You are on page 1of 24

Linsua Sg 1:1~2) ~ 7 ~.

~(~ 2~
~ o r t h - l l o | b n ~ Puhlis~fir~g Cump~ny

INDEFINITE AGENT~ PASSIVE AND IMPERSONAL PASSIVE:


A f'UNC'TIONAI, ~ T U D ¥

Zy'gm ~ F~AJZYNG[]=R ~'


L:rri~z'rsit;ui ('l;h~r~d~

R¢,;¢i'¢¢d D ~ w b c r 191;[

This lial-~r p~[Ite~e3 ~, Ibrrnal n~tL l~aclional difl'~r~:niia[iitn hell~.'~,en imi~-:rs~nal


pa~i~et and ¢~:~r kin,:J~; or pa.-~ive. In p{irdeu/ar, ii will shrew thai wl'k'r~cr there

i~e ~entu,ac~ ffn~ an ~,~[~fi~it~ human a~cnt. This is not a funuLkm ill IhL, pa.~si,,'c
~'olrm of ~r~nsitLvLr~'L':E~. which nol only do nol imply any scmelnlic ~'L:Alllr:~or the
~uhje,:t. bul vcxy s.]f'[Cil09 nol imply cx[[lCl]~:c oF" th',." ~ubj¢ct L[|to~=lllcr,
The i~lp~r .-ltan pro[~t'~ a~ ~phmmi~n for xhe facu TllaL in marLY Itlngu~t~e,~.
c~lLalJ:~ ClClll¢.q[~-elthe pa.%i~: ~0t-ln wc~-u,itiliz.~d 1o i~:,JiuaL,:m; :~lld¢finlt.CIllm.iLtliagc~.

i. Imr~Im:~ion

LI. The data

[11 CLlrl~tlt i~.a~¢, l h c to_rill irnpel-~ollal p a s s i v e ~.l~tnr]s f o r passive forms


0f'verbs in $ca~.cnce~ for which on~ cannot postulalc aa u ~ d c r l y i ~ sobj¢,Lq.
The.~ sentences, like an? other ira~rsonal :~entences, may er may no: h a w
a surface struclure subjqcl, usually 3 p~ sg. p r o n o u n The present papeT will

'.rh.~ wiJTk ml ibis p p e r was partially SUpl~rled by a Gram from the Cotinc~l on Research
and Cr~r~tr~¢ I~'~l'k, UnLvcrsily c)f Cotori~h3. [ ~,o~ lil~ Io Ihsnk David P~rlmur.zer tbr
c'rg~ltlJ~lg~ll~ II1~ tO tmdcrlake a d~rip~Lon cf th= ~unt~,Rc~ ~ndin8 ia -m~ and -re in Polish.
[ ',~uld a l ~ ffk~ la thank Giu[i~ Lep,mby~ W.B Lockw~mL nnd Scot! Delan~y, wilh ~hom
l have di~a~s~d languaEe dala anck lheorefica] i~sucs inwil~ex:l in the .~per. The f o l ] ~ i n g
p~pIw have Ilell:~l a ~ wJ~ LIL¢l a n $ ~ 8 ¢ da~a: Mc;hura~nad AlL Joseph de I-[~, Ham~a
Kh~3h~i[aty, ~ e ~ a t KiLruc~ghL Ahdelelah Kulbl. Gr~ziana Lnzzarinu, Maria Richmond, Hugh
Sr.hrnidt.and ]'i~mas tlol[w~J~, Noa¢ oflh¢ ~o~,'c is in any w,~y respoasii~l~/ur th~ ~n~l~nl
of this paper~

0024-3841/$2/0090-A)(}00/$~2.7~ ,~ 1982 North-Hc, ll~nd


~:g Z. Fro.t:,r~teit, r ' Pa.~'sire urn/huln.,r.somd tm.,.ri~'v

d~:zl ~,'Jlh "~wo t ~ p ~ o f constrllctJ~lL,~ : the ]las,~iv~ f o r m o f intransitive w r b s


~nd passive forms o f transitr, rc verbs in ]arq~ua~,es ,n which the patient or
direct object i~ markzd in the s~,me wa? in which it is markcd in thc active
f,CtKCr[C¢,
The term passh'¢ fbrm designates the form of a verb or sentence which is
a~d in constructions whose primao, fLmction is to indicate |hat the subject
o f a tuansitivc ~rlLl~rh [S pLPt~'~J.mlr~L~r ~:~O'[nt.~otlte: argLn/nent rather Ihan Ihe
cxp,:¢ted agora.' There arc essentially two types of ¢onstrnctions which
wiLl he considered here: constructions i,1 which tim v~'rtxd form is ideqtica]
with t ~ vcrl~l Ibrms ~ : d in tl~¢ passive constrm:uons or constructions
witll the so<ailed reflexive pronouns. (For a description of the diLiTe~entiation
between passives marked by reflexive pronouns and other types of passives,
see I-"rajzyngier 1978.) The passive of intransitive." verbs in st.weral languages
was recently a subject for two papers (Comrie t977; Pertmtttter I978~.
Comrie~s p:~per a l s o d e n i s w i t h t h e Im.';sive f o r m s o f t r a n s i t i v e v e r l ~ in
w h i c h t h e p a t i e n t is m a r k e d in t h e s a m e w a y a s in t h e a c t i v e s e n t e n c e .
S o far. C o m r i e ' s a r i d P e r l m t t t t e d s p a p e r s a r e t h e m o s t i m p o r t a n t a t l e m p t ~
to provide an explanatiott fo~' imlx:rsonal Imssives. The data disettssed in
the two paal~er~ will cun~tituIc the bulk of the materia~ discussed in lhe
present ~ p e r . but data from ottter languages wilt abo be included in
c o n n e c t i ~ r t ~¢ith p o ~ i h ~ e u n i v e r s a l p r t ~ p e r l i e s o f i m p e r s o n a l p a m i v e s .
The above worktt~g dc~inir.ion ol' ~d~sive ~qui~,~ .,t I'ev~eomrr~m~. First, it is not tmiversal
i~l IIIm~¢t,~t t~l" ¢~tph~|uing all I-trots ~bai v,m~u~ tinsuists m diiti~relil, la,~gmsSt~ clinton to
call passive Thu~, the so.called advcrs¢~li~ lmSSi~'ein Japanese certainly does not Ihl[ t.nder
zhc prol~3st:d delimtinn. AIIhtmgh ann conld u~tpkfin why ~t3rn~ I:pguist~ ma:¢ call the' J a p m , ~
advcr~ati~ form "pa~i','e', such an explanation is beyond the scope of Ihi~ paper But lh¢
Japar~¢~© advcrsallv¢ p~i~'~ also i~ bcyon~ ~h¢ scope ol~ this paper For ~t~ time,Kin has
na-'hing ~ct ,'lo with the function of pa~gi~ et3nqructiong as lhe.y "tin known in Indo-Europetm
[or Semili¢l laneu~gcs, whwh g,¢¢ rise to the ~¢rm "p~ss}~¢' to bel~in with, A research
methoda~c,[:)~ sammaraes applied in the ,;catch af tml~tfislic universals in g'hich all the instances
of the ~sil$¢ e f a lincgir4i¢ ~,¢rm gee ~¢l'0dzized, w~lh the aim o£ lintting some univerml
conm~nin~ the language, ig rl.llt a reliable tool Such methotiology df~eg not uncover a
unieersuL d language bat, rather, the a'ea~ons '~hat ]in~luists might ha~e had in classifying a
li~mi Ibtm ~,itliiza a gr~ttizmatimd framework thiamin.! m the slud.y ol s~rae other langtZa¢Cs.
S t t e ' h II'~[h~,doloj~}, r may M ~p,prapriate tfl ahe histor~ oF science but is useless in the s~arch
for u¢i~.[r~lr. ¢.f lana~g¢,
The ~e~.'nt dd~nhion aims to eaptar~ the primary, function aF pasties in [angtr.~tt~ which
~re ~¢1~ s~;IJcd and which set~,ed os models far liaguims in category assigamen1_s of hitherto
rtan-dascrih:d langttages~ Note thai. althoul~h the pa~t~: [brms in lanpaltes d i ~ a t t ~ l ~ilL
al~,~ have such £um:dor~s as topicalization, f~eusirlg, etc,, they atwa¥~ will hate the function
~ i ~ t~ I~lCln ~n t ~ ~ W t r Fur u f¢¢~¢w ~ ¢-vJden~ ¢oh~nting ~ l i flltl¢i|OII of i~Mive,
sc.z Ftaj~:mgier 197g and Ftajz~n~gr and K~ttrie] MS,
t.2. The hrpc~lhe~er

1,2~L d~'e 1~a.l',~'/le.i c:/' tr~ulx{tiv¢' r~,Jb,~ ~i~ld impelw¢mcll pa,~-,#ves 111¢ sat~w

This is the question to ~hich holh Perlmutter and Comric give .'in
allirmal['¢¢ answer {el. Perlmvtter 1078: 167). The pro01em ~ith ~his qaes-
lion is in determining what Conslilute.~ "1he same ph~nt~raenon'. Aprrarcntly,
tbr hol]'l aulhors, it is enough LO show tha~ there arc characteristiea thai
the t,,va cOl'islrtIe[iunS hLl'Ve in common la Ihi~ case, the~ characteristics
are the derivationa[ historie~ of r ~ v e s ol'ir~nsitive verbs (herealh_w referred
to as simply 'ice.ire'),
In lhe preserit paper it will be shown the, the only lhing that passive
arid impersonal p~lssives ha','e in cemmon is the f u r l of the pnssive marker,
he il a ferm blithe verb or tile, refleaive pronoun, It will he shown, however,
Ihal the Iwo eoi]structiorl,~ diff'e.: iiz bath their fuJlCtioas and syntactic
preperties. Thus. an:dygin$ them as 'the aame phenomenon' lor the sake
of one or anvther syntactic theo~ ob~ures the facL thet languages LINgthe
small cliffcrences in form te express differ:n| mc~nings

t.2.2, Conrie's hypodres'Zs"


Comdu's hypothesi~ is that I'mth imperson.',l passives and other types of
pa~,iSil,re$ are derived lhrQugh a demotion of subject. In respecl to impersonal
p ~ s s i ~ Comde says'. "'In the impersonal passives, how~'¢er, where the
underlying aubje¢l t~rns up as an oblique object {X-rated), lhcre does
appellr Io he spol~laneolJs dcmolion of a subj~:t, nol caused hy prt~moticm
of some other nour~ phrase to s~bject'" [1977; 48), He goe~ on to ~how
float sine=, in sotne Nngu.-igc.s, lhe Olyl¢~-'t preserves its position (heeler it~
marking) Ihat il bud in Ih¢ active senlence, fllere could not be promotion of
the object and. therefore, the lack of surface structure ~ubject musl i~
treated u~ speutaneous (~emotiorL In tile preaem puper it will be shown
that the basic assumptian of Comrie's hylmthe~is, ~,iz. the assumption that
the subje¢l hag I~'ev deleted from impersonld sentences, is false for Polish.
Spanish, Ladl;, id, possibly, Welsh; and that data from Dutch and Germ-'m
do nol suppor~ lhis hypolhesis in ~rl uJ]ambiguous way. "/'lie Ihlsity of
Co~rie's h),'l'~hmis will be shr~wn mainly by postulating and providing
evidence for Ihe pestulule lllal the impersonal sentences he quoles do indeed
have a subj~l, which is oVerlly indicated by d~e structme of the whole
~cntcnee,
"+711 ~.. Fr,~jz.,ngier P~.,.~.s/rvand +)nl.~.es.uu/l:..+,+]r'e

Pcrlmu;tcr 1978 postul-zi:s thr~:c Lyp,:s o f strata:


a. A tran_~li,+e stratum which tiroL+fins a I-arc and a 2-arc.
h, An tmaeeusntiv¢ ~trz+ttlrn wt~jch conlzins z 2-are but no 1-are.
c. An uncrgative stfalUnl wllJch cofJt~it+Js a I+-af¢ hut no 2-arc
He furtl~er slates tha~. initially unerg:ltive clauses ¢orre+pond io the tradi-
ti~nal active intransitive clauses ~lnd Ihey contain verbs indic~ding volitimml
~ct~. such as ~play', "work', "~l~a~. ", etc. They also contain verbs indicatin~
in~,oluntatS ht~dily fL,ICtiott+~. +such a~ "sneeze'. 'hiccottgh', etc+
Unaccusnlive cla~se~ correspond lo the traditional predicates expressed
by adjec+.ives in English, prcd+.caics who)so initial nucle:+r term is semantically
Lk patienl, e g "burn'. "fall'. *drop'. "~ink'. "trip'. predicates of exisLing and
haplmnin ~, and dnrativt~s, e.g. "rL,main'. "slay', 'survi~.e', e l c Perlmutler
sit+los tirol c l a u ~ whozc initial stratum is unaccusrqJ,,c +.-armor have an
impersonal passive. He proposes Ihat, it+stead of a st~mtataeous demotion
m =mpersonal pa~';ives, p.'e have ~ case for promotion of d u m m y 2 to the
~ubject relationship L. To quote PcrLmutter: 'According to the proposal
ir PL~Ial and Perlm~Jiter imp.'rsonal passives univers~dly involve a d~mmy
which advances frt+m 2 to I'" (19714: 158)+ This derivation applies only |o
the initially ur~cr,ga~ive clauses, it clots not apply, according to Perlmutter,
to inilially .naecttsacive clauses, Tl~e predictive power of Perlmutter's hypo+
thesis is almost nil for. ~s hc stales, the s~mc verb,s may have differcnl
pror, erlies in different lanl+uage~. Thu~ one verb may be initially tnia,.~eu-
sali.'~q in one language 'out unet~zliv¢ in another,
T o shrew that Perlmt, tler is w~ong (or right) is u bit more difficult, for
how can one 8t~o,,x' that something happens or does not h++ppen to a dummy,
i.c, ;in enti.ty ~,home existence has not been shown and is, in fact. very much
in doubt',' in the same way in which various entities at' this nature were
postulated in ~raditioqal tr++nsformational grammar, u dtmnmy is rt~uired
by cerUtm theoretical assumptions about the underlying form of a scn~cn¢¢+
Neverlhck~ss. there arc ways to ",crify Perlmutter's hypolhes~. In particular,
if Lt can he shown (as it will be} that there are impersonal pas~i.,.,es in which
2 did not advance to I, then impersonal pamiives '_ire not derived lhrough
the 2 t-o I ,ds~an¢cmcmt. it will aLso be shown :that, within a relational
framework, I in the impersonal senlenc~s is an indefinite human subject
rather thun a d u m m y - regardless of what. il any, pronouns occur in the
s~bjecl posiLinn. Thins, it will be shown lhal Perlmnlter~s alternal[s*e In
~pqnt~rteou$ demotion is alst~ wrong.
Instead of two hypotheses which do not lake in:to consid~ralion the
Z, l'i,~4.;zr:~A,~(v ; P~n'silx, eatal im17er,~,,nr¢l ,e~e~',~'ire 27j

funeti~3rt hi' the ]ir~gt~istic ror/tts, I ~ m l d like to l~mpt~e a ft.ntcti(m;il des-


cription of imper$onM pas.~ive,~. By providing cvidcm.'e Ibr tbi~ description.
i will also pm¢ide Ihe nc~ssary evidence agains! both Comrie's und F'erl-
mutter's hypothes~$. The expl|tn,'Lliorl prt~eided will make it possibh: to
predict the fmnctloq of lhe passive ~brm of an inhan~ilive rerh and also,
tx~,'ause of pra~ma:ic considerations, it wilI make it po~,~ihle to predit'l
when a p~ssive form or 0n inlransifive verb t~ possible rand when it is hill,

1.2 4. The"107;a/twW.~ re &, &,/~,mted m ~he pves'em~ paper


1 would like to propo,~e th:~t the primary function of the passive [brm
c~f if:transitive verbs is to iJtdiczt~e t h a i u sentence has un unspecifiqd hum~l~l
agen~ which is also subject of" ~he sentence~ So rat, thi~ claim ha~ n('~t
been made with r~pc'~t to the gcner~tl properlie~ of the form, although
it has beerl made with respect to particular languag~.~, such a.s Spani,~h.
!, t]ni~ rnnction, ahe passive ur hltransilhce ",erbs differs radically from
*he passiw: of transitive verbs whicl~ do not indicate any kind of ~ubjcct,
atnimate or inaj~ima=¢, hlllflarl or r/on-tla, ll~lall. II also d[ffi_'rs fm.n in'~perso~sa]
vcr'os which do nut have ;.lily semantic stlb~.=[, .such a,~ weather iml~'rso=als.
Although the intransitive p~ssi~e is a device to indicutc the indel~ile
ht~man subject, it is by no me,~n~ the only device to perform thi,~ l'~nclion.
The other d e v i l s , howm.,~r, :~re o~tside the scope of this papc,-, ~ Why the
idemity of subject is not indicated can be tmswcred ~,l~rough the .sludles
of pragmatic asp~ct.~ t~r ]a~guage Oa,'hJeh a~so lie out,~ide the .~q-~pe of tl'tis
paperl,. ! will not address the ~mpllcadons that the analysts has for [h~
theury of relational grammur but will leu~'c this uspect to linguisl~ working
within this ~heory.
Since the ertnciid arguments in favor of the hypothesis pre~'v~ted in the
present paper consist ia the irterprctation of the data, the next section
will consist of the preseidation lind interpretation of data from Polish,

cle~ic~ ace: ~ ~f 3 p pl o~ ~ p~ .~. ~ o u n s (~g, Ruffian, EngRsh); ~J~ ~f ~ecil~e


:~d~linir¢ Izu,r~,n ~ub.lecl p r o n ~ n (~.g.-. "en¢' in Kn#i.~hL ~ I¢~rnu Which E~ ~qq,i'~ale~nl Io
English ~man" leg., Fren~. ~l'lT~.[rll, 13,n,leh). In .~ever~l rancgUages there is an ~.tdi~innal
dilg+m,Ulatkm m,++ del~nding or, ~,.h,lhcr or not th+ I p. is incl,+dcLI in the m ~ f i n i t e
s~bjeet c~ n~l. Such larrgur~e~ ar~ Prlish an~ langli.~h Mnflt'Ten~ae hvatw~en ' ~ " and 'lh-S).
Thtl~, i~ P~,ti~F't. the ~iffcren~ beckon II~¢ Fart..~ endi~lg i~ -m, and .to e,n 1he orte hand
uml i-#rms wil~ the dilic ~'ie is I.iml. ~kl~' lult~zr, buz. z~.e the i~.lrmgr, alkr~-~ los the p~.s',.'~t.bjlily
t'~al th¢~ first I~'rscm is ['nducled Jn the inde~L'file sttbj~ag.
27? Z. F~aj:reegn,e .; P~.ssJ='e uml imp,,e,~mal pa,~.~'irz,

Latin, Spanish, Dald~, German, a~d TmkJsh (laitguages discussed in Corer;,


1977 and PerImutter 1978). as well as Italiatt ~ud Arabic (which were not
di~eus~c-J i~: the above papers).

2. Analysis of the data

2 [, PolLth -~

There zrc two reasons ['or which the impersonals in Polish are considered
in the.'present pnper despite the fact that they do not reflect the contempt~rary
passive form of the x,erb in Polish. The lirst reason is that artaiysi.s oi" da~.u
from Polish cons|ilutes part of Cotnrie's evidence for his hypothesis. Tlne
other reason is to show how the passiVE form Evolved in ~he history of
lan~gaage l o hecodqe a a aelive f o r t ~ indicating o n l y indefinite h u m a n agelll,
TI~c following sentences in Polish are ~liven by Comrie (1977) as evidence
lbr the spontaneous demotion rule, i.e., the rule of demoting the sabject:

Diskoauje si9 pram ("pr~z uczanTch )


'The *~orks are being completed ~by Lhe scierttL~tsf
IO,3konano price r-pr~cz u¢~onych}
"The ~'orks have tc=n u~mpl~tg~l Cl":'~/~h¢ ~a~.~hsls)'
Zapukano d~ drT_w~ ("prz_:~ .%,~iadaj
"There was a ~,~uoek au the dome (by a neighbm'f
tdzi~ s¢ s~ybk~ {*pr~. u~z]]i6w)
"OnL? walk~ quickly (h~, ~cim,~]hny~)"

in synchronic descriplions o~" Polish and in historical studies, the above


forms were never described a~ pa~;ive (cf, K]emensi~,icz 1961; Kcmeczrta
[95~: Br~jer~ki ]979), In f~tet, in ~ll the studzes, including the compuralivc
works such as Siatk~wski (1963), Ma~usowicz (1969), and olhers, the forms
ea~lirtg in -n0, -to, or the forms with reflexive s.~¢ are destribed as indicating
an indefinite hmnan subject.
The best e,,iden¢~ for the funeiiov of the forms ending in - n o nnd -to is
the fact that they do nn~ admit any other agent through a prepositional

In the pre6en~hon of the dala. senlenoes zakan iron Comrie, Per]multer, or some otlz~
pu~Ihh=~ som¢¢s ar~ quoted ~'~azim, withou~ eorpec~ions o r word.for, wo£d rra~sl~zion~
New ~ala are i~rev~deg[ v.'it.h Ihe ~rd-for.werd tran~miou whenever this ¢ann~4 be dedug~ed
from ~he ~ranslatiGn of the whol~ s~k*aee.
phrp+e a.q +howrt in the eg+mplcs above, All other passi'v¢ sentences in
Polish, with Ihe e+~.ception d" the non-slalive formed wilh sit. ztdmit an
agonlivc phrase; for example:

~tbt z~lafl polam++rty prze+" wod~Cgo


i;l~.lv l~.~.'ame ~'~rokt.m I~7 jrtni{or
"The ~l~Ie ~w'~s broken by +~j+tnitof

Other evideJ,+e is the rant zh+Jt ti}e leorm.~ ending it+ -no zmd -~+~ cannol
he u~ed with verbs dcno+.irt~ nomht=rmm ~,:rivities. TbJas, if one were to
say in Polish

Zaszczek+'mo de dr--wi
bark ;tl duP~r
'it h~.,i been bllrk+:d ++]~tim d~,~r"

the implication is ttmt there was somebody barking ;at the door, rather than
"There is a d~g harking ai the door'. Similarly. one c~ln] ~ay

K crr:~iki przcgrl+zly sl,Sj


,wi~odwtirnim bite ihmullh It=hi-
T h e w~odworrns }talc gone throug~ ~h¢ table'
S~bt ~si~|~ l~rzegryr.ieny
p r ~ karnikJ
bi tt~zl I h~otJgh
'The tame Ilr..+~be~'n yuined by the +v~mt~dv.'orm~'"

S t ~ zoJia~ p i t . t y r i a n 3 ,
'The table h-'~ been ruiRed."

if, however, one says Pr-egr.;'.-iono •z•L the implication is that ~omeh0dy.
a htJmal~, ha~; nliJlM ~he t=abPe by bitirlg through it.
One more piece of ¢+idene¢ that the forms ending in -no and -to are not
passive in Modern Polish is Ihe differeluee in their abilily !o hm'e t-ert~in
adverbs or manner added. Thus, in the passive sentences above, ~ne ¢..aflnol
udd an ad,+erb hldical~eg the altitude of the agent ioward the ucti(jn, e.g, :

t~fii .~=+:lnipchl~tnte pt~-gr~,zitJny


uJ~willir~j+ly
'Tim: table i:+ ruirtt~ ul=,~,dIit,gty~

The addition o£ such an adverb, however, is perfeetty possible with -no


sentmnce.q ~ontaining an indefinite subject, e.g+ :
274 Z, Fru/,zytl¢it'r , Pa~.~~t" w~;I b,mt'ra'~ul y.,ra-¢v¢

Z ran.'l hi,game ,aieeh@nie


in mor~;ir~gru~
In the mor~iag one w~uld run unwillingl)'"

T.~iz forms ending with -trf~ and -to cannot occur in sentences implying
a sl.atlvc meaning. Tt~ put it in other words, use of the forms ending in
-no and -~o implies a non-~tative meaning, a w,lled action, even when used
v,,ith inherentL), stfitivc verbs. T h a i the senlencc Wisimto na ilptach does 11o/
i~ply a sinister "One ~as hanged on the ropes + bat. rat~er, "One wnu]d
haog on lhq ropes" ~ in a sports demon~;t[adon or compctil~Qn.
C~mrie's claim that these sentences in Polish do not have subjects is
tn,v only in |hut, indeed, Ihere i~ nt~ avert nominal ow pronominal subject
present. It is, however, false tar the semantic struclure: there is a subject
present and the sub)ect is human. The fur~ction of these syntactic cvnstruc-
tions, theref,~:e, is to indicate tt, at the )ndefinite human agent is overtly
marked hy syr~r-'telie struclz~re.
la the passive constructions, on the other han,d, lr the su~;ect is not
overtly marked it canno~ be predioed. Thus. in the sentence

house bcqam~ d~lro3,ed


"l-Ira house ~.,~ dt~t~oyed"

the possible cause of destruction may have been human, or an ~nimate or


non-animate cat,se. However,~ Ihe senlence

Zburz0no dorn

me,ms only th~tt some humall.~ destroyed the house.


The forms ending in -#o ~nd -to represent old passive form~. Ample
evidence for this is pro','id~d by wrilte• documents. These forms also w w e
the subject of many historical and comparative studies. There are many
•:Jteresling I:noblems related to these forms, but in connection with the
present paper, one q,aesdan is important: were the impersonal passives
both transitive and iiltrittlsltlvc in old Polish document~? Notigc that in
the examples (in Comrie 19771 qnotect above, there is a fair number of
impersonal passives with direct objects following them. These examples
arc important for Comrie's hypothesis about passive b.~ing a subject demotion
rule. Brajerski 119791 argues eonvmcingty thai Ihe emergenoe of direct
objects in impcrso~tal constructions followed the emer~nc¢ of impersonal
Z. Fr~#/Lva,g,rc::'P~r~.¢i~'~'and iatpersm~a! i)a,,t~'fvc 27~

passives, In the oldest tegt~, the impersom|] p~)~ves ~r¢ the pa,,,si~,cs of
intransitive verbs. Historically, lherefoc'e, Ihc primary functicm of impe~'sc,nal
passiv¢~ was to indicate the indefinite human ~ub.,/cct. Laler o n , when thi~
construction c c a . ~ to be percei~.'ed as r)assive, il was possible Io u ~ il
in the same function wi~h transitive verbs.

2.2. ~za'k ~,~h

As Perlmut~er (L97[¢) stares, Turkish. does not admit an agentivc phr:a~¢


in rh~ passice eonstrucLio)~, Tills is aIso true for st~nlences lh~l ale p~t~.~enled
as lmpcrson~+l constructions, such as.

I/tsrada ~ah ~n]nr.'o.vn~tin,ii,a.~nr:hr


~He~' it i~ v.'orked..I~laycd.+hotned"

i.e., passive Forms of inlcaasidvc verbs. In Turkish, h~)wevcr, 1he passive


construction o f an intr~l~lmLive verD d o e s nol admit an ~ g e n t i v e phrase
b~ause ~t indic~.t:s the indefinite human subject, jus~ :ts in the indefinite
subjeel ~enlenees described for Poh~h, Thus, the sez~tcmce

K bpck~en kaolmaz
dol~ fl"Oll] lull ~¢~ Fg~):/iv¢
F r n m the thl~, it i,; II~L rL,l! ~¢.':]y'

is understood To concern haman beings oply. Alg~ compare the following

Ihi~ d~s$ leous¢ fr(nn run


=F'r~m 1his d,5~ kcrmd is is run uw~y"

whidl may indicate I~-Opl¢ (sad= as guards) running from 1he dog kennel,
but certainly not dogs. The above sentence has the .~ame subject as

Bin h = p i ~ h a ~ t l kaCnl,r
prls~, from
'l~ro~ thi~, pri~o~ i~ is mn away ~

An il]ler~slil]g lesull has been o~irled in the inlcrprelatiuIL o1" the


following sentence:
Bara~La kar~ kaf~ma yenir
here ~nl; I~+~'k ~';~t
"ll.'r¢ it i+ ¢+rtcn I~;ick ,+]+t+'

The agenl that eals ~hc black ants is understood to be a haman = not cvcn
an anteater. Similarly,

I~ ur:ld;~ ko'~Ltlur
"Flcre it is r . n '

applies Io hutnar=s olll) and cannot normally be used in r e s l ~ d to animals.


Some other sentences which have to h a w a non-hum'~n, i~c., an animal
agem c~,not have lf~ imper~omd p;~s;,.'e~ Compa;'e the following:

}'af~?ler nrm~ilyu.~'j hi~ir,~itrr


r~ls ft*rn lit.re gnaw
'The tats h.ive ~nawod the I'tlmimre ~

M~Mlya hitirildi
T h ~ (utniture i =, ~r~=~,'ed"

bat

'i+urada ~itirila~r
"I~ i,',i ~lT, LIW,,;:,iJ, h¢l-~"

'BLtrdd~J ~ t ~ n g ~ hm, i~nir


" I I ¢ ~ it t~ bnrkcd ~,ll ni~.ht

2,3. Lati~

]n th~ traditional discussions of' the function o f i~L~ivcs, one o f the most
importanl • if nol lhe most imporl~nl - argmncnts against claiming thai
the function o f tSe pa~iv~ k to indicate that the only argument o f the
sentence is patient rather than ~rt expected agenL ~ a claim that in Latitl
there arc passive forms o f intransitive verbs, such as itur 'one goes'.
Th= onty way Io determine IJ~ function o f thmse forms is to check the
q~i~ting ~xts in Latin and find o m whether any o f d'm passives o f intransitive
v e r b s c:~n c,~er o c c t i r with a n o t ~ d ~ u m a n agent, or t o r e l y ot~ g f a t n t n ~ r i a n s
who wcrc c0,~cious o f the cxistiafi distinctions. In the two standard texts
on Latir~ syJ~tax, we find the folluwing descriptions: Ernout and Thomas
(1972L who make a vet7 clear distinction Ixtwccn impersonal (i.e., a form
without an urtderiyJng subject) an,~ ittdefinite suMs+eL translate itur as 'on
~ . . ~ ? . j : j . g i c r ; /~aX*'h'c --,t #,/a.rs.m:! i~.s.~/n ' ~7;

va'. Ihus by the acuve, mdefinitc tbrra in French (p. 20.% [[ofmana and
Szantyr (19~5: 418) sly.t(: that one of the means fur expressing ;l~e indefi.~ite
agent ("die 'man'-Bedeutung") is the impersonal passix,~ wJ'deh, in laler
L+lin. w.~ replaced h } r Jit.l~ll'.J plus . ~,erb in the active form
[n gapporl of these two m~st aulhoritali~c goure¢~, ! woatd like ~o poi~st
out ~hat a random cheek mto some of the ¢oncard~nees !o the Lafi.
,.~dlers did nol turn up any passiees of i[~lran~ilive verbs .ged ~ilh aon-
htlnlall scrlnHllti¢ s.,bjee~, The c×arnp]es quoted in ".'~lrio~s sour¢c~ have
always been human suhjectg: e.~. iexamples flora Olr?b_~k] and Snfarcccicz
1937. quoled after Brajer~ki IoTP: 941:

Dic~ Iqoc[¢sque ~:sluL bil,ilt~ ~Plautl


"One wo.]d ea~ and drink ,h~y aml aigl~c"
~ic !l-jr ~d asm, I~Vir~il)

"Q~e would 1~161 hruvL'l)"

2.4. .S!m.i.sh

The .~o-¢u]led impersonal seater,~.x's in Spanish quoted by Corarie (1977),


s L g h ~t~.

N~. ~¢ habh~ dg~ ramie. ~'*por ~o¢ cst~diame¢l


*Mtl~iC Js n43( ;p,~kt:n of (by the ,~Hiil']e~ts|"

and

P,I~ d~nde ;e ;ale l~pc~r I ~ viajems}?


"Where ,:lo¢s ,0jle e~,il ~'~by the Iravclle¢~l?"

although ~he~ do no~ h a w un over~ subject, aever;,heles~ ind~e~le Ihul lhere


ix an a.ldcrlying, undefined hamar~ subject (cf. the explicit slatcmcnt to
that eBL-'¢t in 5after 1974; I~),
Again, ;~.s in Po~ish and Tatkish. this properly of imperso,LLl passives
is s~pporLed by the f~¢1 that !hey do aol. admit the agent!re phrase hecaage
th=re already is an agea:L implied by Ih¢ grammatical form of 1h¢ scn(enee.
The Fa¢l that this agent muaLbe hura~n is supported by the ungrammulicality
~1" the folloYving sentences :

-I thi~k i~ is har]L~l in the Ibm~L"


FJensn qll.~ se bahia ~rt o[ t~ctsq~
"l think tltat lhere ix seine ~utkif~- in tl~ forest'

AIs,~ compare tl~e following:

"No ~ Itt~,, este "trio

"N~ sv rebecca ~sle ~f~


"Onq dolt not M'ay ~his ~,~2ar"

The fael l h u i o~*e cam'tot ibrm th~ indefinite age,it sejltences with a verb
i t : d i ~ i n g an aet,v~t~,, restnct~'d to animols does not nleall that one e~r|not
form the passive , ~ L e l l C e ' ; with such a verb. Compare the following senlet~ces:

• ~ n ~e r~, ru Ildadef'~l t?~,'l(' afh,'~


'Olin dot'S, l]~,~ glla~ Ih~ "~0~ tl tJli~ y~-a,'

bLJt

La m.',cltta e~L~ ¢oida


"The w~md is gnuwt'd"

Note ~hat the 'impersonal ~ssivcs" a r e formed only with Ihc n o n - s t m i ~


p~sive form. This i.~ an indication that we are dealing with a ~ p a r a t e
grammatical category which i~ only morpholoSically related to the p~$$ivc.
In the proper passi~'e consttuctiofls in Spanish~ the d]stinetitrn between the
s ~ t i w and non-stative types of passive is maintained. The implic'ation of
this analysis is |hat imperson01 passives life funcLionulty not connected with
the other types of pas~i~.¢ in Spanish. S~e Frajzyngier (1978~ for ~ c distinc-
tion betwee.i slalive 8lid non~lalb.'e passives.

2.J, Italiatr

Although the data from [talian were not considered by ,,ither Comrie
([077) or Perlmuttcr (1978L the 7 writ be considered here, For in [t~lliun
]inguislics there are forms labeled qmperson~' (of. l~.l~chy 1974), but
which, in faG. i n d i ~ t e the indefinite hmnan agent.
I will not r e . a t th~ same type of ~,~amplcs as in Spanish; th~ following
two should suffice:

Semo the ~i p ~ l a
"1 hear gc,me,~n¢ talkio~"
Z Fr~f//'__l',~,~h';" . Pre+_tqrr ,'1~1~/hlq:er~rala[ lJ..ra+f'tv/ "r'fl)

FILtLnot

=~:b,;nt#gilt: .~i .',l~l~,ia


"1 Imar .'~a~co~c t*arkit~g"

AftuaIly. the last sentence o0uId be produced, but the impl]¢uti0n wo=ld
be that hmnu.l~s are harking. Some of my Italian ilfformanls could accept
the sentence Qnf st cvr?'e =11 is run here', in reference to animals, s~y,
produced in a ~roo, but they wotdd explain Ihal it i~ ai~,ays produced i+a
jocular I]ll]r]]|1~F.
As in SDaaJsh. no stafive impermmal passives occur in Italian,

2,6. Amh~c

P'-,ssiv¢ comstructior~ in Arabic does not a]low for the agent to occur in
I~e ~amo se~tteJ]ce~ eg.

Active: k.',tahd ~u~,~r'u kilu~Lm "Yu~IL'u wrt~te a hflok"


Pu~vt:: kt~(it,a itl-klt:Ib~, "lItc ]>oo.k Wil~. V,rrilleil]."

The fact that no agent can occur in the sentenc~ camai],n~ a pa~ive
form of the verb indicat~ that tim passi'~ construction is not derivcct t'rnm
the active construction through a transformation, lr|trdn.~itivc v;rbs may
alsa h~,.,~ a passive I~rm. Ag, in. as in the case a f Irans[tive verbs, there
i.s no translbrmational relationship Eetweet| the active and rite passive
term, The subject uFih¢ ~¢~ive intransitive ~;¢rb may not also be ~he subject
o f the passive int~+~m~itiveverb, e.g.

Adiw:~ .jam yu.~u(u 'Yusu~Ur-J,'

A¢li¢c: h:rabd ~;t~0t-u "i'~uf~ rat~ aw4y. ¢~ap~d"

The pas.sivc fo~m o f aa intransitive verb may. however, accept a locative


~_~rnplemene. As prvdicled by the hypothesis slated JJlI the present pal'er.
the subject o f such a sentence is hum~m and un-~p,,-cified, e.~.

LSOI'I1L'(IrjC l'o.ri ElW~y [TO,HI lh~ prisan'

The best evidence for Ib¢ hypothesis that tl'~ pa-~sive form oF a . ip!r~n~iti,.'¢
2g{I 7 FretL~.av~gi,~, i Pa~"~iv,~ eme/ ~t~tpt,e~tuu,~ petrsivt"

verb is ~ marker of an unspecified human subject is provided by senlencc~


in which the '.orb refers specifically to animal ra|her tban hurnar~ activiti¢~
or in which lh~ pragmatic context is such that it m~y indicate an unimtd
subject. Comr~lre the following examples containing tk¢ passive form of
the verbs j a r . "to run' and hara!~= "to run away':

juliy.I li hadiqaii d-h;~.'w-m~l


in Zo,~

huriha rain htLrlit]ati ,'1 hey~vavat


f~'om
'~r~sl;t',qi.~l' tall ;4'fl,id.y Ii'on) I l i ] t g / . o o

And I]naIly a verb thai sl'~cil~catly refers to non-human activity, nab.ha


•to Inlrk'~

Active: nab,ha el-kulbu "a dog b~rl.~d


P l~si~*" nuhiha fi eLmcdina "~'omu~a. bJirked i. I,e'~,n'

The last senle.ce may he used with a de.rog;~tory nma|fing in reference In


humans only. It ea~noL bc ta~cd in rcrcrcll¢¢ tO dogs.
The data from Arabic lhl,s support lira hypothesis about the function
of the !~,~iv¢ of intr;msit~ve ~rbs as proposcd in ~hc prcsen! paper. The
[n~p~rlance of {lie data from Arabic re,~ts not unit in the fact thai there
is one more language Io co.firm the hypothesis, b0t also in the f~c[ 4hat
this language belongs to a differe.t fatuity or languages. Sinc~ also in
Hebrew the impersonal passive indicates the unsp~-'ified human .sent
IRuth Berman. p,e.) we can assume ihut the Facts ill.strated for Arabic
are ;I]SO tt'U¢ for other Semitic lavgua~s,

2, 7 Dwdj and Germ~at

r[~]C data broughl Ibrth by Comri¢ {1977) and Perlmuttcr (J978) lot
Dutch .lid German differ in one major resl~ct from that brought in for
other languages. Some of the sentences which Comri¢ ~nd Perlmuttcr label
"im~rsaaal" may have aa agct~t addLnl to them and. therefore, they are as
umdl "~rson~F us any other s c n t c , ~ with ~1 p~rsonal subject. Compare
th,: followtn~ ,~entenc~s:

~ c r m a n ; Es wurde d~rl= $ch,~h.."r yore D..h[=r guholrcn '~


' T h e ~ ~,~; o. [e.]~hing ,~f the pupi/ by Ihc i['~"hrlr'
DaLch: Er wmdt hicr doo~- d~ j~mgdui v,.,~l gt,~mst
~Jt i~ dunccd I~cre -rL I01 h~,' ~hL" .i,tllnlg ~',.~ple'

Tile F.~ct Ihnl the 0as~i,~c ~r~l¢llCes ,~'ith ¢,~' in German or er in Dutch ,'is ~l
~urfn¢¢ subjcel c~tn ,dmit lhe ugeqti~',: phrase puts into doubt Ibc hylx~thc.si~
~,,~tneed and pra,,'ed for Tud:ish. Ar-ahic, L~fin. Spanish.. und h~tlian al~ou~
the indefinite agorot function oF the pazsivc o~ intra~|sitivc yetis. For, ii' the
hypt~,lhesi~ ~,~re la be Itlle ill ~ e n ] l ; t / ~ ~lld I')ulch. sentcn¢~.~ ~vitl~ age~itive
phrases ,~hould ~ot be grammatical. TEe t~¢t th~tt the "~bo,,,¢ ~entence~ ure
grahal'nal[cal, hOWg%'Cl', t]og'~ I l a [ nec¢~.~arily imply l_h:it the hypothesis is
f~l~e. Thcr¢ [s n strong possibility thul d~e above sentences have a difft'r~'~I
function f'rom their format co~_tnterparts ~la Romance lan~uzg,,s, And whilc
I am net prepared ~o cow,duct a synchronic analysis o1" eithcr Dotcl~ er
Ge'rlrlltn, Jl appears that the fmlelion or senllences with asentJve phrases
may be Ih¢ ~am¢ as the existential ~lhcr¢' scmcnce~ i~ English. vi~: ~tut]ng
t]mt '~uch arld such t2~cl, evcnL prnce~, etc., o~:ctirred'. Cc:tainly the
lr~:a~l~[Jons irtto English o1" the sentences with e~ and er surface 5lJbjec'l'~
~tapport the ~l'opo~ed intcrpret:~ion. Thus, Curme (1922: 33~) give~ tl~e
Ib]lo~'irtg Ir~ndations:

~i rd l,'cbu f~n
'Th-re i~ runni.ng L~t~ing ~n'
F:$ 'd,'LIl'de il'nl'lL~r vie] ~ptaltltllell'l~ ~d~ehe~i~t. trod get;icl~:
' / a c r e v,~ ~la, ay~ .~ .~,ot,d dc-al c f ,;h~fllis~g. j~,ki~g and [auphit~"
f.)l~n wird ~t:laflzl
"Th.ere is d.',r~cing going on up_~airs"

Apart from .~nt~nces ~s'ith =~gcnti~w t0hra.~es, the re~t (~f the ir~p~sonal
l~i~;es in Oerman and DuLch do not differ J'rom Iheir c o l z u [ e r p u H ~ in
Romance languages and clearly indicaie t h e pre.scac¢, o f ~ ~ttb,ju~'! which.,
~s in o~her languages exunnined zo far, ha~i to be human.

Es w ~ ~teTn yon tln~ ~man:,l

This senh~ncL" ap]~.~rs in Har:.llu]li~ (1965l, b,Jt I| 11~.$~,~.cn c~rlego:i¢'ally rejecqed b> the nrrqive
.~pcak¢~ ! have ¢o'~alled /,Itho~gh I do ~ol want to rule o,,t lhe po~ib,l,t~. ~.hat in ~ome
particular context Lhis r~ttlcn~c ;ould ~ a~;l~Zblc. I thi~k ill should he described only
,~ilhin such a ¢o~lleXl f~L~er ih~Ln he- taken ns a neuLr.~I u.~ample tbr Lhc passive COl~.lrtzcti~n.

~'itho~ any prr~hlem,


Curm¢ (1922: 3.~8) sklles a[~OIll Ih~ [mr~rsonal p~sives :

~-lli,~ c=mgru~i~rl ca. oo1~ bL" used v.ilh mtra~sili~.,s wlfigh e~pres~ an aciivil:~- gr ~=u,ldititm
Lhai ~ta~d~ in a relation :o ,', fre~" moral a~,e~t : F_~ wirzl~,'ge.~.vr~, g~,.~,qgcz.Je,i"They (;riddle.the)
;.Lie ealilqz. ~l'-'~:pi,$." hLL[ n<wl~.E,', ~'.','i'g g~:f.~imke."l, ffebth:l'..~t'r,a'l~'a-~qrff '~[llere is a sp.~k[illl~, it
is li~hln[~,~, lhc~' i.~ ru~h,inl~ at ~,,ater.'-~

[~ is difficu[| n~w Io interprel the term "free moral a~ent', but all the
examples ~ha~ C.rm¢ ~ransl~tes inv01"~¢ u human subj¢cl, e.g,.

The trlp~t important s u p ~ r l for this hypothesis COnleS~ however, from


the ~peakers of' GCTI-rlan who. when a~kcd to inlqrprel al! intransitive
scnlcno: with ,~.~a~ ~," surfa~: subjccL invariably gb, c .human~ as the trader.
lying sebj~cL of Lh~ ~ntenc¢.
We can no~ disetL~ the pL'obiem of whether or not the passive of
m~:ransiti,.e sentences rCpl'e~;tts the same phenomenon as the passive of
translti~ senlcnc~. From the ~aoint of view of the forn~ of the verb, the
answer i~ yes. From the palm of view of their fun¢lion, lh¢ answer is no~
The passive farm of intransitive wrbs indicates that i~ subjct't is human,
This is hal the [anctiott of Ihe passive foy]~ of the transitive verb, While
th~ un~rlying ~ubjcct is implied by the form ¢~f the intraasitiv¢ passiv~
verl:,, no impLication con~raing the subject can bc four~ in the I~assive of
transitive verbs.
As in other languages ~.-xamincd so far, there are no stative impersonal
passives in Germas~. Thus. attho.gh one can say Es .'ird gezan=z "I~¢r¢ is
dancing going on'. ¢h¢ slative counterpart of this sentcnc~ wi~h the verb
seil~ w';I not he grammalical: *E~; is| getunzt. $~il~rl~ ungrammuiicu[ is
"Et i~'t A'¢~c./gesdwben. whi~e E,¢ ward,, gegM g~,'lrJ~e, ~TIlete was playing
at ninepins going on' (Cu~me 1922: ~3~,) is grammati~l. ~
I~, mu;! b,= poinLed au~. that l~rlmalt~r (b~751 s z ~ /he fact that "iml~r~oul passives'
ma~ ~- u:~l anl:~ wi~h a verl~ dcneti~g willed activi~.. He does no~+ however, limit ther~
~o the htnn~n agliv[~t.~-=only, ~ C'¢idCt/l=~ b~' hk$ fv0tn~l¢ on p, 18~, [l~t ~wfl ~i~ ~LC~¢fl!
is n,~L ~ c L . ~[I would hi= ~:lt~er to th~ l[mh m say thal the irnlr,maon~l form denotes a
v , i i l ~ ae~ivity, r e ~ r d l c ~ or she .~¢~ used C 0 r a ~ the Polish example quoted earlier whh
the im'~r~onul [':Tm ~1- Lhu '~.~h- 'in Ia~n~' ~ , i s ~ . . which implies a ~, iHed activeLy uilh~u~.k
ih~: m[-mil~w ~~Fed haa n~ su~.h implication,
In Dutch. Impeder. the_ xia~iv~ iml~r~r~al pa~siv~ arc p ~ b l ¢ , ~',8, Er i~ ~¢I ~¢d,~n~
'lhere was much ,l~ahC[Itl]'- I am [psL~f.~ iv a~-i anpnyl]),.~i5 I~Vigw~r p[ I . ~ - ~ f..~r Lhis
in ['ar~at$oa.
Note Ihat in the passiw of' transitive verbs in German, the distinction
I:teLweerl statlve and non-staLive passive is mainlaitzcd. As in Spanish, die
implication of this analysis in German is that the impersonal pa,si~,¢s
repres,nt a distinct grammaLi~a] category l¥om that of other type~ nf passive.
With ~ q x c t to Dutch, Kir.~ner II976 : 389) ,tales th.at impersonal passive
e.~n b¢ ap,~lied only to ,~¢rbs denotiltg ht~man ,tcti~,,itie~ Pedmutter (1~78)
denies this by providi~ the rot|owing sentence, which he cl,aims ~.o be
grammatical:

E.r "~erd! gehhfl't fgehin~ik~Tge~rn~t/~J:rnluuv, d

Such sentences wer~ ¢mphatLcal]y rejected by a Dutch speaker who is not


a linguist. In addition, compare the ['t.,llowing ungraum'mtJcal sentences :

=Er wnrdl aan d~ d . ' - r g~l~lal~


"~t is h~rked at the door'

'~Er wordl doo~ d~ la,,~rLd~:ilvai~ tic: kel~;~¢l v,¢l,~¢lol~',a


I t is r~.m lr0m the kennel ~,y =he do~'

Note, ho.we~,'er, that the non-htzman a#ntivc phrase is allowed in d~c


¢~i~tenf.i0.1 sentence with o passive verb. Thus the following seniem.~e w-'is
accepted by a native s p e a k s :

Er wcrd d~,r de fallen aan de',,e me=,hel,~ geknaagd

b~t no!

*Er ~ M geknaag, d

[nl~r~ling evidence for the fat1; that Ihe impersonal ¢onslrll¢~ion arlually
:cpresents art indefinite human subject is provid~ by the ['act that |be
~ntences wiLh er... w~re spontaneously paraphrased with a men construc-
tion; for example, along with Er wurdr your de ko~ing gebogel= was provided
Men 3lligl Four de l~oma~., bulb meanin$ 'One kneels before the king'.
Virtually the sarrm holds for German. Thus. the impersonal sentences
denote only a human suhjecL and the following are ungrammatical in
C_mrman :
2~4 Z, Fraj:?wgw# i P.,~,~h'e wut imjqrr.~l~ui/ p~,~,~iri,

'P_~ I, urde l~i'il4h


'1~ h=~ Ix'~n h a t t e d ~

From the a l i v e discussion of DuLch and German, one m~,y conclude


tha~ the pv~sive xentence wilh the mirfac¢ subjcc( er and es, res~'~dvcly,
ha~ I~ its primii~ funcuon the ]ndiei~lion oF an nn~peciligd human agent.
This ~on~truetio~, ho~,e~e~, may haw an agenfive phra.~ added; then its
function i~ chanted ~o i n d i ~ c the existence of a liven stale, proc¢.~.% etc.

2,,~. WvAh

Aw~rV {1978) p~sltllatf.~ Ihe e~<isierimof Iwo lype~ of pas~iv~.s in Welsh.


one of which ,~he calls fuel 'g~t" passives, and th~ o~:hcr which sli~ call~
iml~r~nai passiws. The ¢~le# pi~s['~cs i~ivolv¢ the use of Ih~ verb cae/ us
lhe main verb m the ~-ntem:e which is inflect "d for tens~ p~ld a~pe~l. It also
alrccs with the following pronoun in number and I~rson. Sinc~ the agrcc-
rn¢~t ,~'ilh the f=~llowiitg pronoun is the main characteristic of the' ~-i-b and
~ub~ecl m tht: active scn~rlc~, the verb c a e / a r i d th~ following pronoun ure
c~rrespondirlgly called the v.:rb arid subject of the passi~= +¢nicnc~, There-
tore. th= verb of Lh¢ pa~ive sent~cc H different lioni the verb of the ilcti','¢
sentence. The ,a;rb of th~ a~'ii~ -~i.illell0e OC~"ILIP~ in the pa.ssive in th~-
uninl]ect~l form, The following is the Form of active and co~rCsl~ndin!l
vael p~ssive (A'~,l~ry t076: 48):

irll~eclet~

i:~fl~¢:cA -ninfle~tc~l

1"lie irrq~rsonal pas~i~,~ h~ls the l~tssiw form oF the verb that would have
occurred in the active. This verb, how'ever, does not agree with Ih¢ following

" Spe~lccr~ of German [~y r=u m e , h i =LEO'O"id +e+p¢<:l to the ~ ; ¢ p ~ l ~ l i t y of mttC+,i aentcm'cs.
All t)[ the umJic ~?-uakcr+ uP Germau I tiave com, ullP+d [ejected such +ont~¢¢.~. There is.
ho'.,,'c'.~er, a I'~s~ihilhy l'~hilcd O~Lt m me I~y Theo Yc'nnemann t h e stmh ~ntmces could be
~:ccpt=Elc in =p~.~;ia] ~;ir~ulzlslav;~=, ! I ~ l i c l l dial Ih~ ~li~lt~llces v f opinion ;n re+w-el to
tl~me ,~tcncc~ is alread2~ a ~Jgnill~al fact i~icming tlml they dJ[Ter From pu~ive m.~luu,~us
I~ith tr'ansllJvc v~.rL~ in ~hwh thuF¢ i~ ~o ~tl~h d~lTcr=ll~ of ulli~iioil~,
pro~ou~ in number or gen(ler. The pronoun, eherefore, i~ ¢~nsidered Io he
ohiect rather than subject. There is no ¢Icmen~ in this form of the pa~ivc
~,hieh does agree with the verb and, ~herefore~ Awher3, considers that ti|ere
is no sttrlkce ~trueuare subject, k is presumably because or" Ibis proper~y,
~.e., the L~m¢l~o1 the sL~rface strz,cture st, hject, lhat A~,,bery ~mlls this ~ype
of passive the impersonal ~ i v e ~ She, her~]f, h(~,~:ver~ maint,'tins on
purely formal @round~ that thes type of sentence has ~ eornplctely unspecified
subject, which cannot be deduced from the surface structure save for the
specific t'eature~ li~l nnay he required by tile sennanlic properties ~f the
verb. T:]'L¢ following is the form of abe impersonal i~tssive (w~th the corr¢~-
poncling ttctive form) (Awi_~ry IgTfi: 14B):

AL~ive : ~,'] NP~ Iqp a


I

h~a~-f~n~l pa~i've : Vj NP.~ lean NP.

i,llt'etic~lls

The impersonal [:mssiv~iJmWelsh does not t.~rrespond form'tlly or semantic-


ally ~o the impersonal passives ~ts the,' are kno~,n Jn other Innguages, e,g.,
in /~nguages diset~sed in lhe presen! paper. ! could ,o! find in A~bery
(19771 an answer to whether o r nol. the passives of intran~ativc verbs indicate
a haman or no~-human subject. Onfur~unaie]y, I did ~ot have an opporliJnily
to work with ~. native speaker of Welsh. Thus, the d.ata from Welsh neither
conlr~,:lict nor ~Ul:,porl the h,c~the~is about the function or passives of
irz~tratlsi~{ve verbs advan¢c:d [n t[l£- pre~nt paper.

3. lh~hy passi~e ©sate to indicate Ibe indefinite subject

The short review of data in the preceding section has ~hown that [n
atany Indo-Earopea~ languages and in Turkish, the l~-~sive form of the
inttanshive verb is used to express ~h¢ indefinite subject. This fact alone
may be taker, by some as an indieatiolt of cerlaiJi char'~cterislics of the
passive and, in fact, this is ~,hal actually happened Jn the qu¢|ed stodie~
of Comrie and P~rh~utlerJ
The semantic category o f the indefinite subject seems to have been
grnmmuticalized in h~do~Europe~m k'mguases by the use of already existing
devices, such as:3 p. plural, qrg., Russian; tt'3e oFlhe texeme indic~mtingll/al),
e.g.. Late Latin and man~ Gm-n~uuic iungu;~ges, including E,~glis~, at a
c~'rtain stage; and s+vcral languages" ~s~ of the passive constructiQn. Since
the lattm- a¢+tirr~:l in many unrelated dallguages, ane sh+uld rule out the
possibility of borrowir.g and, rather. Ior~k into the condilions which made
this cllat]ge of funelion p,~ssib[e.
In order to do this, it is necessary to start with the function of the
passiv,: con:z~.~etion. Following Kur),iowicz ( ] 946), ~nd sligh|ly ref~rmulatint;
his stateJneftts, I take Ills p;imary f~tnetlon o f the prssix,e construclion It+
ir~dicalClhat I/1¢ only N P that oo~ur~ with a transitive verb i.J 11OI the expected
agenl.
There are several ,.~onsequenees 1o this assumption. The firs~ one i~ Lhe
fact th;=t the ag~rtt, s i n ~ it is not mentiort~d, remains unknown - at least
on ;he basis ol" t]~¢ anforraatic~n provided by the sentence. -[rhc second eon.
seqllence is the fhet thal~ since the construction is used to indicate the rote
of the N P witli !Jar transitive verb. ~! is available, as it were, to be treed in
sora¢ other function with intransitive verbs. Since the implication a f the
passive co;lstruclion with transitive verbs was Ihal the agent :e=nain~ u~,-
known, Ibis implication, or s¢c'ondary f'tmction, becomes t]i¢ primary func-
tion when used wilh the i,~J.ransil_ive verbs. Thus the change of function of
the pus@co ¢onstru~.'ttou is a well-motivated process,
In Polish,-'~ecording I~ K]eiruensiewiez (191~1), the forms t~lt'~ll, riolTEl~ ¥tTl?fflt~ltJ;
etc. ~on¢ spoke'. ~on¢ made', etc., derive from the forms m6u,ioJro if,st
•spoken is', robi,s~e~.je.vt 'made is', at:. They have be~r further extended to
irtclurle transitive verbs, e.g., budoa~tno dora)' "houses were built', rohiotto
plany "pla0s were re'ado', ¢1¢,
The changes in Romance languages involved at least three step~. Fir.~t~
the im~--rsoaai Forms in Latin. i.e.. ~he form used in subjectless senten¢¢~
came to indicate pa~sL~¢. Then, passive with intransitive senten~s came to
iudjcale indefinite agent Laler, in Latin itself, and most of its descendents,
the lexeme h~ntu was ta,~ed to indicate the indefinite a ~ n t . This rt+rm wa,~
abandoned in the Middle Ales in most of the Romar, o: langL=ages, e~¢ept
for French and Calta~an~ =rod was replat~d once again by the non-~tative
passive, ix., by th¢ forms using the retie×ire pronouns s u ~ as se c~tre,
se hahta, et¢~ (for a pal'lial e~planation, s,~=~Nyrop 1925: 368).
A similar p r o e e ~ took place in English, which u;ed ~ realize the indefinite
agent by the pronoun 'man', whi,,h later was replaced by a number o f
grammatical devices, sttch as 'people', "one'. 'they °, 'y~u', and by a passive
construction (Visor 1¢/63 : 511.
In Old Irish, the l~SSiV¢ of intransitive verbs ~rves to express the
ZL ,~?~4i=.lwga+: P,w~ire g, td imperxt,md ptt.~'~ilv 2.',7

in~definile human agenl, e g , t~.6gm" 'let people_ snmeone go" lit. "lel it he
gone', re'hoti~ 'people have Ix:ca', ¢~¢. (Thurn~sen 1946: 328),
Th~ semantic ~tteg~ry of inde~nite ;~geJlt can be an ~ndcp~ndent gram-
nautical ¢,tlegory. i.¢.. realized by its own means r"a:her tha~ by mearts which
ha,,~ some other primary f~mt:tion, as is the case i~) Indo-European lan-
guages, Thus, in Hausa, a Chadic language, the indefinite agent is realized
by a speCiu[ p r o n o u n a and its appropriale v~riznts in diffexnt aspeclua[
and cea~ forms, Thus one can say.

0 ~,uki 3"~ 8udLI

AIi 7~ta k~Udu


'Ali I1='1~; r~lll ~uw[iJ"
[411 gil~ LI
"(.]/'~¢ Ilil~ rLJl) i.lWdy"

The l ~ t senlerlq¢' can be ttsed only whejt Ihe agent who did Ihc running
is human.
As in I,do-Euroix'aa I~.Jlg0tage~, ~he gramr, adcal lorm o f iadefinJtq agenl
cannot h~ used with verbs proper to animat behax, itw only; I hUS the sentence

"An yii Imushi


make I~zrlcit~g
'On~ harked"

is uagrammari,'=l in Hausa~
An ]rttereSlil]g question is whcLhe/ m~c t~u'J claim thai it k a universal
property ol" impersonal passives ~hul they lndi¢~.te an indefinite human
subject. A definite ansx~:r to such a question may ~ obtained only through
an exar~in~lioa of ~ll the languages that hav~ a ~ s i v e laura and thal h a v e
~n innpersona[ passive, There are several factors that have In be t~ken into
co~xsideration ia a study of universal proGcrly. The first is the fact thai
*.here are othe~ de,ices ir~ languages thtlt can b~ tJsed to i,di~lte an indefinite
h a m a n subject angL thc~for¢, even if there were a passive ~onstruction, an
impersonal imssive does not h~ve to nectar. But it is als0 likely that in the
~amc situation an impersonal p:~ssi,.'e would occur, but its function wJ]] be
differ~t. This apl~ar~ to I:~ the ¢~se with passives oF intransilive verbs in
Saltskrit wlxich appear to take the human subject in the form of instrumcmal.
e. 8, (l:h'ajerski 1979 : ~.M-}:
"il is m~ne t~ me'_ '1 ~o"
,~up~,"~tc ; v.'=}.-;t ",t ~s slcpi b)' ~otr. "you ,=iccp"
~il ~-dIg a l i ¢ l i a "it ~.~ ,~¢nl b.~" I]im'. "h= ~.¢l~I"

I. view oF the a b ~ e Sanskrit c~ampte nnd ia view of thc Ju.¢t that n o


exJliaustJ~;e ~ttJdy of all i~mgti~ges l~hut may h~,e an impe~soJtal passive was
made, tile question of the universality of the property of imper~nal passive
of, allot be ~tnswered in zt ye~/no manner. Instead, it is propt~sed that this
properly may be a reason for ~he imp.~onal passives to emerge.
Therelor~ ralher than considerirl 9 a form as having or , o l having ~ given
function, one may' look at it as llavint a potential to realize a given I'anedon.
The data front Arabic, . ~ e r a l lndo-E,ropean langl~nges, and flora Turkish
support the hypotb.esis that the p~sivc of irltrarlsiliv¢ verb~ has (=he potential
of i'~;alizJrig the i'urlctioil of an indefir, ite human subjecl.

4. Conc[minm

It has been ~h~'a,~ tha! the ~:;~lcgorJ,' which Comfit 11977) ~tnd Pcrlmuttcr
~1978) de~'Jbe a.~ impcr,sonal passive is p~ssive onl~ in form. While its

Tah],: ]

.~ltl~'rlC'~ s l r u e t u r c characicrisries
f=,tlicng Function

r['rStl'i&iti¥~" absen~ m u r k e d the ~zm~ indic.ales


Pas.~i~e mOrT~htdogy a'= subject 0r tha~ 1he ~ub.
or r¢ll~xivc F,] GillillJ.ll$ txaa~iLiv¢ ~ j¢Cg fU~-
innraasilive marked de-
menu} is

Pl~sitc ~c~r ph,~l ~.[t~' mnrkc'd ject o4" I r l U l ~ r a ~ e r than


~r [cl]c.~i~: p r o n o u n ti~e ur in- ¢xpt~tcd
tra ~ t i ' ~ e agrnc

Tt~n~r.iv~ ~- i~- ph~cn[ ahr,cne


I h.a,l tJi+
~nssi,~e mori~h ~Iog~ ~ n l e'l't 12g
or rcflcxk~e prc~Koun ha; an, h~-

"I-rNt~ili~¢ ~b~:nq mark¢~ a~ obj¢¢l t ~maJt ~ t


Pass]'~e m~rpholoJzy in an active
nr rellexl'~e pronoun ~nl,Cl'l¢.~
futlCtiOi+ {s 3elive~ il .diff~.¢s ~'rt~ll'~ ~,ther active' .~en:'cacc~ in h+tving an
indctinil¢ human .,=ubj;e¢[; ehe active scnlen~s have ~1 specified subJect,
hUItlall, rtotl~hLllrll~lll~ al~iL,~lRtt or .ott-aJtJmare,
These scnwnces differ from impersoJlal sent~=}~s in [he .'~am¢ |~:uture,
e~¢ept Ib~l itnperson~l sentences indicate the lack of any subjecl in the
semantic structure of th~ s e n t e n c e . They differ from passive sentcnc¢.~ in
Sl:3eC'i.['yirjg s o m e fepturcs o f the aemantlc subject w]tlle the ['EISSiVe Sell teJlL'es
do not specify th= ~rnanti¢ subj~t in any way,
In this w a y it w a s p o s s i b l e IO obtairn is f'unctio|l=d diffcrenliarit}ln I ~ : ~ J ~
the ibur ~_r'~mm~lic~l ~.-,*,~.e~orLcsin ]an~uaBes i~ which u]l tour occur~ Ta ~1o I
slllllfllurize$ t h e rela6onship b e t w e e n the ~tJr~tce StFuctur~ marking oil verbs,
agetlt, a n d patient (direct object) a n d the p r i m a r y f u n c t i o n s Ihnt the variotL~
c o r r e l a t i o n s h a v e in [~lngtlages d=sct~ssed

Refermees

Am'~zy. O.M.. ~Q7~. "l'he ~|IL~ eft Welsh. Londar=: C.rnbrkJ~e |)niv. Pres,~

C'omrie. l~.. I977. |~l del'cn~' ,af ~.'ffgtLI./l~'~u~ ,~ICJliOI~AJ-II~"IhC imper~om~l passive. 2n: Pc'ter
C¢~t¢. J;rr~)ld M. S~d~'k [¢d.~,)~ Syn(~ an~[ ~¢mantics, ~ 1 . 3 . 4 7 5~ N~;w York : A~:adcmic
iI~FL~5.
Curn~.GO,< 1~22. A 7.ra~rl~r of I~¢ G¢~ma~ Jal~$u;ql~. N~- Y~rE'. Mucmilla~.
Kraou|, A.. F. l-[Iol:ia~, 1972. S~,rnl~l~Cel'llinc Paris : Krinck~eek
Fr~jzyng~cr.Z., 1978 An an~lysb er k-.pa~iv;~, kingu~ 46, I J) 156,
Fr~.~V==~e[. Z,. TJ [,Z.alr[el, mbL Prhnary arid ~ecn~dary £|mclJOn in lancua~e ,;h,,~.
Harznung, W_. I'J6~. [lie Pu~sivlran.-~rm~tienc.~ im [.~tscl-cn_ Studlu Gra~nm,tJca ~, 9(i
114,
Halmann, ] B.. A. ~r~nlyr~ ]gE-'i k~t.-l-nis;'b~ SynlHx lind ~l~';~;tJk Mihtchen : ~eL'k
I ~ i r ~ r , l~ ~, 1976 On lhe ~ubjeerrleSs "pseu~0-pa~iv¢' in slpndard Dutch. In; Charles Li
(¢'lJ.~. Subjcd ;llIl~ topJCl N¢~ York; At:ade~' Press.
Klcmen~k'~vk'z. 7~,. 19~,t. Histeria jCzyka poHk~eb,o, '='oJi I. WaI~W.
K,ofl¢l,%qcla. H.. 1 9 ~ . U bU,i;Ivwi¢ ~a]~[~ j~.C[)gfflU paskow%~ ~4,3~' kilEuI/t,. Porad,ik ~ykowy.
Z e ~ t 90, 38:~-.391.
[q,46. ~.r'galivno~[' ] stadisJ'n~K ~ v jazyl~e, |zvc.~tm Akadcmii Nauk SSSR.
~l~.~'~.rl'lO'd~i~,~),,
vO] :;, rlr, ~I ~.77-393.
i,epschyr~ G-, 197d, A~eLnne Cl~lrIjzioni c~n si. [n: ~t,di Lingtlisfiei ~ri .oitor'e di Tr'i~tano
l~fdli~ Pica; P'=¢inL
Ma~i~,~_]wic~ } . V , r,~g. o d~,.~ sir~tuk~J.~-skixfi(mslrukciiax ~ovremenneg<>pal%koge jazyka.
[d; "~le~o'v~qija I:~ p~l'~,k~mu jez.Xku, J[4-13.:1. M~:'~ow Naug~.
Nyrop, K., 1925. ~r[arnm:]Jrc hi_~torLque de lu langue rr~.n~'aise. Copmhagt'.z,_
Perlmmlcr, 13. M.~ [973. Iml~rS0~al passi,,es and the irnaccusntJve hypnlhesis_ Proccec~infs
9r the I:@tlrlll A,rmal Mccti~t$ .of ~,h¢:Bcyk¢loy LJngui:.li~; $~.[¢t~, J57-18~.
$1a','iamkaj-', fi]~-L~lBia~. 267 2 ~
SuV,er. M.. ]974. Where does imper.~nal .~e ~ome ~ m ? In'. R../. Campbell M . G Ge,ldin,
M~ ~.layr~l~ WaJq~ (ed~.). LJJ~guLr~rJ¢s~u<~i~s in Rolnan¢¢ laap,a~s. 146 1~7. Waslfin~t~m.
DE: (~ecrge~w,n Univ P~ess.
Tl,,ro:y~-a, R., 1~46. :~ ;~ramma~ eF Old Irish. D.uh,lia,
Vi~scr. V.T., 19~,3. ,*,n hislorical .~yrxla~ of lhe English lan~l,al~', pan I. Leiden: Brill.

You might also like