This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
fassertion.ThesamecanbesaidaboutBy~e&Fleischman's(l995)do inted outthatahhoughatfixesarcformedfromreducedparticlcs.andfollowtheoriginal particleorder.diachronicaUy.Mwpaniclesmaygranmlaticalilethat!illagapinone oftlleinllectionalparadigms.lftheparticle!iUsagapinoneo(tbemoreccntralcatc· gorics,andthispanicleisreducedintoanatfix.theresultingordcrisconAicting.AI· thoughtheiconic-distanceprinciplemaycausearcanalysisorpotentiallyaredistributionofoneormoreoftheattixesinque
mainepi
ChaplcrZ. Modahty
In the present uudy, a di.
fortheabilitativcmodalityinsomelanguagedescriptions,andinthcdcscriptionof Mixtec(Bradley&HoUenbach 1988)itisuscdforafutureaspe.:t(cftheoccasioual uscofirrealisasanaspectdcnotinguuinitiatedl!\•cnts).Corrcsp<~ndingtothep<~tcn
tial in the m0<1d system. Palmer (2001) uses the term speculative for epi
tivettdoes"notgenerallyindicatep
A dubtlativc [... ] i> u•wl1y dosmbed ""cxpro.,mg an dement o( doubt that the o•·cmdo>cnbedinlhcpr<>p<»ili<>n<>e
Another epistemic concept that is rarely given its own term, and which will play a key rok in the description of rFIWHHHU. is uncertainty. In the description ofNgiyambaa. Donaldson {1980) uses the term "iguorativc" for a clitic with such a muning (Palmer 2001:53).Nahara{2002)alsousesthetermignorati\'einthatscnseinherdescription ofthe[apanesecopula Theignorati,·ealsoexistsasamodalityterminvan(iijn's (2006)dcscriptionofYurakaro!.ltwillbeusedhereaswcU.althoughalwayswithan explicativenote.Analternati\'etermisotfcrcdbylkrbyshirc(l979)inhisdescription ofHixkaryana with it< "uncertainty" sulfu.lastly, the term admirative will be us.:d for a modality of surpri«: or counter-expectation (see DeLlnccy ]997)_ This term wiU be neededinthedescriptionoftheSpanishsubjunctivc One problem with the definition of cpi.
mndalverbsthatbothdcnntecvemandcpistemic-C\identialmodalit)•-Thisisoncofthe reasons why Palmer (2001) chonso:s to subsume prop<~
4. Eg. Aguoruna (Jjo;oro;m, Yr~n l\16)), Crow (Siouan Proper: Groceyk 2007, 395). Awa (Ea
19
~
Mod•hlpndSul>ordmators thesameuni"'rsalsu~rcategory.Howt".·er,thatfactmayha"'otherexplanations,ouchas
gr.unmaticahzal!on. Accordmg to Hopper & Traugott {2003). the grammatical!Zallon pro<e>S goeo from more central. lexical categories to more ~ripheral func!Lonal catego· ries.lntheGermaniclanguages..somemorphemcshawbotheventandepi.StemLC-evtdentialmeamngs.e.g.nrusl,mayandca"inEngl"h.lmportantly.thede"'loprnenthasgone from.-.-enttoef""temLC-evidenl!almodalityandnottheotherwayaround(Traugottl989, Bybee et al. 1994). Thi.s dearly ouggests that epistemK-evidentlal modality" a more peripheral category than ,.,.. nt modality. Other morphemes haw both e"'nt·modal and fu. lure meaning>. e.g. will and shall in Engltsh. Again. the development goes from event mo· dalit)'tofuture{Palmer2001:l041f) AnotherinterestingfactasregardsmndalwrbsmtheGermaniclanguagesis,as "~II be shown m more deta1l in Chapter 7. that modal wrbsexpressing event modality can be con>tructed "'ith modal verbs that express epistemic-evidential modality m German. Dutch and the Mainland ScandinaVIan languages. In any such duster. the most penpheral 1.e. leftmost morpheme "always epistemic-evidential (constructed example from Swedish) (3)
De m
/fl
In these languages, syntactiC sco~ of auxiharies mirrors semantic scope. The epistem· LC·e>~denhal meanings have sco~ owr the deonllc·d}'Ilamic ones Furthermore,intheselanguages.modalwrbsoccurnngtothele!i:oftheperfect auxiliary tend to get an epistemic·evidenllal reading. whereas modal wrbs occurnng tothenghtarealmostalwarseventmodai!Eide2005:30l;exarnplesfromSwedish) (4) a.
b
Banken kan lla agerat innan jag blev inkopplad the.bank can haw acted before l became involved 'Thebankmayhaveactedbeforelwascalledin.'
(GP04)
Fhtiga vedrnt1dningsl.;lsare har k1111nal vad d!hgentreaders.of.weekly.maga•zineshawbeen.able.toguesswhat somvar i giirmngen (GP04) that was in the.making 'Dihgemreadersofweeklymagazineshawbeenabletoguesswhatthere was brewing m the wind'
This has to do "~th the fact that epistemic-evidential modal verbs tend to be finite m the Germanic languages {Abraham 2003). whereas event-modal oneo can either beliniteornon-linite.Thus,evemmodalityisgeneratedmsideoftense(caneltherbedeternuned by tense or not). whereas epistemic·evidential modality is not
Chap!al_ Modahr)-
Cinqu~
( 1999) pres~nts funh~r morphosyntactic eVIdence that epistemJ<·e•·idential modality has s)111ae!k scope over event modality. Remmiscem of the situatmn in the GermaniC languages, the modal panicle kmr in Sr:man (Creole) receives an eplstemicreadingwhenitoccurstnthcleftofthep;>sttensep;>rtideben,butanabilitativereadlngwhenitoccurstntherightofit (5)
a.
A lwn /ten e nyan he may PASTPRU\ieat 'Hemayhavebeeneating'
b. A /ten lw11 nyan. htPASTC3JlUt He could eat.' Yd another interesting example is Chickasaw (Mudwgean). where ep1stemic and de· ontic·d)•nanucmodalityareexpressedbythesamesutlixes.Themodalsutlix-Ol'llihas two uses. When it is located insidetheten.<e suffix -tok, it ha. an abihtallve meamng. When located outside the ten.<e sutlix. however, it has an epistem1c meaning of speculatlonordeduc
a
Hilha -1 -n'n-tok dance-lM>·a'n·PST 'lcouldhavedanced'
b. Hilha -tok -n'tri dance -P~T -a'ni 'Hemustha•·edanced' Lastly. it should be pointed out that Byb« (1985: 20ff. l65ffl, who mvesngated verb denvalionandintlectloninSOunrelatedlanguages(see3.2helow),arguesthatthe ind!Cati\'l:,sUbjuncti\'l:andepistemicmodalityshouldbekeptapartfromeventmodality.Wiulethefnrmercanbesaidtodenotethedeg=ofcornmitrnent!Othetruth ofthepropo51tionandhavescnpe!1Verit.thelatterisrnoreagent-:mdevent·related :mdha.s.copeO\'erthemaineventonly l'..
pcrmis>Oonandobligation.boc~w.clheyd<>enbcccrlaincondthonson
wllh...,ganltothcm>inp"'dkation Byb~'sdivi.>ionwas.>upportedbythefactthatdeonticanddyn:mticmodahty.incon·
•=·
tra.tto ep!Stemic modal it)'. were nrely expreS<ed as inflection on the verb m her ple: "However, intlecti{lnal marker; of obligation. permi.,ion, ab11ity or mtentiott are utremel)'rareinthes=ple"(Byhee !985: l65ff)_lntereotingly,however,shefuund examples where event modality appeared to he derivationaL To mnclude, epistem!C· evtdenllal modality can be separated from event modality on both semantic and for· mal ground>
11
n
Modahry•ndSubord1na1ars A< was pointed out in the previous section, Bybee's ( 1985) and Bybee &Fleischman's ( 1995) definitions of epistemic modality ditfer somewhat from that of Palmer (2001) \\'hereas Palmer (2001)
Onetermthalwillbeofparticularimportancefortheprc<entinvcstigationisthedeclarative.Thisisduetothefactlhalili
Chop~a!. Modohw
onchespeaker'sknowledgt"thactbcpropo
b!mlhertothetruthofthepropruition.Anothersuchcxamplethatcouldbementioned IS Crow (Central Stouan, Gracz}·k 2007· 392). where the deduallve marker -k is !he unmarkedmemberofasetofeplstenucmarker< Thelesscommonasseruve-shrdenotesthatthe
arguesthatdeclarati\-ecompkmentdauses(riMr-dause<)illgenerallackasseruvene« To him.rhal-dausesare merdycontentclauses (propositJOns), SJnce they cannot func\Lona<mdependentutterances.asopposedtodedarauvemainclauses.wbtcharc:asscrttons(see2.l2) Cnstofaro(2003)furtherarguesthatnon-assentontSthcdetinmg charactensncof
terms dedaratl\'e and mdlCative are used mterchangeably One such example ts the fi. nal mood suffix -rm Hua (East-Central Family: East !>;ewGmnea Highlands). Ha1m3n (1980).caUsthissuffixthemdicatl\'e.whcrcasFolq·(l986.162)callsuthcdeclarative Anlllher example ts the mood ~uffix in Nootka AccordJng to Klok1d (1978: 8), Dnidaht(aNontkadiale.:t)hasafinalmoodsullix-a·,whlChhecallsthede.:larali\'e According to David<son (2002: 267tf), howe>'t'r. Nootka ha. a penul!tmate mood suffix -(mJa·- (10Uo"'t'd bysub,.ct agreement) which he call< the indkattve.1\ thud example >
Even more renurkable.perhaps, ts Lo"·e's (1999· 274) dc.cnpltonofNambtquara. where indkattve is the cover term for mood forms that are not tmperauve. Jndtcative verbscanthenbedlVIdedmtodeclarativeanddub!laltveones HerethetermsmdlCatl\'eanddedarattveareusedinanopposnewayaswouldbeexp~cted.Ciearly.thereis noumversallyacceptedcnt~rionthatl
andino;bcauve. Forreasonssuchast~eprc<entedabow.thepresentautboradoptsPalmer's (!986)detinnionofthede.:laraliveasatcrmfortheunmarkedep~
foramodalmarkertbatdenotesfactualnyanddausetypeandcorrespondstotheindiCati\'Cinthemoodsystem.Withthtsmea.mng.itcanalsobeusedasalabelforcomplemenuzersoftheTHAT-type
''
>4
.l.lodohtyondSubordtnota" 2.1.5 "lheconditional Theconditoonalisanothermndalitycategorythatishighlyrrlevantforthepresent mvesligalion.sincetheCiermanictFamongotherthingsh3s3conditionalfunctoon.in the sense that ot introduce~ conditinnal protase< lnPalmer(2001).theconditionali.<subsum«lundrrprnposillonalmodalJty.and in Bybee (1985: 170) it i~ placed together "ith the indtC31l\"e-subjuncllve tfutinctmn and epLStemic modality in the proposed functional (mood) category ··commitment to truthofproposihon~Conccptually.theconduionalcouldbecountedasamemberof
rroposihonal modality ~ince conditional c~u<es art alwap; semantically non-factual; thej·denoteh)"potheticalnrcounterfactualpropo
Another problem with the conditional. which is of particular 1mportance for the present investigation, tsit~statusin conditinnal prntases_ Although conditional prota· •••maybesemanttcall)"hypothetkal.counterfactualetc..Jtcouldbearguedthatthe primary meaning of the conditinnal a.< a protasis nJatker is cirrumstance adwrb1al ("incase').lnfact,insnmelanguagrs,thcmeaningoftheconditionalmprotasisdarne• seems to be temporal ("when"). In e.g. Dongnlese ~ubian. the conditional function is marked by tbe adverbializmg sutfix -k which al
Chapter2_ M<>dahty
trmporalongin,buttherrar~alsomanythathavtepistemic,optativeorinterrogatt"",
i.r.modalorigm. Thestrongestindicationthattheconditionalisindeedamodalitycategof)'talher than a circumstanc~-adverbial one is the fact that in many language,, condttional moodmarkersorparticlesarenotconlinedtotheadverbialprota.>ISdausebutilio occurintheapodom,wheretheydenotethattheeventishypotheticalorcounterfactuaLThisLSe.g.thecaseinWaur;i(Ea.>ternMaipuran:Arawak)withthemnditional rnclillc-mw(Derbyshirel9H6:550tf)andinTigre(EthiopianSemitic)withthepartide wtl (Palmer 1001: 209). Such examples are numerous, partocularl)·when it comes to counterfactualconditionals.lnadausechaininglanguagelikeCentraiPomo,lasllJ·,it "onl)·the apodo,;is clause with the final \'erb that carries the conditmnal marker (Palmer10(Jl:l51tfJ Jtmustalsobeacknowledged.aswassaidabo,·e,thatinsornelanguages,thecondttional and the irrealis or subjuncth~ tna)' he expressed by the oame morpheme. In other languages, the conditional forms a binary diotiru:tion together with etther the subjuru:ttveltrrealis or indicath•e/realis. e.g., in Hungarian "~th the mood suffixes subjunctive -1a andconditional-auaa(Abondolo 1998: 447), Limbu(Eastern Kiranli: Tibeto-Burman: Sino-Tibetan) "~th irrealis -p-ni and conditional -mm (van Driem 1987: 135-141), Mapuche (Araucanian) with indicative -(u)}• and conditional ·(•i)l (Smeets 1989: 226tf), and Mnni (Wissel Lake-Kernandoga: Central and Western Mam Section: Trans-New Guinn) with indicative -l)a and cnndttinnal -.rogo, -ogo (Larson&Larsoni9SH:406-43l).Fnrthe<ereasnns.thecond!llonalshou!dbeseenas amemberofpropositiona!modalityratherthanadrcurnotance-adverbialcategory.
l.l lhe indicative-subjunctive and uali<-irreali< distinctions At the centre of modality lie the di
linctionbelong.stopropositionalmodalityonl}'
15
2.2.1
'!he indicative-subjuncth·e dislinelion (Palmcrl986,2001andNoonan2007)
In many language<. (e.g. the Romance and Germanic ones), morpho:mcs thai arc labellcdthcsubjuncti•·ccanbesaidtoha•·ea
(7) a.
Yo s~bi~ que d csr01M ahi ] knoW.MH that he J>e.3S<;.PAST.INV here 'lknewthathewashcrc'
b. Dt•do que ~prcad~doubt.lst; thatlearn.3S<;_p~~•-•ut 'ldoubtthathe'.
Epistcmicmodalily:spcwlative: (II)
UTIN (Palmer200l:l09) Jan~ apsolutoswucas quominccduntinfeelorcs now paid.olf think.2s<;.PR~:S.su~r when comdn dyers 'Youmaythinkthattheyarcalreadypaidolf.whenincomcthcdycrs'
E>•idmri~l mod~/iry:
(9)
r,·porlarivc:
ITAUAN Si diu chc i «~ldati si~11o !'<'lliti. one says that the «~ldicrsbdrL.PII~<.W~J kfi 'Theysaytbatthe
(Palmer200i:ll4)
Lastly. epistemic modal morphemes licence the subjuncti>'e in some languages in the sense that if a clause contains epi
'Maybe he's coming.'
Chap~a2.
Modoh!)"
Jndeed.a<wassaidabwe,Byb.:e(l9.'15:J70)proposo:d!ha!!heindica!ive-
Relatedtotheu«:nfthesubjunctivealterpredica!esofdoubt.thesubjunc!i'l<"can alsnbeas.
Cuandose trrmi11r Ia guerra.>"Oiwrt a lnglaterra murn.fliT.lS!l!O England when RnL tinish.3st;.rRU.Sii~J !he war 'When !be war ends. I will return to England'
Another type of grammatical marker< tbat often licence the
~7
l1l
Mod•htrondSubordmators n..cessity,abili~·etc.ftodenotethattlleeventLSpotential(non-actuali .. d],wllereas
tlleindicativeisusodforallotherevents (13)
LORI (LkASLAN;Noonan2007:104) Zine va pta xa.·ye bed<>ze. woman from man wantedlSG COMP cllicken·OBJ steaUs<:.sUBI 'Thewomanwantedtllemantostealtllechicken.'
In many other languages (as 111 Rwsian, Peman, and Bemba). the subJunctive is used notonlya!terevemmodalpredicates..butal""afterpredicatesexpressingdoubt,asthe following example from RusSLan shows (r\oonan 2007: l07f (14)
faromneo•aJuS',ltoby Bonspriid. [ doubt COMP Bons come.SUBI 'ldoubtthatBoriswoUcome/carne.'
lntheselanguages..thesubjunctiwisal..,usedinconditionaldausestodenotethat theeventishypothetical.Theindocativeisusedafterprodicatesexpressingknowledge. unerance,belief.andcomment.Therefore,intheselanguages,themdicative·subjunctivediStmctiondenotesthattheeventisfactualvs.non-factual(o.e.propo"tionalmo· dality).Noonan(2007)callsthistherealis-irrealLSdLStmction Lastly.mafewlanguages(•uchasSpanish),thesubjunctoveisnotonlyusodalter predicatesexpressmge>-.ntmodahty.doubt(=7babow).andinconditmnalcon· strucuon.,butalsoalterpred.icatesthatexpresscomment,astllefollowingSpanish examplesllows(Noonan2007:109f (15) Lamwt<> que Juanwlga estanoche. regret.ISGCOMP Johnleave.lsG.SUBtthis mght lregrettllatfohnwillleavetonight' Themdicativeisusedafterpredocatesexpressingknowlodge(•ee7aabovef,unerance and belief. Since complement clauses of commentatove predicates are presupposed to betrue,andthereforecanbesaodtobeequallyunwortllyofbeinga.sertedaspotential, dubitativeand conditional clauses. the indicative-subJunctovediStinctioncan be said tobeoneof...,.rtionvs.non-assertionintheselanguages,Noonan(2007:109,followmgKleml975],argues,althougllheadmitsthatnotallofthemdicativecomplements mtheselanguagescanbeconsideredassertionsmthe"technicalsense"oftheterm. Palmer (1986), who prefers a umfied drliniuon of the mdteati>"O·subjunctive dis· tinctmn,arguesag;nnsttlleanalysisoftlleSpamshindicatove-subjuncuvedJStinction mtermsofassertionandolfersanotherexplanationthatiscompatiblewiththefactualityanaly"s.TheuseoftllesubjunctiveinSpanisllcomplernentdausesofcommenta· live predocatesgives the clause an emotive-evaluative interpretauon, he argues. This willbedealtwothinmoredetailbelow. One problem woth any definihon of the mdicative-subjunctive distinctoon is. as Palmer (2001: Bllf. 1381f) silo""'· that subjullcltve morphemes also tend to denote
Chapter2_ M<>
speech-actnotionssuchaspoliteimperative.jU<siveanddeSJdrrati\'eloptalive,asin thefoUm,~ngexamples:
Polllermprratiw·
(Palmer200l:l38)
(16) Z"fALtAN
''"
enter.3sc.;.PR1S.SiiBJif.yDu.please Desidmllirrloptative· {17)
{Palmer200l:lll)
fi:LA:W1ST·AfRtL:A
nju"IM bald"r be.long.ls(; ..
ltcouldbearguedthatthesefunctionsarepartsofthee>·emmodalsystemlasiti>in Palmer 1986). They are dearly scmantkallj• related to obligative and >-olitive modality and determine that the event is non-actualized. Furthermore, when expressed by e>pe· ciall)·dedkatedmorphemes(ratherthanby!hesubjunctive),theyoccurinmodalor mixed ')'stems in S!ltne languages (Palmer 200L SOtf)_ On the other hand, Palmer {2001:80)arguesthattheimperativeandjussivcshouldbedistinguishedfromob~ga·
uve modaltty in the sense that they are performative {speech-acts): "the speaker actu· aUygtvesthe'command'intheactofspeaking'~Thi
typicallydonotoccurinsubordinatedauses'(Palrner2001:80) {18) a. You must come.
I ~id that you must come •Jsaidthatcomein'
Aregardsthedesiderative/optative.Palmer{200l:l341f)poimsoutthatitJSsrmanti· rally relatedtovolitionalmodality, butthatthereisonerrucialdifferenro. Vohtive morphemes could be interpreted as directive under the right fdinty condittons. wh~reasdesid~rative mnrphemes are more remm-.d from the speech situatmn. Therefore,vohtmnalclausescannotbecnunterfactual,whereasdeSLderativeclausescan {19) a. lwishthathewouldhavecome. b. "!wantlohntoha•-ecnme According to Palmer(200l: 13<1),thcdesiderativ.:shDuld iruteadberegardedasa memb..r of propositional modality, together with markers that express fears (apprehensive),sincethey"indicateallitudestoproposition>ratherthan unrealized
s.
llshouldbepOintedoutthatthertwereanuollyembedd.edtmp
~9
JO
ModahtyondSubordm•tors
evenu" Bybee & FleiSchmann (1995), who a<Sume a tlurd domam of mod :!lay whtch hutodowuhspeechacts.placetheoptallvetheremstead Althoughthede>!derattveloptativeustofthesubjunctwernayha>..,anon-factual explanauon.thet3ctthatthesubjunctivecanbeustdtodenotespeech·actnouons
whtchfunctionalcategory•tbelnngs. PartlvforthLSrea.son,Palmer{200J)chon<e
explanauonthatwa.compahhle"iththet3Ciualityanalyst< Thisw!llbedealtwithm mnredctadm2.2.5 1.1.1 'Jhcreihs-trreahsdt
a good-good-RU,L 'lt~
Chaptorl. b
Mod>lo~·
haat nyt ttpooy -nt -h~ dog.>OII)PL-kJ!l.PlAL:TI01<·IOO·IItR 'ltnughtkllldogstoo"
A
wberea
b
c
rniy="law OJIL-]AULRK·sing Heshould/tssuppasedtosmg." ktiy- til vi=bahw s~u- I AU.IRII· S« Jdon'r.-ebtm' /ti.
l"i=bahw
"lflseett' In some cases. the reahsand trreabsthemseh·esdenotetbe>ecategories{Palmer 2001: 145\f) In e g. Martcopa. tile realt~ denote actuale>-,nt< m the pre.-nt {see lOa) nrpast,whereastbeirrealiscandenotethefuture(:tlthougbttmoreoliensignal
ny-aa)·-ha )/2-gtvC·IRR 'lwtUgiveittoyou'
Furthermore.theirreahscanexpresstheinterrogauvemsomelanguages.e.g.mCaddo {Palmer20Cll:l2ff,Chalel995 354)· (23)
s~i>l-
),=bahw -nah 2AGUtR-SU -HRF "Ha•-,youseenhtm?"
Thepanernthatcategortestllatentailthatthee>·emisnon-actualtzedorirrealare markedasirrealist<notentirelycon.
l'
grammatical markers that are constructed \\ith the irrealis (negative. obligati,·e, prohibitive.conditionaletc.).Wherea$thelanerareprelixcs.theimp.:rati\'eisunmuked andthefutureismarkedbyasutfix.Chafe(l995:359)arguesthatitislikelythatthe future and imp.:rative constructions stem from an older layer of Caddo morphology, before the reali.<-irreali.< distinction \\'a.< grammatkali;r.ed_ Thus. they do not interact "'ith the distinction in the: same way as the: prefixes do. Funhermore. Chafe ( 1995: 358) speculatesthatbothfutureand imperativemaybejudgedtodenoteeventsthatare more likely to o.::curthan sayne~ativeevenuor events in polar questions. Interestingly. negative imp.:rati,·es. i.e. prohibiti,·es. are in the: irrealis in Caddo (Chafe 1995: 356)_ The:;e are obviously more hypothetical than imperatives, as they ask the listeneriiOStorealiuthcpropo
k~Ssal1lyi:b.lhw PM<>H· 2.AIHRR· :«:e 'Don'tlookatit!'
E'·en more interesting is the fact that in some of those
lan~uagcs
that have a choke
betweenreali.
tions. this choke i< not random but also has to do with what is marked as actual \'S hypothetical. In Central Porno. the future is marked as either realis or irrealis dependin~onthesp.:aker'sexpectancyfortheeventtooccur(Mithunl995:370,37lltf):
(25) a. W-nta: Iii w;i--n -hi ?a- q6: be -w :/k"e town: to ~O·IMPf -~AME.IRR LM; tO\\'ard: carry -P~RH' :fU'I TU go to town and brin~ it back·
b. "/i-
f'O. w
l.A
:Ja
P~Rf\' :UJH.SIM.MU~
1\U '/ba --n --
"Aiieri'm~one,youwiUsutfer.'
The first sentence describes events that arc not predetermined to occur. but rather dependonthespeaker.Thesecondreferstotheinel-'itable(Le.presupp
Chop!CI'l Modoiow \'S.non-factual{r~callthatN'oonanl007rder
reali<-JrrealJS dJSuncnon mar also ~h~d some bgbt on why the subJUOl
dmauon (!'.>lmer 2001 5). This follows from the fact that the
used(Palmcr2001.5).Therdore.realis-irreahsmJgbtbehenerterm<mtheselanguagc<.How>:ver,bothpairsrepresenttheuni\'l:~lvalue
mg to Palmer (2001. l4lllf; by capital fromtheonood..withthes.amcname)·
lett~r<,
the unwersal values are dtstinguished
JflndJ
thecross-bnguJstlCcategoTJes Since th~ irreahs, but alro the subjunctive. may co-occur woth or mark anJ' of the othcrmodalityutegoTJesplusMhercatcgories
.13
34
lrlod>.htyondSubotdtnolnr.< •upercategoryModalit)•withRealis-lrruli
tanttoe\..Juateit. As was said above, Palmer (2001) divides Modality into two domams: proposi· tional modality and event modality. Whcrea< propo
npresseshislhersurprise towards what isob\ions to the addressee {Chafe 1995: 357): (26) htis- ba-
?a=sa-yi=k·a,.ih-sa'l AUM· J.8~N.IRR·name· know 'Mygoodness.hekn...,.·smynantc!'
At the sameumeas Palmer(200 l ) wishes to keep the ntoodsystem binary Realis-lrrealis, he admowledges that there are languages where speo:ch-act categories such as the tm· p~rattve and jussive are expresud by especially dedicated morphemrs on the Side of thesubjunctive(Palmer200l:l36ff).Furthermore.inlangnageswiththerealisnreahs dt>tmction, the mealis (or sometimes the realis) co-occur with such markers. Thus, thr mood system is not always binary. Another speech-act mood that Palmer (2001:204tf)touchesuponandthatmayoccuronthesidcoithesnbJuncttveiSthe optattve.r.g.mCiassical
Chop~..-!_
Modaht)·
how~v~r.
Palmer {2001: lMifJ argues that the optati\'e i11 al lea.
gued below that a narrower definition of the di
non·assertionandthatRealis-Jrrulisare\'Jiuesthatcanbeascribedtoallmodality categones. In fact, much ofthe criticism can be properly d~alt ....~th if the Realis-Jrreahs distincttOnisgiventhenarrowerdefinitionreal-irreal(Chafel995).actualJzed-non· actualized (Mit hun 1995), or f~ctual-non-factual (P3lmer 1986. Noonan 2007) and if modahtytsdi\•idedintothethreefunctionalcategoriessp..,ch-actmodahty,pmposi· tmnalmodalityandeventmodality(correspondingtothethreedomaimproposedln Byb~ & Fleischman 1995). Where a.< the speech-act moods imperative. jus.stve, mter· mgative.optativeetc.arentce"arilynon-a"ertivethroughbeingotherspeech-acts. theyarenotnecessari!ynon-factual.althoughtheytendtobemterpretedassuch.They canbemarked.ufactualtnachievccertainelfects
lS
A< was said in 2.2.2. Chafe ( 1995} and Mithun ( !995} argue that the factlhatthere is variation ;u to what categoric~ are marked ;u ~alis or irrealis only demonstrates that thcrcalis-ir~alisdistinctionshouldbeseenas(panof)afunclionalcalegorywhichis
dislinctfromtheseothercategorics.Someoflhet;'J'Oiogical'·arialionmayhavchisloricalexplanalions,e.g.lheuseoflherealistogctherwilhlhefutureandtheimperative in Caddo (Chafe 1995: 359). Furthermore. both the future and the imperative mar bejudgedtodenoteevenuthataremorelikelytooccurthansayprohibili\'C.<.polar questions.andnegali\'CS, whichalldisplaytheirreali
centralfunctionalcategoriessuchastcnscandaspcct.lfaninlerrogativeorttcgalive clauseisinthepastperfective.itismarkedas~alis.whc~asifitisinthefutureitis (typically)markedasir~alis(Mithunl995:381tf)
Mithun {1995) further sugge~u that the ditference betwun the languages where the negative and intcrrogati,·e always co-occur with ir~alis marking and those where the marking of ~alis-irreali< is independent of them may be explained in term.< of scope.ln$0melanguages,thepropositionmaytirstbecharacteri~edasrcalisorirrea lisandthcnb.:negatedorqucstioned;inotherlanguage~,thepropositionmayfirstb.:
negated or questioned and then marked for mood. in which case the mood marking maybesensili,•ctothcnegath·eorintcrrogati\·emcaning.lntheformerlanguages. negationandtheinterrogativehavcscopeovcrtherealis-irreali.<;inthelauerones.il is the other way around. This clearly suggests that the grammatical marker< rep~scnt dislinctfunctionalcategories.rathcrlhanoneandthesame.Chafe(t995:360)otfersa rome"·hat similar explanation for the Northern lroquian languages. where realis-ir~ alismarkingisal
Chapter2_ M<>•or, pmpmnttheditferencebttwuntheindicativeandthcsubjunctiveandshowsthatthe}" areinfuctnotrtdundant. Heretheymarktheadditional'1'istemtc(i.e.factual)di>· tinctionofrelativecertaintyvs.non-certainly. Since certainty-non-certaint~· clearly are epistenuc nottons.. this suggests that the focal meamngofthe Rea!is-lrrealis distinction is proposiltonal modality. Recall that in 2.2.1&2.2.2,itwasshownthatthtsubjunctiveandtheirrea!tscandenoteandoccur togttller witll '1'tStemic-evidential modality. The connection between the sub)unctiveITrealisandproposttionalmodalityismorefundantentalthanthat.though.AsByhe.> (1985) reports. tlu.re are languages "ith epistemk markers that belong to a mood syslem.onthesideofmoodssuchastheindicati•·e.subjunctil"eandcondttiona!(butalso moodssucllastheinterrogaliveandoptative,itseems]_Palmer(2001:162tf)alsotind> languageowlu.retheindicative-subjunctil"(:.realis-lfre;>lisorsimJ!ardistinctionsappear to hal-e intermediateepi.
Senecawiththetrio(actual(directperceptionormemory).future{predt
Another example of an overlap between irrealis and epis~mic markers is ~gt)"ambaa. where the irrea~s marker is found in the ..arne sys~m as the "purprn;itn•e". wbtch in its epi>temic use LS essentiall)" dcducth·e. Palmer (2001. 142) therefore proposes that a better term for the irrealis in such contexts would be the epistemic term dubitative Thesuggestionthattheirrealis-su~juncliveinthemood
thedubttativeintheepistemicsystemishighlyinteresting,astherearemanyexan>ples of languages where '1'i~temk mar~r.; of uncertaintrldoubt function m a similar wa)· a>thesubJunctiveorthcirrealis.Theynotonlro.:curinplainpropositmnal-modali~·
contexts but also m conjunction with thn<e other modality categones that are often found "~th the irrealis or the subjunctive. In Ngipmbaa. Serrano and Hixkarpma. the
l7
ignoralil..,, dubitati~e. and uncertainty markers, respectively. also indicate questions (Palmn2001:S31f,l661fJ: (27) NGIYAMSAA guya ·ga: ·ndu dha ·Yi fish.ASS ·li>N ·2NOM ~31 -~ASl' 'Didyoueatatish?/Youateafish.ldon'tknow_' (28) SERRANO k"'a'i ta ·rn t kih~u:~i eat U1:8·PL)'OUfish.At:\: 1\re)'oueatingfish?' Thisl
This is also the case in Tariana (Aikhen•-ald 2003: 38711'), with the uncertainty marker -da which. can co-occur with the futu.-.,, negation and emphatic interrogative markers. La5tl); m Namia (Sepik: ~pik-Ramu; Feld~u<eh & hldpaU.
ne olirawomi -kale ro? you morning -of go nu~ 'Areyougoinginthemorning?(orlateron!) Thus, there is a special kind of connection between the irrealis-subjunctn-.,and the dubitatw~luncertain. Thedubitati~luncertaincanbesaidtobethemodalcounter·
partofthemoodsirrealisandsubjunctive.Thi.>indicatesthatthefoca!meaningofthe irrealisandsubjunctivet~todenoteuncenainty(i.e.propositiona!moda!ity)
Chapter2. Modohty On~
problem ~>ith the separation of the Realis-lrrealis dmmc!ion from categories suchasthefuture,negative,imperative,jussive.interroga!iveandeventmodahtyJS that the.., categories to some extent alw denote that the ewnt "non-factual. However, itmustbestres
a.
Na: sipala elema -lo ~P~ ls,; crocGdile decei\"C ·HiT -N~UT l'mgningtofoolthecrocodilc.'
b. Pleini iela pu -lo -q~ Plane tomorrow come ·FUT -lt~P '(t'veheardthat)theplanewillcometonwrrow_"
Yesu rna ila -mel pu -lo -pe Jesus again return ·SPAN.S~Q come ·¥UT -on ')e>uswil!(detinitely)comebackagain' d. Dusupi te)'U -lo -Ia Dusupi fall -n:T -possts 'Dusuptmightfall' Sosopa yele-lo -pit/ ~>-otia -It pato Centipede bite·FUT ·COND be.sore -No.,zR btg 'lfacentipedcbitesyou,it"sverysore' Theseexamplesclearlyshowthatthefutureisnotamarkerofnon-factuahtyoruncer· tamtypor>c.butonlycompatiblewithsuch markcrs.ltcanbeaddedthatWinford (2000)wentthroughthemarkingofthefutureandtheirrealisintheCreolelanguages oftheworldandfoundthat"'Thcrealityisthatallcreolesdistingnishfutureten
J9
40
.\lo&lnyondSubonhnators moodandmodaltty':Tocallthefutureanirreahscategory,.therdore.accordingto Winford."dearlymappropnate" Thesamehneofrea•omngcanbcappliedto<Mhcrcategonesthatcanbemarked a
Caddotodenoteapolarquemon,andtheuseoftheepistemtcmarkerstodenoteboth epe
Chapter!. M<>lol\ Aswuh the future,e\·ent modalitytscompaublcwtthmarkers ol non-factu.!lity. but they arcnotmembersofthesamefuncuonalcategoryThepr~maryfunctmnolevtntmodal·
ltytsnottomarkthepropoSJtionunon-aC!ualu.cd.buttoexpreS>semanucallyncher notionssuchasablhty.vohtton,obhgauon,pcrmt«IOnetcltcouldevenbearguedthat eventmodalityha.theabthtyofasstgnmgthcmattcrolestothesub)tct,e.gmalcticiary (mu.
OJtcontheother. In concluston, a diVlston of Palmer's supercategory Modahtymto more fine grained functionalcategonescorrcspondmg tothedomams proposed in Bybee & Fleischman(l995)andanarrowerdefinllionotRealts-lrreah<mlermsotthespeaker'< atmudetowardsthefactual
cenain\Yandrelatednottons.ThefactualdefiniuonofRealis-lrreahshasthebcnefit thatitolfersamorewcll-detinedfuncuonalcategor)'{propo
ahtyandthat tt belongs to prop
41
4>
ModohtyoodSubordtnotoro daus~s.
Recall that this was u~d as a main supporting argument for Palmer's (2001) analysisof!hcRcalis-lrrealisdistinctiona
Example ( J5) above showed how the subjunctive was used in a presupposed clause inSpanish.Anotherexamplefromhalianisthefollowing(Palmer200[; 122) [32) Mi
>orprettd~
che tu diw
questo_
meit.surprisesthatyousa)·.ls
The argument was that if tile subjuncti,·edcnottd non-factuality. it would not be used indau~s that an presupposed tobe\rue.lfit denoted non-ass<,.lton, ho-ver.!he use of the subJunctiVe would be more expected, since what is presupposed doe• not n~d to be asserted. Ontbeotherhand.,Palmer(J986:119)arguedthattheremaybeanotberfactorthat g<»"CTD>theuseofthesubjunctiveinthesedauses.Presupposedcornplementdauses aretypicaUysele
ha.-.anemotiveorenluati\'CCOtllponcntinth<:irsemantin(please,glad,rcgrel.sur· prise etc.). Rather than being presupposed, the comrJement dames may thus be seen as emotive ore\·aluati\'C, Palmer (1981>; l!9)argues: "the speaker or subject does not presmtthefacts,hemerelyt.".'aluatesthem".Thisisinlinewiththedelimtionofthein· dteall\'e·subjunctiveasthespeaktr"sattitudetothefactualttyofthepropostlion.The mdJcatl\•e-subjunctivedoesnotmerdydenotethatthepropositiontsfactualornon· factualbutratherthespcaker"sattitudetowardsthisfactualtty.Thenon-factualsubjunc!tvecanbousedtoindicatethatthepropositionisnotinaccordwiththespeaktrexpectations (the so called admirative; see Delancey !997)_ A teUing example is that "adnut~ which tstmplica!i~andevaluative. can alwtakecomplementdaU5esin the subjunctive inSpamsh: (33) Ad.,J/o que apwula! apremltAdmit thatlearn.3stl.PilESSUII)/t:
meaningo:>f"grudgingadmittance"(Palmer 1986: 145)
Anothcrappropriatelabelwould~rhapsbetheconcesstve
Onestrongindicationthatthesubjunctiveisusedtoachie\-.anemoltveelfectin tltesecontextsisthefactisthatthereisachokebetw«ntbeind.icativeandthesub· juncliveaftercertain factivepredicatesin Spanishaodltalian.e.g.ltorronzar>o'be shocked' and scr im:rtdible "be incredible~ depending on the degree of emotional reaction.withthesubjuncti\'eiiSCdforthcgruteremotion(Palmer2001:122).
Chop!CI'l Madollw
Another
iruer~stmg
fact that Palmer (]986) pomts out"' that corrcspondmg
daU
It bothers rn~ that ~-ou sl1ou/tfcomplain so much
Thl< kind of slrould 1s aLm used to yield a counter-~xpectative rn~aning (Palmer 1986: 120) and oth~r ~mot1ve meamngs such as frustration (the frrntrah\'C), as cal\ be seen in the followmg exclamatwe rlmr-dause:
(35) Thathesltoulddosuchathmg' Furthermore,(35) t<dearlyaparalldtotheltahan"surpme'exarnpleabove(32), lt couldtherdorebeargu~dthatthesubjunctiveinexamplessuchas(32)uusedto)1dd
a counter-upectam·e {adrnJrative) meaning. rather than to deoot~ non-a.<s<:rtion {P;almerl%6 120) kdattdk,. particularly informatwe in this regard !11 kelandtc, complement< of emOii•-e factl\-epredtcat~srnu~talsobem thesubjunctn·e.butonlym thesubjunctwe ofthemodalverbfkultt"shall'(Sigurosson20011·14) (36)
I>.Wcrgmnmtai! tunglid fktsl• brosa!'brost/ ?bros1r. it "'fun that rnoon.theshali.MJ~ smt!el =iles.su~l smdcs.tNU 'ltt
Notethatfktt/ucanalsobeusedmexdarnativedaUSt:s(Sigurosson2008: ll)· (37)
A
Theemotwo:-surpnse readmgcan also explain theC.ddoon example Palmer(200l: II. l77)cites.wher~thetrrealistsusedWlthanadmtrati\"Cprelix
(311) lrcis-ba ?;o:s,a-yi:k'awih·
ll\"C Caddo cumple abll\-e I< made dear from the fl!Ct that content (wlt-)quemotJS in
41
44
ModohtyondSubordm•tors Caddo,whJ(h31epresupposedbutnon-cmouve.3l"etn!herealt>(asopposedtopolar questions wh"h 3l"e m the ureahs, as was shown above, Chafe 1995: 354) (39) diknt- yalt!Y'"hahw-nah wbat>- 2AG R~AL· see 'Wbathaveyouseen? Chafc(l995·354)expl:unsthatthecontentquestion'"presuppose5tha!yi)Uhaveseen somethmg. and the speaker wants only to know what Jt waS' Theseevaluatwe-emonveandnon-factualexpla113tionswouldnot.howl.'Ver,apph· to
tiveclawesgenerall)·donotcontampresuppmedpropoOJnons..butratheradditional mlormanon lnfact,non-restncuverelativeclausesaregenerallyhddtobeassernve (Hooper&Thompsonl973).lfthcsubjuncuvedenotesnon-asseruon.ttwouldtbereforebeunexpectedtofindthesub)Unctwemarkermthesedauses Lunnexplamsthe useofthesubjuncllvemthesedausesa.
Det var ~n otack Jlakt nU at\ fora ovasen it was a nasty fan to to makenol(e "Whatanastyfantomakesomuchnok~'
b
Di lrade vt diskat d~ then had we done.the.dtshes then "OK.nowwchavedonethedtshe<
Literally. both thc.e dau<e< are assernon~ about the past. The speaker chooses to rcmovctheeventfromthespeechslluattunmordertobemoremduectandperbaps moreJntimatewiththehstener. tn!hesensethatwhatapphe5tothepast dD<:<not
ne.:essaril)·applytothepresent(thepluperfectin40bdounothavereleYall(eforthe preoent). Since the truth of the proposition in (40b) is ob1~ouo to the addre>See, it would have been supertlunus to assert it in the present. potentially a violation of Grice"s rnaXJmofquantit}•.Aibeitrernovedfromthespeo:ch->i!Uationthroughtheuseofthe pluperfect.(40b)isstillanassertion.though.whichshowothattheassertive-non·as· sertivedistinctionisnotemplo}"edinthiscase ThesamecanbesaidabouttheuscofthesubjunctiveinSpanishforoldinformauon. As in the case with the polite imperative atld the emotive complement clause. bothtobefoundinSpanish.thes~akerusesthesubjunctivetoremm·ethepropo"·
uonfromthespeechsituation,thepresentreality.intotherealmofthought.inthis casemordernottobesupertluous.Therei<noreasontoaS>umethatthea.>ser\Jve· non·a.>sertl\'edistinctinnplaysaraleinanyoftheseca.>e>.lnfact,mtheimperatiw case,Jtcannotplayarok,>inceimpentivesarennn-assrrtivetobegmwith lncondusinn.theweofthesubjunctiveinpresuppo..,dcornplementdausesin the Romanct languages may have several related explanatmns. It may be emottveleva.luative, admtrative, evidential or simply removed from the s~ech situa· tton.ltmusta.ls.nbeaddedthatthesituationinSpanishisrathurare(lcdaodiclS.a. wa•saidabove.anotherexample).MostlanguageswithanindicativelsubjunctivedlS· tmctionusethesubjunctiveinpresupp<><edsubordinateclauses(~oonan2007:t091f)
Examples of languages of the latter kind are Russian, Persian and German 1411 a. Karl nkuptim. dass Helga abgerdst is//
.,!~JSenbergl00 6 b:Jll)
Karl accepts that Helga depaned be-'""' be.sust "Karl accept~ that Helga has departed' (Noonan2007:109) Sotdlcju. tto [vanuctkr segodnja1·eCcrorn evening regret.Js(; (:UMP [van ltal"t'.3«•-f~T-tNU today 'lregretthatlvan"'illleaV('tonight" If the indLCative-subjuncti\'C distinctio>n deno>ted assertio>n vs. no>n-assenion. thi< would mean that preSUj>posed clauses are asserted in e.g. Russian. Peman and German. whichwouldgnagainstthetraditionalviewofpresnpposttionandasserllona.
4~
~l<>dalnyondSubonhnators
notonl)·occursafterknowle.lgcprcdicate
\amlr'andthat"indkatesthatthestatemcntisafactandthattheretsaconstdcrab(e amount of emottonal involvement br the speaker" (Irwin 1974) The same
Chafe(l995),:\lnhun(l995).Palmer(l986),lthasbeenarguedthatthisdtsttnction d~notes factuality Cactuallly'' or "realtty"J It has been argued that Palmer's (2001) t~iderdefinitionofthedimnctton mtermsofassert!Olllsconceptuall}'IOO broad Rather,theRealis-lrreahsd>stmctionshouldbeseenasa membernfproposittonal modality, u opposed to speech-act modaltty and event modaltly. Thts would explam whytbe interrngauv~. >mperJtwe and JUSSI\'e are marked as lrrealts in many languages buta
Table2. D•tfcrenllyp<•ofmodalny Sp«ch·octm<>dolity lmp
hortott•"< JU<
Propo.,l!onalm<>dahty
E>-entmodal!ly
cp,.l
d~nucm<>dohty
dynom>emod.>lny
intemogatwe
grammatical markers from the speech-act moods In this .ecnon. such evtdence has becnpre.entcd.mparl!cularas~gardsthereahs-trreal"disl!ncl!on
Tht
{sp<:«IJ-ncJmodnlrly{prop<>sitronalmodnlity{ten.e(aspectkwmmoda/rly (vmce(valence(verb))))))))
z.3 Notionsrelatedtomodality Z.J.t Complemenl-lakmg predtcates and the mdtcalt\'e-subjuncl!ve dtstmction (Noonan2007) In rdauon to modalit}; there are certam kinds of pre
Languageswnhanindtcall\'e-subjunctivedistinction{orrelatedmoodsdistinctions) almrutalwaysu,;,:thembJunctive{etc.)afterprcd!catesthatstgnalth.atthepropruition potei\Ual After propostttonal altitude prediCates. the most common mategy is to u
1<
knm•·ledge(e.g.kno~>:d~<mver.n-alize.sceandhrar)areoftencalled,;,:mifactivepred
ICates(frotnKantunenl97J).N'ormaUytheircomplementshaveafacttve(presuppo,;,:d) orratherfactualreadmg Whennegatcd.turnedlntoquesUofl5orcondttionals,however,thecontplementdauseslruethelffactualny (43) a. h
lreallled/dtscovuedthatlhadnottoldthctruth If! realll'.e/disco•;er later that I ha,·enot told the truth, I "~n confc.. u to
T}'ptcally.predlCalesofknowledgesele<:tmdkauvecornplements. When negated or qucsttollcd. how""''· they are often constructed with the subjunctwe (Noonan 2007: J30).Predlcatesofapprehens!on(fearing)areoftenconstructedwnhettherthe indkattve or the •ubjuncm·e depending on the cenamty that the event w~J occur (Noonan2007. t20Jf).Predtcatesofprete11Ce(eg.lluagme.pretmd.fool,rrick),lastly, ..,[ectcomplernentsthatarenotreai.Vnexpcctedly.however,theydonotsele<:t
Noonan(2007:127)attemptstoexplainthl'asamauerofestabltshmganalternative realnyand.lntheSpam•hcase,assenmg"1thinth!Sreal!ty AlthoughsomeofthediVts!ons.analysesandconcluSionsmademNoonan(200i) couldbecalledintoquesllon(2.25abO\'earguedagainsttheanalrstsoftheSpanish md!Cati\'e-subjuncuvedtsuncuonasassertw~:-non-assemve),npresemsvaluablcty
polog!Cal informatron about the relauonship betw~:en modahty and complement selecting predicat(s Funhermore, it olfero a fruitful dwision of compl(ment selecting predicates Howe."Cr, 1n some cases. more transparent Jabdsand e.·en more fine-grained
Chapl
divi5ions nuy be called (or. Rather than u.
Sp-eechactsandper(omlativcs(Austin l962,Searlel969)
Any linguistic investigation that is concerned with modality and subordinator• mu.
cutionary(orce.Thisissobecauseoneo(theiunctionsofmodalityistodenotc•peech act.<. lnthisbook.the specch-actnotiono(assenion has already been treatedextensivelyinregardtothecontrovcr.
of modality Thetermsperformativc. illocutionaryforceanda.<scrtionaU stemfromAu.
49
wasthattheyshouldl>eabletol>epanphras.:dintoanexplicitp
2.3.2.1
Speech·adswitltinrltefieldafmoda/ity
Speech act thcory has had a major induence within the field of modality and its terms have been applied rather freely on various grammatical categories. First. the moods imperativc,interrogativeanddcdarativearegroupedtogetherinafunctionalcategory given the label (iUocutionary) force in many fran\CW
Chop!eJl. Modaht\" wmmitmcnttolruthofproposition). Thc
and Hooper'• (1975) dt\'tston of pr~d1cates ~nd contplement clauses m terms of assertion. The present mwsttgahon dol'< not adopt Hooper & Thomp.sott~ (1973) and Hooper'<(l975)model.Smceithas~enandsuUJsquiteinllucntl3lwttbmthcresearch
of modality and because tl addresses rome of the usue< dealt "~th m thiS book. namdr thesemauucsofcomplementdallSes.ttneedstobepreo;entedandevaluated.howe•·cr Hooper(l975)dtVIde<predtcatesthatselectcomplementdausestnternuofassertion. Byassertl\·epredtcates. Hoopcr(l975)refer
thccornplementdause,Hooper&Thomp
teranceTheydonolexhortthehstenertothtnkthattheproposttionalcontenti<true, nordoesthespeakercommtthimlhcrselftothetruthoftheutlerance.sotluttfll provesfalse,slhe~
predkatcsofbehefexpre«anopinionaboutaproposmon,butdoe<notassenthattt tstrue. Thesuggesllontbattheremaybe"weakasseruons"IStbusdearl)'ltot\'fT)' plaustble Conversely, the claim that main clause< beaded by a Sf'<'Culattw predicate sucha
s•
~
,\lodaluyandSub<>:rd>notor< problematic.A.<parentheticaltagsinthelperson,thisistruc.butnotasmainclauses in3person (44) a. ShcisinBrussels,/l>dicvc b. TheinvestigatorsbdicvethattherCJ'(Irtisaccurate The proposal duu complement c~usc< of predicates sucb as my. d~im and assrr/ are "indirectus.:rtion<""i
sertthem, s/he only rCJ'(Im what someone else (the matrix subjcctl has a<.<ertrd. This isalsotruefortheweaka<serti>-eprrdicate<,onlythatinthi
Chapter2 Modaht)'
Th~ v~ry
ruson why the speaker chooses to embed such 1mportant information., m (45),JS.accordingtoCristofaro(2003:3i).thats/hedoesnotwanttobehe!drupon·
.,b]~forit,u.non·a~rtion.
2.4 Conclusion
lnthischapter.termsanddelinitionspertainingtomoodandmodahtyh.a\·ebeenlfl· troduced.lnsp!red by Bybee & Fleischman ( 1995), ot has been argued that modality is bestdiVIdedintothreecategories:speech-actmodalny.propositoonalmodali!Jand event modalil1'· Focus has been laid on propositional modality. which stands for the speaker'sattitudetothefactualityofthepropositionandwhich!Sdenotedbyeptstem· ic·e\~dential modal morphemes. the conditional. and the indicative-subjunctove and realis·urealisdistincuons.Asrcgardsthelatter,itha
ish,thesubjuncllve!SalsousedinpresupposeddaU<esinnewspaperartideswhenthe propositoonhasbeenmentionedinpre,·iouscditions.lnthisca..,,itwasarguedthat the subjunctive is used to remove the proposition from the SJ"'e<:h-siluation (mto the realmofthought)inordernottobesuJ"'rduous
Sl
CHAPTER 3
The morphosyntactic status of modality
Thepr<:>wuschaplertmroducedthenouonofmodahtyandarguNthatnshouldbest bcdl\idedtntothreedi"'nctfuncuonalcategonesspeech-actmodaluy,propo
how. One way of tesung whether propositional modality. as delined m Chapter 2. 1< a fwKtionalcategorywuhscopeovcrthewholeproposuionosbystudymgtherdatl\·e orderoffun(lionalmorphemesinrclationtothele"calcategorytheyhelongtointhe language< of the world Often, the internal order of morphemes mtrrors thm
dalltyandcomplementtzersbelongtothesamefunctionalcategorycanbeexplored further [fnot.thehypotheSistsfahtlied llybee(l985),Cmque (l9119).and Julien (2002)allstudythe mternal order of fWlctionalmorphem"' indttfercnt languages. These sun·eysWIU be presented and cvaluatedmthegiwnorder.llybo:'C"<(l9115)m•·eshgallonlind•thatwhatcorresponds to propo'
1.
Colle·antnlllolcapttol
leuer.rhetunmtheport<>f•peed>.
;6
:\.lod•htrondSubordmators beconductedtosettlethequostion.F~rst,howevor,mmetechnicalandtheorel!calis·
suesrelevanttothesubJe
).1 Morphooyntax In 1.3.1. tl was brie!ly stated that the explanatory branch wtthin Language TypoiOS)' oftentriestofindconceptual/cogmtiveexplanationsforlinguts!icuni>-ersals.Parallels haveb<.endrawnbetw..encolournanungandcolourpen:eptmn.andbetw.,.,nnoun phr~hieran:hiesandsaltenceofentitiesinpen:eptmn.ThLStscallediconietty.One SJl"cialkindof~
diStance. The~
,..,,b.
Chap<erl Themorphmyntacucstatusotmodolm· "moved" and 'adJom~d" to the lex1cal item that selected It (leavmg a copy or trace m !tsoriginalpostuon) Theresulnngclustercan.inmrn,also~mm·~dandadjoinedto the lexical item that select~d the laneroneetc In such a way, th~ hnear order ts preserved.although
Chomsky (2001 38) pomtsout, hnwe•·er. such an operation" not compaublewnh the Mtmmah
CI>rdance with the order of the funcunnal wDrds !rom whiCh the affix~• h.a•·e develDpcd. theunwersalorderoftheluncunnalcategoriesthattheseaffixesla~caltze,andtheicDn
ic-diStancepnnciple.Theexactexplana\lona
and ~-tl«tW dem~nts m a parucular language plays a pan, smce verb-final language< tendtohavcrnoresuffixesth.anverb-rnWiolandverb-mittall:mguages,andverb-!mllal languagestendtohav~mor~prefixesthanvcrb-rnedialandverb-linallanguages (Bybee eta!. \990) A snmewhat cornphcaungC!rcumstance IS th~ tact that Bybee et al. ( 1990) found that suffixes were overall rnnre prevalent than prefixes, ~v~n m verb-tmuallanguages Th!
57
)8
ModohtyondSubor
between and within the investigations that wiU he presented below. However, as Bybee's{l985)investigationshows(see3.2helow).itappearstha!theimnicdistancets generallyfollowedinthelanguagesoftheworld Sote that the model of alfixation as..
Anotherpointthatisimpor!anttomakeisthatthedassificationsoftheJternsofstudy aredependentontheana!ysesmadehytheindividualresrarchersandthattheremay beconfli.:!inganalpesnfsomeofthemorphernesinquestion.Itistobeexpe
thatarewt!houtanyexceptionandincontrovertiblyonewayortheother.NeYerthe· less, it can s\1!1 be argued to be wonhwhile to <ee what results such an inYes!igation mtghl)uld.
J.1 lhe Semantic-relevance hypothesi5 (Bybee 1985) Ashasbeenmenuonedabove,B)·bee(l985)presentsat)]lOiogicalinvestigattonofthe mternalorderofverbalderivationalandinRecltonalcategoriesmthelanguagesof!he world.Sheexploresthehypothe!Oisthatthe<emanticreleYanceofanaffixtothes!ern determmes tis relative distance from the
acategoryis,themorelikelyitistoberealiledasanaffurandnotasapartide.
Chop!erl Themorphosyntwrcsl•tusofmaod.Jity
By "semantic relevand~ Bybee (l98S: l5) refers to the degree to which the meanmgofafunctronalcategory"atfectsthelexicalcontentoftheverbstem':Bybee(l985:4) propo~•thefollowinghienrctw
(47)
V-Va!ence-Voice-Aspect-Ten~-Mood-Agreement
Thevalencecategory(e.g.tnmitive,causative)istheonethaltsmrutrelevanllolhe verb stem. as ll introduces a new ~ub-event and a new argument A> peel does not alfect thelexicalcontentoftheverbtosuchahighdegree.yetitdetermmesifthecvent •hould be viewed from witlu>ut or within. and this may have con~quences for the meaning(e.g.resultativeorinceptivereadings).Terueose\..,nlessrelevanl:itisdeictic andplaccstheeventonatimeo
ltdetermme>thewholepropositionratherthanjustthewrb;itindica!esthespeaker's commitmentto"-ards the truth of the proposition and what the speaker wants to do withttmthedi.scour~(Bybeel98S:20tf)
Thehypothesisalsomakessnmepredictions.First.themostreln
ageofmoodwa.dueinparticulartoahighprevaknceoftheimperativeasanatftx Asregardstherelah\-enrderoftheinllectionalcategorresaspect,tense,andmood, the re>ults •upported the proposed hierarchy Aspect was closer to the verb stem than tensein8languagesofl8withbotba.pectandtenseintlection.lnnolanguagewas ten•edruertotheverbstemthanaspect;theremainingJOI.anguagesettberhadten~ andaspectsuffixe.onditferent~idesofthe
morpheme• that than mood in 10outof23language'Swith botbaspectandmood:nolanguagebadtbeotherorder. Tense was closer to tile stem than mood in 8 out of20 wttb both tense and mood: one language, Ojibwa, bad the other order {with the dubitati,.., suffix preceding the preter· itesuffix;Bybeel985:331f,l96) Another intereSling obscn-ation that Bj
S9
6o
ModahtrondSul>ordmators
from the unmarked value, tended to be morphologio:aHy null (15 oot of 25 "60%), whereas the subjunctive tended to be morphologicaDy overt (i out of 8; Bybee 1985' 53). As always in linguisticinvestigation.. therewerecortainde!imtional problems. withcategoriesthatpartlyo•;er\apped lnsomelanguages.moodpartlyoverlapped with the future. e.g. the potential or mtentional in Kiwat, Pawnee and Zapot..:. In these parttcular cases. Byb"" (1985: 156) argued that modality was the pnmary function. the futurebemgasecondaryone.Aswasshownin2.3,thefutureosacategorythatisoften assocoatedwiththelrrealis.probablybecausethefutureentatlsuncertamty.hshould also be added that Bybee found that the future occurred. on the sideofothertense markersinmanyotherlanguages.soshesulldecidedtocountitasatensecategory rather than a mood one. The explanation Bybee (1985: 194) offered for the partial overlaps between the fwtCtLDnal categones was that a grammaucal marker denoung a certamfunctionmaywtdenirsmeaningintoalsodenotingafunctLOnbelongingtothe adjacentfunctmnalcategorymthehierarchy.Thus.thefuturepretentoftenhasacon· ditlonal use {e.g. m the German' and Romance language._ Tiwi, and Sterra Miwok). Thecondttional.mturn.hasanoptativeuseinPau"JleeandNahuatl.andthesublunc· tn·ehasantmperativeuseinBasque,MasaiiandTarascan lnsptredbyf"Dley&Van\'alin{l984).Bybee{l985: 1691f)furtherm\..,sttgated whether mood may till t"'"O separate functions. one illocutionary fon:e function {i.e. speech act modality I and one truth-commitment function l!k>culionory fur
The md~eati,.., Bybee saw as a complex category whteh can either mdteate that the clauseisa""dedaratiwassertmnoftruth":orthatthespeakertscomnuttedtothetruth of the proposition ln>om
categonesprobable,potenllalanddubitallwtogetherl'.iththesubjuncllveandcond•·
}.
As was ,.,d 111 footnotol ch_ 2. sh< SOII.l
wh•
t9fl\l)
Chop1er3. Themorphooynto
the>'Orb"alfectsthemorphologicald15tancefromthe5temandthemorpho-phonolog· icalfusmntothestema.regardsthefunctmnalcategoriespertamingtothel-erb.The reoultsdO\'IatedlTomherpredichonsinthreeways.Fmt,tensewa>le>Smorpho-pho· nologicallyfusedthanexpected.Second,thorewerecenainol'erlapsbetweenthecategones. Third, mood behand differently than expected. The Imperative was realLZed asanatfixmorelTequentlythanexpected,andthemterrogatlwhadaditferommorphosyntacti< distribution than the other moods. How.-ver, it could be argued that By· bee'ssamplewaotoosmallforthe>edeviatmnstobeconsideredsignilicant,especially asregardsthemternaldistribul!onofthernoodrnarkers.Mostofthelanguagesthat had mood as an mHectmnal category onl)· had two mood markers. 3.1.1 Asse>Sment of the semant!C·rdevan<e hypothesis Onthewhole.ftybee(l985)supportstheukathatmodalitytsaperipheralfunctional category,withscopemertenseandaspect.lnthatrespect,ltcouldbecomparedto oubordinators,complementizersinpart!Cular.Bybee(l985)alooexploredthepossibilitythat"mood"muldbedi>idedintotwofunctionalcategones,onethathastodowith illocul!onaryforce/speechactsandasecondthatpertainstotruthcommitrnent.That isalsowhat,.arguedunderthepresenthypothesis..althoughitisclaimedthatthelatter categoryhastodowithfactualit)·(prop<»ttmnalmodaltty),ratherthan"degreesofas· oerlton"(see2.2.5&2.3.2).AlthoughBybee(l985)didnotfindmorphruj'IItacticeYi· dencetosupporttheseparation,>hedidlinde>·idencetosuggestthattheintermgahve was a more peripheral categorythanall othermodali~·categories. Thatthe"dedarati,..," functionoftheind!Catincouldnotbeseparatedfromthemd!Catlve,sub)unctive,conditional, and epistemic mood markers is in hne w1th the present ~'JlOtllesi>. which
61
argues that declarative is not an sp.-«h·act modal category but merely the unmarked epistemicYalue. Thestatementthatthe"dedaJlltive"functionoftheindicatl\-eCon· trastswitlltlleinterrogatiwdoesnotseemtobesupportedbyherdata A!.reg.ardssubj..:tagreement(personandnumber).thepresentinvestig.atmndoes notadoptBybre"syJewthatJtshouldbesernasafuncllonalcategory.lnmanylan· guages.e.g. the Romance languages. subJeCt agreement is often the onlymorpllologLCal markerofthesubject(thesubjectpronounbeing""droppN").Sincetllesub)..:tisan argumentofthel·erbratherthanafunctionalcategory.itisreasonabletoassumethat agreementaffi.xesalsolexicalizethLSargumentthroughanagree·relation.ratherthan lexicalizmganotherfunctionalrategof)·.lndeed.asChomsky(2001: 138tf)argues.a functionalcategorycontainingonlythesubjectagreementfeatureswouldbeuninter· pretablemthefunrtionaldomainoftheYerb.h""notonlytmghrnotextst.but«lllll<>l eXJst on ratherplausJbleassumptions" (Cilomsky 2000: l.llltf). Put ditfermtly.thesub· ject"snumberandpersonfeaturesarerompletelyirrdeYanttothepredicateandthe propositionandsllouldthereforenotbeseenasmodifiersofit.Subjectagreementis nomoreafunctionalcategof)·oftheverbtllanthesubjectis.ThisleaYesmoodasthe outmrutfunctionalcategory.
H
Thesplit-inlle.:tinnhypothesis(Cinquel999)
Cinque(l999)isageneratJVeandtypologi
Hsurel. "lhounil·eJ">;J]orderoffunrlionolrategorie•bclongingt<>lhcrnn«tionoldo mainaccordingtoCmqoc(l'l99)
Chopl
are se1·eral points to be noted here. First. Cinqu~ proposes that "Mood ~rreahs'' to whicll he subso:nbes speculath~ panicles such as per!J~ps and the moods rndicalive, subjunctive, reahs and irrealis, should be plac~d closer to th, dause centre than the temporal categori~s past and future (but funher away than antenor). This is because temporaladverbssuchasotiCcandrltellcanbeplacedbeforedausaladverbssuchas perlmpsinthelanguageshecxplored(Cinquel999: 136,152).Second,hedrawsdis· tinctions between "mood irrealis': epistemk probability ("episremic"), epistemic ne-
n...r~
cessityandepistemkpossibUity.ltshouldal«~beaddedrlutCrnque(l9991proposes
that event modal categoriessuchasvolitional. obligative, abthtative,and pernus"ve arelocatedevendosertothedausecentre.insidecenainaspectcategories(habitual, repetitiveandfrequentative) Crnque'sindependentevidencetosuppontheproposalthat"moodtrreahs""a morecentralcategorythanthetemporalcategoriespilStandfuturetsthatrnSamoan (Polyneoian: Central/Eastern Oceanic) and Ndyuka (CrMle; Huttar & Huttar 1994:519)theirrealispartidecomestotherightofthefuturepartideand,inSamoan, loth~ right of the past particle: (48)
(POL¥N~SJA": CENTRAL/EASTERN OClANIC)
SAMOAN
lava pe a sci e aluatu e m lelei but 1MPH ~-"'PII t"HR HIT tRR 2"; go DlR ta" eat weU 'ButifyougoandeatwelL' (49)
(CRWLE)
NOYUKA
I
be o
sa poi
Jfu)ll)'ll"ete' be.abJe for eat yet 'Wouldyoohavebeenabletneatret?' youANTfUT lRR
Furlh,rmore, Cinque (1999: J~Stf) identifies clause adverbs wch as_fra11kly and /1011· osllywith the nouon of speech actllndcr "Speech ad' he also mdudes the mood.. de· darative,imperattVe,andinterrogative.Thefunctionalcategory"Speechact"hasadi· reel parallel in Rizzi's ( !997, 2002) assumed category "Force·: which Rizzi locales even funh,r out in the clause periphery (in the Complementizer domain: see 4.2 below). However,Cinque(l999Jargucsthattheplacementofclausaladverbssuchashomsrly andfrmrklyindicatesthatthec.ategoryisal
).).1 A..e<>tMntoftheoplit-intlectionhypothesis Cmquis(l999)investigationisanimportantone.sinceitopensupthrposstbllttyofa more tine grained structure of inflectional categories within the framework of Generaltv• Grammar. It also has the mength that is uses data from a large \'artet}· of languages. The relevance nf Cinque (1999) to the present inv«trgation is that it makes a mnnectmnbetweenthemoodsindicativc,subjunctivc,realis,rrrealtsandpotentral
6J
64
ModahwondSubordm•tors morpheme< m(has pahaps. Furthermore. It draws ad1stmctton bet....·een what Cinque (1999) •alls '"speech act moods" and "reahHrreah< moods", somelhmg that IS also donemthepre<entsmdy.However,ital
cnnstuuleortobepart..ofddferentfuncunna!categnnes The"realts-ureahsmnnds" areevenp!aced"inside"oftensc,whicharguesagamstthehypotheststhatgenera! subnrdinatorsshou!dbesubsumedunderpropoSJUnnalmndalJI}'.SJncegeneralsubordtnalor<arguablyha>'Cwtder scnpethan ten.<e ThescconduS~onscan be seriously quesunnedthough,whtchwJ!lbednneinlhefo!lowtng First.itmustbeacknowledgedthatmatleastN'orwewanandEnglJsh.theorder ofclausa!adverbial<areopenforvanauon.Cmque(l999)offer
Wewlll•!probablyptr/rapscomeatetght Theeventswill'!probablyp.>ssiblyoccur
ltisthereforehardtomonvatethecalegnncalseparanonbetweenper/mps.probably andpossrbly Theyreprescntditfereotcpl
Chap~r3
Themorphosyn1actK•1aiU-
One reason for regardmg th~m as dllfcrent funcuonal cat~gones os thatthcj•can form compkx teo,.e• m som~ languages. e.g. future in the past. Howev~r. I~}' aU fill the same function. to place the event on a umdinein rdation to a rderence pomt. Indeed, m many languages, on~ morpheme IS the unmarked \'3lue (typically the pre~nt) and maydenoteoneoftheotherfunctionsaswell(ineg.S"<edJSh,thepre~ntcanalso
denote the future) Secondly,asregardstherdationbet,...,.,nternporaladverbssuchas onceandthmmdthep<JSltoonofthefunctionalcategoroespastandfuture.lll~lmpor
tantto note 1bat the lemporal adverbs can also be place
ThelastpointiSeasily
conductedonGoog/em11kwuhEnglishpagesonly. lnordertoobtaintbetemp<Jral meanmgoftitenandonce(thcyalsobaveotherrneaningsnotrelevantbere)andto assure that the adverb1als were clause me
"woUthenperhaps' "woUperh•p•then'
tOto tlSO
%"wlllperhapst~n·
-..,uthrnnee<s=tty "woUnoces>arolytben .,.·.,,unec•"•nlythen" 'woUthenposs:obt)'' ·.,·,Upo>
24.,
"",.,
"wosonceperhaps" 'wo•porhopsonce" %"wasperlupsonce" • ..·o•onc.ne<;<ssont)' "wosneces.anlyonce 'io"wa.<noce,.•nl)onc<" 'U·o•on<epo>
[50]
"!."""'P""'bl)'once"
'"'
710
6s
66
:\lod•htyondSubonh11.:1.to" S.Ome examples at' perhaps, ltfcwari/y and possibly before o11ce and r!rw (53)
a We expect to see an intermediate period where Amencan carmakers otfer HPS sptems but they will pa!rnps then try to reduce the fuel consumption by ~me additional measures b. Sucl'ldistortionsoftl'leMTdata 1--1 canreasonablybetreatedasasto· cl'laslicproces.
(54) a. They lie on a tomb chest with ba,;,:s of pillars at each corner. indicatmg thattherewasperhapsonceacanopy. b This is notta say that there was t~<:ces.sarily mtce an original modd Chilam Balam and a mndeltitle Thts was pos.sibly Oil« an old tinnds blowing House and ts situated most appropriately in the vaUey known as D«p Swmcomhe (valley of the ptgs) The results show that
"""'is equally often placed after prrltap> and pos.sibly (11ccenarily
~ieldedtoofe"•hits)asheforethem.Asforrhett.itcanbenotedthatpcrhapsi>placed
beforett>liglttlymoreoftentltanafterit,whcreaslleces.smilyisplacedbeforerltcllm every fourth case. Only pos.sib/ycan definitely be .aid to be placed after rltw.lhe situ· ationisbynomeansasdear-cutasCinque(i999)mightsuggest Cinque"s(l999}independentevidencethatpastandfutu..,aretno..,peripheralfunctionalcategoriesthan"moodirrealis"canalsobcdisputed.TheanalystsoftheSamoan Example(48)isquestiontdbyPratt(l984:20.262)whomggeststhatsit(set)denotesa mild tmperattve. MortOver, Ct.mpbcll (2000: 1438) classifies,..·; a> optative. These dasstficatmns would t>etter suit the meaning of (48). whkh appears to he a suggestion Theanalysioot'theNdyukaExample(49)canal
ChoplerJ. Thomorphosynlocl>e
Cinque'<:.!dyukltnampleabove.theyareused in
E,·rlijk. dit worstd i< onacccptab.:l llflnestlythi< proposal is unacceptable 'Hone
b. 'Ecrlijk i~ dit voorstel onacccptab.:l honestly isthi< proposal unacceptable
Thcsc: facusuggestthatspeoxh-act ad•"Crbialsareadjoincdto the claus.:. rather than heingpartsofthemostperipheralphrascoftheintlectionaldomain Second.thea«umptionthatthedeclarativeandtheindicativebdongtoditferent functional cateGories was disputed in Chapter 2. The declarative should be<\ b.: <.:en as the unmarked epistcmk value (Palmer 1986. 2001). Third. as regards the propos.:d functionalcategor)'""Speech-aclingeneral.itisdilliculttos.:etheadvantageofas· sumingthatanotationaUyidenticalfunctionalcategoryisrealizedintwoditferent position
67
68
ModahtyandSul>ordtnata" declarative, indicative-subjuncti>~, and epistcmk-evidential modality onto seven d!f· ferent functional categories. and placing some of them "'inside'' pa>t and future, these shouldbeconsideredmembersafthesamefunctionalcategory.propruotionalmodalil)'· as m Palmer ( 19!16, 2001). This nid, Cinque (1999) .rill has the benefit of dividing the inflectional domain into a more fine-grainnl mucture. Furthermore, Cinque draws a di>tmctoon between s~ech-act modality and Realis-Jrrealis, something that is alsodoneinthepresentinvcstigation.La
conceptton th>t inflection on])' belongs to the "inflectmnal domain': and not the "cornplementizer domain': Jntlectionandcomplemeouizersare different forrnalsptems that work on different lewis nflanguage (syntax and morphologj·). Howewr, they can arguablydenotethesamefunctinnalcategory.evcnsimultaneously. Therearenumerousexamplesnffunctionalcategoriesbeingdcnotnlsynthetkallyandanal)1tcaUyat the same time, especially when there IS a shift in a IJnguage from a syntheltc to an analyticalsystem (e.g.thegenerallya.sumedreplacementofmood intlectionbymodal verbs m many Germanic languages). There is no restriction that the intlected \'erb must belocatedinthesyntacticpos.itinnassociatedwiththefunctionalcategorythattheverb ismllectedt0r.Chnmsky(2001)spcculatcsthat\"<:rb-raismgiscondJitonedb)·theaf· tixalcharacterofinllectionalcategorks;howcver.theverbmu•tnotbe raised in all languages. Wbentheintlectedverbisnntraised.tbes}'1li.1Ciicposinonmayalterna· lively be ltxicalized by a particle. nr in the Complememizer case, a complementizer
J-4 Syntactk heads and woed formation (Julien 2002) ]ulien(2002)isanotherlanguagetypologicalandgenerati\·eworkthatmv.,;tigatesthe mternalorderofderivaunnalandinflectinnalcategnne•pertaimngtotheverbandthe clause. Wherea• Bybee (1985) bad a sample of SO languages.lulien (2002) mvestigates themorphemeorderinnolessthan 530languagesfrornl80so-called"genera':Languagegenus1Sasmallerunitthanlanguagefamil}'(inthebroadsense,e.g.lndo-European and Uratic). Following Dryer ( 1992). genera are defined as "genetic units with such a highdegreeofinternalstmilaritythatthe\lllidit)'Oftheunitisuncontrovemal"(fulien 2002:41).Themajnrbrat1Chesoflndn-Eurnpeanareexamplesoflanguagegenera.The
Chapter)
Themorphosyntactlcst•tusotrno&ll~·
dmsionofthel:mguage.mvesugatedmtogenera,.donemordertoavotdgenenc btas.Allmall,fulten'sdatacouldbeconsideredareprrsentanvesdecl!onoltheexmmglanguagesoftheworld lu!Jen(2002)partlysupf'(lrt.
lncontra
·MAl<J)'
-tya I•IJJantyuuy
-PAH·N~O
·SJY
yi
-1nl
IRRh3\'(ll\(TCY-PH'l )'OU-IJAl
'ldtdn'tmtendtnshowyoumercy' Julten
In accordwlthCmque(l9'99),/ulienplaco:sepistenuc :md ev1denualmodality out.<Jdeofte1Ue.She(2002.2&9tflgi•·esthreeexampleswhereep~>lemic-e>idential
morphemes comeout>1de tense morphemes {l.czgian. Wtchlla & Warao): (58)
(HaspdmathJ99J·)48) Bakud-a md mm guUedt-z aijud -ua -Ida Baku -!N~>S seven man.A~sbullet ·PAT takeout -A<JR -~\'lD 'lhesaythatmBakusevenmenwereshot'
HZU!AN
69
70
ModohtyandSubordm•tors (59)
WJ<.HllA khJ?as·kJ)"il· 'Y k1- hirahr-re/cr/a-s poor- QUOT-JNOEf.SURJ·AUR·ground-hck
(RoodJ976·267)
(60)
W>\IIAO (Rom~o-Ftgurr~J985: 129) Warao-tuma amh~·hLiu nahamutu anu ka!Ju -ya -ymna Warao·PL before-SlWULSky O~LQ]h•e -PRhS-H"ARSAY
"Thepooron~•~r~!ickingthcground"
"ThesaytbattbeWarao,long.longago.lh·edinth~sky." A<saidabove,]uhen(2002)apparentlyd~snotcountthemoodspot~utia!anddubJ
tauvcascpJStenuc In funher agrffillcnt With Cmquc ( 1999) but also partly wuh By\>« ( 1985), Julien (2002: 100) separ~s th~ dcdara\1\'C from the mdtcatJVc. md put.> dedaraU\'C together with the sp«ch-:u:t moods. To tUustratc thl<. she P"'
I
(KHOISAN;juhcn2002:J8J)
H
mrw·pt ·hlnt
the}"DE<:LRUiPAHS~
"Thcysawhim' Onthcothcrhamt]ulien(2002:273,289)alsnprcsentstwnl.anguagrs.Gr"'nl.andic(from the same family as liluplaq) and Hua. wherc the md.kaU\'C is SJruatcd outstde of tense (63)
L)Jt""NLA"OtC (fnrtesquel984:)20tf) lppa
(64)
Ht.JA(EAH""WGCtN"AHLI;HLA!S:MAtNn<:TtO".TRAN>-Nl"' GCINU;
Haimml98D-236) Kamant" -Kt' past kzo -gu -e Kamam -LOM letter wrl!C -t"Ul ·11<0 h<; "l wtU make Kaman! wrtl~ a lell~r
Chopterl Themarphmyntwtcst•tu•otmodobl!'
LikeCtnque(l999),Juhen(2002)partlycon
guagc< areShona, Turkish, NonhSaanu. Kan1nro.LnwcrGrand Valle)·-Dant. Gnnniyandi,h'lupiaqandHua\<e: Sl•m~a(Bnnll• /lloger-Congo).AsmmanyoftheBantulanguages,Shonahuafin.:J moodsuffix.euher-cforthesubjunctive,potentialandnegauwor-afnra!lotherfuncnons )uhen,mosthkelycorrectly.analysesthr<eftnalmoodsuffixrsas l'('a]!'andtrreah<. Shonaalsohasten<eprefixe<.AccordmgtoMyer~(t990).the prefixe<mShonaarele>
71
7!
ModalnyondSubon:hnator. an' pan of the stem, whereas the prefixes are more chttc-hke_ Furthermore. the internalorderofthesullixcsmirrorsscope.sothatforanysuffix,tthasscopem•er thesuffixestotheleftofit.ln agreemcntwiththegenerattvehead-movcment modeldescrib,dm3.l abmoe,lulien's(2002 l9llf)analymofthesefuctsisthat thevubhasfirstbecenc}'dicallyleft·ad)oinedtothefunctionalutegoricsrepresetUedbythesullixesunulnhasre~ched"grarnmaticalmood':Themorepenph
eralfunctJOnalcategoneshavethendiududtotheverbasprehxes.Thus."grammaucalmood': as represented b)• the final mood suffixes. JS a more central categorythantensemthislanguage Aswassaidabove.thepresentmvestigaltondoesnotadoptthegenerative "head·nJovemem" explanation of affixation In agreement wtth Chomsky (2001) andB)'b,eetal.(l990).alfixauontsratherseena.saphonologtcalprO<:ess.That. however.doesnotalterthefactthatthesullixesinShonaarernoretiedtotheverb roolthantheprefixes.Thesulfixcsaremorephonolog~callyreducedandmore
incorporatedwtthther(}{II.Thequcsttonisonlyhowthtsshouldbetnterpreted 1\otethatapart from the mood sullixe<descnbedabm... Shonaalsohasa potentialandasubjunclit'eprelix(thelauerashightoneonthesub]Cctagreement prefix)Theseoccurtogethertvnh-e(M)'ersl990) (65)
anga-vereng-e shelhe-Por-read -e 'Helshecouldread.'
Thtssu!llleststhatatleastpamof"grarnmatlCalmood"havewtderscopethan Julien{2002)argues.NocealsothatthepotenualprefixiSmorpholog~eaUyheavter
thantheirrealtssullix.Sincetheprefixismoreindependentandlessfuscdwith thet'erb.a<M)·ers(l990)statesthatthepreftxesingeneralare,tlcouldbeargucd tob,ab,ttermdicatoroftheoriginalfunctionalposittonrelanngtoboththepotenualprefixandthetrrealissulfix.lnasense.-<"hasbecomeamarkero(,•erbal agreementwnhpropostUonalmodaltty,whereasthepotenualrepresentsamore
lnterestingly.mtherelatedSantulanguagesGtkuyuand.Swahth,correspond· mghypotheucalprelixesaremuatedtothele!tofthetemporalprefixcs {Welmersl973:36[)
(66)
nga- lo k1tal;u- HYP- PAST· PAII.TPI-.HSS~· ~O.:T'ifyouhadlookedforn' {67)
(WaldJ990·299)
tai(U\'U
ngi[!;(i- HYP-
Jut search
k~ ~UT-
e~tdl l!>I>H·
'ifJshouldeverseU[n['
seJJ
Chapta3. Themorpho5)TI1a
lncondusion,SIIonacannotbcconsideredardiableindtcatorofpropositmnal modalityasamorecentralcategorythantcnse. Tr~rkish. According to lulien (2002: 290), Citing Brendemoen & Hovdhaugen (1992), Turkish. has a p11tcntialsuffix -cbil (-Abil) wllicllts situated dooer to tile
verb ro11tthan tile tense suffixes. The primary meaning of this suffix. however, is
not to express epitemk possibility, but rather dynamic ability (Johanson & CsatO 1998: 213, K11rnfilt 1997: 376). Nevertheless. Kornfilt (1997: 376) reports that it canalsobeu.edtodenoteepistemicpo«ibility orada o/-abil Hasan there bc-A~tL 'Hasanmavbethere'
(68) Hasan
., ·At>M
In that respect, Turkish is an indication that propositional modaltty,.amore centralcattgorythantense Ontlleotllerhand,)ulien (20fl2: 336)alwnotesthatTurk.ishha.aconditional suffix which is situated outside tense. Thi< ratller indicates that propost· tionalmodalityi
read -H;r "if! will read'
·Y!
-lw
Therefore,Turkishcannntbec!lnsiokredadearindicationthatpropositioualmodalit)'isalessperipheralcategllrythattense.ltisequallyanindLCationoftlleoppositc
North Saam1. According to Julien (2002: 290). the potenual mood is sttuated clos· ertotheverbthantenseinNonhSaami· (70) mii wahkelt-eatza·i -met da -i -d olbmu -i ·d WC h.eJp ·POT ·PA.H ·IH that ·PL ·ACC person -PL ·ACC 'Wemight(llave)helpedthosepcoplc' According to Sammallallti (1998: 7i). the potential
73
74
Mod•hty.ndSubordtnato" .><:opethant~n~.Onthentherhand.thereisalsnaconditionalmoodwithtv;o
markers. -st<· and -wtcte-, which stand in complementary dtstnbu!JOn with the t~nsts. Furth~rmor~. many Saamic languages do nnt have a past potent tal. so that thepotentialisincnmplem~ntarydimibutinnwitbtcruc
Kamoro (Asmat-Kamoro: Cemml mtd South New Guiotea-Kutubmt: Trans-New
Guima). According to Julien·~ snurce of information (Voorboeve 1975). the cnn· dJtinnalmoodissituatedinsidenftcnscinKamoro(Juhen2002:290) -ken! -nko ·IIIII -n em ~REf ·gl\"e ·(:(1)<[) ·HNSE ·)
(71) apa
Ontbeocherband,Boelaars(l950:98)repnnsthatKamoroal•oha•anirrealis cucumfixmapa-•etbat isuscdincomplementdause•after•·erb•ofcognition. The cucumfix is added to the full)' tense- and pennn-mAe.:ted verb. whereby the •ullix pan -"c is situated outside ten.<e.lhe situation in Kamoro i• thu• more complex.anditisdoubtfulifKamorncanbcusedasanargumentforanyorder. Lower Gnmd l'allq-Dani (Grear D
·/! ·ik J.:ill ·fACl ·R~M.PA.H ·hQ llulltd(it)alongtimeagn."
Will
On!hcolhtrhand,theanalysi.
the gerw1d form (and perbap~ also in the iteratwe parhciple). It i5 only w1th a certam amount nf resen·atinn that loii'OT Grand VaUey-03m can be con... dered anindicahontbatpropositionalmodalityi
Chapterl Themorphmyntacttcstatu•otmodollr.·
(73)
ngab
-Ja
-w•
-la
'!want to eat' However, according to McGregor (2002· 364) the verbal system m GooniyandJ canbesaidtoheonthebordube"'•tencnmp!e"verbconstrucbons,cnnststingof umntlected laical ma.m verbs and auxihar)·-hke functional verbs carrymg themfkction.and grnmmaticahud cnnjugattonal class systems. In Goomyandt,there ts aclo.«
these"dassdieri:asMcCiregnrcallsthem.isthattheyareln<:atedtnthenghtof the mood md ten>e affixes. In the example above. the dassdier" the morpheme -~m-la(actually-/'l"LN0.\1' t-a'e,.tend'J S!ncethetenseandmoodaffixesare affil(Csoftheda.>
~t:'
nga)'l ngag' guwu- OJ· neg eat IRR· 312M;· 'Donoteat1'
At:X ·HITIRR
AccordlngtoRumsey(2000),thereareten<eprefixe.,irrealtsprefixes,tensesuflixesandtense/trrealtssulfixesinBunuba'Oneofthetrrealtsprehestsga(cf]ll mGoonyandt).Theluturemdrcauveprefixi
(inonyandicanhardlybecnnstderedanindicahonthatprop!tstltonalmodalityi.s amorecentralcategorythantense /fitrpilUj (lmrrt BJumo. Eskimo-A/eul) lulicn,ctnng Mclean ( 1986) states that the indicati~-e mood ts 5ttuated closer to the verb than the past tense m l~upiaq (lulien2002290)
The ten« and ur~ab.< P"'hxes ore tn compkmenlor)' dutnlmlton So too arc the trreoll'l Wn<eandtcnsosoffi:u• Bothmthe>ndtcallwondtheureahs,thetoto,..,olhx>Snull
4•
7!
76
ModohtyondSubord1natou (75) spinach ·tu -ilatu spinach -eat -Jike·ISI.I 'lhkedea!ingspinach' On the other hand, Mc~an (1986: 32tf) a!«> reports that the future suffix -mll<j" situatedinsideofthcindicatil·csullix-tu/ru (76)
11igi
·lliaq ·IU
·q
eat ·FL'l·ISI> -3st; 'ShewiUeat.' The Ialter is. as was shown above, also the case in Greenlandic, which belongs to the same dialect continuum as lflupiaq (lulkn 2002: 273). Furthermore. Yupik (Esk1mo-Aleut) appears to have the overall suffix order V-T-M, where M among oliler things consists of tbe indicative and T mdudes 1he past (Mithun 1999:409).
tncondusion,i!isdoubtfuliflt\upiaqcanbeusrdasanindicalionlhatprop· osilionalmodalityisamnrecentralcategorythantense.ltisequaUyanindication of the opposite. H..aw (H11awmr). According to Julien's source of information (Slam; & Erickson Hollenbach l%9),tlledubitativeissituateddnsertotheverbthanthefuturem Huave(Julien2002:290): (77)
ap- kc-
rna-
mng
FUl'·UL'B· )St;· do Ma)•behewilldo.' However, Noyer (1993), based on Stairs & de Hollenbach (1981) and Suarez (1975),statesthatthennn-pastaffixmandpastaffixtattachdire
Chapl
~In
-11<1
-~
-ir -11!.
za
wa .z
COIRO ·AOR ·N~(; ·PST ·(:O:
ewer gu -da -j call give -f\;t' -PPl
-wal ur 'lfyouhadn'tcorne,lwouldhavecalkdyou' As regards the declarative, it was condudN in 2.L4. in accordance with Palmer (1986, 200l)thatitisbfst«
n
foraseparationofthe interrogativeandtheremainingmoO
3·5 Two typological surveys of the morphosyntactic status of propositional modality Theprevioussectionsprcsentedandassessedpreviousresearchonthesyntactkstatu< ofpropositionalmodality.ltwasconcludedthatamoredetailedandtargetedsurvcy wasn«dedtosettlethequestionwhetherpropo
beconfik1inganalpe
Within propositional modality are included (ategoricssu(h a< the indkati\·e. subiunc!i\'t',reali<,irrealis,a.<sump!i•·e.deduc1i,•c.probable.po!emial(
isalsoempirkallyhighl)'que<1ionableifoneandthesantelanguagecanbecomidcred of evidence for both orders. U
3·S·I Survey of the internal order betv.'t'en propositional modality andtenscinlulien's{2002)530languages Julien(2002)canform!hebasisofamnredetailed~l'"logicalin'·estigationofthein
ternal order between prnpositional modali~· and tense. since she lists tbe morpheme
ord~roftheS30languagessheinvestigatedinanappendix.Thefocusof[ulien's(200l) inv~stigation
was on tense and aspect mood markers are often not spec1fied but "m·
pl)·labeU~dM(formood).lheaimofthepre<ent
languagesthatarelistedwithMorrelevantmodalitymarkersoutsideorinsidetense, •-~- to consult the source~ and dettrmine what the grams are Since the language~ dealt with in 3.4 were pan of julien's inventory. these must be mdudedamongtheresult~. Julien (2002,330-356)listsanother 13langu.age•"ith Mor.-..levantmodalitymorphemesinsideoftenscin.-..lationtotheverbandanother 33withtheoppositenrder.Puttogcther.thereareonl)·60language•ofthe530she m.-.,sllgated (I L %) that are.-..levant fort he present im•r
ed m 3.4.1, and wher~ two w~re confirmed to have markers of propoSL!mnal modaltty inSLd~ of t~nse, julien lists anoth~r 13 with mood (M) or relevant mo
Dongole.el\ubian{Nubian: Eastern:Ea.
In contrast, four were in fact shown to ha,·e markers of prop051!ional modahtyoutside oftense.Thesewere: I Magi (Mailuan: Eastern Main Section: Trans-l\ew Guinea) 2. TLWL (Tiwian: Australiatl)
Ch•pterl Themorphmyntae!Jcst•tusotmodoJm· 3 4
Niunat/Dmdaht. Nootka & Makah (Southern Wakashan) UpperChehahs(T=nosanSalishan)
Apart from those that have already been treated in the preVIous secuon, Juhen (2002 330-356) list.> anoth"' Hlanguages with mood or relevant modaluy markers ouutdeDftense,mrdatmntotheverb fDrthrttDfthese,theJdenlltyofthemDod markerscDu)dnotbeestablishedmthesDurcesconsulted Anotherse•oencouldnotbe usedasarguntents!oreuherorderaccordmgtothesoorresmnsulted.Either,modality was m complementary dtstnbution \\ith tense or 11 was mtertwmed with 11. The rem.aining23lanGuageswereallconlirmcdtobavemarker.ofpropm•honalmodality outstde of tense However, two of these (Abklw and Karbad) belonG to the .arne genus.andweretherdorecountedasone.The23languagesarepresentedindetailtn appendu:A1,bntarehstedhere: 1 2 3
Ma
81
6>
.\tod.altlyandSubon:hnator> Greenberg (1987) included Chiqui1o under Macro-Ge. bu1 on que5Uonable ground•. i1 seem•. Rodriguez ( 1999' l65tf) doe• nol mclude 11 under Macro-Ge When1helanguages1ha1w.:relrea1edm3.3.lareadded.whichwereal•op.arlof )ulien"•(2002)mvesliga1ton.1heproportmnsare7•4+22"33languageswi1hmarker> ofproposttionalrnodalJ1you1sideof1ensecompared1o2t2=-4language•wtthmark· ersofproposi1ionalmodalityinsideoften5e.lnpercen1.1ha1meansJ3/37,89.2%.Jf oneadoptsthea.. umption1ha1uniYersalca1egorieshaveacertainumversalorder.J1i• almostnine1imesmorelikely1ha1proposi1ionalmodali1ytsarnoreperipheralca1ego· 1)"1han1ense.than1heopprutte J.S.l Independent surYey ohhe mternal order between propositional modality and tense in the languages of the world Jnordertodrawa.:onclusion1hathasany•1a1is1icstgnilicance.anindependen15urvey mus1~conduc1ed. Foran inven1of)"of1helanguage•and language famihe•of1he world. Etlmologue.(om has ~en consulted.~ goal of !he sur.-q has ~en 1o ob1am descrip1mns from "'"f)" language famil)· in a wider sense. and ideally from ~•ry lan· guagegenus.excep1for1he38generaalreadycOYeredby1helastsurvoy.Arestric1ion !hat has nar"""'d down the sample has been !he requiremen11ha11he)· must have a modern. we51ern grammatiCal des.:ription. For some language• m !he more remote ar· eas of !he world. !here are only descriptions made by le" grammatically oriented an· 1hropologi515orlaypersons.EYenwithou15uchares1nction.however.alargeportionof !he language• of the world cannot ~ included •impl)· because !hey lack obtainable grammahcal des.:riptions. Many languages in !he more remote areas of !he world do not even have a wri11enlanguage ...... nle5Sagramm.a1JCaldescnptton.lnparti
ChaplerJ Thomorpho•rnta
n:strktionthattwoormorelanguagesfmmthe
Some of the data come from the handbook< ill Romled~ Langu<~ge Fttmt/y$cries, whkhcomai!lrepresemauve:<elcctio!lsoflallguagespectficdesutpltonsofllloftbe language familiesol the"·orldorgcnuslcvdbnnchesofthem Other senesan: Mol/lou
Grammar Llbrarywhtcb comam grammaiLcaldescription of less well-known la11guag· cs and P<1dji< Lingtuslin which contams language-spectlic grammatical descripuons oflanguago:
tbatbavebeenused.whtchcontamlanguage-spedficgramma!JCaldescnptionsoflanguagesmacertain,..,gton.e.g.HandbookofAwtmlimtUmguages.Thenrdtgenowlm•g"ages of tile Cau~asus. A Ham/book of Amrrt~Oit lndialt Languages. The lat~gU
osiuonalrnodalitywas>Jmatedoutstdcoftenseinrelattontothe\-erb.andfourlanguago:< "'llhthereverseorder.lh=an:prcscntedmdetaJlinAppendtxAl.butarelistedbere:
Lallgl'llg<'SWithmarkersofpropostllonalmodalityour
3
6 7 8 9 10. ll 12 13 l~
15. !6. 17. 18. 19.
Hdi(Chadic:Afro-AstaUC) Arable & Modern South Mabian (Semllic: Afro-Asialic) Pashto(Easternlraman) Sindhr([ndo-Aryan) Burman(Lolo-Burrnese:Tibeto-Burrnan:Sino-Tibetan) Garo(!ingpho-Konyak-Bodo:Tibeto-Burman Smo-T!betan) limbu (Mahak.irantl. H1mala)1Sh: Tibeto-Burntan. Smo-Tibetan) Dravidian luray & )uang (South Munda) & Kherwanan (North Munda} T:uof. Bonont & Nahk (New Ireland: Western Oceamc. Austronestan) Sudest (Papuan T!p. Western Oceamc. Austroneston) Kamula (Central and South New Guinea-Kutubuan. Central and Western: Mam Scctro11· Trans-New Gmnea) Angaacaha & Kapau (Angan. C".cntr~l and \\'estern· Mam Se.:twn. Trans-New Guutca) Nankina (Huon-Fmtsterre: ("~nual and \\'i:stem· MainS..:uon. Trans-New Guinea) Kiwai (Trn!ls-Fly. Trans-New Gumea) Pa•>'aian(Teberan-Pawaian:Trans-NewCiuinea) Bar:u (Cemrnl and South Eastern: Eastern: Mam Sectmn. Trans New Gumea)
83
20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.
Tauya (Brahman: Mandangand Alben Range: Trans New-Guinea} Binandere (Binandercan: Eastern: Main Section: Trans New Guinea} Djamindjungan(Awtralian} Karajarri (South-West: Pama-:.lyungan: Aw~ralian} Siusl.aw (Oregon Penu~ian: Penu~ian} }lisman(Maiduan:Penu~ian} Ku~enai (lsola~e}
Campa and Amuesha (PreAndine Arawak Arawakan} Amahuaca(SouthCentralPanoan} Chibcha(ChibchanProper:Chil>chan) Pilet (lsola~e) Kwaza (lsola~e: RondOnia} Waorani(lsola~e)
Cocama-Cocamilla {Tupi-Guarani Subgroup lll: Tupi}
La11gm~s
with """k.·rs of pr!>posiliot~o/ modality inside rcme
Hindi. Urdu & Punjabi (Central Zone: lnd!>-Aryan} GreatAndamanese(Andamane<e) 3. S
pass.ll'fl-><0><
Since i~ i< very difficult ~o de~ermine bow ~o correctly count sucb languages un ~he level of abs~rac~ion chosen for ~hi>; im·es~iga~ion. ~hey are no~ counted here. Ano~her Languag.: tbat could not be coun~ed w;u Hup. Although Epp< (2008: 381} s~atesthat Hup has the overarcbing ord<:r V-A-T-M (including ~he dccl.ara~ive and conditional), there are exceptions to ~his generaliz.a~iun. This is also ~he situation in Yurakari (van Gijn 2006) where event modality and ~en<e morphemes generally occur closer ~o ~he •·erb than propositional mudality morphemes. and in Epena Pede.: (Choco: Harms 1994). where ~he mood sutlixes are generally situated ou~side of ~he tense sutlixes 3·5-l Cumparison of the,,.,., in•·estigations Since bo~h in•·estigations yielded exac~ly the same resulu. ~hey supp<Jrl une another and show that propo
included cases. Put differently, it is almost nine Urnes more common with propositionalmodalityout
ccnainti•·esutfut(butoutsidethedubitati\•e).Furthermore.therei
que
ChaplerJ Thomorpho•yntoctK•l:ttU-
ten<e Funh~rmon-, m bmh Southeast Ambr)m and Nonh Sarum, there arc
thatproposttionalmodaluy.likesubordinatorshav<:«:opeo\<erthewholcproposition, thcextstetKeofambordinatorwiththeoamemeanmgmapamcular]anguageisa
lyfuturereference"" A
Ill Dongole<e Nubian. furthermore. Armbruster ( 1960) calls the ""col!dtltonal"" suf6x-kanad,·erbralizmgsutfixtb.otbasbothatemporalandacondtttonalmeanmg (r.e.'when"and'tf").Aswassatdtn2.L.5.thecondruonaltsaproblemattccategoryfor theveryreasonthatnoverlapswithtemporaladverb< lnlht<parltcularcase.itl
Thos•mealsoopph<SlotlloselanguagesfoundW>ththeopposneorderwhcrethere>re
~onOtc\lngrn•ty«"Mopuche.rndTuLu
&7
1111
Mod•ht)·andSubordmators that th~ realL'l sulli~ has a non-past meaning. ThiS would be an une~pected meaning. since the typological realis categoryts notallonallya .. ociated "'~th the past. Further· more,th~merecircumstancethat-ll(u)appearstodenotethenon-pastquestionsthe
orderverb·propositionalmodaltt)-··tense. Thus,thevalidttyofalloftheetghtcounterexamplescanbecalledintoquestion (but also, as a con..,quence. the validity of siK ofth~ 66languages in support of the hypothesis).AUmall,thetwoinvesttgatmnsmustbesaidtosll'itainthehypothesisthat propoS>tionalmodahty.hkesubordinators,tsacategoryw•thscopeov~rten..,,inthe
semethatthatanalptsissomuchmorehkelytobecorrectthantheoppositeone
}.6 Conclw.ion Thischapterhasdealtwithprevtousre..,archonthemorphosyntacticstatusofpropo"tionalmodaht)-·andpresentedtwonewtypologicalsurveysonthesubJe
andgeneralsubordmatorshavescopeoverthewholepropos.iuon,ithasbeenimpor· tam to ..,e whether markers of propositional modality also have scope over the finite proposttion,mdudmgtense. Threepreviomtnvestigationshavebeenpresentedand evaluated.AsforpropoS~tionalmodahty.thecondusionsmadeinthe..,,nvestigations
diverged.Bybee{l985)didnotfindanyevidencemsupportofthetdeathatthemoods indicative,sub;unctive.potenhal.probable,dubitahve,andcondiuonalcouldbe•epa· rated from the spe
Choplel'~.
Thomorpho•)'Tl!O
other$.\hercwcrcnotablccxcrptions.la
llg
CHAPTER 4
Subordinators and modality
Thepre•entbookexplore•thehypothe•isthatgeneral•ubordmator>(mrnplernentiz· ers),a•opposedtoadverbial•ubordinator>,shouldbe•ub.mrnedunderpropos:~tlonal rnodali~·.lntheGerrnanlClanguage•.iti•hypothe51ZedthatTHATdenote•theReah•
(declarative: factual and potential), whereas n and WHnH£R denote the lrreali• (hypothetical and uncertain). lnthepreviou•chapter,theworkinghypothes:~•thatpropositmnalrnodal•tyi5a
peripheral functional category with .oope over the fimte pmpruition wa• su•tained by two independent typologJCal inYe5tigatlons. The resuh• •howed that rnarker5 of pmpo· sitlonalmodalityoC(urredouts:~dernarker5oftensemnetlme•moreofienthanthe
oppru1te. A closer look at the counterexample• further showed that most of these may have other explanations (a difference 111 morpho·phonological status between the markers in questmn or incorrect analpe• of them). All this makes a companson be· twernpropositlonalrnodalityandgeneral•ubordinator5posslble.lnthischapter.pre· vtousre..archonsubordinatorsandtheconnectionbetweensubordmatorsandmo· dali~·WJllbepresented.
4.1 Subordination and subordinator. 4.1.1 Subordination Roughlyspeaking.asubordinatedausecanbe•aidtobeaclausethati•(apartof)a constituent in another dau ... More formal definitiorn of subordmation haw p=n diffi.rulttoformulate,sincesubordmationranbeexpre•sedmavarietyofdilferrnt form• m the languages of the world (Chri5tofaro 2003). In e.g. the GermanlC languag· es, the fJ'PLCalsubordinate dau .. IS a fimte dause mtroduced by a subordinator. How· el·er,mdause-rhaininglanguage•.asubordinatedau.. maysimplyberealizedasa wrbinaverb-chain(seebelow).SubordinationtypicaU}'imulYe55omedeg=of"de· ranking~whirhmeansalackofexpressionofoneorsewrall·erbatfunrtional•atego·
riesandauseofspeC1almarker5,notmedmmdependentdau..•(Cristofaro2003:55) However,therei•noformalcriterionthat•anbeformulatedtocaptureth15inade· fimtiwsense Cristofaro (2003: 291f) aUempts to •olYe thiS formal problem b)· giving subordma· tionapragmatirdefinition.Sheproprue•thatsubordmatmnranbedefineda•ala•kof a... rtJVeness,i.e.iUorutlonar}'force.SubordinationlS ..anasymmetriralromrnuni<eatJVe organJzationofthe .. ntenre:'ThemaindauseimposesLtsprofile(dlooutionaryforce)
overthewholescot~nce,\\·hercasthesubordmateclauseisconstruedinthcperspec
tiveofthemainclausc.Onereasonforsubordinatingclauses.Crillofaro(2003:3l) argues. is to a•..,id asserting already known information, •~hich the s~aker wishes to remindthelistenerof.ln
main clause) that the s~aker wishes to (ommunicatc. On the other hand. subordinationcanalsoinvo]vcne\\'information.lnfact.itmayevcncarrythemostimportant int0rmation.asinthefoUowingexample(Crillofaro2003:36) (79) Hesaidit'sraining Themostimportantinformationin(79)isabouttheweather.notthefactthatsomebody s.~idsomething.Ho•~ever,asCri.
raining" i.< not a~ned, a.< Cristofaro's;uscrtiveness tesu reveal (see also 2.3.2 abOV<') (80) a. Hesaidit'sraining,"lst~)lt? b. lsitthccascthathesaidit'sraining'("Didhesayit'sraining?l"lsitraining?) ThepresentinvestigationadoptsCri.
ljustranmtoSusan.who\"
Theb-exarnpleisarestricti\'erdativeclauseofadclinitenoun.lnthiscasc.thepropositionisnota<scnedbutprcsupposcd. Thedclinitenounindicatesthattheidentityofthe referent. specified in the rdative clause. i< known mformation to both the speaker and the listcner.lnthenon-restricti\'ea-cxarnple.however.thcpropositionc:q>reSSedintherebti\'Cclauseisdearlya.<scrted.asthetag·quest:iontcstre\"eai<'.U•ingChrhtofaro's(2003) criteria.onemustthereforecondudethatnon-restrictiverelativeclaUSC$.asopposedto rcstrictiverclati\·eclauses.cannOibcconsideredsubordinatc.lntcrestingly.thisi.
Chapter~-
Subordin•tor
''scmantJcaUysubordinate': Hooper &Thompson (1973) alroargue that resticttve rdative daus..s of mdtfinitt nouns are assened.lhis analy
ampleofadverblalsth~tcannotbeusedasindiGitorsof•Uocutionaryforce,mdhence non·subordinatton, hn,...,1..,r, are modalad1~rb.< such as m fa
lnfact,youare"-TOOg lnfact,lwas"TOng. lnfact,youwereright.
The funcllon of these adverb~ is to signal the speaker's attirude towards lite I milt oftlu: proposilloll(thatitisanunexpectedfact),nothislheranitudetowardstheasscrtion (whichadverbssuchas.frank/yandlttmcstlydo)_ Sur~"enough,theycmbeuscdin pre•upposed•uh
ltoccurredtoherthatshewasacruaUywrong
Notethatonecannotaddthetagquestion wom'tsheto(84). Thesamecanbesaid aOOut modal panide< such wold'[ suppose. probably' in German More problematic are particles such as ja in German. fo denotes the speaker's at· tttudetowardsthetruthoftheproposition,thatiti
a.
Dasistja immer dasselbe lt i~ PART always the.same 'lsalwaysthesame,)'ouknow'
b.
Aber das babe lchja ~sagt' but that ha1·~ I ~ART said 'Butthatiswhatlsaid!'
Typically,italsosignalsanappealtothclistencrforconsent(Abraham2009:2),which isclearlyaspeech-actootion'.However,insuOOrdinatedauses._iadoesnol>eemto 1.
jac~nai>OhO>-.>na~·ourofcountor••J>C<101ion(theadmuatl\-.)onthepMtofthespeak
u and/or the lt>teMr. In thl> <enso. ll 1}'picallr occur< m independent
(<xampkfrootAbraham1009:J) (i)
llusi<:>l Jd
ou•!
I'Oulook rurout "Why,youarel<><>ktnghke>Ome~no 1 "
ad~matJ>-.
dau><>
~l
H
,\\od.oluyondSuOOnhnotor> signal an appeal for consent (example from DM
Dtgilal~ Worl~rbt~eh
der
d~tmcllm
Spt~dr~. wwwdwds_d~)
(86)
urnere Zuwrsteht, da6 das Gluck _in doch emma! kommcn mlisse my irmer confidencethattheluck ro\ltTPo\Rlottce come must 'm}' mnu confidence that my luck must surely change o~~e day' m<:~lle
Onecannotaddthetagquesllonm.:lllwallr'nght'tothesubordmatedause,whkhone reasonablyshouldbeabletodoitj~stgnalledanappealforconsent.Thw.themodal
pantcletntsnotadearurdkatoroftllocuuonaryforce'a.ndnon-subordmauon Theretsyet:motherproblemwiththeuseofsp<:ech-actadverbtalsasmdtcatorsof dlocutionaryforcemsubordrnateclause<.,\swa.<potntedoutm3.3,spee.:h-actadverbial
l'mafratdthat,trank.ly,wearegoingtolose"'Frank.ly.l'mafratdwean:going to lose
ltisnottheembeddedclansethatisanhonestsp<:ech-act,butthea.sseruonthatlhe >p<:a.kcr ts a.fratd Note that when the main clause is nomma.lized. one cannot use the sp<:ech-actadverbtalinthecomplementclause {88)
m)·furthat,"frattkly,n-ean:goingtolose
Subordinationiscloselynedtothenotionofcomplementanon.Snll.thetwoconcept< do not cotnpletdy overlap. In the wider sense of the term, complementahon also refer to non-fimtecomplements,t.e. infinill\'e phrase.,pantctptal phrases etc. Although thesecandearlyfuncuonascomplementsofverbs,nouns,adjecuve:.etc,theyc;m hardly be classified as subordmate dames. A clause must atle~st desmbc a proposiuon,notjwtanevent 4.1.1Subordinators Subordmatedausesaremtroducedh)'
l•
C:hopler~-
Subardtnol<>r
manic subordinate clauses, the dedarative is thought to be lexicaliscd b}' THAT, whereas the interrogath·e is lexicalil<'
lnthepresentinvestigation.theneutraltermgeneralsubordinatori<preferredto the more te.:hnicalterm complementizcr. By general subordinator. the author means subordioatorswithoutrichlexkalconteot.asopposcdtoad\'Crbialsubordinators.The gcneralsubordinatoronlyhasameaninginrelationtothedauseitisheading,whcreastheadverbialsubordinatordenote
subordinateclausei
plement clauses but also conditional protasis clauses in many of the Germanic language$. Like other adverbial clauses. protasisclausesareadjuncts that denote under what circumstances the main events take place ('in case'). On the other hand, protasis clauses are hypothetical or counterfactual, so from a semantic p<1int of \1ew. this is •·
By~ninuoolcopitalleller.thecotegorri•di•nngui
?5
96
Modoht)·andSubordmators
compat>blewt!h!heanalp,.oftfaslrreaiiS.Furthermore,condL!ionaltf·clawesre· semble complement clauses in the ""'"e that they too display a kmd of complementarysyntacllc dis!nbuuon between the subordinator and the finite •·erb m !he Germamc languages(tf·vs.!heso-calledquesllon-formedcondt!ionals).lnfoc!.it"illbeargued in Chapter 10 bolow that conditional subordinators can hke so many other adverbtal subordina!ors m the Germanic languages, be dn·ided into two components: a c~rcum stance-adverbial first par! {"m ca..,"), and a proposttional-modal ""cond part {hypothetical) which is subordinated m relation to the former. This "supported by the complex origin of conditional subordinators, described by Traugott ([985: 2901f): epis!etmc,op!a!iveandinterrogahve,i.e.modal.ontheonehandandadverbtalonthe other 1 . Smce Germanic If can both introduce condiuonal and interrogah>"O subon:h· na!eclauses,i!wtllbearguedthatcondttionaltflexicalizesthehypo!he!icalcomponent.ratherthan!headverbtalone.lndeed, Traugott{l985:290)andothersholdthat Germamc tf denves from a pro!o·Germanic noun meaning 'doubt: whiCh origmaUy funcuoned as a question marker. Therefore. it wtU be argued that tf is a kind of com· plemen!izerevenwhenttmtroducescondt!ionalclauses.lno!herlanguages.thecon· di!ional"subordmator"hasadverbialongmandcan!hereforebethough!!olexicalize the ao:h-.rbial component. e.g. wem> in Gennan. In colloquial Norwegian and Danish. lastly.bothpart.sarele:ucahzed,ltvi.-om'if'. Anotherquestiontha!thebookwilltrytoanS\•"Orishowtoanalyserelativesubordinators.Relam·esubord.ina!orsarealsoratherpoorintheirsemanttcs,andrestrtCtiw relatn-.clawesarecomplernenttzersinthe""n""that!heyarecomplement.sofnouns.lt mayappearthat"relallvity"andrnodaht)'·areratherdtlferentnotions,which.inturn,can be..,enasananomalyto!hehypothesiS!hatgeneralsubordinators(complemen!izers) can be identified with propositional modalit)-•. Hov.-.ver. in Chapter 10 belm.·. tt v.~ll be argued!hatatleastEnglishres!rtC!iwrelativelfidlshouldbeanalysedasaRealissubor· dmalor. Restricllve relative dau""s are ""mantically Real is. As complements of delimte nouns, !hey are presupposed (..., 8 Ib). When complement.. of indefinite nouns, they are notpresupposedbuteitherfactualorpo!enttal(example82aandb,resp«!i>-.ly).Note thatevenwhen!heyarepotenllal,!heproposittonisstiU"i!hintheReaiiSdornain;althoughthereferent!Shypo!hetical,itisimpltedthattherearelnfactsuchreferentsmthe realworld.Alsono!eth.a!!heEnglishrelativelltaliscontined!orestrictiverda!iveclausesandtha!itiShononyrnouswi!hthededarati>-.lltal. Las!ly.inlimtivemarkerssuchasloinEngltshareanalysedascomplernentizersor. more generally. subordinators in some frameworks (e.g. Noonan 2007, CGEL). Yet, theseareclearlydtlferentfromsubordina!.orssuchasTHATandtf,mthesense!hat their complements are non-finite. Withm the framework of Generative Grammar. suchrnorphemesarenotthoughttolextcalizethetnos!peripheralfunctionalcat.egory of the clause, Complementizer. but rather the tense category {as defective; ""e e.g. Chomsky 2000).lndeed, if subordinators are deli ned as introducers of subordmate ;.
l'luscopul•>"
Chopt
Submdmotoro;mdmodalot)
dau<e<. mlimm·e markers cannot be da~stfied as subordmator>,
const
typically not whoUy mduded wuhtn any consmuent of the nudem" (Hop~r & Traugolt 2003:!76) Ioparucular,adverbialclausestendtobemterdcpcndentmthclanguage• of the world. wherea< complrmcnt clause< tend to be true subordmate dawes On thcwholc,thesedtMtncttonsdo notba>-cconse<juenccsforthe hypothc
97
98
.l.lodohtyand
Sul>ordonat~"
4.2 The functional category complemenlizer and modality Aswassaidintheintroduction.thehypothesisthatgeneralsubordinatorsbelongto the same functional category as the moods indicative and subjunctive ami epostemic modal markers os based on the observation that they appear to denote the samethmg. namely whtther the prop<>~ition is factual or hypothttkal de. (the ReaiJS-!rrealis di>· tinction). HoWe\"tr,it isal~ohasedontheobsen·ationthatthegeneralsubordinaton are m complementary syntactic distribution with the finite \"erb in the Germanic Vl l.anguages(exceptforlcelandicandYiddish).ThisisalsothecaseinEnglishpolar (yes-no)questionsandprota~isdauses{tF·V>.question-formrdconditmnalsJ.lnfact.
within the framework of (iencrati,·e Grammar. the analysts of Germanic Vl and VI (e~cept V2 in Icelandic and Yiddish) is thac the finite verb occupoes the Cornplementizerposttion(denBestcnl9&3.Piatzackl9116a.b).lnterestingly.verbraJSingi.solien conne.:ted \\ith intle.:tion. According to Chorn
a.
bakhrnm -lr1111 bat\
l eat
-susrrice
"Jwouldliketoeatrk~'
b. nebar clra ...-r sclhaa garam outsideis ·tNnRvery hot "lsitveryhote>otside?"
ChoJ"eJ4_ Subordin;uorsondmodaln)-
At th~same time, Bhatt ( 1999) argues that the two functions mood and subordmatmn should be sun as separate fun([ional categories. In German, Dmch, and th~ Mainland Scandmavian languages, the finite verb i
(symmetric \'2-languages). ~refore. for these languages. Batt (1999) argu~s that the complementizerdoes not determine mood. but only subordination. whereas the ~rb determmesclauset;~inbnthmainandsubordinateclouses.Thus,thetwofuncuons
subordmation and mood actually represent two separate funcuonal categones. In asymmetric V2-languages. thecomplenu:ntizer lexicalizes both of these categories. To supporttheiduthatsubordinatinnandmoodrepresentseparatefunctionalcatego· nes. Bhatt (1999: 152!1') shows that Korean expres"'' these two funcuons through separatesulfixesnntheverb (90)
Bill-un John-i wa -ss -Ia -ko sayngkakhanta Bili·TOf John-~oM cnme ·PST -o~<:l -w~uRD thinks 'BillthinksthatJohncame.'
llhatt ( 1999) is important to the hypnthesis explored in the present mvestigallon. smce it connects complementizeu and V2 at least partly with modality. On the other hand, thein-depthanalysisactuallyconstitutesacounterargurnenttothehJ'POthe>ts,asit considers mood and subordination to be two separate categories. There are, howeYer, certainproblemswiththatanalysis.First.tbesuggesllonthattheretsaumvcr>alcate· gorywhichonl)·fulfilsthefunctionofsuhordinatingclausesisnotveryplausible.Sub· ordination per sc dncs not correspond to any semanuc feature. Indeed. if one adopts Cristofaro'5 (200l) definition of suhordin~tion. it i< rather idemilied wnh tile lack of a certainsemantic-pragmaticfeatul'l:.namdyassertivene.,_Sincehoththeexplanalory branch of language Wpology and the Minimalist Program of Generattve Grammar ..,sumethatpartsofspeecharecombinationsofphoneticandsemanticfeatures,iti> dtlficulttomoti\'atetheexistenceofacategorywithoutanysernanticfeature..lnfact, "~thm the Minimali>t Program of Generative Grammar. such categories are assumed nottoexist"(Chomsky2001) Second. the as~umption that the finite verb marks thedause I)~ in main clauses. whereas the complementizer marks the clause type in embedded clauses in asyrnntel· nc \'2-languages implies that the finite verb does not mark clause!)~ m embedded clauses m these languages. However, in German. the finite verb carries mood markmg mbothmainandsubordinatedauses.andinalloftheselanguages,theretsadear
6.
lt
•• rrero>~llono heod1ng pre-positional ab)
\Ill
>oo.\lodalnyondSubonhna>or. d>stmction betw«n anon-modal dcdarative form ;md modal wrhs m both main and subordmateclauses Third. the suggesl>on that the dosely related languages kdand>c and Mainlmd Scandinavi;m,YiddishandGermanshouldha\'l!Cog>13teWmplemenhzersthatdonot lexkalllethesamefuncuonalcategontstsnotveryplausihklnfact,aswillbcd!scuss.:d 1n more detad m Chapter 9 hdow, lcelandJc. Norwegian and Dutch aU have the possJbdtt)' of constructing If with rHAT m interrogati\'e md cond>tional claus.:< {thela11cratleastmkeland>candNorwegian,Vangsnes2006 3,Z""artl993 43) {91) a
Einmgva"' ga~nan :Wheyra cfai'J fOlk er ailvinna he.PK~.'-'1!81 fun lo hear 1f that people are to work also \'lil ai'J hi'Ja atlra componenu/modules wilh to trmslate other component<Jmodules 'It would also be fun to hear 1f people are workmg on transJatmg other components/modules'
b_ Ofte kan '~ lure p.i omM endringen •lqer for oftencmwewonderontf thatthechangehappensfor re(ormens egensk}·ld? the reform'• own ~akc 'Oftcn,wccanwondertflhechangehappcnsforthcs;ill,ofthcretorm>UCIP' Pietvroeg of/oldm /an Mane kuste Pietaskffl >filfthat /an Mane kused Pteta•kdtf/mkmedMane' ThiS would be unexpe.:ted >fcomplemcntirerslexkalised both subordination and clause type in h'orwegtan and Dutch but on])' subordinauon m kdand!c. In aU these examples, IF do:arl)' lex>cal~zes both cia~ type and subordmal>on (mterrogatl\~ or condttional)whereasTHATtSbaskall)'redundant ltisapparentthatthesepantion bct•~eendauset)'Peandsubordmauon isnotver)'tcnable f'Ourth.>he
4.1.1lhcsplit-complementizerhypothesis(RtuJI997,2002,Stroh-Wollin2002) AswasmentionedmU.2,manygencrati•·erese>rchers,origmallyRizzL(I997.2002). argue that the functmnal category Complementizer withm Generam<e Grammar should be split into at least two parts· a mor~ peripheral category that det~rmme• dauset}'l't'and(arneotllocullonaryforce{calledForce)andamorecentralonethat
Chopter~-
Suhordin•tor
has to do with the rather vaguely defined notion of fintteneS<. Anordmg to Rtz:u (l997)complementizerslexicalizeboththesecategories.ThishypothesisLSinterestmg, as some elaborations of it (Stroh-Wollin 2002 and Giorgi & Piane5i 2002) conn..: I the fimtene .. cattgorywiththeindkative-subjunctivedistincllonanddrawadJstinctmn between these moods and moods with iUocutionary force (i.e. speech-act moda~ty). Stroh- Wollin (2002) is such an elaboration of the Split-Complementizer hypothesis that removes some of its le!<S well-moth·ated assumptions. based on data ti-om the Scandmavianlanguages.AccordingtoStroh-Wollin(2002:l33tf.l60tf).r..,rcedL"Ier· minesfourdausttypes:dedarative,interrogative,tmperatiwandoptative/hor\alive. ThiSagreeswell\\iththeillocutionaryforcecategor)"PmpruedmBybee{l98S,see3.2) lnoppruitiontoRizzi(J997),Stroh-Wollin(2002:t27tf)furtherarguesthatFon:ecan only()., prestnl m main dawes and subordinate clauses with main clause topolog>•, sinceonlythesecanb..consideredindependentunerances.Embeddeddedaralives andintermgativeswithsubordinateclausctopologymerelyrepresemthepropositmnalcontentofunerancesandquestions,sheargues. Thisalsoagreeswellwiththedelinition ofsubordmationadoptedin the present itn<e"igation (Cristofaro2003.Searle 1969). Crucialforthehypothesisexploredintheprescntin\"estigation.Stroh-Wolhn (200l:l5-l.l62)identifiesthemoreceotrallinitene<ScategorywiththeReahs-Jrrealis diStinction and factuality. "Finiteness-modal".asshecallsthecategory,isbothwhere complementizers of the THAT and tfiWHHHH types give the pmpositmn a truthvalue and where the indicative-subjunctive moods and epistemic morphemes are h· censed.ThiSassumptionissupportedbythecomplementarysyntacticdistnbnltonof complementizers and finite verbs in the Germanic Vllanguage<. One synta~ti<e pt...:e of.,.idenceforalsoincludingcpistemicmodllityisthefactthattheepLStemicparticles kanskcandknnltmtdacanalsooccupythispositioninSwedish,leaVJngthelimtevcrb intheverbphrast,asinsubordinatedausts (92) Han kmuk,aldriglwlast en bok i sinliv he mayb..never has read a bookinhislife 'Maybe,hehasn.,.erreadal>ookinhi.<entirelife" Stroh-Wollinassumestheretobeyet anothcrlinitenesscategorybelow FinitenessmodalwhJChestablishesthenexusrelation.butthatisnotrele>·anthen-. Theideathatthefinitenesscategorycanbeassociatedwithatleastthesnbjnnctn·e hasalsobeenputforwardbyGiorgi&Piane
Manocrede (cite) iill pmno. Manob..lie,<es that be.3su.rlt~'-''--'K1 ldt 'Manob..lie><esthatheleft'
101 ModahtyandSubordm•tors b
Manoha d.tto "(che) e pamto Manoha. saJd that be.3•GPH'l"ll ldl ManosaJdthatheleft'
ThmanalycandthesubJunchvelextcahua modal feature (""Mood"). In the cnmplementiur case. thJS leature "camed by a func· tinnalcat.:gorythatcanbetdentiricdwtththelinuenesscattgory(Rlzzt2002).lnthe subjunctive case wtthom an O\'CTI cnmplementiur. the modal feature IS marked by the subjwtctiw mlkclton. Smce thts marker is morpholog~eally fused wtlh subject agreement{p
clauses but also eqmvalem comparative dau.<es ('as') and <;an lollow the wh·\\'Otd in embedded content que>hons. The Engh~h restrict""" relat""" complemcntizer th~l i• analysedthesantewayasthededaralll't/lmtduetotheircommonongmandhnmonymy. Scandmavian s
ChoJ"eJ4_ Subordin;uorsondmodaln)- 10) 4.:1..1.1 fr
doesnota.scribeaninherentsematUicfeaturctothecategoryfiniteness.sinceitisthe complemenlthatiseithcrfiniteornon-finite.notthecategoryitself.llisdifficultto moti.-atetheeXIStenceofacatcgorywithuutanysemanticcontent.csf'<'
104 ModohtvandSubordm•tors Another problem Mth the Split-Complcmenuzcr model" !ha!Jtas.sumcs that !he dcclarativei
clausesdonothavelllocutmnaryfnrcc.somethmgwhlChStroh-Wolhnacknowle
}lot only the relation between the md~eati\'C and the dedaram.., but also between thesecategonesandli.nllenesscanbemadeclearwhenonelooksatthedescripllonof theKherwarianlanguages(.\1unda:Austro-A
peratweorsubordmate,Anderron2008·13l) lnthedescnpuonofHo.Jastly.-a!
Fmally.itshouldbepmnte
Chort
SuboJdm>!oJ>ondmodall!)' oos
andhighlyreleYantforthepresenthypothesis.ln8.5.further..,.idencefromGerman willbepresentedthatstrengthenstheobserv•!ionth•tthereisaconnection~tween
declarahve mmplementizer omis.sion and the use of the subjunctiYe. Howe>·er. Giorgi &Pianesi's(2002)moredetailedanalysesofthephenomenafailtodJS!ingui•h~·
twernuniversalgrammarandforrnsofspecdiclanguages,andbetweenthetwoforrnal systems inflection and complementJ:<eJ5. The fact that the subJUIKILYe inflection 15 fuzed with the sub)e<:t·agreement inflectmn m halian does not neas.sanlj' mean that theuniven;alfunctmnalcategory""Mood"mthatcaseisfuzedwithsubject-agreement l!JSalanguagespecifi<e.morpho-phonologicalphenomenonthat.annotbegeneralizedintoauniwrsalfeature(unles.srnorelanguagesarepresentedthatshowthesame panern). Furthernmre.thefactthatthesubJuncti\"e·inlkctedverbinhaliandoesnot raise to the r!Je-positmn when r!Je is omined does not mean that ""Mood" in that case isles.s peripheral than when lexi<ealized by c!Jc. As was pointed out m 3.3 above. com· plemenhzers and mtleo!ion are two separate formal systems that work on ditferent le>·elsoflanguage(syntaxandmorphologyrespec!Lwlylandthereisnoconditionthat themflectedYerbmustraLsetothesyntactJCpruitionassociatedwithtlleinflection. althoughitcandosoinsomelanguages.
4-3 Compkmentiurs and modality from a typological penp..:tive Wherfas 4.2 laid emphaSis on the complementary syntachc distribution between complementizers and the timte Yerb m the Germanic \'2 language• and presented research that connected thJS with propoSitional modal it~·. the remaming chapter wiU be de\uted totypologJCalim·estigationsthathawfoundconnectmnsbetweencomplementJzen; and modality. 4·3·' Complementizer5 and the realis-irrealis distmct1on in the language• of the world (Noonan 2007) As was said in 2.4.1. ~oonan (2007) 1s a survey of complernentahon m the languages of the world. As regards the relation between complementation and modallly. Noonan pomtsoutthatthereJSadistinctionbetwernmdJCatJVeandsubjunctive(orrealisand irrealis) complements in many language•. lndJCatJve and subjunctive mmplements can of course be distinguished by mood. as was shown in 2.2.1. but :-loonan point• out that the chmce of complemen!izers can also determme whether the complement 15 indicatJVeorsubjunctive. lnsornelanguages.suchaslangoandlou.mdJCatiYeandsubjunctJVecomplementshaw the same complemrn!izer. ln other languages, how.... er. such as Rumanian andBulgarian.indicatiYeandsub)unctivecomplementshawditferentcomplementLZers
(Noonan2007:62).lntheselanguage.,thecompkmentizersalsodenotetheindica· live·subjunctiw(or~alLs·irrealis)dLSlinction:
(96) a
(Noonan2007:6l) El spuM cd cite11e o carte he says UJMP rnd.3s
b. El nea si1 lJI~sc
(Noonan2007: 105) Dobre.lr te srdtnnx good CoMP you meUS
b. M1..tja d~ ida. think.!S
tioni•credibleornot(Noonan2007:58): (98) a
xal najlnl<> chuluj najpresulente said he CUMP will.come the president "Hesaidthatthepresidentwouldcome."
b.
xal naj dmbJ/chulur naj prestdente satdhe COMP wiU.come the president "Hesaidthatthepresidentwillcome."
lunrehablel
Chopta~-
Subordin;uorsondmodal•tr 11>7
Furth~rmore,
in Kin)'arwanda, the complemcntizer< ko and ngo md1cate a neutral ver· sus negati~epropositional attitude on the p,lrtoftllespraUr(:-loonan2007: 125lf): {99) a
~'ateUrefe ko amazi yari mare-mare think.3s<.;.PAST (:OM~ water be-3•'" d«p "Hethoughtthatthewaterwasdeep"
b. yatekeretellg<:'atnaziyarimare-mare "He(misguided]y)thoughtthatthe"'aterwasdeep." Palmer {1986: l48lf) reports that there i1 yd anotherdubitative cornplementizer. kcmgo.m Kin}'arwanda. Wherea1 ngodenotesdoubt on the Msi1 of direct evtdence. ko11go denotesdoubtonthebasilofindirectevidence. Another highly relevant example nf Noonan"• is in fact if in English. As Noonan (2007:ll41f)pninuout,ifi!u<edincomplementclamestocancrloutpo•illvefactual 1mphcatmns (100) a. AlfknowuharZekecame b. AlfknowsifZekecame(butldon"t) Theuscofijinconditionalprotascsisrdated. ~oonanargues.Itmdicatesthatthe propo>Jtiontsnon-actual.flnthcanbesaidtodenotelrreali•modality. Toconclude,Noonan's(2007)datasupportthehypothesi•thatgeneralsubord.ina· torsdenotepropositumalmnd.ality.Thefactthathepoint•outthatEnglishifdenotes lrreali5ishighlyrelevant.lnChaptcr8.amoredetaoledim'entoryoftheusesofthe G~rmanic THAT and tf wiU be prescntl.'d that lends further support to this claim. 4·3-l Complementizers as modal categories (frajzrngier 1995)
Onelanguage-typolagicalinvestigationthatactuallyexploresthehypothesisthatcomplementizersshouldbe,;eenasmodalmorphemesisFraJzyngoer(l995).Frajzyngier ( 1995) argues that cnmplementizers primarily denote modality and that the functmn of introducing complement dawes is ,;econdary. In support of this. he presents data from a hattdful of languages. One piece oft\•i<.kttce is the Pol1sh cornpl~rnentizerczy "whether" which not only introduc~s embc:dded interrogative clauses but also interrogallvemainclauses(Frajzyngierl995:478) (101)
a. Spytalem
czy
ask.]SG.P~Rf.PAST whether "laskedwhethershecame."
przyszla COIDt.3>~M.,G.PERF.PA
b. Czy przyszla? Whether COme.3>EM.SlLPHEPASl "Did she came?"
toa ModohtyandSubordm•tors Normally.polarqu~snonshavequesuonmtonanon Wh~ndtr.ctpolarquestion•are introducedbyccy.howev~r.thequemonintonationdoesnothav~tobeused. Therefore.ucouldbearguedthatrz)•andthequesuontntonanonfillthesamefunction Se.:ond. French qm "that' can be used a.< an optative marl= (Fra)zyngter 1995 480)
(102)
Qu'ilp~n..,e!
'Mayhepensh!' Here.itcouldbeaddedthatEngltshi/lmlyandSwedishomandM/mracanalsobeuscd a.< optative markers (the b-example i~ the Swedish translanon of the a-example)· (!03)
a !fonlyyou'dtoldmeearher! b
Omandliduhadeberattatlbrmigtidtgare'
The reason why dedoraltve main clauses do not com am complemeunzers ts becau<e theyrepresenttheunmarkedvalue,Frajzyngkr(l995 480)argues.Themdlcativel' not morphologically marked either. A< fnr the marked modahtrvalue, thts t< typically marked byothu modal particles, word order or mtonauon m mam darnes. Asregardswordorder.Fra)Z}11gier(l995:497)connectsmvemnnmmamcl
Hwac~erge nu >ecangoldontreowum' whether ynunnw.eek goldtn trees Dnyounowseekgnldmtrees!'
Furthermore, Fratzyng>er (1995) gwes examples of complementJzers wnh ept
zemPoltsh.whtchdenoteshearsayandindirectevtdence gi}mlele(EastChadtc),whtchdenotcsd!reclpercepnon t~
Fraizyngifr(l995)alsopmmsoutthattheirrealis/hypmheticalclmcbyinPoli.shc3n clnictzetocomplemenltzerswithdilfercntelfects·m-by=dubtlallve,o-by,a-by.3nd mcrhby=desideratn•e,andu-by=oblig3tive{occursali.erobhgatn·epredicates) The!eclusterscanalsomtroducemainclauscs.Totbts,Jishouldbeaddedthatu-by also occurs alter negated predicates ot belkf or predtcates of doubt (Rothstein l993-i35/f).lnthe..,ca.es,u-byratherseemstodcnotethedubnall\-e (105)
a S9dzr.zeJniWJ'Jechal 'ltbmkbe:Salreadyleft'
Chop1eJ4 Suborduu.tor
.'i!esqdzr.ZebyjUZW')'je
'[doubtthatjanckwillmanagcmume' It could al
maucalizauon The fact that main clauses arc normallrnolmtroduced by complementizerssuggeststhatcomplementizcrsnormallydonotlextcalizethemostpenpheral functionalcategorymmamdau~s.byh)-pothe
where they do introduce matndauscs.the complememizers have developed a spc.:chactfunctioll Thl