. 12
( 17)


essential in reference -- in [241], ˜that which would qualify as a token of navy-

[241] Z yt bvtk vfnhjcrb<gen> .
I did not have a navy jacket.

Negated, bv†nm takes the genitive almost obligatorily, over 95 percent of the
time. Predicates that report perception or cognition (d«ltnm ˜see™, py’nm ˜know™)
are weakly existential, in that they report the presence of something in a person™s
cognitive space. They are more likely to use the genitive than other verbs (85%
genitive, as opposed to 60% with other verbs).
At the opposite extreme, predicates like cxbn’nm rjuj ´-kb,j r†v ˜consider
someone as someone (something)™, yfpd’nm/yfpsd’nm ˜call™, yfpy’xbnm/yfpyfx’nm
˜appoint, designate™ are in effect transitive copular predicates. Like other pred-
icatives, they presume the existence and individuation of the entity. If negated,
they use the accusative.

[242] Ntgthm vs e;t ,jkmit yt cxbnftv yfib ghjcnjhs<acc> ,tcrjytxysvb, f yfib
,jufncndf ytbcxthgftvsvb.
Now we no longer consider our expanses endless, or our riches inexhaustible.

These verbs do use the genitive in senses other than the predicative: tckb yt
cxbnfnm tlbycndtyyjuj j,hfpwf<gen> cntrkf ˜if one does not count the lone ex-
ample of this glass™.

Argument individuation: Arguments that are individuated in reference are
more likely to occur in the accusative, while arguments with essential refer-
ence prefer the genitive. At this point it will be useful to refer to a small sample
of ¬fty-one examples of the context yt gj ´vy/ ˜I don™t recall™ in one memoir
Predicates and arguments 325

Table 5.10 Genitive of negation/yt gjvy/

acc gen % gen

animate (proper), without yb 3 2 40
singular concrete 5 12 71
singular abstract 0 7 100
plural 0 11 100
0 11 100
total 8 43 84

(Table 5.10). Since the verb -- a transitive existential -- is held constant, case de-
pends primarily on the referential properties of the argument.38
Proper nouns and common nouns referring to unique animate beings are
likely to use the accusative (around 90%, as opposed to 30% for common inani-
mate). In the sample corpus, three of ¬ve tokens of animate objects (without yb)
are accusative:

[243] D эnb lyb z cjdctv yt gjvy/ Fylh/ie<acc> .
I have no memory of Andriusha at all during those days.

The two tokens with genitive have essential reference -- in [244], ˜no memory
of a person ¬tting the description of a teacher™.

[244] Z exbkfcm ljvf. Gj irjkmysv ghtlvtnfv yt gjvy/ exbntkmybws<gen> .
I studied at home. For academic subjects I don™t remember having any teacher.

At the opposite extreme, abstract nouns and event nouns are very likely to
use the genitive, as are plural nouns (in this sample, exclusively):

[245] B ybrnj, rhjvt vtyz, tt gjke,kbpytwf, yt gjvybn nt[ ktn<gen> tt ;bpyb.
No one, except me, her near twin, remembers those years of her life.

The one context of variation in this sample (Table 5.10) is singular concrete
inanimate common nouns. The genitive is used when there is no memory of
singular entities that have essential reference -- in [246], there is no memory of
whatever the color of the binding was:

[246] Wdtnf<gen> vjtq j,kj;rb yt gjvy/, ,skj bplfyj b[ hfpys[ wdtnjd --
vfkbyjds[, cbyb[, ptktys[.
The color of the binding I don™t remember, it was published in various colors --
raspberry, blue, green.

38 A. Tsvetaeva, Vospominaniia (Moscow, 1971).
326 A Reference Grammar of Russian

The genitive is more usual, but the accusative is possible if there is partial
memory. Example [247] contrasts the one fact the speaker fails to recall (the
name) with the positive memory of other facts about the individual (background,

[247] D jlyjv bp fynbrdfhys[ vfufpbyjd Vjcrds dtkbrbq ryzpm, -- tuj bvz<acc> yt
gjvy/, -- pyfnjr uhfd/h, jy bcrfk nfv xtuj-yb,elm lkz cdjtq rjkktrwbb --
hfccvjnhtk ytljeybxnj;tyyst ghbpyfrb ghbyflkt;yjcnb uhfd/h
Hevzywtdcrjve vept/.
In one of the antique stores in Moscow a grand prince (his name I don™t
remember -- he was a connoisseur of engravings, he was looking for something
for his collection) -- discovered the still not completely obliterated traces of the
fact that the engravings had belonged to the Rumiantsevsky Museum.

5.4.5 Genitive objects: summary
Negation always allows the possibility that the alternative possible state of af-
fairs could be envisaged, but the alternative can be more or less prominent.
When a transitive verb is negated, the genitive is appropriate to the extent that
the import is to negate the whole event unconditionally. For this reason, the
genitive is appropriate when the opposite polarity of the predicate is irrelevant
or precluded. Emphatic negation entertains but dismisses alternative states of
affairs; in gerunds and participles, the currently reported property is presup-
posed, positive alternatives being deemed irrelevant. At the level of argument
reference, the genitive is appropriate when the object has essential reference --
when it cannot be de¬ned independently of the currently reported property, as
is the case with abstract and plural nouns, and all there is to say about the
object is that something ¬tting the de¬nition of being a such and such does not
participate in the event.
The accusative is appropriate with negated transitive verbs to the extent that
the opposite, positive, polarity of the predicate is entertained or implicitly as-
serted, as it is with certain phrases (x´nm yt) or with modalities in which the
opposite polarity is kept in view (in questions, counterfactuals, expectations
in context). When the current negated situation is not the only situation that
might be reported of the world, the communicative focus goes to the contrast of
polarities, and the negation of the event (and the failure of the object™s partici-
pation) are no longer unconditional. At the level of the argument, the accusative
is appropriate to the extent that the reference of the argument is individuated.
When the object is individuated, it is relevant to the text in ways that go beyond
the current negative proposition, and the negative situation being reported can
be viewed as a property of the individual named by the object, rather than as a
property of the verb and object as a whole. For this reason, proper names and
Predicates and arguments 327

nouns referring to animates are generally accusative. In the most general terms,
the accusative is appropriate to the extent that the negated situation is only one
among many things that might be said about the object entity.

5.5 Secondary genitives and secondary locatives

5.5.1 Basics
For most nouns of Declension<Ia> the genitive ends in {-a}. In addition to
this “primary” genitive (or gen1), certain nouns have the possibility of us-
ing a “secondary” genitive (or gen2) that ends in {-u}.39 Also, certain nouns
of Declension<Ia> use a secondary locative form (loc2) ending in stressed
{-ú} instead of the expected locative form {-e} (loc1). For some nouns of
Declension<IIIa> there is variation in the place of the stress in the locative case
form. For the nouns that have the variation, the unstressed {-i} is used in the
same contexts as the primary locative loc1 of Declension<Ia> , while the ending
with the vowel stressed {-í} is used in contexts analogous to those in which the
loc2 in {-ú} is used.
The uses to which the secondary case forms are put are among the regular
functions of the genitive and locative cases: the secondary cases are indeed gen-
itives and locatives. Both secondary case forms are restricted to a small number
of lexical items. For some lexical items, these secondary forms are quite stable,
for others less so. Over time, the secondary cases are gradually becoming more

5.5.2 Secondary genitive
Gen2 is used most freely with mass nouns designating solids or ¬‚uids, portions
of which can be detached and manipulated -- measured, purchased, consumed.
It is used with appreciable frequency only with approximately a half-dozen such
nouns, with less frequency with another dozen. Diminutives retain gen2 well.
With other nouns, gen2 is residual.

[248] Nouns in Declension<Ia> using gen2
SUBSTANCES AND FLUIDS [more frequently]: x’q ˜tea™, c’[fh ˜sugar™ , csh ˜cheese™,
c´g ˜soup™, v=l ˜honey™, nf,’r ˜tobacco™, rd’c ˜kvass™

39 Overview and statistical information in Krysin 1974:165--73, 246--49 (gen2), 174--79, 250--51 (loc2).
40 In the study of Krysin (1974:169), the commonplace comestibles c’[fh and x’q ranked lower
than half a dozen other nouns: rd’c (75%), ndjhju (59%), nf,’r (51%), k’r (50%), csh (49%), djcr
´ ´ ´
(48%), and only then c’[fh (44%), x’q (42%). The low rank re¬‚ects how the questionnaire was
constructed. The ¬rst six were used only with the verb reg«nm, the most favorable context for
gen2. C’[fh and x’q were tested in other contexts, some of which discourage gen2 (dsgbnm cnfrfy
rhtgrjuj xf/<gen2> ˜drink a glass of strong tea™ -- unfavorable; ghjvsiktyyjcnm dsgecnbkf ,jkmit
cf[fhe<gen2> ˜industry produced more sugar™ -- very unlikely). In a contemporary search on the
328 A Reference Grammar of Russian

SUBSTANCES AND FLUIDS [less frequently]: dj ˜wax™, rbgznj ˜boiling water™,
´cr ´r
rjymz ˜cognac™, rj´hv ˜food, fodder™, i=kr ˜silk™, ´rcec ˜vinegar™, g†htw ˜pepper™,
gjhj ˜powder™, l=ujnm (l=un/) ˜pitch, tar™, k’r ˜lacquer™, rthjc«y ˜kerosene™, ndjhj
´[ ´u
DIMINUTIVES OF MASS NOUNS [frequent]: vtlj ˜honey™, rdfcj ˜kvass™
´r ´r
ETHERS [unusual]: dj´ple[ ˜air™, ghjcnj ˜space™
EVENTS [residual]: cv†[ ˜laughter™, dplj ˜nonsense™, i©v ˜noise™, cnh’[ ˜fear, terror™

Gen2 is occasionally used with borrowings ([249], on Washoe™s sign language):

[249] Pyfr: Gbnm Sign: Drink
Description: Hand balled into a ¬st,
Jgbcfybt: Herf c;fnf d rekfr, ,jkmijq
fore¬nger touching the mouth
gfktw rfcftncz hnf.
Context: Asks for water, medicine,
Rjyntrcn: Ghjcbn djls, ktrfhcndf,
kbvjyfle<gen2> . Ghjcz lemonade. Asking for lemonade,
kbvjyfle<gen2> , xfcnj rjv,bybhetn cj often combines with the sign for
pyfrjv ¤ckflrbq≥.

The extension to borrowings such as ijrjk’l ˜chocolate™, l;†v ˜jam™, or kbvjy’l
˜lemonade™ suggests that gen2 has been mildly productive, but overall, its use is
being curtailed. Very recent borrowings such as qju©hn ˜yogurt™ are unlikely to
develop gen2.
The possible contexts in which gen2 appears are these: (a) in the partitive sense
of the genitive, with verbs reporting transfer (reg«nm ˜buy™, ghtlkj;«nm ˜to offer™)
or consumption (dsgbnm ˜drink down™, c(†cnm ˜eat up™); (b) with negation, often
emphatic, especially negation of the same verbs that could elicit gen2 in its parti-
tive sense; (c) with approximate quanti¬ers and quantifying predicates (cnj ´kmrj
˜so much™, [dfn’tn ˜suf¬ces™); (d) domestic measures of quantity (cf[fhe<gen2>
rk/yen cfve/ rhjitxre ˜of sugar they put in a small pinch™; lfkf tq recjr gbhjuf
b xfire xf/<gen2> ˜she gave her a piece of pie and a cup of tea™; gjkkbnhjdfz
,fyrf vtle<gen2> ˜a half-liter tub of honey™; ,tksq kjvnbr cshe<gen2> ˜a white
chunk of cheese™; exceptionally ghbdtp wtksq xtvjlfy jn,jhyjuj, leibcnjuj b
rhtgxfqituj nf,fre<gen2> ˜I brought a whole suitcase™s worth of select, pun-
gent, dark tobacco™); (e) with speci¬c quanti¬ers (gznm rbkjuhfvv nf,fre<gen2> );
and (f) residually, as idioms with prepositions (cj cnhf[e<gen2> ˜from fear™, stylis-
tically marked, ,tp cf[fhe<gen2> ˜without a dose of sugar™, and certain ¬xed
phrases, «p ljve<gen2> ˜away from home™ vs. bp lj <gen1> ˜from out of the
building of the house™. Descriptive modi¬ers reduce the likelihood of using gen2:
dsgbk xf/<gen2> is nearly universal (97%) while dsgbk rhtgrjuj xf/<gen2> is not

web (<02.XI.02>) of collocations with the in¬nitive regbnm, xf/ and nf,fre scored well over 50
percent; cf[fhe and csh were around 50 percent; other nouns occurred too infrequently to allow
even impressionistic judgments about frequency.
Predicates and arguments 329

(55%).41 Contexts that are then excluded from using gen2 are prepositions
(except some residual idioms), transitive verbs other than verbs reporting trans-
fer or consumption when they take the genitive under negation, and adnominal
genitives: gznyf xfz<gen1> ˜spots of tea™, ,tctlf pf gbnmtv xfz<gen1> ˜a conver-
sation while drinking tea™, e/nysq pfgf[ nf,frf<gen1> ˜the pleasant aroma of
It is often said that gen2 can always be replaced by gen1, and in a sense that is
true. Gen2 forms are above all genitive; these contexts are all contexts in which
gen1 can be used. Yet when gen2 is possible, it contributes an extra nuance.
Gen2 is most natural in contexts in which the predicate detaches and de¬nes
a recognizable quantum of the mass; the event creates a dose, a portion -- with
the intention or result that the dose of the mass can be manipulated in a con-
ventional, domestic way. For this reason, gen2 is most frequent in collocations
such as gjg«nm x’/<gen2> , understood as a ritualized event:
[250] {jhjij gthtjltnmcz d ce[jt, gjpfdnhfrfnm, gjgbnm xf/<gen2> , jnlj[yenm.
It would be good to change into dry clothes, have breakfast, drink some tea, relax.

In [251], the purchase de¬nes the portion and its function (to be eaten):
[251] Pf[jntkjcm tcnm, b jy regbk cege<gen2> b ghbcnhjbkcz hzljv c ldevz
He wanted to eat, and so he bought some soup and set himself up next to two

Gen1 is used for the partitive sense if the idea of conventional portion is lacking,
as in [252], where the mushroom-gatherer imagines buying types of things, not
[252] Jyf ghbltn gthdjq d ktc, yf,thtn gjkye/ rjhpbye cfvs[ kexib[ uhb,jd,
ghjlfcn b[ yf hsyrt, regbn ,tkjuj [kt,f<gen1> b cf[fhf<gen1> .
She™ll be the ¬rst in the forest, she™ll collect a full basket of the best mushrooms,
sell them at the market, and buy some white bread and sugar.

With negation, gen2 is used when the corresponding positive sentence would
otherwise use gen2: dsgbkb xf/ ˜they drank some tea, engaged in the ritual of
tea-drinking™, yt dsgbkb xf/ ˜they did not have a chance to engage in the ritual
of tea-drinking™. With y†n, gen1 denies the universal availability of sugar ([253]),
while gen2 denies the existence of the requisite dose of sugar ([254]):
[253] Yt ,skj cf[fhf<gen1> , b c nheljv, pf ,jkmibt ltymub ljcnfdfkfcm cjkm.
There was no sugar, and it was only with dif¬culty, for a lot of money, that salt
could be acquired.

Whole web, <02.XI.02>.
330 A Reference Grammar of Russian

[254] Jyf vyt ghtlkj;bkf xfq, yj bpdbybkfcm, xnj e ytt ytn cf[fhe<gen2> b [kt,
She offered me tea, but apologized that she did not have sugar and the bread was

With quanti¬ers and quantifying predicates, gen2 is used again for actions
that create conventional portions or conventional events: yfgbnmcz dtxthjv
ujhzxtuj xf/<gen2> ˜drink all one wants of hot tea in the evening™; yfkbkf tve
xf/<gen2> ˜she poured him (a portion of, a cup of) tea™, or doses of domestic

[255] {kt,f -- recjxtr, f cf[fhe<gen2> cjdctv ytvyj;rj, c xfqye/ kj;txre.
Of bread, there was a small piece, and of sugar, very little, about a teaspoon™s

Yfhj ˜people, folk™, unusual as a mass noun in that it refers to animate
beings, is widely used as gen2 -- with an explicit quanti¬er ([256]) or even as
subject genitive ([257]):

[256] Vyjuj yfhjle<gen2> ghjdj;fkj vjkjls[.
Many people accompanied the young people.
[257] Yf ltcznm ktn yfhjle<gen2> [dfnbn.
There are enough people for ten years.

To review: gen2 is most natural in events that report that a quantity of the
mass is detached and manipulated in a conventional, even speci¬cally a domes-
tic, fashion.

5.5.3 Secondary locative
Loc2 in stressed {-ú} is used only with the two locative prepositions d and yf,
but not with j, gj, ghb, prepositions that govern the locative case but whose
meaning is less spatial. Less than two dozen nouns in Declension<Ia> use loc2

[258] Nouns in Declension<Ia> using loc2
d†nth ˜wind™, uh©yn ˜soil™, l©, ˜oak™, p©, ˜tooth™, rh/ ˜hook™, k†c ˜forest™, v=l
˜honey™, vj ˜moss™, vsc ˜cape™, j ´ngecr ˜leave™, g«h ˜feast™, cy†u ˜snow™, cj ˜juice™,
´[ ´ ´r
cnj ˜haystack™, [k†d ˜livestock shed™, [j ´kjl ˜cold™, w†[ ˜workshop™, x’y ˜vat™, x’q ˜tea™

The nouns listed in [258] use loc2 with different frequency. In the sociolin-
guistic survey conducted in the 1960s (Krysin 1974), the use of loc2 in d cytu©
˜in the snow™ remained constant at 97 percent for speakers from the oldest
cohort to the youngest. For other nouns, the percentage of speakers who used
Predicates and arguments 331

Table 5.11 Usage of loc2

oldest youngest
cohort cohort tokens total percentage
{-u} {-u}
noun (%) (%) loc sg

97 97 1050 1053 100
d cytue
41 37 91 321 28
d wt[e
35 22 47 68 69
d vtle
25 17 68 658 10
d jngecre

= Krysin 1974
= http://www.lib.ru <04.VI.02>

loc2 declined slightly from the oldest to the youngest cohort (Table 5.11). The
frequency of usage was checked on a website with extensive contemporary
Russian texts (www.lib.ru) for four nouns (Table 5.11). The frequencies of us-
age are comparable to the ¬gures recorded a quarter of a century ago, except
with the idiom d vtl©.
With cy†u, loc2 is used almost exclusively. It speci¬es the kind of medium or
location in which a state or activity is situated ([259--60]):42

[259] Jy c nheljv ,htk d cytue<loc2> xfcf ldf.
He wandered through the snow with dif¬culty for two hours.
[260] B tve gjxtve-nj nz;tktt ,skj ghtlcnfdbnm ct,t nt[ nht[ xtkjdtr, rjnjhst
kt;fkb yf cytue<loc2> .
It was for some reason more dif¬cult to imagine those three people lying on the

At the opposite extreme, with j ´ngecr, loc1 is now almost universal; it describes
an of¬cial status ([261]). Loc2, when it is used, is an informal, less bureaucratic
variant ([262]).

[261] Gjkj;tybt e ytt ,skj nhelysv. Vfnm c jnwjv d jngecrt<loc1> , ,f,eirf tkt
[jlbn . . .
She has a dif¬cult situation. Mother and father are on leave, grandma can hardly
walk . . .
[262] -- Cflbcm b jnls[fq -- Ns dtlm d jngecre<loc2> .
-- Sit down and rest -- after all you™re on leave.

42 Loc1 can be used with cy†u if the noun is understood as an abstract repository of various prop-
erties manipulated by mental processes, as in Jakobson™s (1936/1971[b]) constructed example,
:bdjgbcyjcnb d cytut ytn ˜there is nothing picturesque in snow™.
332 A Reference Grammar of Russian

With rh’q, loc2 has become idiomatic; it is basically restricted to unique loca-
tions that have characteristic properties -- familiarity ([263]), remoteness ([264]),
or extremeness ([265]):

[263] d yfitv rhf/<loc2> in our region
[264] d cfvjv lfktrjv rhf/<loc2> in the most remote region
[265] d nft;yjv rhf/<loc2> the taiga region

In contrast, loc1 is used, for example, to differentiate one region from another
(d Rhfcyjlfhcrjv rhft<loc2> ˜in the Krasnodar region™). In a similar fashion,
d vjpu©<loc2> ˜in the brain™ is used for the brain as seat of consciousness, d
vj <loc1> for the physiological organ. D rheu©<loc2> ˜in the circle™ means a con-
text for something (d rheue<loc2> tuj bynthtcjd ˜in the con¬nes of his interests™),
especially a social context (d ctvtqyjv rheue<loc2> ˜in a family environment™, d
rheue<loc2> nfywe/ob[ ˜among those who were dancing™). D rh©ut<loc1> describes
the geometric ¬gure (c yfhfcnf/otq crjhjcnm/ dhfoftncz d rheut<loc1> ˜he spins
in a circle with ever increasing speed™). In these three instances, the difference
is very much lexical; rhf/ and rheu© evoke different senses of the nouns from
rh’t and rh©ut.
W†[ ˜shop™ is one of the few nouns which has real variation in usage. Loc2 d
wt[©, less bureaucratic than loc1, presumes that the properties of this locus are
known and serves as a background for other events ([266]):

[266] Ghjikj ytvyjuj dhtvtyb, b Jkz dlheu gjxedcndjdfkf, xnj ,thtvtyyf <. . .> D
wt[e<loc2> ;tyobys chfpe gjyzkb d xtv ltkj.
A little time passed, and Olia suddenly became aware that she was pregnant
<. . .> In the shop the women understood right away what was up.

(Understood in generic terms, as a type of livelihood, d wt[©<loc2> can also
characterize a person: ;bpym d wt[e ghjikf ˜she had passed her life in the

[267] Dct jyb hf,jnfkb d эrcgthbvtynfkmyjv wt[t<loc1> , ult cnhjbkbcm vjltkb.
They all worked in an experimental shop where models were built.

D w†[t<loc1> is used as focal information, for example, to differentiate workshops
For nouns of Declension<IIIa> , the difference between loc1 and loc2 is one of
stress: loc1 j rhj but loc2 d rhjd«. Stress on the ending in the locative case
has begun to yield to stress on the stem, to judge by warnings in manuals of
usage. As the stress changes, the stressed and unstressed variants can acquire
different senses analogous to the senses of d rheu© vs. d rh©ut, etc. As shown in
Predicates and arguments 333

Table 5.12 Loc2 in Declension<IIIa>

bare noun and preposition novel collocation comment

(a) d uke,« ˜at a depth™, d lfk« ˜at a d ytj,sryjd†yyjq lfk« ˜at consistent end stress,
distance™, d rhjd« ˜in the blood™, d an unusual distance™ all contexts
uhzp« ˜in ¬lth™, d cty« ˜in the
shade™, yf wtg« ˜on a chain™, yf jc«
˜on the axis™, d yjx« ˜in the night™
(b) d/yf gtx« ˜in/on a stove™, d nty« ˜in ´vtyyjq g†x« ˜in a blast end stress in bare
d lj
the shade™ furnace™; d эktrnhjg†xb noun, some variation
˜in an electric oven™, d in novel collocations
gjken†y« ˜in half-shade™
d cntg« ˜in the steppe™∼ yt d cn†gb d hfcr«yeditqcz cn†g« ˜in
[3 sources] the ¬‚ung-out steppe™
d cdzp« ˜in connection™ [5 sources] ∼ d ythfphsdyjq cdz
´ ´p«
d cd„¦p« [1 source] ˜in unbroken contact™
[2 sources]
(c) d ctn« ˜in the network™ [1 source] ∼ d fyukjzpsxyjq c†n« ˜on stem stress common in
d c†n« [3 sources] ∼ d c†nb the English-language novel collocations,
[1 source] net™ occasional in bare
noun [substandard]
d otk« ˜in the slit™ [4 sources]∼ d
o†k« [2 sources]
(d) [only] ’yutk dj gkjn« ˜angel in the d xtkjd†xtcrjq gkj ˜in end stress only in
¬‚esh™ ∼ [usual] d gkj ˜in the human form™ idioms

c†n«, etc. = variation in the position of stress

Table 5.12,43 there is a gradation of possibilities: nouns that still have the end
stress of loc2 consistently; nouns that have loc2 without modi¬er but occa-
sional variation in novel collocations with modi¬ers; then nouns with variation
between loc1 and loc2 in novel collocations, sometimes even in phrases with
no modi¬ers; and nouns that have generalized loc1 (stem stress) except in ¬xed
phrases. Over the very long term, the use of loc2 is on the decline, but each
noun has its own preferences. Among nouns of Declension<Ia> , loc2 is still close
to automatic with cy†u and k†c, but is little used now with most of the other
nouns listed in [258]. Loc2 is also on the wane among nouns of Declension<IIIa> .
When both variants are possible, loc1 is used for novel combinations that de-
¬ne a new individual on the spot, while loc2 presumes that the entity and its
properties are familiar.

43 Sources: Rozental and Telenkova 1976, Ageenko and Zarva 1967, Gorbachevich 1973, Avanesov
and Ozhegov 1959, Ozhegov 1989, Zalizniak 1977[a].
334 A Reference Grammar of Russian

5.6 Instrumental case

5.6.1 Basics
The instrumental is the one case other than the genitive that is used in a wide
range of contexts. Though heterogeneous, these contexts have some similarities
and connections.44

5.6.2 Modal instrumentals
Closest to the use of the instrumental for predicatives (§5.2) is the use of the
instrumental to state a simile, dsnm djkrjv ˜howl like a wolf™ or to name a
function of an individual, hf,jnfnm by;tythjv ˜work as an engineer™. In both
uses, the construction identi¬es the subject entity as being like unto a certain
type (˜wolf™, ˜engineer™) in some respect, but stops short of saying that it is to be
identi¬ed completely as belonging to that type.
Certain idiomatic phrases with the instrumental case describe the medium of
an event -- the location ([268]) or time ([269--70]).

[268] Ljhjuf ikf nj ktcjv<ins> , nj gjkzvb<ins> , xthtp lthtdyb b ctkf.
The road went through the forest, over ¬elds, through villages and settlements.
[269] Gjt[fkb ghjljnhzljdws relf-nj d etpl ljcnfdfnm [kt,, b b[ cgzob[ yjxm/<ins>
The provision brigades went off into the hinterlands to get grain, and then they
were killed at night as they slept.
[270] F d Dtkbrbq xtndthu gjckt cke;,s ldtyflwfnb tdfyutkbq pdtplyjq yjxm/<ins>
vs ytckb pf;;tyyst cdtxb.
And on Maundy Thursday after the service of the Twelve Gospels we carried lit
candles through the starry night.

The instrumentals identify a type of medium in which a certain activity is ap-
propriate -- a type of road in [268], a type of time (nighttime, with overtones of
mystery in [269--70]).45

5.6.3 Aspectual instrumentals
A characteristic feature of Russian is the use of the instrumental with predicates
that describe activities in which a human agent moves a body part of the sub-
ject or an immediate extension of the body: vf[yenm {herjq ∼ nhzgrjq} ˜wave
{with the hand ∼ a rag}™, nhzcnb {ujkjdjq ∼ herjq ∼ gbcnjktnjv} ˜shake with

44 The contexts discriminated by Jakobson 1936/1971[b], 1958/1971[b] have been decomposed into
syntactic structures by Worth 1958, restated by Wierzbicka 1980, and translated into cognitive
grammar by Janda 1993.
45 Giusti Fici 1989:64: the instrumental “est fonctionnel par rapport au mouvement de passage en
soi, et [. . .] il sert à le caract†riser.”
Predicates and arguments 335

{the head ∼ hand ∼ a pistol}™. The body part is synecdochic to the aspectuality
(change) of the predicate. When the mobile entity is a separate, external entity,
rather than a body part or an extension of a body part, these predicates are
transitive and use the accusative for the mobile entity:

[271] Jlyb wtkjdfkb tt, lheubt vjkxf nhzckb here<acc> .
Some kissed her, others silently shook her hand.

Many of these predicates are intrinsically cyclical, and so form semelfactive
perfectives in {-nu-}: vf[’nm ˜wave [continuously, repeatedly]™, vf[y©nm ˜give a
single wave™. Some have re¬‚exive transforms in which the mobile entity is the
subject: herb nhzcencz ˜hands shake™, vs nhzckbcm ˜we shook™. The full range of
constructions is attested with ld«ufnm(cz) ˜move™. It uses the instrumental for
synecdochic parts ([272]), the accusative for separate entities ([273]), the re¬‚exive
transform for spontaneous motion of body parts ([274]) or autonomous agents

[272] Z ldbufk kjrnzvb<ins> d ,jrf.
I moved (with) my elbows into people™s sides.
[273] Rnj-nj cnfk gkzcfnm, ldbufkb c ievjv vt,tkm<acc> .
Some started to dance, they moved the furniture noisily.
[274] Tuj otrb<nom> ,scnhj ldbufkbcm.
His cheeks moved quickly.
[275] Jyf<nom> pf;ukf e;t cdtxb b ntgthm ldbufkfcm r cnjke.
She had already lit the candles and now was moving towards the table.

Other predicates use the instrumental in a similar fashion, although they do
not have the same range of options as nhzcn«(cm), ld«ufnm(cz). Some predicates
occur only with a synecdochic body part, and therefore consistently use the in-
strumental: vbufnm\vbuyenm ukfpfvb ˜blink with the eyes™. Verbs reporting the
emission of a sensory signal express the locus of the signal in the instrumental,
crhbgtkb djhjnjv ˜they squeaked with the winch™, rjgsnf ,ktcntkb gjlrjdfvb
˜the hoofs gleamed with the horseshoes™, ,ktcnbn kfrjv yjdtymrfz ,fkfkfqrf
˜the new balalaika gleams with lacquer™, or else the locus of the signal is nom-
inative, crhbgtkb rjktcf ˜wheels squeaked™, pe,s ,ktcntkb ˜her teeth gleamed™.
Similar is g’[yenm ˜smell™. Its instrumental is metonymic to the general aspectual-
ity of the predicate, which is the emission of a smell: gf[ytn {lsvjv ∼ jdwfvb ∼
cdt;tcnm/ ∼ vtljv ∼ ctyjv} ˜it smells of {smoke ∼ sheep ∼ freshness ∼
honey ∼ hay}™.
A small set of verbs that report launching projectiles (vtn’nm\vtny©nm ˜toss™,
´cbnm/,hjc’nm ˜throw™, idshz ´nm\idshy©nm ˜chuck™) can take either the ac-
cusative ([276--77]) or the instrumental ([278]):
336 A Reference Grammar of Russian

[276] Bp gjxnjdjuj dfujyf rblfkb vtijr<acc> c gbcmvfvb b ufptnfvb.
Out of the postal car they would toss a bag with letters and newspapers.
[277] Yt cgjcj,ys[ ,hjcfnm rfvyb<acc> ltntq djhjys ybrjulf yt ,jzkbcm.
Children incapable of throwing stones were never feared by crows.
[278] :tyobys heufkb rjnf, ht,znbirb rblfkb rfvyzvb<ins> .
The women cursed the cat, the lads chucked stones.

The accusative reports a directed change in the aspectual argument, the instru-
mental a type of activity affecting the missile, such as the pelting with stones
that befell the tomcat ([278]).
And there are also instrumentals that specify the nature of the mobile
entity that affects an (accusative) patient of a transitive predicate, as in
cyf,l«nm/cyf,;’nm ˜provide™: jyf cgjcj,yf cyf,;fnm эktrnhjэythubtq<ins> ujhjl
˜it is capable of providing a city with electrical energy™, cyf,;tybt athv
nt[ybrjq<ins> ˜provision of farms with technology™. This is the normal valence
of this verb and of verbs like pf,«nm: pf,bdfkb hs,jq ,jxrb<ins> ˜they stuffed
barrels with ¬sh™, or [279]:

[279] Ubvyfcnthrb pfcntuyekb yf dct geujdbws b yf,bkb cjkjvjq<ins> .
They buttoned their coats all the way up and stuffed them with straw.

This and similar verbs sometimes use an alternate valence, in which the mo-
bile entity is accusative and the domain is a (directional) prepositional phrase:
yflj pf,bnm d uytplf lthtdzyyst ghj,rb<acc> b dyjdm ghbrhenbnm iehegfvb gtnkb
˜you have to drive wooden plugs into the holes and screw in the hinges again™.

5.6.4 Agentive instrumentals
Consistent with its name, the instrumental case is used to express instruments --
that is, metonymic extensions of the subject™s agentivity:

[280] F. Rfhgjd gthtitk kbyb/ if[vfnyjuj эrdfnjhf dnjhsv cdjbv rjytv<ins> .
A. Karpov crossed the chess equator with his other knight.

The instrumental case can be used (though in practice infrequently) to express
the displaced agent of a passive construction (§5.8). The instrumental is used in
a construction somewhat like a passive, in which a transitive verb in the neuter
singular lacks a subject and reports an act of nature that affects the patient,
expressed in the accusative. The instrumental expresses the metonymic force of
the event™s agentivity:
[281] Rjvyfne pfkbkj djljq<ins> .
The room got ¬‚ooded by water.
[282] Vtyz cbkmyj elfhbkj njrjv<ins> .
I got a hard shock.
Predicates and arguments 337

Table 5.13 Types of instrumental constructions

context example interpretation

predicative ,snm ghbdktrfntkmysv<ins> ˜be property holds in one predicate
attractive™; ghbqnb history, fails to holds in a
hfpjxfhjdfyysv<ins> ˜arrive parallel history in another
disillusioned™ time-world
simile dsnm djkrjv<ins> ˜howl like a wolf ™ predicate history holds in
imagined world of comparison,
though not in actual world
t[fnm {ktcjv<ins> ∼ yjxm/<ins> }
medium medium in which event, as type,
˜travel {through the forest ∼ at
(location / time) is embedded
manner ujdjhbnm htprbv njyjv<ins> ˜speak entity characteristic of activity as
in a harsh tone™ type of activity
aspectual nhzcnb herjq<ins> ˜shake a hand™, entity synecdochic to aspectuality
,ktcntnm evjv<ins> ˜shine by
means of the mind™, ,hjcbnm
rfvyzvb<ins> ˜throw stones™,
cyf,;bnm ujhjl эythubtq<ins>
˜provide the city with energy™
instrument gthtqnb kbyb/ rjytv<ins> ˜cross the entity synecdochic to agentivity
line with the knight™
pseudo-passive of rjvyfne pfkbkj djljq ˜the room entity synecdochic to agentivity
natural force got ¬‚ooded by water™
passive agent entity synecdochic to agentivity
,thtuf pf[dfxtys lfxyjq
ge,kbrjq<ins> ˜the shores were
occupied by the dacha-goers™

An instrumental is often used to express the manner of an activity:

[283] U. Rfcgfhjd xtnrbvb vfytdhfvb<ins> abueh dsyelbk hfpvty b jn,bk fnfre.
G. Kasparov with precise movements of the pieces forced a trade and repelled the

Here the instrumental seems to be intermediate between an instrument in the
strict sense and a circumstantial instrumental like yj
´xm/ ˜at night™.

5.6.5 Summary
The basic uses of instrumental are summarized in Table 5.13, with examples. In
predicative constructions (§5.2) the instrumental case imputes two alternative
predicate histories. In one the property holds, in another, the property does
not. More broadly, the instrumental can be said to manipulate two situations. It
both differentiates them and also connects them as part of a larger picture. In
338 A Reference Grammar of Russian

the contexts listed in the second half of Table 5.13, the entity expressed by the
instrumental is tangentially involved in the progress of the event; the entity is
in a relationship of synecdoche to some other more central agent or aspectual
element, or to the general idea of agentivity or aspectuality. For example, the
contribution of the instrumental entity as agent or aspectual argument is only
partial, incomplete; it is connected to agentivity or aspectuality, but that entity
is not identi¬ed completely as the primary agent or aspectual argument. In the
middle of Table 5.13 are constructions in which the instrumental is used in
adverbial functions. They describe a history that has one shape -- harsh-speaking
or forest-traveling at night (through the medium of forest) -- and that shape is
linked to but differentiated from other imaginable types of histories. If there
is a unity in the constructions employing the instrumental, it is the way in
which two alternatives are proposed, where the asserted history is viewed as a
synecdochic part of a larger history.

5.7 Case: context and variants

5.7.1 Jakobson™s case system: general
In two studies twenty years apart, Roman Jakobson developed an analysis of
the case system of Russian that is both of historical and continuing interest.46
The analysis, formulated in the spirit of the structuralist intellectual climate of
the period between the two world wars, consists of the following interlocking

Invariant meaning (Gesamtbedeutung): Jakobson proposed that each case has
a consistent meaning, or value. That value is present in all contexts in which a
case is used -- with verbs, with prepositions, with adjectives or nouns.

Binary feature analysis: Collectively, the cases form a tightly structured system
in which each case can be speci¬ed by positive or negative values of a minimal
number of binary features. Over the whole system, the features are utilized as
fully as possible.

Markedness: The binary features are asymmetric: for each binary feature, one
value is marked (more narrowly de¬ned and restricted in usage), the other un-
marked (broader in de¬nition and usage).

46 For useful translations of Jakobson 1936/1971[b], 1958/1971[b], see Jakobson 1984. Commentary
and application: Neidle 1988.
Predicates and arguments 339



Fig. 5.1 Jakobson™s feature analysis of Russian case

Maximalization: Jakobson included in his analysis the two secondary cases.
(Fig. 5.1 and the discussion below ignore the second genitive and locative.)

Syncretism and iconicity: The binary feature analysis of case predicts the
occurrence of syncretism (the same morphological expression of different case
endings). Syncretism occurs between cases that are similar and share features;
that is, similarity in value is matched in an iconic fashion by similarity of mor-
phological form.

5.7.2 Jakobson™s case system: the analysis
Jakobson™s analysis of the six basic cases can be represented as in Fig. 5.1.
Nominative and accusative are [ ’p e r i p h e r a l ] , inasmuch as these cases are
used for the major arguments of predicates. The distinction of [±p e r i p h e r a l ]
¬ts with contemporary theories of syntax that distinguish between syntactic
cases (= [’p e r i p h e r a l ] ) and semantic cases (= [+p e r i p h e r a l ] ), except that
in contemporary theories, syntactic case is automatically derived from a syn-
tactic structure and is thereby devoid of meaning, whereas Jakobson exactly
wanted to argue that all cases in all contexts have value. Even if the features
of Jakobson are utilized in a contemporary approach as notational devices anal-
ogous to phonological features, the spirit in which the features were intended
differs radically. The accusative and dative are a class and share the feature value
[+d i r e c t i o n a l ] , since they express the direct and indirect objects; both can
be said to occur with arguments to which activity of the predicate is directed.
The locative is transparently [+p e r i p h e r a l ] . It is less than obvious in what
sense the locative is [+q ua n t i f y i ng ].
The genitive and instrumental (§5.6) are the cases where the issues of invari-
ance and binary features come to the fore. Both are used in a wide range of
The genitive is used with prepositions, with verbs, with quanti¬ers, and as
internal arguments of noun phrases. These uses, claims Jakobson, all re¬‚ect a
restricted quantity of participation by the argument marked with the genitive
340 A Reference Grammar of Russian

This formulation makes sense with contexts which measure the quantity in
some way -- with quanti¬ers and verbs that govern the genitive. Quantifying verbs
like [dfn«nm ˜be suf¬cient™, yfcjk«nm ˜salt up a whole lot of™, yfcvjnh†nmcz ˜look
at to one™s heart™s content™ measure quantity of participation against an implicit
standard. The partitive usage is quantifying (dsgbnm x’z<gen1> ˜drink some tea,™
§5.5). The genitive of negation could be viewed as restricted participation (§§5.3,
5.4). What this formulation means with respect to the internal arguments of
noun phrases -- possessors -- is less clear, unless one takes this to mean that
the possessor participates only by virtue of serving in a limited role relative to
another entity -- the head noun of a noun phrase. But this is a rather different
sense of limited participation from the genitive used with quanti¬ers.
In a similar fashion, the uses of the instrumental can be seen as related. The
instrumental of simile and the predicative instrumental propose that an identity
or property holds of something, but only partially (§§5.6, 6.2). Similarly, the
agentive instrumental (true instruments, instrumental in pseudo-passives) and
the aspectual instrumental (in nhzcnb herjq ˜shake [with] one™s hand™, idshznm
rfvytv ˜throw [with] a stone™, gf[yenm jdwfvb ˜smell of sheep™) identify an entity
that participates in the event in a certain way -- as agentive or aspectual -- while
at the same time the entity is synecdochic to agentivity or aspectuality in general
(vtyz elfhbkj njrjv ˜I was hit with a shock™, vf[fnm herjq ˜wave with the hand™).
What really characterizes the instrumental, then, is synecdoche: it indicates
an entity that is part of the larger agentivity or aspectuality of the predicate.
As Jakobson de¬nes the instrumental, it is positively de¬ned for the feature
[+peripheral], and it is negatively de¬ned for other features. The de¬nition is
not suf¬ciently re¬ned to get at what is involved in the instrumental: an entity
is part of the whole, but not the whole story.
Whether these various contexts of the genitive, and the various contexts for
using the instrumental, reduce to a single invariant meaning (Gesamtbedeutung)
is ultimately a question of how one conceptualizes grammar. Jakobson seems to
assert complete unity, but does so exactly by exhibiting the heterogeneous con-
texts in which a case is used -- for example, the contexts of the instrumental case
listed in §5.6. No matter what, a grammatical description will have to contain a
list of the various contexts in which a case is used. Wierzbicka™s exposition (1980),
intended as a defense of Jakobson, does exactly this; it recognizes a set of con-
texts and gives somewhat different paraphrases for each. Inevitably one comes to
a network model, a model that describes a set of partially distinct but partially
related contexts or constructions. Once the network of contexts is spelled out,
the question of whether there is an invariant meaning (Gesamtbedeutung) fades
in importance.
Predicates and arguments 341

Where Jakobson™s de¬nitions of case have some special insight is in contexts
in which there is synchronic variation. For example, saying that the genitive
is quantifying does get at something of the variation between accusative and
genitive in the context of the genitive of negation: the genitive is indeed used
when the utterance denies participation -- that is, when participation of an
entity is quanti¬ed negatively.
The assessment is then mixed. The various constructions or contexts
(Sonderbedeutungen) of each case have to be distinguished and described in some
way, as partially distinct constructions. The fact that there is some similarity
is inevitable, since the various constructions have developed from common his-
torical sources. If one attempts to generalize over all contexts, the resulting
overarching, Platonic de¬nition will be vague. Yet an invariant value proves use-
ful as a way of interpreting the sense of ad hoc variation of cases in contexts in
which there is active variation.

5.7.3 Syncretism
While Jakobson formulated his analysis primarily in order to account for the
meaning (value) of cases, he also attempted to demonstrate that syncretism
matches meaning -- that is, that cases which have the same morphological
expression have similar meanings, and speci¬cally that all instances of syn-
cretism -- the same (or similar) morphological expression for different cases
-- occur between cells that are adjacent in Fig. 5.1.47 Similarity in form oc-
curs only when there is similarity in meaning. For example, nominative and
accusative form a class because they merge in the singular of inanimate nouns
of Declension<I> ; this small class can be de¬ned as [’peripheral, ’quantifying].
When the genitive and locative plural of (inanimate) adjectives merge -- Jakob-
son™s gbdys[<gen=loc pl> ˜alehouses™ -- that syncretism can be described simply
as the merger of [+quantifying, +peripheral] cases. In this way syncretism ap-
pears to be iconic of meaning.
While it is true that all instances of syncretism occur between cells that
are adjacent in the pictorial representation, it turns out to be dif¬cult to de-
¬ne that concept of “adjacent cells” in terms of features; complex manipu-
lations are needed. For example, accusative and genitive are merged in ani-
mates. To state this, one has to say, as in [284](a): among [’peripheral] cases,
[+quantifying] (genitive) syncretizes with [’quantifying] if the [’quantifying]
case is also [+directional] (accusative), but not if it is [’directional]

47 On case geometry and Fig. 5.14, see Chvany 1982, 1984, 1986, McCreight and Chvany 1991.
342 A Reference Grammar of Russian

[284] syncretism featural statement
® 
(a) acc = gen [’peripheral]
± 
 [+quantifying] 
 
 
° »

 [+directional]  
® 
(b) dat = loc [+peripheral]
± 
 [+quantifying] 
 
(/Declension<II> )
 
° »
 [+directional] 
 
± 
(c) gen = loc = dat [+quantifying] 
® 
 [+peripheral] 
 
(/Declension<III> )
 
° [’quantifying] »
 
 
 
(d) gen = loc = dat = ins [+quantifying]
(fem sg adj; numerals) [+peripheral]

A similarly complex statement has to be used with the dat and loc, which
merge in the singular of Declension<II> : the [’q ua n t i f y i ng ] case syncretizes
only if it is also [+d i r e c t i o n a l ] ([284](b)). A disjunction of features is required
to state the syncretism of the set {gen, loc, dat}, which occurs in the singular
of Declension<III> ([284](c)), or the syncretism of the set {gen, loc, dat, ins},
which occurs in feminine singular adjectives and some numerals ([284](d)).
Thus the patterns of syncretism between cells adjacent in Fig. 5.1 are not
actually predicted directly and transparently from the feature de¬nitions; extra
statements are needed. In fact, it has been pointed out that, if the six basic cases
(not the secondary cases) are arranged in the linear order: nom, acc, gen, loc,
dat, ins, then only cells that are adjacent in this one-dimensional list tolerate
syncretism, as marked by shaded cells in Fig. 5.2.48 This linearization makes it
clear that the patterns of syncretism have their own logic that is not directly
tied to the featural de¬nitions of Fig. 5.1.

5.7.4 Secondary genitive and secondary locative as cases?
Jakobson included in the discussion the secondary genitive and locative. Jakob-
son™s inclusion of the two secondary cases has attracted some attention, the more
so since he changed the featural de¬nitions of these cases from the ¬rst study
in 1936 to the second in 1958. In 1936, to characterize the distinction of two
genitives and locatives, Jakobson invoked a special feature not otherwise used;
gen2 and loc2 were said to be marked as [+shaping] with respect to gen1 and
loc1. (In Jakobson™s language, gen2 and loc2 indicate “etwas Gestaltendes oder

48 Chvany 1982.
Predicates and arguments 343

Fig. 5.2 Syncretism and linearization of Russian cases

zu Gestaltendes,” meaning that gen2 shapes a mass quantity and loc2 serves as
a container, thereby shaping something else.) In 1958, the analysis was changed,
and gen2 and loc2 became [’directional], like nom and ins, while gen1 and
loc1 became [+directional], like acc and dat.
As has been noted, the revised analysis of 1958 is the less appealing.49 Gen1
and gen2 are equally directional or non-directional, since both can equally be
used for objects of predicates in the partitive meaning (regbnm cf[fhf<gen1> ∼
cf[fhe<gen2> ). In the older two-dimensional ¬gure, some features, speci¬cally
[±shaping], were of limited utility. Now in the revised analysis, the system of
eight units makes a cube in which all features are used to the maximum. The
1958 analysis seems motivated less by patterns in language than by the desire
to produce an elegant geometric ¬gure.
Are gen2 and loc2 separate cases?50 Perhaps the question is misguided. Per-
haps we should not be forced to declare either that they are cases (if so, why
are they so limited?) or that they are not cases (if not, why is there nevertheless
some small difference in meaning between the secondary and the primary cases,
some of the time?). It might be preferable not to put the question in terms that
require one to choose yea or nay.51 What these secondary cases are is alternate
morphemes for the basic genitive and locative cases used under some conditions,
with elusive semantic and stylistic overtones, with certain lexical items. Here, as
in other instances of change, the older form is retained in the older, idiomatic,
abstract uses, the newer form is employed for novel combinations not learned

49 Worth 1984, 1998.
50 See Comrie 1986[a], 1991 on the theoretical problem of de¬ning cases.
51 Despite Mel chuk™s pronouncement (1986:56): “One cannot, however, talk about ˜variants of a case
2™ or about ˜case allomorphs that differ semantically™ (as is sometimes done): these expressions
are logically absurd.”
344 A Reference Grammar of Russian

as conventional phrases. This picture of the dynamic development is lost if one
is forced to answer a binary question.

5.8 Voice: reflexive verbs, passive participles

5.8.1 Basics
Most verbs take the same cases in their arguments in all contexts in which they
occur. In the few instances in which one verb allows arguments in different
cases -- accusative and genitive for objects of negated verbs, accusative or in-
strumental (idshy©nm {r’vtym<acc> ∼ r’vytv<ins> } ˜toss a stone ∼ engage in
stone-tossing™) -- the verb has the same form; only the case of the argument
differs.52 There are, however, two productive patterns for modifying valence in
which the verb changes shape: re¬‚exive verbs and passive participles of verbs.53

5.8.2 Functional equivalents of passive
The passive constructions of Western European languages do several things at
once. The agent, which in the active construction would be the (nominative) sub-
ject, is downgraded to an oblique case, if it is mentioned at all; more commonly,
it is not mentioned. The patient, which in the active construction would be the
(accusative) object, is given more prominence in its new role as the (nomina-
tive) subject. Together, these two criteria amount to a re-weighting of the two
arguments of a transitive predicate. In addition, a passive construction often
has different aspectual connotations from the active; in particular, a participial
passive reports a resulting state rather than an event.
Something of the effect of a European (speci¬cally English) passive is achieved
by other means in Russian. To avoid assigning explicit responsibility for an event,
Russian uses the third-person plural form of the verb, whether transitive or

[285] Z tuj [jhjij gjvybk, tuj c[dfnbkb<pl> ghzvj yf ,fpfht.
I remember him well, [they] seized him right there at the bazaar.
[286] Vfif jnrfpsdfkfcm, tq uhjpbkb<pl> fhtcnjv.
Masha refused, [they] were threatening her with arrest.

The construction fails to specify the identity of the individuals responsible for
the event, even if something more speci¬c could be said. In [287], the woman
speaks of being observed in general, though her grounds are that she knows
that one individual is observing:

52 Apresian 1974. For a broad conceptualization of valence, see Kholodovich 1970, Ruˇiˇka 1986.
53 The exposition here makes use of Hudin™s (1990) analysis of the functions of the three passive (or
passive-like) constructions (and examples [287], [298], [307]).
Predicates and arguments 345

[287] J,thyekfcm. Gthtl ytq cnjzk ve;xbyf, dscjrbq, [eljq.
-- Jy [=jcnhjd] lfdyj jnrhsn, jndtnbkf jyf b cenekzcm, yterk/;t, pyfz, xnj pf
ytq yf,k/lf/n<3pl> , djikf d djle.
She turned. In front of her stood a man, tall and thin.
-- It™s been discovered for a long time, she answered, and stooping,
uncomfortably, aware that she was being watched, got in the water.

The construction with unspeci¬ed third plural, then, establishes the existence
of individual(s) responsible for an event, but refuses to name them. In this con-
struction the object is often placed before the verb (§7.3.6), where it is linked to
the prior discourse ([285]). This has the effect of foregrounding the object, an-
other typical function of the passive.

5.8.3 Reflexive verbs
Many verbs include a morpheme descended from the historical enclitic re¬‚exive
pronoun, -cz in its fuller form (after consonants, but after both consonants and
vowels in active participles), -cm in reduced form (used after vowels, except in
active participles). There is a number of recognizably distinct, albeit related,
types of re¬‚exive verbs.54

Reflexivum tantum: Some verbs are only re¬‚exive, ,jz ´nmcz ˜fear™, ck©ifnmcz
˜listen to™ (though related to other verbs from the same root), ,jhj ´nmcz ˜¬ght, to
struggle™, cvtz ´nmcz ˜laugh™, yfl†zncz ˜hope™. The verbs tend to be semi-transitive;
there is often another argument that is involved in the activity, similar to a direct
object, but less directly affected. Historically the argument could not occur in
the accusative. In recent years, these verbs that formerly governed the genitive
(,jz ´nmcz, ck©ifnmcz) have begun to allow the accusative ([218]).

“True” reflexives: Certain verbs seem to be literally re¬‚exive, in that the subject
acts on the self. Such verbs are now limited to conventionalized, domestic activ-
ities involving contact with the self™s inalienable body: vsnmcz ˜wash (oneself)™,
,h«nmcz ˜shave (oneself)™. As a rule, except for this small set of verbs, an action
performed on the self is expressed by an argument pronoun ct,z Where both
re¬‚exive verbs and re¬‚exive objects with ct,z exist, they differ in meaning.
Vsnmcz is an intransitive activity, not directed at an object, whereas vsnm ct,z
´ ´ ´
is an activity directed at an entity, which could be any thing (such as a horse),
but in context happens to be the same entity as the subject:

[288] Z ,sk d ,fyt. Z vsk ct,z otnrfvb, rfr rjyz.
I was in the bathhouse. I scrubbed myself with brushes, like a horse.
54 For the basic typology, see Ianko-Trinitskaia 1962, Gerritsen 1990.
346 A Reference Grammar of Russian

Certain common re¬‚exive verbs have acquired the sense of engaging in an
activity intensively with the self and all parts involved: ,hj
´cbnmcz ˜throw oneself,
to lurch™; ,«nmcz ˜beat against (for example, snow against a window)™, not the
same as ,bnm ct,z d uhelm ˜beat oneself on the breast™.

Reciprocal reflexives: A number of actions that portray individuals (or groups)
mutually acting one on the other are expressed as re¬‚exive verbs: lh’nmcz ˜en-
gage in ¬ghting each other™, vbh«nmcz ˜reconcile™.

Habitual reflexives: With a very small number of verbs, the re¬‚exive implies
a predisposition to an activity, the classic collocations being cj,frf recftncz
˜the dog bites™, rjhjdf ,jlftncz ˜the cow butts™, rhfgbdf ;;tncz ˜the nettle

Phenomenological reflexives: Some intransitive verbs report a manifestation of
color and its perception:

[289] F dyenhb Rhtvkz gjlybvfkcz dscjrbq ,tksq cnjk, rjkjrjkmyb, hzljv
,tktkb ldf gznbukfds[ cj,jhf.
Inside the Kremlin there raised up a tall white bell tower, alongside showed
white two cathedrals with ¬ve domes.

The corresponding re¬‚exive attenuates the manifestation of color and its per-
ception. It is partial, or is visible through obstacles, or is unstable and runs the
risk of disappearing:55

[290] Djpybwf gjrfpfk yfv ryenjv: -- F djy b Rbhbkkjd ,tkttncz.
Our driver pointed with his knout: -- Over there Kirillov is just beginning to show

To this type belong: ptkty†nm(cz) ˜become (show) green™, ntvy†nm(cz) ˜become
(show) dark™, cby†nm(cz) ˜become (show) blue™, rhfcy†nm(cz) ˜become (show) red™.
In practice the re¬‚exive forms are not frequent.56

Modal impersonal reflexives: Some intransitives form a re¬‚exive that treats its
argument as a domain to which some attitude or inclination to perform an
activity is ascribed, often an inclination that is negative or inhibited: (vyt) yt
cgbncz ˜it isn™t sleepy to me, I don™t feel sleepy™, yt hf,jnftncz ˜doesn™t feel like
working™, (yt) [jxtncz ˜it isn™t appealing to™.57

55 Ianko-Trinitskaia 1962:227, Gerritsen 1990:40--46.
56 On the web (<01.XI.02>), non-re¬‚exive past-tense forms of ,tk†nm outnumbered the re¬‚exive
forms by a ratio of 96 to 1, those of rhfcy†nm by a ratio of 153 to 1.
57 Mr’zek 1971.
Predicates and arguments 347

[291] Gfyjd cjdctv lhtvktn. Pfnj Gjgjde yt cblbncz yf vtcnt.
Panov is very drowsy. In contrast, Popov can™t sit still.

This type of re¬‚exive verb is formed mostly from intransitives, and so the re-
sulting valence is impersonal. But some verbs occur with a clause that has the
role of an argument: [jxtnmcz dthbnm ˜one would like to believe™; levftncz, xnj
˜one is inclined to think that™; yt dthbncz, xnj ytn nt,z cj vyjq ˜I don™t want
to believe that you™re not here with me™.58 {j ´xtncz can even take a nominal
argument: ybxtuj tve e;t yt [jxtncz ˜he no longer wants anything™; vyt nj;t
[jxtncz cxfcnmz ˜I also would like happiness™.

Quantifying reflexives: In combination with certain pre¬xes, the re¬‚exive af¬x fo-
cuses on the quantity of the subject™s participation. The productive formations in
yf- . . . -cz -- such as yfujdjh«nmcz ˜speak much, to one™s heart™s content™, yf†cnmcz
´csnf) ˜eat one™s ¬ll™, yfjnls[’nmcz ˜rest fully™ -- assert that the quantity of the
subject™s participation reaches some limit of satisfaction or suf¬ciency:
[292] Cnjkmrj jy gskb yfukjnfkcz yjxm/!
How much dust he swallowed during the night!

The formation in hfp- . . . -cz means intense involvement by the subject exceeding
a norm ([293]), lj- . . . -cz activity that leads up to a boundary signaling a change
in the world ([294]):
[293] Ghb xe;jv xtkjdtrt jyf hfpsuhfkfcm, hfpdtctkbkfcm, hfpujdjhbkfcm.
In the presence of a stranger, she became thoroughly playful, merry, talkative.
[294] Hjlvfy jgznm ljbuhfkcz, edjkty bp rke,f ¤Lfkkfc Vfdthbrc≥.
Rodman again has fooled around [to that point that he has been] released from
the Dallas Mavericks.

Intransitivized reflexives: The most productive function of the re¬‚exive af¬x is
to make intransitive verbs from transitive verbs. A transitive, by de¬nition, has
a subject (an argument responsible for change or imbalance) distinct from an
object (an aspectual argument whose change or imbalance is reported). Detran-
sitivized re¬‚exive verbs formed from transitives lack a distinction of responsible
and aspectual arguments. They report a change that befalls the sole participant:
ghtrhfn«nmcz ˜cease™, jndk†xmcz ˜be distracted™, yfx’nmcz ˜begin™, cjdvtcn«nmcz
˜coexist, overlap™, bpvty«nmcz ˜change™, juhfy«xbnmcz ˜be restricted™, ek©xibnmcz
˜get better™, cj[hfy«nmcz ˜be preserved™. The interpretation of individual verbs
varies depending on whether the sole argument is inanimate, in which case
the event is spontaneous ([295]), or animate, in which case the event can be

58 Gerritsen 1990:153--60.
348 A Reference Grammar of Russian

understood as instigated by the subject ([296]:
[295] Yfcnegbn vbhjdfz htdjk/wbz, b dct itltdhs dyjdm r yfv dthyencz.
Once the world revolution arrives, all the masterpieces will return to us.
[296] D rjywt fduecnf vs dthyekbcm d Vjcrde.
At the end of August we returned to Moscow.

Reflexive passives: When, for a given verb, the roles of modal and aspectual
argument cannot be understood as merged and the change is induced externally,
the re¬‚exive intransitive verb might be called “passive.” The passive sense of
imperfective re¬‚exive verbs presents an activity as a generic situation that is the
property of the aspectual argument:
[297] E vyjub[ bpdtcnys[ vyt gjkbnpfrk/xtyys[ vyjubt vtczws gjlhzl
rjyabcrjdsdfkbcm dct gbcmvf.
Among many of the political prisoners I knew, all letters used to be con¬scated
for many months at a time.

In [297], the re¬‚exive passive presents a static fact as a property of the
letters -- they persist in the state of inaccessibility; using the third plural
(rjyabcrjdsdfkb) here would focus on the active participation of the unnamed
agents. Exceptionally, the responsible party is actually expressed in the instru-
mental case ([298]):
[298] Эnb cbufhs rehbkbcm vjbv lzltq<ins> dc/ ;bpym.
These cigars used to be [= were of the type that were] smoked by my uncle all his

The passive use is related to the detransitivizing function of re¬‚exive predi-
cates mentioned above, but differs in certain respects. The detransitivized re-
¬‚exives presume that change can occur spontaneously without an external
agent, and they are formed from both aspects: cjdvtcn«nmcz/cjdvto’nmcz ˜be
compatible with™, bpvty«nmcz/bpvtyz ´nmcz ˜change™, juhfy«xbnmcz/juhfy«xbdfnmcz
˜be limited™. The passive sense presumes that change would not occur without
an external agent (in [299], the room needs an agent to instigate ventilation),
and it is only formed from imperfectives.
[299] Tckb d gjvtotybb vyjuj k/ltq b jyj lkbntkmyj yt ghjdtnhbdftncz,
cjlth;fybt d djple[t eufhyjuj ufpf edtkbxbdftncz.
If there are many people in a dwelling, and it fails to be aired out for a long time,
the content of carbon dioxide in the air increases.

Since Fortunatov (1899), there has been an impulse to see a unity in the
overall group of re¬‚exive verbs: they have reduced valence.59 While there is
59 For a notational account, see Babby 1975[b]; for a semantic account, see Schenker 1986.
Predicates and arguments 349

some uniformity, there is considerable lexical diversity among re¬‚exive verbs.
There are recognizable groups that have distinct meanings and properties and
different degrees of productivity (compare the productive intransitives such as
´bnmcz/ecgjr’bdfnmcz ˜calm down™ with the residual true re¬‚exives vsnmcz
ecgjrj ´
or the restricted re¬‚exive of proclivity cj,’rf rec’tncz).

5.8.4 Present passive participles
A present-tense passive participle is formed from the present stem of imperfec-
tive transitive verbs with the formant {-m-}, to which are added adjectival end-
ings. These passives in {-m-} can be used as attributive modi¬ers (,jkmibycndj
ghjbpdtltybq, njulf ,tpelth;yj hfc[dfkbdftvs[ ˜the majority of works, praised
at that time without restraint™) or as predicatives, as in [300], which conjoins a
present participle with a past passive participle:

[300] {hjvjcjvs vjuen ,snm xtnrj bylbdblefkbpbhjdfys b jgjpyfdftvs<prs psv> c
gjvjom/ ljcnfnjxyj ghjcns[ vtnjljd.
Chromosomes can be clearly differentiated and recognized through rather simple

This construction belongs to technical, journalistic, or bureaucratic styles.
The suf¬x {-m-} forms derivatives that are used as ordinary adjectives, with
a modal connotation of possibility or proclivity to be involved in an activ-
ity: ytevjk«vsq ˜implacable™ describes an individual who cannot be molli¬ed;
thus also k/,«vsq ˜beloved™, ytdjj,hfp«vsq ˜unimaginable™, ytbcghfd«vsq
(jgnbv«cn) ˜incorrigible (optimist)™, ytekjd«vsq ˜ineluctable™, ytpf,sd’tvsq
˜unforgettable™. These derivatives can be formed from perfectives as well as imper-
fectives: ytdjpven«vsq ˜imperturbable™ from djpven«nm<pf> ˜bother, perturb™.

5.8.5 Past passive participles
Past passive participles, or simply passive participles, are formed from the past-
in¬nitive stem of perfective verbs by means of a formant ({-n-} or {-ón ∼ ¬n} or
{-t-}), to which adjectival endings are added. As a verbal adjective, the passive
participle can be used in the positions in which one expects to ¬nd adjectives:
as a preposed attributive modi¬er ([301]), as a postposed modi¬er ([302]), or as a
predicative in a transitive construction ([303]):

[301] <. . .> d cnhjuj htukfvtynbhjdfyyjv<psv> j,(tvt
<. . .> in a strictly regimented volume
[302] <. . .> cj cdjbvb ldevz ,hfnmzvb, nj;t jce;ltyysvb<psv> gj njve ;t ltke
<. . .> with his two brothers, who were also sentenced in the same matter
[303] Gj gthhjye tuj ghjdtkb crjdfyysv<psv> yfhexybrfvb c j[hfyybrjv.
They led him along the platform bound with handcuffs with a guard.
350 A Reference Grammar of Russian

But above all, the passive participle is used as a predicative with be. The participle
is the predicating element, and it agrees with the subject (ry«uf in [304]):

[304] Rybuf<\fem sg> ¤Hswfhb Rheukjuj cnjkf≥ jcnfdktyf<psv d Vjcrdt.
fem sg>
The book Knights of the Round Table got left in Moscow.

As in other predicative constructions (§5.2), the participial passive distinguishes
three tenses; no overt forms of be are used in the present. Because perfectives
have only two tenses but predicative constructions distinguish three, the par-
ticipial passive cannot be derived simply by transforming an active sentence
structure into a passive one.60
In some respects the predicative construction with a passive participle is not
entirely analogous to a predicative construction with an adjective. The sole ar-
gument of a passive participle can be affected by quantifying operations, such
as the genitive of negation ([305]) or approximate quanti¬cation ([306]):61

[305] Yfv j,(zdbkb, xnj ybrfrb[ gjktds[ yfuhepjr<gen> yt gjkj;tyj<psv> .
We were informed that no extra pay for being in the ¬eld had been established.
[306] <skj hfccnhtkzyj<psv> xtkjdtr cjhjr.
There were shot some forty people or so.

Such constructions with adjectives are inconceivable (— ybrfrb[ gjktds[ yfuhepjr
yt dscjrj [as if: ˜no ¬eld salary was high™]; — ,skj ;bdj b pljhjdj xtkjdtr cjhjr
[as if: ˜were hale and hearty some forty people™]. Unlike predicative adjectives,
then, passive participles tolerate quanti¬cation, and the subject of a passive is
not necessarily individuated.
When is the predicative construction with a passive participle used? Passive
participles describe states that characteristically result from prior perfective
events. In context, the idea that a speci¬c event is the source of the current
state can be more or less prominent. Often, in descriptions of scenes, no spe-
ci¬c event is understood to be the source for the state, though the subject has a
property that characteristically arises from an event ([307]). Sometimes, however,
the state can be understood to derive from a speci¬c event ([308]):

[307] Ptvkz, djple[, vtczw, pdtpls crjdfys<psv> dvtcnt, crktgfys<psv> vjhjpjv . . .
The earth, air, moon are forged together, riveted by frost . . .
[308] Dct dsikb ghjdjlbnm tuj d ghb[j;e/, gjwtkjdfkbcm. Jy eitk. Yf lheujq
ltym C. Rjdfktd ,sk fhtcnjdfy<psv> .
We all accompanied him out, took leave. He left. On the next day S. Kovalev was

60 See Babby and Brecht 1975, further Brecht and Levine 1984, Babby and Franks 1998.
61 Lobanova 1975.
Predicates and arguments 351

Table 5.14 Properties of passives and near-passives

re¬‚exive imperfective perfective passive non-speci¬ed third
passive participle plural
(rjyabcrjdsdfkbcm, (edjktys, hfpjhdfy) (c[dfnbkb, edjpbkb,
ghjdtnhbdftncz) uhjpbkb, pf ytq
expression of agent extremely rare (3%) rare (9%) ---
view of event repeated activity event presented as responsible agent
presented as stative stative resultative asserted to exist,
property of patient/ property of but remains
subject patient/subject unnamed

Common to all uses of the passive participle is that a potential event is pre-
sented as a static property of the subject. Accordingly, the passive participle is
used in description or summary rather than narrative (as in [309]). The unspe-
ci¬c third plural, in contrast, presents a pure event -- in [310], one event in a
narrative series.

[309] <jktt gjkjdbys b[ jrfpfkbcm yfheibntkzvb lbcwbgkbys. Vyjubt bp yb[
edjktys<psv> pf ghjueks.
More than half of them turned out to be violators of discipline. Many of them
were removed for absenteeism.
[310] Pfvtifyyst d “ltkj” dspsdfkbcm yf ljghjcs, bv euhj;fkb. Jlyjuj bp yb[
ghbcelbkb r ghbyelbntkmysv hf,jnfv. Ytcrjkmrb[ edjkbkb<pst pl> .
Those who were mixed up in the “affair” were called in for interrogation, they
threatened them. They sentenced one of them to forced labor. They removed
several from work.

5.8.6 Passives and near-passives
Use of passive and near-passive constructions is summarized in Table 5.14.

5.9 Agreement

5.9.1 Basics
Finite verbs agree with their subjects.62 Verbs express gender--number in past--
tense forms and person--number in present-tense forms (including the present-
tense forms of perfective verbs). In [311], the past-tense verb is feminine, in
agreement with the overt subject, a feminine singular noun.

62 See Crockett 1976, Corbett 1979[a], Corbett 1983[a], 1983[b], 1988[b], Robblee 1993[b].
352 A Reference Grammar of Russian

[311] <f,eirf ?kz<\fem sg> et[fkf<fem d Gfhb;.
Grandma Iulia left for Paris.

If the subject is not expressed as an overt argument phrase, the predicate ex-
presses the features of the implicit referent of the subject, as in the continuation
of [311] as [312]:

[312] <. . .> b ghj;bkf<fem pst> nfv ,jktt ltcznb ktn.
<. . .> and lived there for more than ten years.

If the sentence is impersonal -- that is, lacks the possibility of a subject -- the
verb appears in the neuter in the past ([313]) or in the third-person singular in
the present ([314]) and future:

[313] Vtyz njiybkj<nt pst> , ljvjq tkt lj,hfkcz.
Nausea overcame me, I hardly made it home.
[314] Vtyz njiybn<3sg prs> , ljvjq yt lj,thecm.
I™m overcome by nausea, I won™t make it home.

5.9.2 Agreement with implicit arguments, complications
Agreement is largely without problems in Russian, but there are some contexts
of interest.

Collectives: Subjects that are nouns cause little uncertainty. Even singular nouns
with collective sense elicit singular number, with the gender appropriate for the
noun (fem sg for rjv’ylf in [315]):

[315] Rjvfylf<\fem sg> ,skf<fem> ljcnfnjxyj cbkmyf<fem b [jhjij
gjlujnjdktyf<fem sg> .
The team was quite strong and well-prepared.

The subsequent context maintains the singular number (gjlujnj ´dktyf) unless
the individual members of the collective are explicitly named (for example,
buhjr« ˜players™). Plural is not used in Russian, unlike in varieties of English:
The Barcelona player said his side were not prepared for the vociferous support given by
Korean fans.

Implicit gender of personal pronouns: When the subject is a ¬rst or second
person, a past-tense verb re¬‚ects the gender of the referent of the pronoun,
though the pronouns themselves do not distinguish gender:

[316] Ns yt vj;tim ct,t ghtlcnfdbnm, rfre/ [jhjie/ dtcnm ns vyt gthtlfkf<fem> .
You cannot imagine what good news you™ve given me.
Predicates and arguments 353

Universal second-person singular: Second-person singular agreement in the
predicate is used without any overt subject noun phrase in a universal sense
of any possible addressee:

[317] Gjckt djqys z epyfk, xnj ,sk nfv ecnhjty kfuthm lkz pfrk/xtyys[. Lf, vtcnj
ds,hfkb gjl[jlzott -- ukeim, ,tpljhj;mt, rheujv ktcf b ,jkjnf. Yt
e,t;bim<2sg> .
After the war I learned that a prison camp had been built there. Yes, they picked
a good place -- wilderness, no roads, nothing but forests and swamps all around.
You won™t escape.

Formal second-person plural: The second-person plural pronoun ds is used in
formal address to a single individual (the formal “B-form”: §4.6). In agreement
with formal ds, verbs are second-person plural, and predicative (“short”) adjec-
tives are plural ([318]):

[318] -- Bhbyf, ds vjkjls<pl> , rhfcbds<pl> , vyjuj buhftnt<2pl> b, yfdthyjt,
joeoftnt<2pl> ct,z ljcnfnjxyj rjvajhnyj.
-- Irina, you are young, beautiful, you get many roles, and no doubt you live rather

Used as predicatives, long-form adjectives re¬‚ect the referential gender and num-
ber of the subject -- singular when a single person is addressed by formal B, even
as the verb is plural:63

[319] Gjxtve ;t ds ,skb<pl pst> {nfrjq uhecnysq<msc sg> ∼ nfrfz uhecnyfz<fem sg> }?
Why indeed were you such a sad one?
[320] Ds {gthdsq<msc sg> ∼ gthdfz<fem sg> } yfxfkb<pl pst> .
You started it ¬rst.

Other predicatives, such as g†hdsq ˜¬rst™ ([320]), are likewise singular if the ad-
dressee is singular ([320]). If the addressee is plural, predicatives are plural: ds
gtdhst<pl> yfxfkb ˜you (all) started it ¬rst™.

5.9.3 Agreement with overt arguments: special contexts
With overt subject arguments, agreement is unproblematic most of the time:
singular agreement is used with a singular noun, plural with plural. Complica-
tions arise in three contexts. All three contexts have subject phrases that could
be understood as referring to multiple entities. Though there are differences
among the three contexts, there are also general principles that apply to all

63 Comrie 1975:408, 410.
354 A Reference Grammar of Russian

Table 5.15 Agreement of conjoined nouns,
animacy, and word order

SV order (%) VS order (%)

animates 100 84
inanimates 85 28

three. In general, singular agreement in these contexts indicates that the group
of elements is understood as a whole, as a unit, and the fact of the existence of
some number of things is more signi¬cant than the activities of the individuals
involved. Plural agreement means that the elements that make up the group are
viewed as potentially distinct individuals. At the level of the argument phrase,
animates are more likely to be individuated, and occur with plural, than inan-
imates; abstract nouns are unlikely to be individuated, and unlikely to trigger
plural agreement. At the level of the predicate, since existential predicates are
interested in the fact of existence, they are more likely to occur with singular
agreement than individuating predicates such as transitive verbs or predicative
constructions. At the level of discourse, the word order in which the subject is
postposed is the order used for establishing the existence of a situation. Accord-
ingly, a verb that precedes its subject is more likely to have singular agreement
than a verb that follows its subject.
The three speci¬c contexts are the following.

5.9.4 Agreement with conjoined nouns
With noun phrases composed of two or more conjoined singular nouns, the verb
can appear in either the singular or the plural. (If any of the conjoined nouns is
plural, agreement is plural.) As can be seen from Table 5.15,64 plural agreement
is preferred with nouns that refer to animates. Also, plural agreement is usual
when the subject precedes rather than follows the verb, as is especially visible
with conjoined nouns referring to inanimates.
When variation is possible, plural is appropriate when the conjoined elements
are distinct entities, and the predicate is independently valid for each. Entities
can be distinct if they are different kinds of things ([321]) or two distinct indi-
viduals of one type ([322]):

[321] Gjhfpbkb<pl> tuj rhfcjnf<\fem sg> bp, b pfntqkbdjcnm<\fem sg> htpm,s.
The charm of the huts and the intricacy of the carving amazed him.

64 Corbett 1983[b]:181. For conjoined nouns, Corbett 2000:207 cites 67 percent plural agreement in
literature, 96 percent in press (there without differentiating animacy or word order).
Predicates and arguments 355

[322] Tt cj,cndtyyfz ;bpym<\fem sg> b ;bpym<\fem sg> dtrf crkflsdfkbcm<pl> nfrbv
j,hfpjv, <. . .>
Her own life and the life of the era had taken shape such that <. . .>

Singular agreement assumes that the elements are not distinct and are intrin-
sically associated; the predicate applies to all the elements together. The entities
can amount to a higher order abstraction -- life and reason together de¬ne a
universe ([323]):

[323] Dyenhb rf;ljuj эktrnhjyf cghznfyf dctktyyfz, ult
ceotcndetn<3sg> ;bpym<\fem sg> b hfpev<\msc sg> , rfr d yfitq.
In each electron is hidden a universe, where there exists life and reason just as in

Or one element can be understood as a concomitant of the other:

[324] <. . .> jcnfkfcm<fem> ;tyf<\fem sg> b vfkmxbr<\msc sg> .
[After he died] there remained his wife and boy.

Or two abstract ideas are (nearly) synonymous:

[325] Yfcnegbkj<nt> ecgjrjtybt<\ntsg> b leitdyfz nb[jcnm<\fem sg> .
There came a calm and spiritual quiet.

When singular agreement is used in the past tense, thereby expressing gender,


. 12
( 17)