A plain negationis emphasizedwith a focusedindefinite.

b. Semanticweakening:Theemphaticnegationbecomesnoncompositional,losesits “even” meaning,andbecomesa plainnegativepolarity item.

Theseprocessesinteractto generatethe complexof changesknown as Jespersen’scycle.

Strengtheningandweakeningare functionally antagonistic,in that oneaddsanexpressiveresourceto the language,while the other eliminatesit. Thereforetheiretiologynecessarilydiffers,andtheyarealsoformally quitedistinct. Yet,aswe shallsee,botharegroundedin thenormaluseof scalarevaluativeexpressions.

Our proposalpartly returnsto the traditionalview that thecycle is drivenby theexpressiveuseof language.In contrastto traditionalphonology-drivenaccountsandrecentsyntax-centeredaccounts,we treatthecycleasfundamentallya processof se-manticchange,to besurewith phonologicalandsyntacticconsequences.

TheGreekdataprovideanimmediateempiricalreasonfor pursuingthisapproach.Theevolutionof negationfrommedievalGreekto themoderndialectsinvolvesseveral

6

roundsof thecyclewith no accompanyingsyntacticchangewhatsoever, andfor thatmatterwith no relevantphonologicalchangeeither. What doescharacterizeall thechanges,however,is aninvariantpatternof semanticshifts.

This is of coursenot to saythat thecycleneverhassyntacticrepercussions.Theweakeningphaseof the semanticshift canactuallybe associatedwith two kinds ofsyntacticreanalysis.Thefocusedindefinite,onceit becomesanegativepolarity item,canbecomea negationheadof its own — the familiar case— or undergoanotherdevelopmentwhich is describedherefor thefirst time: it canbecomeanounor indef-inite pronounacceptablein positivecontexts.This happenedin four separateGreekdialects(section3.2).Sothesyntacticaspectof Jespersen’scycleis quitecomplex.3

Also, theweakeningmay, but neednot, leadto phonologicalreductionof oneorbothof its parts,asaresultof which it caneventuallybecomemonomorphemicagain.

In additionto the inflationarymechanisminvokedabove,the causalexplanationof Jespersen’scyclerequiresa secondassumption,which concernsnegationsystems,andis alsoindependentlymotivated. This is that any languagehasthe resourcestoexpressbothboth plain andemphaticnegation.This is certainlytrue for Greek: asalreadynoted,all dialectsat all timesdistinguishformally betweenthe two typesofnegation.Analysesthat postulateemphaticnegationonly for intermediatestagesinthetrajectoryreducethisto amereaccident.As far astheyareconcerned,thelanguagemayor maynothaveemphaticnegationin its repertoirebeforethechangeis initiated,andagainafter it is completed. If a strengthenermustalwaysbe available,then itfollows necessarilythatweakeningandstrengtheningmustgo handin hand.As soonasanegationis lost, it is renewedby anotherroundof strengthening.

Why mightlanguages“need”bothplainandemphaticnegation?Probablyto servethe very samerhetoricalfunctionsthat causeit to be overused.At leastthreemainfunctionsof emphaticnegationcanbeidentified.Thefirst functionof emphaticnega-tion is to markcontradictionof a (possiblyimplicit) assertion.

(5) A: Obviouslyheatetheporridge.B: No, hedidn’t eattheporridgeat all.

A secondfunctionof emphaticnegationis to denya presumptionor anexpecta-tion.

(6) A: Whatdid it costyou?B: I didn’t costmea thing.

Henceit canalsoconveyanimplicit expectation;for example,(6) couldbeusedin acontextwherethecostof theitemhasnotcomeup in orderto conveytheideathattheitemcouldhavecostsomething.

Third, emphaticnegationstrengthensa negativeassertionby lifting contextualrestrictionson anindefinitein thescopeof negationor by forcinga ‘totality’ reading

3On theissueof unidirectionalityin general,seeKiparskyMS.

7

ona definiteargumentof a gradablepredicate.A clearinstanceof this latterfunctionof emphaticnegationis aspectualdisambiguation,andspecificallydistinguishingtelicand atelic readingsof predicates.For example,(7a) is ambiguousbetweena telicreadingandanatelicreading

(7) I haven’teatentheporridge.

• Telic reading:‘I haven’teatenanyof theporridge.’

• Atelic reading:‘I haven’teatenall theporridge.’ [‘. . . thoughI mighthaveeatensomeof it.’]

Adverbialemphaticnegationdisambiguatesthesentencein favorof thetelic reading.

(8) I haven’teatentheporridgeat all.

Weassumethatthesefunctionsaresobasicthatanylanguagemusthavethemeansto expressthem.

Supposingthata languagemusthavesomemeansof distinguishingplainandem-phaticnegation,andthatemphaticnegativeelementsmaybecomeweakenedthroughnormalusage,it follows directly that negationmustbe subjectto the characteristiccyclic courseof changethatMeillet andJespersenidentified.

We arenow in a positionto solveanotherpuzzle. Jespersen’s cycle countsasaclassicinstanceof grammaticalization.Grammaticalizationisconsideredto beUNIDI -RECTIONAL grammaticalchange(whethertrivially by definition,or in aconsequenceof somedeeperprinciples,Kiparsky MS). How, then, can a CYCLIC trajectoryofchangebeaninstanceof grammaticalization?

Givenwhatwe havesaid,oneanswermight be thatonly onephaseof thecycle,theweakeningphase,instantiatesgrammaticalization.It consistsof the“bleaching”ofanemphaticnegativeinto aplainnegative,with lossof compositionality, andtypicallywith phonologicalreductionaswell. In the strengtheningphase,the lost expressiveresourceis formally renewed.But (in termsof thetraditionaltypologyof change)thisis notgrammaticalizationbutordinaryanalogicalchange.A newemphaticnegativeisbuilt compositionallyin accordwith thelanguage’smorphologicalandsyntacticrules.Theiterationof reductivegrammaticalizationandconstructiveanalogyyieldsa cyclictrajectory.

Self-evidently, all so-called‘unidirectional’ changesmustbe part of suchcyclictrajectories,thoughpossiblyof extremelylongduration.For,if theinputsof unidirec-tional changewerenot renewable,theywould no longerexistanywhere,becausethechangewould havetakenits courseeverywhere.Moreover,becauseof theuniformi-tarianprincipleit wouldbepuzzlinghowtheyevercouldhavearisenin anylanguageatanystage.

8

2 Jespersen’scyclein Greek

2.1 Descriptivesummaryof thetrajectories

The documentedhistory of Greekhasthreecompletedroundsof the cycle, plus afourth which is underwayin a numberof dialects. All consistof a mutually linkedsemanticstrengtheningandweakeningprocess;theweakeningphaseof cycleI is alsoassociatedwith a syntacticargument-to-headreanalysis.Thediachronyof thenega-tion systemsin (1) is shownin (9). Thefirst columnof arrowsin thechartrepresentthemorphosyntacticstrengtheningby theadditionof afocusedindefinite,andthesec-ondcolumnof arrowsrepresentthe corresponding“inflationary” weakeningsof thenegation’s force. Keepin mind that the weakeningsarepurely diachronicreanaly-ses,whereasthe strengthenings,in additionto beingdiachronicinnovations,form asynchronicoppositionbetweenemphaticnegationandplainnegationin thegrammar.

(9) PLAIN STRENGTHENING WEAKENING

(I) o>u . . . ti- −→ oÎ dè én −→ oÎdèn

(II) (oÎ)dèn . . . ti- −→ dèn . . . tÐ-pote −→ dèn . . . tÐpote

(III) dèn . . . tÐpote −→ dèn . . . prama −→ dèn . . . prama

(IV) dèn . . . prama −→ dèn . . . dros�, �pantoq . . .

At stageI, theplain,non-emphaticnegationo>u(k) strengthenedtooÎ-dà e>ic (oÎdeic)‘not evenone’,andoÎdeic in turn lost its emphaticmeaningandbecameaplainnega-tive ‘no-one’. ThecorrespondingneuteroÎdèn cameto serveasa clausalnegation,atfirst emphatic(‘not at all’), latersimplymeaning‘not’, andbecomingphonologicallyreducedto dèn (seeRobertsandRoussou2003:157-160for ananalysisof thischangein the contextof their approachto grammaticalizationbasedon minimalist syntax).Thiscyclewascompletedby theearlymedievalperiod.

At stageII, theplain indefinitetÐ- is strengthenedin negativecontextswith potè‘ever’ in theneuter. In themasculineandfeminine,its emphaticcounterpartis kaÈ �nãnac, kaÈ �n mÐa ‘evenone’. Viz. dèn . . . tÐ ‘not anything,nothing’→ oÎdèn . . . tÐ-pote ‘nothingatall’, dèn . . . tÐc/tin�c ‘not anybody, nobody’→ kaÈ �n ãnac ‘nobodyat all’. TheresultingtÐpote andkanènac arethenin turn weakenedto plain negativeindefinites,in fact, to negativepolarity items.Thisdevelopmentwascompletedin themedievalperiod.4

4It mayhavebeenatwo-stageprocessfrom astrongnegativepolarity item(anindefiniteacceptableonlyin negativecontexts)to a weaknegativepolarity item (acceptablein otherlicensingenvironments,suchasantecedentsof conditionalsor questions).

9

As emphaticnegativesareweakened,newonesareagainformedto replacethem(stageIII). Dependingon thedialect,this is donein oneof two ways. Thenegationcanbereinforcedby theadditionof astrengthenersuchask�n ‘evenone’,eitherbareor addedto anindefinite(includingtÐpote andkanènac):

(10) StageIII: Strengtheningby k�n

a. dànnot

êqwhave

k�(n)atall

(ywmÐ)(bread)

‘I don’t haveany(bread)at all’ (Mani, Blanken160)

b. dènot

meme

poneihurts

k�at all

‘I don’t feelpainat all’ (Mani, Georgacas106)

c. k�gganenac ‘no-oneatall, notasoul’ (Cappadocian,Danguitsis1943)

d. k�gkanac (< k�n kanènac) ‘no-oneatall’ (Macedonian,Kretschmer273,Hoeg201)

e. k�ntÐpoutac = ‘tÐpota �polÔtwc’ (Naousa,Kontosopoulos181)

f. kadÐbouda = ‘tÐpota ântelwc’ (Samothraki,Kontosopoulos188)

In Pontic, the renewalof emphasisis achievedjust by ınas ‘one’. The result is aninterestingreversalwherekanıs, etymologically‘evenone’,is usedfor plainnegationandthebareınas‘one’ is usedfor emphaticnegation.

(11) a. kanıskh erjen ‘nobodycame’

b. ınaskh erjen‘not evenonepersoncame’(emphaticnegation,Drettas281)

Thesecondsourceof newemphaticnegativesatstageIII is strengtheningby lexi-cal indefinitessuchasprama ‘a thing’.

(12) brÐskeifinds

mi�na

kopèllagirl

. . . poÌwho

dànnot

¢xereknew

pramathing

‘finds a girl whohasnoclue’ (Thera,Kontosopoulos166)

In yet a fourth cycle,someof theseemphaticindefiniteslosetheir emphaticchar-acterandbecomeweaknegativepolarity items. The emphaticnegationis thenre-newedby other lexical items. In the Cretandialect,prama ‘anything, something’,which wasintroducedat stageIII, becomesa weakpolarity item (capableof appear-ing in questions,see(13)),andis replacedin its emphaticfunctionby wordssuchasdros(i)� ‘dewdrop’. Theseexamplesarefrom the copiousinventoryof minimizersandgeneralizersfrom everystageanddialectof Greekcompiledby Andriotis 1940(pp.86-87).

(13) a. >Ed¸kasÐgive-3Pl

souyou-Dat

prama?thing

— >Apantoq !hope

‘Did theygiveyouanything?Nothing!’ (‘Not ahope!’)

b. ^Efaeceat-2Sg

prama?thing

— Dros�!dewdrop

‘Did youeatanything?Nothing!’ (‘Not a dewdrop!’)

10

c. ^Eqete‘eat-2Pl

nato

f�meeat-1Pl

tÐbotec?anything

— Dwroc!present!’

‘Do youhaveanythingfor usto eat?Nothing!’ (‘Not evenfor free!’)

3 Strengtheningby k�n- ‘even’ and -pote ‘ever’

3.1 MedievalGreekkanènac andtÐpote

In Medieval Greek, the inheritedplain indefinite tÐc, tin�c, originally ‘someone’,‘anyone’,beginsto bedisplacedin referenceto humansby thenewemphatickanènac‘(not) evenone’ (from kaÈ �n énac). In modernGreekkanènac hastakenover,butreflexesof tÐc, tin�c survivedialectally. In this sectionwe documenta previouslyunnoticedintermediatestageof the trajectoryin Late MedievalGreektexts. In thefollowing sectionwe examinethe survivalsof indefinitetÐc in CalabrianGreekandtheremarkablereflexesof k�n- ‘even’.

At thestageof medievalGreekrepresentedby suchtextsasMakhairas, theem-phaticforceof kanènac hasbeenattenuatedbutnotyetwholly lost. Its meaning‘(not)evenone’hasacquiredapartitivecomponentandnormallyrefersto somecontextuallysalientgroup. Thetranslation‘(not) evenoneof them’ (German‘keiner’), is usuallyappropriate(andindeedthat is how Dawkinssometimesrendersit). In virtue of thismorespecializedmeaningit contrastswith themorenondescripttin�c ‘(not) anyone,no-one’(German‘niemand’).Theexamplesin (14)-(15)illustratethecontrast.

(14) kanènac ‘anyone(of somegroup)’,‘(nicht) einer/keiner’

a. kaÈand

dànnot

âglÔtwsenescaped

kanènacanyone

‘andnotoneof themescaped’(Makhairas16.15, tr. Dawkins)

b. nathat

. . . mànnot

âlemonhj|hcpity-2Sg

kanènananyone

‘that youhavemercyonno-oneof them’ (Makhairas16.1)

c. kaÈand

�nÐswcif

. . . �f sùcleave

zwntanänalive

kanènananyone

‘and if you leaveanyoneof themalive’ (Makhairas16.5)

d. oÉthe

�rqonteclords

âsÐghsan;fell-silent

kaneÈcno-one

lìgonspeech

oÎnot

dÐdeigive

‘the lordsfell silent;noneof themuttersaword.’ (Belisarios153)

(15) tin�c ‘(not) anyone’,‘(nicht) jemand/ niemand’

a. dànnot

�fhkenallowed

tin�nanyone

n�to

p�gùgo

eÊcto

tänthe

soult�nonsultan

‘would allow nooneto go to thesultan’(Makhairas126.4)

11

b. täthe

koursocpillaging

dànnot

e>inaiis

tinäcto anyone

di�forocadvantageous

par�except

touto the

fous�touarmy

‘pillaging is of advantageto nooneexceptto thearmy’ (160.36)

c. mhdànthat-not

êqùhave

tin�canyone

fwtÐantorch

‘so thatnooneshouldhavea torch’ (362.11)

d. kaÈand

dànnot

âglÔtwsenescaped

tin�canyone

‘andnooneescaped’(546.4)

e. n�so-that

m�not

täit

barejhis-bored-with

tin�c,anyone

�mm�but

ålwnwnto-all

n' �rèshpleases

‘so thatno-onegetsboredwith it but everyonelikes it’ (Threnos11)

f. dànnot

prèpeishould

�njrwpìcperson

tinacany

n�to

sàyou

kathgor shaccuse

‘No personshouldaccuseyou’ (Threnos165)

g. tin�c �g�phn met' aÎtän m� boulhjh poÐshanyonelovewith him notwantdo‘so thatno-onewouldmakepeacewith him’ (Threnos459)

h. kaÈand

n� kratounkeep-3Pl

täthe

dÐkaion,just

tin�cno-one

m�not

�dikhtaiis-done-injustice-to

‘and theyshouldkeepjusticeso thatno-onegetsunfairly treated’(Belis-arios384)

Similar semanticcontrastscanbefoundin theindefinitesof otherlanguages,e.g.Germanniemandandkein, keiner. Thesearenominalandadjectival,respectively,andthe partitive readingof the adjectivalkein, keiner is presumablydueto an im-plicit complementthatdenotesa contextuallysalientgroup. It is likely thata similarsyntacticdistinctionis responsiblefor the differencein meaningbetweentin�c andkanènac.

Thecontrastappliesalsoto interrogativessuchasEnglishwhich(of them)vs. who,orGermanwelcherversuswer, andto regularindefinites,whereit seemsto correspondto the well-known contrastbetween‘partitive’ and ‘nonpartitive’ indefinites. It isthereforeinterestingthatMedievalGreekexpressedthecontrastformally in all threeof thesepronounseries,that thedistinctionwaslost in all of themin ModernGreek,andthatin eachcaseit wasthemorespecificindefinitethattookover.

(16) TheintermediateLMG system

indefinite interrogative articlekanènac ‘anyone(of agroup)’ poiìc ‘which’ ãnac (specific)tin�c ‘anyone(atall)’ tÐc ‘who’ ∅ (nonspecific)

12

In ModernGreek,theoriginallyspecificindefiniteskanènac, poiìc, and(in mostuses)ãnac havedisplacedthe indifferently specificor nonspecifictin�c, interrogativetÐc,andthenull indefinitearticle.

3.2 Theliberationof kanènac: from ‘nothing’ to ‘something’

In the familiar scenario,whennegativepolarity itemslosethe requirementthat theymustbein thescopeof negation,theybecomenegatorsin theirown right. This is thepathby which, for example,Frenchpersonnechangedfrom ‘a person’to ‘no-one’.Greekalsooffersa few casesof this type,whichwediscussin section(3.4)below.

More interestingly, however,the “liberation” of negativepolarity itemscanhaveanotheroutcome,which to our knowlegehasnot beendescribedbefore: they canturn into regularindefinitesor nouns. This hashappenedin four widely separatedperipheraldialectsof Greek. In Cappadocian,kanènac functionssimply asa nounmeaning‘person’,thereverseof thedevelopmentundergoneby prama.

(17) a. Êtìthis

patis�qking

s�lsesent-3Sg

d�the

�skèri“a

soldierst,his

nato

¢brounfind-3Pl

dekeinìthat

dothe

kaneÐcperson

‘The king senthissoldiersto find thatman.’ (Ulagatsh,Dawkins384)

b. ¢rtecame

énaone

kaneÐcperson

‘therecamea person’(Ulagatsh,Dawkins366:2)

c. >htonwas

�ll�other

naone

kaneÐcperson

‘therewassomeoneelse’(Ulagatsh,Kesisoglu1951,citedfromAthanasiadis1976:164)

d. ât�that

kaneÐcperson

‘aÎtäc å �njrwpoc’ (Axos,Mavroxalividis& Kesisoglu94)

e. polÌvery

ærtosincere

kaneÈzperson

�naiis

‘polÌ eÊlikrin�c �njrwpoc e>inai’ (Axos,M&K 114)

In at leastsomeof theseCappadociandialects,it continuesto occurundernegationaswell.

(18) tthe

îrgowork

m�mine

’cto

kaneÐnano-one

dènnot

dìit

qairinÐskwentrust

‘I entrustmy work to no-one.’ (Axos,M&K 182)

In UkraineandCalabria,kanènac is a non-polarityindefinite,either‘someone’,or moreparticularly‘someonespecific(butnotnecessarilyknown)’, ‘a particularper-son’. In view of the specificreadingof kanènac in MedievalGreekdocumentedintheprecedingsection,it is temptingto seeit asanarchaismpreservedon themargins

13

of theGreek-speakingterritory, ratherthananinnovationwhich justhappenedto takeplacefour separatetimes.

In the Tauro-Romeicdialect of Mariupol (Ukraine),kanıs is apparentlyboth aspecificindefinitein affirmativecontexts,anda weakpolarity itemsundernegation.At anyrate,Sergievskij(1934:562)glosseskanıs as(a) Russiankto-to' ‘someone(specificbut not necessarilyknown)’, asopposedto `kto-nibud~' and(b) `nekto'‘(not) anyone’in thescopeof overtor impliednegation,andcitestheexamples:

(19) a. kanıs ırtin ‘someonehascome’(`kto-to prix�l')

b. den fenit bdina-pa kanıs ‘there doesn’tseemto be anyonearoundany-where’(`nigde nikogo ne vidno')

Theusesof kanènac in Calabrian,judgingfrom theexamplescitedbelow, arealsocompatiblewith themeaning‘someonespecific(butnotnecessarilyknown)’.

(20) a. kanenamuipe ‘someonetold me’

b. ırte kammıa ‘someone(fem.) came’

c. jir eguokanena‘I’m looking for someone’(Rohlfs1949:122)

In Cretan,kanènac maybea simplepositiveindefinite(Pitikakiss.a.).

(21) a. Kianènacsome

perastikìcpasserby

d�here

pèrasepassed-by

kiand

¢koyecut

t�the

portok�liaoranges

‘Somepasserbymusthavepassedbyandcuttheoranges.’ (Cretan,Pitikakis)

b. åthe

basili�cking

kat�labeunderstood

ítithat

Íp�rqeiexists

kaneÐcsome

gèrwnold-man

‘The king understoodthattherewassomeold man.’ (ibid. 412)

c. â�nif

dènot

moume

eÒpetetell-2Pl

poiìcwho

êqeihas

qwsmènonhidden

kanenasome

gèrontaold-man

‘if youdon’t tell who(amongyou)hassomeold manin hiding’ (ibid. 412)

Let us mentionin this connectionthat Cretanalso haspositiveprama ‘something’(Pagkalos1983,420-1).

(22) KaÐand

trwneeat-3Pl

pramasomething

‘and theyeatsomething’

It is hardto saywhetherpositivekanènac in thesedialectsderivesfrom theme-dieval specificindefinitekanènac discussedin the precedingsection,or developedfromtheordinarymodernnegativepolaritystage.In alanguagewith negativeconcordsuchasGreek,thedistinctionbetweenindefinitesof the ‘some’ typeandindefinitesof the ‘no/any’-typeis neutralizedin thescopeof negation.Soa ‘no/any’ indefinitethatpassesthrougha negativepolarityphasecouldchangeto a ‘some’ indefinite.

14

3.3 Continuationof tÐc, tin�c ‘someone’

TheconservativeItaliot dialectof Bovain Calabriahastıspo(< tÐspote) ‘(not) any-body’, ‘nobody’. Evena reflex of the bareunsuffixedpronounis preservedin theAccusativetino (Rohlfs1949:123).

(23) dopusince

tithat

egoI

enot

xxorosee

tin oanyone

odossuinside

‘sinceI don’t seeanyoneinside’ (Taibbi& Caracausi411)

In Bova, tıspoalternateswith kanena, asseenin thesetwo parallelversionsof aproverbialsaying:

(24) a. mebetween

portadoor

ceand

porandajamb

minot-should

valiput

kanenaanyone

tathe

dastilafingers

‘nobodyshouldstickhisfingersbetweendoorandjamb’ (T&C 374)[‘Tra impostae stipitenonmettaalcunole dita.’]

b. mebetween

portadoor

ceand

poranda,jamb

tıspoanyone

minot-should

valiput

tathe

dastilatufingers-his

‘nobodyshouldstickhisfingersbetweendoorandjamb’ (T&C 374)[‘Tra impostae stipitenonmettanessunole propriedita.’]

However,thereis a syntacticdifferencebetweenthem. In the texts,kanenaoccursonly undernegation,whereastıspois oftenfoundasanindependentnegation.5 Thus,thesyntaxof tıspoin Bovadiffersmarkedlyfrom thatof othernegationselsewhereinmodernGreek.It is, in fact,essentiallythatof ClassicalGreek.Whentıspoprecedesthefinite verb,noothernegationis required:

(25) a. tıspono-one

paigoes

stoto the

ParadisoParadise

mewith

tinthe

garrottsacarriage

‘nobodyentersParadisein acarriage’(Taibbi& Caracausi1959:384)

b. mabut

tıspono-one

efaniappeared

nato

mume

tathem

ttsiporeiknew

dyamer´ısiinterpret

‘but therewasnobodywhoknewhowto interpretthemto me’ (T&C 444)

c. EsuYou

tisp’no-one

ad.d.oother

kathan

emmename

gapailove

‘Youdon’t loveanyoneotherthanme’ (T&C 345)

Whenit follows, thefinite verbrequiresanadditionalovertnegationsuchase(n).

(26) a. enot

sseyou-Acc

passeggisurpass

tıspoanyone

‘andnobodywill overcomeyou’ (T&C 1959:406)

b. tonthe

geroweather

ennot

doit

xxorısees

tıspoanyone

‘no-oneseestheweather’(T&C 385)

5Otherthanthis syntacticproperty,kanenaandtıspoappearto beequivalent.At leastwe candetectnodifferencein meaningbetweenthemin thetexts.

15

c. sentsawithout

nathat

tonhim

ıvvrifound

tıspoanyone

‘without anybodyfindinghim’ (T&C 410)

Theruleseemsto bethatanegativeargumentthatfollows anegativehead‘reinforces’it (negativeconcord),while a negativeargumentthat follows a negativeargumentisindependent.Thepatternin (25)–(26)is familiar from ClassicalGreek:

(27) a. oÎk o>iden oÎdeÐc ‘no-oneknows’ (Aesch.Ag. 632)

b. oÎdeÈc oÎk o>ide ‘no-onedoesn’tknow, everyoneknows’

c. oÎdeÈc o>ide ‘no-oneknows’

Couldthisperipheraldialecthavepreservedafeatureof ClassicalGreek?Morelikelythesyntaxof tıspois simplycalquedonthatof Italiannessuno.

(28) a. nessunomi piace‘I don’t like anyone’

b. nessunononmi piace‘I like everyone’

c. nonamonessuno‘I don’t loveanyone’(*amonessuno).

Althoughthe Italian editorsdon’t makethis point explicit, theysurelynoticedit be-causetheyregularlytranslatekanenawith alcunoandtıspowith nessuno(asin (24)above).TheseItalian counerpartsform a similar syntacticpair: Chi hai visto? ‘Whohaveyouseen?’Nessuno/*Alcuno.‘Nobody’.

3.4 k�(n) asanewstrengthener,andasa head

In Mani, Macedonia,andThrace,k�n (< kai �n) ‘even’, ‘at least’servesasageneralstrengthener(Blanken160,Georgacas106,Andriotis88).

(29) a. dànnot

êqwhave

(ywmÐ)bread

‘I don’t haveany(bread)’(Blanken160)

b. dànnot

êqwhave

k�(n)atall

(ywmÐ)(bread)

‘I don’t haveany(bread)at all’ (Blanken160,Georgacas106)

c. dènot

meme

poneihurts

k�at all

‘I don’t feelpainat all’ (Georgakas106)

d. dànnot

tänhim

�f kasilet

n�to

p�reitake-3Pl

tänthe

�nasasmìbreath

touhis

k�atall

‘they didn’t evengive him a chanceto catchhis breath’ (Mani, Kon-tosopoulos173)

e. k�neatall

gn¸shunderstanding

dànnot

êqeihas

‘he hasnobrainsatall’ (Selybria,Andriotis88)

16

RelatedemphaticindefiniteNPIsincludeN.W. Peloponneseki�c (Kalavritaetc.,Geor-gacas107),andtheSouthernislandtype(sv)ki�(c) (CreteandKarpathos,Georgacas106-7,Kontosopoulos32).6

NorthernGreekk�(n) is itself in turn combinedwith other indefinitesto formcompoundemphaticnegativepolarity items.

(30) a. k�ntÐpoutac = ‘tÐpota �polÔtwc’ (Naousa,Kontosopoulos181)

b. kadÐbouda = ‘tÐpota ântelwc’ (Samothraki,Kontosopoulos188)

c. kanèmou ‘toul�qiston’ (Alonistaina,Kontosopoulos172)

d. k�gganenac ‘personne!,pasuneame’(Demirdesi,Danguitsis1943)

e. k�gkanac< kai an k�nac (Macedonian,fromk�nac< kaènac< kanènac,Kretschmer273,Hoeg201)

f. kaniw ‘nothing’ (Aenos,Thrace,Andriotis88)

Thiscyclehasgoneto completionin theCorsicanManiotof Cargese.Herek� hasbecomea particlewhich normallyaccompaniesdà(n), without emphasis(“sansqu’ilsubsistela moindreideed’un renforcementde la negation”,Blanken159-161). Inshortsentences,suchas(31a),it is nearlyobligatoryfor mostspeakers;theemphaticnegationis kajìlou asin (31b).(Examplesfrom Blanken159).

(31) a. dènnot

êqwhave-1Sg

(k�)k�

‘I don’t haveany’ (je n’enai pas’)

b. dènnot

eqwhave-1Sg

kajìlouat all

cairìtime

‘I havenotime atall’ (emphatic)

c. �the

gl¸ssalanguage

dènnot

âq�netouget-lost

k�k�

‘la langueneseraitpasperdue’

d. dènot

g�n′eimake-3Sg

k�k�

pol′uvery

krÔocold

‘il nefait pastresfroid’

e. dènnot

ênaiis-3Sg

k�k�

kakìcbad

�jrwpocperson

‘ce n’estpasunmauvaistype’

f. k� qreÐa ‘pasbesoin’

g. xerw lig�i, m� k� pol′Ô ‘j’en saisunpetitpeu,maispasbeaucoup’

In theplaceof k� thisdialectalsouses�ra ‘tÐpota’ ‘nothing’ (from �rage?).

6Cretan(sv)ki� is furtherstrengthenedto (sv)kiaouli�c, miaouli� (< ki� goule�) ‘(not) even(one)sip’ (Georgakas109-110).Cypriot Greekhasbareên tzaÈ (< dàn kaÈ). Blanken161andKontosopoulos1994:24glossit asaneutralnegation:ên tzaÈ dÐnei mou = dàn mou dÐnei, ên tzaÈ e>ida ton = dàn täne>ida. However,our Cypriot informantstell usthat it retainsits statusasanemphaticnegation.

17

(32) a. dànnot

êqwhave-1Sg

�raanything

‘I don’t haveanything’(Blanken109)

b. Hthe

gunaÐkawoman

ìpouwho

dennot

èjeihas

�raanything

nato

k�nhdo

p�eigoes

stoto the

perÐpatowalk

‘A womanwho hasnothing else to do goesfor a stroll’ (AcademyofAthensfield notes,citedfrom Nicholas534)

TheCargeseuseof k� maybea Frenchcalque.But Blanken160notesthat thesameusehasdevelopedindependentlyfor thecognatemaka in OtrantoGreek,whichhasbeensubjectonly to Italian influence:7

(33) enot

plonnosleep

makamaka

‘I’m notsleeping’(‘je nedorspas’)

A parallelwould be the dialectof Thasosoff the coastof Thrace,wherek�(n) canfunction as a one-wordemphaticnegation‘not at all, not in the least’ (PanagiotisPappas,voce).

4 Lexical renewal of emphatic negation

4.1 Replacementof tÐpoc, tÐpote

Thereplacementof tÐpote andits cognatesby otherlexical itemsis acharacteristicofEasternGreek.TÐpo(c) survivesin Farasa,Amisos,andin thePonticdialectsof OfandSourmena(Oikonomidis252,Athanasiadis167).

(34) a. tÐpoc tzì poika se ‘I didn’t doanythingto you’ (Farasa)

b. doul eec 'kè jè n� poÐshc tÐpoc ‘You arenot going to do any of thechores’(Amisos)

c. � k�ta tÐpo 'kÐ âxèr� ‘the catknowsnothing’ (Surmena)

Elsewherein EasternGreekit is oustedby a varieryof otherexpressions.Ponticdialectshavedèn, tidèn ‘anything’ (Nikopolis kesvki, Papadopoulos121) insteadoftÐpote asan“anaphorevidee” (Drettas).

(35) a. âg°I

dànnothing

�lloelse

�kÈnot

yalafwrequest

seyou

‘I requestnothingelsefrom you’ (Argyroupolis,Balabanis1937:102)

b. >Egnèf�senwaved

åthe

r�ftectailor

kaÈand

íntecthen

tereisees

nàneither

qalÐncarpet

nànor

dànnothing

�lloother

‘The tailor wavedandat thatpoint seesneithercarpetnor anythingelse’(DivanKerasounda,Balabanis1937:108)

7Rohlfs(1930:845)termsmak� a “Fullwort derNegation.”

18

Other lexical strengthenersinclude(36) ksaj ‘at all’, (37) xıc, xits ‘nothing’, ‘atall’, ‘zilch’, ‘never’ (< Turkish hic < Persianhec), (38) dip, jip ‘totally’, and(39)enase (< Turkish sey< Arabic say? ‘(any)thing’), skrap(Sarakatsan,Hoeg278),svkr�ou (Maced.Vlasti, Andriotis 89), anda hostof minimizer-derivedexpressionssuchaséna cÐj’, éna kokkÐ, éna nÔj, éna kourtj�, éna tzÐgkra, éna koutsoÔj’, énakrizÐ, éna stali

“� (Axos,Mavroxalividis& Kesisoglu78).

(36) a. ksaj kh-enenkasta‘I’m not tiredatall’

b. paradasksaj kh-exo‘I havenomoneyatall’

c. ksaj epıessin-ardean‘haveyoueverbeento Ardhea?’

(37) xıc, xits ‘nothing’, ‘at all’, ‘zilch’, ‘never’ (< Turkishhic< Persianhec)

a. trizthree

mèrecdays

qÐczilch

qizurÐcholy-man

rennot

galajèvgisays

‘for threedaystheholy mansaysnothing’ (Silli, Dawkins288)

b. qÐcatall

rènot

sk¸'nitirises

‘he doesnot riseat all’ (Silli, 300)

c. qÐcatall

nato

f�goueat

ywmÐbread

rènnot

èqou,I have,

nato

for¸souwear

roÔqaclothes

rennot

èqouI have

‘I havenota pieceof breadto eat,I havenoclothesto wear’ (Silli, 290)

d. son.in the

gozmoworld

qıczilch

denot

gelanemade-laugh

‘nothingin theworld madeherlaugh’(Sılata,452)

e. ¢dounewas

�nif

grÔo,cold

qÐcnever

poÔthat

’dènot

joÔdounewas-before

‘it wascold,suchasneverwas’ (Dawkins557,cf. Demirdesi,Danguitsis1943:119)

f. xıczilch

kh-e rkanot-understands

‘il necomprendabsolumentpas’(Pontic,Drettas402)

Although dıp ‘totally’ is not a negativepolarity item, it functionsto strengthennegationin virtueof havingwidescopeoverit.

(38) a. aftosinedıp ftoxos ‘totally poor’

b. dıp filotimodenexi epanotu ‘completelylacksfilotimo’ (Babiniotis181,cf. Sarakatsan,Hoeg278)

c. dÐp-tÐpoutac ( Kozani,Kontosopoulos182)

As mentionedbefore,we analyzethe type enase to bea combinationof a mini-mizeranda generalizer:

19

(39) a. énaone

sèthing

mènot

lal csay-2Sg

‘don’t saya thing’ (Ulagatsh,Dawkins376)

b. énaone

svèjthing

dèmnot

bìrzecould-3Sg

n�to

boÐk�do

‘therewasnothinghecoulddo’ (Axos,M&K 216)

5 Conclusion

Jespersen’s cycle is due to the interleavingof two processes:the strengtheningofnegationby morphosyntacticmeans,andits lossof compositionalityandweakening(grammaticalization,bleaching). Although they havefunctionally oppositeresults,theyarenot formal conversesof eachother. Oneis paradigmatic,the othersyntag-matic. An emphaticnegativealwaysweakensby itself (it is “bleached”),neverinvirtue of beingcombinedwith someotherelement. On the otherhand,negationisstrengthenedonly by combininga simplenegativewith anindefinite.A simplenega-tive, or a simpleindefinite,neverbecomesanemphaticnegativeon its own.

Both the weakeningandstrengtheningphasesof thechangearegroundedin therhetoricalfunctionof boundedscalarevaluativeexpressions.But theyhavedifferentcauses.Theweakeningof theexpressivenegationto anordinarynegationgoeshandin handwith, andis causedby, increasedfrequencyof use.Its endresultis thelossofa necessaryexpressiveresourcein thelanguage.Thestrengtheningthatrenewsit is aconsequenceof theneedfor this expressiveresourcein a language.Suchalternationbetweenweakeningandstrengtheningprocessesis an instanceof the largerdialecticof productionandperceptionin theeconomyof language.

20

References

Andriotis,N. 1940.Die Ausdrucksmittelfur “gar nichts”, “ein wenig” und“sehrviel” im Alt-,Mittel undNeugriechischen.Byzantinisch-neugriechischeJahrbucher16:59-155.

>Ajanasi�dhc, S. 1977.To Suntaktikì thc Pontiak cDialèktou. Kastani�: >Ekdìseic

SwmateÐou PanagÐac Soumel�.

Balab�nhc, >Iw�nnhc. 1937.ParamÔjia. >Arqeion Pìntou 7.83-124

Blanken,Gerard.1951.LesgrecsdeCargese(Corse). Leiden:Sijthoff.

Dahl, Osten.InflationaryEffectsin LanguageandElsewhere.In JoanBybeeandPaulHopper,eds.FrequencyandtheEmergenceof LinguisticStructure. Philadelphia:Benjamins.

Danguitsis,Constantin.1943. Etudedescriptivedu dialectede Demirdesi. Paris: Maison-neuve.

Dawkins,RichardM. 1916. ModernGreekin AsiaMinor. Cambridge:CambridgeUniversityPress.

Dawkins,RichardM. 1931. Folk tales from Sourmenaand the valley of Ophis. >Arqeion

Pontou 3: 79–122.

Makhairas,Leontios.1932.RecitalconcerningtheSweetLandof Cyprusentitled‘Chronicle’.Ed. RichardM. Dawkins.Cambridge:Oxford: ClarendonPress.

Dahl, Osten.2001. Inflationaryeffectsin languageandelsewhere.In Bybee,JoanandPaulHopper(eds.)Frequencyandtheemergenceof linguisticstructure,471-80.Amsterdam:JohnBenjamins

Dawkins,RichardM. 1940.“The dialectsof ModernGreek.” Transactionsof thePhilologicalSociety,1–38.

Dawkins,RichardM. 1950. Forty-fivestoriesfrom theDodecanese.Cambridge:CambridgeUniversityPress.

Drettas,Georges.1997.Aspectspontiques.Paris:Arp.

FwsthropoÔlou, Dèspoina. 1938.ParamÔjia >Imèrac. >Arqeion Pìntou 8.181-202.

Georgacas,Dimitrios. 1938.Glwssik�. ByzantinischeZeitschrift38.99-110

Gordon,MatthewJ.2002. Investigatingchain shifts and mergers. In J.K. Chambers,PeterTrudgill, and NatalieSchilling-Estes(eds.) The handbookof languagevariation andchange.Malden,Mass:Blackwell.

Hatzidakis,G.N. 1917.GlwssikaÈ parathr seic. >Ajhna 29:118.[Non vidimus.]

Horn,LawrenceR. 1989.A naturalhistoryof negation.Chicago:Universityof ChicagoPress.

Horrocks,Geoffrey. 1997.Greek:A History of thelanguageandits people. LondonandNewYork: Longman.

Jespersen,Otto . 1917.Negationin Englishandother languages. Copenhagen:Host.

Katsoyannou,Marianna.1999.Theidiom of Calabria.In A.F. Christidis,ed.,113–119.

Kiparsky, Paul.MS. Grammaticalizationasoptimization.Forthcomingin D. Jonas(ed.) Pro-ceedingsof DIGS2005.http://www.stanford.edu/˜kiparsky/Papers/yalegrammaticalization.pdf

Kontosìpouloc, Nikìlaoc. 1994.Di�lektoi kaÐ >Idi¸mata thc Nèac <Ellhnikhc. >Aj na.

Kretschmer,Paul.1905.NeugriechischeDialektstudienI: derheutigelesbischeDialekt. Wien:AkademiederWissenschaften.

21

Krifka, Manfred.1995. The semanticsandpragmaticsof polarity items. Linguistic Analysis25:209–257.

Ladusaw, William A. 1992.ExpressingNegation.SALT2:237–259.

Ladusaw, William A. 1993.Negation,indefinites,andtheJespersenCycle.BLS19:437-446.

MabroqalubÐdhc, G., I.I. KesÐsoglou. 1960. Tì glwssikì ÊdÐwma thc ^Axou. [Thelo-cal dialectof Axos.] >Aj na: Kèntro Mikrasiatikwn Spoudwn.

Mpampini¸thc, G. 1998.Lexikì thc Nèac Ellhnik c Gl¸ssac. Aj na: Kèntro LexikologÐac.

Nicholas,Nick . 1999.Thestoryof pu: Thegrammaticalisationin spaceandtimeof a ModernGreekcomplementiser. Ph.D.Dissertation,Melbourne.http://ptolemy.tlg.uci.edu/˜opoudjis/Work/thesis.html.

P�gkaloc, Ge¸rgioc. 1955.PerÐ tou Glwssikou Êdi¸matoc thc Kr thc. Tìmoc 1oc.

[OntheCretandialect.Vol. 1] Athens.

P�gkaloc, Ge¸rgioc. 1983.PerÐ tou Glwssikou Êdi¸matoc thc Kr thc. Tìmoc 7oc.

T� Laografik�. [OntheCretandialect.Vol. 7.] Athens.

Papadìpouloc, ^Anjhmoc. 1955. <Istorik  grammatik  t hc pontikhc dialèktou. [His-torical grammarof thePontic dialect.] Athens.

Pituk�khc, Man¸lhc. s.a.Tä Glwssikì >IdÐwma thc >Anatolikhc Kr thc. [TheDialectof EasternCrete.] ^Ekdosh Politistik hc kaÐ Laografik hc <EtaireÐac >Ap�nw

Merampèlou Ne�polic Kr thc.

Roberts,Ian,andAnnaRoussou.2003.Syntacticchange.A minimalistapproachto grammat-icalization.Cambridge:CUP.

Rohlfs,Gerhard.1949. HistorischeGrammatikder unteritalienischenGrazitat. Sitzungs-berichtederPhilosophisch-historischenKlassederBayerischenAkademiederWissenschaften.Munchen.

Rohlfs,Gerhard.1962.NeueBeitragezur Kenntnisder unteritalienischenGrazitat. Sitzungs-berichtederPhilosophisch-historischenKlassederBayerischenAkademiederWissenschaften.Munchen.

Salvanos,G. 1918.Melèth perÈ tou glwssv. >Idi¸matoc twn ân KerkÔrø >Argur�dwn.

Aj na. [Non vidimus.]

Sergievskij,M. V. 1934. Mariupol~skie greqeskie govory. Izvesti� Akademi@i

Nauk SSSR. Otdelenie Obwestvennyh Nauk, 533-587.

Taibbi,G. andG. Caracausi,(eds.)1959.TestiNeogrecidi Calabria. TestieMonumenti:Testi3. Palermo:Istituto Sicilianodi StudiByzantinie Neogreci.

22