Stephen M. Dickey 
University of Kansas 
S-12- and the Grammaticalization of Aspect in Slavic 
Avtor zagovarja stalibte, da  je v zahodnih skupini slovanskih jezikov (tebtina, slovabtina, luiibti- 
na, slovenbtina) sovpad odrazov * s ~ -  in *j1,z- v eno predpono povzrotil (delno) gramatikalizacijo 
nastale predpone s - Iz  kot osnovni preverbe vide v sistemu glagolskega vida. V drugih slovanskih 
jezikih do te gramatikalizacije ni priglo (hrvabtina, srbbtina) ali pa je bila predpona po- gramati- 
kalizirana kot edini prkverbe vide (rubtina, bolgarbtina, makedonbtina, ukrajinbtina, belorubtina). 
V enem slovanskem jeziku (poljgtina) izkazujeta predponi s-Iz- in po- skoraj enako produktivnost. 
It is argued that the phonetic coalescence of *s-6 and * j b z  into a single prefix resulted in the 
(partial) grammaticalization of innovative s-Iz- as the primary prkverbe vide of the aspectual sys- 
tems in a group of western languages (Czech, Slovak, Sorbian, Slovene). The other Slavic lan- 
guages either did not grammaticalize a single prefix (CroatianISerbian) or have grammaticalized 
p o  as their sole or primary prkverbe vide (Russian, Bulgarian, Macedonian, Ukrainian, Belarusi- 
an); one Slavic language has near equal productivity of s-Iz- and po- (Polish). 
0. Introduction 
This paper examines verbal prefixation with s-12- in the Slavic languages, focusing 
primarily on West Slavic and Slovene, where this prefix has been the most product- 
ive.* A description of prefixation with s-lz- is presented, and it is shown that the 
languages in which s-lz- has the highest productivity as a perfectivizing prefix cor- 
respond largely to the languages belonging to a "western" aspectual type according 
to Dickey (2000). The development of s-lz- from a coalescence of the prefixes *sa-  
and *jhz- (as well as *v ,~z -  to varying extents) in these languages is discussed, and 
on the basis of the historical development of this prefix it is suggested that the sem- 
antic nature of s-lz- has played a significant role in the particular development of 
the aspect category in the languages making up the western aspectual type. Aspect 
in the other Slavic languages (most of South Slavic as well as East Slavic) is in con- 
trast argued to have been shaped by other developments, notable among them pro- 
ductive prefixation with po-. 
1. Preliminaries 
The analysis presented here is based on the overall east-west division in Slavic as- 
pect established in Dickey (2000) on the basis of data from Cz, Slk, Sln, Pol, 
*This  article was supported in part by a Fulbright-Hays Faculty Research Abroad Award re- 
ceived from the U.S. Department of Education, as well as a New Faculty General Research Fund 
Grant awarded by the University of Kansas. This support is acknowledged with gratitude. I would 
also like to thank Marc L. Greenberg, who provided me with comments that greatly improved the 
presentation of Slovene and South Slavic. Finally, I am grateful for the efforts of two anonymous 
reviewers that helped improve various aspects of this article. All shortcomings and inaccuracies, of 
course, are mine alone. 
4 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
SrbICro, Blg, Rus and Ukrl concerning variations in aspectual usage in habitual 
contexts, the general-factual, the historical present, the impf in sequences of events, 
performatives and other cases of coincidence, as well as differences in the derivation 
of verbal nouns. The result is an overall division of Slavic into two distinct aspectual 
types: an eastern type (Rus, Ukr, Blr, Blg) and a western type (Cz, Slk, Sln). Pol 
and CroISrb are transitional zones between these two groups; however, Pol tends to 
pattern more like the east and CroISrb more like the west. On the basis of the ob- 
served differences, Dickey (2000) constructs a theory of the meanings of the pf and 
impf aspects in each group (for convenience referred to here as the east-west as- 
pect theory), according to which the meaning of the pf aspect in the western 
group is TOTALITY, whereas the meaning of the pf in the eastern group is a con- 
cept labeled TEMPORAL DEFINITENESS. Totality, which is familiar from the as- 
pectological literature (e.g., Comrie 1976), refers to the construal of a situation as an 
indivisible whole in time, including its beginning, middle and end; the obligatory fo- 
cus on the endpoint of the situation is ordinarily due to the spatial(-temporal) pro- 
file2 of the prefix of a perfective verb, which includes a limiting endpoint as a 
landmark. In a temporal sense, the endpoint marks the transition to a new state of 
affairs, and this is why pf verbs are so closely associated with the transition from 
an initial state to a new state (i.e., the TRANSITION FROM S1 TO S2; see the dis- 
cussion of Shull's 2003 theory of prefixation below). Temporal definiteness is a dif- 
ferent kind of concept and requires some explanation. A situation is temporally def- 
inite if it is unique in the temporal fact structure of a discourse, i.e., if it is viewed 
as both (a) a complete whole and (b) qualitatively different from prior andlor sub- 
sequent states of affairs. This notion has theoretical motivations which need not con- 
cern us here (for details, see Leinonen 1982 and Dickey 2000, and for a more recent 
analysis of Rus aspect which is very compatible with this approach, cf. Zel'dovit 
2002); of primary relevance for the present discussion is the fact that temporal def- 
initeness has as a practical effect the limitation of pf verbs in the eastern languages 
to contexts of (explicit or implicit) sequentiality. Following Barentsen's (1995) version 
of this model, we may capture the profile of sequentiality of the Rus pf nicely by 
schematizing the relationship of the predicate to the prior and subsequent states of 
affairs as X + S + Y, where S is the predicate expressed by a pf verb, X is some 
other prior state of affairs, and Y is some other subsequent state of affairs. This 
schema is in fact one way of representing the location of a situation S in time rela- 
tive to intervals characterized by its absence, i.e., -S + S + -S (this will become 
relevant in section 5). 
Space considerations preclude a detailed explanation of how the hypothesized 
meanings for the pf aspect in the respective groups motivate the differing aspectual 
usage in the parameters considered by Dickey (2000). As an illustration, let us brief- 
ly consider two of them, taking Rus and Cz as representative of the eastern and 
western languages (respectively). The first parameter is habituality: as shown in (I), 
'The following abbreviations are used in this article: Blr (Belarusian), Blg (Bulgarian), Cro 
(Croatian), Cz (Czech), Mac (Macedonian), OCS (Old Church Slavic), OCz (Old Czech), OPol (Old 
Polish), OUkr (Old Ukrainian), Pol (Polish), ORus (Old Russian), Rus (Russian), Slk (Slovak), Sln 
(Slovene), Sor (Sorbian), Ukr (Ukrainian), impf (imperfective), pf (perfective). 
The term profile is used here in the sense of the work of Ronald Langacker (e.g., 1987), i.e., 
the particular elements of a base structure that are designated by a particular linguistic unit. 
S. M. Dickev. S-12- and Grammaticalization of Asvect 5 
the eastern languages strongly prefer the impf in habitual utterances, whereas the pf 
is quite common in the western languages. 
(1) a. Kaidyj den' on *vyp'etPlvypivaeti po odnoj rjumke vodki. 
'He drinks a glass of vodka every day.' [Rusl 
b. Vypijep jednu sklenitku vodky dennE. [czl 
The analysis is fairly simple: in the west, a habitual situation viewed on the basis of 
a single representative instance event can be coded pf if that situation is viewed in 
its totality (as a complete whole); in (1) the quantification of the object facilitates the 
total view of the situation. Thus, what the western pf expresses in (1) is merely the 
action producing the transition from a full glass to an empty glass. In contrast, the 
temporal definiteness of the Rus pf renders it unacceptable in cases of habituality, 
because a habitually repeated situation cannot be viewed as uniquely locatable in 
time; nor is there any sequentiality present on the level of the single representative 
instance, i.e., (la) contains no reference to any X andlor Y in the schema X + S + 
Y. This analysis is supported by the fact that the pf in all the eastern languages is 
generally acceptable in the expression of habitual sequences of events: 
(2) On vsegda tak-vyp'etp kofe i pojdetp na rabotu. 
'He's always like that-drinks his coffee and goes to work.' [Rusl 
In (2), the drinking situation is presented as the first of two sequential situations on 
the level of the representative instance. This fulfills the uniqueness condition (b), 
and the pf is acceptable. 
Another illustrative parameter involves the impf general-factual. Although the 
impf aspect occurs in the general-factual function in all Slavic languages, differen- 
ces do exist: one is that in the western languages, the impf is unacceptable in the 
denotation of a single achievement in the past; in the eastern languages, however, it 
is acceptable. Compare the examples in (3): 
(3) a. NaBelpl*Nachizeli jsem vtera tu knihu, kde je ted'? 
'I found that book yesterday, where is it now?' [czl 
b. Ja naxodili k u  knigu vtera, gde ~ n a ? ~  [Rusl 
In the west, the meaning of totality expressed by the pf renders it acceptable in the 
denotation of a single achievement, which is necessarily a totality, regardless of the 
overall context. As for Rus, the general-factual function is inherently incompatible 
with the temporal definiteness of the pf aspect, as the situation in question cannot 
be viewed as unique in the fact structure of the discourse; note also the lack of any 
explicit sequentiality to motivate the pf in (3b). 
Let us now turn to the impf aspect. According to the east-west aspect theory, 
the impf in each group has its own distinct (positive) meaning. In the west, the impf 
expresses QUANTITATIVE TEMPORAL INDEFINITENESS: the assignability of a 
situation to more than one conceptual point in time in the fact structure of a dis- 
This Rus example is taken from Styreva (1992: 176), where it is given with its Slk equival- 
ent. 
6 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
course. In (3a), this meaning contradicts the context of a single achievement, which 
must be assigned to a single (conceptual) point in time, with the result that the 
impf is unacceptable in Cz. The meaning of the eastern impf is QUALITATIVE 
TEMPORAL INDEFINITENESS: the non-assignment of a situation to a single, u- 
nique point in time relative to other states of affairs. Habitual events obviously 
cannot be located at a single, unique point in time and are thus qualitatively tem- 
porally indefinite, hence the acceptability of the eastern impf in (1) above. Gener- 
al-factual contexts such as (3b), in which a single achievement is not uniquely locat- 
ed relative to other states of affairs, also sanction the qualitative temporal indefi- 
niteness of the eastern impf. (This brief description gives only a broad outline of 
the analysis; for details, see Dickey 2000.) 
The east-west aspect theory as presented by Dickey (2000) primarily concerns 
aspect usage, as opposed to the morphology of aspect. However, certain differences 
between the eastern and western groups are evident in the morphology of aspect and 
the derivation of various kinds of Aktionsart verbs, very few of which have been 
discussed in detail. Dickey (2000: chapter 7) does discuss the presence of a class of 
specifically ingressive verbs prefixed in za- in the languages of the eastern group (as 
opposed the languages of the western group, where za- is simply a totalizing prefix). 
Dickey (2001b) examines east-west differences in the scope and function of the re- 
flexes of *-nq-, and Dickey and Hutcheson (2003) discuss east-west differences in 
the derivation and function of delimitatives in po-. The analysis presented here at- 
tempts to interpret the productivity (or lack thereof) and function of the innovative 
prefix s-Iz- in the Slavic languages in terms of the east-west aspect theory. 
Before beginning the discussion, however, it is necessary to lay out a basic 
theoretical approach to aspectual prefixation in Slavic languages. As a grammatical 
process, aspectual prefixation itself (as opposed to Aktionsart prefixation) is rarely 
discussed in detail as a system, perhaps due to the array of prefixes that perfectivize 
verbs in any given Slavic language. The approach taken here is based on that de- 
veloped by Shull (2003), who discusses the relationship between the spatial mean- 
ing(~) of a prefix and its telic and perfectivizing functions. She draws a clear distinc- 
tion between the spatial and abstract uses of prefixes, and argues against the view 
that abstract prefixation is based on directly metaphorical mappings of the spatial 
meanings of prefixes to abstract domains. Shull (184-5) also suggests that all prefix- 
es, whether involving a landmark that is a SOURCE, a PATH or a GOAL, and re- 
gardless of their particular trajector and landmark configurations, share an abstract 
profile of two states S1 and S2 (i.e., the initial state and the resultant state respective- 
ly), and that all prefixes tend to become goal prefixes profiling the attainment of S2 
due to the "goal orientation of language," which "effectively neutralizes the distinc- 
tion between Source, Path and Goal prefixes when they are used abstractly" (185). 
Thus, all prefixes share as an abstract schema the TRANSITION FROM S1 TO S2, 
i.e., the transition from an initial state to some different resulting state. Shull (225) 
suggests that the spatial prototypes of prefixes are in fact subcases of the abstract 
SOURCEIPATHIGOAL schema and that the former "possess a richer structure and 
thus occupy a privileged position in the semantic network of individual prefixes". 
(Her analysis of Slavic prefixation is very much in the spirit of Langacker (1988), 
who argues that speakers derive abstract schemas from lower level, more highly spe- 
S. M. Dickev. S-/Z- and Grammaticalization of Asvect 7 
cified meanings (one of which may be a prototype), so that in addition to a proto- 
type, speakers may also access a highly abstract schema from all nodes of the sem- 
antic network of a linguistic unit.) Shull's approach to abstract (i.e., telicizing, per- 
fectivizing) prefixation is important for the discussion of s-/z- presented below be- 
cause it emphasizes the elements of meaning that are shared by otherwise spatially 
disparate prefixes, which in turn can help us to understand the semantic develop- 
ment of s - /~ -  from two distinct prefixes *sa- and *jbz-. Shull's analysis presents a 
number of other interesting findings which shall not be elaborated here, but will be 
mentioned as they become relevant in the discussion. 
2. A Description of s-/z- in Slavic 
This section presents a brief description of the prefix s - / ~ - ~  in the West Slavic lan- 
guages and Slovene. Section 2.1 discusses *sa- and *jbz- in OCS and the coalescence 
of the two prefixes with the advent of the fall of the jers. Section 2.2 discusses the 
resulting situation in West Slavic and Slovene; the productivity of s-/z- or lack there- 
of in the East Slavic languages and the remaining South Slavic languages is dis- 
cussed in section 2.3. 
The productivity of s-/z- as a perfectivizer in the West Slavic languages is well 
known in the aspectological literature. Yet this prefix is problematic, as it is general- 
ly recognized as having two sources: *sa- and *jbz-. The original source of its voice- 
less allomorph s- was the prefix *sa-. Its voiced allomorph z- developed from the 
erosion of Common Slavic *jbz- during and after jer-fall (i.e., in principle *jbz- + 
* j ~ - ~  + Z-). Thus, Cz, Slk, Sor6, and Pol have no prefix iz- comparable to the South 
Slavic languages, only z- (cf. SeligEev 194111969: 91, 234, 3177). 
The same basic coalescence of *sa- and *jbz- into s-/z- has taken place in all 
these languages, but the details differ from language to language according to the 
degree of regressive voicing assimilation that has taken place (and orthographic tra- 
dition). Though Cz, the language that has utilized s-/z- to the highest degree, has 
apparently established voiced z- as the default (cf., e.g., zosnovatp8 'endow'), its orth- 
ography has made a half-hearted attempt to maintain s(e)- and z(e)- as separate pre- 
fixes. Nevertheless, Janka (1997: 102) points out that both sbZhnoutP (+ *sa-) 'run 
down' and zbZhnoutP (+ *jbz-) 'flee' are pronounced [z-1, whereas both skopatp (+ 
*sa-) 'dig away' and zkopatp (+ *jbz-) 'dig up' are pronounced [s-1, and remarks 
that in Cz "tendencies toward the unification of s(e)- and z(e)- have been (and are) 
at work." Moreover, the effort to maintain the distinction between etymological *sa- 
and *jbz- has resulted in a certain amount of confusion in dictionary treatments 
The reflexes of *ss- and *jbz- shall be collectively referred to as s-/z-, regardless of the de- 
tails of voicing in the individual Slavic languages (and whether or not individual scholars group 
them together); where such details become relevant, the language-specific forms are referred to. 
The intermediate stage jz- is attested in OCz. 
Sorbian data included in this analysis are taken from Upper Sorbian. 
SeliSEev in fact only mentions the corresponding prepositions, but the developments are 
identical. 
An anonymous reviewer points out that "in formal speech, especially in Bohemia, the prefix 
z- followed by a vowel is often devoiced with a glottal stop dividing the prefix and the following 
vowel." 
8 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
(cf., e.g., the competing spellings skazitp and zkazitp, both 'spoil'). Janka's remark 
seems in my view to be quite an understatement, in light of Trivnitek's (1923: 175, 
183-5) observations on how artificial the modern Cz orthographic distinction be- 
tween s- and z- is. As Poldauf (1954: 64, fn. 4) points out, dialectal differences play 
a role as well: central Bohemia prefers s(e)-, e.g., seiitp 'sew' (except in bookish 
words such as zemdlitp 'become exhausted'), whereas Moravia prefers z(e)-, e.g., 
zeiitp 'sew'. Thus, we are justified in concluding that spoken Cz has merged *sa- 
and *jbz- into a single prefix, with different defaults depending on the dialect, and 
with various artificial exceptions stemming from prescriptive efforts in dictionaries. 
In Pol the default (unmarked) allomorph of the prefix has become z(e)-, which 
occurs not only before voiced obstruents but also before s, sz, resonants and vowels, 
cf., e.g., zgrupowaP 'group together', zszydp 'sew together', zrnbwiP sie 'agree', 
zebra@ 'collect' (+ *sa-), and s- occurs only before voiceless obstruents, e.g., skleidp 
'paste together' (+ *sa-)? Likewise, in Sor, Slk and Sln the unmarked allomorph is 
voiced. In Sor it is z(e)-, cf., e.g., zeh? so 'come together' and zrnjasdp 'crumple' to- 
gether (+ *sa-), whereas s- occurs before voiceless obstruents, e.g., styknydp 'put to- 
gether' (+ *sa-). In Slk, the default is z(o)-, cf. e.g., zobratp 'collect, assemble' (+ 
*sa-), zrnetatP 'sweep together' (+ *sa-) and splynftp 'flow together' (+ *sa-). 
Though the coalescence of *sa- and *jbz- into s-/z- is generally treated as a 
West Slavic phenomenon, it must be pointed out that this same development took 
place in the westernmost South Slavic language, Slovene, cf., e.g., Bajec (1959: 112). 
But in contrast to the West Slavic languages, standard Sln has also kept the prefix 
iz-, so that it disposes of both iz- and z- (+ *jbz-). In Sln, many verbs would indi- 
cate an unmarked allomorph of z- cf., e.g., zedinitip 'unite' and zoiitip 'make nar- 
row', where z- is in prevocalic position. However, as in Cz, if the jer in *sa- pro- 
duced a modern fill vowel, se- is common, e.g., segnitip 'rot'. Note also that s- is 
common in verbs containing reflexes of *sari- retaining the nasal, either in the old 
centripetal meaning of the prefix (e.g., snitip se 'come together'), or in its resultative 
meaning (e.g., snestip 'eat up'). However, verbs with the more recent perfectivizing 
s-Iz- (see below) show z- as the unmarked allomorph before vowels and resonants, 
cf., e.g., zindustrializiratip 'industrialize' and zrornantiziratip 'romanticize'. (One 
might be tempted to see z- before vowels and resonants not as the voiced allomorph 
of S-/z-, but as a reduced form of iz-, as the SSKJ gives doublet forms of some re- 
cent loans, e.g., zniveliratip-izniveliratip 'level'. But many such recent loans prefixed 
with z- have no doublet in iz-, e.g., zindustrializiratip 'industrialize'. Note also that 
the press tends to prefer forms in z-, as does the colloquial language. In view of 
these facts, I consider recent loans in z- to be evidence of the productivity of an in- 
novative-though not really new-perfectivizing suffix s-/~-.) Otherwise, the allo- 
morphs of s-12- occur strictly according to voicing, cf., e.g., spitip drink (+ izpitip) 
and zbratip 'collect together' (+ *sa-). 
Thus, the distribution of voiced z- and voiceless s- in Cz, Slk, Sor, Pol, and Sln 
often does not follow the etymological sources at all (cf. in this regard also SeligEev 
A palatalized allomorph S- appears before 6, e.g., SciqgnqCP 'pull together'. Note also that 
according to Smiech (1968: 266), Pol attests similar dialectal differences regarding the default al- 
lomorph. Voiceless s-, for instance, is the default in many areas of Malopolska, Silesia and south- 
ern Wielkopolska. 
S. M. Dickev. S-12- and Grammaticalization of Asvect 9 
194111969: 182, who makes the same point regarding the corresponding prepositions s 
and z in West Slavic). The descriptions given above show that in Cz, Slk, Sor, Pol, 
and Sln *sa- and *jbz- have in fact phonetically merged into a single prefix s-lz-. 
Although s- and z- are distinguished in the orthography of the western Slavic lan- 
guages to varying degrees, this is ultimately irrelevant to the semantic analysis pre- 
sented here, which considers the consequences of the rise of s-/z- as a single prefix 
for the Slavic system(s) of verbal aspect. 
2.1. The Prefixes *s%- and *jbz- in OCS and the Coalescence 
In this section the hybrid nature of western Slavic s-lz- is discussed from a semantic 
point of view. Such an examination is, however, not a simple issue. Since diachronic 
developments are the focus of the analysis, the best point of departure is a review of 
the semantics of *s5- and *jbz- in OCS, the oldest recorded Slavic language. The 
following description of OCS s5- and iz- is based on the information given by Slon- 
ski (1937). 
According to Stonski, sa- had the following meanings in OCS: (1) the CENT- 
RIPETAL meaning (i.e., motion from many directions to a single landmark), e.g., 
saiitip 'sew together';" (2) the DOWNWARD-ABLATIVE meaning (i.e., motion 
downward from a landmark), or "generally motion from any point" (225), e.g., 
satrestip 'shake off'; (3) the CONCOMITANT-ACTION meaning (225), e.g., sapoiitii 
'live [with]'; (4) the RESULTATIVE meaning (Stonski, 5), e.g., sazbrztip 'ripen'. 
Slofiski (253) notes that the centripetal and resultative meanings are the most fre- 
quent meanings of sa-, followed by the downward-ablative meaning. A count of 
Slofiski's verb list shows that the centripetal meaning and the resultative meaning 
are each expressed in 47 verbs (31.5% each); the downward-ablative meaning is re- 
presented in only 23 verbs (15.4%), and the concomitant-action meaning is repre- 
sented by only 10 verbs (6.7%):' Thus, the two primary meanings of sa- in OCS 
were the centripetal meaning and the resultative meaning. 
The latter is most important for the history of s-lz- as a perfectivizing prefix. 
The resultative meaning of OCS sa- is already an abstract aspectual meaning, 
which even by that time had apparently become semantically detached from the 
concrete spatial meaning(s) of the prefix. In other words, by the time of OCS, sa- 
was already functioning in a manner anticipating the priverbes vides of the modern 
Slavic languages. Given the well documented semantic development from concrete 
spatial meanings via non-spatial telicity to abstract perfectivization that Slavic pre- 
fixes tend to undergo, the question that naturally arises is which of the two spatial 
meanings of sa- served as the starting point for its resultative meaning. The situa- 
tion in this regard is not entirely clear. Some verbs, such as saiitip 'sew together', 
satvoritip 'create' and sazbdatiP 'build' would indicate the possibility of deriving the 
resultative meaning from the centripetal meaning (cf. in this regard Agrell 1908: 
lo Regarding data from OCS and other older stages of Slavic languages, the superscripts P and 
indicate "proto-perfective" and "proto-imperfective" respectively; it is doubtful that fully gram- 
maticalized aspect systems existed before the fifteenth century, though the time at which it can 
be said that grammaticalized aspect existed probably differs among the individual Slavic lan- 
guages. 
l1 A few of Slofiski's judgments are perhaps debatable, but his numbers are valuable as a re- 
flection of the overall tendencies at work. 
10 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
87). In my view, an interesting piece of circumstantial evidence for this hypothesis 
is the fact that the OCS deadjectival/denominal inchoative verbs which according to 
Sloliski express this resultative meaning - samilitip se 'take pity on', samraznqtip se 
'freeze', and sastarZtiP se 'age"' - manifest the very same reflexive derivational 
model as the very concrete centripetal sanitip se 'come together'. This is important 
because such inchoative predicates have no inherent predisposition towards one kind 
of (spatial) telicity over another: the fact that they were derived according to the re- 
flexive centripetal model suggests that this meaning of the prefix was productively 
metaphorized to create resultative verbs. 
On the other hand, some of Slonski's resultative verbs can be plausibly derived 
from the downward-ablative meaning, e.g., sakratitip 'shorten' (cf., German ab- 
kiirzen, English cut downloff) or sakrytip 'hide' (cf. German abdecken 'cover'), 
though there are fewer such cases. Overall, it does seem that the centripetal mean- 
ing is the better candidate for the source of the abstract, resultative meaning of 
OCS sa-. We might, however, qualify this view with the observation that individual 
verbs could have a predilection for developing a resultative meaning from one or the 
other sense depending on the lexical content of the verb; thus, kratitii 'shorten' 
would naturally form a pf resultative by combining with the (downward-) ablative 
sense of sa- (inasmuch as shortening is effected by cutting away), whereas zbdatii 
'build' would form the same by combining with the centripetal meaning (inasmuch 
as building involves assembling things in one place). The notion that the lexical con- 
tent of a prefix overlaps with the base meaning of a source verb to the point where 
the prefix becomes apparently "semantically bleached" and creates a "lexically iden- 
tical" pf verb is known as subsumption, and has been a real part of the develop- 
ment of prefixal perfectives in Slavic (cf., e.g., Poldauf 1954 and Niibler 1990). But 
again, the inchoative data are convincing enough for me to assume that the centri- 
petal meaning was the meaning most closely related to the resultative meaning of 
sa-. Accordingly, we may assume that by OCS, the individual spatial meanings of 
sa-, i.e., the centripetal meaning and the downward-ablative meaning, had each 
produced abstract resultative meanings simply profiling the TRANSITION FROM S1 
to s2. 
How conceptually distinct the resultative meaning was from the centripetal 
meaning is open to speculation, but this analysis does not depend on how this ques- 
tion is answered in any direct way.'3 Again, Shull (2003) argues that abstract mean- 
ings of prefixes need not be considered to be metaphorical extensions, but rather in- 
dependent meanings, as with abstract meanings there are rarely good candidates for 
the roles of trajector and landmark. For purposes of this discussion, I will assume 
that Common Slavic *sa- had a distinct resultative sense, more or less "de-etymolo- 
gized," and thus very likely standing in some sort of semantic network relationship 
with the spatial centripetal meaning (cf., in this regard, Bajec 1959: 112, who also 
links these two meanings). Figure 1 shows the assumed network relationships. (The 
network diagrams given here and below bear no claim of psychological reality, but 
are one way of representing the semantic structure of the prefixes in question. The 
nodes of the semantic network are represented by the circles, and relatively more 
Cf. also ssssxn~tip sq 'dry up', given in the Staroslavjanskij slovar'. 
l3 Note that Slosar (1981: 91) also assumes a separate resultative meaning for ss- in OCS. 
S. M. Dickev. S-12- and Grammaticalization of Asvect 11 
salient nodes are represented by thicker circles. Where it is doubtful that there is 
any significant semantic link between two nodes, no line connects them, as in the 
case of the concomitant-action meaning and the resultative meaning. We may as- 
sume a general semantic link between the centripetal and downward-ablative mean- 
ings inasmuch as they both profile the TRANSITION FROM S1 TO S2.) 
Figure 1: Basic Semantic Network for OCS s%- 
resultative 
n 
concomitant- centripetal downward- 
action ablative 
Let us now turn to OCS iz- as a basis for assessing the semantics of Common 
Slavic *jbz-. Sloliski (1937) gives three main meanings for OCS iz-: (1) the ELAT- 
IVE meaning (out of), e.g., istaknqtip 'pluck out'; (2) the EFFECTIVE meaning, 
signaling that an action is carried out 'th~roughly"~, e.g., isaxnqtip 'dry up'15; (3) the 
RESULTATIVE meaning, e.g., izmznitip 'change'; and (4) the DISTRIBUTIVE 
meaning, signifying that an action affects all of a set of objectslsubjects one after 
the other, e.g., izvezatip 'bind together [one after another]'. According to Stonski's 
data, equal numbers of verbs prefixed with iz- have the elative and effective mean- 
ings (49 in each case, or 39%). The resultative meaning (15 verbs) is clearly related 
to the effective meaning, inasmuch as thoroughness of completion is usually in- 
volved in producing a result. The distributive meaning is also closely related to the 
effective meaning: an action carried out thoroughly on a set of objects will affect 
all of them. This is why most of the verbs Sloliski gives as having distributive 
meaning are also characterized as expressing the effective meaning; in fact, the only 
verb Sloliski gives as simply distributive is izvezatip. The others (a total of 9) are 
characterized as distributive-effective (dystrybutywno-efektywny), e.g., izbitip 'kill 
[several]', izmrztip 'die [of several]'; the relation between the notions of thoroughness 
and distributivity can likewise be seen in the English kill off, die off/out. Summing 
up, we may conclude that the original spatial meaning of *jbz-, out of, produced a 
large number of verbs that may be broadly characterized as "resultative", for the 
pure resultatives (izmznitip), the effective verbs (isaxnqtip) and the distributive verbs 
(izvezatip) all profile the "absolute completion" (to use Jakobson's term) and accom- 
panying result of an action. The resultative meanings of iz- may be viewed as meta- 
phorical extensions of its spatial meaning 'out of', as Bajec (1959: 112) has suggested 
("what proceeds out of a verbal action is completed"), but need not be, cf. Shull's 
views discussed above. Figure 2 illustrates the network for OCS iz-: 
l4 Cf. Agrell (1908: 109); Agrell is the source for Sioriski's Aktionsart meanings. 
l5 Note the same meaning with the synonymous prefix aus- in German: austrinken 'drink 
up', auspendeln 'swing to a standstill'. 
12 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
Figure 2: Basic Semantic Network for OCS iz- 
resultative effective 
elative distributive 
Thus, *sa- and *jbz- both had very prominent resultative/perfective meanings 
in addition to their spatial meanings (from which the former were originally meta- 
phorically derived) by the time of OCS. In this respect, they are similar, and also 
stand (alongside po-, which had also developed some of the same aspectual mean- 
ings by this time, cf. Sloliski 1937) apart from other prefixes, whose semantic func- 
tions were more clearly dominated by their primary spatial meanings. It is indeed 
interesting to note that in the case of some predicates, OCS attests resultatives pre- 
fixed with both prefixes, e.g., iztstip and s.6nbtiP 'eat up', ispttip and s5pttiP 'sing'. 
It is not my intention here to determine the precise difference in meaning between 
such doublets; all that is important for this discussion is that both *s5- and *jbz- 
could in principle produce resultative verbs from one and the same source verb. This 
situation is not unique to OCS, but has been observed in other Slavic languages. 
Witkowska-Gutkowska (1999) discusses many such pf doublets in OPol and analyzes 
the reasons in each case for the elimination of some in favor of one in the modern 
lang~age!~ According to Witkowska-Gutkowska, the elimination of such redundant 
doublets involved the loss of some meanings of individual prefixes as well as the 
semantic specialization of various prefixes. For example, OPol attests nagotowadp, 
przygotowaP, ugotowaP, and zgotowaP, all meaning 'prepare, ready'; of these, 
modern (standard) Pol has retained only przygotowadp. While an analysis of the fac- 
tors influencing the elimination of such doublets is quite interesting, equally interest- 
ing is the issue of why such doublets would appear at all, given their apparent re- 
dundancy. In my view, the cause is to be found simply in the early metaphorization 
of the spatial meanings of various prefixes to produce telic pf verbs. For instance, 
the Pol prefixes na-, przy-, u-, and z- all had spatial meanings involving trajectors 
and landmarks which could be metaphorized to profile the telicity of a given action 
(i.e., to profile merely the TRANSITION FROM S1 TO SZ); thus, it should not come 
as a surprise that a certain amount of "overproduction" resulting in "redundant" 
doublets would take place. (This is by necessity an oversimplification of the semantic 
issues involved. It would be erroneous to assume an absolute degree of synonymy be- 
tween nagotowadp, przygotowadp, ugotowaP and zgotowadp: it is hard to believe that 
many if not all such verbs each had a particular lexical nuance added by the prefix. 
But the slight differences in meaning are irrelevant to the immediate concerns of 
this discussion. This also applies to any possible difference between OCS iztstip and 
smtstip.) Returning to OCS iztstip and smtstip, it should surprise us even less that 
s'b- and iz-, which had already developed salient resultative meanings, should pro- 
duce "competing" or "redundant" doublets. 
l6 Nefed'ev (1994: 78) briefly mentions the same process in the history of Rus. 
S. M. Dickev. S-12- and Grammaticalization of Asvect 13 
This brings us to the fall of the jers (b, a), which resulted in the reduction of 
*jbz- to Z- in West Slavic, Ukr and Blr, which in turn was undoubtedly the catalyst 
for the coalescence of s- and z- into one prefix. 
Slosar (1981: 105) discounts the possibility of the replacement of *jbz- by *sa- 
in Common Slavic. According to Hujer (1922/1961: 125), the final reduction of jz- + 
z- occurred in Cz in prehistoric times (i.e., before the fourteenth century, the time 
of the oldest Cz texts). The reduction of *jbz- + z-, as well as the resulting phonetic 
merger of s- (+ *sa-) and z- (+ *jbz-), are simple to comprehend as sound changes 
in a system with voicing assimilation. Less clear are the consequences for the sem- 
antic meanings of the prefixes and their subsequent development. 
In my view, it may be safely assumed that after the phonetic merger of s- and 
z- there was some period of time during which the spatial meanings of the prefixes 
were essentially unaffected, i.e., the semantic networks of the prefixes were un- 
changed. This is illustrated in Figure 3, which presents an (admittedly considerably 
simplified) "compound" semantic network for the new s-Iz-. 
Figure 3: Compound Semantic Network for Early s-/z-l8 
distr. effective 
centripetal 
elative 
resultative 
downward- 
ablative 
Thus, the new z- retained the elative meaning as one of the salient (and inde- 
pendent) nodes in its network. However, it is clear that this situation did not con- 
tinue (and it is possible that the elative and downward ablative meanings combined 
into a single ablative meaning), because in West Slavic (as well as Ukr and Blr) the 
function of expressing the elative meaning was taken over by vy- (at a relatively 
Figure 4: Compound Semantic Network for s-/z- after the Spread of vy- 
dE effective 
centripetal - (elative) 
resultative 
downward- 
ablative 
l8 Here the concomitant-action meaning is ignored, as it is no longer relevant to the analysis. 
14 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
later time in Blr and Ukr),'9 cf. Slosar (1981: 105; see section 2.2.1 for a discussion of 
the spread of elative vy- as it relates to s-Iz-). Thus, the semantic "basis" of the co- 
alescence were the resultative meanings shared by both s- and z-. After vy- largely 
took over the elative meaning, the elative meaning lost its salience in the network of 
s-Iz-, and the resulting network is illustrated in figure 4. 
2.2. Innovative s-/z- in West Slavic and Slovene 
The following sections describe the coalescence and especially its semantic conse- 
quences in more detail. Cz is discussed in 2.2.1, and the description of Cz serves as 
the basis of the presentation. The descriptions of Slk, Sor and Sln in section 2.2.2 
and the description of Pol in 2.2.3 focus primarily on the degree to which the devel- 
opment, scope and function of s-Iz- in those languages resemble or differ from that 
of s-/z- in Cz. 
2.2.1. Czech 
After jer-fall and the spread of vy-, we may say that in Cz s-/z- had three principle 
meanings, the centripetal, downward-ablative, and the cluster of resultative mean- 
ings. The resultative meanings became very productive in Cz (and Slk), and further 
development produced verbs that were not specifically resultative, but arguably ex- 
pressed perfectivity without any additional semantic nuance. Slosar (1981: 106) ob- 
serves that by the end of the fifteenth century (as evidenced by the psalter of the 
Bible bendtskd, printed in 1506), "a situation closely resembling that in modern 
Czech had crystallized," i.e., s-Iz- had developed from a primarily resultative prefix 
into a prefixing expressing simple perfectivity. Thus, according to Slosar the Bible 
benatska attests more simple perfectives in s-Iz- (e.g., zmZnitiP 'change7, ztlfcip 'beat', 
zstaratip 'grow old', ztvrdnftip 'harden') than effective resultatives in s-/z- (e.g., 
zeiratip 'devour', zemdlttip 'become exhausted'). The productivity of s-/z- character- 
istic of the fifteenth century must be a contributing factor-in one way or anoth- 
er-to Vintr's (2001: 214) general view that "it is fifteenth-century Old Czech that 
first shows the full grammaticalization of aspect."20 
Before going on, it should be pointed out that this account of the development 
of Cz s-/z- as a single prefix differs considerably from the way s- and z- have been 
treated in the Bohemistic literature (in which different scholars have treated s- and 
z- differently). For instance, KopeCnf (1962: 120-1, 126-8) treats s- (+ * s ~ )  and z- 
(+ *jbz-) as completely distinct prefixes in modern Cz. This is clearly wrong. Slosar 
(1981) distinguishes between s- and z- in Cz, on the basis of their spatial meanings 
(the centripetal and downward-ablative, and elative meanings respectively). However, 
in his treatment of s- he admits some degree of coalescence due to voicing assimila- 
tion, and suggest that the resultative meanings of s- and z- were "a point of con- 
tact" between the two prefixes and that "the former at some point during historical 
times changes into z(e)-" (91). In his treatment of z-, he observes that "during his- 
l9 Pol wy-, Sor wu-. 
20 One may agree or disagree with Vintr's opinion that aspect was "fully" grammaticalized in 
fifteenth-century Cz, but it seems rather clear that by that time some state of affairs existed that 
fairly closely approximated the present-day aspect system in Cz. See section 5 for a discussion of 
the grammaticalization of aspect in Cz. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 15 
torical times z- was reinforced in its resultativ: function by forms that originally 
had the prefix s-" (105). In other words, what Slosar is suggesting is that with the 
advent of vy- in the expression of the elative meaning, z- remained solely in its re- 
sultative meanings, and afterwards s- moreYor less continually hemorrhaged its resul- 
tative meanings in the direction of z-. Yet Slosar (104) acknowledges "sporadic" cases 
of z- with the centripetal meaning (e.g., zbZhnoutiP se 'come together') and the 
downward-ablative meaning (e.g., zloiitip 'put awaylinter') which he considers to be 
minor tendencies in the other direction. 
Though Slosar's (1981) views are characteristically well-thought out, in my view 
it is useful go back to TrhvniCek (1923: 173-6) in an investigation of s- and z-, i.e., 
S-12-, in the history of Cz. TrhvniCek discusses in detail the confusion of the prepo- 
sitions s and z as well as the corresponding prefixes s- and z-. He notes that the 
prepositions s and z were confused only on the basis of the downward-ablative (s) 
and elative (z) meanings (both governing the genitive, i.e., s with centripetal mean- 
ing was not confused with z), cf., e.g., pdstip s nebes, pdstip z nebes 'fall from the 
heavens'. TrhvniCek suggests that it was the semantic proximity of the down- 
ward-ablative meaning of s and the elative meaning of z that facilitated the con- 
fusion of the two prepositions (in terms of Shull 2003, both are SOURCE preposi- 
tions, and their schemas differ only in the precise relationship of trajector and 
landmark). Moreover, TrhvniCek (174) points out that the alternation between s 
nebes and z nebes cannot be a consequence of a change in default voicing, as 
modern Cz distinguishes [sn-] and [zn-1, cf., e.g., sntstp 'eat up' and zndti 'know'. 
According to TrhvniCek (175-6), the confusion of the prefixes was more extens- 
ive: etymologically elative z- alternated with etymologically downward-ablative s-, 
cf., e.g., zpastip and spastip 'fall', as well as with etymologically centripetal s-, cf., 
e.g., zvyknutip and svyknutip 'get used to' (for numerous examples, see TrhvniCek 
1923: 155-63). Though TrhvniCek does not directly speak of a coalescence of s- and 
z-, he does observe that "the precise differentiation between the prefixes and prepo- 
sitions s and z ,  which the modern Cz literary language has introduced into its orth- 
ography, in many cases never existed in the [OCz] language." Thus, distinctions in 
the current orthography bear little, if any relevance to the issue of whether etymo- 
logical s- and z- coalesced into a single prefix, and in my view there is no reason to 
view s-/z- in historical Cz as anything but a hybrid prefix with allomorphs condi- 
tioned by voicing assimilation, and thus we are justified in consistently referring to 
orthographic s- and z- together as s-lz-. 
The other point that TrhvniCek makes which is relevant for the issue of the 
coalescence is his view that the confusion of the prepositions s and z in their 
downward-ablative and elative meanings (respectively) reflects "a coalescence of two 
concepts, or meanings [i.e., the downward-ablative and the elative meanings-SMD] 
into one" (174). What TrhvniCek's remark amounts to is a suggestion that Cz seman- 
tically merged its two primary SOURCE prefixes. Considered from Shull's (2003) 
approach, in which downward-ablative s- and elative z- are to be viewed as both 
sharing a SOURCE tr3jector-landmark schema, this development seems quite plaus- 
ible, especially given Slosar's (1981: 92) observation that s- expressed "motion down- 
ward or away [my italics-SMD]", i.e., s- was apparently already a generic SOURCE 
prefix.21 Note also TrhvniCek's (175) observation that in modern Cz a phrase such as 
Circumstantial evidence for this view is Sioriski's (1937: 225) observation that already in 
16 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
ze stromu is itself ambiguous between the downward-ablative and elative meanings, 
and that this ambiguity is removed by the accompanying verb (e.g., sletZtP ze 
stromu 'fly down from the tree' versus vyletZtP ze stromu 'fly out of the tree'). Thus, 
it is quite possible that after the fall of jers, Cz merged its downward-ablative and 
elative SOURCE prefixes into a single generic SOURCE prefix, in which case Figure 
3 (and Figure 4) should have a single spatial SOURCE node in place of the down- 
ward-ablative and elative nodes. 
In my view, the nature of the semantic coalescence of the spatial source mean- 
ings of s - / ~ -  is relevant for a proper understanding of the spread of vy- as an elative 
prefix. Everything else being equal, it is difficult to figure out whether vy- "dis- 
placed" z- as the elative prefix in Cz (as suggested by Slosar 1981: 107), or whether 
vy- filled a semantic vacuum left when z- lost its elative meaning. If soon after 
jer-fall and the coalescence of *sa- and *jbz- Cz merged its two main source pre- 
fixes s- and z-, the most reasonable assumption is that the elative meaning was tak- 
en over by vy- only afterwards, as the new source prefix s-/z- was semantically un- 
derspecified. Another reason for taking this view is that OCz and Cz attest remnant 
verbs prefixed with z- which arguably express the original elative meaning of *jbz- 
(e.g., OCz zdechnutip 'die', zvltcip sZ 'extricate oneself', Cz ziicip se 'renounce, fore- 
go', zhostitp se 'get rid of', zbZhnoutP 'flee'; cf., Slosar 1981: 107 and the references 
cited there); this is why vy- did not entirely replace z(e)- in the elative meaning  lo- 
sar, 107). Moreover, according to Slosar (98-9) the productivity of elative vy- has 
been increasing in Cz, and abstract, metaphorical meanings have only begun to de- 
velop in historical times, which in my view indicates a relatively recent proliferation 
of elative ~y- .~ '  A final reason is that had vy- replaced *jbz- very early, the elative 
meaning of *jbz- would not have been able to develop its resultative meanings in the 
first place. 
With regard to the dynamic development of s-/z- as a perfectivizing prefix in 
Cz, I again find Shull's (2003) analysis the most useful. After the coalescence of 
*s5- and *jbz-, each of which had arguably already developed independent abstract 
resultative meanings (i.e., the TRANSITION FROM S1 TO S2), the morphological re- 
organization of so many resultative verbs prefixed either with s- or Z- as being pre- 
fixed with allomorphs of the new prefix s-/z- resulted in one way or another in the 
establishment of a prominent semantic node of resultativity in the network of s-/z-. 
These meanings of resultativity are best viewed as independent semantic meanings 
(or as submeanings of a single abstract meaning of change of state), and not as 
secondary meanings derived via metaphor (cf. Shull 2003: 184). The development of 
s-/z- in Cz as its primary prdverbe vide is a quite logical (though not necessary) 
consequence of the disassociation of its resultative meanings from the spatial mean- 
ings that ultimately produced them. 
Indicative of the role of s-/z- as a resultative prefix in Cz is the fact that it was 
the first prefix to function as a specifically distributive prefix, signaling that a situ- 
ation affects all of a set of objects (or involves all of a set of subjects). Slosar (1981: 
OCS s6- expressed not just the downward-ablative meaning but also ablativity in general, "gener- 
ally motion from any point". 
'' KopeEnf (1962: 123) observes that vy- in fact never functions as a perfectivizing prefix 
without some nuance of its lexical meaning (i.e., elativity). 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 17 
106) observes that s-/z- performed this function in OCz, giving examples such as 
zjimatiP 'take [all of]', zoblatitip 'dress [all of]', zprobijetip 'punch through [all of]', 
etc. Though KopeEnjr (1962: 128) gives examples of distributive verbs in z- in the 
contemporary language, e.g., zotviratp 'open [all of]', the prefix is not very product- 
ive in this function"(and po- has taken over as the primary distributive prefix in 
historical times, cf. Slosar 1981: 120-1). 
Decisive for the status of s-Iz- as a priverbe vide in OCz was the extension of 
its perfectivizing role to the class of inchoative verbs, e.g., zhdstndtip 'thicken', 
zeslabnutip 'become weak', ztvrdnutip 'harden', as this considerably diversified the 
predicate types that utilized s;/z- as a priverbe vide, and thus contributed to its 
universality as a perfectivizer. Slosar (1981: 106) points out that in OCz, s-Iz- quickly 
became productive in the perfectivization of such inchoative verbs, and (130) makes 
the important suggestion that it was the (deadjectival) inchoative verbs that served as 
an intermediate step in the spread of s-Iz- to various other kinds of verbs. As point- 
ed out in section 2.1, inchoative predicates have no inherent predisposition towards 
any one kind of metaphorical telicity over another, and the fact that in OCz such 
verbs began to be prefixed with s-Iz- is a strong indicator of the productivity of per- 
fectivizing s-/z- in Cz from a relatively very early date. I suggest that the new pro- 
ductivity of s - / ~ -  as a perfectivizer of inchoatives in OCz was the result of the newly 
created (or consolidated) node of abstract resultativity which was central to the sem- 
antic network of s-/z-; I see no compelling reason at all to view the perfectivization 
of inchoatives with s-Iz- as the result of a direct metaphorization of one of the spa- 
tial meanings of the prefix. In other words, the semantic mechanism behind the 
perfectivization of inchoatives in OCz was the new abstract meaning of s-Iz-: the 
TRANSITION FROM S1 TO S2. Here it is worth pointing out that KopeEnf's (1962: 
120-1, 126-8) approach of attempting to make a strict distinction between s- and z- 
is misguided in that it entails that z- (+ *jbz-) became the perfectivizer of inchoa- 
tive verbs in Cz, which does not accord with the etymological evidence from any of 
the other Slavic languages. As pointed out in 2.1, in OCS the prefix that perfectiv- 
ized inchoatives was sa-, not iz-. Though data from contemporary Slavic languages 
shows that some inchoatives (mostly involving colors) have been prefixed with reflex- 
es of *jbz-, e.g., Cro izblijedjetip 'fade', others have been prefixed with reflexes of 
*sa-, e.g., Cro zgusnutip 'thicken'. Even Mac, in which iz- has become wildly pro- 
ductive as a perfectivizer, prefixes many inchoatives with s- and not iz-, as is evi- 
dent from Ugrinova-Skalovska's (1960) description of Mac prefixes, cf., e.g., zbesnep 
'become furious' and snemeep 'become mute'. The problem disappears if one assumes 
the coalescence of *sa- and *jbz- into S-Iz- in Cz. 
At this point it is important to mention a nuance of completeness of the 
change of state expressed by inchoatives in s-/z-, which becomes clear when one 
compares such verbs with inchoatives prefixed with po- that also exist for some 
verbs in Cz. The latter generally express attenuation of the result. Thus, bZleti 'turn 
white' has as its "default" zbZletP, whereas pobZletp means 'turn somewhat white' 
(note that in the S S J ~  the latter is defined in terms of the former, i.e., pobZletP is 
defined as trochu zbZletP). Other such doublets with the same difference in meaning 
are poblednoutPlzblednoutP 'turn pale', poternatplzternatp 'turn black', and poter- 
venatplztervenatp 'turn red'. The verbs in z- are regularly defined as the "perfec- 
tives" of their source verbs in the S S J ~ ,  which gives a very good indication of their 
18 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
default status vis-i-vis their correlates prefixed with po-. Here one can see the ori- 
ginal resultative meaning of s-/z-: in order for inchoatives prefixed with s-Iz- to be- 
come established as the default perfectives in contrast to their correlates prefixed 
with po-, the nuance of thoroughness must have been salient. It should be men- 
tioned that not all inchoatives prefixed in po- attested in the SSJC are currently 
used with attenuative meaning; some are simply rare verbs (e.g., podbinitttp as op- 
posed to standard zdttinitttp 'become childlike'). What is important is that s-Iz- has 
established itself as the sole productive prefix in the perfectivization of inchoatives 
in Cz. 
Prefixation with s-Iz- has become the main way perfectivizing a related class 
of Cz verbs, the factitives. Thus, btliti 'color white, whitewash' has as its pf zbtlitp. 
And here one sees the same relationship to po-: zbtlitp is the default pf of btliti, 
whereas the SSJC' defines pobtlitp as 'make somewhat white'. Another such doublet 
is pokroutitPlzkroutitP 'distort'. Here as well it should be pointed out that some verbs 
in po- are simply archaic (e.g., pohasitp 'extinguish') or otherwise stylistically mar- 
ginal, which is again an indication of the ascent of s - / ~ -  as a productive perfectiviz- 
er in Cz. 
Not only did s-Iz- take over the perfectivizing function of its two source pre- 
fixes *sa- and *jbz-, but it also made inroads on another old telic perfectivizing 
prefix, vz-, which again is an indication of its productivity. Hujer 192211961: 125 ob- 
serves that vz- essentially became z- in OCz (cf. in this regard also Gebauer 1963: 
435-7). For example, 'lift up' is now zdvihatilzdvihnoutP (+ vz-). Gebauer gives 
many other examples, suggesting that old vz- was more likely to reduce to z- if it 
was followed by one or more consonants (cf., e.g., zdvihnoutp) than if it was fol- 
lowed by a vowel. An example of the retention of vz- in prevocalic position is vzitip 
'take', vzalp 'took' [masc.]. But some of Gebauer's (435-6) examples, e.g., Modern Cz 
zoratip (+ vzoratip) 'dawn', zeptatip se (+ vzeptatip s 4  'wonder', and the OCz doub- 
lets wzechcePlzechczeP 'wanted', wzebranyePlzebranijP 'chosen', etc., indicate that vz- 
became z- quite readily in prevocalic position. In fact, vziti is more likely to be ex- 
ceptional than zoratip and zeptatip se, as its present tense adds a jer reflex to the 
prefix: vezrnernp 'I will take', etc. On the other hand, literary Cz has retained vz- 
before one or more consonants, e.g., vzhltdnoutp 'look up', vzklititp 'sprout up' 
vzkiiknoutp 'shout out'. If in some cases vz- became z- before vowels and in other 
cases remained vz- before consonant clusters, then we are justified in asking exactly 
what role phonetics played in the change from vz- to z- in some Cz verbs. 
The phonetic resemblance no doubt played some role, but to reduce the change 
to phonetics as Gebauer does leaves us with no real motivation for the change where 
it did take place?3 given the inconsistency with which it occurred. In this respect it 
is important to point out TrhvniCek's (1923: 179-80) observation that zblfditip 'err' 
existed alongside vzblfditip before the phonetically conditioned change vz- + z- took 
place and that it is incorrect to assume that verbs in z-, e.g., zbuditip wake up, de- 
veloped only from those prefixed in vz-, e.g., vzbuditip, etc. In other words, it is 
likely that s-Iz- was replacing vz- at a time before the morphophonemic change vz- 
23 Gebauer (435) observes that the standard language (obecnd teftina) regularly changed vz- 
to Z- before consonants, which gives us something more closely resembling a morphophonemic 
change, but this still does not explain the cases in which vz- + z- before vowels. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 19 
+ z - . ~ ~  Thus, I think it is more likely that there were two processes that occurred, a 
morphophonemic change vz- + z- and a spread of the new productive prdverbe vide 
s-Iz- at the expense of its older counterpart vz-. Note that Trhvnitek's (177-82) ac- 
count of the alternations s-Iz-/vz- emphasizes the disappearance of slight lexical dif- 
ferences between verbs containing the various prefixes, as well as the fact that the 
status of s-, z- and vz- as prdverbes vides was responsible for the alternations. 
Thus, we are justified in concluding that, regardless of a morphophonemic 
change vz- + Z-, s-/z- began to replace vz- as a perfectivizing prefix in OCzF5 If 
S-12- spread at the expense of vz- in OCz, then we must be able to semantically 
motivate the switch. This can be done quite easily. Here again we see the relevance 
of Shull's observation that SOURCE, PATH and GOAL prefixes, regardless of the 
configuration of trajector and landmark, share an abstract schema of the 
TRANSITION FROM S1 TO SZ, which they profile when used abstractly. If in OCz 
s-Iz- was already becoming established as an abstract perfectivizer, then it is possible 
that it would take over the perfectivization of verbs from other telic prefixes profil- 
ing the TRANSITION FROM S1 TO S2 on the basis of that shared semantic mean- 
ing. 
Another indicator of the productivity of perfectivizing s-/z- in Cz is the current 
situation regarding the perfectivization of loan verbs. As is well known, loan verbs 
offer an ideal way to assess which prefixes are currently productive as perfectivizers 
in a given Slavic language. According to Rusinovh (2001: 220), z- is undoubtedly the 
most productive perfectivizing prefix in Cz, cf., e.g., the following prefixed pf loan 
verbs: zbojkotovatp 'boycott', zdiagnostikovatp 'diagnose', zdigitalizovatp 'digitalize', 
Fformdtovatp 'format', zmonitorovatp 'monitor', znormalizovatP 'normalize', zrecidivo- 
vatp 'have a relapse', zrelaxovatp 'relax', zunitarizovatp 'make unitarian', etc. Rusino- 
vh (224) observes that perfectivizing z- has little or no spatial or other concrete 
meanings. So the fact that z- is used to perfectivize loan verbs makes sense, given 
the fact that many new verbs loaned into a Slavic language will provide no motiva- 
tion for telicity resulting from a particular spatial configuration. 
One last important indicator of the productivity of s-Iz- as perfectivizer is evi- 
dence of a weak tendency to generalize s-Iz- as a perfectivizer at the expense of 
other prefixes, at various times. In particular, s-/z- apparently forms a number of 
variant pf verbs alongside other perfectivesF6 Some examples of such doublets are 
pofantitp se-Ffantitp selfantiti se 'go crazy over', nafilmovatp-FfilmovatP/filmovati 
'film', p?efiltrovatP-FfiltrovatP/filtrovati 'filter', vyformovatp-FformovatP/formovati 
'form', ponZmtitP-znZmtitP/nZmtiti 'Germanize' and umiitp-zemiitPlmiiti 'die'. The 
point is nicely illustrated by the following examples containing nafackovatp and 
Ffackovatp, which reported the very same crime in different newspapers: 
24 TrAvniEek (180) points out as well cases in which vz- appeared alongside original z-, e.g., 
vzkusitiPlzkusitiP 'try', and suggests that such pairs were created in analogy to the pairs where vz- 
was the original prefix. Note that in order for this to happen vz- and z- must have been "compet- 
ing" before any morphophonemic change. 
z5 Note also that one might consider it irrelevant whether a morphophonemic change vz- + z- 
was the mechanism for the replacement of vz- by z-, as the end result is the same: the spread of 
S-/z- as a perfectivizing prefix. I would suggest that in fact both processes probably occurred in a 
kind of feedback loop, each encouraging the other. 
26 Cf., Slosar (1981: 130). 
20 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
(4) a. Jeho partnerka v osmkm mEsici tEhotenstvi se ho zastala a Sobek ji 
nafackovalp. 
'His spouse, who is eight months pregnant, stood up for him and Sobek 
slapped her .' [Cz; Dnes] 
b. Muie surovE kopal i do obliteje, ienu v osm6m mEsici tEhotenstvi, 
zfackovalp. 
'He also kicked the husband brutally in the face; he slapped the wife, 
who is eight months pregnant.' [Cz; Blesk] 
If s-Iz- is or has been competing with other prefixes to form pf verbs, the most 
reasonable conclusion is that such competition is the result of a tendency to general- 
ize s - / ~ -  as the grammatical marker of perfectivity in Cz. 
The historical development of a purely abstract perfectivizing prefix s-Iz- in Cz 
is very interesting with regard to the grarnmaticalization of its category of aspect. 
Though a more detailed discussion of issues involved with grarnmaticalization is 
presented in section 5, some preliminary remarks are appropriate at this point. We 
may say that aspect is a grammatical category in Cz because the great majority of 
predicates are expressed in (perhaps contextually determined) pairs of verbs, and the 
impf/pf aspect distinction is not restricted to a particular tense (e.g., past) or even 
finite forms. Though Shull's point that the schema TRANSITION FROM S1 TO S2 
may be extracted from virtually any source or goal prefix is surely correct, in my 
view the grarnmaticalization of aspect in the individual Slavic languages will be ac- 
companied by efforts to generalize a single prefix to varying degrees in each. In Cz, 
it is clear that the prefix which has been generalized to the highest degree is s-/z-. 
Not only is it the most productive prefix in the perfectivization of loan verbs, but it 
also has found the most diverse application as far as transformative predicate types 
are concerned: s-/z- perfectivizes ordinary transitive verbs (e.g., budovati/zbudovatP 
'build', hotoviti/zhotovitP 'produce', lomitiIzlomitP 'break', etc.), inchoatives (e.g., 
zfamfrnZtP 'go crazy', zchladnoutp 'become cold', zeslabnoutP 'weaken', etc.), and 
factitives (e.g., zfamfrnitp drive crazy, zchladitp 'make cold', zeslabitp 'make weak', 
etc.). According to ~losar  (1981: 106-7), s-/z- is currently the "primary" perfectiviz- 
ing prefix in Czech. 
Inasmuch as s-Iz- has played a role in the grarnmaticalization of aspect in Cz, 
it is important to keep in mind that the initial catalyst for the development of s-Iz- 
as an abstract perfectivizing prefix was the purely accidental coalescence of *sa- 
and *jbz- resulting from jer-fall. Therefore, the evident steps towards the grammati- 
calization of aspect in Cz should not necessarily be viewed as a teleological process, 
and the post-Common Slavic grarnmaticalization of aspect in the individual lan- 
guages is to be viewed as contingent on developments specific to individual Slavic 
languages or groups of them. 
To sum up, the development of s-/z- as an abstract perfectivizing prefix has 
been important in the grarnmaticalization of aspect in Cz, and has arguably played 
a significant role in the formation of the semantic meaning of the pf aspect in Cz. 
As abstract perfectivizing s-Iz- had as its origin the resultative meanings of *sa- and 
*jbz- which were based on the schema TRANSITION FROM S1 TO S2, which was in 
turn based on the telicity abstracted from the centripetal and elative meanings of 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 21 
*s5- and *jbz-, its pf meaning has necessarily been one that is very telic. In the fol- 
lowing sections, the other western Slavic languages are considered, and it is shown 
that s-12- has played an important role in the perfectivization of verbs in Slk, Sor, 
Sln (section 2.2.2) and Pol as well (section 2.2.3). 
2.2.2. Slovak, Sorbian and Slovene 
Let us start with Slovak. The situation in Slk closely resembles that of Cz. Smirnov 
(1970: 90, 99) observes that s-1.- is quite productive in the perfectivization of loan 
verbs, cf., e.g., alarmovatilzalarmovatP 'alarm', duplovatilzduplovat~ 'double', redigo- 
vatilzredigovatP 'direct'. Like Cz, loan verbs tend to be biaspectual and are prefixed 
with s-1.- to make them unambiguously pf. Smirnov (92) also points out that s-/z- is 
productive in the perfectivization of inchoative verbs, cf., e.g., zbohatnutp 'become 
rich', zhasnutp 'goldie out', zintenzivnietp 'intensify', zmlknutp 'fall silent', etc., and 
that such pairs are quite numerous in Slk. Note also that factitives are prefixed with 
s-/z-, cf., e.g., zboiititp 'make poor', zdraiitp 'make expensive', zhoriitp 'make worse', 
zmladitp 'rejuvenate', etc. It is worth pointing out that in contrast to Cz grammari- 
ans, Horeckf (1959: 200) states that s-, z- and zo- are allomorphs of a single prefix 
in Slk (this is also the position taken by the MSJ: 414). Given the overall close re- 
semblance between Cz and Slk with regard to s-/z-, this description of Slk will suf- 
fice. 
Sor shows the same development as Cz and Slk. Though Werner (2003: 144) for 
some reason excludes z- as a reflex of *jbz-, it is hard to believe that the coales- 
cence of *s%- and *jbz- described in section 1 did not take place in Sor as well. The 
HEWONS (22: 1718-19) clearly gives etymological indications that the preposition z 
out is a reflex of *jbz- (e.g., ze wsy 'from a village'; cf. also SeliSCev 194111969), and 
there is accordingly no reason to assume a different origin for the prefix. Though 
almost no remnants of elative *jbz- remain in Sor, ZUCP SO 'take off one's shoes' 
seems to be just that (cf., Cro izutip se), and zhubiP 'lose' very likely contains a re- 
flex of original *jbz- (cf., OCS izgoubitip, Cro izgubitip). Another piece of evidence 
for assuming that *jbz- was a source for Sor s-/z- is the fact that s-/z- is the form- 
ant for the distributive Aktionsart (cf., Fasske and Michalk 1981: 120-2), e.g., spo- 
wrbdedP 'turn around [of all]', zesydadp so 'sit down [of all]', zrozdawaP 'distribute 
[all of]', zwotpisowaP 'copy [all of]'; recall from section 2.1 that iz- was a distribu- 
tive prefix in OCS (on the basis of its resultative meaning), and currently is in 
Cro/Srb, Mac and Blg, and also that *jbz- was the source of distributive s-/z- in Cz, 
Slk and Pol (cf., Kucala 1966: 63). Nowhere in Slavic have clear reflexes of *s5- and 
*VW- been productive in the derivation of distributives. Note that Sor never began to 
derive distributives with po- like the majority of the other Slavic languages. As in 
Cz and Slk, the elative spatial meaning of old *jbz- was taken over by *vy- (+ wu-). 
Otherwise, s-/z- in Sor shows the basically the same productivity that it does in 
Cz and Slk. For example, it has been productive in the perfectivization of inchoative 
verbs, cf., e.g., schudnyCP 'become poor', zbohatnydp 'get rich', zblZdnyCP 'become 
pale' and zmjerznyCP 'freeze', etcZ8 It also perfectivizes factitives, cf., e.g., zbarbiCP 
27 Smirnov labels the prefix s-lz-lzo-, including the allomorph with the jer reflex. 
28 A difference between Sor and CzISlk is that some inchoatives are prefixed with wo-/wob- 
(t o-lob-), e.g., wobhorknykp 'become somewhat bitter', wochl6dnyCP 'become cold', wornl6dnykP 
22 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
'dye', zbtaznidp 'make crazy', zbohadidp 'make rich', etc. The suffixal inchoative-fac- 
titive distinction seems to exist in Sor as well, cf., e.g., zblaznjedp 'become foolish' 
and zbtaznidp 'make a fool of someone', zhorbadedp 'become buckled' and zhorbadidp 
'make buckled'. Sor appears not to make a consistent distinction between a thorough 
and attenuated change of state expressed by s-/z- and po- respectively. One attested 
pair is schiliP 'bend' vs. pochilidp 'bend a little', but the doublets zhorSiLPlpohorSidP 
'worsen', skrbtSidPlpokrbtSiCP 'shorten' and polZpSidPlzlZpSiiCP 'improve' appear to be 
synonymous (Werner, 128, observes that for some doublets the correlate in po- oc- 
curs in figurative contexts). The surface-contact meaning of po-, however, produces 
a related distinction in meaning: pobarbidp means 'cover a surface with paint', 
whereas zbarbidp expresses a more "thorough" result, i.e., 'dye'. A similar distinction 
exists between sEorniP 'blacken' vs. poCbrnidP 'blacken the surface of', and zbtlidp 
'color completely white' and pobZlidP 'whitewash'. Finally, s-/z- has also been produc- 
tive in the perfectivization of loan verbs, e.g., skondensowadP 'condense', zbankroto- 
wadp 'bankrupt', zelektrizowadp 'electrify', zrnechanizowadP 'mechanize', which we 
may consider the sine qua non of its status as a prdverbe vide in a given Slavic lan- 
guage. 
Werner (144) points out that Sor s-lz- has taken over the domain of wz-, and 
that at earlier times spelling with wz- was artificially recommended on an etymolog- 
ical basis. However, today etymological wz- is z-, cf., e.g., zbudiidp 'wake up' for ol- 
der wzbudiidp. Werner does not take up the issue of how *s%- and *v%z- coalesced 
into Sor s-lz-, i.e., whether there was a purely phonetic erosion of wz- or not. This 
issue will not be pursued here. What is important is that as in Cz and Slk, s-/z- 
took over old perfectives prefixed with wz-, which added to its distribution and thus 
salience as a prdverbe vide in the Sor aspectual system. 
It makes sense that Cz, Slk and Sor all pattern very similarly regarding s-Iz-, 
as they are closely related to each other genetically (non-lekhitic West Slavic lan- 
guages). But as pointed out in section 1, the rise of s-/z- as a perfectivizer is clearly 
evident in Sln as well (note that this fact parallels the phonological links between 
Sln and West Slavic discussed by Greenberg 2000: 40-1). Moreover, s-Iz- has basic- 
ally the same scope and functions that it does in the West Slavic languages.29 The 
time frame of the coalescence, or the rise of s-/z- as a perfectivizer, is not clear; the 
retention of iz- in many sources in historical Sln obscures the picture. Merge (1995: 
168) observes that verbs prefixed with s-/z- entered into aspectual pair relationships 
'become freshlyoung'; these correspond in Cz to verbs prefixed with s-Iz- (cf., e.g., Cz zhorknout 
'become bitter'), or to doublets in s-Iz- and o- (cf., e.g., Cz ochladnoutPlzchladnoutP 'become cold'). 
Werner (2003: 147) notes that in Sor a few similar doublets exist, e.g., womjelknyC~lzmjelknyCP 'be- 
come silent', and points out that the correlate in wo- often expresses a figurative meaning (e.g., 
wobl2dnyCP 'grow pale (die)' versus zbl2dnyCP 'grow pale'). Occasionally Sor inchoatives in wo- ex- 
press a meaning of 'again' (cf., Werner 2003: 153), e.g., wosylniCP 'become strong (again)' vs. zesyl- 
niCP 'become strong'. Note that o-lob- occasionally has the same meaning in Cz, cf., e.g., omlad- 
noutp 'become young (again)' vs. zrnladnoutP 'become young'. It appears that Cz has generalized 
s-Iz- as a perfectivizer of inchoatives to slightly higher degree than Sor. According to Anan'eva 
(2003: 9), prefixation of inchoatives with o-lob- is an archaism in Pol (see section 2.2.3), and I 
think we may safely view prefixation of inchoatives with wo-lwob- in Sor as an archaism as well. 
29 The following description is drawn partly from Dickey (2003), which discusses the product- 
ivity of s-Iz- in Sln. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 23 
in the sixteenth century.30 Bajec (1959: 103) suggests that the original catalyst for the 
coalescence, the change *jbz- + z- resulting from jer-fall, has been continued in 
modern times by the Sln reduction of unstressed -i-, thus producing the change iz- 
+ z-. This may well be true, but in my view it is important to avoid any confusion 
as to the older nature of the coalescence (i.e., that its origin lies in the fall of the 
jers). There is clear phonetic evidence in favor of this view. The Prekmurje dialects 
have not reduced unstressed -i-, yet in these dialects *jbz- has nevertheless been re- 
placed by z-, resulting in a single prefix s-Iz-, which is clear from the following ex- 
a m p l e ~ : ~ ~  zgubitip 'lose' (+ *jbz-), zravnatip 'even' (+ *jbz-), skopatip 'dig up' (+ 
*jbz-), spisatip 'compose' (+ *s%-), zebratip 'collect' (+ *s%-). 
According to Bajec (1959: 112), the contemporary prefix s-/z- developed out of 
a coalescence of *s%- together with the eroded remnants not only of *jbz- but also 
of *VW- and *roz-. Evidence for this process involving the latter two prefixes is the 
existence of doublets such as razdrobitip-zdrobitip 'crumble/break apart' and vzra- 
stip-zrastip 'grow up' in sixteenth-century Slovene, cf. MerSe (1995: 167-168; 192). In 
my view, it is unlikely that coalescence of these prefixes in Slovene was the result 
of the phonetic erosion of vz- and raz- as opposed to a spread of z- at the expense 
of the other prefixes. I think this is particularly true in the case of raz-, as there is 
little, if any other evidence of the phonetic erosion of that prefix. Note that in Cz 
and Slk (where roz- seems to be quite stable) a few synonymous doublets in roz- and 
z- exist as well, e.g., Cz rozdrdsatp-zdrdsatp 'scrape up' (these verbs are synonymous 
according to the SSJC). It is more likely that in these cases s-/z- replaced raz- and 
roz- as a perfectivizer on the basis of the shared schema TRANSITION FROM S1 
TO S2. I suggest that this process has occurred in Sln, Cz and Slk only in a few 
verbs where there was a fair amount of semantic overlap between the base verb and 
the prefix (e.g., 'crumbling' and the idea of 'apart') and the lexical semantic loss 
resulting from the elimination of raz- or roz- in favor of s-Iz- was minimal. 
As in Cz, Slk and Sor, Sln, s-/z- has been highly productive with loan verbs. 
An older example is zglihatip 'even out'. A search of the OSSJ shows that s-/z- is the 
most productive prefix in the perfectivization of loan verbs, counter to Plotnikova's 
(1971: 35) view that iz- is the most productive prefix in this function. (The OSSJ at- 
tests 105 pf loan verbs prefixed with s-/z- compared with only 19 prefixed with iz-.) 
Examples are individualiziratii/P/zindividualiziratiP 'individualize', manipulira- 
tii/PlzmanipuliratiP 'manipulate', pakiratii/Plspakiratip 'pack', etc. A particular charac- 
teristic of Sln is that s-/z- often competes with iz- as the perfectivizing prefix for 
one and the same biaspectual loan verb, producing doublets such as balanciratii/p/iz- 
balanciratip-zbalanciratip 'balance', dijierenciratii/PlizdiferenciratiP-zdijierenciratip 
'differentiate', niveliratii/p/izniveliratiP-zniveliratip 'level'. When loan verbs occur with 
a prefix in the contemporary press, I have found them to be prefixed with z- and 
not iz-, e.g., zblaniiratip 'blanch', ziritiratip 'irritate', zmiksatip 'mix', etc. 
The same situation exists regarding native Slavic verbs. Where doublets exist for 
native Slavic verbs, for example, izpitiplspitip 'drink [up]' (+ pitii 'drink'), the con- 
temporary press prefers the variant in s-/~-; compare for example the following (re- 
30 It is worthing noting that Merge (1995) does not take the issue of aspectual pairs lightly. 
31 These examples were kindly provided to me from the card files of the Slovar stare knjiine 
prekmurctine by Dr. Majda Merge. 
24 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
duced) headline from the newspaper Novice: 
(5 )  Spilp kislino in si prerezal vrat. 
'He drank acid and slit his throat.'32 [Sln; Novice] 
The Sln informants I have spoken with are clear in their view that the (stan- 
dard) colloquial language only uses the variants in s-Iz-. These facts indicate not on- 
ly that iz- as an "empty" perfectivizer is at the very least restricted to very formal 
registers (or perhaps an artifact of prescriptivist tendencies in Sln dictionaries), but 
also that *s%- and *jbz- have in fact coalesced into s-/z-. The only exception is 
clearly elative iz-, which remains in the standard language, e.g., izhajatiilizitiP 'go 
out', etc. 
Sln s-/z- has also been productive in the perfectivization of inchoative verbs, 
cf., e.g., shujiatip 'lose weight', stemnitip se 'darken', zboletip 'fall ill' and zoglenetip 
'carbonize'. However, in Sln it has nevertheless not been as productive in this role as 
it has in Cz and Slk. Various inchoatives prefixed with s-/z- in Cz have equivalents 
in Sln that are prefixed with other prefixes, cf., e.g., Cz zesurovZtP and Sln posuro- 
vetip 'become wild', Cz ztichnoutp and Sln utihnitip 'quiet down'. Notably, many in- 
choative verbs that are prefixed with s - / ~ -  in Cz are prefixed with o-lob- in Sln, cf., 
e.g., Cz zestarnoutp and Sln ostaretip 'age', Cz zchromnoutp and Sln ohrometip 'be- 
come lame', Cz zvadnoutp and Sln ovenetip 'wilt', etc. Just as in the case of Sor (see 
fn. 28), we may consider the prefixation of inchoatives with o-lob- to be an archa- 
ism, and therefore as an indication that s-lz- has not been used to prefix such verbs 
to the extent that it has in Cz. On the other hand, it seems that newer (loan) verbs 
that fall into the inchoative class are prefixed with s-lz-, e.g., skristaliziratip 'crystal- 
lize'. Some older inchoative verbs have doublets in s-lz- and po-, primarily those de- 
noting the gaining of various colors, e.g., zbeletip-pobeletip 'whiten', and strne- 
tip-potrnetip 'blacken', though according to informants there does not seem to be 
any semantic distinction between them. 
The same situation holds regarding the prefixation of factitives. Some factitives 
in Sln are prefixed in s-/z-, as in Cz, cf., e.g., Cz zkiivitp and Sln skrivitip 'bend', 
whereas others utilize other prefixes, notably o-lob- (cf., e.g., Cz zkrvavitp and Sln 
okrvavitip 'bloody', Cz zlehtitp and Sln olajiatip 'make easy') and u- (cf., e.g., Cz 
zladitp and Sln uglasiti 'tune'). It is interesting that Sln does not have the tidy sys- 
tem of suffixation that Cz does (-2- for inchoatives, -i- for factitives), so that many 
Cz inchoatives in z- ...- 2- have Sln equivalents that are reflexive factitives, cf., e.g., 
Cz zbystietp and Sln zbistritip se 'become clear', Cz zkrvavZtP and Sln okrvavitip se 
'become bloody'. Likewise, doublets exist for some factitive verbs, e.g., zboljia- 
tiP-poboljSatiP 'improve', zdaljSatiP-podaljiatip 'lengthen', and zmanjiatip-pomanjia- 
tip 'reduce', though there does not seem to be any semantic distinction between these 
either. 
In any case, the diversity of predicate types that utilize s-lz- as a perfectivizer 
is noteworthy: s-12- has not only expanded its sphere of productivity at the expense 
of other prefixes (iz-, raz-, and vz-) in the class of transitive accomplishment predi- 
32 A search of the files of the Besedilni korpus for the newspaper Delo confirms this im- 
pression for izpitiPlspitiP: spitip yielded 17 hits, izpitip 0; among past-tense forms, spilp yielded 76 
hits, izpilp only 8. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 25 
cates, but has also been productive as a prdverbe vide for inchoatives and factitives. 
Furthermore, as mentioned above, it is now the most productive prefix for the per- 
fectivization of biaspectual loan verbs. In my view, these facts taken together indi- 
cate that Slovene has been taking steps towards grammaticalizing slz- as its perfec- 
tivizing prefix. 
2.2.3. Polish 
As shown in section 2, the coa!escence of *s%- and *jbz- occurred in Pol as well 
(note that Pol treatments, e.g., Smiech 1986, treat s-Iz- as a single prefix, like the 
Slk and unlike the Cz literature). Klemensiewicz, Lehr-Splawiriski and Urbanczyk 
(1981: 251) observe that the phonetically-conditioned alternation between s- and z- 
existed in the sixteenth century. This puts the coalescence in Pol more or less at the 
same time as Cz, though if, as Slosar (1981: 106) suggests, s-Iz- was already estab- 
lished as a prdverbe vide in Cz by the beginning of the sixteenth century, it is pos- 
sible that the development took place slightly later in Pol. 
Szelesifiski (1972: 226-7) discusses doublet formations involving wz- and s-Iz-. 
He notes that there are a few dozen such doublets, e.g., wspuchnqkp-spuchnqkp 
'swell up'. He points out that the fact of the existence of a doublet for a given lex- 
eme does not necessarily mean that the doublets arose due to the phonetic erosion of 
wz-; rather, some doublets arose due to prefixation of the same lexeme with both 
wz- and s-Iz-. Szelesinski (227) makes no specific claims about which process was at 
work with individual verbs, but points out that the existing doublets in which the 
prefix adds a meaning of ingressivity, e.g., wzbakp sie-zbakp sic 'begin to fear', 
probably arose as the result of the phonetic reduction of wz- to z-, as the meaning 
involved (ingressivity) is one characteristic of wz- in the various Slavic languages. On 
the other hand, he suggests that the doublet wzbudzikp-zbudzikp arose not due to 
erosion (i.e., wzbudzikp + zbudzikp), but from parallel prefixation (budzikp + wzbu- 
dzikp, zbudzikp); here the aspectual meaning is one of resultativity (and not in- 
gressivity), for which according to Shull's hypothesis one can expect different telic 
prefixes will each profile the TRANSITION FROM S1 TO S2. 
Pol s-Iz- has been very productive as a perfectivizer. Agrell (1908: 85) suggests 
it is the primary prdverbe vide in Pol. It is clear that s-Iz- is the most productive 
prefix in the perfectivization of loan verbs: Guiraud-Weber (1998: 75) observes that 
S-12- is the currently most productive prefix in the perfectivization of loan verbs in 
Pol, and also points out that there are many loan verbs prefixed with s-12- are at- 
tested in various publications but which are not included in contemporary dictionar- 
ies. This again is-an indication of the productivity of s-Iz- as a perfectivizer in Pol. 
But in contrast to Cz, Slk, Sor and Sln, s-Iz- does not appear to have been the most 
productive perfectivizing prefix in modern Pol overall. Smiech (1986: 9) lists po- as 
the most productive prefix. Anstatt (2003b) also suggests that po- has in fact been 
the most productive perfectivizing prefix in Pol. She bases her opinion on Cockie- 
wicz (1992: 96ff.), but adds that he limits his statistics to "the most frequent Polish 
verbs" and that an examination of Saloni (2001) reveals that z- is used to perfectiv- 
33 One area that shows the higher productivity of po- in Pol than in Cz is the relatively very 
productive derivation of delimitative verbs (e.g., poczytaCP 'read for a while'; cf., Dickey and 
Hutcheson 2003). 
26 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
ize loan verbs "much more frequently" than po-. Such a state of affairs is somewhat 
odd: one would expect that s-/z- would attain a very high degree of productivity in 
a language before it became productive in the perfectivization of loan verbs. In any 
case, it is clear that s-Iz- and po- have both been quite productive in 
Another important difference between Pol and Cz (and Slk) is that in Pol s-/z- 
has not been quite as productive in the perfectivization of inchoative verbs, which is 
reminiscent of the situation in Sln. Though many Cz inchoatives (preliminary 
counts indicate about 50%) in s-/z- have Pol equivalents in s-/z-, in a considerable 
number of cases Cz s-Iz- corresponds to another prefix in Pol. The following short 
lists give an impression: 
(6) a. Cz Pol 
zblednoutp zblednqCP 'grow pale' 
ztervenatp sczerwienieCP 'turn red' 
zd?evnatttP zdrewnieCP 'become woody' 
zhrubnoutp zgrubieCP 'coarsen' 
ztichnoutp ScichnqCP 'become quiet' 
b. zbohatnoutp wzbogaciP sic 'get rich' 
zesrnutnttP posrnutnieP 'become sad' 
zrnlknoutp urnilknqdp 'fall silent' 
zobecnttp upowszechniP sic 'become widespread' 
zpustnoutP opustoszeCp 'become desolate' 
zeSiletP oszaledp 'go crazy' 
As these few examples suggest, in addition to s-Iz-, Pol inchoatives tend to be pre- 
fixed with u-, po- and o-. Anan'eva (2003: 9) observes that in OPol the prefix 
o-/ob(e)- perfectivized denominal inchoatives to a greater extent than today. Thus, 
the perfectivization of inchoatives in Pol with o-lob- (as was suggested for Sor and 
Sln) is an archaism, and innovative prefixation with s-/z- is in fact one of the 
reasons for its lower frequency in contemporary Pol. 
Like Cz, Pol attests doublets inchoative doublets prefixed in s-Iz- and po-. 
Agrell (1908: 734) suggests that s-Iz- expresses an absolute change and po- a relative 
change. Yet Anstatt's (2003b) description of Pol po- indicates that this is not entirely 
true; she points out that according to Piernikarski (1975: 61) Pol inchoatives in both 
po- and s-/z- may in principle occur with troche 'a little' and zupelnie 'completely', 
thus expressing either a relative or absolute change. Thus, it appears that Pol does 
not make the systematic distinction between attenuativity expressed by po- and 
completeness of change expressed by s-/z- that one finds in Cz. Anstatt also observes 
that po- tends to have a resultative meaning with factitive verbs, so that here as well 
there seems to be no consistent semantic opposition between po- and s-Iz-. 
Thus, we may conclude that, while Pol shares the development of s-Iz- with the 
western languages, it departs from the western type slightly by virtue of the fact 
that s-Iz- does not appear to be its clearly dominant priverbe vide, but competes in 
this regard with po-, which has been more or less as productive as s-/z-. This situ- 
ation recalls not only the transitional patterning of Pol in respect of aspect usage 
but also its transitional status in two other areas of aspect morphology: as observed 
by Dickey (2000: 227) it patterns in between the eastern and western languages 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 27 
with regard to the derivation of ingressive verbs with za-, and as shown by Dickey 
(2001b) it also patterns between the western and eastern groups with regard to the 
role -nq- has played in the derivation of perfective verbs. And so the transitional 
patterning of Pol with regard to s-/z- should not be surprising. 
3. s-/z- in East Slavic and the Other South Slavic Languages. 
Section 2 described a fairly compact group of western Slavic languages in which 
s-Iz- has enjoyed very high productivity as a perfectivizing prefix: Cz, Slk, Sor, Sln. 
Pol also belongs to this group, though the productivity of s-/z- there is not quite as 
high as in the other languages. The following sections show that in the East Slavic 
languages and as the other South Slavic languages (CroISrb, Blg and Mac) s-/z- is ei- 
ther less productive than in the western group (Ukr and Blr) or the coalescence nev- 
er took place, with iz- functioning as a resultative and perfectivizing prefix to vary- 
ing degrees (Rus, Cro/Srb, Blg, Mac). 
3.1. East Slavic 
In Ukr and Blr *jbz- became z- as in West Slavic and Sln, creating the conditions 
necessary for the coalescence and the creation of s - I z - .~~  Rusanivs'kyj (1978) treats 
the development of *jbz- and *sa- in Ukr from the eleventh to the twentieth centu- 
ry. In Old Rusian (from the eleventh to the thirteenth centuries) iz- and sa- were 
still phonetically distinct and thus separate prefixes. Though minor differences surely 
existed, for purposes of this discussion I see no harm in assuming that iz- and sa- 
in Old Rusian were essentially as they were in OCS (cf., section 2.1). In OUkr (the 
fourteenth-fifteenth centuries) the process of coalescence was ending, so that accord- 
ing to Rusanivs'kyj (250) iz-/is- was a phonetically conditioned alternate of z-1s- be- 
fore roots beginning in a consonant, cf., e.g., sljubilip (1433) vs. isljubujemai (1352) 
and smolviviiP se (1404) vs. ismovivSiP se (1434). It is hard to interpret the overall 
significance of the few forms that Rusanivs'kyj gives; in my view they indicate that 
in the fourteenth and fifteenth centuries iz- and s- were still in the process of coa- 
lescing. As in the western languages s-/z- has spread at the expense of original 
*vaz- in Ukr (cf. Bogdanova 1963: 301 as well as ~ e r e x  1951, who lists 'upward mo- 
tion' as one of the meanings of z-), cf., e.g., zbrojityp 'arm', zbudytyp 'wake up', zle- 
tityp 'fly up'. Thus, Ukr displays the same basic coalescence of original *sa-, *jbz-, 
and *vaz- that West Slavic and Sln do. 
However, it seems that innovative s-/z- has played a less significant role as a 
perfectivizer in Ukr. This is evident from the fact that s-Iz- has not become the 
primary perfectivizer of inchoative verbs in Ukr. Cz inchoative verbs in s-Iz- often 
have correlates in Ukr prefixed with other prefixes: cf., e.g., Cz zbohatnoutp vs. Ukr 
rozbagatityp 'become rich', Cz ztervenatp vs. Ukr. potervonityp 'turn red', Cz zter- 
natp vs. potornityp 'turn black', zdrsnztp vs. Ukr poierxnutyp 'become course', Cz 
zeslabnoutP vs. Ukr oslabnutyp 'weaken', Cz zestarnoutp vs. Ukr postarityp 'age', Cz 
zeiloutnoutP vs. poiovtityp 'become yellow', Cz zhnzdnoutp vs. Ukr poburityp 'be- 
come brown', Cz zmladnoutp vs. Ukr. pomolodityp 'become young', etc. Ukr diction- 
aries attest some of the above inchoatives prefixed with s - /~ - ,  e.g., zbahatityp 'be- 
come rich', ztervonityp 'turn red', zmolodityp 'become young', etc., but internet 
34 The same coalescence occurred with the prepositions as well, yielding z. 
28 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
searches show them each to be used much less than their correlates in other prefix- 
es, esp. in po-. The picture is much the same for factitives, though here s-Iz- seems 
to be even less common in Ukr: cf. Cz zdivotitp vs. Ukr zrobytyp dykym 'make 
wild', Cz zdraiitp vs. Ukr zdoroiytyp 'raise the price', Cz zeslabitp vs. Ukr oslabytyp 
'make weak', Cz zesmziitp vs. Ukr osmijatyp 'mock', etc. A few factitives are attested 
in s-Iz- as well as another prefix; an example is zgiriytyp and pogiriytyp 'make 
worse', and in this case the former is used much less frequently than the latter. 
Thus, it is not surprising that in contrast to Cz, where s-Iz- had become firmly 
entrenched as a perfectivizer by the beginning of the fifteenth century, Rusanivs'kyj 
(251) observes that in the sixteenth and seventeenth centuries the frequency of s-lz- 
lagged behind that of several of other prefixes, namely u-lv.6-, po-, pri-, nu-, za-, 
and vy- (here Rusanivs'kyj is speaking of both as a lexical and aspectual prefixa- 
tion). He also observes (252) that at this time the voicing assimilation of s-Iz- (in- 
cluding zo-, iz-/is-) was still an ongoing process. This suggests that Ukr completed 
the phonetic coalescence of *s5- and *jbz- somewhat later than Cz and Pol (and Pol 
completed it no earlier than Cz, perhaps slightly later).35 
As far as perfectivizing prefixation is concerned, Rusanivs'kyj (278) observes 
that in the fourteenth and fifteenth centuries the most productive perfectivizing pre- 
fix was U-IVE, followed by po-, vy- and only then s-Iz-. Rusanivs'kyj (280) also ob- 
serves that during the sixteenth-eighteenth centuries po- rose to become the most 
productive perfectivizing prefix, followed by u-lv-; s-Iz- was only the third most 
productive in the role of a perfectivizer. Note again that this situation contrasts 
greatly with the situation in Cz, in which s-Iz- was very productive as a perfectiviz- 
ing prefix from the fourteenth century on, whereas according to Slosar (1981: 128) 
the productivity of po- as a perfectivizer declined after the fifteenth century, and 
as a consequence, as Poldauf (1954: 64) observes, po- is not currently productive as 
a perfectivizing prefix in Cz. According to Rusanivs'kyj (1978: 280), it is only in the 
latter half of the nineteenth century that s-Iz- becomes the most productive perfec- 
tivizing prefix, followed by po- and za-. Note that there is some disagreement here, 
as Serex (1951: 278) and Pugh and Press (1999: 205) consider po- to be the currently 
most productive Ukr prefix (followed by s-Iz-); it is possible that western and eastern 
Ukr differ in this respect. If, as suggested above, *s5- and *jbz- coalesced in Ukr 
later than in Cz, it is possible that the lower productivity of Ukr s-Iz- in comparison 
to Cz is a result of the later date of the coalescence, as po- could have strengthened 
as a prdverbe vide in the meantime. 
I am not aware of any historical examination of prefixation in Blr, but accord- 
ing to AtraxoviC and Bulaxaw (1962: 328) as well as Bulyko, et al. (1990: 199) pa- is 
the most productive perfectivizing prefix in Blr, followed by s-Iz-. This indicates 
more or less the same situation as in Ukr, i.e., that the productive perfectivizing po- 
has prevented s-Iz- from clearly dominating the system of empty perfectivization, or 
35 Marvan (2000: 293-302) assumes the existence of a "North-Slavic Sprachbund" in which 
certain features were gradually transmitted by language contact from west to east, and among 
them includes "z- as a prkverbe vide". Though he does not elaborate this idea, it does seem plausi- 
ble, at least in the case of s-Iz-. Interestingly, the SSM lists a few verbs prefixed in z- that were 
loaned into Ukr from OPol, e.g., zvesty 'take to' ( t  OPol zwieic?, zrnertyp 'die' ( t  OPol 
zernrzekp), zrnyslytyp 'imagine' ( t  OPol zrnySliCP). 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 29 
even becoming the primary prkverbe vide at all. Important in this respect is Atra- 
xoviE's and Bulaxaw's observation that Blr pa- is particularly productive in the per- 
fectivization of inchoatives, e.g., pabahacec'p 'get rich', pagrubec'p 'become coarse', 
paslabec'p 'become weak', etc. This situation is clearly different from the situation 
in Cz, where, as has been pointed out repeatedly, s-Iz- is the dominant perfectivizer 
of inchoatives. 
As for Rus, it never underwent the coalescence of *s5- and *jbz-. Common 
Slavic *jbz- remained as iz-, which in its elative meaning has been replaced by vy- 
in historical times. However, iz- has not been as marginal in Rus as is sometimes 
suggested, nor has it been strictly a product of South Slavic influence. Avilova (1964: 
48-50) observes that vy- finally replaced iz- relatively late, in the nineteenth centu- 
ry, and that iz- remained productive in the creation of certain kinds of resultatives: 
verbs with a nuance of intensity, e.g., izbit'p 'beat up'; verbs expressing a harmful 
result, e.g., izglodat'p 'gnaw'; verbs expressing the full affectedness of an object by 
the action, e.g., iscelovat'p 'kiss all over'. On the other hand, po- has been very 
productive in Rus in historical times. According to Xlebnikova-ProkopoviE (1956: 
154), in the seventeenth century po- was among the three most productive perfectiv- 
izing prefixes in Russian, and she points out (140) that "in the language of the 
second half of the seventeenth century, the prefix po- also enjoyed widespread use 
as a purely perfectivizing prefix, occupying a significant position among the prefix- 
es that served as simple perfectivizers." Note that Xlebnikova-Prokopovii: (139) expli- 
citly excludes delimitatives as pf partner verbs, which naturally lowers her assess- 
ment of the productivity of po- as a prkverbe vide. Contemporary Rvs is also char- 
acterized by high productivity of perfectivizing po-: according to Certkova (1996: 
123-24), pc- is the most productive perfectivizing prefix in contemporary Russian 
(note that Certkova does include delimitative po- in her assessment). Thus, although 
Rus did not develop s-/z- as a prkverbe vide along with Ukr and Blr, it does share a 
high productivity of perfectivizing po- with them, and we may consider productive 
perfectivizing po- to be one of the defining features of East Sla~ic.3~ 
3.2. CroatianISerbian, Bulgarian and Macedonian 
In sections 2 and 2.2.2, it was shown that Sln has perfectivizing s-/z- like the West 
Slavic languages. In this respect Sln differs from the other South Slavic standard 
languages. However, among Cro dialects the Kajkavian dialects show the same coa- 
lescence of original *s%- and *jbz-, which is not surprising given the overall similar- 
ity and proximity of Kajkavian Cro to Sln. Peco (1991: 256) observes that a feature 
shared by all Kajkavian dialects is the change "iz + z", cf., e.g., z hiie 'out of the 
house' and zoraP 'plowed' [+ izoral]. Examples of forms that show the complete coa- 
lescence of *s%- and *jbz into s-12- in the Bednjanski Kajkavian dialect can be 
found in the texts given by Jedvaj (1956; I have omitted the vocalic diacritics): zver- 
SiliP 'finished' [+ *s5-1, zvejolp 'winnowed' [+ *jbz]-, skaopolip 'dug outlup' [+ 
*jbz]-. Interestingly, a few Cakavian dialects have also merged *s5- and * j b ~ - ? ~  
36 Thus far I have not fully addressed the function of po- to derive delimitatives with regard 
to its status as a pre'verbe vide; this issue is taken up in section 5. 
37 Whether the Cakavian dialects that have merged *sa- and *jbz- (e.g., the dialects of Kastav, 
cf. Peco 1991, and OrbaniCi in Istria, cf. Kalsbeek 1998) all originated in the northern region of 
30 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
Thus, apart from Kajkavian and isolated Cakavian dialects, Cro and Srb have 
not merged *sa- and *jbz-; they did not reduce the latter to z- as a consequence of 
jer-fall, rather *jbz- becomes iz-, as in Blg and Mac. Although the South Slavic lan- 
guages are therefore largely irrelevant for this discussion, it is worth noting that in 
Cro, Srb, Blg and Mac iz- has remained a very productive resultative and perfectiv- 
izing prefix. For instance, loan verbs that are biaspectual in standard Srb and Cro 
readily derive pf forms prefixed with iz-, e.g., istreniratip se 'train', istuiiratip 'show- 
er', izmanevriratip 'maneuver', izmanipuliratip 'manipulate'. Indicative of the produc- 
tivity of iz- as a resultative perfectivizer is the fact that many are colloquial and not 
listed in dictionaries, e.g., Cro izlifratiplSrb izliferovatiP 'deliver', izluftatip 'ventilate', 
izorganiziratip 'organize', etc. Note that such loan verbs prefixed with iz- have strong 
distributive senses (akin to distributives prefixed in po-), so that they occur chiefly 
in contexts where all of a set of objects are to be affected, e.g., izorganizirati sve 
'organize e~erything ' .~~ Blg and Mac also make wide use of iz- in various aspectual 
functions (cf., e.g., Stojanov 1993: 218-9 and Ugrinova-Skalovska 1960: 659). Ivanova 
(133) observes that iz- is statistically the most frequent "empty" prefix in Blg. 
While Blg and Mac have not developed s-/z- as a prdverbe vide, the situation 
regarding po- in these languages is more complicated than in East Slavic. Though 
in Blg po- is productive to some degree as a resultative prefix (cf., Stojanov 1993: 
222), Ivanova (1966: 124, 133) suggests that po- does not function as a prdverbe vide 
as frequently as in some other Slavic languages (e.g., Rus). For instance, po- is not 
as productive in the derivation of inchoatives in Blg as it is in Rus, cf., e.g., Rus 
pozelenet'p and Blg pozelenejap 'turn green', but Rus postaret'p vs. Blg ostarejap 
'age'. On the other hand, however, Ivanova (124) points out that Blg po- is "semanti- 
cally markedly bleached" (she does not characterize any of the o t h e r - ~ 1 ~  perfectiviz- 
ing prefixes in this way), and (132) presents word counts indicating that more verbs 
are prefixed with po- in Blg than any other perfectivizing prefix. Though the dis- 
crepancy can be accounted for by discounting various procedural verbs (e.g., attenu- 
atives, delimitatives, etc.), it seems that Blg po- must have primarily aspectual func- 
tions. (It is puzzling that she does not include a verb as common as popitamp 'ask' 
as a case of empty perfectivization in her short list of po- perfectives.) While it 
seems clear that po- is not as productive in Blg as it is in East Slavic, it nevertheless 
occurs much more frequently as a perfectivizer than in Srb. For this reason, and al- 
so because of its productivity as a delimitative prefix (cf., Stojanov 1993: 221, and 
Dickey and Hutcheson 2003) I will consider Blg, to be a language which is or has 
been characterized by a relatively high productivity of po- as a perfectivizing pre- 
fix. As for Mac, Ugrinova-Skalovska does not specifically address the productivity of 
po- as a prdverbe vide, but points out that "with an enormous number of verbs po- 
functions simply to perfectivize them, as a resultative [prefix]" (88). Ugrinova-Ska- 
lovska (86) also observes that po- is highly productive as a delimitative prefix, to 
the point where "almost any verb, either simplex or derived, may be used with po- 
in that meaning." Thus, I likewise consider Mac to be a language characterized by 
Croatian linguistic territory in close proximity to Kajkavian and Slovene is an issue which cannot 
be investigated here. 
38 This is not to say that these verbs do not occur without objects quantified by sav 'all', but 
when they do they nevertheless have a distributive sense, so that izorganiziratiP konferenciju 'or- 
ganize the conference', means in fact something like 'organize the conference from a to z'. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 31 
productive perfectivizing po-. 
Section 3 has shown that East and South Slavic have utilized s-/z- as a prd- 
verbe vide to a considerably lesser extent than Cz and the other western languages 
(including Sln). Though Ukr and Blr do have s-/z- as a hybrid prefix, it has not 
played the role in these languages that it has in Cz or even Pol. The remainder of 
South and East Slavic, i.e., Cro, Srb, Mac, Blg and Rus, has not developed s-/z-. 
Ukr, Blr, Rus, Blg and Mac all share productive perfectivizing po-, though East 
Slavic appears to make use of it slightly more than Blg and Mac. In Srb and Cro, 
on the other hand, po- is not particularly productive as a perfectivizing prefix, cer- 
tainly much less so than in Blg and Mac. 
4. The Distribution of s-/z- and the East-West Aspect Theory 
In the previous sections it has been shown that s-/z- became the primary prdverbe 
vide in a western group of languages, i.e., Cz, Slk, Sln and  SO^.^^ The coalescence 
also took place in Pol, Blr and Ukr, but in these languages s-Iz- nevertheless did not 
attain primary status as a prdverbe vide very early, if at all, as it faced rather stiff 
competition from the other major Slavic perfectivizing prefix, po-. Pol may be 
characterized as "mixed" with regard to productive prdverbes vides, since in Pol 
s-Iz- acquired considerable productivity early on (though not to the degree of Cz and 
Slk) while po- also enjoyed considerable productivity. In Blr and Ukr the total coa- 
lescence of s- and iz- was apparently completed later than in the west, and in any 
case the competition with po- was even stiffer, so that s-/z- became the most pro- 
ductive prdverbe vide only very late (i.e., in the latter half of the nineteenth cen- 
tury in Ukr), or has nevertheless remained less productive than po- (i.e., in Blr). As 
pointed out in section 3, Rus is also characterized by highly productive perfectiviz- 
ing po-. 
The languages characterized by s-/z- as a prdverbe vide are basically the same 
languages that belong to the western aspectual type according to Dickey (2000). Pol, 
while also having s-Iz-, is not considered by Dickey to be a member of the western 
group, but rather an aspectual transitional zone. As suggested in section 2.2.3, we 
may make sense of this in terms of the correlation with s-/z- by keeping in mind 
the fact that in Pol s-/z- has not been the only productive prdverbe vide, but has 
competed in this role with po-, which has also been very productive, if not more so. 
The languages of the eastern group, i.e., Rus, Ukr, Blr, Blg and Mac, are disting- 
uished by the fact that po- has been highly productive as a perfectivizer in each 
since the sixteenth century. This is regardless of whether they developed s-/z- (Ukr, 
Blr) or not (Rus, Blg, Mac4'). 
In this respect it must be kept in mind not only that s-/z- has generally been 
39 Sor has either dismantled (according to Breu 2000) or transformed (according to Toops 
2001) its aspect distinction in the relatively recent past. Though originally perfective prefixed 
verbs now exhibit patterns of usage that may be characterized as "biaspectual" with regard to 
typical Slavic aspect usage, I consider it beyond doubt that before the dismantling/transformation 
of the its aspect category Sor belonged to the western aspectual type. Therefore it is grouped in 
with the western languages in this discussion. 
40 Mac is not treated by Dickey (2000), but my informal examinations of its aspect usage 
confirm that it displays basically an eastern pattern of usage. 
32 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
more productive in the west, but also that po- has not been more productive than 
s-lz- in any of the western languages. As pointed out above, Cz po- is currently not 
productive as a perfectivizing prefix; its productivity has steadily declined since the 
fifteenth century. Slovak sources differ slightly: the MSJ (414) lists po- last among 
the prefixes that serve as prkverbes vides, i.e., s-lz-lzo-, u-, nu-, -za-, o-, po-, 
whereas Smirnov (1970: 78) ranks it fourth after s-lz-lzo-, za-, and u-. In any case, 
po- is not productive enough to compete with s-lz- at all. Merge (1995: 197) observes 
that in sixteenth-century Sln po- was second to s-lz- as a prkverbe vide; my im- 
pression is that s-lz- is currently far more productive than po- in Regarding 
the productivity of po- in the western languages, Pol and CroISrb, it should be 
pointed out that, as discussed by Dickey (2001a), in these languages po- has re- 
tained its spatial SURFACE-CONTACT meaning, e.g., Cz poblinkatp 'puke all over', 
Slk pomotitp 'pee all over', Sor pocokrowakp 'sugarcoat', Sln pokozlatip 'vomit all 
over', Pol popluskakp 'splash all over', CroISrb popiiatip 'pee all over'. This is in 
contrast to the eastern languages, where po- has lost the SURFACE-CONTACT 
meaning, and no longer has any clear spatial or lexical meaning (see section 5). 
We may therefore generally oppose the productivity of s-12- in the extreme 
western languages to the productivity of po- in the languages of the eastern group, 
again regardless of whether they too merged *s5- and *jbz- (Blr, Ukr) or not (Rus, 
Blg and Mac). The remaining languages, Cro and Srb, are not characterized by 
either s-Iz- or po- as a dominant prdverbe vide. In this respect, Cro and Srb are 
aspectually conservative. If we take into account that with regard to aspect usage 
Cro and (to a lesser extent) Srb closely resemble the western type, it becomes clear 
that, inasmuch as we can correlate aspectual morphology to usage type, it is in fact 
productive po- that positively correlates to the eastern pattern of aspect usage, 
whereas the western pattern includes languages with productive s-12- (Cz, Slk, Sln, 
Sor) and without it (Cro, Srb). Note that the actual current situation regarding the 
productivity of perfectivizing po- is not of primary importance; rather, what is 
important is that it has been the dominant productive prkverbe vide in the relatively 
recent past (for instance, since the sixteenth century). Prefixation with po- may be 
tied to eastern aspect usage in another respect as well: as Dickey and Hutcheson 
(2003) demonstrate, the languages of the eastern type are the same languages that 
have developed po- as a productive delimitative prefix. The following section will 
consider these overall correspondences with regard to the grammaticalization of 
aspect in the various languages. 
5. Innovative s-/z- and the Grammaticalization of Slavic Aspect 
As Mende (1999: 286) observes, the rise of grammatical aspect systems in Rus and 
the Slavic languages in general is not typical of the phenomena customarily exam- 
ined in grammaticalization studies. Most case studies of grammaticalization (cf., 
41 NO assessments are available of the relative productivity of the prefixes, other than Plotni- 
kova's (1971: 356) observations on the prefixation of loans verbs. She suggests that po- is the 
fourth most productive prefix in the perfectivization of loan verbs, after iz-, s-/z-, and pre-. My 
disagreement with her on the relative productivity of iz- and s-/z- is irrelevant here. I would point 
out that her observation that po- primarily derives delimitatives from loan verbs in fact indicates 
that po- is not particularly productive with loan verbs, as po- is on the whole not very productive 
in the derivation of delimitatives in Sln (cf. Dickey and Hutcheson 2003). 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 33 
e.g., Hopper and Traugott 1993, Lehmann 1995) trace the development of single lexi- 
cal units into markers of grammatical meaning (a representative case might be the 
evolution of English go from a verb of motion to a future auxiliary). Studies of the 
development of aspect in Rus, e.g., Bermel (1997), Nsrgird-Ssrensen (1997) as well 
as Mende (1999) must necessarily deal with the rise of a grammatical category 
which is not expressed by a single marker, but by a whole array of them (for exam- 
ple, Mende lists 12 prefixes which commonly function as prkverbes vides in Rus). In 
this respect, the very nature of Slavic aspectual systems defies a simple application 
of the principles and precedents of grammaticalization theory. Thus, according to 
Mende (286-7), 
[tlhe grammaticalization of aspect consists of the rise of a whole category by means of 
derivational processes. At the risk of overstating the point, we may say that the develop- 
ment of Russian aspect involves a process whereby there are as many grammaticalized 
units as there are verbs-based on the derivational "doubling" of (almost) all verbs. The re- 
sulting grammatical status of these verbs is solely a consequence of the functional opposi- 
tion evident between the source verbs and their derived correlates, and not of any regular 
morphological paradigm. Thus, there is no single unit that is grammaticalized, but rather 
numerous lexical units (i.e., all verbs) together come to comprise a hitherto nonexistent 
grammatical category without losing their lexical status at the same time. [Original emph- 
asis, translation mine-SMD.] 
Here Mende is following the approach taken by Bermel, who, in a meticulous exam- 
ination of Rus verb usage and morphology in texts ranging from the eleventh to the 
seventeenth centuries, demonstrates that Rus aspect developed into a fully fledged 
grammatical category by the spread of the pf-impf opposition from the class of 
non-punctual telic predicates (i.e., Vendler's accomplishments) to punctual telic predi- 
cates as well as atelic predicates (cf. Bermel 1997: 463). Bermel, Nsrgird-Ssrensen 
and Mende all emphasize imperfectivizing suffixation as a key step in the estab- 
lishment of aspect as a grammatical category in Rus (Slavic), following Maslov 
(1961). While I agree wholeheartedly with this view, it must be emphasized that the 
mass extension of the distribution of the aspect opposition in the Slavic languages by 
means of imperfectivizing suffixation firmly entrenched the impf-pf opposition qua 
opposition, but played no exclusive role in determining the precise meanings of the 
semantic categories being created. In this respect, I think the role of prkverbes 
vides in the grammaticalization of aspect has been somewhat neglected, as I would 
argue that such prefixation has played a crucial role not only in formally creating 
one member of the aspect category (i.e., the pf aspect), but has also finely tuned the 
meaning of the pf aspect in the various languages.42 
In order to give this hypothesis adequate consideration, I think it is necessary 
to reexamine some assumptions of recent work on the grammaticalization of Rus 
aspect. Bermel, Nsrgird-Ssrensen, Mende and Lehmann (1999) study the grammati- 
calization of Rus aspect by focusing on the gradual rise of the pf-impf aspect oppo- 
sition as a grammatical category, and are thus concerned with the timing and over- 
all processes which led to the establishment of aspect as one of the obligatory 
42 This is not meant to deny the role played by the semantic nature of the productive imper- 
fectivizing suffixes in the respective languages. But since all of them (e.g., Rus -yva-I-iva-, Cz 
-dva-) have parallel origins as iterative morphemes, they may be ignored for some examinations of 
the east-west aspect division. In any case, they are beyond the scope of the present discussion. 
34 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
grammatical categories of Russian. They do not approach the rise of the impf-pf 
aspect opposition in terms of the grammaticalization of individual morphemes. As 
far as I can tell, each of these investigations tacitly assumes that the current func- 
tional division between the pf and impf aspect in Rus (i.e., the impf aspect predom- 
inating in the expression of events in process, habitual events, the simple confirma- 
tion of events in the past, whereas the pf aspect is largely restricted to the ex- 
pression of single, completed events) is somehow the expected result of the (near) 
maximum grammaticalization of an aspect opposition expressing a totality distinc- 
t i ~ n ; ~  though this idea is strongest in Mende's and Lehmann's discussions. In my 
view, this line of thinking is open to question, because it almost invariably leads to 
the conclusion that aspect in Slavic languages such as Cz (in which the functional 
division between the pf and impf aspect differs in some important ways from Rus, 
as pointed out in section 1) has not been grammaticalized to the degree of Rus 
aspect. Such a view may well eventually prove accurate or useful, but it does not 
change the fact that Cz has a stable system of aspect usage based on a distinction in 
totality-its system of usage is simply different from that of Rus. It is understandable 
that the system of usage in modern Rus would be considered the "finish line" in a 
grammaticalization study examining only Rus, but Lehmann (1999: 225) suggests 
that this kind of model is valid for all of North Slavic (though he nevertheless con- 
siders only Rus data). 
To illustrate the problem, let us consider the fact that western languages such 
as Cz allow the pf in habitual contexts to a much higher degree than eastern 
languages such as Rus (cf., section 1, as well as Dickey 2000: 49 and the references 
cited there). Similarly, older stages of Rus allowed the pf in habitual contexts to a 
greater degree than modern Rus, gradually limiting it in this function (cf. Bermel 
1997: 204, 280, 404, as well as Dickey 2000: 282 and the references cited there). If 
the overall similarity of the situation regarding aspect and habituality in modern Cz 
and ORus may be taken as likewise indicating an overall similarity of their aspect 
systems (which I think is a fair assessment, given several other points of similarity 
between aspect in ORus and modern Cz, cf., Dickey 2000: 282-3), then it is arguab- 
ly misguided to assume an acute lack of systematicity in the ORus aspectual system. 
This should not be understood as a reactionary approach favoring older views of a 
"fully developed" aspectual system in Late Common Slavic or the early historical 
Slavic languages; rather, it simply means that one should be cautious of going very 
far in the opposite direction, assuming a kind of purely lexical category in ORus 
(and the other Slavic languages). In other words, we are not limited to viewing the 
grammaticalization of Rus aspect as the progression from a lexical totality-based 
category to a grammatical totality-based category; we may also assume the exist- 
encelrise along the way of grammatical meanings distinct from the meanings of the 
impf and pf in the final, grammaticalized state. In the case of Rus, an alternate 
approach would be to assume the progression from a lexical totality-based category 
to via an increasingly grammatical totality-based category to a grammatical cate- 
gory based on a distinction in temporal definiteness (sequentiality). 
Thus, my concerns do not amount to a rejection of the approach shared by 
Bermel, Norgird-Sorensen, Mende and Lehmann-or more precisely, the picture that 
43 That is to say, by expanding the system of aspectual pairs of verbs to a maximum. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 35 
emerges when these studies are taken together. Rather, I would simply argue that in 
a theory of the grammaticalization of Rus (Slavic) aspect emphasizing the lexical 
expansion of both the pf and impf aspects on the one hand, and the overall expan- 
sion of the aspect opposition on the other, there ought to be room for a focus on 
the particular semantic development of the (proto-)pf aspect in these languages, es- 
pecially in the context of the east-west aspect theory. Accordingly, I suggest that it 
may prove fruitful to analyze the semantic development of the pf aspect in the west 
and east in terms of the perfectivizing prefixes dominant in each group (respective- 
ly). This approach involves the concept of grammaticalization in its more traditional 
sense, i.e., investigating the distribution of individual morphemes to see which, if 
any, have been grammaticalized as markers of the pf aspect, and leads to the fol- 
lowing question: what significance did the grammaticalization of s-Iz- as a marker 
of perfectivity have for the development of aspect in western Slavic? 
In order to shed light on this issue, I believe a comparative cross-Slavic ap- 
proach is still most productive. Therefore, let us briefly recapitulate the evidence re- 
garding the status of s-/z- and po- as the dominant prkverbes vides in the respective 
languages. Data from Cz and Rus will be taken as more or less representative of the 
situations in the western and eastern groups, respectively. The prefixes s-Iz- and po- 
share some important traits, which are summarized in (7-8): 
(7) In Cz, s-/z- displays the following traits: 
a. It is the only prefix which does not have a central spatial meaning (cf., 
e.g., Rusinovh 2001: 224). 
b. As pointed out in section 2.2.1, it forms many alternate pf verbs along- 
side other prefixes, e.g., zemiitp-umiitp 'die'. 
c. As pointed out in section 2.2.1, it has perfectivized the largest range of 
predicate types, including ordinary telic verbs (e.g., zbudovatp build), 
inchoatives (e.g., zfamfrnttP 'go crazy') and factitives (e.g., gamf rnitp 
'drive crazy'). 
d. It remains highly productive, notably in the perfectivization of loan 
verbs (e.g., znormalizovatP 'normalize'). 
(8) In Rus, po- displays the following traits: 
a. It is the only prefix which does not have a central spatial meaning (cf., 
Voloxina and Popova 1997: 379, Camus 1998: 101, and Shull 2003: 
158)?4 
b. As observed by Certkova (1996: 123-4), po- is productive with the 
greatest number of predicate types, including ordinary telic verbs (e.g., 
postroit'p 'build'), inchoatives (e.g., poveselet'p 'become cheerful'), facti- 
tives (e.g., poveselit'p 'make cheerful'), and delimitatives (e.g., po- 
guljat'p 'stroll'); another class might be atelic verbs occurring with parti- 
tive objects (e.g., pokurit'p 'smoke'; cf. Anstatt 2003a). 
c. According to data given by Tixonov (1998: 476), po- enters into more 
44 Recall from section 3.2 that according to Ivanova (1966: 124) Blg po- has no spatial mean- 
ing. Serex (1951: 292) observes that po- has no spatial meanings in Ukrainian. 
36 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
variant prefix alternations than any other Russian prefix, (cf., e.g., 
pomeret'p-umeret'p 'die'). 
d. It remains productive (cf., Certkova 1996: 123-4) in the perfectivization 
of loan verbs, (e.g., poformulirovat'P 'to f~rmulate')?~ 
On the basis of (78) I suggest that s-lz- and po- occupy analogous positions in the 
aspectual systems of Cz and Rus respectively; their positions in the remaining lan- 
guages of the respective groups are more or less identi~al.4~ 
Accordingly, we may classify the Slavic languages into three types with regard 
to the systems of aspectual prefixation that they have developed: (1) the subsump- 
tive type, (2) the s-12- type and (3) the po- type. In the subsumptive type, no 
prefix has clearly lost its spatial meanings and developed into prdverbe vide, acquir- 
ing primacy in the system of perfectivization. The aspectual system of OCS de- 
scribed in section 2.1 is of this type, as are the systems of Cro and Srb. This is not 
to say that no prefixes develop abstract resultativelperfectivizing functions in these 
languages (cf. resultativeldistributive iz- and distributive po- in Cro and Srb), but 
that such functions appear to be fairly clearly derived from their spatial prototypes. 
The s-lz- type consists simply of the western languages (basically Cz, Slk, Sln and 
Sor) in which s-lz- has become the prdverbe vide. Similarly, this should not be mis- 
understood as a claim that no other prefixes function as perfectivizers, but rather 
that the others do so primarily on the basis of subsumption. In the po- type, the 
prefix po- has become the prdverbe vide and shaped the meaning of the pf aspect 
regardless of whether s-lz- has also developed as a prdverbe vide in that language 
(Ukr, Blr, and perhaps Pol) or not (Rus, Blg, Mac). This information is summarized 
in (9): 
(9) Breakdown of Slavic According to Perfectivizing Prefixation: 
a. SUBSUMPTIVE Cro, Srb 
b. S-12- Cz, Slk, Sor, Sln, (Pol), (Ukr), (Blr) 
c. PO- Rus, Ukr, Blr, Blg, Mac, (Pol) 
In this breakdown, Ukr and Blr are listed under the s-lz- type in parentheses, be- 
cause as was pointed out above the presence of po- as a prdverbe vide in a lan- 
guage is maximally distinctive, (i.e., it supersedes s-lz- in the determination of the 
45 Certkova observes that po- follows pro-, za-, s- and ot- in the perfectivization of loan verbs 
(109). However, internet searches show that po- prefixes many of the verbs that she lists as pre- 
fixed with other prefixes. Thus, one can find poformatirovat'P 'format' in addition to otformatiro- 
vat'p, poformulirovat'P 'formulate' in addition to sformulirovat'P, etc. It is not only a matter of de- 
riving delimitatives in po- alongside telic perfectives in other prefixes, as many of the competing 
perfectives in po- attested on the internet are not delimitative, e.g., [elsli mne skufete, kak pofor- 
matirovat'P, ja podgotovlju polnyj spisok ... 'if you tell me how to do the formatting, I will pre- 
pare a full list' [nlo kakie-to vyvody poformulirovat'P poprobuju 'but I will try to formulate some 
conclusions', etc. A more detailed treatment and data cannot be attempted here. Note that if de- 
limitative~ prefixed in po- were included in the loan verb statistics, po- would certainly be higher 
in the list, as the vast majority of loan verbs can derive delimitatives in po-. 
46 AS in the case of delimitatives (cf., Dickey and Hutcheson 2003), Pol demands a special 
treatment due to its clearly mixed status regarding the productivity of s-/z- and po-, but this lies 
beyond the scope of the present discussion. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 37 
meaning of the pf aspect); Pol is listed under both prefixes in parentheses to indi- 
cate its mixed patterning. 
In my view, an adequate analysis of the morphology of Slavic aspect should in 
principle have consequences for a theory of the semantic distinctions expressed by 
aspect (and therefore accord in some non-trivial way with the facts of aspect usage 
in a given Slavic language), and vice-versa. What follows is an attempt to make 
sense of the basic correlation of the productivity of s-/z- with the languages that 
exhibit western patterns of aspect usage according the east-west aspect theory and 
the productivity of po- with the languages exhibiting eastern patterns of usage on 
the other, with reference to grammaticalization theory. From the standpoint of 
grammaticalization, it is very interesting that in a given Slavic language there are at 
most two candidates for a "grammaticalized" priverbe vide (Pol, and perhaps Ukr, 
Blr); in the remaining languages there is either only one (Cz, Slk, Sor, Sln, Rus, 
Blg, Mac), or none (Cro, Srb). It is important to reiterate that by grammaticalized 
priverbe vide I mean a prefix which has either largely lost its primary spatial 
meanings (po-) or at least restructured its network in such a way that its abstract 
meaning appears to be a central, independent node in its network (s-/z-). I do not 
think this fact has ever even been recognized, let alone fully appreciated in Slavic 
aspectology. Given that in an individual Slavic language only one prefix tends to be 
"semantically bleached" while a variety of others also function as perfectivizers, a 
natural question that arises is whether this fact has any particular significance. I 
would argue that it does: the establishment of a single prefix as a prkverbe vide 
"semantically organizes" the perfectivizing function of the remaining prefixes that 
function as perfectivizers on the basis of subsumption. 
In this respect, recall again Shull's (2003) view that source, path and goal pre- 
fixes, in addition to their particular spatial configurations, all share an abstract 
schema, the TRANSITION FROM S1 TO S2. What occurred in Cz and the other 
western' languages was that the establishment of s-/z- as a prdverbe vide grammati- 
calized one prefix as the marker of a category of perfectivity based directly on teli- 
city, i.e., the TRANSITION FROM S1 TO S2. I suggest that the introduction of 
grammaticalized s-/z- into the originally subsumptive system of perfectivizing pre- 
fixation in Cz reinforced the TRANSITION FROM S1 TO S2 as the aspectual mean- 
ing contributed by all other perfectivizing prefixes, which in turn helped to stabilize 
totality as the meaning of the pf in the western languages (recall again the 
east-west aspect theory outlined in section 1). In other words, what took place was a 
kind of semantic restructuring, or analogy, whereby the meaning contributed by 
s-Iz- comes to define the semantic relationship between the pf and the impf: the 
semantic difference between OCz bledZtii and zbledZtiP 'turn pale' begins to serve as 
a model for the semantic difference between verbs such as OCz psatii and napsatip 
'paint'. In other words, with regard to aspectual semantics, the relation napsatip : 
psatii is reanalyzed according to the model zbledbip : bledbii. Given that in a com- 
pletely subsumptive system napsatip already profiles the TRANSITION FROM S1 TO 
S2, one is justified in asking how the grammaticalization of s-Iz- as the systematic 
priverbe vide in fact changed the system, i.e., how it contributed to the grammati- 
calization of aspect in the western languages. In my view, it did not radically alter 
the nature and development of western aspect, but set the western pf firmly on a 
38 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
path to maintain TOTALITY as its categorial mear1ing.4~ 
If we consider the grammaticalization of s-/z- as a marker of perfectivity with 
regard to the criterion of maximal regularity in signaling an opposition (cf. Leh- 
mann 1999: 208)-in this case, the affixation of s-Iz- to all possible verb stems-it 
becomes clear that despite its high productivity, s-/z- has not come close to establish- 
ing itself as the sole marker of perfectivity in the western languages; for instance, 
in Cz, the language with the highest productivity of s-Iz-, za- has nevertheless also 
enjoyed considerable productivity in historical times. So s-/z- does not pass the max- 
imal distribution test. There can be no doubt that the status of s-/z- as the leading 
perfectivizer of loan verbs is evidence of an impulse towards grammaticalization, 
but given the enormous inventory of verbs in which other prefixes signal perfectivity 
it is also highly unlikely that s-/z- will ever attain "maximal distribution" and thus 
be grammaticalized in the traditional sense of the term. 
And yet it does seem that the development of s-/z- must have some significance 
for the establishment and evolution of (western) Slavic aspect as a grammatical cate- 
gory. In this respect, Nichols' and Timberlake's (1991: 129) suspicion that processes of 
grammaticalization "may be less straightforward and obvious than is usually as- 
sumed" seems quite appropriate. Because if the suggestion made above is correct, 
i.e., that what is required to determine the grammatical meaning of the perfective 
aspect in a Slavic aspect system is a single priverbe vide, then it is probably mis- 
guided to assess the grammaticalization of s - / ~ -  as a marker of perfectivity accord- 
ing to the usual criteria of obligatoriness, regularity of expression, etc. And the 
same is true for the grammaticalization of the aspect opposition as a whole: if one 
accepts Shull's view of an abstract schema of the TRANSITION FROM S1 TO S2 ex- 
tracted from and shared by all spatial prefixes, and if it only takes one priverbe 
vide to semantically organize the perfectivizing function of prefixes, then the array 
of perfectivizing prefixes (cf. the remarks by Mende cited above) in a Slavic lan- 
guage seems much less arbitrary or cha0tic.4~ This point cannot be stressed enough. 
Taking this line of thought further, I suggest that what was decisive for the estab- 
lishment of s-Iz- as a grammaticalized priverbe vide was the development that re- 
sulted in it functioning to perfectivize verbs of a maximum number of predicate 
types, which is not quite the same as simply perfectivizing some maximu& number 
of ~erbs .4~  (Again, I view inchoatives as particularly important in determining which 
prefix is a priverbe vide in a given Slavic language.) To all appearances, in the 
western languages it was the purely phonetic accident of jer-fall that precipitated the 
distribution of s-/z- as a perfectivizing prefix over such a broad range of predicate 
types: ordinary telic verbs, inchoatives and factitives. 
I think the spread of s-/z- in Cz and the western languages may be success- 
fully analyzed as a case of what Nichols and Timberlake (1991) term retextualiza- 
47 The precise evolution from TRANSITION FROM S1 TO S2 to totality as the meaning of the 
western pf certainly involves the interrelation between prefixed pf verbs and the innovative de- 
rived impf verbs; this issue is not taken up here. 
48 By the same token, the fact (mentioned in 2.2.3) that s-/z- competes with po- in Pol but is 
nevertheless currently the productive prefix with loan verbs is also less puzzling. 
49 Accordingly, one may consider the high productivity of s-/z- with loan verbs in the west- 
ern languages as the consequence of its grarnmaticalization and not a process of grammaticaliza- 
tion per se. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 39 
tion in their analysis of the expansion of the Rus predicative instrumental. Retext- 
ualization refers to a two-part process of innovation and conventionalization. The 
innovation involves a new usage (token) of a linguistic unit which is analogous to 
but not identical to preexisting exemplars of that linguistic unit; the new token is 
then conventionalized as an exemplar which serves in turn as a model for newer an- 
alogical innovations. I see no reason why this model of grammatical development 
should not be applicable to changes in the distribution and functions of aspectual 
morphology as well, and accordingly suggest that what happened with western Slavic 
s-lz- is that this original accidental innovation was conventionalized, or retextual- 
ized, as a prdverbe vide in those languages. Moreover, I would argue that the anal- 
ogy outlined above whereby the pf meaning of all the other prefixes (e.g., OCz 
psdtii/napsatiP) was slightly reconfigured on the model of s-/z- (e.g., OCz bledt- 
tiilzbledbip) is likewise a case of retextualization. 
The retextualization hypothesis makes sense with regard to Slosar's (1981: 
130-4) account of the growth in the productivity of S-IZ-.~' He surmises that in ear- 
ly OCz s - /~ -  spread more or less mechanically through semantic classes of OCz 
verbs. Thus, s-Iz- originally functioned as an empty perfectivizer with verbs of de- 
struction/consumption and loss, e.g., zbbstip 'stab to death', zemiietip 'die', zhubitip 
'devastate, kill', zjdstip 'eat up', etc. From this group, s-Iz- spread to an antonymous 
group of verbs denoting "emergence", e.g., aevitip (st?) 'appear', zmnoiitip 'multiply', 
zmyslitip 'think up', etc. Subsequently it spread to a group of verbs expressing a neg- 
ative result, e.g., zdraiditip 'irritate', zkalitip 'cloud, muddy', ztlatitip 'crush under- 
foot'. From here it then spread to the deadjectival inchoatives (e.g., ztvirdttip 'hard- 
en') and factitives (e.g . , zmZktitiP 'soften7). Accepting Slosar's chronology of the 
spread as an approximation of what really happened, I suggest the retextualization 
proceeded as follows: jer-fall and the phonetic coalescence of *sa- and *jbz- pro- 
duced a class of resultative verbs denoting the destruction/consumption of the object, 
most of which were originally prefixed with *jbz- (e.g., zbbstip 'stab', aestip 'eat 
up'). The accelerated disassociation of z- (t *jbz-) with its etymological meaning of 
elativity led to its productivity as a resultative prefix with such verbs, where the re- 
sult is very tangible. Verbs of destruction/consumption originally prefixed with *s%-, 
e.g., sidcip 'burn up', reinforced this productivity, as their resultativity (telicity) was 
reanalyzed as a case the new resultative prefix z- and their connection to the etymo- 
logical basis of their telicity (either the centripetal or downward-ablative meaning of 
*s%-) was lost. This in fact was the coalescence creating s-/z-, and the resultative 
verbs of destruction/consumption were the first "exemplars" of resultative s-lz-. Sub- 
sequently, verbs of emergence, many of which according to Slosar (1981: 132) had 
been originally prefixed with *sa-, e.g. zmyslitip 'think up' (cf. OCS samyslitip) were 
retextualized as resultatives analogous to the verbs of destruction/consumption. Verbs 
of negative result were probably reanalyzed as resultatives in s-/z- simultaneously to 
the aforementioned process (I do not think the time of the assimilation of this class 
of verbs into the new resultatives in s-/z- is crucial for the analysis, and since ac- 
cording to Slosar they functioned as synonyms of the verbs of destruction, I consider 
it possible that they were retextualized relatively early). In other words, I suggest 
Slosar refers only to z-, but this does not affect the hypothesis given here; his data for s- 
(cf., Slosar 1981: 124) are incorporated easily enough, as they involve the same classes of verbs. 
40 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
that the retextualization of resultative verbs prefixed in s- (t *sa-; their resultativi- 
ty was originally metaphorically derived from the centripetal or downward-ablative 
meaning of the prefix) as resultatives prefixed with the new abstract prefix z- was 
crucial for the semantic coalescence. This ongoing process can account for Slosar's 
(1981: 105) observation, mentioned in section 2.1.1, that z- as a resultative prefix was 
reinforced by verbs that originally had s-. 
As pointed out several times in the preceding sections, the next and most sig- 
nificant step was the establishment of s-/z- as the inchoative prefix. This occurred 
by virtue of the fact that s-/z- was becoming firmly established as an abstract resul- 
tative prefix, which was ideal for the perfectivization of inchoatives, given the fact 
that they had no natural association with any kind of metaphorical spatial telicity 
(cf. section 2.1) as well as the fact that various exemplars already in s-/z- had al- 
ready been created by jer-fall, e.g., zsechnutip 'dry out' (cf., OCS isaxnqtip). This 
then led to diverse verbs being prefixed with s-12- to form resultatives and in time, 
simple perfectives. To sum up, jer-fall resulted in a nascent group of resultative 
verbs in s-/z- which served as exemplars for the retextualization of other verbs con- 
taining the prefix; as verbs of other classes prefixed with the new s - / ~ -  began to 
serve as exemplars of resultative s-/z-, the growing group then generally became a 
productive derivational model for perfectivizing prefixation. 
The second component of the retextualization process involves s-/z-, on the one 
hand, and verbs containing other prefixes on the other. According to the analogy 
hypothesized above, the aspectual values of the pair psdtii/napsatiP 'paint' were re- 
analyzed according to the values of the pair bledttii/zbledttiP 'turn pale'. Such a re- 
analysis may be easily analyzed as a case of retextualization, in which verbs pre- 
fixed in s-/z- serve as exemplars for a retextualization of pf verbs prefixed with 
other prefixes. As pointed out above, the shift in meaning is slight, and is probably 
better viewed as a process leading to the reinforcement of resultativity as the proto- 
typical meaning of the western pf as a opposed to anything like a seismic shift in 
aspectual semantics. The emergence of resultativity as the meaning of the perfective 
would easily lead to a subsequent development of totality as the meaning of the wes- 
tern pf.51 The fact that the grammaticalization of s-/z- as the dominant prdverbe 
vide resulted in a reinforcement of an already existing meaning (telicity/resultativity) 
as opposed to a semantic shift allows us to make sense of the fact that the subsump- 
tive languages, Cro and Srb, pattern very closely to the languages of the western 
group, i.e., the s-/z- type, Cz, Slk, Sor, Sln. 
The hypothesis outlined above may seem trivial, given the "minimal" semantic 
impact of innovative s-/z- on the system of western aspect. However, I think it is 
more realistic than hypothesizing a more significant and "teleological" grammatical- 
ization process within the system of perfectivizing prefixation in the western lan- 
guages. The western languages merely took steps to grammaticalize the meaning of 
51 At this point it must be stressed that the hypothesis advocated here should not to be under- 
stood as a claim that it was s-/z- alone that determined the semantic development of the western 
pf. There were other important factors involved, including the role of -nq- as a productive perfect- 
ivizer (cf., Dickey 2001b) and-perhaps most importantly-the role of German (and very possibly 
also Romance) language interference in the western languages. These issues cannot be discussed 
here. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 41 
the perfective that was already shared by telic prefixes, as opposed to creating a 
new semantic opposition by means of that grammaticalization. In my view, the sem- 
antic organization of the pf aspect by means of s-/z- may help to explain why pre- 
fixes in the western languages have retained a relatively strong spatial character (for 
instance, all the western languages have retained the surface-contact meaning of 
po-). This has been pointed out by Shull (2003: 228-30) concerning prefixation in 
Cz, which she concludes is more noticeably spatial than prefixation in Rus, which is 
on the average more abstract. If western languages such as Cz grammaticalized s-Iz- 
as the marker of the TRANSITION FROM S1 TO S2, which established telicity as the 
abstract aspectual profile of their perfectivizing prefixes, the fact that the aspectual 
profile of these prefixes was an element of meaning compatible with their original 
spatial meanings would allow them to retain these spatial meanings at no cost to 
any additional grammatical function(s) they might have or develop. 
This view of the development of the western pf also helps to explain why de- 
limitative~ in po- never became highly productive in the western languages (cf. 
Dickey and Hutcheson 2003), for the following reasons. If the grammaticalization of 
S-12- reinforced the TRANSITION FROM S1 TO S2 and thus resultativity as an early 
prototypical meaning of the western pf, then it makes sense that delimitatives in 
po- comparable to those in Rus never became highly productive in the western lan- 
guages, as their profile, the absencelnegation of a situation (-S) followed by the sit- 
uation in question (S) followed by the absence of the situation (-s), i.e., -S + S + 
-S, differs considerably from the TRANSITION FROM S1 TO S2. The difference be- 
tween the profile -S + S + -S and the TRANSITION FROM S1 TO S2 is too great 
to allow such delimitatives to be derived on a large scale as straightforward instanti- 
ations of the TRANSITION FROM S1 TO S2 hypothesized as the shared schema of 
perfectivizing prefixe~.~' Further, the assumption that grammaticalized s-/z- rein- 
forced resultativity as the primary meaning of the western pf accords with the "te- 
lic" nature of delimitatives in po- in the western languages. For instance, unlike 
eastern delimitatives, delimitatives in the western languages may easily stress the 
full extension of a predicate over a particular period of time, as shown in (10): 
(10) Dostal horkou nemoc a poleielp si nEkolik tfdnG, nei se uzdravil. 
'He came down with typhoid fever and  la^ for several weeks before he 
recovered.' [czl 
Thus, western delimitatives have retained a prominent telic sense, which is arguably 
supported by the telic nature of the western pf as influenced by s-Iz-. 
Despite the admittedly minimal impact of s-/z- on the meaning of the pf in 
the western languages, if the hypothesis of the grammaticalization of s-/z- advo- 
cated here is correct, we should in principle be able to find other differences be- 
tween the aspectual systems of the languages of the s-Iz- type on the one hand and 
those of the other languages (belonging to the subsumptive and po- types) on the 
other. One subtle yet important difference involves the productive derivation of pf 
52 This semantic incompatibility does not render the derivation of such delimitatives im- 
possible. Rather it simply means that they will be marginal subtypes of the perfective in the 
western languages: delimitatives do not typically denote a change to a new, positively character- 
ized state, but only the (trivial) change from a situation to its absence. 
42 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
verbs of abstract change of state. Cz and Slk, the epicenter of the spread of s-Iz-, 
have developed deadjectival inchoatives and factitives as classes of verbs much more 
than the other Slavic lang~ages.5~ For example, Cz derives zeviednttp 'become dai- 
lylquotidian' directly from the adjective viedni 'daily' (i.e., the clearly adjectival stem 
viedn-; cf. Slk zoviednietp with the same meaning), whereas this notion must be ex- 
pressed periphrastically in other Slavic languages, e.g., Rus stat'p obydennym, 
CroISrb postatip svakidainji; cf. also Cz zvodnatttp 'become watery' ( t  vodnaty' 
'watery'; Slk zvodatnietp) as opposed to Rus stat'p vodjanistym and CroISrb postatip 
vodenast. This model is productive with loaned adjectives, cf., Cz znervbznttp 'be- 
come nervous' (+ nervbzni 'nervous'; Slk znervbznietp), as opposed to Rus stat'p 
nervnym, CroISrb postatip nervozan. The derivation of deadjectival factitives with 
s-Iz- is also quite productive in Cz and Slk, in contrast to the other Slavic languages: 
cf., for example, Cz zpiistupnitp 'make accessible' ( t  piistupny' 'accessible'; Slk 
spristupnitp), as opposed to Rus sdelat'p dostupny'm, CroISrb utinitip dostupnim, as 
well as Cz znervbznitp 'make nervous' (Slk znervbznitp), as opposed to Rus privestip 
v nervnoe sostojanielrasstroit'P, CroISrb iznerviratiPlutiniti nerv0znim.5~ Though for 
individual predicates other Slavic languages will have inchoative or factitive verbs 
equivalent to the Cz and Slk verbs in s-Iz- (e.g., Ukr zvodnityp 'become watery', 
CrolSrb iznerviratip 'make nervous'), the examples given above accurately reflect the 
status of inchoatives and factitives as a systematized element of the CzISlk aspectual 
system as opposed to the other Slavic languages. 
Further evidence of the integration of inchoatives and factitives in s-Iz- into the 
CzISlk aspectual system are verbs prefixed with ~ n e - . ~ ~  Examples of inchoatives are 
znemravnttp 'become immoral' ( t  nemravny' 'immoral'; Slk znemravnietp) and zne- 
hybnttp 'become motionless' ( t  nehybny' 'motionless'; Slk znehybnietp); representat- 
ives of factitives are znemravnitp 'make immoral' (Slk znemravnitp) and znehybnitp 
'make motionless' (Slk znehybnitp). Such verbs have few direct equivalents in East and 
South S l av i~?~  where periphrastic translations are required, cf., e.g., Rus stat'plsde- 
lat'p beznravstvennyrn 'becornelmake amoral', stat'Plsdelat'P nepodviinyrn 'be- 
comelmake motionless' and CroISrb postatip nernoralanlutinitiP nernoralnirn 'be- 
53 Pol and Ukr have also developed some inchoative verbs of this type, but not to the extent 
of Cz and Slk, as pointed out in 2.2.3 and 3.1. 
54 Sor also appears to derive deadjectival inchoatives and factitives in a manner resembling 
Cz and Slk, cf., e.g., zbrunjeCP 'become brunette', zbruniCP 'dye brunette', spFistupniCP 'make ac- 
cessible', etc., though it is impossible at this point to assess the productivity of s-/z- in this role 
more precisely relative to Cz and Slk. 
55 Sor derives some such verbs, e.g., znjemjernikp 'upset' znjesmjertnikp 'make immortal'. Pol 
derives some verbs in znie-, e.g., znieksztakiCP 'deform', etc., but only a few are actually deadjec- 
tival, e.g., znieczuliCP 'make numb' (t czuly 'sensitive'); in Pol this derivational model seems more 
marginal than in Cz and Slk. 
56 Ukr derives some verbs in zne-, e.g., znestjamytysjaP 'lose consciousness', zneslavytyP 'de- 
fame' (both denominal), but very few clearly deadjectival verbs, of which an example is zneru- 
xomytyp 'make motionless' (t neruxomyj 'motionless'). One item which indicates a lack of sem- 
antic consistency as far as the element of negation is concerned is zljudnityp, which means 'be- 
come depopulated' and not 'become populated' (+ ljudnyj 'populated'); I think the only plausible 
way of arriving at such a meaning is to assume that here z- expressed ablativity or elativity (in a 
fashion similar to English depopulate), which is an indication that s-/z- in Ukr is not quite as free 
of spatial meaning as it is in Cz. 
S. M. Dickev. S-/Z- and Grammaticalization of Asvect in Slavic 43 
comelmake immoral', postatip nepokretanlutinitip nepokretnim 'becornelmake mo- 
tionless'. Hauser (1995: 201) observes that in Cz inchoatives and especially factitives 
in zne- are increasing in productivity, and observes that contemporary usage attests 
verbs in zne- not given in dictionaries, e.g., zneprlchodnitp 'make impassable', zne- 
splavnitP 'make unnavigable', znerovnopravnitp (Slk znerovnopravnitP) 'make une- 
qual', zneviditelnitp (Slk zneviditelnitp) 'make invisible'. Such productivity contrasts 
with the almost complete lack of such verbs in East and South Slavic. 
The fact that these models of derivation are productive in precisely in Cz and 
Slk lends support to the hypothesis advocated here that the grammaticalization of 
S-/z- as a prkverbe vide privileged the TRANSITION FROM S1 TO S2 as an inde- 
pendent, abstract aspectual meaning in the western languages. Moreover, I suggest 
that the salience of this abstract meaning in the aspectual systems of Cz and Slk is 
responsible for the creation of verbs in zne- referring to very specific domains, as in 
the case of zneschopnitp 'declare unfit for worklgive someone a sick note' (+ ne- 
schopny' 'unfit'). To recapitulate, Cz and Slk are producing a diversity of predicate 
meanings involving inchoatives and factitives with s-/z- and zne-: they derive incho- 
atives and factitives involving one and the same notion, e.g., Cz zpiisnttp 'become 
strict' and zpiisnitp 'make strict', as well as positive and negative correlates, e.g., Cz 
zrovnoprdvnitp 'make equal (before the law)' and znerovnoprdvnitp 'make unequal 
(before the law)'. In my view, this productivity of change of state verbs in s-Iz- and 
especially zne- in Cz and Slk indicates that these languages are extending the scope 
of S-12- in its meaning of the TRANSITION FROM S1 TO S2, which would be un- 
likely if this meaning were not relatively salient in the network of the perfective as- 
pect. On the other hand, the fact that other Slavic languages derive such verbs to a 
lesser extent or not at all makes sense if we assume that their aspectual systems 
lack a grammaticalized prefix signaling the TRANSITION FROM S1 TO S2 (e.g., 
CroISrb), or are oriented around another conceptual category (e.g., Rus; see below). 
Another consequence of the grammaticalization of s-Iz- can be seen in some 
subtle differences in the relative importance of prefixation for the aspectual systems 
of the languages of the s-/z- type and those of the subsumptive type, i.e., Cro and 
Srb,57 regarding the status of biaspectual verbs in the respective aspectual systems. 
Grickat (1957: 66, 104-5) observes not only that newer loans are quite resistant to 
prefixation in CroISrb (e.g., niklovatii/P 'plate with nickel', denunciratii/P 'denounce') 
but also that CroISrb has a higher number of older biaspectual verbs (e.g., krstitii/P 
'baptize', testitatii/P 'congratulate') than any other Slavic lang~age.~'  In my view, 
Grickat (116) rightly attributes the high level of biaspectuality in CroISrb to the lack 
of sufficiently abstract perfectivizing prefixes in the language(s), and recognizes this 
57 Likewise, there ought to be the same kind of differences between Cro and Srb on the one 
hand and the po- languages such as Rus on the other, as prefixation is just as important for as- 
pect ( i f  not more so) in the latter as in the s - / ~ -  type, and indeed there are (cf.  Grickat 1957: 
119-28); but the focus here is on the consequences of s-/z- as a grammaticalized pre'verbe vide in 
the western languages as opposed to languages lacking that pre'verbe vide as well as any other 
(i.e., PO-). 
58 IvanEev (1971: 170, fn. 1) disputes Grickat's conclusion that biaspectual verbs are more 
numerous CroISrb than in Blg, ascribing to Blg "first place" in this regard. The issue cannot be 
resolved here, but I think there are several reasons for accepting Grickat's conclusions, or at least 
for the assumption that biaspectuality is more prominent in the aspectual system of CroISrb than 
44 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
as an archaic feature of CroISrb (128). In contrast, languages of the s-lz- type dis- 
play a relatively low level of biaspectuality, both in older verbs and newer loan 
verbs, the latter of which are consistently, if slowly, affixed with aspectual morpho- 
logy, primarily, in fact s-lz-, cf., e.g., dormatovatp 'format' and zregistrovatp 'regis- 
ter'. 
If we assume that s-Iz- contributed significantly to the determination of the 
meaning of the pf in the western languages, we can do the same for the develop- 
ment of po- as a prdverbe vide in the eastern languages. It was shown above that 
Rus po- shares some important qualities with Cz s-Iz-, which make it unique within 
the Rus system of prefixation. Another important indication of the unique status of 
po- among Rus perfectivizing prefixes is the fact that, as Camus (1998: 101) points 
out, po- is the only Rus prefix for which the aspectual pairs consisting of an impf 
simplex and a prefixed pf verb outnumber the pairs consisting of a prefixed pf 
verb and its suffixed impf correlate. The importance of this fact cannot be stressed 
enough in an evaluation of the status of po- in the Rus aspectual system. In view of 
these facts, I consider it possible that the grammaticalization of po- as the dominant 
prdverbe vide in the eastern languages has had systemic effects analogous to those 
of S-12- in the western languages, the difference being that its impact on the devel- 
opment of semantic category of the eastern pf has been greater. And like the spread 
of S-12-, the spread of eastern perfectivizing po- may be analyzed as a case of re- 
textualization. The following remarks outline a hypothesis of the grammaticalization 
of po- in the eastern languages; as the development of perfectivizing po- is to all 
appearances a considerably more complex phenomenon, a detailed treatment is im- 
possible. 
Dickey (2000: 282-7; see also the references cited there), suggests that the ori- 
gin of the east-west aspect division lies chiefly in changes that have occurred in 
eastern aspect usage approximately since the seventeenth century. The current pat- 
tern of eastern aspect usage described in section 1 is to a considerable extent the 
result of changes that involved on the one hand the expansion of the impf in the 
non-actual present and the general-factual function, and on the other the increasing 
restriction of the pf aspect to contexts of sequentiality, as well as the loss of aspect- 
ual pairs of verbal nouns. (It is unlikely that these changes in usage took place at 
the same time; rather, the elimination of the pf aspect from contexts of habituality 
was probably a slow process that began relatively early, whereas other changes, such 
i n  the Blg system. First, some very common CroISrb biaspectual verbs have aspectually distinct 
cognates i n  Blg, cf.,  e.g. CroISrb Eestitatii/P vs. Blg testitjap 'congratulate', CroISrb jebatii'P vs. Blg 
ebai 'fuck', and CroISrb vidjetii/P vs. Blg vidjap 'see'. Second, though both CroISrb and Blg have 
"impf" simplex verbs that are used like pf verbs i n  sequences o f  events, e.g., CroISrb jestii, Blg 
jami 'eat', there are paired simplex verbs i n  CroISrb for which this is true i n  contrast t o  their 
equivalents i n  Blg; an example is CroISrb pitatii-(paired with upitatip), which occurs i n  past-tense 
sequences o f  events quite easily, whereas Blg clearly prefers the popitamp 'ask'-(paired with 
pitami) i n  such narrative sequences. Third, Blg does prefix loan verbs t o  a limited extent, e.g., 
otreagiramp 'react', as does CroISrb, e.g., izorganiziratiP 'organize', but also shows a tendency t o  
s u f f i x  loan verbs i n  order to create derived i m p f  verbs, e.g., ekranizirvami 'produce for the screen', 
registrirvami "register' and servirvami 'serve'. Finally, it is worth pointing out that since Blg re- 
tains the aoristlimperfect distinction for biaspectual verbs, such verbs are arguably less conse- 
quential for a n  assessment o f  the Blg aspectual system than they are for CroISrb, where most con- 
temporary urban speakers have reduced the past tense system t o  a single preterite. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 45 
as the expansion of the impf general-factual and the loss of pf verbal nouns, oc- 
curred later, as late as the eighteenth century.) Dickey and Hutcheson (2003) suggest 
that the productive derivation of delimitatives in po- in the eastern languages since 
the seventeenth century stands in some connection with the innovations in eastern 
aspect usage, as well as the grammaticalization of aspect in those languages. It is 
interesting that Bermel (1997) concludes that Rus aspect was grammaticalized much 
later than is commonly assumed, around the sixteenth century. Bermel's conclusions 
accord with those reached independently by Nergird-Serensen (1997), who argues 
that aspect crystallized as a grammatical category in Russian in the seventeenth 
century. The temporal coincidence of the rise of delimitatives in po- with the re- 
cently suggested chronology of the grammaticalization of Rus aspect cannot be ac- 
cidental. 
Let us first discuss the place of delimitatives in the aspectual system of the 
eastern languages (much of this discussion is drawn from Dickey and Hutcheson 
2003, to which the reader is referred for details). Though Russian/Slavic aspectology 
has traditionally viewed delimitatives in po- as just one of several Aktionsarten, it is 
clear that they have a special status as pf verbs which qualify as "aspectual part- 
ners" of their impf source verbs (when these are construed as atelic predicates), and 
this fact has been recognized in some recent work on the subject (e.g., Lehmann 
1988 and Certkova 1996). The development of the productive derivation of delimit- 
ative~ in po- in East Slavic and Blg began in the sixteenth century and resulted in 
the spread of the aspect opposition to a whole predicate class, atelic activities (e.g., 
sidet" 'sit'). Before the advent of delimitatives, the aspect opposition was largely lim- 
ited to telic predicates (accomplishments, e.g., stroit" 'build', and achievements, e.g., 
skakat" 'jump'), i.e., predicates that are readily viewed as completed events. The 
spread of the aspect opposition to atelic activity predicates thus represents an im- 
portant step in the grammaticalization of the aspect opposition, as it considerably 
extended its distribution throughout the verbal inventory. 
Sigalov (1975) documents the spread of delimitative po- in Rus. Delimitatives 
were originally derived from stative verbs, as early as Common Slavic; ORus exam- 
ples are poleiatip 'lie for a while' and pobolttip 'be ill for a while'. Sigalov (171) hy- 
pothesizes that the delimitative meaning developed next in po- derivatives of inde- 
terminate verbs of motion, e.g., ORus pobtgatip 'run for a while', poxoditip 'walk 
for a while', and "verbs of psychological activity", e.g., ORus pomolitisjap 'pray for 
while' poveselitisjap 'be cheerful for a while', approximately from the sixteenth to 
the seventeenth centuries. Only subsequently (in the seventeenth and eighteenth cen- 
turies) did the delimitative meaning develop in po- derivatives of verbs of speech 
(e.g., pobestdovatip 'converse for a while', pogovoritip 'talk for a while'), verbs of 
sound (e.g., pogremttip 'thunder for a while', pogudttip 'drone for a while') and 
verbs of physical activity (e.g., pokopatip 'dig for a while', pokositip 'mow for a 
while'). Note that all these verbs are attested in ORus in other meanings (usually re- 
sultative), so what we are dealing with is the reanalysis of such derivatives as delimi- 
tative~, and not usually their new derivation. I suggest that part of what happened 
was that these various classes of verbs were successively retextualized as delimitat- 
i ve~  on the basis of already existing delimitative exemplars. 
Why delimitativity would become so productive in the eastern languages is a 
question which cannot be anwered here in any definitive way. However, Sigalov 
46 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
(1975) and Dmitrieva (2000) assume that the origin of modern delimitative po- lies 
in the loss of the resultative meaning of the prefix with individual verbs, which was 
derived more or less directly from older spatial meanings of the prefixlpreposition, 
its surface-contact meaning and/or its allative meaning.59 Some spatial meanings of 
the preposition po that existed in ORus, such as its allative meaning (e.g., po Roga- 
ticju 'all the way to Rogatica'), were lost well before the eighteenth century, so it is 
possible that the loss of the allative meaning of the prefixlpreposition deprived the 
old resultative po-derivatives of the spatial metaphorical basis for their telicity, at 
which point they began to be reanalyzed as atelic delimitatives in po-. 
I believe that an important exemplar for the aforementioned processes of re- 
textualization were in fact determinate verbs of motion in po-, e.g., ORus poitip 'go', 
which are not commonly associated with delimitatives. Yet data from ORus show 
that poitip could at one time express the duration of the motion in fashion re- 
sembling ordinary delimitatives, as shown in (lo), one of many such sentences oc- 
curring in Afanasij Nikitin's Journey across Three Seas (fifteenth century): 
(11) A is cjuvilja poBliP esmja do Pali 8 dni, do indtijskyja gory. 
'And from Chaul we eight days to Pali, to the Indian mountain[s].' 
[BLDR 7: 3541 
Note that the emphasis of the duration of the situation here resembles that in the 
modern Cz ex. in (9). Thus, it is not at all misguided-especially in a diachronic 
analysis-to seek a semantic element shared by determinate motion verbs and other 
types of verbs prefixed with po-. Shull (2003: 147-80) argues that the profile of po- 
with determinate motion verbs in modern Rus is not merely ablativity, but the be- 
ginning of the motion as well as some indeterminate amount of it (cf. Shull 2003: 
153). Shull's profile for po- with motion verbs is given in Figure 5. 
Figure 5: Shull's (2003: 153) Schema for Rus po- wi th  Motion Verbs 
I LM (background TRY) 
According to this schema, a trajector (TR) traverses some amount of a trajectory 
(TRY), for which the landmark (LM) is not an ordinary landmark in space but in 
fact the full trajectory of the motion expressed by the verb, i.e., the normal extent 
of the action itself. According to Shull (173), this schema unites the meaning of po- 
in determinate verbs of motion (e.g., pojtip 'go'), inchoatives (e.g., pokrasnet'p 
'blush'), delimitatives (e.g., potitat'p 'read for a while') as well as telics (e.g., postro- 
it'p 'build)), in the sense that the contribution of po- is the meaning that "some a- 
mount of the action has been completed", and the individual interpretations of the 
verbs are functions of the predicate types themselves (for details see Shull 2003: 163). 
Shull's synchronic view that the aspectual meaning of determinate motion verbs in 
po- is essentially the same as that of delimitative verbs is illuminating, and is an- 
other reason for linking determinate motion verbs in po- and delimitatives in a dia- 
chronic analysis. 
59 Note that this is also a claim made by NEmec (1954). 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 47 
An oddity of the prefix po- with determinate motion verbs, e.g., ORus poitip 
'go', is that the original spatial meaning of the prefix in this case (something akin 
to the contact meaning of the modern preposition po 'along') was redundant inas- 
much as motion always occurs along a path. As pointed out in (8a), Shull (158) ob- 
serves that the prefix currently has no spatial meaning in motion verbs. I suggest 
that the spatial meaning of the prefix was lost relatively early in determinate motion 
verbs, so that ORus poitip 'go' took on a meaning of temporal ingressivity (as dis- 
tinct from spatial ablativity), yielding the current meaning of 'beginning of the mo- 
tion plus some indefinite amount'. I consider it possible and in fact very likely that 
determinate motion verbs (which are very common in discourse) were very salient 
exemplars in the process of retextualization of older po- resultatives as delimitat- 
i v e ~ . ~ ~  
Let us now consider the consequences of the grammaticalization of po- as the 
eastern prkverbe vide. The hypothesis advocated here is that, parallel to the case of 
s-Iz- in the western languages, the loss by po- of its spatial meanings (cf. 8a) as 
well as its role in perfectivizing diverse predicate types (cf. 8b) resulted in it per- 
forming a similar role, by contributing to a redefinition of the prototypical meaning 
of the pf aspect, i.e., of the semantic distinction between the pf and impf in verbs 
with other prefixes. Delimitative verbs are important in this respect, because their 
profile of a limited, indefinite duration of a predicate in time, i.e., -S + S + -S, 
reveals the core meaning of po-: abstract limitation in time irrespective of telicity 
(completion), recall in this regard Shull's (163) view that po- perfectives are resulta- 
tive or delimitative based primarily on the telicity or atelicity of the source verb. 
Thus, when po- combines with a predicate that is conceptualized as telic (e.g., 
~tro i t '~ )  it creates a resultative, whereas when it combines with a predicate concep- 
tualized as atelic (e.g., ~to ja t '~  'stand') it creates a delimitative. As suggested above, 
this effect is much different from the TRANSITION FROM S1 TO S2 expressed by 
s-Iz- (which is why the latter does not derive delimitatives in the western languages). 
Why exactly is this the case? Certainly the delimitative schema -S + S + -S is not 
incompatible with resultativity; rather, it is the inclusion of two temporally contigu- 
ous situations in the profile base that are not the profiled situation (S) that consti- 
tutes the important difference. I contend that these temporally contiguous situations 
are in fact the prior and subsequent states of affairs in the definition of temporal 
definiteness of the eastern pf given in section 1; thus, the schema -S + S + -S is 
in fact equivalent to X + S + Y (recall section 1, where temporal definiteness was 
schematized as X + S + Y, where S is the situation profiled by the verb and X and 
Y represent the temporally contiguous, qualitatively different states of affairs). Thus, 
it was the profile of po- of a situation sequential in time to two contiguous, quali- 
tatively different states of affairs X and Y that played an important role in the re- 
structuring of the eastern pf aspect. If we assume that most other prefixes (Rus data 
will be taken as representative) profiled a (spatially-based) telicity, i.e., the 
TRANSITION FROM S1 TO S2, then we may say that the following analogy took 
60 Note that the eastern languages and Pol, all of which have productive delimitatives, also 
have pf determinate verbs of motion prefixed in po-. Blg has lost poitip 'go', but Middle Blg did 
have it (cf. Lilov 1964: 110-11); note that Blg still has other ingressive verbs of motion, e.g., po- 
bjagnap 'run'. So there is a definite correlation between the development of delimitatives and the 
existence of determinate verbs of motion prefixed in po- in a given Slavic language. 
48 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
place: the relation napisat'p : pisat'" i.e., a resultativity distinction based on the 
TRANSITION FROM S1 TO SZ is reanalyzed on the basis of the relation posidet'p : 
sidet", a distinction based on X + S + Y, with the result that the relation between 
the telic pair napisat'p : pisat" is now a distinction based on X + S + Y. In other 
words, napisat'p went from being a pf verb profiling a writing event that produces 
some result to being a pf verb which profiles a writing event that produces a result 
and which is located between a preceding situation X and a subsequent situation Y 
(cf. the definition of temporal definiteness given in section This analogy is in 
fact a second retextualization, parallel to the restructuring of the aspectual seman- 
tics of the western prefixes to s-12-. However, in the eastern languages the restruc- 
turing is more significant. And I propose that it was a restructuring of this kind of 
the prototypical meaning of the eastern pf that produced the changes in aspect 
usage mentioned above that are described by Dickey (2000: 282-7), which basically 
involve an increasing restriction of the pf to contexts of sequentiality . 
Of course, this hypothesis is highly speculative (as will be any hypothesis of 
historical semantic development). However, it has the advantage of accounting for 
the nature of perfectivizing prefixation in the eastern languages which, like the 
western languages, have developed only one priverbe vide. Moreover, this hypothesis 
allows us to account for the differences between the eastern and western groups 
with the same descriptive and theoretical concepts. Further, the hypothesized de- 
velopment and effect of perfectivizing po- in the eastern languages fits in chrono- 
logically to produce the changes in eastern aspect that are known to have occurred 
in the last four hundred years. Finally, it might be added that a semantic restruc- 
turing of all the eastern perfectivizing prefixes on the model of the temporal profile 
hypothesized for po- can account for the fact that, as Shull (228-30) concludes, Rus 
perfectivizing prefixation is on the average relatively abstract compared to Cz pre- 
fixation. 
6. Conclusion 
This article has presented a description of the distribution and origin of the innovat- 
ive hybrid prefix s-Iz- in the Slavic languages. S-12- developed primarily in a group 
of western languages-Cz, Slk, Sor, Sln, Pol, Ukr and Blr (as well as Kajkavian dia- 
lects of Cro). It has been argued that s-Iz- played an important role in shaping the 
meaning of the perfective aspect in a group of western languages (Cz, Slk, Sor, 
Sln), which have already been demonstrated to comprise a western aspectual type by 
Dickey (2000). The remaining Slavic languages did not develop s-12-. Cro and Srb 
have been argued to be languages that have not developed a priverbe vide, so that 
the aspectual systems of these two languages are still based on the mechanism of 
subsumption; nevertheless, the semantic difference between aspect in CroISrb and the 
western s-12- languages is not great, and CroISrb pattern closely to the languages of 
the western group as far as aspectual usage is concerned (cf., Dickey 2000). Rus, 
61 AS in the case of s-/z- in the west, this is not meant as a claim that the rise of delimitat- 
i v e ~  was the sole cause of the change of the meaning of the eastern pf from totality to temporal 
definiteness. It is very probable that the eastern telic pf verbs were developing increased associa- 
tions with temporal localization anyway, and that the process described here was a kind of final 
stage or reinforcement of that development. The lack of German language interference in the eas- 
tern languages is surely important in this respect, but as pointed out in fn. 51, this issue cannot 
be dealt with here. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 49 
Mac and Blg did not develop s-/z- either, but did develop another prkverbe vide, per- 
fectivizing po-. Po- also became highly productive in Ukr, Blr and Pol. East Slavic, 
Mac and Blg all share an eastern aspectual system (cf. Dickey 2000), and it has 
been argued here that the eastern system was semantically shaped to a great extent 
by perfectivizing po-, regardless of whether s-Iz- was also developed in an individual 
language (Ukr, Blr) or not (Rus, Mac, Blg). Pol was also significantly affected by 
the development of po-, but not to the degree of the eastern languages, so that as- 
pectually it is a transitional zone that patterns fairly close to the eastern languages. 
The approach taken here, that in the respective language groups either s-Iz- or 
po- has come to be the dominant prkverbe vide, is innovative in that unlike previ- 
ous approaches it assumes that prefixation has played a more active role in the 
semantic formation of the pf aspect in Slavic aspect systems: s-/z- has played this 
role in the west, whereas po- has played it in the east. It has been suggested that 
the developments of s-Iz- and po- in the respective language groups have been simi- 
lar in many ways: they have both emerged as abstract perfectivizing prefixes, and 
have both become productive in the perfectivization of verbs of diverse predicate 
types, in addition to other minor similarities. The grammaticalization of s-/z- as a 
prkverbe vide in the western languages contributed to the development of totality as 
the prototypical meaning of the pf aspect, whereas the grammaticalization of po- as 
a prkverbe vide in the eastern languages contributed to the development of the pro- 
totypical meaning of the pf aspect from totality to temporal definiteness. 
It has further been argued that these prefixes have participated in atypical 
grammaticalization processes involving their spread through different classes of 
verbs by means of a process called retextualization (cf. Nichols and Timberlake 
1991), as well as the restructuring of the pf semantics of the other perfectivizing 
prefixes in the respective language groups. In this way, s-/z- and po- have each ex- 
erted a great influence on the meaning of the pf aspect in the respective language 
groups without having been completely generalized as markers of perfectivity. 
Though the hypotheses of their developments (especially in the case of the spread of 
perfectivizing/delimitative po-) are admittedly speculative, they accord well with the 
facts as we know them concerning the east-west aspectual division, and have the 
advantage of correlating differences in aspect morphology with differences in usage 
between the western and eastern groups. 
References 
Agrell, Sigurd. (1908) Aspektanderung und Aktionsartbildung beim polnischen Zeit- 
worte. Lund: Hikan Ohlssons Buchdruckerei. 
Anan'eva, N. E. (2003) "Rol' dannyx glagol'nogo slovoobrazovanija v sovremennyx 
zapadnoslavjanskix dialektax dlja izuCenija istoriCeskix processov slavjanskoj 
glagol'noj derivacii". A. M. Moldovan, ed. Slavjanskoe jazykoznanije. XIII 
Meidunarodnyj s"ezd slavistov. Ljubljana, 2003 g. Doklady rossijskoj delegacii. 
Moscow: Indrik, 7-18. 
50 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
Anstatt, Tanja. (2003a) "Die Quantelung des zweiten Arguments im Russischen: Der 
Typus s'kst' jabloko - poest' supu". Thomas Daiber, ed. Linguistische Beitrage 
zur Slavistik IX. Munich: Otto Sagner, 7-30. 
- -. (2003b) "Das Verbalprafix po- im Polnischen". Zeitschrift fur slavische Philo- 
logie 62: 359-385. 
Avilova, N. S. (1964) "Razvitje pristavotnogo glagol'nogo slovoobrazovanija". V. V. 
Vinogradov and N. Ju. Svedova, eds. Oterki po istoriteskoj grammatike rus- 
skogo literaturnogo jazyka XZX veka. Glagol, naretie, predlogi i sojuzy. Mos- 
cow: Izdatel'stvo "Nauka", 45-104. 
Atraxovit (Krapiva) K. K. and M. B. Bulaxaw, eds. (1962) Hramatyka belaruskaj 
movy. Tom I: Marfalohija. Minsk: Vydavectva Akadbmii navuk BSSR. 
Bajec, Anton. (1959) Besedotvorje slovenskega jezika. ZV: Predlogi in predpone. 
Ljubljana: Slovenska akademija znanosti in umetnosti. 
Barentsen, Adrian. (1995) "Trexstupentataja model' invarianta sovergennogo vida v 
russkom jazyke". Stanislaw Karolak, ed. Semantika i struktura slavjanskogo 
vida I. Cracow: Wydawnictwo Naukowe WSP, 1-26. 
Bermel, Neil. (1997) Context and the Lexicon in the Development of Russian As- 
pect. Berkeley: University of California Press. 
BLDR 7 = D. S. Lixatev et al., eds. (2000) Biblioteka literatury drevnej Rusi. Tom 7. 
St. Petersburg: Nauka. 
Bogdanova, V. A. (1963) "0 pristavkax vy- i iz- v slavjanskix jazykax". I. A. Vinni- 
kova, ed. Voprosy slavjanskoj filologii. K V Meidunarodnomu s'kzdu slavistov. 
Saratov: Izdatel'stvo Saratovskogo universiteta, 297-308. 
Breu, Walter. (2000) "Der Verbalaspekt in der Obersorbischen Umgangssprache im 
Rahmen des ILA-Modells". Walter Breu, ed. Slavistische Linguistik 1999. Re- 
ferate des XXV. Konstanzer Slavistischen Arbeitstreffens, Konstanz 
7.-10. 9. 1999. Munich: Otto Sagner, 37-76. 
Bulyko, Aleksandr Nikolaevit, et al. (1990) Sopostavitel'noe opisanie russkogo i belo- 
russkogo jazykov. Morfologija. Minsk: Navuka i tbnika. 
Camus, Remi. (1998) "Quelque considCrations sur le prCverbe po- en russe contem- 
porain". Revue des ~ t u d e s  Slaves LXX(1): 101-112. 
Certkova, Marina. (1996) Grammatiteskaja kategorija vida v sovremennom russkom 
jazyke. Moscow: Izdatel'stvo Moskovskogo universiteta. 
Cockiewicz, W. (1992) Aspekt nu tle systemu slowotwbrczego polskiego czasownika i 
jego funkcyjne odpowiedniki w jezyku niemieckim. Krak6w: Naklad Uniwersy- 
tetu Jagiellonskiego. 
Comrie, Bernard. (1976) Aspect. Cambridge: Cambridge University Press. 
Dickey, Stephen M. (2000) Parameters of Slavic Aspect: A Cognitive Approach. 
Stanford: CSLI. 
- -. (2001a) "The Temporalization of the Prefix po- in Slavic". Paper presented at 
the Second Annual Conference of the Slavic Cognitive Linguistics Association, 
University of Virginia, Charlottesville, VA. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 51 
- -. (2001b) "Semelfactive -nq- and the Western Aspect Gestalt". Journal of Slavic 
Linguistics 9(1): 25-48. 
- - . (2003) "Verbal aspect in Slovene". Sprachtypologie und Universalienforschung. 
Dickey, Stephen M. and Hutcheson, Julie. (2003) "Delimitative Verbs in Russian, 
Czech and Slavic". Robert A. Maguire and Alan Timberlake, eds. American 
contributions to the Thirteenth International Congress of Slavists. Columbus, 
Ohio: Slavica, 23-36. 
Dmitrieva, Ol'ga. (2000) "Formirovanie sistemy russkix delimitativnyx glagolov". T. 
V. KoEetkova, ed. Predloienie i slovo: paradigmatiteskij, tekstovyj i kommuni- 
kativnyj aspekty. Saratov: Izdatel'stvo Saratovskogo pedagogiCeskogo instituta, 
28-33. 
Gebauer, Jan. (1963)yHistorickd mluvnice jazyka tesktho. Dil I: Hldskoslovi. Prague: 
Nakladatelstvi Ceskoslovensk6 akademie vEd. 
Greenberg, Marc L. (2000) A Historical Phonology of the Slovene Language. Hei- 
delberg: Carl Winter Verlag. 
Grickat, Irena. (1957) "0 nekim vidskim osobenostima srpskohrvatskog glagola". 
Juinoslovenski filolog XXII: 65-128. 
Guiraud-Weber, Marguerite. (1998) "La prkfixation des emprunts verbaux en russe et 
en polonais". Revue des Etudes Slaves LXXI1: 67-77. 
Hauser, Pfemysl. (1995) "Slovesa se zdvojenjrmi pfedponami zne-, zane- (zneklidnit, 
zanedbat)". Petr Karlik, Jana Pleskalovi and Zdenka Rusinovi, eds. Pocta Du- 
Sanu Slosarovi. Sbornik k 65. narozenindm. Boskovice: Nakladatelstvi Albert. 
HEWONS = (1978-89) Historisch-etymologisches Worterbuch der ober- und nieder- 
sorbischen Sprache. Bautzen: Domowina-Verlag. 
Hopper, Paul J. and Elizabeth Closs Traugott. (1993) Grammaticalization. Cambridge: 
Cambridge University Press. 
Horeckjr, Jin. (1959) Slovotvorna sustava sloventiny. Podstatnt meno, pridavnt 
meno, sloveso. Bratislava: Vydavatelstvo Slovenskej akadkmie vied. 
Hujer, Oldfich. (192211961) "Praslovansk6 vaz-, jbz- v EeStinE". In PiispZvky k histo- 
rii a dialektologii tesktho jazyka. Prague: Nakladatelstvi Ceskoslovensk6 aka- 
demie vEd, 116-125. 
IvanCev, Svetomir. (1971) Problemi nu aspektualnostta v slavjanskite ezici. Sofia: 
Balgarska akademija na naukite. 
Ivanova, Kalina. (1966) Desemantizicija nu glagolnite predstavki v savremennija 
balgarski kniioven ezik. Sofia: Balgarska akademija na naukite. 
Janka, Wolfgang. (1997) "Zur Unterscheidung der tschechischen Verbalprafixe s(e)- 
und z(e)-". Anzeiger fur slavische Philologie XXV: 101-11. 
Jedvaj, Josip. (1956) "Bednjanski govor". Hrvatski dijalektoloSki zbornik 1: 279-330. 
Kalsbeek, Janneke. (1998) The cakavian Dialect of Orbanic'i near ~ m i n j  in Zstria. 
Amsterdam: Rodopi. 
Klemensiewicz, Z., T. Lehr-Splawiriski and S. Urbariczyk. (1981) Gramatyka histo- 
ryczna jezyka polskiego. Warsaw: Panstwowe w ydawnictwo naukowe. 
52 Slovenski jezik - Slovene Linguistic Studies 5 (2005) 
KopeEnf, FrantiSek. (1962) Slovesny' vid v CeitinZ. Prague: Nakladatelstvi Ceskoslo- 
venskk akademie vEd. 
Kucala, Marian. (1966) Rozwbj iteratiwbw dokonanych w jezyku polskim. Wroclaw: 
Zaklad Narodowy im. Ossolinskich. 
Langacker, Ronald. (1987) Foundations of Cognitive Grammar: Volume 1. Stanford: 
Stanford University Press. 
- -. (1988) "A Usage-Based Model". B. Rudzka-Ostyn, ed. Topics in Cognitive Lin- 
guistics. Amsterdam: John Benjamins, 127-61. 
Lehmann, Christian. (1995) Thoughts on Grammaticalization. Munich: Lincom Eu- 
ropa. 
Lehmann, Volkmar. (1988) "Der russische Aspekt und die lexikalische Bedeutung 
des Verbs". Zeitschrift fur slavische Philologie 48(1): 171-81. 
- - . (1999) "Sprachliche Entwicklung als Expansion und Reduktion". Tanja Anstatt, 
ed. Entwicklungen in slavischen Sprachen. Munich: Otto Sagner, 160-254. 
Marvan, Jifi. (2000) Jazykovd mildnium. Slovanskd kontrakce a jeji Cesky' zdroj. 
Prague: Academia. 
Maslov, Jurij S. (1961) "Rol' tak nazyvaemoj perfektivacii i imperfektivacii v processe 
vozniknovenija slavjanskogo glagol'nogo vida". V. V. Vinogradov, ed. Issledo- 
vanija po slavjanskomu jazykoznaniju. Moscow, 165-95. 
Mende, Julia. (1999) "Die Grammatikalisierung des russischen Aspekts". Tanja An- 
statt, ed. Entwicklungen in slavischen Sprachen. Munich: Otto Sagner, 285-325. 
Merge, Majda (1995) Vid in vrstnost glagola v slovenskem knjiinem jeziku 16. sto- 
letja. Ljubljana: Slovenska akademija znanosti in umetnosti. 
MSJ = RuiiCka, Jozef, ed. (1966) Morfolbgia slovenskdho jazyka. Bratislava: Vyda- 
vatel'stvo Slovenskej akadCmie vied. 
Nefed'ev, M. V. (1994) "SemantiEeskaja evoljucija glagol'nyx pristavok nu- i ob- v 
istorii russkogo jazyka XI-XVIII vv.". Voprosy jazykoznanija 4: 73-83. 
Nichols, Johanna and Alan Timberlake. (1991) "Grammaticalization as Retextualiza- 
tion". Elizabeth Cross Traugott and Bernd Heine, eds. Approaches to Gram- 
maticalization 1: Focus on Theoretical and Methodological Issues. Amsterdam: 
John Benjamins, 129-46. 
N~rglrd-Ssrensen, Jens. (1997) "Tense, Aspect and Verbal Derivation in the Lan- 
guage of the Novgorod Birch Bark Letters". Russian Linguistics 21: 1-21. 
Niibler, Norbert. (1990) "Zum Begriff der Subsumptionsprafixe in der Aspektfor- 
schung". Anzeiger fur slavische Philologie XX: 123-34. 
OSSJ = Odzadnji slovar slovenskega jezika. Ljubljana: Slovenska akademija znanosti 
in umetnosti. (CD-Rom version.) 
Peco, Asim. (1991) Pregled srpskohrvatskih dijalekata. Belgrade: NauCna knjiga. 
Piernikarski, Cezar. (1975) Czasowniki z prefiksem po- w jezyku polskim i czeskim 
nu tle rodzajbw akcji w jezykach slowiariskich. Warsaw: Panstwowe wydawnic- 
two naukowe. 
Plotnikova, Ol'ga. (1971) "Dvuvidovye glagoly inostrannogo proisxoidenija v sloven- 
skom literaturnom jazyke". Vestnik Moskovskogo universiteta 1: 28-36. 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 53 
Poldauf, Ivan. (1954) "Spojovhni s pFedponami pFi tvoFeni dokonavfch sloves v EeSti- 
n?'. Slovo a slovesnost XVl2: 49-65. 
Rusanivs'kyj, V. M. (1978) "Dieslovo". V. V. NimEuk, ed. Istorija ukrajins'koji movy. 
Morfologija. Kiev: Naukova dumka, 235-341. 
Rusinovh, Zdenka. (2001) "Slovesnk neologismy a problkm vidu". Mieczyslaw Balow- 
ski and JiFi Svoboda, eds. Jezyk i literatura czeska u schytku XX wieku. Wal- 
brzych1Ostrava: 219-225. 
Saloni, Z. (2001) Czasownik polski. Odmiana. Slownik. Warsaw: Wiedza Powszechna. 
SeliSEev, L. M. (194111969) Slavjanskoe jazykoznanie 1: Zapadnoslavjanskie jazyki. 
Moscow: Narkomprosa. (C. H. van Schooneveld, ed. Slavistic Printings and Re- 
printings 106. The Hague: Mouton.) 
Serex, Jurij. (1951) Narys sutasnoji ukrajins'koji literaturnoji movy. Munich: Molode 
iyttja. 
Shull, Sarah. (2003) The Experience of Space: The Privileged Role of Spatial Pre- 
fixation in Czech and Russian. Munich: Otto Sagner. 
Sigalov, P. S. (1975) "Istorija russkix ogranicitel'nyx glagolov". Trudy po russkoj i 
slavjanskoj filologii: Serija lingvistiteskaja 23: 141-81. 
Slonski, Stanislaw. (1937) Funkcje prefiksbw werbalnych w jezyku staroslowiariskim 
(starobutgarskim). Warsaw: Naklad Towarzystwa Naukowego Warszawskiego. 
Slosar , DuSan. (1981) Slovotvorny' vjvoj teskdho slovesa. Brno: Univerzita J. E. 
PurkynE. 
~miech,  Witold. (1986) Derywacja prefiksalna czasownikbw polskich. Zaklad Naro- 
dowy im. Ossolifiskich. 
Smirnov, L. N. (1970) Glagol'noe vidoobrazovanie v sovremennom slovackom jazyke. 
Moscow: Nauka. 
SSJC' = (1989) Slovnik spisovndho jazyka teskdho. Prague: Nakladatelstvi Eeskoslo- 
vensk6 akademie vEd. 
SSKJ = (1970-91) Slovar slovenskega knjiinega jezika. Ljubljana: Slovenska akade- 
mija znanosti in umetnosti. (CD-Rom version.) 
SSM = (1977) Slovnyk staroukrajins'koji movy XZV-XV st. Kiev: Naukova dumka. 
Staroslavjanskij slovar' = R. M. Cejtlin, R. VeEerka, E. Blagovoj, eds. (1994) Staro- 
slavjanskij slovar' (po rukopisjam X-XI vekov). Moscow: Russkij jazyk. 
Styreva, E. D. (1992) "K voprosu o vzaimodejstvii kategorii vida s glagol'noj leksi- 
koj". L. N. Smirnov, ed. Zssledovanija po slovackomu jazyku. Moscow: Rossij- 
skaja akademija nauk, 166-78. 
Szelesifiski, Iwo. (1972) "Wyrazy z przedrostkiem wz- (ws-, wez-, wes-) w polszczy- 
inie". Rozprawy komisji jezykowej XVIII: 211-29. 
Tixonov, A. N. (1998) "Vidovye korrelacii v sovremennom russkom jazyke". Marina 
Ju. Certkova, ed. Tipologija vida: Problemy, poiski, reienija. Moscow: 466-77. 
Toops, Gary H. (2001) "Aspectual Competition and Iterative Contexts in Contempor- 
ary Upper Sorbian". Journal of Slavic Linguistics 9(1): 127-54. 
54 Slovenski iezik - Slovene Linguistic Studies 5 (2005) 
TrivniEek, Frantigek. (1923) Studie o teskkm vidu slovesnkm. Prague: Niklad Cesk6 
akademie vEd a umEni. 
Ugrinova-Skalovska, Rada. (1960) Znatenjata nu glagolskite prefiksi vo makedon- 
skiot jazik. Skopje: Institut za makedonski jazik. 
Vintr, Josef. (2001) Das Tschechische. Hauptziige seiner Sprachstruktur in Gegen- 
wart und Geschichte. Munich: Otto Sagner. 
Voloxina, G. A., and Z. D. Popova. (1997) "Kategorija glagol'nogo vida v svete se- 
mantiteskogo ustrojstva glagol'nyx pristavok". Marina Ju. Certkova, ed. Trudy 
aspektologiteskogo seminara Filologiteskogo fakul'teta MGU im. M. V .  Lomo- 
nosova 3. Moscow: 34-41. 
Werner, Eduard. (2003) Die Verbalaffigierung im Obersorbischen. Bautzen: Domo- 
wina-Verlag. 
Witkowska-Gutkowska, Maria. (1999) Staropolskie prefiksalne dublety czasownikowe 
i ich wspbtczesne odpowiedniki. L6di: Wydawnictwo Uniwersytetu Mdzskiego. 
Xlebnikova-Prokopovit, E. N. (1956) "Glagol'noe formoobrazovanie pri pomogEi pri- 
stavok v pamjatnikax russkoj pis'mennosti vtoroj poloviny XVII veka". Doklady 
i soobitenija Instituta jazykoznanija 10: 130-55. 
Zel'dovit, Gennadij. (2002) Russkij vid: Semantika i pragmatika. Torun: Uniwersytet 
Mikolaja Kopernika. 
Prispelo decembra 2004, sprejeto februarja 2005 
Received December 2004, accepted February 2005 
S-/Z- in gramatikalizacija glagolskega vida v slovanSEini 
Prispevek obravnava nastanek p~edpone s-/z- kot posledico sovpada predpon 
*s5- in *jbz- po izgubi polglasnikov. Ceprav se je enotna predpona razvila v s-lz- v 
EeSEini, slovagtini, luiigtini, slovengEini, ukrajingEini in belorugEini, pa je imela naj- 
veE posledic za glagolski vid v Eegtini, slovagtini, 1uiigEini in slovengtini, kjer je 
predpona postala najbolj produktivna (izkljutno) kot prkverbe vide. Drugi slovanski 
jeziki izkazujejo drugatne sisteme predponske dovrgnosti. V hrvagtini in srbgEini se 
ohranja arhaiten sistem, v katerem se leksikalno enakovredni dovrgni glagoli tvorijo 
iz nedovrgnih z razliEnimi predponami. V poljgEini se je razvila druga vrsta prkver- 
be vide, namret s predponama po- poleg inovativne s-lz-, enako tudi v ukrajingEini 
in belorugtini. V preostalih slovanskih jezikih (rugtina, bolgargtina, makedongtina) 
se je kot prkverbe vide razvila predpona po- (tu ni priglo do sovpada *s5- in *jbz-). 
Vseslovanska slika o predponah s-/z- in po- v preteini meri ustreza Dickeyjevi 
(2000) razvrstitvi slovanskih jezikov v dve skupini glede na glagolski vid: vzhodno 
(ruSEina, ukrajingEina, belorugtina in bolgargEina) in zahodno (Eegtina, slovagtina, 
luiigtina in slovengtina) z dvema prehodnima conama (poljgtina in hrvagEina ter 
srbgEina). Avtor zagovarja staligEe, da se je predpona s-/z- gramatikalizirala kot prk- 
verbe vide, pri Eemer je v leksikalnih vrstah, v katerih s-lz- nastopa kot prkverbe 
vide, priglo do delitve s preubeseditvijo (retextualization). Ta postopek je pripomogel 
k ustalitvi celostnosti (totality) kot osnovnega pomena zahodnega dovrinega vida, iz 
S. M. Dickev. S-12- and Grammaticalization of Asvect in Slavic 55 
katerega izvira sodobni zahodni sistem vidske rabe. Vzporedni postopek s predpono 
po- v vzhodnih jezikih naj bi igral pomembno vlogo pri razvoju pomena vzhodnega 
dovrgnika v kategorijo Easovne doloEnosti (temporal definiteness) in ustvaril sodobni 
sistem vidske rabe v teh jezikih. 
S-12- and the Grammaticalization of Aspect in Slavic 
This paper examines the rise of s-Iz- as a coalescence of in the prefixes *sa- 
and *jbz- resulting from jer-fall. Though s-/z- was developed in Czech, Slovak, Sorb- 
ian, Slovene, Polish, Ukrainian and Belarusian, it has had the most consequences 
for aspect in Czech, Slovak, Sorbian and Slovene, where this prefix has arguably 
been most productive (and exclusively so) as a priverbe vide. The other Slavic lan- 
guages have developed different systems of perfectivizing prefixation. Croatian and 
Serbian are argued to represent an archaic system of subsumption, relying largely on 
the semantic overlap of various prefixes with imperfective source verbs to create lex- 
ically identical perfective partner verbs. Polish has developed another priverbe vide, 
po-, alongside innovative s-/z-, as have Ukrainian and Belarusian. The remaining 
Slavic languages, Russian, Bulgarian and Macedonian developed po- as a priverbe 
vide (and never merged *sa- and *jbz). The cross-Slavic picture regarding s-Iz- and 
po- corresponds in large part to Dickey's (2000) division of Slavic into two aspectual 
groups, an eastern group (Russian, Ukrainian, Belarusian and Bulgarian) and a 
western group (Czech, Slovak, Sorbian and Slovene), as well as two transitional zones 
(Polish and CroatianJSerbian). It is argued that s-Iz- underwent a grammaticalization 
process whereby the lexical classes of verbs for which it functions as a priverbe vide 
were diversified by a process of retextualizaton, which helped to stabilize totality as 
the meaning of the western perfective, producing the current western pattern of 
usage. A parallel process involving po- in the eastern languages is hypothesized to 
have played an important role in the development of the meaning of the eastern 
perfective to a category called temporal definiteness, which has produced the cur- 
rent eastern pattern of aspect usage.