BY ILI~jI~ SEMANTIC CL~k%</IFICATtONS IN TIE LEXI(XIN 
Eler~ V.P/ff)USTIEYA and Ekaterina VoRAKHILINA 
Institute of" ~-' c, ooientifll~ 
and Technical Inf'ormation, 
Academy of Sciences of' the USSR 
\[~ievicha 20a 125315 Moscow 
Abstract 
In this paper we investigate 
general principles of 
constructing semantic 
classifications that yield 
useful predictions 
combinatory options 
Several semantic 
Russian words are 
concernin~ 
of words. 
classes of 
discussed, 
implemented in an expert system 
named "Lex i co~raphe r", the 
"Lexicographer" is supposed to 
provide its users with all kind 
of information concerning some 
15.000 most common Russian 
words. Alon~ with morphological, 
syntactic 
infor.~tion 
conventional 
system 
information 
and semantic 
usually stored in 
dictionaries, the 
should contain 
about referential 
characteristics of words and 
about restraints in 
combinability with other words 
in syntactic constructions of 
different types. In its final 
version "Lexicographer" should 
provide the users with all sorts 
of bibliographical information 
system beinff conceived as an aid 
both in the area of natural 
language processin~ arid in 
traditional lexicography. 
Semantic Features proposed 
regulate co-occurence of verbs 
with their non-obli?atory 
dependents - such as Modifiers 
of place or time; Instrumental 
and Benefactive objects and the 
like. 
( concern i n~ both i nd i v i dual 
words and semantic classes of 
words) and with concordances 
made on the bas i s of a 
suffi c i ent ly re presentat i ve 
corpus of Russian texts. 
One of the basic components of 
the system is its lexicon; the 
lexicon contains information not 
only about individual lexer~s, 
but also about sen~ntic and 
syntactic classes of lexemes. 
Thus, for nominal lexe~es such 
features are ~iven as: "NATURAL 
CLASS", "ARTEFACT", "MASS TERM", 
"SET", "BODY PART" and the like. 
For predicates the lexicon 
i yes the values of such 
I 2.31 
features as Vendlerian classes 
(STATE, ACT IV ITY, 
ACCOMPL I SHMENT, ACH I EVEIvENT), 
of. \[I\]; 
NONCONTROLLED 
possibility 
subordinate 
\[g\], etc. 
It is argued that 
cases combinability of 
CONTROLLED vs. 
action \[2, p. 32\]; 
off having a 
indirect question 
in many 
a. word 
can be predicted on the basis off 
the semantic class(es) it 
belongs to. Thus, semar~tic 
classifications will be the main 
topic off our exposition. 
It turned out that altogether 
different approaches ar'e 
required for semantic 
classifications off predicates 
and For lexemes with non 
predicative neaning. In this 
paper we give central attention 
to the predicative zone off the 
lexicon. 
Semantics of verbs has been 
studied intensely in recent 
year's, of. works by Ch. Fillmore 
\[3\], Z. Vendler \[i\], L. Karttunen 
\[3\]. it was a long tradition for 
linguists to lay the strongest 
emphasis on the 
se,emtic properties 
verbs. Our ~oal 
will be to attract 
individual 
off separate 
in this paper 
attention to 
those semantic properties of 
verbs that affect their 
combinability and still are 
common to more or less numerous 
verb classes. 
Thus, combinability of verbs 
with subordinated noun phrases 
or adverbials designating place, 
time~ re&~on~ purpose~ 
Instrun~nt and other parameters 
of the situation denoted by a 
predicate (taken together with 
its arguments) was usually. 
assumed to be non-restricted. 
This is definitely not the case, 
and the problem deserves special 
attention. There are some well 
known important observations 
made in this area. Thus, as is 
known, one of the Vendlerian 
classes - accomplishment verbs - 
is singled out by the fact that 
some verbs combine with a 
special kind of time adverbials, 
such as in two hours; these 
adverbials designate a period of 
ti~re dur'in~ which 
went on and finally 
inFerent limit. 
In what follows 
the process 
reached its 
we shall 
demonstrate several sen~nt ic 
cl&sses of Russian verbs with 
the correspond i ng rules of 
sen~nt i c co- ocourence 
restraints. We ar,~ue that the 
restr.ai nts d i soussed are 
semantic in their nature; thus, 
it is natural to suppose, that 
some analogous or. co~nate 
r-estrai nts would hold For 
corresponding sel~ntic classes 
of verbs in other" languages. CF. 
the look by A. Wierzhbioka \[8\] 
where the author strives for 
selm%ntic explar,ations in ~r,a~zr 
in a broad sense of the word, 
namely, tries to find semantic 
mot i vat i on for every 
co-occurrence restriction in the 
cho ice of morphemes or 
grammat i cal construct ions. 
1. There ar'e classes of 
J.ooatiorl JYl tJ.rr~ riOT' ih :gp~-~CO i:~ 
boss i b 1 e. 
One off such c\].&sses is 
c;onstituted by the so-called 
we i~ht", B;tIa~egb "to possess", 
~HaTb • 1:0 know • , Hafffe,ffTZ-,c~z ' to 
hope • , nO~}{~Tb • to rerrember" 
etc. (Stab:re states are opposed 
t:.o the so--cal led temporary 
states of. such pr.edicates 
~eoea~TbC~ • to make Fun • , Ob~Tb 
E.oao~,¢ • to be hungry • , rmaB~ 
• drunk • etc for which 
localization i.n time is 
possible. Another" cl,._~s with the 
s,m~e combinatorial properties is 
constituted by causative verbs 
denoting" situations that ~ive 
states a~s their, results, cf. pa- 
~O~aT~ "to cause j oy • , o~opqaTb 
• to grieve", pe~oMeH~oP, aTb "to 
reoo~t~llend" , 8St'PITh " to l:rk:%ke 
an\[ry" etc. Note especial ly 
verbs denot ins changes in social 
state such as OS;flOBeTb : to 
become, a widow'. 
Verbs of these semm\]t ic 
cl~¢sses do not a\].low of place or' 
time adver'bials (of all tirre 
adverbials only those cooccur, 
with such verbs that denote 
super-lon~ intervals, such as 
pa~e "before", B MOJIO~\[OOTH " in 
the youth", l{oa~a-TO " long 
a~o', cf. \[5\]). 
In order to 
explanation 
restraints of 
~ive semantic 
oF '' ~ '~ CO-OCOUr~.r~c~ 
a verb it is 
sometin~s necessary to use a 
more exact classification of 
time modifier's. Thus, the ve~'b 
...o o~uUI With ~fl°l J.dv~t bJ.~il 
expr'essing exact t ir~ (~OH oi~o~J.- 
Z#a;/i ~ 14. ~:~0), \[hough for other, 
Perfect ive verbs in Russian such 
trod i Fi c,~at- i on usual 1V. i s not 
precluded (0~ npza~e~ .B I4.30), 
o~t. different iat ion oF exact 
t i ~e arid e~r~rac i n%" t i ~. i h \[ 5\] o 
T~ L. ver'b OFiOS/~aTb i ;_-, hot 
unique in its serr~\]tic c, lass, 
cF. su.~-" c' - verbs as- yc:neTb. • to be 
in ti,e •, OTCTaT~, aaTs~yrrbc~ 
( 21o~at{ 8aTSHyaOzO, aa,lfep~aTbC,~, 
COXpSH~fTbOS arid so\[0e other's. 
~.°° There is aJ-~ interest_ in~ 
dist.inction that has r, et~ined 
unnoticed up till now, nm~ely, a 
distir~t.ion'r' - between c:oncr.~.t~._ .. _ and 
non-concrete verbs. A verb is 
¢x~r~,-ete if it designates (in a 
proper context) a sit.uation that 
is supposed to go throufi'h a 
series of phases in its natural 
development each of which can be 
fi'iven a detailed descr~p.ion. 
Indeed, for a situation 
described by a verb de;waTb •to 
run • or, peaaTz~ •to cut" (both of 
them concrete) it is always 
possible to say, for every fixed 
~on~nt, what the subject 
involved in the situation is 
doing at this moment. On the 
other hand, there are verbs that 
p~esent the situation in such a 
way that they emphasize the 
final result of the action 
described while the rrea~s used 
in order to achieve this result 
remain in the shadow; and tt~is 
semantic feature characteristic 
of" their use ~'~d understanding" 
233 
must be reflected in the ir 
lexicographic representation. We 
call such verbs r~-~rete. 
Thus, OTOMOTMTb means " to 
reven~e oneself for" an evil 
committed eaz'lier" ; but the 
means used in order to attain 
this goal are irrelevant. As a 
consequence, no concrete phases 
can be discerned in the action 
denoted by this verb: the verb 
OTO~CT~Crb does not express "an 
unfolding a~t ion", even when 
used in the Imperfective. The 
same is true for the verb ncnop- 
T~ "to spoil", which is also 
non-concrete: ~c no pTzT~ means 
"to deprive of the possibility 
of normal functioning'; but the 
way in which this deprivation is 
fulfilled is irrelevant for the 
meartin~ of the wor'd~ so it+ has 
no phases, it does not denote a 
developin\[ process. 
Non-concrete verbs very often 
are aspectual ly defective; 
namely, they either have no 
Imperfective or, bein~ used in 
the Imperfect ive, they lack 
Progressive meaning. Thus, amon~ 
the verbs that, accordin~ to Ju. 
Apresjan \[7\] belong to the class 
of the so-called momentary verbs 
(achievements in Vendler' s 
terminology) and thus lack a 
Pro~r.essive meaninff there is a 
major group of verbs that owe 
the i r momentar i hess to non 
-concreteness; cf. noSe;~TaT~ "to 
win", "to succeed' ; aaOTM8aTb, 
8aOTaBaTb ; PIBMeHSTb < po1\[PIHe> ; 
sapywaTh rlpaBZgIO "to violate the 
r'ule" ; o~xoT~fPb 8al~OH " to 
234 4 
violate f"+~. ...... + _ +t t.. J.,LI, W , H~+t~+~6ti+,~Tb < TO- 
S,.'<D+~e~b t,o fai \], ; paas+agm2v~ 
<Tc~I,~Hy> ; C,rie/qOSaTb <OOB6~TX> " \[.O 
fo\]. low the adv ice" :, Co}epWaTL, 
spex "to sin', et.c. 
(1) a. Oe~a~ c .d?a~e~oM "r'm~ 
with a tor'ch <in his 
herod> ; 
b. *MOT~Ia c nOpT(.lSeaeM <B 
py~_.> "r'eveng'ed hi rr~e 1F 
bag" < in his with a 
r tal l,.J ;- 
( 2\] a. F;~oOt4JI qaObI MOJIOTP~DM 
"broke the watc, h with 
the hammer" ; y~ap~a 
naam-J~ "hit. with a 
st t,.k ; 
.~icrlop'z!4Ji <qaObl> MO~OTI-ff~M 
"spoiled the wa,tc-h with a 
ha~rer'" ; ~ocmppdH3/ nad~o# 
"reveng'ed himself with a 
st. i ok". 
3. Not every predicate can c,jo- 
occur' with such a c,,ot~mo~t- - tvoe. of 
verb-complement as Benefac-tive 
(expressed in Russian with the 
help of the proposition ~\[Jzs 
"for'). Thus, it is impossible 
to say ~dO~TbO~ ,~dZS I<O~O-Jn4dO 
"to be aZr.aid for' stub,' *'Sa~CTbc~-' 
smb ~.rl~ l-eoso-JII4~o 't.o swe~" for' - ' 
etc. These restrictions are 
easily deducibly from serremtic 
decompos i t ion of the Benefact ive 
construction: X P ~a~ Y = "X did 
P having" in mind to provide Y 
with the result of P'. It, 
follows from this decomposition 
that a predicate for which this 
construction is possible must 
denote art action that is both a 
contr, olled one and ar~ action 
havin~" some_, definite result. 
Th i s i s ~hy Uene fach i re 
complement cannot co-occur with 
the verb dOaTbCa, de not i n~ 
non-control led state, or with 
the verb ~wi~cTbca, denotin\[ an 
action with no result. 
It is a well-known fan~t that 
verbs denotin~ non-controlled 
act ions normally do not co-occur 
with a modifier of purpose; 
thus, such quest ion m-~ *Sa~e~ 
~\]~OT c~ea? ~a~eM Mope ~,y~}fT? 
~e deviant (questions about the 
cause of such states are normal: 
flov, e~y H\]IeT cries? l-lo~e~v Mope 
Thus, if the diction~'y 
contains informat ion about the 
feature CONTROL, then the 
behavior of a verb with :respech 
to a modifier of" purpose becomes 
pred i ctabl e. 
(3) *flO~eMy OH a~aem? 
.Ik)qe~y OH do~eeT? 
.Ilo~e~y OH MOgm~O~ 
(4) Ilor~eMy Tb! CTOMmD ? llo\[\[eMy 
Tbl MOJtql,IIff~ ? 
Predicates in (3) denote states 
that are cal led determinate i n 
\[6\]: these states develop in 
t i 1~ accord ins" to the i r owr~ 
laws~ ~-~cl it. is impossible f'or 
the subjecst to chan~e his state 
of his own accord, And this 
se~nt.ic feature "" " expia~ ns t.he 
d ifferenc, e between (S) awl (4), 
Summary 
It is customary to think that 
co-ooourence possibilities are 
determined only by individual 
properties of words: 
dependencies that do not 
correspond to predicate-argument 
relations are not semantically 
~tivated, Our research shows 
that. there are no syntactic 
relationships without semantic 
back~roundin~, though serr~%ic 
features of words that r,e~utate 
co-occurence ew, e not always easy 
to discover,. 

References

1,. Vendler Z. Verbs and times. - 
Philosophical reviews, 1957, 
v. 64, p. 143-160. 

2. Dik S. Functional ~rammar. 
-Amsterdam etc: NorthHolland P. 
C., 1978. 

Karttunen L. 
constructions 
complement 
Langu~es, 
80. 
La lo~ique des 
an~laises a 
predicatif. - 
1973, N 30, p. 86- 

4. Fi 1 lmore Ch. How to know 
whether you' re comi rL8 or 
~_oins:, - In: Essays on deixis, 
/ Ed. by Rauh G, Tubins-en: 
Naz, r, 1983. - p. 219-£27. 

Wierzbicka A. The sel~nt.ics 
of ~rammar: A~ster'dam, etc. : 
Jho~n Senjamins, 1988. 

IIa~ty~eBa E, B, I~ aeMaHTH~eoKo~,\] 
F~acc~Ka~H B!DeMeHHb~X ~eTep- 
MHHaHTOB rlpe£Jlo~tieHH£, - B I<H, ; 
HANK: CHCTeMa I~ ~yHKI~HOHHpoBa- 
HHe. 16: Hay~a, 1988. 

7. Ha~y~eBa E.B. ~mO~Ma ~H 
OnOOOOHOCTB HO~HaTB KOCBeH- 
HBlfl Bo~poc H3 ceMaHTHEH O~oBa~ 
- B KH.: 8HaH~e H MHeH~e. ~: 
IIayKa, 1988, c. 33-46, 

8. Pax~Ha E.B. Ke~ ~ np~HHa. 
P~npoeH ~S~KOSHaHHa, 1.989, 
N6. 

9. AnPec~H ~ ~. P~aro~bl MOMeH- 
Ta~HOrO ~e~CTB~ R nep@opMa- 
THB~ B pyccKoM aa~e. - B KH.: 
Pyc~cTH~a cero~H~. ~: HayKa, 
1988
