The European linguistic area: Standard Average ... - uni-freiburg

i ;.1,ijr! jri:j...li'' ljl

LanguageTypologyand LanguageUniversals und Sprachtypologie universalien sprachliche La typologiedeslangueset lesuniversauxlinguistiques An InternationalHandbook/ Handbuch/ Manuelinternational Ein internationales von / Edite par Editedby I Herausgegeben Martin Haspelmath' Ekkehard Konig Wulf Oesterreicher' WolfgangRaible Volume2,12. Halbband/ Tome2

Walter de Gruvter ' Berlin ' New York 2001

i*.'

1492

XIV. Typologicalcharacterizationof languagefamiliesand linguisticareas

107.The Europeanlinguisticarea:StandardAverageEuropean

li rlltt' rili i. l' li lt lt

I' r, I

l . r

; I {

i

f

,,

1,ii t ;',r, til;Ir

L 2. 3. 4. 5. 6. 7.

lntroduction The major SAE features Some further likely SAE features Degreesof membershipin SAE How did SAE come into being? Abbreviationsof languagenames References

1.

Introduction

This article summarizessome of the main piecesof evidencefor a linguistic area (or Sprachbund)in Europe that comprises the Romance, Germanic and Balto-Slavic languages,the Balkan languages.and more marginally also the westernmost Finno-Ugrian languageslthesewill be called coreEuropean languagesin this article). This linguistic area is sometimescalled Standard AverageEuropean (abbreviated SAE), following Whorf (1941)[956: 138].The existenceof this linguistic area is a relatively new insight (cf. Bechertet al. 1990,Bernini & Ramat 1996, Haspelmath1998,van der Auwera 1998,Konig & Haspelmath1999). While the close syntactic parallelsamong the Balkan languageshave struck linguists since the l9th century and the existenceof a Balkan Sprachbundhas been universally accepted,the European linguistic area has long been overlooked. This may at first appear surprising,becausethe membersof the Sprachbundare among the best studied languagesof the world. However, it is easy to understandwhy linguists have been slow to appreciatethe significanceof the similarities among the core European languages:Since most comparative linguists know these languagesparticularly well, they have tended to see non-European languagesas special and unusual, and the similarities arnong the European languageshave not seemedsurprising.Thus, it was only toward the end of the 20th century, as more and more had become known about the grammatical properties of the languagesof the rest of the world, that linguists realizedhow peculiar the core European languagesare in some ways when seen in the world-wide context. From this perspective.StandardAverageEuropeanmay "exotic even appear as an language"(Dahl 1990). A linguistic area can be recognizedwhen a number of geographicallycontiguouslan-

guagessharestructuralfeatureswhich cannot be due to retention from a common protolanguageand which give these languagesa profile that makesthem stand out among the surroundinglanguages.There is thus no minimum number of languagesthat a linguistic area comprisesQtaceStolz 2001a). In principle, there could be a linguistic area consisting of just two languages(though this would be rather uninteresting), and there linguistic are also very large (continent-sized) areas (Dryer 1989a). Likewise, there is no minimum number of structural featuresthat the languagesmust share in order to qualify as a Sprachbund.For instance, Jakobson (1931) establisheshis "Eurasian linguistic area" on the basis of just two phonological features,but of course an area that shares more featuresis more interesting.As will be shown below, Standard Average European languagesshare over a dozen highly characteristicfeatures,so we are dealingwith a very interestingSprachbund. A linguistic area is particularly striking when it compriseslanguagesfrom genealogically unrelated languages(like the South Asian linguistic area (* fut. 109), or the Mesoamericanlinguistic area (+ Art. ll0)), but this is not a necessaryfeature of a Sprachbund.The Balkan languagesare all Indo-European, but they are from different families within Indo-European (Romance, Slavic. Greek. Albanian). and not all languagesof thesefamiliesbelong to the Baikan linguistic area, so nobody questionsthe validity of the Balkan Sprachbund(- tut. 108). In the case of SAE, three entire branches of Indo-European(Romance,Germanic and Balto-Slavic) belong to the linguistic area. However, here too it is clear that we are not dealing with a genealogicalgrouping, becausenobody ever proposed a branch of Indo-European that consists of precisely these three families. On the contrary, IndoEuropeaniststypically assumea particularly closegenealogicalrelationshipbetweenItaiic and Celtic (and sometimeseven an ItaloCeltic protolanguage),but Romance(the sole descendantof ltalic) is inside SAE, while the Celtic languagesdo not belong to SAE. And sinceso much is known about the grammatical properties that Proto-lndo-European must have possessed, it is fairly easy to test whether an SAE feature is an Indo-Euro-

107. The Europeanlinguisticarea: StandardAverageEuropean

peanismor not. As was shown in Haspelmath (1998), most of the characteristicSAE features (also called Europeanismshere) are not Indo-Europeanismsbut later common inno_ vations. Thus, what needsto be shown in order to .. demonstrate that a structural feature is a \ Europeanismis (D that the great majority of core European languagespossesses it; (ii) that the geographically adjacent lan_ guages lack it (i. e. Celtic in the west, Turkic, eastern Uralic, Abkhaz-Adyghean and Nakh-Daghestanian in the eist, and perhaps Afro-Asiatic in the south): (iii) that the eastern Indo-European languageslack it (Armenian, Iranian. In_ dic); and (iv) that this feature is not found in the majority of the world's languages. Par ilarly the last point is not easy to demonstrate for many features because there are still far too few representativeworld-wide studies of grammatical structures, so to the extent that our knowledgeabout the world's Ianguagesis incomplete and biased. we cannot be sure about the European linguistic area. In this article, I will cite whatevlr in_ formation is available, and sometimes I will have to resort to impressionistic observations. The designation "core European lan_ guage" for members of,SAE is diliberatelv vague, becausethe European linguistic area does not have sharp boundariei. It seems possibleto identify a nucleus consistins of continental West Germanic languages(i.g. Dutch, G_erman)and Gallo-Romance (e. g. French, Occitan, northern Italo-Romancej. For this set of languages,van der Auwera (1998a:-824) proposesthe name Charlemagne Spr )und. Of the other languages, th6se which are geographically furtier from this center also seem to share significantly fewer SAE features, i. e. Ibero-Romance. insular Scandinavian(Icelandic and Faroese),East Slavic(Russian,Ukrainian, Belorussian)and Baltic. Even English, a West Germanic lan_ guage, is clearly not within the nucleus. Of the non-IndolEuropean languagesof Europe, the-westernUralic languages(i. e. Hungarian and Balto-Finnic) are at least marginal-mem_ bers ofStandard AverageEuropea-n;they are ln many ways strikingly different from east_ ern Uralic. Maltese also exhibits a number of Europeanisms not shared by other Arabic

1493

varieties,but Basqueseemsto show very few of them. Somewhatfurther to the east.Georgian in the southernCaucasus(and perhaps the other Kartvelian languages) shares a surprising number of featureswith the core European languages.These impressionistic statementsshould eventually be quantified. but since it is not clear how much weisht shouldbe attachedto eachfeature.this is iot straightforward. All ofthe featuresdiscussedbelow are svntactic. or concern the existenceof certiin morphosyntacticcategories.I am not aware of any phonologicalpropertiescharacteristic of the core European languages(cf. Jakobson l93l: 182;"do six por ne udalos'najti ni odnogoobsdeevropejskogo ... poloZitel'nogo fonologideskogopriznaka [so far not a sin[le Europe-widepositive phonological feature lias been foundl"). Perhaps phonologists have not looked hard enough,but at leastone major recent study of word prosody in European languageshas not found any phonological evidencefor StandardAverage-Europein (van der Hulst et al. 1999,especiallyMaps I -4) (but cf. Pisani 1969).A few eeneralizitions are discussedby Ternes(199-3),but he finds that in most respectsEuropean languagesare unremarkablefrom a world-wide perspective.Perhapsthe only featureswortlr mentioning are the relatively large vowel inventories(no 3-vowelor 4-vowel inventories) and the relativelycommou consonantclusters (no restriction to CV syllables).In these respects,Europeanlanguagesare not average. Dut they are by no meansextremeeither. 2.

The major Standard Average European features

In this section I will discussa dozen srammatical featuresthat are characteristic6f the core European languagesand that together define the SAE Sprachbund. ln each case I will briefly dehne the feature and sive a few examplesfrom SAE languages.Thin a name map, which indicatesthe approximate location of languagesby the arrangementof (abbreviated)languagenames,shows the distribution of the various feature values within Europe. In each caseit can be observedthat the nuclear SAE languagesare within the SAE isogloss, and that the marsinal languagestend to be outside the isoiloss to a greateror lesserextent. (part of thE material presentedhere was already included in Haspelmath1998.)

linguisticareas XlV. Typologicalcharacterizationof languagefamiliesand

t494

rtttt' f; r l . t-

2.1. Definite and indefinite articles Both a definite and an indefinite article (e' g' 62) exist in the bookla book; 'tut' English -Romance lanGermanic all almost and all guagesplus some of the Balkan languages iUoaern Greek, perhapsAlbanian and Bulgarian), but not outside Standard Average Enropean. To be sure, their forms and syntactii behavior show considerablediversity (see Nocentini 1996 for an overview)' but their very existenceis characteristicenough' The distribution of articlesin European languagesis shown in Map 107.1.(Abbreviitions of languagenamesare given in the Ap' pendix.)

'i':

1. li

t

Dut

EU

l

ll

r

Tal

Fr

tz\

Sftl

2.2. Relativeclauseswith relativepronouns The type of relativeclausefound in languages such as German, French or Russianseemsto be unique to Standard Average European languagis. It is characterizedby the followine four features:The relative clauseis postno-minal, there is an inflecting relative pronoun, this pronoun introduces the relative clause,and ihe relativepronoun functionsas a resumptive,i. e. it signals the head's role within tlie relativeclause(cf. Lehmann 1984: 103-109, Comrie 1998)'In English,a relative construction like the suspiciouswoman whom I describedalso displays all these features. Furtherrnore,in most SAE languages the relativepronoun is basedon an interrogative pronoun (this is true of all Romance,all Slavii and some Germanic languages,Modern Greek, as well as Hungarian and Georgian). (Languageslike German, whose relaIiu. p.onoun is basedon a demonstrative,or Finnish, which has a special relative pronoun, are not common.) The geographical distribution of the relative pronoun strategy is shown in MaP 107'2.

CB l

r

f

:r'

t :ii ! it'ir, ririlil

Trk

Am

definiteand indefinitearticlepresent - - - - only dehnitearticlepresent \ Map 107.1:Definiteand indefinitearticle

RG

Pol

Ddt

Eng

In large parts of eastern Europe there are no articlel at all (East Slavic, West Slavic, Finno-Ugrian other than Hungarian, Turki-c' Nakh-Daghestanian,Ikrtvelian). Someneighboring non-SAE languagesdo have definite articles (e.g. Celtic, Semitic,Abkhaz' Mordvin). and Turkish has an indefinite article, but no neighboring non-SAE language has both definite and indefinite articles.The only exception among Germanic languages,Icelandic (which only has definite articles like nearby Celtic), is also the most peripheral Germanic language geographically.We can also be certain that the existenceof definite and indefinite articles is not an Indo-Europeanism: The Iranian and Indic languages i,au. g.n.tully lacked articles throughout their history. World-wide, articles are not nearly as common as in Europe: According to Dryer's "it (1989b:85) frndings, appearsthat about a itti.d of the languagesof the world employ articles"(125 out of a sampleof about 400 languages).Only 3l languagesof those in Drler'J sample(i. e' lessthan 8%) have both definite and indehnite articles.

Cz

Gm

HnE

Fr

Uk

Sln It

-

SCr

relativeclausewith introducing relativepro-

- - - - only particlerelativeclause Map 107.2:Two relativeclausetypesin Europe The only other type that is widespread in Europe is the postnominal relative clause introduced by a relative particle (Lehmann 1984:85-87), which often occursin the same languagebesidethe resumptiverelative prono,in tlp" just -the described(an Englishelam.nle radio that I bought)' Particle would'be relativesof this type exist in most Slavicand Romance languages,as well as in Scandinavian languag.s unA Modern Greek, but also in Welstian-dIrish (Lehmann 1984:88-90)' The relative particle is sometimesdiffrcult.to distinguish from a degenerate resumptlve pronoun, and in many European languages

107. The Eulolxan

lirrguistic ateir: Statrdarcl Avct':tgc Littrtlpcatt

it developedfrom a relativepronoun thlouglr the graclual loss of iuflectional distinctions. However. this also rneans tlrat the relativc clauseloscs its specificallyEuropean tlavor. bectruseparticle relatives are also ilttested widely elsewherein the world (e. g. in l'cr'sian, Moderu l-leblew. Nahuatl. Inclonesian. Yorubar,and Thai, cf. Lehnrann 1984:8597). l'Iowever. 1he relativc prolloun stfategy clear'lyis typically liulopcau. It is not founcl in the easterrrInclo-Europcanlanguages,irncl as Comrie (1998:6l) notes."relativeclauscs fonned usilrg the relative pronoult stlatcgy ate quite exceptioualoutside Europc. cxccpt as a recent tcsult'of the influcnce of Euro'Ihe peall languages... relative pronoun strategythus seemsto be a lcmarkable areal typological fcatulc of Er:r
t?g

cei lrk

Map 107.3:'11nys'-perfects in Eulope

r495

ln contlast to the languages iust metrtioned. iu Slavic. l:inno-Ugliau and Arrneuian thc perlcct is usually baseclon a participial constrr"rctiou with ln activc participlcand a coprrla (c. g. Finnislt ole-n see-nu! [be-lsc 'l lcceivc-nrcr] havc rcceived').llungarian seemsto lack a pcrfcct completcly.Irt sotuc Nukh-Daghestanian languages(e.g. I-ez-gian and Goclobeli;. tlre pcrfcct is folrnecl on the basis of the past convclb plus thc colrula. Geolgiancoruescloscstto tlrc SAlr prototypc in that its transitivepcrl'cctis bascdon a passivc palticiple.but this is conrbineclwith thc copulu rathcr than the transitivevcrb 'have'. so (hat t.lrcpcrlcct has a qulsi-pirssivcstluc[ulc, with thc agcnt in tlrc dative case('The 'l lcttcr is-writtcuto-mc'. r'atlrcrtban havcwlitten thc letter'). In Welsh. tlre ;rell'ectis fornrcdrvith thc prcpositionrlerli'aftcr' ('SIrc is aftcl sclling thc housc' for' 'She has solcl -l-hc tltc housc'). casl.cruIndo-Eulopeanlarrguagcs ulso lack a 'lravc'-ltcrfcct{ lbl instancc.botlr Pcrsianand I-lindi/tjrdu havc a perlcct ba.r^cd on a participlcplus thc copula, sornervhat likc Slavicantl Armenitrn). Dahl t1995.199(r3 : ( r 5 ) .t a k i n g a g l o b a l pcrspcctivc. uotcsthat thc'have'-pcr-fect is :rlmost exclusivelylbund in liuropc. Nou' one rnight ob.jectthat this is not a primitive feature of EulopeanInnguage.s. Many languages do not usc a transitivc'havc'-vcrblilr intlipossc'ssion clting prcclicirtivc at all. and it hirs in lact bccnsuggcstccl that the vely existeuce of a transitivcvcrb of preclicativc possession (c.g. I-irzarcl1990:l4(r-.47; is a liut'opcanism "1.,'cxprcssion Ilcnvcnistc1960[966: l9-51: la plus couranteclu rapport indiquc clansnos langucspar nrurr s'inoncc il f invclse par [trc ri ... Tellces1la situationdansla nra.jolitcdcs 'fhe langues.") restrictionof a 'havc'-perfcct Lo Europc woulclthen bc.jursta collsequcnce of this (cf. Dahl 1990:7). l-lowever.so far no publi.shccl lcscatclrlrasclocullrcntecl an arcirl lcstliction lirr' 'havc' vcrbs. Frcaclclresscd by {'r"rlthcr' lescarch.Il''hervc'-vclbsturn out to be typical of Eulope. tlrat woulcl fit with tlre tenclcncyof lJulopcan languagesto have norninativeexpeliencers in expclientialverbs(sec thc rrextscction). 2.4. Nominativccxperiencers 'l'hcrc arc tw() witysol'cxplcssingcxpcricuccr illguntct'lts of verbs of sensation.cl'uotiotl. cognition and pcrception:The exlrericncer

t496

I 1ll!l :lill "':

XIV.

'fypological characterizatiou of languagefamiliesand linguisticarcas

may be assiurilatedto agentsand coded as it nourinativcsubicct(e.g. / /iAcit), or it ntrv be assiurilated to a paticnt or goal, so that the stimulusargumentis codcd as thc nominative subject (e.g. 1/ pleuseslrc,). In Bossong's(1998)typology,the first typc is called generali;ing,and the secondtypc is called irtverting.Bossongstudiesthe expressionof ten common expcricntial predicatesin 40 Europeanlauguages.llc cornputcsthc rclation betweeninverting predicatesand generalizingpredicates, arrivingat ligurcsbctween 0.0 fbr English (where all predicatesare generalizing) and 5.0 for Lezgian(wherc all predicates are invertingl.By arbitrarilydividing the languagesinto those showing predorninantgeneralization(ratios betwecn0.0 and 0.8) and those shorving prcdourinant inversion(ratios betrveen0.8 and 5.0). we arrive at the geographicalpattcrn shown iu Map 107.4.

Europeanis fairly clear: lndic languages are rvcll-kno*'n tor thcir "dative subiects"of expericncervcrbs, so agaiu the l'caturcis not genetic(seealso Masica 1976,especially Map 6. for the areal distribution of dative subjectsin Eurasia and northern Africa). (See Haspelmath2001 for rnore discussion of experiential predicatesin European languages.) 2.5. Participialpassive StandardAverageEuropcanlanguagestypically have a canonicalpassiveconstruction (* Art.67) formed rvith a passiveparticiple plus an intransitivecopula-like verb ('be', 'becorne', or tl.relike). In this passivethe original direct object becomesthc subjectand the original subjcct may be ornitted,but it may also be expressed as an adverbialagent phrase.Such constructionsoccur in all Ronlauceand Germaniclanguages,but also in

l:irno87) N(r() l:)

Swdo I

S.rm(il.8l)

L\lru.83) I-tv(t s)

f

Lil(ori:r)

l

t .ii

t;,ili

IJB()0) llrt(01{)

I'ol(0.881 Cz(o 76)

F r ( ou ) &il(0.10)

tliiii,,

I f{iiiti,

t)ut (o fl) Cnn(t;rt

I lngon) It(r{8) SCr {}ri

sl'nor:t

Ilonr(2.b)

I ' r t ( ol J l

Blg()s)

Ulilllr

Crktr:zlt

Trkru

Map 107..1:Prcdontinantgcrrcrrlization (cel)ter)vs. inversion(pcriphcry)

Thus, Bossong'sstudy basically confirms earlier claims (Lazard 1990: 246-47, Dahl 1990:7) that the gencraliziugtypc is characteristic of SAE. although some of the figuresare perhapsa bit surprising(e.g. the tact that Hungarian turns out to be more SAE than German or Dutch, and the inclusionof Turkish, but not Romanian or Albanian. with respectto tiris feature).It is not possible to explain everythinghere, but rve evidently have before us a fairly typical SAE pattem with French and English at tl'recenter,Celtic (plus Icelandicthis tirne)at the rvesternmargin, Balto-Slavic,Fiuno-Ugrian and Caucasiatrat the easternmargin,and lairly gradual transitionswithin the macro-areas.No systcmatic world-widestudieshavc beenmade. but at least the behavior of easternIndo-

all Slavic(includingEast Slavic)and Balkan languages,as rvell as in lrish. The geographical distributionof suchparticipialpassives is s h o w ni u M a p 1 0 7 . 5 .

Lt!

Ir

ht Dut

Fog

Cnn

Brt

RE

Pol Cz I lng

ljr

Lkt

5ln It

SCt llun

Srd

Spn

Alb

l'rt Nllt

Bl8

Crk

Map 107.5:I'articipialpassives in Europc

V.E

t491

l0?. The Europeanlinguistic area: StandardAverageEuropean

No passivesexist in Nakh-Daghestanian and in }iungarian, and passivesof different formal types are found in Turkic, Georgian, and and in Armenian (stem suffrx), in Basque, 'get'-passive: 'Terry got Celtic (cf. the Welsh his hitting by a snowball' for'Terry got hit by a snowball'). Finnish and Irish have passivesofa different syntactic type: In this construction, only the subject is backgrounded, while the direct object remains in its place. Participial passives are very rare in lansuasesother than Standard Average Euroa i"ui. ln Haspelmath (1990) I surveyed world-wide sample of eighty languagesand found that a passive exists only in the minority of the languages(thirty-one)' Of these thirty-one languages,only four have.a passive iormed from a participle plus an intransitive auxiliary and two of them are European languages (Latin and Danish). The most common formal type of passive is the stem sufftx (found in twenty-ftve languages)' Syntactically, the possibility of an adverbial agent phrase is also by no means universal, but it is characteristic of SAE languages(Lazard 1990:246). It must be admitted that the SAE status of this feature is less evident than that of the frrst two features because the eastern lndoEuropean languages also tend to have passives of this type. In fact, in my 1990 study' the two non-European languageswith participle-auxiliary passiveswere Baluchi (an lranian language) and Maithili (an Indic language).Thus', one might say that this feature is an Indo-European genealogical feature' However, at least the Celtic languages and Armenian, two non-SAE branches of IndoEuropean, do not have such passives, and Maltese is a non-Indo-European language with such a passive(calqued from Italian)' 2.6. AnticausativeProminence There are three ways in which languagescan expressinchoative+ausativealternations such 'get losVlose', 'break (intr.)/break (tr')', as 'rise/raise'.One is by means of a causative derivation (- Art.66), i.e' a derived verb based on the inchoative member of the alternation, e. g. Mongolian xail-uul''melt (tr)', . from xajl- 'melt (intr.)'. The second is by means oi an anticausative derivation, i. e. a derived verb based on the causativemember, 'change (intr')', from e. g. Russian izmenit''sja tchange (tr.)'. (The third type, in izimi{ which neither member is derived from the other, i. e. non'directedalternations,will not

be consideredfurther here.) In Haspelmath (1993),I examined3l verb pairs in 2l-J1nguages and found that languages differ palrs greatly in the way inchoative-causatlve are expressed:Some languagesate anticaupreferring anticausativesto sative-protninent, causatives,while others are causalive-prominent. lt turns out that anticausative-prominence is a characteristicfeature of SAE' In my sample, German, French, Romanian, Russian, Modern Greek and Lithuanian show the highestpercentagesof anticausative verb pairs (between 100% and 74'h of all pairs that do not belong to the third,-nondirected, type). The percentagein the European languagesof my sample are shown in Map 107.6. Fin 47"k Lit

Jng 44% Rom 96%

lzg 40'/"

65'/"

70- 100%anticausatives - - - - 50-70% anticausatives pairs of anticausative Map 107.6:Percentage By contrast, Asian languages show much lower percentagesof anticausatives,preferring causativesinstead(e.g. Indonesian:0'%. Mongolian: 1l%, Turkish: 34%, Hindi/Urdu 35u/o,Lezgian: 40'%). An intermediateposition is occupied by the Finno-Ugrian languages of eastern Europe (Finnish 47'%' Udmurt 46oh, Hungarian Muk) as well as Georgian (68%) and Armenian (65'lu). In a study involving more languagesfrom Asia, Africa and Europe but less language-particular detail. Masica (1976) found a clear distinctive pattern for Europe: few causatives, heavy reliance on anticausatives(see especially his Maps 2 and 3). ln a recent worldwide study of 18 verbs from 80 languages, Nichols et al. (to appear) report that in inpairs involving inanimate choative-causative participants (i. e. the most typical subtype)' ihe causativeis generallyfavored worldwide and is strongly disfavoredonly in Europe.

r498

XIV. Typologicalcharacterizationof language familiesand linguisticareas

is not an Indo_ _ Anticausative-prominence turopeanism: Older lndo_Europeanhad a productive causative formation, which losi its productivity in the European b;.;;;; but continued to be produciive in .urt.in Indo-European(cf. the low ngrr. of:jy"-un_ ticausatives in Hindi/Urdu).

,1ll!l titi ,'i l;,i

t

;

t:,ii f i.ii;, [iiili.

2.7. Dative externalpossessors In Kdnig & Haspelmath(199g)and Hasoel_ math (1999),we studiedthe clistribution of external possessorsin thirty European lan_ guages(- Art. 73). We found three main lan_ guage types in Europe: (i) those with dative ex.ternalpossessors. e. g. Gernran Die Mutter t'uscttt dent Kind die Huure .The mother is washingthe child'shair., (ii) thosewith loca_ tiv_eexternal possessors, e. g. SwedishNrigon briit armen pd honont ,soireone brokJK arm (lit. on him)', and (iii) those that laci externalpossessors and must expressporrar_ sors NP-internally, e. g. Englislr. me SAf teature.externalpossessors in the dative, is lound ln Ronrance.ContinentalWest Ger_ manic, Balto-Slavic,Hungarian anO Salkan languages (Greek, Albanian). North Germanic and Balto-Finnic languagesnave tocative external possessors,i. e-.th--eyura ,or"_ what.peripheral SAE languages*itt, ..rp".t to this feature.The geographical distribuiion l s s l r o w ni n M a p 1 0 7 . 7 .

pol

D\]t Cm

Ru

Cz

F.

Hng

Bq

hyena ate the hare's fish'). This type is not found in Europe at all. Conve.srty,'Outiue fxternal possessors seemto be very rare outside Europe (the only caseI am awire of is E;.. cf. Ameka 1996),so this is a u".y .oburr"î ample of an SAE feature. 2.8. Negativepronouns and lack of verbal negation The areal distribution of negationin Eurooe '& has been studied in detaif by Bernini Ramat (1996) (see also Ramat & Sernini 1990).Here I. rvill single out just on. urp..i ol negatlon.the cooccurrence ofverbal nisa_ tion with negativeindefinite pronounr. i dlrtinguish rwo main types:(i) V + Nt (verb I negative indefinite), e. g. German Niernand kontmt 'nobody comes', and (ii) NV + NI (negated verb + negative indefinitet. e. e. Modern Greek Kandnasdhen irxete ,noboiv (lit. not) comes'.A third, *l^.a tvp. rnintl be distinguished in which verbai n.nuiion cooccu.rswith negativeindefinitesonl/when the indefinite follows the verb but noi when it precedesit. e.g. Italian Nessunoyiene,no_ body comes'.but Non ho vistonesszro.Not I have.seen.nobody'. For our purposeswe can crassrlythls type as a subtypeof (i), V + NI. The Standard Average European typ. L V + NI (cf. Bernini & Ramat f SgO:f S+,Has_ pelmath 1997:202).It is found in French(if we disregardthe particle .,le),Occitan and all \rennantc tanguages.as well as (in the mixed vanety) in Ibero- and ltalo_Romanceand Al_ banian (but not in Romanian or other Bal_ kan languages).The geographicaldistribu_ tlon of the typesis shownon Man 107.g.

Ukr

stn It

spn Pd

fft

Ronr Alb

Nllt

Fin

scr

srd

BtB

Lfu Lir

Grk Pol

Map 107.7:Dativeextemalpossessor In the far west (Welsh.Breron.English) and ln tne southeastlTurkish.Lezgian)ol-Europe thereare languages which do*not fruu. .îJr_ nal possessors at all. The easternlndo_Euro_ pean languagesKurdish, persian and Hindi/ Urdu alsobelongto this type.OutsideEurope a fourth rype enjoysconsiderable popularify: the "relation-usurping"type. where'he pos_ sessor"usurps" the syntacticrelation of the possessum (e.g. Chichewa,a Bantu language, 'The has hyena are the hare the fish' i"i .firJ

H"g

Rom

Brs

I zo

aG)-"

_.\\--l Irk

M

Map 107.8:.Lu1ey"_g.r lacking verbal negation wlth a negativeindefinite All the eastern European languages (Balto_ Jlavrc. Frnno-Ugrian. Turkic, Nakh-Daghes-

of Georgianiand llniol).wirh the exceprion the Celtic languages in the west ihow the NV + NI type.This typeis alsothat of the

107. The Europeanlinguistic area: StandardAverageEuropean

eastern Indo-European languages (lranian and lndic), as well as that of the clear majority of the world's languages:Kahrel (1996) his studiednegation in a representativeworldwide sampleof 40 languagesand found only five languages with V + NI negative patterns, one of which is the SAE language Dutch (the other four are Mangarayi (Australia), Evenki, Chukchi (Siberia)'and Nama (southern Africa)), as against 4l NV + Nl patterns,and sevenothers. I found a very similar pattern in my (non-representative)sample of 40 languages(Haspelmath 1997: 202). 2.9. Particlesin comparativeconstructions Comparativeconstructionswere investigated by Stassen(1985)in a world-wide study of l9 languages(- fut.75). Stassendistinguishes six main ways in which the standardof comThree kinds of locparisonmay be expressed: 'bigger ('bigger from X', itive comparatives 'bigger at X'), the exceedcomparative to X', ('Y is big exceedingX'), the conjoined comparative ('Y is big, X is little'), and the particle comparative('bigger than X')' The particle in this latter type is often related to a relative pronoun (cf. English thanlthat' Latin quamlqui),and the casemarking of the standard is not influencedby the particle (so that 'I it is possibleto distinguish love you m-ore than she' from'I love you more than her'). fu Heine (1994)notes,the six typesare not evenly distributed among the languages.of the world. Of the l8 particle comparatives in Stassen's\ample,l3 are in Europe, and of the l7 European languagesin the sample,l3 have a particle comparative.The distribution within Europe again conforms to our expectations: Particle comparativesare found in Germanic, Romance, Balto-Slavic, the Balkans,Hungarian, Finnish and Basque,so this is the SABtype. The distribution is shown in Map 107.9.

,'uby t", " '.

Ttk

particle comparative - - - - locative comparatlve Map 107.9:Comparativetypesin Europe

t499

The locativecomparativesare all at the western fringe (Breton) or the easterntiinge of Europelf innish, Russian,Nenets. Ubykh' Turkish. Laz). The other two types do tlot exist at all in Europe the exceedcomparative is founcl particularly in Africa. and the conjoined comparative occurs only in the Americasand Oceania. q u a t i v ec o n s t r u c t i o n s 2 . 1 0 .R e l a t i v e - b a s e d Comparisonof equality (equativeconstructions) is discussedlessoften than compartson a of inequality,and nobody has undertakenstudy of equativeson a world-wide scale' Still, there are good reasonsto think that equativeconstructionsprovide evidence.for Sfandard AverageEuropean (Haspelmath& Buchholz1998).ln Europe.tnany languages have an equativeconstructionthat is based constructlon' on an adverbialrelative-clause For example,Catalanhas /or Z corn X'as Z as X' (where Z is the adjectiveand X is the standard).Catalan cont is an adverbialrelative pronoun, and tan is a correlativedemonstratlve.A very similar constructionis found elsewherein Romance (Portugrteselrio Z contoX, Occitan tan Z cornal'), in Germantc (German so Z tt'ie,f . in Slavic (Czechtak Z jako X, Russiantak(oi) le Z kak .l,), in Romani (katle Z sar l). in Hungarian (olvan Z mint X), in Finnish (niin Z kuin X), and in Georgian (isetiveZ rogorc-l'). ln the English constiuction, the relative-clauseorigin of cs is not fully transparent synchronically.but diachronicallyas derivesfrom a demonstrative (eull slld > a// so) that was also used as a relativepronoun. In some Balkan languages.the correlative demonstrativeis not 'as used (e.g. Bulgarian xubaw koto lebe pretty as you'), but the standardmarker is clearly of relative-pronounorigin. (Thereis probably some connectionbetweenthe relaiiue-p.onounorigin of equativemarkersand the relative-pronoun origin of comparatlve standardmarkersthat we saw in $ 2.9.). Non-SAE languageshave quite dift'erent Many SOV languages equativeconstructions. in eastern Europe have a special equative standard marker (Lezgian -ti:, Kalmyk iitrg: also Basque be:ain and Maltese daqs), and the Celtic languageshave a special (nondemonstrative)marker on the adjective(e.g. Z with X'). ln lrish chornhZ le X'EQUATIvE 'equword the languages, Scandinavian the ally' is used on the adjective(e.g. Swedish lika Z som X'equally Z as X'). The distribution of the relative-basedequative con-

107. The Europeanlinguisticarea:StandardAverage Europeurr

, in theeastern Caucasus, and indee.d in manv otherpartsof the world,but they*uy nru.", .havehadsubjectpersonagreement mirking.) 2.12.Intensifier-reflexive differentiation Intensifiers are wordslike Englishself, German sclbsl,Frenchnftnrc and Russiin ,ralr that characterize a noun phraserel.erentas centralas opposedto an implicit or explicit periplrery(e.g. Thepope hir)selfgow i, i,i audience,i. e. not just the ior*ainolsfArt.57: Kcinig& Siemund1999).In manv Ianguages, the intensifie,.^pr.rrion i, ;i;6 used. as a reflexivepronoun,for inrtan." in Persran(xod-ai 'himself': Huiang xocl_oi 'Hushang himself', and Huiang xoias_ru2 did self-eccsaw] ,Hushing saw hini [Hu-shang self). However,a featurethat is"typicai of SAElanguages is thedifferentiation oi..ff.^_ lvepronounsand intensifiers (Konig & Has_ peln'-'h1999). For instance, German'nas srclr (refir".ve)vs.selbst(intensifier), Russiantras sebjavs.sall, Italianhassi vs.rre$o, Greck ltaseaft6vs.idtrjos.Map I 07.I 2 ,f,o*, if,.l"n_ guages in Europewith specialrcflexivepronounsthat arenot identificalto intensifiers.

Brt

u6g sPn

Rm Alb

Lzg,

striking, but which nevcrthelcss secnrgood candidate-s for Europeanisms. No maps-rvill be given for theset'eatures,ancl the .ui.l.n.. will be summarizedonly briefly. 3.1. Verb fronting in polar interrogativcs l n t h c l l l g c r n l . i o r i t vo l ' l u u g u i r u e . s . folur irrtcllogativcsart urar.kedby interrogativein_ tonation or illl interrogativcparticli or both (- Art. 77). ht his sanrpleof' 79 languages.

Ultan (1978)tbundonly ,.u.n tu,rtu,ft"i

showing the altcrnative srrategy oi uJ,.U fronting (oftcn called,'subject_i,erbin.",er_ !lon"). Of these.six are European(English. !pn9h, Rornanian. Russian. Hungo-ii,,,r, Finnish; the sevenrhlanguageis lufafiyj. i" that the SAlr status of verb fronting ieenrs beyond doubt. In lirct, the large rn,ri-o.ityoi Germanic, Romance and Slir-victa,rguog., (plus.Mo.dcrn Greek) appear to hauJ virb t r o n t l r ' t gr n p o l u r q u c s t i o n si r r o n e f o r . n r or another.The three Europeaulanguageslor which Ultan explicitly reports thit rio verb rrontlng_occurs are pcr.iplrcral: Basque.Girc_ I r c a n d L i t h u l n i a n . l : u r t l r c r r r r o r cS.n E l u , . r _ guagesare characterized by tlreabsence of an interrogativeparticle. In UItan.s Ootn, tir. nine . E u r o p e a nl a n g u a g c se x h i b i t i n gu ' p , i r _ t r c l ci n p o l a r q u c s t i o n sa r c a l l p c r i p h c r i i l to a greateror lesscrextcnt:Ilasque,Irish, Scot_ tish Gaeiic, Albanian, Hungarian, Litlrua nian. Russian,Finnish, Turk]sh (and I can add Nakh-Daghestanian). Vcrb fronting iir polar questionswas suggestecl as a Eiiro_ peanism already by lleCkman ( 1934) (c.f. Dahl 1990).

c.8

Blg

Trk

Am

Map 107.l2: Intensifier-reflexive differentiation

Intensil'rer-reflexive differentiationis not an ropeanism,becauseeasternIndolldo European languages havethe ,ua. .*o..._ sronlor intensifiers and reflexives (e.g. per_ sian.r-or/-ai, Flindi aap).Thereare nJ oJ_ Irshedworld-widestudiesyet, but it seenrs that non-differentiation ii very .ornrnon aroundtheworld,andwhilediffeientiation is alsofoundelscwhere, it is not io"n,l i,i;;;;; unmedrately adjacent to European languages. l.

r50l

Somefurther likely SAE features

n thissection,I will mentiona few features vhicharelesswell-documented thanthoseiu :2, or whosegeographical distribution is iess

3.2. Conrparativemarking of adjectives Most Europeanlanguagcshavespecialforrns lor ac|cctlvesoccurringin cornparative col.l_ structlons. For instancc. En-clishuscs thc sulllx -crîn (The dog-isbigg_er tltr-,,t .thisway Iltecat).Suchan inllcctionalmarkei of adiec_ trvesis not cotlrntouin the world's Iancuaces outs.ide^ of_Europc.Somc languagcsrri roirl. ot adverbialparticlenrodilyingthe ad_ l{lnd Jecrlve1'rnore').but perhapsthe most cont_ mon type is represented by Japanese. rvhcre the comparativescmanticsis iar.iea Uy tlre starrdardrnarkeralone (e.g. itru_gu,,rki ,.u,li ookli [dog-sunrcat from big] .rhJ dog is Uig_ ger than the cat'). Specialcomparativeforms are found in all ^ Germanic. Balto-slavic and Balkan lau_ qy3Sel(wirh the exceptionof Ronraniauand A l b a n r a n ) ,a n d r n o s t R o r n l n c c l a n g u a g e s preserveat leastfour suppletivefonni (elc.

r 500


in the world's languages,and strict subject agreementis characteristicof a few European languages,some of which happen to be wellknown. In her world-wide sample of 272 Nnts languages,Siewierska(1999) finds only two strict-agreementlanguages,Dutch (an SAE Kom language) and Vanimo (a Papuan language of New Guinea). Siewierska further notes R6 Udm Pol Dut Eng that outside of Europe, she is aware of only Cz Gm Tat Uk Llng Fr two additional strict-agreementlanguagesthat Sln are not in her sample (Anejom and Labu, two It SCr Oceanic languages).Gilligan (1987) reached LzB a similar conclusionon the basisof a sample crg of 100 languages.The distribution of strict subject agreementmarkers in some European equative-constructions languages Map 107.10:Relative-based is shown in MaP 107.11.

structionin Europeis shownin Map 107.10' following Haspelmath & Buchholz (1998: 297\.

i'''tll!t tiil :r:.

', ,': !

equatives relative-based Impressionalistically, seemto be rare in the world's languages,and the easternIndo-Europeanlanguagesdo not seem to use them in general (however, a counterexampleis Punjabi).

Fin Est Lfu

ur

@u*

Pol

I

2.11. Subjectpersonaffixesas strict agreementmarkers The majority of the world's languageshave bound personmarkerson the verb that crossrefer to the verb's subject (or agent). When thesesubjectafftxescooccur with overt subject NPs (full NPs or independent subject pronouns), they are called agreementmarkers. However,in most languagesthey can occur on their own and need not cooccur with overt subjectNPs. For example,in the Bul'you (pl.) work', we garian phrasevierabotite -ite (2nd person plural) seethe subjectsufftx cooccurring with the independent subject pronoun lie 'you (pl.)', showing that -ile is an agreementmarker. But in Bulgarian it is equally possibleand probably more common to say just rabotite'you (pl.) work', i.e. the subjectsullix can have a referentialfunction on its orvn. In German, by contrast. this is 'you work' \s ihr arbeit-el.Since not possible: the agreementsufftx -el does not have such an independentreferentialfunction, the subject pronoun rlr cannot be omitted' Lan"nonguageslike German are often called pro-drop languages", and languages like "pro-drop languages"; Bulgarian are called "strict-agreementlanbetter terms would be "referential-agreement languages". guages"vs. It has sometimesbeeu thought that strict agreement,as exhibited by German, English, and French, is the norm and that referential agreementis somehow special. But in fact, referentialagreementis far more widespread

LJk

Hng 9n

eq It

scr Rrn

Srd

Sf

Alb

Prt Mtt

Grk

GB

Bl8

"lte: "

Trk

Alm

[6nguagsswith strict subject agreement - - - - languageswith obligatory subject pronouns' lacking verb agreement Map 107.11:Obligatory subjectpronouns

The map showstwo non-contiguousareasin which subject agreementsuffrxescannot have a referential function: Germanic and GalloRomance languageswith Welsh on the one hand, and Russianon the other.Perhapsonly the western European area should be thouefit of as being relevant for SAE; in Russian, past-tenseverbs do not have subject person affrxes,so Russianis not a very good example of a strict-agreement language. In the easternNordic languages(Norwegian,Swedish, Danish), the subjectpronounsare obligatory as they are in English, German or Icelandic, but the languageshave lost agreement distinctions on the verb entirely (cf' Swedish 'llyoulhe bite(s)', jag biterldu biterlhan biter Icelandic 69 bftbrt biturlhann bitur). T\ese "non-pro-drop" in a languages are thus sense,but they are not strict-agreementlanguages.English is approachingthis type' as the only remnant of subjectagreementis the 3rd person singular present-tensesuflix -s. (There are also some languagesof this type

r!;l,i :,na

I 502

'lll l I llil

: t

l

I i, f:',rt, l.;:,1

h;;rl, I

rl:ttli;

EEIIIi


Italian maggiore'bigger', ntinore'smaller', migliore'better').Comparapeggiore'worse', tive forms also exist in Basque(e.g. haundiago'bigg-er'),Hungarian(nagy-obb'bigg-er')' Finnish (iso-mpi'bigg-er'),and other FinnoUgrian languages. Comparativeforms are not completelyunknown outside of Europe. Arabic has a special comparativefbrm (e.g. ?akbar'bigger', 'big'), but it is unique among from kabiir Afro-Asiatic languagesin this respect.Old Indo-Iranian languages had comparative forms, and the modern Iranian languages have preservedthem to someextent(e.g. Persian-ter. Zaza -tr\. But further east,in modern Indic, the comparative does not exist anymore,and languageslike Hindi-Urdu and Bengali use a construction analogousto the Japaneseexamplejust cited. Similarly, in the Uralic languages,the further east we go, the fewer comparativeswe find. For instance' Khanty (a Finno-Ugrian languagespokenin westernSiberia,i. e. outside of Europe) does have a comparativeform in 'sak (e.g.iant-sak 'better'), which is used when no standard is present.But in a completecomparativeconstruction. no marking is found on the adjective (e.g. narl ke:se:-ne:x'eltiarn [you knife'better than your knife'. Ni2sc from goodl kolaeva 1999:21). Thus, although this featureis not contined to Europe, it is typical of a SAE feature in that it is robustly present in western [ndoEuropean and Uralic languages,but gets rarer the further eastwe go in thesefamilies. " A a n d - B "c o n j u n c t i o n 3.3. The feature discussedin this section is less distinctivethan the others mentioned so far' but I hope to show that it is not at all devoid of interest. Stassen(2000) offers the first world-wide typologicalstudy of NP conjunction strategies,based on a sample of 260 languages(- Art.82). He distinguishestwo basic types, and-languages(using a symmetric particle) and l'illr-languages (using an asymmetric comitative marker). Two thirds of Stassen'ssample languages are and-lan' guages,and sinceSAE clearly belongsto this type, too, it is not a very distinctiveproperty. cover all of northern Eurasia, And-languages South Asia, the Middle East and northern Africa, Australia, New Guinea, and parts of Central and South America. Mllr-languages are encounteredin sub-SaharanAfrica, East and SoutheastAsia, the islands of Oceania, and large areasof North and South America.

However, within the and-languagesthere are position of severalsub-typesaccordingto the"A the particle, which we may call "A and-B", "A-ind 8", "A-and B-and", and B-and" "and-A (of the remaininglogical possibilities' "and-A andB" seemsto be inexistent,and B" occursonly as a secondarypattern).Most all European languages, and in particular "A SAE languages,belong to the sub-type"A"A-and B-and" and and-B". The types and B" are found in some languagesof the Caucasusand in some Turkic languagegas well as scattered throughout northern Eurasia and South fuia (e.g. in Abkhaz, fuchi, Persian.Sinhalese,Tamil, Burmese,Korean accordingto Stassen;Stassenalso points out that there is a correlation with verb-final word order here). Furthertnore' some peripheral European languagesmake restricteduse (e.g. Russianmy s toboj of the ruitft-strategy 'we with you', and also Old 'I and you', lit. Irish, Lithuanian, Polish and Hungarian, these according to Stassen).Taken together, "A anddata do show that belongingto the B" type is not a trivial feature of the SAE linguistic area. 3.4. Comitative-instrumentalsyncretism In all SAE languages,the preposition that accompaniment(: comitative)also expresses servesto expressthe instrument role (e.g. English with: w'ithher husbqndlrith the hamrrer). Such languagesare said to exhibit comitative-instrumentalsyncretism.Stolz (1996) studiedcomitative and instrumentalmarkers in a world-wide sampleof 323 languagesand found that this kind of syncretismis typical of Europe. Non-European languagesmore commonly possessseparatemarkersfor these 'with two semantic roles (e.g. Swahili na As (instrumental)'. (comitative)', krvc'with of Stolz's thirds two about shows, Table 107.1 samplelanguagesare non-syncretic,and only one quarter is syncretic.(The remainingl1nguugesbelong to a mixed type, which I ignore here for the sakeof simplicity;thus, the percentagesdo not add uP to 100%.) Two areas diverge significantly from the generaltrend: Oceaniahas far lesssyncretism ihan the world average,and Europe has far more syncretism than the world average' When we look at the pattern within Europe' it becomeseven clearer that we are dealing with an SAE feature (as Stolz recognizes,cf' languages 1996:120).Of the l6 non-syncretic in Europe, l0 are Caucasianlanguages,i.e' they are clearly outside of SAE, and one is


I 503

Table 107.1:comitative-instrumental:syncreticand non-syncreticlanguages

syncretic(e.g. English) languages percentage Europe Africa , Americas

tuia Oceania World

25 20 l6 l2 6 79

49% 3t% 2t% 18% t0% 24%

only politically, not anthropologically, in Europe (Greenlandic). Four of the remainine five languagesare also otherwisenot typica-i instancesof SAE (Basque, Finnish, tr,tattese. Mari). And when we look at the 38 IndoEuropean languagesin Stolz's sample, we seethat syncretismcannot be regarded as an Indo-Europeanism:Of the eight Indo-European languagesnot spoken in Europe, only th show syncretism,while five show nonsyncretism.Thus, in Asia Indo-Europearr languagesbehavelike Asian languages,and there is no generalpattern for Indo-European. 3.5. Suppletive second ordinal Most languageshave a suppletive form of the ordinal numeral 'first', i. e. a form not derived from the cardinal numeral .one'. An exampleis Gerrnan,where'lst' is erster(unrelated to eins'l'), contrastingwith other ordinals such as zweiter'2nd' (cf. zwei ,2'\, vierter'4th' (cf. vier'4'), and so on. In Stolz's (2001b)study of 100 ldnguagesworld-wide, there are 95 languages with special ordinal numerals, and of these, 78 have a suppletive word for 'hrst'. Thus, languages that say (literally) 'oneth' for 'lst' are not common. However, the same sample has only 22 languages in which the word for .2nd,, too, is suppletive and not derived from..2' (e.g. E1 ;h second).Thus, most languageshave (literally) 'twoth' for '2nd'. The 22languages that have a suppletive '2nd'word are lieavilv concentratedin Europe: 17 are European languages,and this type is clearly the mijority within Europe (which is representedby 27 languagesin Stolz's sample).Of the l0 European languagesthat do not have a suppletive second ordinal, six are clearly outside SAE (Basque,Turkish, Armenian, Georgian, Lezgian, Greenlandic). Among SAE linguages, only some Balkan languages(Romanian, Albanian, Romani) and German lack a suppletive secondordinal.

non-syncretlc (e.g. Swahili) languages percentage

l6 38 54 4t

54 209

3r,yn 58% 69% 7lo 86% 65"1,

This is clearly a very marginal feature in grammar, but it is intriguing that it should show such a clear geographicaldistribution. 3.6. Some other characteristicsof SAE The features examined so far present the most striking evideucefor StandardAverase Furopean,but there are probably -uny -oi" featuresthat will turn out to be characieristic of the core European languagesin one way or another. In this subsection,severalsuch candidateswill be mentionedbrieflv. The first few featuresin the following list ire purely negative:At first glance,this may seemodd, but of course the lack of a category that is widespread elsewhereis no lesJ sienificant than the presenceof a category that is rare elsewhere. (i) Lack of an alienable/inalienable opposition in adnominal possession(' Art. jil. ln Nichols's(1992)world-wide sample.almost half of the languagesshow such an opposition, but no EuropeanlanguagedoeslillZ: 123).More generally,this opposition is rarer in the Old World and common in the New World, but in Europe it is even lesscommon than in Africa and Asia. (ii) Lack of an inclusive/exclusive opposition in.first personnon-singularpronouns.Again, this opposition is commonest in the New World and in the Pacihc region, but in Europe it is even rarer than in Africa and Asia, as was shownby Nichols(1992:123). (iii) Lack of reduplicatingconstructions.I have no systematicevidenceto back uo the claim that this is a characteristicfeature of European languages,but reduplicationis so common acrosslanguagesthat its almost to_ tal absencein the core European languages becomesstriking. (Interestingly,reduplication existedin older Indo-Europeanlanguagesat least in one construction, the perfect, but even here it was lost entirely by the Middle Ages.)

I 504

Typologicalcharacterizationof languagefamiliesand linguisticareas

(iv) Discourse pragmatic notions such as topic and focus are expressedprimarily by sentencestress and word order difl-erences (Lazard 1998:I I 6). Only the Celticlanguages and French give a very prominent role to clefting, and particles rnarking discourse pragmaticnotionsare virtually unknown. (v) SVO basicword order at the levelof the clause.This feature is of course found elsewhere in the world, but in Europe it correlates particularly well with rhe orher SAE features. The Celtic languagesin the west have VSO order (exceptfor Breton, which is also otherwisemore SAE than Irish and Welsh),and the easternlanguageshave SOV word order. Interestingly,Balto-Finnic(Finnish, Estonian,etc.)and (lessunequivocally) Hungarian have SVO word order, whereas the easternUralic languages haveSOV.Similarly, the eastern Indo-European languages tend to show SOV word order. (See Drver 1998 ior more on word ordcr in the linguagesof Europe.) (vi) European languagestend to have just one converb(- Art. 83) (cf. Nedjalkov 1998). For instance, Romance languageshave the gerundioIgt ronrlrf, English has the -lng-form, and Slavicand Balkan languageshave their adverbial participle. The Celtic languagesin the westcompletelylack sucha fonn, and the languageseast of SAE tend to have more than one converb. Otherwisethe core European languagestend to have adverbial conjunctions (' Art.63) to make adverbial clauses.According to Kortmann (1997: 344\, they have "a large, semanticallyhighly differentiated inventory of free adverbial subordinators placed in clause-initialposition". More generally,they tend to have finite rather than non-finite subordinate strategies (r Art. 100), though a multi-purpose infinitive usually exists (except lbl the Balkan languages). (vii) Europeanlanguagesusually have a special construction for negative coordination, e. g. English neitlrcr A nor B, Italian ni A nt .8, Russianni A ni,B, Dutch nochA noch B, Hungarian sem A sem B. Again, no worldwide study has been published,but such a negative coordinating construction is rarely reported from languagesoutside Europe (cf'. Haspelmathto appear). (viii) SAE languageshavea largenumberof characteristicpropertiesin the area of phasal adverbials(expressionslike alread1,,still, no longer, not yet) (van der Auwera 1998b). Theseare rather well documented.but for the

detail I have to refer the reader to van der Auwera'sthorough study. (ix) "Preterite decay": the loss of the old preterite and its replacementby the former present perfect. This is a change that occurred in the last milleniumin French,German and northern Italian, as well as in some other adjacentEuropeanlanguages(cf. Thieroff 2000: 285). Its distribution is far narrower than that of the other Europeanisms, but it is the only feature of those studiedby Thieroff whose geography comes close to Standard Average European (cf. also Abraham 1999). Quite a l-ew additional features have been mentionedin the earlier literature as characteristic of SAE, but earlier authors have sometimesneglectedto make surethat a proposedEuropeanismis not also common elsewhere in the world. Most of Whorf's original examplesof SAE featuresseemto be of this kind. For instance,he notes that in contrast to SAE, Hopi lacks "imaginary plurals" (such 'ten as days', according to Whorf a "metaphorical aggregate").But of course,we have no evidencethat such plurals of time-span nouns are in any way characteristicof European languages.It may well be that they are common throughout the world. (To give Whorf his due, it must be added that he was not interested in demonstrating that SAE Ianguagesform a Sprachbund.He just used this term as a convenient abbreviation for "English and other European languages likely to be known to the reader", without necessarilyimplying that theselanguagesare an exclusiveclub.) 4.

Degrees of membership in SAE

Membership in a Sprachburrdis typically a matter of degree.Usually there is a core of languagesthat clearly belong to the Sprachbund, and a periphery of surrounding languagesthat share features of the linguistic area to a greateror lesserextent. In order to quantify the degreesof membership in SAE, a simple proceduresuggests itself that was first applied to areal typology by van der Auwera (1998a).In addition to individual maps in which the linesdenoteisog l o s s e (sa si n M a p s 1 0 7 . 1 - 1 2 ) w , e cancombine ditlerent features in a single map and show the number of isoglossessharedby the language.Map 107.13showssuch a "cluster map" in which the linesstandfor "quantified


isoglosses"(or "isopleths"). The map combinesnine featuresof$ 2.: definiteand indcfinite articles,relativeclauseswith relativepronouns,'have'-perfect,participial passive,dative external possessors,negative pronouns and lack of verbal negation, relative-based equativeconstructions,subjectpersonafllxes as strict agreementmarkers, and intcnsitierreflcxivediflbrentiation.The languagesin the nucleus(Frenchand German) show the SAI: value in all nine of these features.The languagesin the next layer (Dutch, other Romance, Albanian) show eight features, tire nextlayer(English,Greek,Romanian)shows sevenfeatures,and so on. In this map, the resulting picture is actually very clcar, becausethe SAE area with at least five SAE, featuresstandsout from thc remaininglanguages,which have at most two SAE fcatures. I:in --rL.t

Nnls

Lt Rrl

Rustl

Llttrn

I uLr

/

Irk

T.rr

. .. v.l: crs I Ann

Map 107.13: A clustermap conrbining ninefclturcs Suchclustermaps are thus a fairly direct representationof degreesof membershipin a linguistic area. But of course, the cluster rnao dircctly reflectsthe choiceof featuresthat arc combined,and this choice is always sornewhat arbitrary. Of the twelve featuresin g 2, onlv ninewereselectedherebccauseinforrnatic n the other three was incomplete.Icleally, the featuresof g 3 should have bcen added,too. But it seemsto me that thc mairr resultsof Map 107.13would uot be changed (this rnap can also be compared to thc vcry similarmap in van der Auwera (1998a:823), rvhich combines five adverbial featurcs or l-eatureclusters).The ntost striking featurcs o f M a p 1 0 7 . 1 3a r e : (a) The nucleusof Standard Average European is formed by French and German (a hnding that led van der Auwera (1998a:824) to proposethe term Clnrlemagne Sprachbund lor the nuclearareaof SAE). In viervof the

r 505

historical role played by speakersof these trvo languages both in thc earlyrr.redieval history of contiuentalE,uropeand !n tlie vety recentattempt at Eulopcan unillcation,this is of courscau extrcrlelyintriguingresult. (b) The southcrnEuropcanlanguages(both Romanccand Ilalkan languages) arc at lcast a s c l o s c ,i f n o t c l o s c rt o l h c n u c l c u st h a n t h c northcrn languagesaud l-uglish.This means t h a t i t i s n r i s l e a d i n gt o c a l l S A E l c a t u r c s "Westcrn Iiuropcan lcatur.es",as is sonrctirnes donc. It is truc that the Slavic languagesin the cast lack many SAE features. but thc Balkan langr.rirgcs ar.cgcncr.allv urorc SAE than Slavic,altlioughthcy are not wcstem European. (c) Englanclstandssonrcwhatapart ll-onithc E u r o p c a nn u c l c u s( a s n o t c d a l s o b y v a n d c r Auwcra 1998a:823), althou_qhit is closcly related gencalogicallyto Gcrnran ancl has beenthoroughlyinfluenccdby Frcnch.Sincc English is cr.rrrcntlythc donrinant languagc throughout thc rvorld, it is rvorth pointing o u t i t s s o n t e $ , h amt a r g i n a ls t a t u sa m o n g i t s Euroltcansistcrlanguagcs. -1 It is inrpclltantto kcclt irt rnind that thc l'cat u r c so n w h i c h M a p 1 0 7 . 1 3i s b a s e dh a v en r . - r t becn sclcctcdrandomly and are thus by no mcansrcprcscntative. of the ntorphosyntactic li'itturcsol' Luropcan languagcs.thcl' rvcrc includedpreciselybccausethcy werc knowrr to show a distributionthat suppor.ts the SAI: hypotlresis.Thus. no clainr is rnadc that all (or cvcn thc nrajorityol-)fcuturesrvill shorvu siurilar distribution. It is perl'cctlypossiblc that rve u,iil some day discover anothcr Spracltbuntl,bascd on a difi'crent set of t'catures,thal hasRussianat its coreand extends all the way to wcsternSiberiain the eastancl centralAsia in the south, but within EuroDe conrpriscs ouly the Slavic,Balkan.anclScaidinavianlanguagcs. This area would overlao w i t h S A E . b u t i t r v o u l d n o t c o n t r a d i c ri i . Thus, a languagenray in principlebelongto dill'crcnt linguistic arcas. aud dif]'cr-entlinguistic arcas uray cocxist "on top of' eaclr othcr. Sinceareal typology is only in its infancy, rve do not knorv how comnron such situaLionsare, but nothing in the logic of a Sproclbuttdimplicsthat thc rvorld shouldbc exhaustivelydivisible into non-overlapping Sprachbtindc. In fact, a nuurber of srnaller linguistic areaswithin Eulope have bcen proposedin the literature(apart li'om the Balkan arca. rvhoscimportanceis ltot doubtcdby anyonc).

l 506

XIV. Typologicalcharacterization of languagefamiliesand linguisticareas

e . g . b y L e w y ( 1 9 4 2 ) .W a g n e r( 1 9 5 9 ) .D e c s y (1973),Haarmann(1976),and Ureland(1985) (cf. also Wintschalek 1993on a Volga-Kama area).Currentlythe most thoroughlystudied areas are the Circum-Balticarea (cf. Stolz 1991, Dahl & Koptjevskaja-Tamm(eds.) 2001) and the Mediterraneanarea (cf. Cristofaro & Putzu (eds.) 2000). However, no strongclaimsabouta Circum-Balticor a Mediterraneanlinguistic area seemto have been madeas a resultof thesestudies. 5.

How did SAE come into being'l

Linguisticareasarisethrough languagecontact, but preciselywhich contact situation gave rise to Standard Average E,uropeanis not inmediatelyclear.And what is the source of the variousEuropeanisms: Who borrowed from whom? A full discussionof the sociohistorical,cultural and sociolinguisticissues is beyond the scopeof this article,so I will restrict myself here to mentioningjust hve possibilities:

'iil!l riI

(i)

(ii)

(iii)

(iv)

(v)

L

retentionof Proto-Indo-European structures and assimilationol some nonIndo-Europeanlanguages to lndo-Europeanlanguagestructure; intluence from a common substratum of a pre-Indo-Europeanpopulation in Europe: contacts during the great transformations at the transition from late antiquity to the early Middle Ages in Europe; the oftlcial language (Latin) and the common Europeanculture of the Middle Ages; the common Europeanculture of modern times. liom the Renaissance to the Enlightenment.

The tlfth possibilitymust be rejectedbecause a time depth of 300-500 yearsis not suffrcient to account for grammaticalcommonalities of the kind discussedabove. If lexical similaritiesbetweenthe European languages are discussed- tor instance neoclassical compounding (socio-lpaleo-lortho-ldemo-, -graphyl-log1,l-cracy, etc.) or idiomatic structure (e.g. ivorl' to*'erltorued'avoriolElferrbeintunn, as poor as a clurc'lt tnouselpattvre conune un rat d'igliselarnr v'ie eine Kirchennraus)then the last severalcenturiesare the appropriate time frame lbr explainingthe historical links. but the basicsyntacticstructurescommon to SAE languagesmust be older.

The first possibility must be rejectedbecause the great majority of Europeanisms are innovationswith respectto Proto-lndoEuropean.For instance,as far as we know, did not havearticles,a Proto-lndo-European 'have'-perfect. "A and-B" conjunction,strict subject agreement.particle comparatives,or relative clauseswith relative pronouns (cf. Lehmann 1974,Haspelmath1998).With reand alsowith spectto Proto-lndo-European, respect to the oldest Indo-European languagesattested in Europe (Ancient Greek, Old Latin, Gothic), StandardAverageEurop e a ni s c l e a r l ya n i n n o v a t i o n . The second possrbility,a pre-Indo-European substratumin Europe causingthe SAE features,would be extremelydiffrcult to demonstrate,but it might be worth pursuing.It is intriguing to note that the geographical spaceoccupied by SAE languagescoincides lairly preciselywith the areaof the Old European hydronymy, i. e. the homogeneouslayer of river names discoveredby Hans Krahe (seeVennemann1994for recentdiscussion). Vennemann(1994) proposesthat these Old Europeanhydronymswerenot coinedby an early prehistoricIndo-Europeanpopulation, peoplewhich he but by a pre-lndo-European calls Vasconic (the only surviving Vasconic language being Basque). Furthermore, the Old European hydronymy is hardly attested in the British lsles, where the Celtic languagesare spoken,i. e. they could not have been influencedby the Vasconicsubstratum. This is in perfect harmony with the well-motivated hypothesisthat the Celtic languages acquiredsome of their striking t'eaturesfrom a different substratum related to the AfroAsiatic languages(Pokorny 1927-30, Gensler 1993). The main argumentagainstthe substratum view is that the SAE featuresseemto be gaining ground too late for a pre-Indo-European substratumto have causedthem. Some SAE features appear only in the first millenium CE, but also the earlier featuresusuallycome fairly late, so that the earliestrecordsof IndoEuropean-languagesin Europe still show traces of the Proto-Indo-Europeanpatterns (e.g. causatives, relativeclauses,locativecomparative. "A B-and" conjunction).If these SAE featureswere causedby a substratum, then we should have much more evidenceof the population speakingthis substratumlanguage.Moreover, a Vasconicsubstratumcan hardly account for the SAE featuresbecause modern Basqueis in most relevantways very much unlike the SAE languages.

I 507


Of the remaining two possibilities,we can probablyexcludeoption (iv) (tne influenceof Latin in the Middle Ages),becausemost SAE featureswere absentin Latin and developed only in the Romance languages.There are only two featuresfor which Latin influence is a likely factor: negation and relative pronouns. In the caseof thesetwo features.the standard languagessometimesshow deviations from the vernaculardialects,so at least the written standard languages may have beeninfluencedby Latin, the Europeanwritten languagepar excellencefor many centuries. Thus, non-standard English has constructionslike I won't do nothing('I won't do anything'), and similarly in non-standard German and French (cf. Haspelmath 199'l: 205). Analogously, Latin-type relative pronounsoccur widely in the standardlanguages of Europe, but vernacular speechoften prefers relative particles (Lehmann 1984: 88, ' ^). However, Latin probably only helped t' reinforce these structures in those languageswhere they existed already independently as variants. Thus, we are left with option (iii), the time of the great migrations at the transition between antiquity and the Middle Ages. This seemsto be the appropriate time frame at 'have'-perfect, least for articles,the the participial passive, anticausatives,negative indefinites, nominative experiencersand verb fronting. The rise of these constructions can be observedonly with diffrculty becausethey wereby and large absentin the written classical languagesbut seem to be well in place once the vernacularlanguagesappear in the written record toward the end of the first millennium CE (cf. also Fehling 1980).This hypothesisderives some further plausibility from the fact that language contact must have been particularly intensive and effective during the great migrations, and in the case lrench and northern Italian we have ample recordsof the lexicaleffectsof thesecontactsfHowever,it is not so easyto fit features su'a'|fas particle comparatives, ,,A and-B" conjunction and relative pronouns into this picture, becausethese features seem to have developedaround the middle of the first millenium BC or even earlier (cf. Haspelmath 1998). Of course, we must always reckon with the possibility (or even likelihood) that different SAE features are due to different historicalcircumstances, and the correct picture is likely to be much more complicated than we can imagineat the moment, let alone discussin this article.

6.

Abbreviationsof languagenames

Alb Arm Blg Brt Bsq Cz Dut Eng Est Fin Fr Gae Grg Grk Grm Hng Ice Ir It Kom Lat Laz Lit Ltv Lzg Mar Mlt Mrd Nnts Nor Pol Prt Rorn Rus SAE Sam SCr Sln Spn Srd Swd Tat Trk

Ubv Udm Ukr Wel T.

Albanian Armenian Bulgarian Breton Basque Czech Dutch English Estonian Finnish French ScotsGaelic Georgian Greek German Hungarian Icelandic lrish Iralian Komi Latin Laz Lithunian Latvian Lezgian Mari Maltese Mordvin Nenets Norwegian Polish Portuguese Romanian Russian StandardAverageEuropean Saami Serbian/Croatian Slovene Spanish Sardinian Swedish Tatar Turkish Ubykh Udmurt Ukrainian Welsh

References

"Preterite Abraham, Werner. 1999. decay as a European areal phenomenon". .Folla Linguistica 33.1:11-18. Ameka, Felix. 1996."Body parts in Ewe grammar". In: Chappell.Hilary & McGregor.William (eds.).Ile grannnr oJ inalienabilitt,:A tt'pological

I 508

,. /illll' / rlt:r

I I tili : 1 . : .

.

i

perspectiveon bod,-part ternl.lcnd the part-x.lnle relution.Berlin: Mouton de Cruyter, 78J-8,10. Bechert,Johannes & Bernini,Giuliano& Buridant. Claude (eds.). 1990.Totard a t1,polog1, of European (EmpiricalApproachesto LanguageTylanguages. pology, 8.) Berlin: Mouton de Gruyter. Beckman, Natanael. 1934. Vtisteuropeisksvntox; Nagru nvbildningur i tnrdiska och undt.u rii.staurttpeiska sprdk. [West European syntax: Some innovatlve constructionsin the Nordic and other West European languages.]G
tr{-K K

XIV. Typologicalcharacterizationof languagelamiliesand li'guistic areas Fehling. Detlev. 1980. "The origins of European syntax". Folia LinguisticaHktorica, l: 353-387. Feuillet. Jack (ed.). 1998.Actanceet valencedans les longues d'Europe. (Empirical Approaches to Language Typology-EUROTYR 20-2.) Berlin: Mouton de Gruyter. Garvin, Paul. 1949."Standard AverageEuropean and Czech". Studia Linguistica3: 65-85. Gensler, Orin D. 1993.A typological evaluationof CelticlHanito-Senitic syntactic parallels. Ph. 6. dissertation,University of California at Berkeley. Gilligan, Gary M. 1987.A cross-linguisticapproach to the pro-drop parilneter. Ph. D. dissertation,

usc.

Haarmann, Harald. 1976.Aspekteder Arealtypologie: Die Problenntik der europciischen Sprachbtinde. Tiibingen:Narr. Haspelmath,Martin. 1990."The grammaticization of passivemorphology". Studiesin Language,l4.l: 25-7t. Haspelmath, Martin. 1993. "More on the tvoology of inchoative/causative verb alternationr;. In, Comrie, Bernard & Polinsky,Maria (eds.).Ccl.rativesand transitivy'.Amsterdam: Benjamins,87t20. Haspelmath. Martin. 1997. IndeJinitepronouns. Oxford: Oxford University Press. Haspelmath,Martin. 1998."How young is Standard Average European?"Language Sciences20.. 27t-81 . Haspelmath,Martin. 1999."External possession in an European areal perspective".ln: Payne,Doris L. & Barshi, Immanuel (eds.)External possession. (Typological Studies in Language, 39.) Amsterd a m : B e n j a m i n s1. 0 9 - 1 3 5 . Haspelmath,Martin. 2001. "Non-canonicalmarking ofcore argumentsin Europeanlanguages".In: Aikhenvald. Alexandra & Dixon, R. M. W. (eds.) Non-canonical subjects and objects. (Typological Studiesin Language)Amsterdam:Benjamins. Haspelmath,Martin. To appear."Coordination". In: Shopen. Timothy (ed,.)Languagetypologv and linguistic descriptiott. 2nd ed. Cambridge: Cambridge University Press. Haspelnrath, Martin & Buchholz, Oda. 1998. "Equative and similativeconstructionsin the languages of Europe". In: van der Auwera (ed.), 2 7 7- 3 3 4 . Heine,Bernd. 1994."Areal influenceon srammaticalization".In: Piitz. Martin (ed.) Lonluageco,ttact and languageconJlict.Amsterdam: Benjamins, 5 5- 6 8 .

X

lleine, Bernd. 1997. Possession: Cognitivesources, /brces,and grannaticuli:alror. Cambridge:Cambridge University Press. Jakobson.Roman. 1931.*K xarakteristike evrazijskogo jazykovogo sojuza". [Characterizingthe Eurasian linguistic area.] Reprintedin: Jakobson,

107.The Europeanlinguistic area: StandardAverageEuropean Roman. l9'll. Selectedwritirtgs,vol. 1. The Hague: Mouton, 144-201. Kahrel,Peter.1996.Aspectsof negation.Ph. D. dissertation:University of Amsterdam. K
"lntenKonig, Ekkehard & Siemund,Peter. 1999. perspective." In: a typological sihersand reflexives: Frajzyngier, Zygmunt & Curl, Traci S. (eds.) Reflexives: Forms and funclions (Typological studies in language,40.). Amsterdam:Benjamins,4l-74. A Kortmann, Bernd. 1997.Adverbialsubordination: t.vpology and history of adverbial subordinators hasedon European languages.Berlin: Mouton de Gruyter. Gilbert. l990."Caract6ristiqueactancielles Laz: de l''curop6en moyen type'." ln: Bechert et al. {eds.),241-53. Lazard,Gilbert. 1998.,,D6finitiondesactantsdans les langueseurop6ennes".In: Feuillet (ed.), llt46. Lehmann, Christian. 1984.Der Relativsat:. Tibin' ;en: Narr. Lehmann, Winfred P. 1974. Proto-Indo-Europeatr ;yntax. Austin: University of Texas Press. Lewy, Ernst. 1942. Der Bau der europtiisclten \prachen. (Proceedingsof the Royal Irish Acad'my, Vol.48). Dublin: Hodges& Figgis. vlasica, Colin P. 1976. Defining a linguistic area. )hicago: University of Chicago Press. ',ledjalkov,Igor' V. 1998. "Converbs in the lan:uagesof Europe". In: van der Auwera (ed.), 2t-455. .lichols, Johanna. 1992. Linguistic ditersitlt in oace and time. Chicago: The University of Chiago Press. lic ,, Johanna& Peterson,David A. & Barnes, "Causativizingand decauonathan (to appear). .rtivizing languages". Ms., University of Calirrnia, Berkeley, Jikolaeva,lrina. 1999. Ostyak. Munich: Lincom .uropa. "Tipologia e genesidellocentini,Alberto. 1996. articolo nelle lingue europee".Archivio Glottoloico Italiano 8l.l: 3-44. isani, Vittore. 1969."Tipi d'accentonelle lingue :ll'Europa". ln'. Actesdu Xe CongresInternational 's Linguistes,Bucuresti,vol. Il,57-60. Bucuresti. "Das nicht-indogermarkorny, Julius. 1927-30. sche Substrat im Irischen". Zeitschrift filr celti' hePhilologie16:95- | 44,231- 266, 363- 394: | 7: 3-388;l8:233-248.

I 509

"Area tnRamat.Paolo& Bernini,Giuliano.1990. Westeru Europe: in drift fluenceversustypological the case of negation". ln: Bechertet al. (eds.). 25-46. "From anaphoricpronoun Siewierska.Anna. 1999. to grammatical agreententmarker: Why objects 33.1*2:225-51. don't nrakeit". R)lia Litrguisrit'u Stassen.Leon. 1985. Contparisorttuttl universal grcrntnwr.Oxford: Blackwell. 'rrNo-languages and wtutStassen,Leon. 2000. 4.1: l-54. T1'polttgt' Lingui.stic languages". Stolz, Thomas. t991. Sprachbundint Baltikurt? Estnischund Lettisch intZentrtnt eincrsprat'hlit'lrctt Konlergen:landschali.Bochum: Brockmeyer. "Someinstrumentsare really Stolz,Thomas.1996. good companions some are not: On syncretism and the typology of instrumentals and conitatives". Theoretit'alLinguistics23.1-2: ll3*200. "No Sprachbund beyond Stolz, Thomas. 2001a. this line! On the age-olddiscussionon how to deund Unircr' fine a linguisticarea". Spracht.tpolog,ie salienJbrschung54. "Ordinalia- Linguistisches Stolz,Thomas.2001b. auf die Beziehung Typologenblick Neuland: Ein zwischenKardinalia und Ordinalia und die Sonderstellung von EINSund snsrrn". In: Igla. Birgit & St
r39-152. "On the areal distribution of Thieroff. Rolf. 2000. categoriesin Europe". In: Dahl (ed.). tense-aspect 265- 305. "Somegeneral characteristics Ultan, Russell.1978. of interrogative syslems". In: Greenberg,Joseph H. (ed.) UniversalsoJ humanlanguage,ro/. l. Stanford: StanfordUniversityPress.2l l-248. "Sprachkontaktund GlotUreland.P. Sture.1985. togenesein Europa". In: P. Sture Ureland (ed.) Entstehungvon Sprachenund Viilkern: Glotto- und etlnogenetisclteAspekteeuroptiischcrSprachen.Tibingen:Niemeyer,7-43. "Conclusion".In: van der Auwera,Johan. 1998a. v a n d e r A u w e r a( e d . ) ,8 1 3 - 3 6 . "Phasaladverbials van der Auwera.Johan.1998b. in the languagesof Europe". In: van der Auwerzr (ed.)2 , 5- 145. van der Auwera, Johan (ed.). 1998.Adverbialconstructionsin the lttnguagesof Europe. (Empirical Approaches to Language Typology-EUROTYP. 20-3.) Berlin:Mouton de Gruyter. van der Hulst, Harry & I{endriks.Bernadet& r'an "A survey of word prode Weijer. Jeroen. 1999.

1510


In: van der sodicsystemsof Europeanlanguages". in the lanHulst. Harry (ed.) ll/ordprosodics1'stenrs guagesof Europe.(Empirical Approachesto LanguageTypology-EUROTYP.20-4.) Berlin:Mouton de Gruyter,425-475. Vennemann,Theo. 1994."Linguistic reconstruction in the context of European prehistory". Ti'ansactionsof the PhilologicalSociety92:213-282. Wagner,Heinrich. 1959.Dns Verbunritr den SpraInseln:ein Beiftag:ur geographichender briti.schen schenTypologiedes Verbunrs.Ttibingen: Niemeyer. "The relationof habitWhorf, BenjaminLee.1941. ual thoughtand behaviorto language".In: Spier. Leslie(ed.) Language,t'ulture,and personality: Es-

sa,-s it1tnennrv oJ Edx'ard S-apir.Menasha, Wis.: Sapir Memorial PublicationFund. 75-93. [Reprintedin Whorf (1956),134- 159.1 Whorf, Benjamin Lee. 1956. Language,thought, ond realitl': Selected +tritirtgs of Beniatnin Lee ll4nrl'. Edited by John B. Carroll, Cambridge/MA: MIT Press. Wintschalek.Walter. 1993.Die Areallinguistikanr innrungenin llblgaBeispiel$,ntaktischer Abereinst Kann-Areal. (Studia Uralica, 7.) Wiesbaden:Harrassowitz. Martin Haspelmath, MPI Leipzig ( Deutschland)

llil l

flj i

108.Aire linguistiquebalkanique

t t.

r t.

t

i

t:,ll

[,iliil;] I sl;iri;

$iliiiir

l. 2. 3. 4. 5. 6. T. 8.

G6n6ralites Phonologie Systdmeverbal Systdmenominal Autresunites Relationsphrastiques Subordination R6l-6rences

1.

G6neralites

La linguistiquebalkaniqueest une discipline relativementr6cente,bien que la ddcouverte de traits communsentre les languesbalkaniques remonte ii la premidremoiti6 du XIX" sidcle.Les specialistes(Asenova 19'19:5-45; Schaller1975:37-45) s'accordentd diviser I'histoire de la disciplineen trois periodes: une pdriode prtlinmnire, ou I'on cherche ir expliquer les traits communs par I'influence oi la lindu substrat, une p6riode classicpte guistique balkanique acquiert ses lettres de noblessegrdce 2r la publication en 1930 de Linguistique balkanique. Problimes et rtsultats de Sandfeld, qui repr6sentela premiere synthdsecompldte, et une periode modente, marqueepar le polycentrismeet I'internationalisationdes recherches(nombreusesrevues specifiqueset organisationde congres). La linguistiquebalkaniquene cottsistepas ir juxtaposerdes descriptionsde languesdiversesdont le seul lien serait la contiguit6 g6ographique:il faut que ces langues forment une < union linguistique> (Sprachbundl. MCmesi certainesvoix s'6ldventencorepour nier la r6alit6de I'union balkanique(Andriotis & Kourmoulis 1968),la plupart des linguistessont convaincusde son existence. En

effet, les traits communssont trop nombreux pour qu'ils soient le fruit du hasard. ll est vrai que les sp6cialistesdiscutent encore de la notion de < balkanisme), que I'on dehnira ici comme un trait typologique propre ir au moins trois languesde I'union. Ce trait n'a pas besoin d'€tre unique en son genre(ainsi, I'article d6fini postposeexiste dans les languesscandinaves, le < redoublement> de I'objet se retrouve dans les languesromanes);il doit 6tre le r6sultat d'une convergencequi aboutit I un r6sultatidentiqueou quasi identique, alors qu'il n'existait pas ir des stades plus anciens. Les tAches de la linguistique balkanique sont consign6esdans l'histoire de la discipline. Ellesont un triple aspect:synchronique (description) panchronique (extension) et diachronique (formation et 6volution). Bien que I'essentieldu travail descriptif semble avoir et6 acompli (la monographiede Sandfeld a 6te compl6tee,souvent am6lioree,par des centainesd'articles et d'6tudesde d6tail qui ont permis d'accroitre et d'approfondir les donn6es), il reste toujours beaucoup ir faire. L 6tude de I'extensiondes balkanismes n6cessitele recoursir la geographielinguistique (ou linguistique areale)pour determiner avecexactitudele lieu d'apparition de chaque balkanismeet son extensionr6ellesur le terrain. Enfin, la perspectivediachroniquen'est jamais perdue de vue par les balkanologues, malgr6 les nombreusesdifficult6s auxquelles ils sont confront6s,faute de documents6crits. Trois aspectssont ir prendre en consid6ration: l) La gendsede I'union linguistique balkanique;2) La genesedes balkanismes;3)

The European linguistic area: Standard Average ... - uni-freiburg

Recommend Documents