Professional Documents
Culture Documents
Slavic Domain
Edited by
Dagmar Divjak
Agata Kochanska
Mouton de Gruyter
Cognitive Paths into the Slavic Domain
≥
Cognitive Linguistics Research
38
Editors
Dirk Geeraerts
René Dirven
John R. Taylor
Honorary editor
Ronald W. Langacker
Mouton de Gruyter
Berlin · New York
Cognitive Paths
into the Slavic Domain
Edited by
Dagmar Divjak
Agata Kochańska
Mouton de Gruyter
Berlin · New York
Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter GmbH & Co. KG, Berlin
앪
앝 Printed on acid-free paper
which falls within
the guidelines of the ANSI
to ensure permanence and durability.
ISBN 978-3-11-019620-7
ISSN 1861-4132
쑔 Copyright 2007 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin
All rights reserved, including those of translation into foreign languages. No part of this book
may be reproduced or transmitted in any form or by any means, electronic or mechanical,
including photocopy, recording, or any information storage and retrieval system, without
permission in writing from the publisher.
Printed in Germany
Preface
The present volume contains a selection of research on Slavic languages
carried out within the Cognitive Linguistics framework. Most of the papers
were originally presented at the 3rd Slavic Cognitive Linguistics
Conference, which took place in September 2004 in Leuven (Belgium).
Contributions by Barbara Dancygier, Laura Janda and Elżbieta
Tabakowska were later solicited specifically for this volume. All of the
papers have been written or revised with one goal in mind: presenting a
volume of interest to both Slavic linguists and cognitive linguists.
To the extent that we have succeeded in achieving our goal, we are
greatly indebted to the many linguists who have lent us their expertise, be
it on Slavic or non-Slavic languages, in cognitive or descriptive linguistics,
i.e. Neil Bermel, Mario Brdar, Alan Cienki, Steven Clancy, Hubert Cuy-
ckens, Östen Dahl, David Danaher, Barbara Dancygier, Larry Feinberg,
Dirk Geeraerts, Elżbieta Górska, Stefan Gries, Gaëtanelle Guilquin,
Tuomas Huumo, Laura Janda, Robert Kirsner, Wojciech Kubiński, Ron
Langacker, Geoffrey Nathan, Elena Petroska, Anna Siewierska, Michael
Smith, Elżbieta Tabakowska, Willy Van Langendonck and Margareth
Winters. We would also like to express our gratitude to René Dirven, Dirk
Geeraerts, Laura Janda and Ron Langacker for aiding us in taking our idea
from book proposal to final product. Torkel Uggla and Ines Van Houtte
prepared the manuscript for publication – a task that was financially
supported by the Science Foundation Flanders (Belgium) – while Birgit
Sievert walked us through the production process. Last but not least, warm
thanks go to our respective husbands, Torkel and Andrzej, for supporting
our work, both in theory and in practice, as well as to Agata’s mother
Elżbieta Horszczaruk for her assistance in matters of everyday life.
While working on this volume we were guided by the strongly-felt hope
that the wide spectrum of cognitively-oriented research on Slavic data
presented would contribute to the appreciation of both the beauty of the
notoriously complex Slavic languages and the power of the theoretical
tools developed within the cognitive framework. These tools allow a
linguist to tackle language phenomena in all their wonderful complexity
and to enjoy subtleties and intricacies without any need to disregard non-
conforming facts or to force natural language into artifical shapes.
Preface ..................................................................................................... v
Why cognitive linguists should care about the Slavic languages and
vice versa ……………………………………………………………… 1
Dagmar Divjak, Laura A. Janda and Agata Kochańska
Part two. The verbal system: the meaning of tense, aspect and
mood
What makes Russian bi-aspectual verbs special? ................................. 83
Laura A. Janda
From its early days, cognitive linguistics has attracted the attention of lin-
guists with research interests in Slavic languages (to name but a few,
Cienki 1989; Dąbrowska 1997; Janda 1993a; Rudzka-Ostyn 1992 and
1996). In recent years this interest has rapidly expanded, as can be wit-
nessed by the establishment of the Polish Cognitive Linguistics Associa-
tion, the Russian Cognitive Linguistics Association, and the Slavic Cogni-
tive Linguistics Association, as well as by the many Slavic Cognitive
Linguistics conferences held at various venues in Europe and North Amer-
ica over the last seven years.
This is not surprising, for at least two reasons. First, one of the founding
assumptions of cognitive linguistics has been present in Slavic linguistics
all along: Slavic linguists have always recognized the fundamentally sym-
bolic nature of language and hence the fact that diverse formal aspects of
language exist for the purpose of conveying meaning. One striking illustra-
tion of the close affinities between cognitive linguistics and ideas formu-
lated within traditional Slavic linguistics comes from the relatively early
days of modern linguistic research on Slavic languages. In a study devoted
to the nature of the contrast between the perfective and the imperfective
aspect in Polish, a German Slavicist, Erwin Koschmieder (1934), proposed
two conceptualizations of time which could easily be paraphrased as in-
volving either the MOVING TIME metaphor for the perfective or the
MOVING EGO metaphor for the imperfective (for a discussion of the two
time metaphors see Radden 1991: 17ff). Other examples abound. Tradi-
tional analyses of Polish case by Kempf (1978), Klemensiewicz (1926) and
Szober (1923 [1963]) aimed to provide a full-fledged semantic analysis of
Polish case. This type of work with its emphasis on psychologically realis-
tic explanations, has always been “a characteristic feature of Polish
(Slavic?) linguistics” (Tabakowska 2001:12; translation AK), and contin-
2 Dagmar Divjak, Laura A. Janda and Agata Kochańska
In the next section we will briefly discuss the main theoretical concepts
developed so far within the cognitive paradigm, with special emphasis on
those assumptions and ideas that are most directly relevant to the analyses
offered in the present volume.
Over the last three decades in mainstream linguistics the conviction has
grown that language is not a purely formal, algorithmic system processed
in a separate language faculty. Instead, our language capacity is considered
an integrated part of human cognition. The description of language is thus
a cognitive discipline, part of the interdisciplinary field of cognitive sci-
ences. One of the fundamental qualities of human cognition that is most
pervasively present in language is categorization.
semantics of tense, aspect, and mood, to name but a few grammatical cate-
gories.
An important aspect of the conceptualistic view on meaning is the rec-
ognition of the imagistic component of semantics, that is, of the fundamen-
tal role construal plays in meaning. A precise characterization of its dimen-
sions allows an analyst to offer detailed and rigorous characterizations of
meaning contrasts among linguistic structures which are equivalent in
truth-conditional terms, but nevertheless exhibit subtle yet important dif-
ferences in meaning, resulting in otherwise unexplainable differences in
discourse behavior. A principled account of construal is a necessary pre-
requisite for developing a full-fledged symbolic approach to grammar:
grammatical meaning is by necessity abstract and can hardly be character-
ized in terms of specific conceptual content. It may, nevertheless, be in-
sightfully analyzed in terms of the type of construal it imposes on con-
ceived scenes, as demonstrated, for example, by the highly revealing
notional characterizations of nouns and verbs proposed by Langacker (cf.
e.g. 1987b).
Tomasello 2003). At the same time, a usage-based view provides the right
perspective for the full appreciation of corpus studies in linguistic research
that no longer asks whether a certain phenomenon is possible or impossi-
ble, but instead focuses on how likely or unlikely the pattern is to occur
(see Gries and Stefanowitsch 2006). Last but not least, the adoption of the
usage-based model is important for the study of language change, as it lays
the ground for recognizing the role that is played in historical linguistic
evolution by factors such as frequency and mechanisms such as context-
bound pragmatic inferencing.
The purpose of the present volume is twofold. On the one hand, we want to
investigate to what extent the theoretical framework and analytic tools
developed within cognitive linguistics can be insightfully applied to the
study of Slavic languages. As may be apparent from the brief discussion in
section 2 above, Slavic languages, with their rich inflectional morphology
in both the nominal and the verbal system, provide an important testing
ground for a linguistic theory that seeks conceptual motivation behind
grammatical phenomena. On the other hand, the specific observations and
insights arrived at in the course of cognitively-oriented analyses of diverse
phenomena in Slavic languages may enrich the understanding of already
established aspects of the cognitive model of language and serve as cata-
lysts for their further development and refinement.
This volume is important for a number of reasons. First, as far as its de-
scriptive range is concerned, the volume deals with a variety of empirical
phenomena that are of major interest to any linguistic theory. As men-
tioned above, the topics discussed include the semantics of case, tense, and
aspect, complex event conceptions, voice phenomena, word order, sound
symbolism, and language change. Secondly, the analyses address a variety
of theoretical issues that are important for cognitive linguistics in general.
Among them the reader will find: the role of virtual entities in language,
the importance of subjectification in motivating both synchronic polysemy
and diachronic language change, different ways of conveying the speaker’s
epistemic attitude, various kinds of non-prototypical event conceptions and
their grammatical reflections, the role of metaphor in grammaticalization,
and the influence exerted by local, contextual factors of pragmatic nature
in diachronic morphosyntactic change. Topics of general theoretical inter-
est also include the issue of iconicity in language and the idea that overtly
12 Dagmar Divjak, Laura A. Janda and Agata Kochańska
Notes
References
Apresjan, Ju.D.
1995 Izbrannye Trudy. Tom 1, Leksičeskaja Semantika [Selected Writings,
Part 1, Lexical Semantics]. Moskva: Škola “Jazyki Russkoj
Kul’tury”.
Apresjan, Ju.D.
1974 Regular polysemy. Linguistics. An international review 142:5–32.
Arutjunova, Nina D.
1999 Jazyk i mir čeloveka [Language and the world of man]. Moskva:
Škola “Jazyki Russkoj Kul’tury”. [2nd corrected edition].
Cienki, Alan
1989 Spatial Cognition and the Semantics of Prepositions in English, Pol-
ish, and Russian. Munich: Verlag Otto Sagner.
Cutrer, Michelle
1994 Time and Tense in Narrative and in Everyday Language. Unpub-
lished Ph.D. dissertation, University of California, San Diego.
Dancygier, Barbara
1998 Conditionals and Prediction. Cambridge: Cambridge University
Press.
Dancygier, Barbara, and Eve Sweetser
2005 Mental Spaces in Grammar: Conditional Constructions. Cambridge:
Cambridge University Press.
Dąbrowska, Ewa
1997 Cognitive Semantics and the Polish Dative. Berlin/New York: Mou-
ton de Gruyter.
2004 Language, Mind and Brain: Some Psychological and Neurological
Constraints on Theories of Grammar. Edinburgh: Edinburgh Univer-
sity Press.
Fauconnier, Gilles
1985 Mental Spaces. Cambridge, Mass.: MIT Press.
1997 Mappings in Thought and Language. New York: Cambridge Univer-
sity Press.
16 Dagmar Divjak, Laura A. Janda and Agata Kochańska
Geeraerts, Dirk
1988 Cognitive Grammar and the History of Lexical Semantics. In Topics
in Cognitive Iinguistics, Brygida Rudzka-Ostyn (ed.), 647–677. Am-
sterdam/Philadelphia: John Benjamins.
1997 Diachronic Prototype Semantics. Oxford: Clarendon Press.
Gries, St.Th., and A. Stefanowitsch
2006 Corpora in Cognitive Linguistics: Corpus-based Approaches to Syn-
tax and Lexis. Berlin/New York: Mouton de Gruyter.
Janda, Laura A.
1986 A Semantic Analysis of the Russian Verbal Prefixes za-, pere-, do-,
and ot-. Munich: Verlag Otto Sagner.
1993a A Geography of Case Semantics: The Czech Dative and the Russian
Instrumental. Berlin/New York: Mouton de Gruyter.
1993b Cognitive Linguistics as a Continuation of the Jakobsonian Tradition:
The Semantics of Russian and Czech Reflexives. In American Con-
tributions to the Eleventh International Congress of Slavists in Brati-
slava, Robert A. Maguire and Alan Timberlake (eds.), 310–319. Co-
lumbus, Ohio: Slavica.
Janda, Laura A., and Steven J. Clancy
2002 The Case Book for Russian. Bloomington, IN: Slavica.
Kempf, Zdzisław.
1978. Próba teorii przypadków [An attempt at a theory of case]. Wrocław:
Zakład Narodowy im. Ossolińskich.
Klemensiewicz, Zenon
1926 Orzecznik przy formach osobowych być [The predicate with personal
forms of ‘be’]. Prace filologiczne [Philological works] XI: 123–181.
Koschmieder, Erwin
1934 Nauka o aspektach języka polskiego w zarysie. Próba syntezy. [An
attempt at a comprehensive theory of Polish aspect. An outline].
Wilno.
Lakoff, George
1987 Women, Fire, and Dangerous Things. What Categories Reveal about
the Mind. Chicago: University of Chicago Press.
1990 The Invariance Hypothesis: Is Abstract Reason Based on Image-
Schemas? Cognitive Linguistics 1: 39–74.
Lakoff, George, and Mark Johnson
1980 Metaphors We Live By. Chicago: Chicago University Press.
1999 Philosophy in the Flesh. The Embodied Mind and Its Challenge to
Western Thought. New York: Basic Books.
Langacker, Ronald W.
1987a Foundations of Cognitive Grammar. Vol. 1. Theoretical Prerequi-
sites. Stanford, California: Stanford University Press.
1987b Nouns and Verbs. Language 63: 53–94.
Why cognitive linguists should care about Slavic languages 17
Vaňková, Irena, Iva Nebeská, Lucie Saicová Římalová, and Jasňa Šlédrová
2005 Co na srdci, to na jazyku: Kapitoly z kognitivní lingvistiky [What's
on the heart is on the tongue: Chapters from cognitive linguistics].
Prague: Karolinum.
Wierzbicka, Anna
1972 Semantic Primitives. Frankfurt: Athenäum.
ȱȱ
ȱȱ¢DZȱȱȱȱ
ȱ
Nominative and instrumental variation of adjectival
predicates with the Russian copula byt': reference
time, limitation, and focalization
Alina Israeli
Abstract
1. Introduction
Černov (1983: 91), who analyzes the same constructions, observes that
in the case of the “actualization of the temporal plan”, particularly with
time expressions emphasizing distance in time, the instrumental is usually
used. Černov labels this usage ‘pluperfect’ (the following examples are
from Černov 1983: 91):
The past form byl may represent three different types of narrative time, the
first two of which involve comparison:
1. The time of the narrated event tn is compared with the time of speech (or
writing) ts. I will call this use of byl tn / ts. This is the case in (5):
2. The time of one narrated event tn1 is compared with the time of another
narrated event tn2. I will call this use of byl tn1 / tn2. This is the case in (6),
where pre-war memories of seventh grade (togda ‘at that time’) are com-
pared with post-war ones of a class reunion (‘lived to have grey hair’):
3. The time of the narrated event tn is not compared with another moment
in time. I will call this use of byl tn0. This is the case in (4a) and (4b) above
as well as in (7) below. It differs from type 1 by having to present events as
frozen in the moment: unlike Russia and mother in (5), raspberry in (7) did
not exist at a later moment thus not allowing even an implicit comparison.
In (4), it is the choice of the narrator to present events as frozen in time and
defying comparison.
When an object ceases to exist, its feature ceases to exist along with it,
and the information conveyed by the instrumental is that the comparison tn1
/ tn2 indicates a change (cf. Filip 2001). A Google search <29.V.2006>
produced eighty-seven examples of the nominative dom byl kirpičnyj and
five examples of the instrumental dom byl kirpičnym, four of which per-
tained to one and the same event of the explosion and the subsequent col-
lapse of an apartment building in Moscow. So in order to motivate this use
of the instrumental, the apartment building had to cease to exist:
In (11), the person in question has died, and consequently the feature
ascribed to his face has gone with him:
Case variation of adjectival predicates 27
Similarly, in (13) the speaker sees the openness of the face as a fleeting
momentary feature:
As we have seen in the case of otkrytyj ‘open’, the same feature could
be a permanent characteristic or a non-permanent one. This is also true for
adjectives like bol'šoj ‘big’. Comparing two types of elephants, the African
ones vs. the Asian one, while explaining the nominative vs. instrumental
variation in the predicates, Ionin and Matushansky (2002: 4) suggest that in
(14) “Nominative requires the subject to vary from situation to situation
28 Alina Israeli
In fact, it is not that the nominative case that requires the subject to
vary, but the quality of the feature that the nominative case represents: this
is what renders (14b) incongruent. Co-reference of the subject in (14b), as
opposed to non-co-reference in (14a), can be explained via reference to
permanence/ non-permanence of the charecteristic and to temporal com-
parison. In (14b) v Afrike ‘in Africa’ and v Azii ‘in Asia’ mean ‘while in
Africa’ and ‘while in Asia’ respectively, thus involving a temporal com-
parison of the type tn1 / tn2 and yielding a pragmatically impossible sen-
tence with non-permanent feature. In (14a) bol'šoj ‘big’ is a permanent
characteristic (pertaining not to individual elephants but to a species) not
allowing for temporal comparison.
The narrative time tn1 / tn2 de-facto means change or comparison. This
type of narrative time, however, offers two possibilities. Analyzing evolv-
ing reference, Moeschler (1996: 21) suggests that it can be construed either
as temporal sequencing (TS) or via a retrospective perspective (RP). In TS
events are viewed sequentially, that is, tn1 is perceived and described as a
point in time earlier than tn2, and tn2 is compared to tn1; in other words, we
are dealing with tn1 / tn2 proper. In RP the narration can view previous
points in time via flashbacks, or view the preceding moment tn1 as com-
pared to the “present” (tn0) or to a later point tn2, so this is strictly speaking
a reversed narrative time tn2 / tn1. In terms of temporal perspective, TS
views the narration as “then and later”, while RP views it as “now and
before”. The TS perspective, as in (15), calls for the nominative in the ad-
Case variation of adjectival predicates 29
jectival phrase describing the state at t1, whereas the RP, as in (16), calls
for the instrumental.
The stative (non-evolved) view, as in (15), and RP, as in (16), can be com-
bined and juxtaposed, as in (17):
The first occurence of staren'kij ‘old’ is stative, the kind that Nichols
describes as having descriptive force and Zel'dovič as observed, while the
second occurence with teper' ‘now’ (cf. Mel'čuk 1985) implies compari-
son, a change of state.
In (18), which refers to a meeting of two school friends some twenty
years after graduation and the war, there are two instances of the nomina-
tive case:
One instance pertains to the moment tn0 (on byl lysyj ‘he was bald’) and
the other to a moment twenty years earlier at tn1 (oni byli nebesno-golubye
‘they were sky blue’). By using the phrase “I remembered”, immediately
before the second instance, the narrator justifies the transposition into a
different time which allows him to present the quality of eye color as sta-
tive, despite the change. However, the features of kindness and gullibility
are presented as changed, and consequently the message is that Venya no
longer believed everything he heard or read. The speaker is thus able to
interweave two different narrative times.
While future tense examples are not discussed here in the same detail
due to their infrequency (relative to past tense examples), it is worthwhile
to mention an example discussed in Zel'dovič (2005: 141). The author
states that (19a) rather than (19b) suggests that the speaker intends to put
the sweater on or empathizes with someone who intends to wear the
sweater due to the observation factor:
I believe that the fact that the speaker (or someone else) is planning to
wear the sweater tomorrow makes the process of drying and implicit com-
32 Alina Israeli
parison between the current wet and future dry state of the sweater immate-
rial; what matters most is the state it is in when worn.
Conversely, in (20) it is the result of drying that is important, and thus
implicit comparison is brought to the fore:
There exist nouns and attributes that cannot pertain to any particular
moment in time, but only to a period, yet the features that they describe
cannot be considered permanent and described statively. This is the case
with the nouns ‘searches’ and ‘influence’ in (21) below:
3. Non-temporal comparison
The use of the instrumental may signal not only a comparison between two
different states of a single entity at two different points in time, but also a
comparison between two different entities, or different parts of the same
entity. In (22), the author compares two generations, and his verdict (using
the instrumental) is that they are different, not alike:
4. Limitation
Limitation in scope refers to a feature either not being consistent, not mani-
festing itself constantly, apparent only within a given frame of reference, or
not being assumed to have impact beyond the speaker. Consider (24), in
which an intelligent person behaves stupidly, thus setting a limit to his wit.
Case variation of adjectival predicates 35
In (25a) the quality of being made of brick is limited to one floor (as
opposed to the rest of the house which is made of wood); in (25b), the limi-
tation is geographic (i.e. restricted to Tula) as well as in scope: the author
deliberately states the limit of their house’s worldliness, which in other
places, e.g. St. Petersburg, might not have been considered worldly at all.
Going back to (4a) and (4b), repeated below, we can see that in (4a) it is
the grandfather’s oldness and its ramifications that are the speaker’s con-
cern, while in (4b), only the implication of the oldness with respect to his
remembering the Morse code is of note:
36 Alina Israeli
i bespečnyj.
and carefree.M.SG.NOM
(È. Xeminguèj [Hemingway]. Ostrova v okeane)
‘Yes, but I remember when he was cheerful and carefree.’
5. Focalization
po odežde prisutstvujuščix.
by clothes present-people
(L.P. Korsakov, T. V. Kirpičenko. Redkie fotosnimki…
http://www.fessl.ru/publish/grodek/kor.shtml)
‘First snapshot… The day was sunny, but apparently rather cool,
judging by the clothing of those present.’
In contrast, the example in (34) does not describe the house in objective
terms, but rather subjectively, and accordingly the case used is the instru-
mental.
In contrast to (39a), where the fate of the oldest child is the same as that
of the others, in (40) the elder is assuming responsibilities similar to those
of his parents; he is focalized in contrast to the other children. If we com-
Case variation of adjectival predicates 45
pare the types of consequences of being the oldest in the contexts in which
the instrumental and the nominative are respectively employed, the instru-
mental identifies the oldest child as simply one of the children, as part of
the group, whereas the nominative sets him apart. So it is despite being the
oldest that the narrator gets an education in (39b), whereas in (40) it is
because of being the oldest that he assumes responsibilities closer to those
of his parents: he must take care of the other children, wash floors, acquire
a profession, and work along with his parents.
In fact, if the speaker sets Karandash apart from other clowns, he will
use the nominative, but if the message is that Karandash was comparable to
other clowns despite being more cheerful, the speaker will use the instru-
mental.
As was shown earlier, in the case of permanent features the use of the in-
strumental case is possible only when the object itself ceases to exist, as in
(9) repeated below:
Some attributes are “observable”, so that the speaker can witness them,
and yet subject to change, such as bol'šoj ‘big’ or molodoj ‘young’. The
natural obstacle there is the nature of things or the “world order” (in the
sense of Cooper and Ross 1975): things and people progress from being
small to getting larger, from being young to growing old, and not in the
reverse direction. Consequently it is possible to find examples such as (43)
but not (44), except in some magic fiction:
And then there are adjectives that express qualities that are by defini-
tion available to and “knowable” only by the experiencer. Some such quali-
ties are purely interactive and inherently relative. A case in point is nesgo-
vorčivyj ‘intractable’. In order to be described as a person who does not
easily come to agreement with others, one has to be viewed through the
prism of those who perceive him as such. As a result, we do not find any
examples with the nominative, for example on Google <30.V.2006> byl
nesgovorčivyj, only with the instrumental case byl nesgovorčivym or byla
Case variation of adjectival predicates 49
Among the examples with vosxoždenie found on the Internet, there are a
large number with the instrumental (24 with trudnym, 4 with očen' trud-
nym, 3 with dolgim, 26 with složnym, 24 with nesložnym, and so on). There
is only one example with the nominative:
7. Conclusion
The choice between the nominative and the instrumental case of adjectival
predicates is governed by a number of factors. The ability of the speaker to
represent events as or as if observed triggers the nominative of the predi-
cate; however, if the speaker is a participant or if he focalizes the narration
on one of the participants other than the one described with the adjectival
predicate, the use of the nominative is blocked. In addition to focalization
away from the participant described, time comparison and time limitation
trigger the instrumental.
This is an area where grammar and lexicon interact. Adjectives repre-
senting permanent features are more likely to be used in the nominative;
the instrumental is likely to be used when the object or person (and thus the
feature) no longer exist. Properties expressed by adjectives used with ab-
stract nouns designating events cannot be observed at a particular moment
in time; adjectives that describe qualities that affect the participant are
more likely to be used in the instrumental.
52 Alina Israeli
Notes
References
Nichols, Johanna
1981 Predicate Nominals: A Partial Surface Syntax of Russian. Berkeley:
University of California Press.
1985 Padežnye varianty predikativnyx imen i ix otraženie v russkoj gram-
matike [Case variants of predicate names and their reflection in Rus-
sian grammar]. In Novoe v zarubežnoj lingvistike XV [New Research
in Foreign Linguistics XV], T. V. Bulygina and A. E. Kibrik (eds.),
342–387. Moscow: Progress.
Timberlake, Alan
1982 Invariance and the syntax of Russian aspect. In Tense-Aspect: Between
Semantics and Pragmatics, Paul J. Hopper (ed.), 305–331. Amster-
dam/ Philadelphia: John Benjamins.
1985 Invariantnost' i sintaksičeskie svojstva vida v russkom jazyke [Invari-
ance and syntactic properties of aspect in the Russian language]. In
Novoe v zarubežnoj lingvistike XV [New Research in Foreign Linguis-
tics XV], 261–285. Moscow: Progress.
2004 A Reference Grammar of Russian. Cambridge: Cambridge University
Press.
Zel'dovič, G. M.
2005 Russkoe predikativnoe imja [Russian predicative name]. Toruń:
Wydawnictwo Uniwersytetu Mikołaja Kopernika.
Why double marking in the Macedonian dativus
sympatheticus?
Liljana Mitkovska
Abstract
This paper discusses the possible reasons for and implications of marking
possession explicitly when it is already marked by dativus sympatheticus
(DS) in Macedonian. The term ‘dativus sympatheticus’ denotes a dative
construction that implies a possessive relationship between the dative refer-
ent and another participant in the situation. The DS/explicit possessive
variation illuminates the specific construal that each of the two structures
imposes on the possessive relation expressed. Since the dative object al-
ready implies possession, the introduction of an explicit possessive marker
in the DS construction has been deemed as superfluous and anomalous.1
However, such constructions have been attested frequently, both in Mace-
donian and crosslinguistically.2 It is argued here that the double marking of
possession is motivated pragmatically, i.e. by the need of the speaker to pre-
sent the possessive relationship from the perspective of the possessor (thus
the dative object is chosen), and at the same time to highlight the possessed
(which triggers the possessive pronoun). It will be demonstrated that the es-
tablishment of the construction is supported by structural and semantic fac-
tors: with certain verbs that require a dative object the pragmatic goal to put
the focus on the possessed results in double marking, as the omission of the
dative either produces an ungrammatical clause or affects the meaning con-
siderably. The application of such doubling to contexts that would equally
accept either DS or a possessive pronoun produces an effect that is clearly
different from the meanings of either of the structures taken separately.
1. Introduction
The starting point for the study of this phenomenon is the Cognitive
Linguistic premise that people are able to construe a situation in various
ways and that different grammatical structures are usually available to en-
code the same situation (Langacker 1991). In particular, two significant
theoretical assumptions underlie the present analysis.
First, it is presumed, in accord with Cognitive Linguistics (Lakoff 1987;
Langacker 1991 and others), that language units are symbolic representa-
tions of image schemas at various degrees of abstraction associated with
the usage events they categorize. Used in a variety of contexts, these sche-
mas must often be expanded and modified to accommodate the usage
events that do not fully match the conventional specification, which gives
rise to new conventional units that constitute an expansion from more basic
ones. As a result, grammatical categories are complex, containing a net-
work of nodes united by a prototype principle. Consequently, the indirect
object construction in Macedonian is viewed as a polysemous category
exhibiting a number of related meanings (Rudzka-Ostyn 1996). The most
central among the meanings are the recipient of concrete and abstract ob-
jects and the recipient of effects (experiencer). ‘Possession’ is also under-
stood as a complex notion, involving a broad array of relationships clus-
tered around the prototypes: ownership, part whole, and kinship
relationships (Langacker 1991 and 2000; Topolinjska 1997).
The second assumption concerns the theory of grammaticalization. Ac-
cording to some findings (Sweetser 1990; Heine et al. 1991; Hopper and
Traugott 1993) a contextually inferred meaning can be spread to other con-
texts by analogy. What starts as a contextual and conversational implica-
Why double marking in the Macedonian dativus sympatheticus? 57
ture can be conventionalized by recurrent usage and give rise to new mean-
ings through contextual manipulation and metaphor. The implication of
possessive relationships in an IO construction is no surprise since the da-
tive and genitive are conceptually related, the former expressing transfer of
an object and the latter the locus of the resultant state. Thus a reanalysis in
appropriate contexts can be imagined. In fact, the conceptual transfer from
dative to genitive has been attested in many languages (Heine et al. 1991:
167).
The paper is organized as follows. In the second section the so called
dativus sympatheticus is defined in comparison with explicit possessive NP
constructions. Then the syntactic, semantic and pragmatic conditions under
which double marking occurs are examined, followed by a short analysis of
analogous developments in reflexive situations. Finally a short conclusion
is presented.
The dative in many Slavic languages has been shown to imply posses-
sive meaning. There is no consensus among linguists, however, as to the
status of DS, i.e. whether it is, actually, a separate construction. Kučanda
(1996: 330), for example, argues against it: “The distinction between sev-
eral subsets of the free dative is quite unnecessary because there is no con-
clusive syntactic and/or semantic evidence to support such a division. The
interpretation of the free dative depends on the semantics of the verb and
the meanings such as advantage, disadvantage and possession are addi-
tional layers of semantic interpretation.” Other linguists, on the other hand,
support the view that DS should be regarded as an independent construc-
tion. King (1998), rejecting the contentions of some linguists that the da-
tive of possession in Czech (‘dative of interest’ in her terminology) is only
a contextual variant of the free dative, insists that it differs functionally
from all other dative constructions. When the speaker chooses to use this
construction the speaker knows exactly what s/he wants to do, namely to
present the situation from the point of view of the possessor of the in-
volved entity. Fried (1999) also holds the view (within the Construction
Grammar framework) that there is a separate construction, which, besides
sharing properties with the other dative constructions, has its own specific
configuration. “In particular, it must be made explicit that EPRs (external
possessors) represent interest overlaid with possession and that the co-
occurrence of these two semantic features is subject to specific grammati-
cal constraints, rather than being a matter of pragmatic plausibility.” (ibid.:
494)
60 Liljana Mitkovska
The analysis in this paper is based on the assumption that although for-
mally DS does not differ from other dative constructions it is associated
with a unique semantic-pragmatic construal, namely to underline the af-
fectedness of the possessor. In fact, this construal has been conventional-
ized, which is most evident in less prototypical examples with stative
verbs, where even though no real affectedness exists, the construction con-
strues the situation as affecting the possessor.7 Example (8) may be uttered
as a warning to the addressee that the situation described in the sentence
may be unpleasant for her. Example (9) suggests that the subject’s knowl-
edge of the hiding places of some people may be dangerous for those peo-
ple.8
that the event has a positive or negative effect on him or her. King (1998)
(among others) has argued convincingly that the distinct function of the
dative of possession is to express empathy with the possessor: “When the
speaker chooses the Dative over any other available construction, he ex-
presses a certain kind of knowledge. He shows that he is aware of a posses-
sive relationship between two referents, one of which is animate, and most
likely human. The speaker also knows that the referent, [A], is physically
or psychologically affected, either positively or negatively, by something
that happened to or with the other referent [B].” (ibid.: 250–251)
(2) Explicit possessive constructions, on the other hand, focus the atten-
tion on the possessed, hence the effect of detachment mentioned by Bally.
Examples (10) to (12) below illustrate this.
In example (10) the sentence pictures the legs of the woman abstracted
from the rest of her body. It resembles the image shown by a camera focus-
ing on them as she moves away. In (11) the first part of the sentence fo-
cuses on the man’s face and what happens to it. Notice how the attention
switches to the person in the second half with the dative. Explicit posses-
sive constructions are often used when the possessed is singled out by
some specific properties, as in example (10) or contrasted, as in (12).
In prototype situations the possessor is in reality also affected, but ex-
plicit possessive constructions provide information about the possessed and
what happens to it, rather than about the state of the possessor. Where the
verb does not express affectedness, this feature will not be present in the
adnominal possessive construction (as in example 10). In such situations
the affectedness can only be imposed by the dative, and the sentence with
DS may differ considerably in meaning from the one with the possessive,
as illustrated by the examples in (13). In (13a) DS ensures an interpretation
that the duke is aware and positively affected by the activity, while in (13b)
the activity is presented as independent of him. Notice that the latter may
be used in situations when the duke is not at all present, he may have been
abducted, while his boots were the only thing found and brought to the
castle.
(14) Ti si mi ja zemal
you.SG be.2SG.PRES IOCL.1SG DOCL.3SG.F take.PAST
mojata kniga.
my.DEF book.F.SG
‘You have taken my book.’
Many scholars dealing with the dative of possession (DS) have noticed
this phenomenon, often characterizing it as awkward or not quite accept-
able. Sometimes this possibility is regarded as proof that dative and ex-
plicit possessive do not have the same function (see for example Kučanda
1996: 326). In the same vein, Janda (1993: 86) claims that: “[t]he conten-
tion that the dative expresses affectedness via possession rather than pos-
session per se is upheld by the fact that there is no rule preventing the da-
tive from co-occurring with other possessives.” On the other hand,
Velazquez-Castillo (1999: 92) maintains that such co-occurrence is “com-
pletely inadmissible” for standard Spanish, although it is possible for some
varieties of Spanish (colloquial Paraguayan Spanish, for example). She
assumes that “there must be something in the meaning of the dative that
motivates this incompatibility” and concludes that “[s]ince the dative al-
ready subsumes the concept of possession, the possessive adjective is not
necessary to establish the possessive relation.”
64 Liljana Mitkovska
b. Mu go zede kajčeto.
IOCL.3SG.M DO.3M.DEF take.PAST boat.DEF.DO
‘He1 took his2 boat.’
‘Only one half of his body is above, the torso without legs.’
Next, there are contexts in which the dative argument is not obligatory,
but its omission may affect the meaning of the clause. In examples (20) and
(21), if the dative argument was dropped the elimination of the affected-
ness component would cause the situation to acquire a general interpreta-
tion which leaves the possessor out of the scene. The difference is also
obvious in the English translation. So in these sentences, in order to convey
the intended meaning and to emphasize the possessed, the speaker needs
both the dative and the possessive pronoun. Naturally, the dative itself will
be enough to imply possession, but that alone does not satisfy the speaker’s
wish to stress the possessed.
(20) (Jas navistina uživav igrajќi vo prodolženijata na 007 vo šeesettite, no)
‘(I really enjoyed playing in the 007 series in the sixties, but)
a. ќe vi se zabeležuva za
PART.FUT IOCL.2PL REFL.DO criticise.PRES for
vašeto odnesuvanje. (V)
your.PL.DEF behaviour
you will be criticized because of your behaviour.’
Finally, there are constructions in which the dative and the possessive
express the same factual situation, the difference being only in emphasis
(cf. examples (26a) and (27a)). Examples (26b) and (26c) and (27b) and
(27c) show that clauses with either the dative or the possessive are per-
fectly well formed and express the same facts. However, the speaker de-
cides to use both markers, thus keeping the possessor’s perspective with
the dative, while highlighting the possessed at the same time by adding the
possessive pronoun.
c. da ni go minira rabotenjeto?
part IOCL.1PL DOCL.3SG.M undermine.PRES work.DEF
by undermining the work to us?’
70 Liljana Mitkovska
(29) Nemoj da si mi gi
don’t.IMP PART be.2SG.PRES IOCL.1SG DOCL.3PL
oblekla moite čevli! (AC)
put.PARTC on my.DEF shoes.DO
‘Don’t you dare put on my shoes!’ (warning)
4. Reflexive situations
Situations involving parts of the body require neither the dative nor the
possessive pronoun for establishing the possessive relation. The sentence
in (35a) is a typical example, where the possessive relationship is clearly
conveyed without any marker –one cannot hold back somebody else’s
tears. In such contexts the introduction of the dative reflexive si is already
felt as emphatic, as in (35b). Still, we encounter examples such as the one
in (32c), where the presence of both the dative and the possessive intensi-
fies this feeling. It seems, however, that in the latter case the construal of
double marking as defined in the previous section dominates. Namely, they
stress the possessed, but also present the situation from the perspective of
the possessor as the affected entity.
74 Liljana Mitkovska
5. Conclusions
Notes
5. This is the most common term, but it is sometimes confused with the adnomi-
nal dative with possessive function, such as the one used in Bulgarian for all
types of possession (e.g. knigata mu – literally: ‘the book to him’ (‘his book’))
and in Macedonian for kinship relations only (e.g. brat mi – literally: ‘brother
to me’ (‘my brother’)) .
6. The term unergative is used here to indicate intransitive predicates whose va-
lence contains an agentive argument, but no direct internal argument in con-
trast to unaccusative predicates, which contain an internal argument, but do not
introduce any agentive argument.
7. Langacker (1991: 294) notes that “linguistic structure embodies conventional
imagery and thus imposes a certain construal on the situation it codes”.
8. The comments here refer to the given examples, in which the affectedness is
negative. It is true, though, that in other situations it could also be understood
as positive.
9. The same tendency has been noticed by King (1998: 164) for Czech, especially
in situations where the modifier is an unusual one.
10. The source is cited fully in the list below.
References
Bally, Charles
1996 The expression of concepts of the personal domain and indivisibility
in Indo-European languages. In The Grammar of Inalienability: a ty-
pological perspective on body-part terms and part-whole relation,
Hilary Chappel and William McGregor (eds.), 31–64. Berlin/New
York: Mouton de Gruyter.
Chappel, Hilary, and William McGregor (eds.)
1996 The Grammar of Inalienability: a typological perspective on body-
part terms and part-whole relation. Berlin/New York: Mouton de
Gruyter.
Fried, Miriam
1999 From Interest to Ownership: A Constructional View of External
Possessors. In External Possession, L. Doris Payne and Immanuel
Barshi (eds.), 473–504. Amsterdam/Philadelphia: John Benjamins
Publishing Company.
Heine, Bernd, Ulrike Claudi, and Friederike Hünnemeyer
1991 Grammaticalization: A Conceptual Framework. Chicago and Lon-
don: The University of Chicago Press.
Hopper, Paul, and Elizabeth Closs Traugott
1993 Grammaticalization. Cambridge University Press.
Why double marking in the Macedonian dativus sympatheticus? 77
Janda, Laura A.
1993 A Geography of Case Semantics: The Czech Dative and Russian
Instrumental. Berlin/New York: Mouton de Gruyter.
King, P. Katerina
1998 The Czech Dative of Interest: The Hierarchical Organization of Pos-
session in Discourse and Pragmatics. Ph.D. diss., Harvard University.
Kučanda, Dubravko
1996 What is the Dative of Possession?. Suvremena lingvistika 41/42:
309–318.
Lakoff, George
1987 Women Fire and Dangerous Things: What Categories Reveil about
the Mind. Chicago/London: The University of Chicago Press.
Langacker, Roland W.
1991 Foundations of Cognitive Grammar. Vol. 1: Theoretical prerequi-
sites. Stanford University Press.
2000 Grammar and Conceptualization. (Cognitive Linguistics Research 14.)
Berlin/New York: Mouton de Gruyter.
Levine S. James
1986 Remarks on the Pragmatics of the ‘Inalienable Dative’ in Russian. In
Case in Slavic, Richard D. Brecht and James S. Levin (eds.), 437–
451. Columbus, OH: Slavica.
Minova-Gurkova, Liljana
1982 Kratkite zamenski formi za indirekten objekt i posvojnite zamenski pri-
davki [The pronominal clitics for indirect object and the possessive
pronominal adjectives]. Literaturen zbor XXIX, Kn. 2: 108–110.
Mitkovska, Liljana
2000 On the possessive interpretation of the indirect object in Macedonian.
Linguistica Silesiana 21: 85–101.
Najčeska-Sidorovska, Marija
1973 Sintagmi so Dativus Sympatheticus i negovoto razgraničuvanje od drugite
značenja vo ruskiot, makedonskiot, srpskohrvatskiot i bugarskiot jazik
[Constructions with Dativus Sympatheticus and its differentiation from
other types of meaning of the dative in Russian, Macedonian, Serbo-
Croation and Bulgarian]. Makedonski jazik XXIV: 119–130.
Payne L. Doris, and Immanuel Barshi (eds.)
1999a External Possession. Amsterdam/Philadelphia: John Benjamins Pub-
lishing Company.
78 Liljana Mitkovska
Laura A. Janda
Abstract
1. Introduction*
Russian is famous for its aspectual distinction between Perfective (the se-
mantically marked value1, in this paper signaled by a superscript “p”) and
Imperfective (signaled by a superscript “i”). Russian aspect is obligatorily
expressed by all verb forms, and furthermore is formally marked by a sys-
tem of aspectual affixes (prefixes and suffixes). For example, ‘write’ can
be rendered in Russian as an Imperfective simplex verb pisat’i, or can be
prefixed to yield the corresponding perfective napisat’p. A few hundred
“Bi-aspectual” verbs can express aspect in the absence of these affixes,
thus constituting exceptions to the overall system of morphological mark-
84 Laura A. Janda
This section will describe the Russian aspectual system in its prototypical
instantiation, in other words, excluding discussion of Bi-aspectual verbs.
There is nothing unprecedented about discussing aspect in Russian while
ignoring the Bi-aspectuals; indeed Glovinskaja’s recent book (2001) on
Russian aspect makes no mention of Bi-aspectuals (either individually or
as a phenomenon), and aside from one article in 1998 by Čertkova and
Čang and another one in 1999 by Jászay, there have been little more than
perfunctory remarks on the topic since the 1960s. I will look first at the
morphology of Russian aspect and then give a brief overview of its seman-
tic expression.
A given verb in Russian is either entirely Perfective (semantically
marked) or entirely Imperfective (semantically unmarked) in all tenses and
forms. Both the semantic and the morphological expression of aspect are
obligatory for all forms of all verbs. Russian is thus unlike most other lan-
guages with Perfective and Imperfective aspect, for which aspect is ex-
pressed only in restricted contexts (for example only in the past tense), and
Imperfective is the marked value (as in French).2 In other words, Russian
verb forms are always either Perfective or Imperfective, regardless of tense
or other grammatical categories, and their aspectual status is also signaled
morphologically. As a consequence most (but not all) scholars consider
aspect to be a derivational, not an inflectional, phenomenon in Russian (cf.
Isačenko 1960; Dahl 1985; Zaliznjak and Šmelev 2001; see also discussion
in Janda forthcoming b).
The morphological expression of aspect is achieved primarily by means
of Perfectivizing prefixes (pro-, za-, s-, ot-, na-, po-, vy-, o-, raz-, ob-, u-)
and Imperfectivizing suffixes (-yva-/-iva-, -va-, -a-).3 There is also a Per-
fectivizing semelfactive suffix (-nu) which is more restricted in its use.
This morphology yields the following types of verbs: verbs with no as-
pectual affixes, which are typically Imperfective like pisat’i ‘writei’; verbs
with prefixes, which are typically Perfective like napisat’p ‘writep’,
perepisat’p ‘revisep’, popisat’p ‘writep (for a while)’; and verbs with suf-
fixes, which are typically Imperfective like perepisyvat’i ‘revisei’. There
are some deviations to this pattern, such as unaffixed Perfectives like dat’p
‘givep’ and prefixed Imperfectives like nasledovat’i ‘inheriti’, verbs with
multiple prefixes like poperepisyvat’p ‘revisep (for a while)’, and semelfac-
tive suffixed Perfectives like čixnut’p ‘sneezep (once)’, but generally there
are aspectually related verbs that morphologically disambiguate such cases,
such as davat’i ‘givei’, unasledovat’p ‘inheritp’, perepisyvat’i ‘revisei’, and
86 Laura A. Janda
čixat’i ‘sneezei’. There are also a few suppletive verbs that break these
rules, such as kupit’p ‘buyp’ vs. pokupat’i ‘buyi’. However, most Russian
verbs fit into the pattern of Imperfective simplex, to which one can add a
prefix to get a Perfective verb, to which it might be possible to add a suffix
to get an Imperfective verb. Overall this is a fairly robust and reliable sys-
tem (cf. Timberlake’s 2004: 401–7 “tripartite system”).
The meaning of Russian aspect is the subject of a vast literature which
cannot be adequately surveyed in this article (cf. Janda 2004). Suffice it to
say that this literature has primarily focused on describing aspect in terms
of semantic features, and since Perfective is taken to be the marked mem-
ber of the opposition, most features describe Perfective aspect, leaving
Imperfective as the default. The most common feature labels are “bound-
edness” (Forsyth 1970; Avilova 1976; Jakobson 1957/1971 and Padučeva
1996; cf. also “delimitation” in Bondarko 1971; “closure” in Timberlake
1982;, and “demarcatedness/dimensionality” in van Schooneveld 1978),
“totality” (Comrie 1976; Dickey 2000; Durst-Andersen 1992; Smith 1991;
Isačenko 1960 and Maslov 1965; cf. also “completion” in Vinogradov
1972), “definiteness” (Bondarko 1971 and Dickey 2000; cf. also “change”
and “sequencing” in Durst-Andersen 1992 and Galton 1976), “punctuality”
(Čertkova 1996; Mazon 1914), and “resultative” (Čertkova 1996 and Vi-
nogradov 1972). If the Imperfective aspect is assigned a feature at all, it is
most often “durativity” (Bondarko 1971; Padučeva 1996; Čertkova 1996).
What the tradition of feature analysis tells us is that a Perfective verb de-
scribes a single, unique event viewed in its entirety at a given point in time,
whereas the Imperfective describes all other events, especially those that
are extended in time. What this tradition doesn’t tell us is that the Perfec-
tive and Imperfective aspects have several dozen, often seemingly contra-
dictory uses that features are inadequate to account for. For example, the
general-factual and “polite” imperative uses require Imperfective verb
forms to describe unique whole events, whereas the Perfective is required
for habitually sequenced events, despite the fact that these events are re-
peated over long and indefinite periods of time. Although features do pin-
point the most significant facts about the semantics of aspect, they gloss
over the untidy realities of a very complex phenomenon.
Feature analysis has another relevant by-product, namely the notion of
the “aspectual pair”, consisting of one Perfective and one Imperfective
verb with the same denotation. The assumption that the “aspectual pair” is
the exclusive or dominant pattern observed in the Russian aspect system is
entrenched and pervasive in Russian linguistics (cf. Vinogradov 1938;
What makes Russian Bi-aspectual verbs special 87
This section will define borrowed and Bi-aspectual verbs, discuss the
status of borrowed and Bi-aspectual verbs in Russian, and outline the vari-
ous paradigmatic, semantic, syntactic, and morphological constraints on
Bi-aspectual verbs.
Borrowed verbs are verbs that cannot be traced to the common origins
of Slavic languages and have entered Russian since it commenced its de-
velopment largely independent of other Slavic languages. Although this is
a long period (reaching from the 12th century to the present), there was
very little activity in borrowing verbs until the Petrine period (late 17th
century), when contact with and knowledge of western European languages
became common among the Russian elite (Avilova 1968). All borrowed
verbs in Russian contain the suffix -ova-, which integrates them into the
inflectional system of Russian (without, however, providing any aspectual
designation). The suffix -ova- is itself Slavic, but is most often used with
various extensions in the presence of foreign stems, appearing as -izova-, -
irova-, -izirova-, -ficirova-. Although the -ova- suffix is indispensable –
without it a foreign word would lack inflection and also the ability to func-
tion as a verb in Russian – it is claimed that a desire to preserve the foreign
flavor of borrowed verbs caused them to resist further affixation (Mučnik
1966; Avilova 1968). This meant that foreign verbs would resist aspectual
prefixation and suffixation, supporting their recognition in Russian as Bi-
aspectual verbs (both Mučnik 1966 and Gladney 1982 compare this to the
importation of indeclinable foreign nouns such as kino ‘cinema’).
88 Laura A. Janda
Bi-aspectual verbs are verbs that can express both Perfective and Imper-
fective aspect with the same morphological form, without recourse to the
Perfectivizing and Imperfectivizing affixes described in section 1. Esti-
mates of the number of Bi-aspectual verbs vary. Mučnik (1966) Gladney
(1982) and Anderson (2002, based on combined listings in dictionaries)
suggest that there are approximately 600. Čertkova and Čang (1998) rec-
ognize 412 Bi-aspectual verbs, only 289 of which are “true Bi-aspectuals”
(the other 123 verbs are claimed to behave as both Bi-aspectuals and as
“paired” verbs). Wheeler 1972/1992 lists 348 verbs as Bi-aspectual. In-
deed, as Jászay (1999: 169–170) laments, dictionaries do not agree on the
identification of verbs as Bi-aspectual (cf. also Čertkova and Čang 1998:
24). Nor do scholars. Mučnik (1966: 69), for example, rejects Isačenko’s
(1960: 144) assignment of Bi-aspectuality to certain forms of or-
ganizovat’p/i ‘organizep/i’. And, as Jászay’s (1999) research shows, native
speakers also vary in their acceptance of Bi-aspectual forms, leading him to
coin the term “častičnaja dvuvidovost’” (“partial Bi-aspectuality”).
There is even less clear information on the status of borrowed verbs in
Russian and their relationship to Bi-aspectuality. Avilova (1968: 66) opens
her article with the statement “Принято считать, что глаголы с
заимствованной основой в русском языке являются двувидовыми” (“It
is commonly assumed that verbs with borrowed stems in Russian are Bi-
aspectual”), an assumption that she never challenges, although she repeat-
edly acknowledges the existence of borrowed verbs that are merely Imper-
fective (Avilova 1968: 66, 70, 71, 72, 73, 76). Čertkova and Čang (1998:
15) recognize 48 Imperfective verbs, nearly all of them foreign, among
verbs (erroneously in their view) listed as Bi-aspectuals in dictionaries. It is
common for scholars to note that some Bi-aspectual verbs are developing
prefixed Perfectives, which means that the borrowed simplex verb is Im-
perfective (Isačenko 1960; Mučnik 1966; Jászay 1999; Zaliznjak and Šme-
lev 2000). However, with the lone exception of Avilova (1968; and these
are merely parenthetic remarks), no one mentions the fact that some bor-
rowed verbs do not enter the Russian lexicon as Bi-aspectuals, and no one
has investigated the relationship between Bi-aspectual and Imperfective
verbs among borrowed verbs in Russian. Avilova (1968) furthermore
claims that the foreign verbs that do enter Russian as Imperfectives are
imperfectiva tantum (Imperfective isolates with no aspectually related
Perfective verbs listed in dictionaries).
Mučnik (1966: 64) challenges the prevailing assumption that foreign
Bi-aspectual verbs are predominantly scientific, technical, or professional
terms. He claims that only 35% of bi-aspectual verbs fall into those three
What makes Russian Bi-aspectual verbs special 89
groups, and that the rest belong to the common lexicon of the literary lan-
guage. Unfortunately there is no corresponding data on the status of non-
Bi-aspectual foreign verbs.
It is true that there is some correlation between Bi-aspectuality and for-
eign origin among verbs, but this fact has always been examined only by
looking at the frequency of foreign borrowings among Bi-aspectuals.
Čertkova and Čang (1998: 13) state that only about 10% of Bi-aspectual
verbs are part of Russian’s Slavic heritage. Anderson’s (2002) combined
list shows that 95% of Bi-aspectual verbs in Russian are foreign. Isačenko
(1960: 144) points out that in addition to native and foreign Bi-aspectual
verbs, there is some marginal tendency to build new Bi-aspectual verbs
using Russian stems and the -ova- suffix: zvukoficirovat’p/i ‘equip with
soundp/i’. But no one has ever asked how many foreign verbs are Bi-
aspectual. Nor has anyone pointed out the significant role of foreign verbs
in building parts of the Russian lexicon other than the Bi-aspectuals. For
example, foreign borrowings account for 20%4 of Imperfective only verbs
and are the sole productive source of such verbs in Russian, but there is no
mention of foreign verbs in Zaliznjak and Šmelev’s (2000: 85–6) discus-
sion of Imperfective isolates.
One thing that most scholars do agree on is that Bi-aspectual verbs are
not ambiguous in their expression of aspect. In other words, Bi-aspectual
verbs do not express a neutral aspect or a lack of aspect. Every use of a Bi-
aspectual verb like likvidirovat’p/i ‘liquidatep/i’ is either Perfective or Im-
perfective, as disambiguated by context (Isačenko 1960: 143–44; Mučnik
1966: 61; Avilova 1968: 66; Galton 1976: 294; Gladney 1982: 202;
Čertkova 1996: 100–109; Jászay 1999: 169; Zaliznjak and Šmelev 2000:
10), just as number is disambiguated when English fish is used in context.
The one significant exception is Timberlake (2004: 407–9) who suggests
that Bi-aspectual verbs are “anaspectual”, and do not express aspect. Due
to the prevailing assumption that Russian verbs come in “aspectual pairs”
(see section 2), most scholars consider Bi-aspectual verbs to be syncretic
examples of “pairs” where the two verbs in each “pair” are homophonous.
This means that Bi-aspectual verbs are deviant in their formal morphologi-
cal marking of aspect, but not in their semantic expression of this linguistic
category.
Bi-aspectual verbs (both foreign and native) display a range of some-
what unusual phenomena, most of which can be stated as constraints on
their paradigmatic, syntactic, and semantic expression of aspect. Although
most forms in the paradigm of Bi-aspectual verbs can express both Perfec-
tive and Imperfective, some forms are of necessity monoaspectual, namely
90 Laura A. Janda
This section will present an alternative to the “aspectual pair” model that is
based on metaphorical motivations for aspectual behavior in Russian. This
alternative model, the cluster model, highlights the differences between Bi-
aspectual and Imperfective simplex verbs on the basis of both semantic
content and morphological behavior.
Given the predominance of the “aspectual pair” model in general (cited
in section 2), as well as the fact that Bi-aspectual verbs are assumed to be
syncretic “aspectual pairs” (cited in section 3), it is necessary to examine
the concept of the “aspectual pair” in some detail. On the basis of a large,
stratified sample representing the full range of the Russian verbal lexicon
(Janda forthcoming a), I have advanced the cluster model of Russian verbal
aspect. This model acknowledges the existence of the aspectual partner-
ships traditionally labeled “pairs”, but at the same time recognizes that
these partnerships are usually embedded in larger clusters of aspectually
related verbs. On the motivation of three metaphors, the Russian aspectual
system differentiates Imperfectives from not one, but four distinct types of
Perfective verbs in Russian, each of which has a specific semantic and
morphological profile. I will look first at the three metaphors and then at
the four types of Perfective verbs.
The three metaphors that govern the Russian aspect system compare the
temporal contours of events to the behavior of concrete objects in three
source domains involving physical matter, motion, and granularity. The
second of these metaphors, the one comparing events to motion, will prove
most important in the analysis of borrowed and Bi-aspectual verbs. The
metaphors (in keeping with traditions of cognitive linguistics, cf. Lakoff
1987) will be stated in capital letters:
1) A PERFECTIVE EVENT IS A SOLID OBJECT, AN IMPERFECTIVE EVENT IS
A FLUID SUBSTANCE. This metaphor distinguishes Perfective from Imper-
What makes Russian Bi-aspectual verbs special 93
rowed verbs in the original study, this correlation can only be stated as a
hypothesis, and this hypothesis will be tested empirically in section 5.
0.5% of the database. These three verbs were each treated as if they were a
single entry, despite multiple meanings. There was one verb (or group of
verbs) that had to be excluded from the study altogether due to the fact that
the differentiation between reflexive and non-reflexive uses as well as dif-
ferent meanings associated with different stress made it impossible to clas-
sify it as either Bi-aspectual or non-Bi-aspectual in such a way that it could
be distinguished by search engines: pikirovat’/sja. This verb is Bi-
aspectual when it is not reflexive, and has two meanings: ‘dive, swoop’
(aeronautical) when stressed on the second syllable; ‘thin out’ (agricul-
tural) when stressed on the last syllable. When reflexive, this verb is Im-
perfective only and means ‘exchange caustic remarks, cross swords’. Dis-
ambiguation of these uses proved to be impossible given the remaining
design of the study. There was also one unprefixed Perfective foreign verb,
atakovat’p ‘attackp’, which was not included in the study because it could
not be classified as either Bi-aspectual or non-Bi-aspectual.
555 relevant foreign verbs were culled from Wheeler 1972/1992, 349
(63%) of which were designated by that source as Bi-aspectual, and 206
(37%) of which were designated as Imperfective. These numbers alone
make a compelling case for comparing borrowed Bi-aspectual and non-Bi-
aspectual verbs, since if nearly 40% of the borrowed verbs are non-Bi-
aspectual, then one cannot assume that all (or nearly all) borrowed verbs
are Bi-aspectual. At this point I have established that Russian has at least
two types of borrowed verbs and they are attested in roughly similar num-
bers. This fact begs the question of why some borrowed verbs are Bi-
aspectual whereas others are not. It should also be noted that these figures
probably underreport the rate of non-Bi-aspectual verbs, since scholars
(cited in section 3) frequently comment that dictionaries list verbs that are
no longer Bi-aspectual as Bi-aspectual. It is likely therefore that some of
the items classified in our list as Bi-aspectual are actually Imperfective.
The hypothesis from section 4 is that one should expect a strong ten-
dency for the Imperfective borrowed verbs to form Complex Act Perfec-
tives, whereas these Perfectives should be rare or non-existent for the Bi-
aspectual borrowed verbs. To test this hypothesis I needed to document the
use of Complex Act Perfectives for all verbs in the study. Because the most
common prefix found with Complex Act Perfectives is po-, it was decided
that these forms would be searched (in Janda forthcoming a, only one verb
was found that formed a Complex Act Perfective with a prefix other than
po- in the absence of any po- form). Searching for po- forms of course
carried with it the liability that in addition to Complex Act Perfectives I
would collect po- Natural Perfectives. Descriptions in the literature indi-
98 Laura A. Janda
cate that the vast majority of po- forms associated with borrowed Bi-
Aspectuals would not be Natural Perfectives. For example, Avilova (1968:
67) lists po- among the prefixes used least frequently in prefixed Natural
Perfectives of borrowed Bi-aspectual verbs (she ranks po- eighth in a list of
ten prefixes used by these verbs). Overall use of po- to form Natural Per-
fectives in Russian is very high (cf. Dickey 2006: 3; Čertkova 1996), so it
was expected that confounding data would be located particularly among
the borrowed Imperfectives. To control this, examples of uses of po- forms
were manually checked and verified for the presence of delimitative mean-
ing (signaled by adverbials such as nekotoroe vremja ‘for a while’). When
po- forms were found, it was nearly always possible to verify the meaning
in at least some of the “hits” as delimitative.
Searches of the verbs were performed on www.yandex.ru in June 2006,
and data was collected on the number of hits turned up for each verb both
without a po- prefix and with a po- prefix. Table 1 gives basic data on the
numbers gathered in this study. Of course data collected on a search engine
is notorious for being unstable and unreliable. However, given the low
frequencies of many of the verbs involved, it was impossible to collect
sufficient data for all but a handful of these verbs from corpora, so a search
engine was the only option.
The most striking observation to be made in Table 1 is that the average
number of hits for po- prefixed verbs is more than five times higher for the
Imperfective borrowed verbs than that for the Bi-aspectual borrowed verbs.
This can be restated as the proportion of po- prefixed forms, which is
2.52% for Bi-aspectual verbs and 12.67% for Imperfective verbs. A logistic
regression model using Pearson’s statistic to adjust for hit rate heterogene-
ity among verbs – this is needed as a different number of examples was
collected for each verb and the probability that a verb will be prefixed with
po- differs even for verbs that belong to the same class – reveals that there
is a demonstrable positive correlation between Imperfective borrowed
verbs and the presence of po- prefixed forms [χ2= 107.37, df=1, p<.0001].
The two types of verbs give very different data, with the Bi-aspectual
verbs yielding obvious differences at the low end of the scale and the Im-
perfectives yielding interesting numbers at the high end of the scale.
Whereas only 10% of Imperfective verbs yield zero po- forms, the absence
of po- forms is more than three and a half times as frequent (36%) for Bi-
aspectual verbs. The presence of small numbers (≤ 25) of po- forms is very
common for Bi-aspectual verbs (44%), but thereafter it drops precipitously
and high numbers of po- forms are vanishingly rare. 35% of Imperfective
verbs also yielded low numbers of po- forms (≤ 25), but this may have
What makes Russian Bi-aspectual verbs special 99
been due in part to the presence of many verbs with very low overall fre-
quency in this group (an effect corrected for in the discussion of Table 2
below).
At the other end of the scale, verbs listed as Imperfective were ten times
more likely (20%) to provide over 500 “hits” for po- forms than Bi-
aspectual verbs (with only 2% of verbs having over 500 po- “hits”). 80% of
Bi-aspectual verbs have 0–25 po- form attestations, but 54% of Imperfec-
tive verbs have 26 or more po- form attestations. At the low end of the
scale, cutting the scale of po- attestations between 0-to-5 and 6-and-up,
there is an even more revealing result. 63% of Bi-aspectual verbs have
between zero and 5 po- form attestations, whereas 67% of Imperfective
verbs have 6 or more po- form attestations. The two types of verbs, bor-
rowed Bi-aspectuals and borrowed Imperfectives, thus behave as mirror
images of each other in terms of their tendency to create po- prefixed
Complex Act Perfectives.
As mentioned above, there are some very low frequency verbs in the
data, particularly as pertains to Imperfective verbs that have only a few po-
attestations. For example, brasovat’i ‘bracei’ yields only 38 unprefixed
“hits” and 5 po- prefixed “hits”. In general, it is hard to say how much
small numbers (or small differences in numbers) of “hits” might mean on a
search engine, since they can result from typos or reduplicated returns. And
it is hard to say how significant effects are when dealing with low-
frequency verbs. In order to correct for these problems, the focus was nar-
rowed to verbs that are robustly attested by the search engine, namely
verbs with over 1000 “hits”. All of these verbs are well-established in Rus-
sian and should be familiar to any native speaker. To compensate for the
100 Laura A. Janda
Table 2(a). High-frequency Bi-aspectual verbs with either zero or over 1000 po-
prefixed hits
Bi-aspectual verbs with over 1000 unprefixed hits
Zero po- prefixed hits: 70 Over 1000 po-
prefixed hits: 3
abonirovat’/sja ‘subscribe’ inkrustirovat’ ‘encrust’ massirovat’ ‘amass;
massage’
angažirovat’ ‘book, engage’ intensificirovat’ rekomendovat’/sja
‘intensify’ ‘recommend’
anglizirovat’ ‘anglicize’ internacionalizirovat’ remontirovat’
‘internationalize’ ‘renovate’
assignovat’ ‘allocate, budget’ internirovat’ ‘place in
internment’
debetovat’ ‘debit, charge’ kanonizirovat’ ‘canonize’
deblokirovat’ ‘unblock’ kanonizovat’ ‘canonize’
decentralizovat’ ‘decentralize’ kapitalizirovat’ ‘reinvest’
dekretirovat’ ‘decree’ kollektivizirovat’
‘collectivize’
demaskirovat’ ‘unmask’ kolonizovat’ ‘colonize’
demilitarizovat’ ‘demilitarize’ kristallizovat’sja
‘crystallize’
demobilizovat’ ‘demobilize’ kvalificirovat’ ‘qualify,
test’
denacionalizirovat’ ‘privatize’ latinizirovat’ ‘latinize’
denaturirovat’ ‘denature’ litografirovat’
‘lithograph’
rowed verb like flirtovat’i ‘flirti’ can form a Complex Act Perfective; pof-
lirtovat’p ‘flirtp (for a while)’ exists and is well-attested.
Table 2(b). High-frequency Imperfective verbs with with either zero or over 1000
po- prefixed hits
Imperfective verbs with over 1000 unprefixed hits
Zero po- prefixed hits: 5 Over 1000 po- prefixed hits: 26
fetišizirovat’ ‘fetishize’ aplodirovat’ ‘applaud’ manipulirovat’
‘manipulate’
fraxtovat’ ‘freight’ buksovat’ ‘skid’ marinovat’ ‘marinate’
legirovat’ ‘alloy’ èksperimentirovat’ masturbirovat’
‘experiment’ ‘masturbate’
vualirovat’ ‘veil’ èkspluatirovat’ ‘exploit’ praktikovat’/sja
‘practice’
èrodirovat’ ‘erode’ fantazirovat’ ‘fantasize’ psixovat’ ‘act flaky’
fextovat’ ‘fence’ redaktirovat’ ‘edit’
flirtovat’ ‘flirt’ regulirovat’
‘regulate’
improvizirovat’ šantažirovat’
‘improvise’ ‘blackmail’
interesovat’/sja šinkovat’ ‘chop’
‘interest’
intrigovat’ ‘scheme’ spekulirovat’
‘speculate, gamble’
ironizirovat’ ‘mock’ tancevat’ ‘dance’
komandovat’ ‘be in trenirovat’ ‘train’
charge’
kritikovat’ ‘criticize’ žonglirovat’ ‘juggle’
It would not be possible to discuss all 555 borrowed verbs in this sur-
vey, but Table 2 provides us with an opportunity to look more closely at a
more manageable (and probably more representative, due to the fact that all
these items are high-frequency, familiar verbs) fraction of the data. The
high-frequency Bi-aspectual verbs in the left column do not form Complex
Act Perfectives. It is fairly easy to see how this is motivated by the mean-
ings of these verbs. These Bi-aspectual verbs for the most part describe
actions that one cannot engage in without having an effect and/or making
progress toward a goal. One can’t just do these things for a while and stop
and have no result. These activities are strongly telic. For example, one
can’t spend some time guillotining (gil’otinirovat’p/i) or immunizing (im-
munizirovat’p/i) without having guillotined or immunized someone. There
What makes Russian Bi-aspectual verbs special 103
verify all attestations, but it appears that the po- forms of massirovat’? are
all (or nearly all) based on the ‘massage’ meaning, which is probably not
Bi-aspectual (Ožegov 1949/1989 notes that the Bi-aspectual use is obso-
lete). Čertkova and Čang specifically cite rekomendovat’/sjai ‘recommendi’
as an example of a verb that is erroneously entered in dictionaries as Bi-
aspectual despite the fact that it is really an Imperfective. Remontirovat’i
‘renovatei’ is likewise listed by Ožegov1949/1989 as an Imperfective.
Among the Imperfective verbs in Table 2, two are listed as Bi-
aspectual by Ožegov 1949/1989: legirovat’p/i ‘alloy p/i’, which yielded no
po- forms, and improvizirovat’p/i ‘improvizep/i’, which yielded over 1000
po- forms. The Imperfective verbs with over 1000 po- forms behave much
like typical native Russian simplex Imperfectives. In addition to readily
forming Complex Act Perfectives, a number of them form other types of
Perfectives in various combinations:
Natural Perfectives: sfantazirovat’p ‘fantasizep’, skomandovat’p ‘give a
commandp’, zamarinovat’p ‘marinatep’, otredaktirovat’p ‘editp’, ureguli-
rovat’p (also ot-, za-) ‘regulatep’, stancevat’p ‘dancep’, natrenirovat’p
‘trainp’
Specialized Perfectives: podregulirovat’p ‘adjust, resetp’, raskritikovat’p
‘tear to piecesp’, vytancevat’p ‘obtain by dancingp’
Single Act Perfectives: psixanut’p ‘crack upp (once)’ (from psixovat’i
‘act flakyi’), spekul’nut’p ‘gamblep (once)’ (from spekulirovat’i ‘gamblei’)
Semantically the typical Imperfective verbs describe human behaviors
associated with various social and professional settings. Most of these
verbs can be used intransitively, and typically these are verbs that can be
used in expressions like on po-X-oval nemnogo i brosil ‘he X-ed for a
while and then stopped’, implying no result. The verbs in this group do not
fit into the subtypes listed for typical Bi-Aspectual verbs above. On the
contrary, these verbs are easily construable as Non-Completable.
The trends in this empirical study are strong and compelling enough to
outweigh whatever shortcomings there may be in the data. Clearly the Im-
perfective and Bi-aspectual borrowed verbs of Russian are very different
groups of verbs, both objectively (quantified in attestations of Complex
Act po- Perfectives), and subjectively (given the semantic groups of verbs
observed). Brief conclusions follow in section 6.
What makes Russian Bi-aspectual verbs special 105
6. Conclusions
Notes
* The author would like to thank John Korba for collecting data and Chris
Wiesen for assistance with statistical analysis. Thanks are also due to the re-
viewers for this volume, to Tore Nesset who commented on an earlier version
106 Laura A. Janda
of this article, and to Alexander Berdichevsky who spotted an error in the data.
All remaining imperfections are to be attributed to the author.
1. Although nearly all scholars agree that Perfective is semantically marked in
Russian, Galton 1976 gives the opposite assignment and Padučeva 1996 con-
siders Perfective vs. Imperfective to be an equipollent rather than asymmetri-
cal relationship.
2. Dahl (1985: 71–72) and Smith (1991: 277) point out this typological correla-
tion concerning the markedness values of Perfective and Imperfective.
3. This list simplifies the picture a bit, citing only the affixes relevant to Bi-
aspectual and foreign verbs. Švedova et. al. 1980 lists 17 Perfectivizing
prefixes, but only the 11 given here are attested with Bi-aspectual verbs by
Avilova (1968) and Čertkova and Čang (1998), and they are given in descend-
ing order of frequency. These prefixes are those common to both Bi-aspectual
and non-Bi-aspectual verbs in Russian.
4. 85 of the 428 verbs listed as Imperfective only in Wheeler 1972/1992 have
foreign origins. This figure includes 29 verbs other than those in the study de-
scribed in section 4, since it includes items with suffixes other than -ovat’,
such as špionit’i ‘be a spyi’ (which is not strictly a borrowed verb, but is built
from a borrowed noun).
References
Anderson, Cori
2002 Biaspectual Verbs in Russian and their Implications on the Category
of Aspect. Honors Thesis, University of North Carolina-Chapel Hill.
Avilova, Natal’ja S.
1968 Dvuvidovye glagoly s zaimstvovannoj osnovoj v russkom literatur-
nom jazyke novogo vremeni [Bi-aspectual verbs with foreign stems
in the modern Russian literary language]. Voprosy jazykoznanija:
66–78.
1976 Vid glagola i semantika glagol’nogo slova [Verbal aspect and verbal
semantics]. Moscow: Akademija nauk SSSR.
Bertinetto, Pier Marco, and Denis Delfitto
2000 Aspect vs. actionality: Why they should be kept apart. In Tense and
Aspect in the Languages of Europe, Östen Dahl (ed.), 189–225. Ber-
lin: Mouton de Gruyter.
Bondarko, Aleksandr V.
1971 Vid i vremja russkogo glagola [Russian verbal aspect and tense].
Moscow: Prosveščenie.
What makes Russian Bi-aspectual verbs special 107
Jakobson, Roman O.
1957/1971 Shifters, verbal categories, and the Russian verb. In Selected
Writings II, 130–147. The Hague: Mouton.
Janda, Laura A.
2004 A metaphor in search of a source domain: the categories of Slavic
aspect. Cognitive Linguistics (15): 471–527.
forthc. a Aspectual clusters of Russian verbs. Studies in Language.
forthc. b Inflectional Morphology. In Handbook of Cognitive Linguistics, Dirk
Geeraerts and Hubert Cuyckens (eds.). Oxford: Oxford University
Press.
Jászay, László
1999 Vidovye korreljaty pri dvuvidovyx glagolax [The aspectual
correlates of bi-aspectual verbs]. Studia Russica (17): 169–177.
Lakoff, George.
1987 Women, fire, and dangerous things. Chicago: University of Chicago
Press.
Maslov, Jurij S.
1965 Sistema osnovnyx ponjatij i terminov slavjanskoj aspektologii [The
system of basic concepts and terms in Slavic aspectology]. Voprosy
obščego jazykoznanija: 53–80.
Mazon, André
1914 Emplois des aspects du verbe russe. Paris: Librairie ancienne Honoré
Champion.
Mehlig, Hans Robert
1994 Gomogennost’ i geterogennost’ v prostranstve i vremeni
[Homogeneity and heterogeneity in space and time]. Revue des
etudes Slaves (66): 595–606.
1997 Nekotorye zamečanija po povodu opisanija kategorii vida v russkom
jazyke [Some notes on the description of the category of aspect in
Russian]. Russian Linguistics (21): 177–193.
Mučnik, I. P.
1966 Razvitie sistemy dvuvidovyx glagolov v sovremennom russkom
jazyke [The development of the system of bi-aspectual verbs in
modern Russian]. Voprosy jazykoznanija: 61–75.
Ožegov, Sergej I.
1989 Reprint. Slovar’ russkogo jazyka [Dictionary of Russian]. Moscow:
Russkij jazyk. Original edition, 1949.
Padučeva, Elena Viktorovna
1996 Semantičeskie issledovanija [Semantic investigations]. Moscow:
Jazyki russkoj kul’tury.
What makes Russian Bi-aspectual verbs special 109
Abstract
This paper attempts to re-examine the nature of the Croatian present tense in
the light of the cognitive linguistic view of grammatical structure. In order to
account for what Croatian traditional grammars describe as the usage of the
present tense for the so-called ‘non-real present’, we will use the naive char-
acterization approach suggested by Langacker (1991: 250). We will con-
sider the following issues: tense as an epistemic category, the nature of per-
fective and imperfective processes (cf. e.g. Langacker 2001b) and the
relevance of this classification of process types for aspectual phenomena in
Croatian, as well as the role of extreme subjectification (Langacker 2003:
13) in the characterization of non-temporal uses of the Croatian present
tense. Rather than hypothesizing distinct meanings for the present tense
morpheme, we recognize the present tense morpheme as a consistent marker
of proximity, indicating immediacy of the process in relation to the speaker.
In doing so, we will discuss the high degree of subjectivity present in cases
where the speaker describes ‘virtual processes’ that occur as a result of her
conceptualizing activity. Finally, we will demonstrate how general cognitive
models proposed for examining the grammatical category of tense can yield
situation-specific models that incorporate process types, as well as aspects
of viewing arrangement. Having taken into consideration the nature of proc-
ess types, the epistemic value of the present tense marker and specific depar-
tures from the canonical viewing arrangement (Langacker, 2003:18), we ar-
gue that non-temporal usages of the Croatian imperfective and perfective
present are as ‘real’1 as their temporal usages, or more specifically, that, just
like in English, their non-temporal or non-present uses can be accounted for
in terms of the inherent value of indicating present time (cf. Langacker
2001b: 267).2
1. Introduction
jede-Ø kokic-e.
eat.IMPF.3SG.PRES popcorn.ACC
‘Whenever she goes to the cinema, Iris eats popcorn.’
Both the imperfective and the perfective present are possible in hypo-
thetical situations located at different points on the realis–irrealis contin-
uum, as illustrated by (3a, b) below. In an example such as (3a) all four
combinations are acceptable: (i) the perfective present tense in both the
main and the subordinate clause, (ii) the imperfective present tense in both
clauses, (iii) the perfective present tense in the subordinate and the imper-
fective present tense in the main clause, and finally (iv) the reverse. All
four examples will be discussed in greater detail in section 3.
b. Da uči-m/nauči-m,
if study.IMPF/PERF.1SG.PRES
prošao bih7 na ispit-u.
pass.PERF.1SG.M.COND on exam.LOC
‘If I studied, I would pass the exam.’
The perfective present can be used to express future time but only when
it is preceded by a modal adverb:
The above examples reveal that the Croatian present is indeed used and
favoured in a number of contexts which are traditionally not identified as
expressing ‘real’ present time. However, we would like to propose that the
cognitive, non-objectivist, view of semantic structure provides a frame-
work within which we can attempt to reanalyze the instances of the so-
called ‘non-real’ Croatian present as designating present “conceptual oc-
currences” (Langacker 2003: 15, original emphasis), thereby defending the
idea that the Croatian present is consistently used to locate the process at
the time of speaking.9 In order to avoid terminological confusion, we
should mention that the term ‘non-real present’ refers to ‘non-temporal’
uses of the present tense.
In the following section, we will present some relevant features of the
Croatian tense and aspect, relying on the descriptions offered by Croatian
grammarians. Then, in section 2, we will give a short description of Lan-
gacker’s epistemic models, the notion of epistemic grounding, the classifi-
cation of process types, and the notions of canonical viewing arrangement
and subjectification, that is the fundamental models and ‘tools’ within the
theory of cognitive grammar necessary for a fresh analysis of the Croatian
present. Finally, in section 3, we will use the examples from the introduc-
tion to suggest a new characterization of the Croatian present tense guided
by relevant cognitive linguistic assumptions.
Perfectives, imperfectives and the Croatian present tense 117
1.2. General remarks on Croatian tense and aspect – the traditional view
Croatian verbs are marked for person, number, gender, tense, aspect, mood
and voice. Since the categories of tense and aspect are crucial for our
analysis, we will give a brief overview of the traditional definition of these
two categories in Croatian grammars.
1.2.1. Tense
Croatian has three synthetic tenses, i.e. one present tense and two past
tenses (imperfekt and aorist), and four analytic ones, i.e. the perfect, plu-
perfect, future and perfective future. Tenses operate in a framework that
suggests a threefold division of time into past, present and future. Since the
Croatian term “vrijeme” is used to denote both time and tense, it is difficult
to tell the two apart in traditional grammatical descriptions. Katičić simply
states that the category of “vrijeme” can be grammatically marked as pre-
sent, past and future (Katičić 1991: 45), and makes a further opposition
between so-called completeness and incompleteness, a distinction that
differs from the aspectual distinction. Thus, he suggests the following sys-
tem: present, past, future; and complete present, complete past, complete
future.10
1.2.2. Aspect
(a)
Irreality
(Known) reality Immediate
reality
(b) (c)
Non-Reality
Unknown reality
Known reality
Immediate
reality
C G
Figure 1. The basic epistemic model, the elaborated epistemic model and the time
line model (taken from Langacker 1991: 242–244)
Within the basic epistemic model, the English tense morphology is de-
scribed in terms of the notion of proximity. Thus, instead of “present” vs.
“past” we speak of a “p r o x i m a l / d i s t a l contrast in the epistemic sphere”
(Langacker 1991:245). The conceptual import of the zero tense morpheme
is i m m e d i a c y and the import of its marked counterpart is n o n -
i m m e d i a c y . Thus, the proximal morpheme contributes to the process
being construed as immediate to the speaker, whereas the distal morpheme
always conveys non-immediacy.
The proposed models offer three notions that will prove to be instru-
mental in the description of the Croatian present tense: ground(ing) and the
conceptualizer’s vantage point, the notion of proximity in the epistemic
sphere and the notion of reality. In the course of the work these notions
will be used and contrasted in terms of their relation to other relevant no-
tions: ground and its relation to the profiled process (Langacker 1990,
2003) in section 2.5., the notion of actual reality and its relation to its vir-
tual counterpart in section 3.2., and proximity as a prototypical value of the
Perfectives, imperfectives and the Croatian present tense 121
The semantic content of the verb plivati ‘swim’ can be further con-
strued in different ways depending on the auxiliary elements that can be
‘built into it’, such as aspectual markers. As far as aspectual marking is
concerned, it is needless to stress that the English and Croatian aspectual
122 Renata Geld and Irena Zovko Dinković
Langacker (1987: 258–261, 2001b: 255–258, 2003: 17) argues that the
aspectual distinction for verbs is easily compared to the count/mass distinc-
tion for nouns. Let us imagine a stone painted against a plain white back-
ground. It is bounded within our immediate spatial scope (Figure 2a). The
basic difference between count and mass nouns is that with a count noun
the boundaries fall within the conceptualizer’s immediate scope, whereas a
mass noun profiles a thing whose substance leaves the boundaries out of
the immediate scope (Figure 2b). Furthermore, a mass is internally con-
tractible/expandable in the sense that a portion of a mass of whatever size
constitutes a valid instance of this mass. Thus a handful of sand remains an
instance of sand. Masses are construed as internally homogeneous and they
cannot be replicated. Count nouns, on the other hand, exhibit at least a
minimal degree of internal heterogeneity and they profile things which
cannot be further divided and continue to represent a valid instance of the
category.
Langacker’s comparison of the count and mass distinction with the as-
pectual characteristics of verbs is straightforward: a perfective verb pro-
files a process construed as being bounded within the scope of predication,
whereas an imperfective verb profiles a process that can completely “oc-
cupy the scope of predication and even overflow its boundaries” (Lan-
gacker 1987: 261).15 Furthermore, imperfective processes represent the
perception of constancy through time and all the component states of the
imperfective process are identical. Thus, imperfective processes exhibit
homogeneity, contractibility and non-replicability (Langacker 1991: 21).
On the other hand, perfective processes always involve a change through
time and it is this change that contributes to their prototypical character.16
Perfectives, imperfectives and the Croatian present tense 123
IS IS
space space
stone sand
There are three classes of Croatian verbs that are traditionally classified
exclusively as imperfectives: these are the so-called stative verbs, rela-
tional verbs and evolutive verbs (Silić 1978: 60–62). Stative verbs are said
to denote static duration, like spavati ‘sleep’ or nadati se ‘hope’, whereas
relational verbs denote relationships, e.g. sadržavati ‘contain’, znati
‘know’, imati ‘have’. Evolutive verbs differ in that they involve a quantita-
tive or qualitative change (Silić 1978: 61). Typical evolutive verbs would
be šetati ‘walk’, pušiti ‘smoke’, živjeti ‘live’, and so on. These three groups
are said to occur only in the imperfective form because their aspect cannot
be changed morphologically without a change in meaning (e.g. znati
‘know’ when prefixed changes into doznati ‘find out’) (Silić 1978: 61).
Generally, it seems that the central notion in the above described classi-
fication of Croatian imperfectives is the notion of homogeneity that is ob-
servable if the process is viewed from a considerable distance.17 It over-
rides the presence of change that we believe is inherent18 in a great number
of imperfective verbs such as plakati ‘cry’, šetati ‘walk’ or spavati ‘sleep’.
In terms of process types, these processes are internally homogenous but
they occur in bounded episodes and thus involve change with regard to
their initiation and termination (Langacker 1990a). This same “inherent
perfectivity” (Langacker 1987: 259) is fundamental in classifying such
verbs in English as perfectives (Figure 3a). This classification is related to
the fact that the English grammatical inventory contains the imperfectiviz-
ing progressive construction that neutralizes this change. The endpoints of
a bounded process are excluded from the immediate scope, which conse-
quently encompasses only a homogeneous internal portion of the process.
It is only this portion that is profiled (cf. Figure 3b). The same ‘internal
124 Renata Geld and Irena Zovko Dinković
view’ (Langacker 2001b: 259) of the event is taken in the following Croa-
tian example:
MS/IS MS MS/IS
IS
t t t
It seems that the verb spavati can easily instantiate both aspectual cate-
gories. In (10) it profiles an imperfective process; that is a stable situation
within the immediate temporal scope (Figure 4a). The internal view on the
situation is reinforced by the imperative in the preceding clause. It indi-
cates that we are changing our perspective and narrowing down our view to
a temporary situation coinciding with the speech event. Sleeping normally
occurs in bounded episodes, and in this particular case the boundaries of
one of those episodes are cancelled and the immediate temporal scope co-
incides with the speech event. However, it is important to stress that such
construal does not depend on the imperative used in the preceding clause.
It is the interplay between the aspect of the verb and the present tense mor-
pheme that allows for the construal of a stable situation within the immedi-
ate temporal scope.
Although the verb spavati ‘sleep’ does not imply internal variation, it
designates a process that implies limits in terms of its beginning and its
end. Thus, if we use the same imperfective present of the verb spavati with
an adverbial of frequency we construe each event as an episode coinciding
with the immediate temporal scope, as in example (11):
Perfectives, imperfectives and the Croatian present tense 125
Such inherent perfectivity that enables us to view the event in its totality
is present in a large number of Croatian imperfectives and is evident in
their iterative usage. Therefore, we will call them non-canonical imperfec-
tives. An analogous example of a non-canonical imperfective is given in
(12):
MS MS MS
IS IS IS
t t t
Both English and Croatian perfectives are largely unacceptable in the pre-
sent tense with the present-time meaning because the conceptual configura-
tion their felicitous use requires is simply difficult to achieve. There are
two problems with the profiled perfective event coinciding with the speech
event: the durational problem and the epistemic one (Langacker 2001b:
263). The durational problem was already presented in example (14): the
present tense morpheme should signal that the profiled perfective process
coincides with the speech event. The event of writing, however, is longer
than the speech event describing it. The epistemic problem resides in the
fact that in order to be described, an event first needs to be observed.
These problems do not arise with imperfectives such as know ‘znati’
(see fig. 4b) because any portion of the process counts as an instance of the
process type. As far as present perfectives are concerned, the situation in
which none of the above-mentioned problems arises are cases in which the
speaker is “in control of the event” (Langacker 2001b: 263) rather than
simply reporting it. In the case of English, typical examples of control are
performatives. The speech event and the profiled bounded process are one
and the same, as in (15):
In (15) the viewing arrangement is such that the speech event is put on-
stage and profiled, as represented in Figure 5. In Croatian, the use of the
imperfective in contexts involving this particular viewing arrangement
supports the idea of the episodic nature of non-canonical imperfectives.
The non-canonical imperfective obećavati ‘promise’ is construed as a se-
ries of bounded occurrences. The speaker’s control is evident in the fact
that the speaker is capable of simultaneously performing and reporting a
single episode of an imperfective process. The episode coinciding with the
speech event is internally homogeneous, which contributes to the speaker’s
promise being construed as equal at any point during the speech event. The
128 Renata Geld and Irena Zovko Dinković
The perfective form of the same verb, that is obećati, cannot fulfill the
performative function because it cannot profile the process identical with
the speech event. It involves substantial change and it is always bounded
within the immediate temporal scope.
MS
IS
Generally, the issue of the speaker’s control over the event is relevant
for the Croatian perfective present, that is, cases involving morphologically
marked perfectives. As already mentioned, Croatian grammars claim that
the morphologically marked perfectives do not express present time.
Again, the problem resides in the default-viewing arrangement whereby the
speaker simply reports on what happens. However, we can easily imagine
the speaker actually doing what she is saying and thus making each event
coincide with the sentence that describes it, as in (17):
Actual
plane
MS
IS
3.2. Virtuality
Virtual worlds are conceptual worlds distinct from the actual world.19 They
include a number of mental constructions (cf. Langacker 2001a: 33) such
as, for example, schedules, scripts, mental replay, and the so-called struc-
tural plane (Langacker 2003: 19), which comprises generalizations repre-
senting the world’s structure. Virtual worlds represent conceptual reality
consisting of virtual events that “occur” in them by virtue of being men-
tally accessed by the conceptualizer whenever she wishes to do so. The
virtual events are ‘fitted’ into the immediate scope imposed by the present
tense morpheme. The term virtual world will be used to refer to any kind
of virtual plane in Langackerian sense. Thus, it is going to host various
virtual events: those representing particular facets of the real world’s struc-
ture and those pertaining to virtual documents and schedules related to
specific events and situations in the actual world.
Let us now reconsider example (1c) from the introduction, here repeated as
(19):
The point of view assumed in this situation is not that of a direct de-
scription of what is given in the actual world at the present moment: the
process of building a castle can be observed and described in its totality
because it does not actually occur at the time of speaking. The virtual
world holds the event representation corresponding to an indefinite number
of actual instantiations and representing the generalization pertaining to
their habitual nature (Figure 9).
Perfectives, imperfectives and the Croatian present tense 133
Structural
plane
Actual world
t
The profiled event coinciding with the speech event is a virtual instance
of the perfective process of building a sand castle. With the change in-
volved in the process and the noun dvorac ‘castle’ used in the singular, the
clause designates Barbara’s habitual activity of building (at least) one sand
castle every summer. This activity is a part of the regular course of events
stored in the structural plane.
If, instead of the morphologically marked perfective, we use the non-
canonical imperfective graditi ‘build’ and keep the time expression svako
ljeto ‘every summer’, as in example (1d) repeated below as (20), the qual-
ity of the process within our immediate temporal scope is changed, but the
occurrence of the process is again conceptual rather than actual (Figure
10).
Structural
plane
Actual world
t
At this point let us re-examine examples (2a), (2b) and (2c) repeated here
as (21a), (21b) and (21c) respectively:
going to the cinema. Thus, the imperfective in the main clause can be eas-
ily “fitted” into the frame. The perfective present in the subordinate clause
in (21a) might indicate that Iris does not go to the cinema so often, and that
when she does, she spends all her time eating popcorn instead of watching
the film. By designating a process involving substantial change, perfectives
in this kind of subordinate clauses often contribute to the meaning con-
strual involving something less ordinary and actually not very frequent, but
habitually related to what follows. Thus, the perfectives in (21a) and (21b)
do not express the habitual actions of going to the cinema, but precondi-
tions for the event that necessarily follows. If a morphologically marked
perfective is used in both clauses, as in (21b), the sentence expresses al-
most a ritualistic course of events whereby the two perfectives designate a
habitual sequence. The sequence in (21c) is a kind of neutral habitual. The
two imperfectives simply state that both going to the cinema and eating
popcorn while in the cinema are something Iris does from time to time.
However, due to the imperfective present in the subordinate clause, the
sentence has both a literal, i.e. a non-metonymical, and a metonymical
reading. Thus, going to the cinema can be interpreted either as the whole
frame of going to the cinema or the action of being on the way to the cin-
ema.
Process Process
Structural
plane
t Actual world
the speaker ‘reads’ the generalization available in the structural plane. This
special viewing arrangement excludes the durational and epistemic prob-
lems and both clauses coincide with the respective speech events, no mat-
ter what type of process is profiled by the clause.
There is another non-temporal use of the Croatian present tense that can
be analysed in terms of the conceptual configuration containing a virtual
sequence, as sketched in Figure 12. This use is illustrated by example (2d),
repeated below as (22):
Process Process
Virtual
world
t Actual world
Examples (23a) and (23b) below – which are, respectively, repeated exam-
ples (3a) and (3b), represent the use of the Croatian present tense in condi-
tional clauses. The causal relationship between the two processes is analo-
gous to the relationship sketched in Figure 11. The situation in the main
clause results from the event given in the subordinate clause.
As already mentioned in the introduction, ako ‘if’ and da ‘if’, as well as
kad ‘when’, prije ‘before’, nakon ‘after’, and others, are space builders that
overtly establish mental spaces and shift the viewpoint to the space they
establish. Thus, the content of the protasis is viewed and apprehended from
a fictive vantage point (Figure 13). When used with the present tense, the
space builders delimit spaces within the virtual world in which the epis-
temic value of the present tense is realized relative to that world.22 The
time of speaking is fictively located in the mental space established by the
subordinator and the corresponding clause, hence the conceptual occur-
rence of the process in that clause coincides with the speech event.
b. Da uči-m/nauči-m,
if study.IMPF/PERF.1SG.PRES
prošao bih na ispi-tu.
pass.PERF.1SG.M.COND on exam.LOC
‘If I studied, I would pass the exam.’
The cognitive distance, on the other hand, is marked by the use of sub-
ordinators. We believe that da marks a higher degree of cognitive distance
than ako. The role of the subordinators is twofold: they mentally frame the
event and they reflect the speaker’s relationship to whatever is described
by the present tense event that follows. The mentioned relationship as-
sumes the character of neutral, positive or negative epistemic stance (Fill-
more 1990). The present tense itself is not used to express any cognitive
distance. On the contrary, we wish to propose that the ‘presence’ of the
present tense stands in a strong conceptual opposition to the hypothetical
subordinators, especially da. The subordinator da creates a mental space
that is construed as epistemically more distanced from the speaker’s con-
ception of reality than the one established by ako, but the present tense
morpheme consistently places the event proximal to the speaker’s vantage
point (be it real or fictive).
Process
Virtual
t world
G'
Actual world
t
G
The scheduled future in Croatian, just like in English, favours a time ex-
pression which indicates the expected occurrence of the actual event, as in
(6), repeated below as (24):
Virtual
world
Actual
t world
Figure 14. The scheduled future (adapted from Langacker 2001a: 31)
This particular use of the present tense strongly supports the idea of an
epistemic background status for present tense meanings as suggested by
Brisard (1999). The nature of the present tense is such that it can never be
used to describe experiences that cannot be anticipated on the basis of what
we have learnt about reality. Cases of the scheduled future represent an
expected extension from generics and habituals. In the virtual world based
on the speaker’s experience and knowledge about particular events, future
events are scheduled and read off before the moment of their actualization.
Since canonical imperfectives in the present tense profile processes that
are stable and continue indefinitely, the use of the present tense for sched-
Perfectives, imperfectives and the Croatian present tense 141
The last case we will consider exemplifies the historical present. In the
case of historical present the relevant virtual world is built from the knowl-
edge of what we have seen and experienced in reality, and the knowledge
of what we have learnt to believe the world is made of. The mental replay
is a unique mental construct that makes possible immediate access to spe-
cific past occurrences that are conceptually occurring at the moment of
speaking. Thus, in example (7), repeated below as (27), the speaker uses
the present tense to revive actual events from the past:
indicating that the utterance is an event situated in the past reality, the con-
tent of the utterance is in the same tense as it was in the original statement.
In cases where the content of the utterance is in the present tense, the re-
ported statement is ambiguous as to the temporal duration of the situation
described. It might or might not extend through the actual speech event.
Such ambiguities are resolved by examining the nature of mental spaces
and other facets of construal. Yet, the present tense will always function as
“the most unmediated way of referring to the ground” (Brisard 1999: 365).
Virtual
world
t
Actual world
4. Conclusion
The aim of this paper was to demonstrate that the analytical models and
tools developed within Cognitive Grammar can be employed to analyse the
nature of the Croatian present tense. More specifically, we have applied
Langacker’s naïve characterization of the English present tense to the
analysis of the Croatian present tense and have provided evidence for the
claim that the Croatian present profiles a process that fully coincides with
the time of speaking.
First, we have re-examined some elements of Croatian verbal aspect and
its relation to process types. Then, we have emphasized the importance of a
particular, non-canonical viewing arrangement, which implies that what is
being viewed is not the actual occurrence of events in the world but their
virtual occurrence in the virtual world. We have proposed that all non-
temporal uses of the Croatian present tense are cases involving extreme
subjectification as a special kind of semantic shift. Furthermore, we have
144 Renata Geld and Irena Zovko Dinković
Notes
1. This specific term is used here to indicate the label used in Croatian traditional
grammars.
2. We sincerely thank Agata Kochańska, Dagmar Divjak and two anonymous
reviewers for their truly generous comments on different aspects of the paper.
3. This refers to the English simple present tense.
4. The abbreviations used in the glossing are as follows: PERF = perfective,
IMPF = imperfective, PRES = present, FUT = future, COND = conditional,
IMP = imperative, GEN = genitive, LOC = locative, ACC = accusative, DAT
= dative. Croatian inflectional morphemes always carry more than a single
meaning, i.e. they are a combination of inseparable features. With certain
nouns (eg. dvorac 'castle') these features are expressed by means of a zero
morpheme which marks not only case, but also gender and number.
5. A short description of the relationship between Langacker’s classification of
process types and Croatian aspectual phenomena will be offered in section 2.2.
6. The Croatian future tense is analytic and formed with the present tense of the
unstressed form of the auxiliary htjeti ‘want’ and the infinitive of the main
verb.
Perfectives, imperfectives and the Croatian present tense 145
7. The present conditional of Croatian verbs is formed with the aorist of the aux-
iliary biti 'be' and the participle form of the main verb.
8. It is important to point out that not all mental spaces in Fauconnier's sense are
virtual worlds in Langacker's sense. For example, temporal expressions evoke
mental spaces which have the epistemic status of past reality.
9. The terms ‘real’ and ‘non-real’ should not be confused (or identified) with
similar notions often used to describe real as opposed to hypothetical situa-
tions.
10. The Croatian term for completeness (cf. Katičić 1991: 45) is gotovost, and it is
described as a relevant grammatical category, which is claimed to be morpho-
logically marked. Croatian grammarians (Katičić 1991; Barić et al. 1995) put
forward the existence of the six conceived time categories and the tense system
is neatly fitted into these categories. For example, Katičić (1991:50) interprets
the perfekt form of the perfective verb vratiti (‘return’) as expressing the com-
pleted present. According to his interpretation, the action was completed in the
past, but it has certain ramifications for the present (for a cognitive reanalysis
cf. Stanojević and Geld 2005).
11. If we were to approximate the meaning of Croatian aspectual pairs we might
consider the following translations into English: for example, for umri-
jeti/umirati ‘die’ and ‘be dying’, for čitati/pročitati ‘read’ and ‘finish reading’,
for govoriti/progovoriti ‘speak’ and ‘start speaking’ or ‘speak out’, and so on.
However, in most cases the translation would be misleading because more ac-
curate translations of aspectual nuances are only possible in context.
12. The elaborated epistemic model plays a central role in describing various uses
of modals (Langacker 1991: 269–281). Thus, there will be no direct reference
to this particular model. However, since it is one of the three fundamental
models that enable us to grasp and describe the tense-modality system, we have
decided to include it here.
13. The Coratian perfekt is formed using the imperfective present tense of the verb
biti (‘to be’) and the l-participle of the content verb.
14. It is important to point out that we are not attempting to simplify the semantic
complexity of Croatian aspect. However, we argue that Langacker’s “quasi-
universal definitions of the perfective and imperfective” (Dickey 1997: 34) can
be easily applied crosslinguistically in the analysis of specific aspectual phe-
nomena.
15. For a discussion of the count/mass distinction and its relation to Slavic aspect
see Janda (2004).
16. Some perfective processes involve only a limiting case of change, namely
change residing only in the fact that they are temporally bounded.
17. The term distance employed here is related to Langacker’s (cf. 2006) discus-
sion of the notion of discreteness. An important aspect of our everyday experi-
ence is that a large group of discrete entities appears to us as a continuous sub-
stance when viewed from a distance. We would like to argue that the
146 Renata Geld and Irena Zovko Dinković
References
Barić, Eugenija, Mijo Lončarić, Dragica Malić, Slavko Pavešić, Mirko Peti, Vesna
Zečević, and Marija Znika
1995 Hrvatska gramatika [Croatian Grammar]. Zagreb: Školska knjiga.
Brisard, Frank
1999 A Critique of Localism in and about Tense Theory. Ph.D. diss., Uni-
versity of Antwerp.
Comrie, Bernard
1985 Tense. Cambridge: Cambridge University Press.
1976 Aspect. Cambridge: Cambridge University Press.
Cutrer, Michelle L.
1994 Time and Tense in Narratives and Everyday Language. Ph.D. diss.,
University of California, San Diego.
Perfectives, imperfectives and the Croatian present tense 147
Dickey, M. Stephen
2000 Parameters of Slavic Aspect: a cognitive approach. Stanford: CSLI
Publications.
Fauconnier, Gilles
1997 Mappings in Thought and Language. Cambridge: Cambridge Uni-
versity Press.
Filipović, Rudolf (ed.)
1978 Kontrastivna analiza engleskog i hrvatskog ili srpskog jezika [Con-
trastive Analysis of English and Serbo-Croatian] Volume 2. Zagreb:
Institute of Linguistics, Faculty of Philosophy, University of Zagreb.
Fillmore, Charles J.
1990 Epistemic Stance and Grammatical Form in English Conditional
Sentences. In Papers from the 26th Regional Meeting of the Chicago
Linguistic Society. Volume 1, 137–162. Chicago Linguistic Society.
Janda, Laura
2004 A metaphor in search of a source domain: the categories of Slavic
aspect. Cognitive Linguistics 15 (4): 471–527.
Kašić, Bartol
2002 Reprint. Osnove ilirskoga jezika u dvije knjige [Foundations of the
Illyrian Language in Two Volumes]. Zagreb: Institut za hrvatski jezik
i jezikoslovlje. Original edition, Rome, 1604.
Katičić, Radoslav
1991 Sintaksa hrvatskoga književnog jezika [Syntax of the Croatian Liter-
ary Language]. Zagreb: Jugoslavenska akademija znanosti i umjet-
nosti.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Volume 1: Theoretical Prereq-
uisites. Stanford: Stanford University Press.
1990 a Concept, Image and Symbol. Berlin/New York: Mouton de Gruyter.
1990 b Subjectification. Cognitive Linguistics 1: 5–38.
1991 Foundations of Cognitive Grammar. Volume 2: Descriptive Applica-
tion. Stanford: Stanford University Press.
2000 Why a mind is necessary: Conceptualization, grammar and linguistic
semantics. In Meaning and Cognition: A multidisciplinary approach,
Liliana Albertazzi (ed.), 25–38. Amsterdam/Philadelphia: John Be-
jamins.
2001 a Cognitive linguistics, language pedagogy, and the English present
tense. In Applied Cognitive Linguistics I: Theory and language Ac-
quisition, Martin Pütz, Susanne Niemeier, and René Dirven (eds.), 3–
37. Berlin/New York: Mouton de Gruyter.
2001 b The English present tense. English Language and Linguistics 5 (2):
251–272.
148 Renata Geld and Irena Zovko Dinković
Agata Kochańska
Abstract
1. Introduction*
Verbs in Polish, just like verbs in other Slavic languages, are either perfec-
tive or imperfective. Prototypically, perfective verbs profile processes
which are conceptualized in their totality, that is, as temporally bounded
within their immediate scope of predication. On the other hand, imperfec-
tive verbs designate processes which are conceptualized as ongoing, that is,
temporally unbounded within the adopted immediate scope. Such proto-
typical uses of, respectively, the perfective and the imperfective are illus-
trated in (1) below.
In (1a), the complete event of writing down an address, together with its
beginning and end, is profiled. On the other hand, in (1b) the immediate
scope of predication (whose temporal expanse is specified by the adverbial
clause of time gdy nadjechał autobus ‘when the bus came’) encompasses
only a portion of internal phases of the process in question, i.e. the portion
which is coextensive with the temporal boundaries of the immediate scope.
It is only these internal phases that are actually designated, although the
notion of the complete event figures in the conceptual base of the predica-
tion. In other words, the profiled phases are specifically construed as inter-
nal stages of the process in question, which is evoked in the base of the
conceptualization as evolving from its start towards its desired completion
point. Hence, even though in (1b) the temporal boundaries of the event are
excluded from profiling, they form a part of the maximal scope.
However, it is a well-known – and perhaps notorious – fact about Slavic
aspect that the choice between the perfective and the imperfective may be
motivated by many other factors, in addition to the temporally bounded or
unbounded nature of the conceived event. The motivating factors are often
subjective in character: they may have nothing to do with the objective
properties of conceived events, but rather with their subjective construal on
Conflicting epistemic meanings of the Polish aspectual variants 151
the part of the speaker. I believe that the speaker’s epistemic attitude with
respect to what he is talking about is one of such subjective factors that
motivate aspectual usage in Polish. In the present paper I would like to
consider precisely those uses of aspectual forms in Polish in which the
speaker’s choice of either the perfective or the imperfective is meant to
convey an epistemic meaning.
Specifically, I would like to consider and compare the respective epis-
temic meanings of the perfective and the imperfective in the past tense and
in the non-past tense used in reference to future events.1 My aims in the
present analysis are: first, to characterize the epistemic meanings of the
perfective and the imperfective in the past and in the non-past tense; sec-
ondly, to investigate how the epistemic meanings in each group of exam-
ples relate to the prototypical values of the aspectual variants under con-
sideration; and finally, to consider how the epistemic values of the two
aspectual variants in the past tense relate to their epistemic meanings in the
non-past tense.
The overall discussion will be organized in four sections. Section 2 will
present an analysis of the perfective/imperfective contrast in non-past tense
sentences referring to future events. The reason for starting the discussion
with perfectives and imperfectives used in reference to future events is that
in examples of this kind the aspectual choice seems to have a relatively
obvious epistemic import. The epistemic motivation behind choosing either
the perfective or the imperfective in past-tense sentences is less straight-
forward and hence, the discussion of this topic will be relegated to section
3. In both sections, I will start with describing the prototypical meanings of
the perfective and the imperfective in the relevant tense. Then I will turn to
a characterization of their respective epistemic meanings and a discussion
of how the proposed epistemic values may be related to the prototypical
meanings. Section 4 will be devoted to a discussion of the relationship
between the epistemic meanings of the perfective and the imperfective in
the past tense and their respective epistemic values in the non-past tense.
Finally, section 5 will present a brief summary and conclusions from the
overall discussion.
The theoretical framework of the overall analysis will be that of cogni-
tive grammar, with special emphasis on its idea of meaning understood in
terms of both conceptual content and construal imposed upon this content
(cf. e.g. Langacker 1987a, 1991). Equally important will be cognitive
grammar’s recognition of the relevance of subjectively construed aspects
of the conceived scene in language (cf. e.g. Langacker 1990), its idea of
virtuality (cf. e.g. Langacker 1999, 2001), and its characterization of the
152 Agata Kochańska
cognitive model of the control cycle, especially the control cycle mani-
fested in the epistemic sphere (cf. e.g Langacker 2002, 2004, ms)
Let me finish the introductory part of the present paper with a few re-
marks concerning the data to be considered. First, all examples under dis-
cussion are sentences referring only to non-repeated, internally heteroge-
neous, and complete events.2 The aim of limiting my considerations to data
of this kind is to exclude non-completion or repetitiveness as possible fac-
tors motivating the choice of the imperfective. Since in Polish non-repeated
complete events are prototypically designated by perfective verb forms,
when the speaker decides to refer to such events by means of the imperfec-
tive, his aspectual choice is not motivated by the objective properties of the
events in question, but rather by more subjective factors. Epistemic evalua-
tion of what is being talked about seems to be one such factor.
Secondly, all the examples under consideration are made-up sentences.
The reason for this is that by using made-up examples I could control the
parameters of contrast between the perfective and the imperfective. In this
way I could create contexts in which the choice between the perfective and
the imperfective clearly has a purely epistemic import. This, in turn,
greatly facilitates the ease of exposition. It should be stressed, however,
that the epistemic uses of the Polish perfective and imperfective, which are
illustrated in the present work by made-up examples, have all been attested
in the corpus that I have collected for a more extensive study of the seman-
tics of aspect in Polish (cf. Kochańska 2002a). Moreover, my own intui-
tions concerning the epistemic import of the two aspectual variants have all
been checked with at least twenty other native speakers of Polish (mostly
students at the courses in cognitive grammar that I have taught) and none
of them have expressed any reservations about the proposed semantic char-
acterizations.
The prototypical uses of the non-past perfective and imperfective are ex-
emplified in (2) below:
Even if the durational and the epistemic problem mentioned above are
disregarded, the viewing arrangement involved in conceptualizing proc-
esses coextensive with the speech time is still far from optimal, at least in
the temporal dimension. When the profiled process is coextensive with the
time of speaking, the object of observation is maximally close in time to
the observing subject. In turn, as Langacker (1993: 456) notes, the viewing
distance
shows a positive correlation with the size of the field of view. When the fo-
cus of visual attention is on a distant object, both the maximal field of view
and the viewing frame (i.e. the general locus of attention) subtend large por-
tions of the surrounding world. But if [we – A.K.] look at something very
close – e.g. the palm of [our] hand – [our] visual horizons shrink drastically
and [we] see but a limited portion of [our] immediate environment. This cor-
relation is mirrored in general conception.
Furthermore, “a decrease in [viewing – A.K.] distance does not invariably
enhance perception. As we approach an object, there is a point beyond
which any further approximation actually makes it harder to observe – we
are just too close to see it well” (Langacker 1993: 457).
One of the contexts in which Langacker’s durational and epistemic
problems do not arise (for a discussion of this point see Langacker 1987b:
83) is that of explicit performatives, whose temporal expanse – by defini-
tion – coincides with that of the speech event and whose occurrence is
initiated and controlled by the speaker. However, even in this case the ob-
ject of conception, that is, a process identified with the speech event, is so
close to the observing subject that its endpoints coincide with the bounda-
ries of the immediate scope, instead of falling properly within them. Thus,
the profiled process completely fills the conceptual onstage region. Since
in a configuration of this kind the processual endpoints fall at the fringes of
the onstage region, they are beyond the area of maximal conceptual acuity
and are not clearly “visible” to the conceptualizer. This situation is analo-
gous to the one in which we are approaching a cow and at some point we
come so close to the animal that its spatial expanse completely fills our
visual field. At that point the animal’s spatial boundaries are no longer
clearly visible to us – in such a case our viewing experience should be
more appropriately described as seeing cowhide, rather than seeing a cow.
It may thus be suggested that explicit performatives involve a concep-
tual configuration that is ambivalent in certain respects. On the one hand,
the profiled process is identified with the speech event and is by definition
conceptualized as temporally bounded. On the other hand, its endpoints –
Conflicting epistemic meanings of the Polish aspectual variants 155
tional phrase ‘for half an hour’, thereby “filling” it completely. One inter-
pretation of (6a) which satisfies this conceptual configuration is that the
clause designates internal phases of a process “cut” from a temporally
more extended complete event by the temporal przez-phrase. Under the
second interpretation – which is more relevant to the present discussion –
the profiled process is construed as starting at the beginning of the period
specified by the prepositional phrase and being completed at the end of this
period. In other words, (6a) may be understood as profiling a full instance
of a temporally bounded process whose endpoints coincide with the
boundaries of the immediate scope adopted by the conceptualizer and
specified by the prepositional phrase of time. Examples analogous to (6a)
are used with this kind of meaning when the conceptualizer specifically
focuses on the extended duration of the profiled process, especially if this
duration is longer than expected. It may be argued that focusing on the
longer-than-expected duration of the profiled process involves conceptual-
izing its temporal expanse as completely filling the adopted temporal view-
ing frame. In other words, in a conceptualization of this kind, the proces-
sual endpoints have to be construed as coinciding with the boundaries of
the immediate scope, rather than falling properly within them. The impor-
tant thing is that, under such an interpretation, it is still imperfective verb
phrases that have to be used in reference to temporally bounded processes.
As illustrated by (6b), the use of perfectives is not felicitous when the end-
points of the profiled process are specifically construed as coinciding with
the boundaries of the temporal immediate scope.
On the other hand, when – as in (7) above – the temporal extension of
the profiled process is indicated by a phrase containing the preposition w
‘in’, instead of przez ‘across’, the process in question is construed as fal-
ling within the specified time frame together with its endpoints, as the im-
port of using a temporal w-phrase is specifically that the profiled process
was completed within the specified time span. When a conceptual configu-
ration of this kind is evoked, the use of the perfective is not only permissi-
ble, but in fact the only possible option: the imperfective in (7b) is unac-
ceptable if interpreted as profiling a one-time complete event (although, of
course, its use is fully sanctioned in the case of the habitual interpreta-
tion).3
In cases analogous to those in (6a) above it is past processes that are put
in profile. Since they are conceptually viewed from a relatively distant
present-time vantage point, the reason for the coincidence between the
endpoints of the profiled event and the boundaries of the temporal immedi-
ate scope is not the “close-up” perspective, which brings about the shrink-
158 Agata Kochańska
Let me start this part of the present discussion with the observation that the
present-time meaning is not the only semantic value that the non-past im-
perfective can assume. It may also be used in reference to future complete
events, just like the perfective. This is illustrated in (8) below.
I would like to argue that in uses of this kind the perfective and the im-
perfective are not fully equivalent and that the semantic contrast between
them is epistemic in nature. The analysis that I will propose for the Polish
perfective/imperfective contrast in the future-time uses is analogous to the
analysis proposed by Langacker (cf. e.g. 1999, 2001) for the contrast in
English between the will future construction and the simple present tense
in its future use. My suggestion is that the perfective simply designates a
future event, whose occurrence is – by its very nature – only potential and
Conflicting epistemic meanings of the Polish aspectual variants 159
the future evolution of reality – if you want to know what is going to hap-
pen in the future, the only thing that you need to do is to have a look at the
relevant script and see what is there. I would like to suggest, therefore, that
the epistemic value of the perfective is the one associated with futurity by
default: from the present vantage point future events are construed by the
conceptualizer as only potential, uncertain, and beyond control. On the
other hand, the imperfective may be used in reference to those future
events which are specifically conceptualized as currently being under the
speaker’s epistemic control, for example due to being scheduled or
planned.
In (9b) above, we are dealing with the script of a schedule for future
events. However, as illustrated by the examples in (10) below, different
kinds of scripts for the future evolution of reality may be involved on dif-
ferent occasions.
The example in (10a) involves the script of a personal plan for future
actions. The scripts invoked in sentences (10b) and (10c) involve different
aspects of our conception of how the world generally works or is struc-
tured, rather than of what happens in it as a matter of contingent occur-
rence.6 In (10b), under its habitual reading, the relevant part of our idea of
how the world is structured is our conception of what is done habitually.
Sentence (10c), on the other hand, seems to invoke our mental representa-
tion of the calendrical cycle.
As has already been mentioned, the use of the non-past perfective is a
default way of referring to future events. Since it is a perfectly sensible
default attitude to view a future event as potential, uncertain, and beyond
control, this epistemic import of the perfective may reasonably be consid-
ered part and parcel of its prototypical meaning. The question, however, is
Conflicting epistemic meanings of the Polish aspectual variants 161
how the epistemic value of the non-past imperfective may be related to its
prototypical use, which profiles an ongoing present process. My suggestion
in this respect is that the apparent future-time use of the imperfective is, in
fact, a special type of the present-time use. When the perfective is used to
refer to a future event, it is this event that is actually designated. However,
in the case of the imperfective the profile is shifted to an entry in the con-
ceptualizer’s mental script for the future. At the speech time, the conceptu-
alizer consults his mental script and activates the relevant entry within it,
as a means of seemingly exercising epistemic control over the future evolu-
tion of reality. My claim is that when the imperfective is used to refer to a
future event, what is put in profile is such an activation, that is, a present-
time mental occurrence of an event, rather than its future actual occur-
rence.7 The corresponding actual future event is part of the overall con-
ceived scene, but is relegated to the conceptual base. Of course, despite
directly profiling entries in mental scripts which are activated at the time of
speaking, the uses of the imperfective illustrated by the examples in (9b)
and (10) are taken as pertaining to future events. This results from our ide-
alized cognitive models concerning the role of such mental scripts in shap-
ing the future course of events. We often seem to conceive of plans, sched-
ules, or what we view as patterns of how things are in the world as having
a causal or even a determining force with respect to future events. The
future occurrence of actions that are planned is commonly taken as being
nearly guaranteed by the very fact that we have made a plan to perform
them. Scheduled events are also taken as being bound to occur. Finally, our
experience that certain patterns have so far been regularly recurring in real-
ity leads us to the extrapolation that they are also bound to recur in the
future.
To sum up the main points of the analysis offered above, I suggest that
when the non-past imperfective is used in reference to a future event, the
profile is restricted to the present mental activation of an entry in a script
for future events. However, the overall conceived scene incorporates in its
maximal scope also the conception of the actual future event. The occur-
rence of this event is viewed as being almost guaranteed by the existence
of the mental script under consideration. This mental script is viewed as
providing the conceptualizer with a measure of “epistemic control” over
future events and it is the meaning of “epistemic control” that all uses of
the imperfective in reference to future processes seem to share.8
It may also be noted at this point that one aspect of the conceived scene
that the future-time uses of the non-past imperfective seem to share with
the prototypical present-time meaning is that in both cases the profiled
162 Agata Kochańska
tion). The very fact that the actor can potentially interact with an entity
creates a situation of tension. Any entity that appears in the actor’s field
has to be dealt with in some way. In turn, since any actual interaction with
an entity may be viewed as leading to the actor achieving at least some sort
of control over it, the potential for interaction also creates the potential for
gaining control. This potential is directed at the entity within the actor’s
field (the target). Hence, in this phase of the control cycle (called the po-
tential phase) the actor is being “pushed” towards gaining control over the
target which happens to appear within the field. In the case of the control
cycle on the epistemic level – which is my present concern – we may fur-
ther distinguish the following three stages within the potential phase: (i)
the formulation stage, when an event first presents itself to the actor’s
mind; (ii) the assessment stage, when the status of the event relative to
reality becomes an active issue; and (iii) the inclination stage, when the
actor inclines towards either accepting or rejecting the event as part of his
conception of reality, but is not yet able to resolve the matter in a definite
fashion (cf. Langacker 2002: 200).
In the control cycle manifested at any level (be it physical, perceptual,
or epistemic), the tension in the potential phase leads to the action phase,
when the actor actually does something to gain control over the target in
his field. Finally, in the result phase of the cycle, as a result of the preced-
ing action, the actor’s dominion is expanded to incorporate the target,
which is now – in one way or another – under the actor’s control. It should
be emphasized at this point that the notion of control seems to be under-
stood by Langacker in a maximally general way: in a sense, you gain
physical control over another living creature in your field both when you
catch this creature and when it irrevocably escapes you, you gain percep-
tual control over a thing in your visual field both when you manage to see
that thing clearly and when you lose sight of it for good, you gain epis-
temic control over an event both when you accept this event and when you
definitely reject it from your conception of reality. In all those cases we
have to do with the actor’s gaining control over the target in the sense that
the target in the actor’s field has been “dealt with” in some manner, the
tension of the potential phase has been resolved in one way or another, and
the target has been definitely placed with respect to the actor’s dominion
(of course, it remains definitely placed either inside or outside the actor’s
dominion only until its status becomes an active issue again).
In the case of clausal grounding predications, the relevant epistemic
dominion is the speaker’s conception of basic reality, that is, of “what the
conceptualizer accepts as having happened or obtained up to the present
164 Agata Kochańska
The next step in the present discussion is an analysis of the respective epis-
temic values of the perfective and the imperfective in the past tense. I will
start with a characterization of the prototypical meanings of the aspectual
forms under consideration in section 3.1. Then, in section 3.2, I will ana-
lyze the respective epistemic meanings of the two tense-aspectual forms
under consideration and how they are related to the prototypical meanings.
The prototypical meanings of the past perfective and the past imperfective
are illustrated by the examples in (11) below.
In (11a), where the perfective is used, the temporal immediate scope lo-
cated in the past contains all the component phases of the event eat, with
its beginning and ending. Hence, a complete process is designated. In
(11b), on the other hand, only selected internal phases of the event fall
within the temporal immediate scope and only those internal phases are
actually profiled. The profiled internal phases, however, are conceptualized
specifically as phases of a complete event. Hence, the profiled portion of
the event eat is conceptualized against the background of a temporally
more extended configuration, the idea of the entire event of eating, which
falls within the maximal scope of the overall conceptualization. In other
words, the past perfective prototypically profiles a temporally bounded
past event, which is conceptualized as a self-contained entity, without ref-
erence to anything else. On the other hand, the past imperfective prototypi-
cally designates only those internal phases of a complete process which fall
166 Agata Kochańska
within the adopted temporal immediate scope, with the idea of the com-
plete process figuring in the conceptual base.10
Let me now turn to the epistemic meanings of the perfective and the imper-
fective in the past tense. Apparently, cases of a purely epistemic contrast
between the two aspectual variants in the past tense are not easy to find.
The group of uses in which epistemic meanings seem to surface and which
I am going to discuss in the present paper are certain sentences referring to
past events of verbal communication. Example of the uses in question are
given in (12) below.
Let me note first that the semantic contrast illustrated in (12) is indeed
very subtle. In fact, both the perfective and the imperfective are actually
possible in both (a) and (b). Moreover, when the speaker wishes to refer to
a past event of verbal communication, there are multiple factors beside the
conceptualizer’s epistemic attitude which may motivate the choice of one
aspectual variant over the other. Hence, the question marks in (12) merely
signal that the aspectual variant not marked by them will be preferred by
the speaker who specifically wishes to emphasize his epistemic attitude
expressed in the first clause of the utterance and who conceives of this
attitude as being directly shaped by the profiled past event of verbal com-
Conflicting epistemic meanings of the Polish aspectual variants 167
that is, a tendency “to continue its evolution along certain paths in prefer-
ence to others” (Langacker 1991: 277) – given what we already know
about the world, we expect to learn about it some new things, but not oth-
ers. One consequence of this is that we are not equally ready to accept as
true just any proposition entering our epistemic field at the formulation
stage – for example, we do not automatically take for a fact just any state-
ment about reality that a stranger might wish to utter. Propositions aug-
menting our conception of reality in ways which are compatible with its
evolutionary momentum seem to be accepted as true in a virtually auto-
matic fashion. However, our minds may also be confronted with proposi-
tions which in one way or another conflict with the inherent dynamism of
our idea of reality. To be accepted as true, such propositions have to be
invested with an epistemic force which is sufficient to counteract the in-
herent evolutionary momentum of our conception of reality and push its
development along previously unexpected paths.
The notion of an epistemic force of a proposition is meant to capture
all kinds of contextual factors which accompany the presentation of a
proposition to the conceptualizer’s mind and which contribute to the likeli-
hood of the conceptualizer accepting the proposition in question into his
idea of reality. One of the ways in which propositions may present them-
selves to the conceptualizer’s mind is via other people’s verbal actions. In
the case of propositions presented to the conceptualizer’s mind in this par-
ticular way the relevant aspects of the proposition’s epistemic force in-
clude, for example, the reliability of the speaker who presents the proposi-
tion to the conceptualizer’s mind or the degree of his own epistemic
commitment to what he is talking about. To put it differently, people differ
with respect to the degree to which they are recognized as authorities on
particular subject matters. In turn, a proposition expressed by an authority
in the relevant domain will naturally have a greater epistemic force than
the one conveyed by a layman. Also, the person expressing a proposition in
an utterance may invest his utterance with varying degrees of conviction
with respect to what he is saying, thereby endowing the expressed proposi-
tion with varying degrees of epistemic force.
The evolutionary momentum of our conception of reality may either
work in tandem with the epistemic force of a proposition presented to our
mind or counteract that force. In any case, a proposition becomes accepted
into the conceptualizer’s epistemic dominion only when its epistemic force
manages to overcome any potential resistance caused by the evolutionary
momentum of the conceptualizer’s dynamically changing knowledge sys-
tem. It may also happen that the proposition presents itself to the conceptu-
Conflicting epistemic meanings of the Polish aspectual variants 171
alizer’s mind with an epistemic force that is too weak to overcome the
evolutionary momentum of the existing knowledge. In this case, the evolu-
tionary momentum of our reality conception may make us reject the propo-
sition which is presented to our mind. Another possibility is that this evolu-
tionary momentum may be too weak in the relevant direction to counteract
fully the epistemic force of the proposition. The two forces may thus re-
main in a sort of dynamic equilibrium, a state of tension in which the epis-
temic status of the proposition remains unsettled in the conceptualizer’s
mind.12
In the context of the present analysis of the epistemic uses of Polish
past perfective and imperfective verbs of verbal communication, it may be
interesting to note that the way in which the above-mentioned aspects of
the proposition’s epistemic force figure in the conception of the profiled
past verbal action seems to correlate with the level of acceptability of using
either the perfective or the imperfective. This is illustrated by the examples
in (13) and (14) below:
The final issue in the present analysis is the question of the relation be-
tween the epistemic value of the perfective/imperfective contrast in the
past tense and its counterpart in the non-past tense. It turns out that in the
past tense the perfective tends to be associated with epistemic certainty and
commitment, and the imperfective with the lack thereof. On the other hand,
in the non-past tense the epistemic values of the two aspectual variants
seem to be reversed: here the imperfective is correlated with a greater de-
gree of epistemic commitment, while the perfective may go together with
the epistemic uncertainty typically associated with the future. The epis-
temic behavior of the Polish perfective and imperfective in the past tense
seems to support the hypothesis put forward by Fleischman (cf. 1995: 539),
who claims that there is a cross-linguistic “attraction” between the imper-
fective and the irrealis, since the notions of irrealis and the lack of epis-
temic certainty seem to be closely related. However, the epistemic values
associated with the perfective and the imperfective in their future-time uses
directly contradict Fleischman’s suggestion: it is the imperfective which
Conflicting epistemic meanings of the Polish aspectual variants 173
portrays the designated future event as almost bound to occur, and there-
fore – as nearly part of reality. On the other hand, when the perfective is
used, the future event is conceptualized as potential, not yet real, and there-
fore epistemically uncertain.
The question therefore is: Is this seemingly inconsistent epistemic be-
havior of the Polish perfective and imperfective a vagary of grammar? Is
this an illustration of the fact that grammar is extravagant and whimsical, at
least as far as its semantic contribution is concerned?
I believe that the answer to these questions should be in the negative. In
my view, there is nothing extravagant or whimsical about the epistemic
behavior of the Polish perfective and imperfective, and any inconsistency
in that behavior is only apparent. However, to see this one needs to take
into consideration two factors. The first is that a common property of vari-
ous uses of the Polish perfective is that it presents the profiled event as a
self-contained whole, which is complete and independent of anything else.
The imperfective, on the other hand, evokes a conceptualization in which
the profiled phases of the process are embedded within a temporally more
extended configuration. In the case of the prototypical use of the imperfec-
tive, the profiled phases are conceptualized specifically as phases of a
complete process (as e.g. in Piotr zapisywał mój adres, gdy nadjechał
autobus ‘Piotr was writing down my address when the bus came’ (cf. (1b)
above). In turn, when the imperfective is used in reference to complete
events (as, e.g. in (12b) above Piotr mówił, że go [artykuł] skończy ‘Piotr
said that he will finish the paper’), these events are often thought of as
being conceptually linked to the time of speaking, thereby being part of a
temporally more extended configuration – if not in the objective, then at
least in the subjective sense.13
The second factor is that the past, the present and the future are not
simply different slices of the time line. On the contrary, as suggested by
Brisard (1999: 11–12, fn.9), they are experienced by the human conceptu-
alizer as having quite different epistemic properties. Events in the past
have already happened and thus, can be known to us. Their epistemic
status has already been decided upon. Events in the present are in the proc-
ess of happening – hence, the present is where potentiality turns into reality
and where unknown turns into known. The epistemic status of events in the
present is in the process of becoming resolved. Finally, events in the future
are as yet only potential and awaiting possible future realization. Thus,
they are not yet real and cannot be known, but only anticipated or imag-
ined.
174 Agata Kochańska
5. Concluding remarks
In the present paper, I have tried to characterize the epistemic values of the
Polish perfective and the imperfective in the past and in the non-past tense.
I have argued that the epistemic values under consideration may be viewed
as relatively straightforward extensions from the prototypical meanings of
the two aspectual variants in the two tenses, once it is recognized that the
objectively construed future- vs. present-time realization or past comple-
tion vs. non-completion may have subjective counterparts in the conceptu-
alizer’s epistemic sphere. The conceptualization evoked by the epistemic
use of the non-past imperfective is that at the time of speaking the concep-
tualizer mentally activates a virtual analog of an actual future event and
thereby exercises a measure of epistemic control over that event. On the
other hand, the absence of such present-time mental activation, which is
indicated by the use of the non-past perfective, conveys the construal in
which the future event in question remains beyond the conceptualizer’s
epistemic control. In turn, the choice of the past perfective may evoke a
conceptualization in which not only the objectively construed profiled
process, but also the subjective epistemic control cycle is viewed as com-
pleted, with the relevant proposition placed under the speaker’s epistemic
control. On the other hand, the notion of non-completion evoked by the
past imperfective may be construed as pertaining not to the objective non-
completion within the immediate scope of predication, but rather to the
non-completion of the subjective epistemic control cycle, in which the
epistemic status of a proposition remains an open issue. The final claim of
the present paper has been that the conflicting epistemic values of the per-
fective/imperfective contrast, respectively, in the past and in the non-past
tense – far from being a whim of the Polish grammar – are in fact strongly
motivated by the interaction between the conceptual configurations evoked
176 Agata Kochańska
Notes
* I would like to thank Elżbieta Górska and Ronald W. Langacker, who gener-
ously offered me their time and commented on earlier versions of the present
analysis. Both my analysis and the way it is presented benefited greatly from
their wise guidance. My gratitude is also due to Dagmar Divjak for all her
helpful comments. Needless to say, all the remaining flaws and errors are en-
tirely my own.
1. As will be illustrated later on, the Polish non-past perfective designates a
future event, while the non-past imperfective may profile either a present
process (the prototypical use) or a future one. The present study will be con-
cerned with the respective epistemic values of the two aspectual variants in
the non-past tense, when they are both used in reference to future processes.
Conflicting epistemic meanings of the Polish aspectual variants 177
2. Note that in what follows I will use the terms ‘complete events’ or ‘complete
processes’ in reference to what is profiled by the examples under considera-
tion. These terms are meant to capture the fact that the examples in question
evoke the idea of a full instance of temporally bounded and internally hetero-
geneous processes. I decided to use the term ‘complete process/event’ instead
of ‘completed process/event’, as the latter term seems to bring about connota-
tions with the past (it is only a past process, a process that already occurred
that might or might not have been completed). In turn, the examples consid-
ered in the present paper evoke the idea of both past and future processes
which are, in both cases, conceptualized in their totality.
3. For a more detailed discussion of the semantics of przez- and w-phrases in
their temporal uses and their relation to the spatial ones, see Kochańska
(1996).
4. Such scripts are instances of what Langacker (cf. e.g. 1999) calls virtual
documents.
5. Note that, in addition to the habitual reading, sentence (10b) may also have an
interpretation pertaining to the speaker’s plan for future actions (‘I’m having
my breakfast in an hour’). However, it is only habitual reading that is relevant
at this point.
6. This conception is what Langacker calls the structural plane (cf. Langacker
1999: 95–97 and 2001:270, as well as the discussion of the struc-
tural/phenomenal distinction in Goldsmith and Woisetschlaeger 1982).
7. Note that the characterization proposed here involves just a slightly different
phrasing of how Langacker characterizes the English contrast between the
will-future and the future use of the simple present tense in terms of the actu-
ality/virtuality distinction (cf. e.g. 1999, 2001). The idea is that both the will-
future in English and the non-past perfective in Polish profile a future event in
the actual plane. On the other hand, the future uses of both the English simple
present tense and the Polish non-past imperfective designate a present-time ac-
tivation of an entry in a mental script, that is, a present occurrence of the rele-
vant event in a virtual plane. In this context it might be observed that, for ex-
ample, in (9a) above the Polish non-past perfective, which designates an actual
future event (Samolot odleci, jeśli mgła opadnie), is translated into English
with the will-future (‘The plane will depart if the fog falls down’). On the
other hand, the Polish non-past imperfective in (9b), which profiles a present
virtual occurrence (W rozkładzie samolot odlatuje za godzinę) is translated
with the English simple present tense (‘According to the timetable, the plane
departs in an hour’). It may also be noted that the English translations of the
remaining Polish examples considered in this section (the examples in [10])
contain present-tense verbs, instead of the will-future. Both (10b) and (c) are
translated with simple present verb forms, while (10a) is translated with the
present progressive. A discussion of why the English equivalent of (10a) be-
haves differently than the translations of (10c) and (d) is beyond the scope of
178 Agata Kochańska
this paper. What is important for the present analysis is that all the English
sentences in question contain present-tense verb forms, instead of the will-
future, as all of them profile present-time activations of entries in a mental
script, rather than a corresponding future actual event.
8. This last way of thinking about the uses of the imperfective under considera-
tion has been suggested to me by Ronald Langacker (personal communica-
tion).
9. This way of thinking has been suggested to me – independently and on differ-
ent occasions – by Elżbieta Górska and Ronald Langacker. See also the dis-
cussion on the same point in section 3.1 and section 4 below.
10. The same point is made for non-past perfectives and imperfectives in section
2.2 above, and in reference to the Polish perfective and imperfective in general
in section 4 below.
11. It has been suggested to me by Ronald Langacker that the use of the perfective
in (12a) shows an affinity with the so-called factive predicates in that it also
involves strong identification of the actual ground with the virtual ground in-
voked by the complement clause (for a discussion of factivity cf. Langacker
2004 section 5). The term strong identification is meant to convey the idea
that the virtual ground and grounding relation conjured up mentally in the
conceptualization evoked by the complement clause are fully identified with
the actual ground and the actual grounding relation, with the consequence that
the conceptualizer not only formulates in his mind the proposition in the com-
plement clause, but also embraces it as part of his own conception of reality
(cf. Langacker 2004: 548–549). One difference between the epistemic use of
the perfective and factive predicates seems to be that factive predicates just
presuppose the conceptualizer’s commitment to their truth, while the epistemi-
cally used perfective seems to specifically convey this commitment.
12. I would like to note at this point that my use of terms such as ‘resistance’ or
‘force’ in the above description of the cognitive model governing human func-
tioning in the epistemic domain is by no means accidental. It was first ob-
served by Talmy (1988) that there are numerous abstract concepts which are
understood metaphorically in terms of forces, counterforces, presence or ab-
sence of barriers, etc. Such force-dynamic elements present in our understand-
ing of abstract notions also have linguistic manifestations. It should come as
no surprise that the ideas of force and resistance are present in the characteri-
zation of a cognitive model of how humans function with respect to issues of
knowledge – in fact, this conceptual area was one of the clearest cases in
which the existence and importance of force-dynamics in cognition and lan-
guage was first demonstrated (cf. the analysis of English epistemic modals in
Talmy [1988], as well as in Sweetser [1990]).
13. As mentioned above, this way of characterizing the Polish perfective and im-
perfective in schematic terms has been suggested to me – independently and
Conflicting epistemic meanings of the Polish aspectual variants 179
References
Brisard, Frank
1999 A critique of localism in and about tense theory. Ph.D. diss., Univer-
sity of Antwerp.
Fleischman, Suzanne
1995 Imperfective and irrealis. In Modality in Grammar and Discourse,
Joan Bybee and Suzanne Fleischman (eds.), 519–551. Amster-
dam/Philadelphia: John Benjamins.
Goldsmith, John, and Erich Woisetschlaeger
1982 The logic of the English progressive. Linguistic Inquiry 13 (1): 79–
89.
Kochańska, Agata
1996 Temporal meanings of spatial prepositions in Polish: The case of
przez and w. In The Construal of Space in Language and Thought,
Martin Puetz and René Dirven (eds.), 491–508. Berlin/New York:
Mouton de Gruyter.
2002a Selected issues in the semantics of the Polish imperfective. A cogni-
tive grammar account. Ph. D. diss., Warsaw University.
2002b A cognitive grammar analysis of Polish non-past perfectives and
imperfectives: How virtual events differ from actual ones. In
Grounding. The Epistemic Footing of Deixis and Reference, Frank
Brisard (ed.), 349–390. Berlin/New York: Mouton de Gruyter.
Langacker, Ronald W.
1987a Nouns and verbs. Language 63: 53–94.
1987b Foundations of Cognitive Grammar. Vol. 1. Theoretical Prerequi-
sites. Stanford, California: Stanford University Press.
1990 Subjectification. Cognitive Linguistics 1 (1): 5–38.
1991 Foundations of Cognitive Grammar. Vol. 2. Descriptive Application.
Stanford, California: Stanford University Press.
1993 Universals of construal. Proceedings of the Annual Meeting of the
Berkeley Linguistic Society 19: 447–463.
1999 Virtual reality. Studies in the Linguistic Sciences 29 (2): 77–103.
2001 The English present tense. English Language and Linguistics 5 (2):
251–272.
2002 The control cycle: why grammar is a matter of life and death. Pro-
ceedings of the Second Annual Meeting of the Japanese Cognitive
Linguistics Association: 193–220.
180 Agata Kochańska
Sweetser, Eve E.
1990 From Etymology to Pragmatics. Metaphorical and Cultural Aspects
of Semantic Structure. Cambridge: Cambridge University Press.
Talmy, Leonard
1988 Force dynamics in language and cognition. Cognitive Science 12:
49–100.
Conjunctions, verb forms, and epistemic stance in
Polish and Serbian predictive conditionals
Abstract
In this paper, we use data from temporal, conditional, and coordinate con-
structions in Polish and Serbian to establish basic formal and semantic pa-
rameters defining conditional meaning in the two languages. Relying on the
mental spaces analysis of conditional constructionsin English (Dancygier
and Sweetser 2005), we describe relative salience of different constructional
parameters in Polish and Serbian. Specifically, we establish, through an
analysis of a broad range of conjunctions and verb forms, that, similarly to
English, predictive meaning is central to conditionality in both Polish and
Serbian. However, in contrast to English, the Slavic languages under scru-
tiny rely to a lesser degree on specific conditional conjunctions and on
clause order iconicity, and in fact favor different parameters (e.g., tense,
mood, and aspectual forms, or overt markers of sequentiality) as central to
predictiveness at a constructional level. At the same time, compositional
mechanisms which are responsible for the emergence of predictive condi-
tional meanings, even in absence of overt markers of conditionality, remain
very similar to English, thus giving support to the general concept of con-
structional compositionality.
Recent work within the broad area of cognitive linguistics devotes signifi-
cant attention to the correlation between syntactic form and meaning. More
specifically, work in Construction Grammar seeks new theoretical solu-
tions to numerous questions related to the correlation between meaning and
form on the sentential level.1 However, recent work within the Mental
182 Barbara Dancygier and Radoslava Trnavac
of the sentence He got up, went up to the window, and waved, discussed by
Barentsen (1998:52), requires all three verbs to be used in the past form.
Also, observations made in Trnavac (2006) confirm the link between per-
fective aspect and ‘sequenced connection’ in the cases of Serbian and Rus-
sian. A more thorough discussion of the role perfective aspect plays as a
marker of ‘sequenced connection’ in the construction of conditional mean-
ings in Polish and Serbian requires further investigation, but the examples
we discuss below support the hypothesis.
In contrast to Polish and Serbian, English seems to rely primarily on
tense and modality in its construction of conditional meaning. Among oth-
ers, the contrast between the present tense and the past tense emerges as a
primary marker of epistemic stance. Interestingly, while past tense can also
be used to mark negative stance outside of conditionals (as in I thought you
were married, when the speaker is expressing uncertainty, rather than de-
scribing a past belief), the present tense may be used as a marker of neutral
stance in paratactic constructions such as (2) and (3),4 but it also comple-
ments other features of the sentences (such as the possibility of sequen-
tial/causal link between p and q) in suggesting a predictive interpretation:
that in English, but the formal means establishing these meanings are dif-
ferent in interesting ways. Furthermore, both of the languages we consider
confirm that a whole range of conditionally interpreted constructions is
prompted by mechanisms of constructional compositionality.
In the present section, we will outline the use of major temporal conjunc-
tions in terms of the stance they represent and the typical verb form pat-
terns they correlate with. While Polish and Serbian conjunctions are rela-
tively flexible in their co-occurrence with verb form patterns, there are
some restrictions which point to epistemic stance as a major distinguishing
category. In our examples below, we will highlight the most common pat-
terns and the important exceptions, but an exhaustive discussion of all the
possible combinations and their meanings would exceed the limits of this
paper.
Epistemic stance, as first introduced by Fillmore (1986, 1990a, b), is a
category which also applies to temporal conjunctions. In Fillmore’s origi-
nal discussion, a temporal conjunction like when marks the speaker’s posi-
tive stance – according to what the speaker knows at the moment of
speech, the future space being set-up is likely to become factual. The verb
forms used in a futurate construction like When they get married, they will
move to France, the same as the ones in (1), also indicate the predictive
interpretation of the whole construction. The similarity between if and
when constructions in English is primarily apparent among futurate con-
structions, but the difference in stance distinguishes them well: while when
typically marks positive stance, if does not, correlating with neutral and
negative stance instead. In Polish and Serbian, for comparison, the bound-
ary between temporal and conditional conjunctions is less clear, and fur-
ther complicated by the possibility for temporal conjunctions to be used in
constructions marking negative stance through conditional mood. Conse-
quently, the primary stance contrast is not between positive stance on the
one hand, and neutral and negative on the other, but rather between the
negative stance on the one hand, and the neutral and positive ones on the
other, with the latter two not being distinguished formally.
Kada is the primary temporal conjunction in Serbian. As example (4) il-
lustrates, it introduces a predictive futurate construction with a positive
stance, very much in the same way as its English equivalent.5 The contrast
between the non-predictive protasis and the predictive apodosis is also
Polish and Serbian predictive conditionals 187
marked similarly to English – the present tense marks p, while the future
tense marks q.6 However, the temporal clause itself usually requires perfec-
tive aspect – a grammatical feature which, as we show below, plays an
important role in predictive constructions in the languages under investiga-
tion here.
ti knjigu.
you.2SG.DAT book
‘If you came, I would give you a book.’
Both (6) and (7) represent standard predictive reasonings, where the
prediction of q depends on the prior occurrence of the non-predicted p, and
both are significantly more common with the perfective present form in
both clauses. This is not surprising in view of the fact that the perfective
present form in Polish is the only available expression of future perfective
events (the only other verb form with future meaning is the periphrastic
future imperfective).8
In Polish, contrary to the case of Serbian kada, the association of kiedy
with positive stance is strong enough that conditional mood is not possible
in predictive constructions with kiedy as a space builder. This makes kiedy
more similar to English when than to Serbian kada. However, the restric-
Polish and Serbian predictive conditionals 189
The verb forms in a gdyby construction such as (8) do not clearly mark
temporal reference, so the sentence can be interpreted, depending on the
context, as a polite offer of a future action or a regret about a past missed
chance. Although there are no clear-cut distinctions, sentence (8a), with an
imperfective verb in p, is more likely to be interpreted as present or future,
while the perfective aspect in the protasis of (8b) is more readily under-
stood as referring to the past; still, a time adverb such as jutro (tomorrow)
or wczoraj (yesterday) could be used in both versions and would push the
interpretation towards the future or the past respectively (the future inter-
pretation is then more strongly marked for politeness). It seems, then, that
the presence of conditional mood (negative stance) is the overriding con-
structional feature in these cases.
Another conjunction in Polish is jak, whose temporal sense is best de-
scribed as when and after put together (see Bielec 1998) – it does imply
completion of p prior to q.11 Jak is thus most common in futurate sen-
tences, occasionally used in past contexts (often with tylko (‘only’), which
then stresses the immediacy of q following p), and practically unacceptable
in construction referring to the present (unless they are generic). The as-
190 Barbara Dancygier and Radoslava Trnavac
sumption of the jak-space event being completed before the main clause or
q-clause event makes jak an ideal space builder of predictive constructions,
mainly with neutral stance:
soon as the mood marker is added. In rare cases, as in the case of Polish
gdy/gdyby, the rise of the conditional conjunction eventually obliterates the
underlying commonality of the temporal and conditional uses. As regards
gdy, its underlying neutral stance seems to have supported an emergence of
a conjunction which further increases its distance from positive stance,
especially since the salience of the neutral versus negative contrast typi-
cally exceeds that of the contrast between neutral and positive.
In conditional sentences with temporal conjunctions aspect seems to
play a secondary role. However, it does support the positive/neutral versus
negative contrast mentioned above. There is a general tendency in both
languages for sentences with the positive/neutral epistemic stance and fu-
ture reference to occur with perfective aspect (see use of kada with the
future events in Serbian, as well the occurrence of the perfective present in
the case of kiedy and gdy with positive/neutral stance in Polish), while
sentences with negative epistemic stance seem to freely occur with both
aspectual forms in both languages.
It could be concluded that in Polish and Serbian the joint contribution
of temporal conjunctions and aspectual forms is overridden by the role of
the contrast between indicative and conditional mood. By the same token,
negative stance emerges as the central parameter of conditionality in the
two languages.
3. Conditional conjunctions
While da (as in [12] and [13]) marks counterfactuality as the only avail-
able interpretation of the construction’s protasis, it also does not require
the use of the conditional mood in the protasis; consequently, bi, the pri-
mary marker of negative epistemic stance, appears in the apodosis only.
Also, sentences with da are partially sensitive to the preference for the
imperfective aspect – the protases referring to the present require an imper-
fective form. The case of da as a marker of counterfactuality also raises
interesting questions about the nature of counterfactual meanings as such.
While earlier analyses of English, such as Dancygier (2002), Dancygier
and Sweetser (2005), make a very strong argument for counterfactuality
being constructed on the basis of negative stance and the structure of the
base space, Serbian seems to have a construction specializing in counter-
factuality as such. For comparison, the construction with ako in (14),
which refers to a future situation (and is thus distanced, but not counterfac-
tual), employs bi, the marker of conditional mood/negative stance both in
the protasis and in the apodosis.
Polish and Serbian predictive conditionals 193
Polish has no verb forms which are clearly associated with predictive-
ness. At the same time, the perfective present, as in Serbian, suggests two
meaning components crucial to predictive meanings – boundedness and
sequentiality. One can thus conclude that futurate conditionals in Polish
rely to a lesser degree on the contrast between the unpredicted p space and
Polish and Serbian predictive conditionals 197
Although, as (25) and (26) suggest, both aspects can occur in the prota-
sis and apodosis of this construction, conditional reading is more salient in
(25), with two perfective sequenced events. Sentence (26), for comparison,
represents two parallel (non-sequential) actions and its reading is better
described as temporal. Furthermore, the verb in the apodosis of (25) has to
be marked by the future tense, since perfective present cannot mark future
reference in a coordinate construction Presumably, then, coordinate con-
structions like (25) can receive a conditional interpretation because they
rely heavily on the verb forms – the imperative in p and the future in q.
This constructional pattern is thus quite similar to English, where the im-
perative, tenseless, but suggesting a future non-predicted situation desired
by the speaker, plays the role of a protasis marker in predictive reasonings.
The imperative is thus a verb form which, when used in protases, composi-
tionally prompts for a predictive interpretation of the construction both in
Serbian and in English, but in Serbian it also has to be matched by a future
form in the apodosis.
In Polish the verb form selection in coordinate constructions is less sig-
nificant. As could be expected, imperfectives are extremely rare, whether
with future or present reference, and express a correlation or a generic
statement. Perfectives, on the other hand, which play an important role in
conditionals, are also common in coordinate constructions, and play a simi-
lar role of establishing a sequence of two causally related events.
Such constructions in Polish typically require that both p and q use the
perfective present forms, which refer to the future while profiling a se-
quence of two bounded events. In many cases, however, the perfectives in
p and q do not ensure the construction’s acceptability if a conjunction is
absent, as in (28):
???
(28) Przeczytasz tę książkę, zobaczysz że miałam rację.
read.PRES.PF.2SG this book see.PRES.PF.2SG that have.PRET.1SG right
‘You read this book, you’ll see that I was right.’
Polish and Serbian predictive conditionals 203
What these examples show, then, is that in Polish the verb forms and
the iconic sequence of p and q play a role in the interpretation of construc-
tions as predictive, but they are not sufficient. The presence of a conjunc-
tion is required to help establish a link between the clauses. Without it, as
in (30a), the clauses are treated as independent utterances.
The examples in this section suggest that coordinate and conjunc-
tionless constructions in Serbian and Polish rely on very much the same
constructional features as their standard conditional counterparts. While
Serbian is more dependent on its verb forms pattern, Polish relies to a
higher degree on the selection of an appropriate conjunction. In both lan-
guages, the sequenced perfective forms prompt conditional reading and
neutral epistemic stance, as in the previously analyzed types of construc-
tions. However, as in the cases of other constructions, the perfective forms
in coordinate/conjunctionless sentences are not sufficient to mark condi-
tionality without the presence of other constructional features. In both lan-
guages these sentences need the support of a coordinate conjunction or a
Polish and Serbian predictive conditionals 205
additive function of and still seems to prevail (as in [32a]), and the con-
structional contribution of the imperative (in [32b]) does not override it:
7. Conclusions
The picture which emerges from the contrastive review of major parame-
ters of predictive conditionality suggests many similarities between the
constructional and semantic features of the English sentences and those of
Serbian and Polish. Not surprisingly, all three languages are sensitive to
the parameters of epistemic stance and express it through the choice of
conjunctions and verb forms. Also, all three languages use verb forms to
mark the respective roles of p and q clauses in prompting for the predictive
reading and to signal the speaker’s negative stance. Furthermore, all three
select constructional forms which best express aspects of predictiveness
such as sequentiality or unpredicted status of p. However, the differences
in the degree to which a language relies on any of the parameters men-
tioned is what distinguishes them in significant ways.
In English, if is the primary conditional conjunction; it correlates with
the neutral or negative stance of constructions, while when is primarily
used as a positive stance space builder. In Polish and Serbian, for compari-
208 Barbara Dancygier and Radoslava Trnavac
Notes
References
Ašić, Tijana
2000 Le présent perfectif en Serbe: temps, mode ou puzzle? Cahiers de
Linguistique Française 22: 275–294.
Athanasiadou, Angeliki, and René Dirven (eds.)
1997 On conditionals again. Amsterdam: John Benjamins.
Barentsen, Adrian
1998 Priznak “sekventnaja svaz’’’ i vidovoe protivopostavlenie v
russkom jazyke [The property of ”sequential connection” and as-
Polish and Serbian predictive conditionals 213
Dagmar Divjak
Abstract*
1. Introduction
Broadly defined, verbs express actions, processes and states, often sub-
sumed under the cover term “events” (Pustejovsky 1991) or “situations”
(Smith 1997).1 When two events or situations need to be expressed, typi-
cally, two clauses are used, be it two independent clauses, two coordinated
clauses or a main and subordinated clause (Talmy 2000 ch. 6; Cristofaro
2003 ch 5–7). Yet, the Russian language includes 293 verbs,2 such as de-
cide or urge that combine with another event or situation in the form of an
infinitive, yielding She decided to study Russian or He urged her to par-
ticipate in that discussion. These verbs, for which the [VFIN VINF] structure
is a point of intersection, are the subject of this paper.
On a motivational iconic view the [VFIN VINF] pattern signals a high de-
gree of interrelatedness of the events involved. I will investigate whether
two events that are captured in a [VFIN VINF] construction are indeed interre-
lated more closely than events that are divided over two full fledged
clauses. The degree of clause integration will be taken as indicative for the
degree of event integration (cf. Givón 2001 ch. 12); clause integration will
be measured by analyzing properties from the argument- and temporal
event structure of the verbs that combine with an infinitive (section 3).
Each of these properties will be interpreted in terms of its cognitive and/or
functional import. The result is a “binding scale” (cf. Givón 2001: 41) from
looser to tighter integration, that comprises eight degrees (section 4). This
scale contributes to our understanding of why certain categories of verbs
might be more prone to grammaticalization than others (section 5).
Data on the 293 verbs’ argument and temporal event structure was col-
lected by means of elicitation tests with 15 native speakers of Russian.3
The elicitation experiment with these native speakers of Russian was set up
as a small number design. Five native speakers judged the constructional
possibilities of the 293 verbs on a three-point scale.4 To guard against lexi-
cal effects, the tests were carried using pro-forms (cf. Smessaert et al.
2005),5 which ensures that the mutual effect of lexical items in a construc-
tion is minimized as much as possible; as a result, the acceptability or un-
acceptability of a particular construction is very unlikely to be influenced
by the lexical compatibility or incompatibility of the words used. More-
over, to check for repetition effects in judgments of grammaticality 10
control judgments were collected for every verb in every construction type
from an ever varying pool of native speakers.6 Finally, the results obtained
were systematically compared against information contained in dictionar-
ies, against utterances found in the Amsterdam Corpus (Barentsen, s.d.) as
well as on the Internet.
Systematization of the results was aided by the surprising consistency
of the elicitation results. Radically opposed judgments did not occur and a
verb was considered to be acceptable in a construction if 10 out 15 judg-
ments were positive and the construction was attested in corpora and/or on
the Internet. If, on the contrary, 10 out 15 judgments were negative and the
224 Dagmar Divjak
construction was not attested in corpora and/or on the Internet, it was con-
sidered infelicitous. Less clear cut cases were assigned a “?”.
2.3. Parameters
There are three important parameters along which verbs that combine with
an infinitive vary. These parameters relate to the function verbs typically
fulfill: verbs express events or situations that have participants and take
place at a certain moment in time. The first parameter zooms in on the fact
that the main participants of events or situations are encoded in the verb’s
argument structure that is the basis of the simplex sentence (section 3.1).
The second parameter checks whether the two verbs expressing these
events can be linked alternatively using a complex sentence consisting of a
main clause and a subordinate clause (section 3.2). The third parameter
investigates how the events interact with time, as the events each of these
verbs expresses take place at a specific moment in time and have a specific
temporal contour, i.e. an imperfective and/or perfective aspect (section
3.3). These three syntactic parameters have cognitive-semantic dimensions
that can be interpreted as encoding the degree of integration between the
finite verb event and the infinitive event.
In the following section I will sketch how the three parameters differen-
tiate between the 293 verbs studied.
Support for considering both (1) and (2) instantiations of one and the
same ‘plan’ sense can be found at the more abstract level. The something
versus do something test can be seen as an operationalization of the distinc-
tion between nominal things and atemporal verbal relations: infinitives can
be referred to as doing something, whereas, nouns can typically be cap-
tured under the cover-term something as (1’) and (2’) show. Example (2’)
then reveals that both a noun designating a thing and an infinitive profiling
an atemporally construed process are possible answers to the question
‘What is he planning?’.
The situation is quite different with verbs that are often named as near-
synonyms to planirovat’ ‘plan’, i.e. dumat’ ‘intend, think (of)’, namere-
vat’sja ‘intend, mean’ and sobirat’sja ‘intend, be about’ (for a full account
see Divjak 2006). Namerevat’sja ‘intend, mean’ lacks the possibility of
combining with non-verbal entities altogether, thus being restricted to
combinations with an infinitive. For dumat’ ‘intend, think (of)’ and sobi-
rat’sja ‘intend, be about’ a difference in construction, i.e. a combination
with an infinitive or with a noun, goes together with a sharp difference in
the meaning of dumat’ and sobirat’sja themselves. Take the example of
dumat’ (taken from Apresjan and Pall 1982: 389–390, vol. 1). With a
preposition and a nominal object dumat’ roughly means ‘think’, as is ex-
emplified in (3); in this sense, dumat’ has a delimitative perfective coun-
terpart podumat’. The other lemma, illustrated in (4), is restricted to com-
binations with an infinitive, and can be translated ‘intend’. In this sense,
dumat’ does not have a perfective counterpart.
and complement clauses share their referents, the more likely they are to be
semantically integrated as a single event; and the less likely the comple-
ment clause is to be coded as an independent finite clause”. Thus, finite
verbs that are restricted to combinations with infinitives in which morpho-
logical and syntactic information is to a large extent shared across the
verbs and in which strict co-reference rules apply, signal a higher degree of
dependence on a second verb than finite verbs that can combine with both
infinitive and full-fledged complement clause. Although the latter con-
structions also consist of two events, both events exist to a certain extent
independently of one another and the infinitive event can be made subordi-
nate to the finite verb event.
Apart from morphosyntactic integration, Givón (2001: 45) identifies co-
temporality as a necessary cognitive pre-condition for considering two
events a single (though complex) event. Cristofaro (2003: 120) takes the
comparable “degree to which the boundaries between these SoAs [States of
Affairs] are eroded or kept intact” as the basic component of semantic in-
tegration. The next section will look into how boundaries can be imposed
on Russian verbs that combine with an infinitive and how these verbs deal
with the co-temporality requirement.
3.3. On time
The verbs ljubit’ ‘love, like’ and osmelit´sja ‘dare’ demand overlap in
or tight sequentiality of time. This requirement is illustrated in examples
(12) and (13).
three parameters and its implications for the meaning of the verbs that oc-
cur as finite verbs in the [VFIN VINF] pattern.
From a logical point of view, there are eight possible ways to combine the
positive and negative results of the three tests, the thing-test, the that-
complementation test and the time test. These eight logical combination
options are illustrated in Table 1, an “idealized” model that – for ease of
presentation – imposes clear-cut boundaries on a phenomenon, graded in
nature (cf. Divjak 2007). In addition, Table 1 abstracts away from the for-
mal details distinguished for each parameter, e.g. the case marking on the
pro-nominal paradigm or the type of complementizer.19 The verbs that
populate each column are presented in Table 2 in the Appendix
On the first row in Table 1 the number of verbs that qualify for a certain
column is provided in square brackets. Rows two, three and four provide
the exact value of each parameter. The shorthand “thing” refers to the
something versus do something test presented in section 3.1. If a finite verb
opens up an argument structure slot in which both a noun and a verb can be
used, the value “+” is assigned. If infinitives cannot be referred to by
means of a pro-noun, the finite verb gets a “-” for the thing-test. The results
of testing for that-complementation, explained in section 3.2, are summa-
rized under “that”: if the infinitive can be distanced from the finite verb by
means of a that-complementizer, the finite verb gets “+that”. If the finite
verb – in a particular sense – is restricted to infinitives, the value “-that” is
assigned. Finally, “time” summarizes the results for the test with conflict-
ing temporal specifications from section 3.3. Combinations of finite verb
and infinitive that do allow two conflicting temporal modifiers in the con-
struction get “+time” whereas combinations of finite verb and infinitive
that reject two conflicting temporal modifiers in one construction are
marked “-time”.
The thing-test, having the largest coverage (70,6% of all verbs checks
positively for this test) is listed first. Besides, this parameter gives a good
view on the direction of the integration: it makes explicit whether the in-
finitive is drawn into the finite verb’s argument structure or whether the
238 Dagmar Divjak
Table 1. Eight patterns for verbs that combine with čto/ėto (s)delat’ ‘do what/that’
1 [80] 2 [10] 3 [80] 4 [37] 5 [8] 6 [14] 7 [0] 8 [64]
P-verbs M-verbs
Although there are eight combinatory possibilities, 89% of all verbs that
combine with an infinitive find a place in columns one, three, four or eight.
These columns are the centers of gravity in this table, and I will discuss
them in turn: one and three each contain about 27% of all 293 verbs that
combine with an infinitive. Column eight collects 21.8% and column four
represents 12.6% of all verbs. On the basis of the cognitive-semantic inter-
pretations given for each parameter in the previous sections, these four
columns can be said to represent the four major degrees of verb integration
for [VFIN VINF] patterns. Column 1 verbs are strong and form [VFIN VINF] se-
quences in which two independent verbs are put in contact, whereas the
weak column 8 verbs form [VFIN VINF] patterns the two verbs of which are
characterized by the highest degree of integration. The other four columns
A binding scale for [VFIN VINF] structures in Russian 239
– two, five, six and seven – each represent less than 5% of all verbs that
combine with an infinitive.
teresting starting point for an elaborate and precise study of the meaning of
that-complementation in Russian.
In columns five and six the finite verb does not have the strength to
pull the infinitive event into its argument structure.22 This indicates a
changed balance between the two events expressed, as is claimed for col-
umn eight. Both in column five and six the events expressed are temporally
independent, but the infinitive event with column five verbs can enter into
a subordinate relation with the finite verb, whereas it cannot do so with
column six verbs. Infinitive events combined with column six finite verbs
are temporally independent and equally important: the finite verb event
cannot dominate the infinitive event, but it does not need that event to be
temporally contiguous. Table 2 in the Appendix displays some of the verbs
belonging to these columns; they contain an interactive element of thinking
or speaking that can be confined to a particular moment in time, compare
here zastavit’ ‘force’ from category five and objazat’sja ‘bind oneself,
pledge, undertake’ from category six.
In sum, the four columns in Tables (1) and (2) that contain most ele-
ments check positively or negatively for either referential inclusion and
that-complementation or that-complementation and conflicting temporal
specifications or for all three parameters.
The aim of the last section is two-fold. On the one hand, I will summa-
rize the main conclusions that can be drawn from research into the con-
structional behavior of the 293 verbs that combine with an infinitive in
Russian; on the other hand I will present typological support for the de-
grees of integration between two verbs in the [VFIN VINF] pattern and the
semantic type of verbs that displays the highest integration.
the infinitive into one of their argument structure slots. The events ex-
pressed by this type of finite verb and the infinitive are linked up to the
extent that, together, they are presented as a COMPLEX EVENT. The overall
import of the finite verb is weakened whereas at the same time, the infini-
tive is foregrounded.
The distinction in the value of the semantic import made by each verb is
stressed by Dixon (1996: 176–179) who works with Primary A and B
verbs, and Secondary verbs. Primary verbs do not need to relate to any
other verb and are always expressed as verbal lexemes. Secondary verbs,
on the other hand, “encode meanings that relate to some other verb, and
which can be expressed as verbal lexemes or as affixes or verbal modifiers
or clausal particles. (…) Semantically, the infinitives are the core concepts
of the sentences with the [secondary] finite verbs providing semantic modi-
fication.” Dixon (1996: 187–188) enumerates as “secondary items” nega-
tion, modals (such as must and can) and aspectuals (like begin, finish, but
also try and attempt), as well as the so-called wanting- and making-type
verbs. The Russian counterparts can be found in category eight that con-
tains M-verbs.
There is a critical difference between Primary and Secondary verbs:
Primary verbs – which do not need to relate to any other verb – are always
expressed as verbal lexemes. Secondary verbs encode meanings that relate
to some other verb, and there are four main ways in which languages ex-
press secondary concepts. This can be done as Secondary verbs which have
essentially the same array of derivational and inflectional possibilities as
Primary verbs, e.g. take complement clauses. Other possibilities for ex-
pressing secondary concepts are offered by verbal affixes, modifiers to a
verb (including both adverbs and modal verbs) and non-inflecting particles
within a clause (Dixon 1996: 178).
Dixon (1996: 177) argues that the finite verbs are syntactically the main
verbs, with the infinitives functioning as verbs of embedded clauses, but
claims at the same time that the infinitives are the core concepts of the
sentences with the finite verbs providing semantic modification. According
to Dixon, “these sentences have the same surface syntax but different se-
mantic interpretations. (…) Here, as at other places in the grammar, we
find one syntactic mechanism being used for tasks which are semantically
rather different”. Yet, would a language deceive its speakers? All [VFIN
VINF] sequences may look similar, i.e. consist of a finite verb and an infini-
tive, yet depending on the finite verb, they take part in fundamentally dif-
ferent argument and temporal event structures, as was revealed by the three
parameters examined above (section 3). And it is the network of construc-
244 Dagmar Divjak
tions as a whole that defines the function and meaning of an element (van
den Eynde 1995: 116ff; Croft 2001: 27–28). In other words, verbs that
belong to different networks of constructions fulfill different functions and
signal differences in meaning, even though the individual constructions
they take part in may look similar.
The difference in degree of binding and its relation to semantics is
prominent in Givón’s (2001: ch. 12) analysis of complementation. Verbs
that display the strongest degree of binding are “modality verbs”. For
Givón modality verbs code “inception and termination, persistence, suc-
cess and failure, attempt, intent, obligation or ability vis-à-vis the comple-
ment state or event”. Although I object to using the term “complement” in
this context for the reasons just mentioned, the Russian category eight
verbs in Tables (1) and (2) that show the highest degree of integration with
the infinitive, express exactly Givón’s “modality” concepts, i.e. they are
“Modal” and “Intentional”, “Tentative”, “Resultative” and “Phasal” verbs.
A third source of support for my position on the weakened syntactic and
semantic import made by the finite verb can be found in Croft (2001: 216–
220 and 254–259). Croft refers to this “non-iconic mapping of arguments”
as a grammaticalization process called “clause collapsing”: a complex sen-
tence structure with a main verb and a complement verb is being reana-
lyzed as a single clause with a tense, aspect, and/or mood indicating form
(the former main verb) and a main verb (the former complement verb).”
The infinitive is chosen as main verb because it is the Primary Information
Bearing Unit (PIBU), the semantic head or “most contentful item that most
closely profiles the same kind of thing that the whole clause constituent
profiles”. In the process of grammaticalization head status gradually shifts.
“The construction’s profile will shift towards that of the PIBU, and the
profile equivalent that is not the PIBU may expand its extension – that is,
bear even less information than it did before. At this point, the formerly
dependent structure becomes the head. If the grammaticalization process
continues, the functional elements end up losing their status as autonomous
syntactic units, becoming affixed to the lexical head. (…) This diachronic
change is a gradual process (…) for some languages, one of the last steps
in this process is the reassignment of syntactic arguments to the former
complement verb.” It is this step we currently seem to be witnessing in
Russian.
A binding scale for [VFIN VINF] structures in Russian 245
Appendix
Table 2. Productively used, stylistically neutral verbs that combine with an infini-
tive in Russian
Cat. # Thing- That- Time- Verb Translation
test test test
1 yes yes yes велетьip&pf Order
1 yes yes yes вынуждатьip- Force
вынудитьpf
1 yes yes yes (приpf)готовитьсяip Get ready, prepare
1 yes yes yes (приpf)грозитьip Threaten
1 yes yes yes допускатьip- Allow, permit
допуститьpf
1 yes yes yes (поpf)желатьip 2 Desire, want, wish
for someone else
1 yes yes yes запрещатьip- Forbid, prohibit
запретитьpf
1 yes yes yes (поpf)клястьсяip Swear, vow
1 yes yes yes обещатьip&pf Promise
1 yes yes yes обязыватьip-обязатьpf Oblige
1 yes yes yes отказыватьсяip- Refuse, denounce
отказатьсяpf
1 yes yes yes подговариватьip- Put up, instigate
подговоритьpf
1 yes yes yes позволятьip- Allow, permit,
позволитьpf afford
1 yes yes yes поручатьip-поручитьpf Guarantee, vouch
for
1 yes yes yes постановлятьip- Decide, resolve
постановитьpf
1 yes yes yes предлагатьip- Offer
предложитьpf 1
1 yes yes yes предлагатьip- Suggest, propose,
предложитьpf 2 ask, order, tell,
invite
Notes
*
This research was carried out with the financial support of the Science Foun-
dation – Flanders (Belgium). I would like to thank Agata Kochańska for her
careful comments on passages relating to Cognitive Grammar theory and the
two reviewers for a variety of useful suggestions. The usual disclaimers apply.
1. Following Pustejovsky (1991) I will use “event” as a cover term for what is
expressed by one verb, be it in Smith’s terms (1997) a “non-dynamic state” or
a “dynamic event” such as an activity, an accomplishment, an achievement or
a semelfactive event.
2. Russian has 293 verbs that combine with an infinitive, disregarding imper-
sonal verbs and verbs of motion. I have culled all verbs for which combination
with an infinitive is attested in one general, explanatory dictionary (in its two
most recent editions) and four specific verb dictionaries, i.e. Ožegov and Šve-
dova (1995 and 1999), Daum and Schenk (1992), Benson and Benson (1995),
Denisov and Morkovkin (1978) and Apresjan and Pall (1982). Table 2 in the
appendix contains the approximately 100 productively used verbs that are sty-
listically neutral; if one stem combines with several prefixes (e.g. učit’ that
takes vy-, na-, ob-, pod-), only the base verb (i.e. učit’ ‘learn’ has been in-
cluded.
3. Although the absence of a specific pattern in a sufficiently large sample of the
language can provide support for the infelicity of that pattern (Stefanowitsch
2006), research into “alternations” is generally not corpus-based. Typically,
corpora are used to provide information about which constructions are attested
for a particular verb, whereas native speaker intuitions are used to decide
whether a particular construction alternates with another construction (com-
pare Levin 1993). As a consequence, the few available approaches to auto-
matic extraction of alternations are concerned with case studies conducted on
the basis of English language data (Schulte im Walde 2007).
4. The experiment was conducted over a period of three months in the form of a
weekly interview during which each native speaker was presented with ap-
proximately 25 verbs. Several measures were taken to minimize the negative
side-effects of this set-up. Among other things, native speakers were asked
both to judge ready-made sentences and to form sentences using particular
constructional devices; these sentences were on a later occasion presented to
the participant who had constructed them as well as to other participants.
250 Dagmar Divjak
Vse utro mat’ zastavljala syna s’’ezdit’ večerom k babuške ‘all morning
mother was forcing her son to go by grandmother in the evening’. Important
here might be the fact that the finite verb has imperfective marking, i.e. has not
been taken to its natural ending.
18. My use of the term “bounded” does not imply reference to telicity.
19. Constructional data can be used to delineate semantically coherent categories
(cf. Croft 1999: 69–74; Divjak 2004, Divjak 2006, Levin 1993). If we look at
elements categorized as “1” in the first column of Table 2 in the Appendix, we
see that taking into account the details of the argument structure (i.e. preposi-
tion and case) yields subgroups that share meaning at a coarse-grained level.
For example, in category four we find verbs that express a positive or negative
attitude towards an event, e.g. bojat’sja ‘be afraid’, želat’1 ‘desire for oneself’,
izbegat’ ‘avoid’, opasat’sja1 ‘fear’, ostereč’sja ‘beware of’, stesnjat’sja ‘hesi-
tate’, stydit’sja ‘be ashamed’. For category 8 verbs, the constructional possi-
bilities of which are limited to combinations with an infinitive, there are no
conspicuous formal characteristics that can be used as the basis for further
subcategorization. Therefore, the verbs were sub-classified with the help of
data on their linear distribution, i.e. their mutual combinatorial possibilities in
verb triples. This discussion, however, exceeds the scope of this paper and the
reader is referred to Divjak (2004).
20. In Russian, construction kernels are sometimes termed predikator ‘predicator’.
Since this term is problematic in English I use the shorthand “P-verb”.
21. Adamec (1968) was the first to describe this class of verbs and termed them
modifikatory ‘modificators’. This term is fine in Russian and contrasts nicely
with predikatory ‘predicators’. Again, it is problematic in English and there-
fore a shorthand, “M-verb”, is used. Givón (1973: 894) uses the term M-verbs
(but attributes it in turn to Karttunen) to refer to “verbs requiring a sentential
complement and, in addition, requiring an equi-subject condition between the
main and complement verb”. I will restrict the term M-verb to verbs that, in
combination with an infinitive, form co-referential [VFIN VINF] constructions in
which the infinitive does not fill up one of the finite verb’s argument structure
slots, the infinitive cannot be extended to form a full-fledged complement; in
addition, clause and both verbs cannot take conflicting temporal modifiers.
22. One of the verbs that belongs to this category is ubedit’ ‘persuade’; interest-
ingly, Dal’ (1996: 575 vol. 2) does not mention the possibility of using ubedit’
‘persuade’ with an infinitive. A century ago, ubedit’ could only mean ‘force to
believe (in) something on the basis of facts, arguments, evidence’. In other
words, the verbs in this category might be loosing influence over the infinitive
they combine with, but they might also be in the process of acquiring such in-
fluence. A diachronic perspective could clarify this situation.
A binding scale for [VFIN VINF] structures in Russian 253
References
Adamec, P.
1968. K voprosu o modifikacijax (modal’nyx transformacijax) so znače-
niem neobxodimosti i vozmožnosti [On modifications (modal trans-
formations) expressing necessity and possibility]. Československa
Rusistika 13 (2): 88–94. (k VI. Mezinárodnímu sjezdu slavistů).
Apresjan, Jurij Derenikovič, and Erna Pall
1982 Russkij glagol – vengerskij glagol: upravlenie i sočetaemost’ [The
Russian Verb – the Hungarian Verb: Government and Collocation].
Budapest: Tankën’vkiado.
Atkins, B.T. Sue, and Beth Levin
1995 Building on a corpus: a linguistic and lexicographical look at some
near-synonyms. International Journal of Lexicography 8 (2): 85–
114.
Barentsen, Adriaan
s.d. The Amsterdam Corpus of Russian. Universiteit Amsterdam.
Benson, Morten, and Eveline Benson
1995 Russko-anglijskij slovar’ glagol’nyx slovosočetanij [Russian-English
Verbal Collocation Dictionary]. Moskva: Meždunarodnaja Škola
Perevodčikov.
Broccias, Christiano, and Willem B. Hollmann
(forthc) Do we need summary and sequential scanning in (Cognitive) Gram-
mar? Cogitive Linguistics 18 (4).
Chrakovskij, Vladimir S. et al.
1983 Kategorii glagola i struktura predloženija: konstrukcii s predikat-
nymi aktantami [Verbal categories and sentence structure: construc-
tions with predicative actants]. Leningrad: [s.l.].
Cristofaro, Sonia
2003 Subordination. (Oxford Studies in Typology and Linguistic Theory.)
Oxford: Oxford University Press.
Croft, William
1999 Some contributions of typology to cognitive linguistics, and vice
versa. In Cognitive Linguistics: Foundations, Scope, and Methodol-
ogy, Theo Janssen and Gisela Redeker (eds.), 61–93. (Cognitive Lin-
guistics Research 15.) Berlin/New York: Mouton de Gruyter.
2001 Radical Construction Grammar: Syntactic Theory in Typological
Perspective. (Oxford Linguistics.) Oxford: Oxford University Press.
Dal’, Vladimir I.
1996 Reprint. Tolkovyj slovar’ živogo velikorusskogo jazyka [Explanatory
Dictionary of the Great Living Russian Language]. Sankt-Peterburg:
Diamant.
254 Dagmar Divjak
Langacker, Ronald W.
1987a Nouns and verbs. Language 63 (1): 53–94.
1987b Foundations of Cognitive Grammar: Theoretical Prerequisites.
Stanford: Stanford University Press.
1991 Foundations of Cognitive Grammar: Descriptive Application. Stan-
ford: Stanford University Press.
Levin, Beth
1993 English Verb Classes and Alternations: a Preliminary Investigation.
Chicago: University of Chicago Press.
Ožegov, S.I., and N.Ju. Švedova
1995 Tolkovyj slovar’ russkogo jazyka [Explanatory Dictionary of Rus-
sian]. Moskva: Az’’.
1999 Tolkovyj slovar’ russkogo jazyka [Explanatory Dictionary of Rus-
sian]. Moskva: Azbukovnik.
Pustejovsky, James
1991 The syntax of event structure. In Lexical and Conceptual Semantics,
Beth Levin and Steven Pinker (eds.), 47–81. Cambridge MA/Oxford
UK: Blackwell Publishers.
Quine, Willard Van Orman
1964 From a Logical Point of View. Nine Logico-Philosophical Essays,
1–19. Cambridge MA: Harvard University Press.
RNC
2004 Russian National Corpus (www.ruscorpora.ru)
Schulte im Walde, Sabine
(2007) The induction of verb frames and verb classes from corpora. In Cor-
pus Linguistics. An International Handbook, Anke Lüdling, Merja
Kytö and Anthony Mc Enery (eds.). Berlin: Mouton de Gruyter.
Smessaert, Hans, Bert Cornillie, Dagmar Divjak, and Karel van den Eynde
2005 Degrees of clause integration. From endotactic to exotactic subordi-
nation in Dutch. Linguistics. An Interdisciplinary Journal of the
Language Sciences 43: 471–530.
Smith, Carlotta
1997 The Parameter of Aspect. Second Edition. (Studies in Linguistics and
Philosophy 43.) Dordrecht: Kluwer Academic Publishers.
Stefanowitsch, Anatol
2006 Negative evidence and the raw frequency fallacy. Corpus Linguistics
and Linguistic Theory 2 (1): 61–77.
Talmy, Leonard
2000 Toward a Cognitive Semantics. Vol. 1. Concept Structuring Systems.
(Language, Speech and Communication.) Cambridge MA: MIT
Press.
256 Dagmar Divjak
Wierzbicka, Anna
1975 Why ‘Kill’ Does Not Mean ‘Cause To Die’: The Semantics of Action
Sentences. Foundations of Language 13: 491–528.
1988 The Semantics of Grammar. (Studies in Language Companion Series
18.) Amsterdam/Philadelphia: John Benjamins.
The ‘impersonal’ impersonal construction in Polish.
A Cognitive Grammar analysis
Anna Słoń
Abstract
1. Introduction
For decades Polish impersonal constructions have been the subject of lin-
guistic analyses, both theoretically-neutral, descriptive in character (see
Doros 1975; Grzegorczykowa 1996; Bogusławski 1984; Wolińska 1978)
and more generatively-oriented (e.g. Jackiewicz 1992; Kubiszyn-Mędrala
1994; Węgiel 1994). The descriptive analyses yielded many an insight into
the nature and the semantics of Polish impersonals; they failed, however, to
capture the more general characteristics of impersonals, to see impersonals
258 Anna Słoń
po nim spodziewać.
after him.LOC expect.INF
‘You never know what to expect of him.’
The problem is made even more complex due to the variation within a
particular construction. A given construction may be further analysed as a
category of subconstructions resulting from the merger of the construction
with a particular type of verbal predicates, the interplay of the impersonal
construction with some other construction, or the interaction with (widely
understood) context. Consider, for instance, example (1b). Normally it is
used as a generic statement describing the accepted way of behaving or
doing something. It is possible, however, to provide a context (e.g. a
teacher showing to a pupil how to produce a particular letter shape) in
which the sentence would mean ‘This is the way how it is being written’.
If, instead of the present tense, the future tense is used, the sentence Tak się
to napisze (so MM this.ACC write.3SG.FUT) would lose its generic meaning
and be roughly equivalent to ‘This is the way how it’s going to be written’
(with a complete lack of commitment on the part of the speaker). Similarly
non-committal is the statement Tak mi się to napisało (so me.DAT MM
this.ACC write.3SG.NEUTR.PAST.PERF)‘I just happened to write it like this’,
where the dative participant is seen as experiencing the situation rather
than as being responsible for it.
The present paper analyses one particular Polish impersonal construc-
tion, namely the 3rd SG NEUTR construction and its subtypes from a Cogni-
tive Grammar point of view (Langacker 1987, 1990, 1991, 1999). Cogni-
tive Grammar (henceforth CG) can handle what has been neglected in
previous approaches and provide a fuller, more adequate analysis of Polish
impersonals, taking into account their form, their meaning and the interre-
lation between the two. In general, CG, which treats meaning as conceptu-
alisation and claims that every morpheme and the sequence thereof is
260 Anna Słoń
b. Widziałeś tę parę na
saw.2SG.MASC.PAST.IMPERF [this couple].FEM.ACC on
ławce? Ona/*Ø kiedyś u nas
bench.FEM.LOC she.NOM/Ø once at us.GEN
pracowała, a on/*Ø to
work.3SG.FEM.PAST and he.NOM/Ø this.NEUTR.NOM
były poseł.
has-been.SG.MASC.NOM MP.MASC.NOM
‘Have you seen the couple on the bench? She used to work
with us and he’s an ex-MP.’
c. ‘Przepraszam, poniosło
apologise.1SG.PRES carry-away.3SG.NEUTR.PAST.PERF
mnie.’ ‘Co niby cię
me.ACC what.NOM like you.ACC
poniosło?’ ‘???’
carry-away.3SG.NEUTR.PAST.PERF ‘???’
‘Sorry, I got carried away.’ ‘You got carried away by
what?’ ‘???’
When the trajector6 is not elaborated due to its referent’s high accessi-
bility and/or low attention-worthiness, the referent’s identity is nonetheless
recoverable from the context. This is illustrated in (3a), where one of the
interlocutors uses an unelaborated trajector assuming that its referent’s
identity is known to the other interlocutor. If it turns out that this is not the
case, the full nominal can be provided. A partially similar case is exempli-
fied in (3b). Here, however, the choice of referents is limited contextually,
and the expression itself is highly conventionalized. In contrast, (3c) illus-
trates a situation where the trajector cannot be elaborated because the iden-
tity of the trajector’s referent is difficult or even impossible to pin down. It
is this last type of trajector non-elaboration, and partly the trajector non-
elaboration accompanying the weather verbs, like the one in (3b), that oc-
curs in impersonal constructions.7
In sum, the absence of the subject nominal in Polish is not sufficient to
make a sentence impersonal. In impersonal constructions, like those in (1)
The ‘impersonal’ impersonal construction in Polish 263
3. The analysis
In (6a) the only profiled element of the event is the process alone, while
in (6b), the profile encompasses not only the process but also quite a non-
prototypical instigator which is at the same time part of the setting and a
reification of the process. The English example in (6c) shows another con-
ceptual strategy of handling a diffuse, setting-like instigator, and is in-
cluded here for the sake of comparison with the Polish examples. English
it, providing an obligatory subject in weather expressions is treated by
Langacker (1991) as profiling a default setting. The imagery evoked by the
three sentences in (6) is shown schematically in Figure 1:
a Inst
setting b Inst
setting
tr
c Inst
setting
tr
To conclude this section, let us point out that the availability of the
cognate-subject construction is limited by morphological considerations
The ‘impersonal’ impersonal construction in Polish 267
and is thus language specific. The dawn is dawning example has as its Pol-
ish counterpart the sentence ?Świt świta, but Rain is raining or Snow is
snowing do not have such counterparts because Polish uses the verb padać
‘fall’, which is morphologically unrelated to Polish nouns for ‘rain’ deszcz
or ‘snow’ śnieg. The only available Polish version is Pada deszcz/ śnieg
‘The rain/ snow is falling’.
Sentence (4b), repeated below as (7a) illustrates another subtype of the 3rd
15
SG NEUTR construction. Here the instigator is likewise defocused, but the
instrument is elaborated, which opens up a possibility of treating it as an
instigator, as in sentence (7b). When an ‘additional’ participant is intro-
duced, it will be viewed as the wielder of the instrument,16 as shown in
(7c):
The three construals are presented in Figure 2. In sentence (7a) and Figure
2a, the only elaborated participants are the patient and the instrument with
the instigator left obligatorily unelaborated. The 3rd SG NEUTR form of the
verb implicitly agrees with a diffuse, unidentified force. It seems likely that
when the causal element is difficult to identify and by virtue of its diffuse-
ness ‘occupies’ a considerable portion of the setting, it may be metonymi-
cally equated with the setting itself.17 The shading of the instigator and the
268 Anna Słoń
a setting IS
tr lm2 lm1
b Inst IS
Instr Pat
tr lm
c IS
tr lm2 lm1
Figure 2. a. The 3rd sg neutr impersonal construction – the instigator remains obli-
gatorily unelaborated; b. The narrowing of the immediate scope of predi-
cation resulting in the conferring of the trajector–instigator status on the
instrument; c. The personal, instigator–subject construction.
b. *Coś go świerzbiło
something.NOM him.ACC itch.3SG.NEUTR.PAST.IMPERF
by jej o tym powiedzieć.
to her.DAT about it.LOC tell.PERF.INF
c. Język go świerzbił
tongue.MASC.NOM him.ACC itch.3SG.NEUTR.PAST.IMPERF
by jej o tym powiedzieć.
to her.DAT about it.LOC tell.PERF.INF
‘He felt an urge/ was itching to tell her about it.’
(11) a. Świerzbiło ją by mu
itch.3SG.NEUTR.PAST.IMPERF her.ACC to him.DAT
przyłożyć.
beat.INF
‘She was itching/ felt an itch to box his ears.’
b. *Coś ją świerzbiło
something.NOM her.ACC itch.3SG.NEUTR.PAST.IMPERF
by mu przyłożyć.
to him.DAT beat.INF
c. Ręka ją świerzbiła
hand.FEM.NOM her.ACC itch.3SG.FEM.PAST.IMPERF
by mu przyłożyć.
to him.DAT beat.INF
‘Her hand was itching to box his ears.’
In example (9) with the verb kusić ‘tempt’, the temptation is neither
seen as coming entirely from outside, nor entirely from inside the experi-
encer participant; thus the situation can be conceptualised both ways. Sen-
tence (9a) suggests a more internal source of temptation, sentence (9c),
with its specified instigator a more external one, while (9b) is ambiguous
between the two, although it is more likely to imply an external source.
The verb korcić ‘tempt, itch’, does not allow the external interpretation
with a specified instigator – see (9c), but it does allow alluding to some
external force, as in (9b). The verb świerzbić resembles swędzieć (in their
physiological sense they are near synonyms, meaning ‘itch’), but in exam-
ples (10)–(11) the former is used metaphorically to render compulsion. The
source of the compulsion cannot be identified linguistically: sentences
(10b) and (11b) with coś are odd, whereas sentences (10c) and (11c) are
272 Anna Słoń
natural, though they do not specify the instigator, but like (8c), feature an
active zone within the experiencer participant. Note the acceptability of
(8b) with the indefinite subject: here, however, the meaning is not that
some unspecified force was making someone feel an itch, but that someone
felt an itching in some unspecified place in his body. With (10b) and (11b)
such an interpretation is impossible due to the idiomaticity of expressions
in (10c) and (11c).
Figure 3 illustrates the contrasting construals for the impersonal vari-
ants in (8a), (9a), (10a) and (11a), versus the active zone personal variants
in (8c), (10c) and (11c):
a Exp b
Exp
AZ
tr
lm lm
In Figure 3a, the experiencer constitutes the overall location for (or par-
ticipant affected by) the event whose instigator is not specified. Because it
is sometimes unclear whether the instigating force originates inside or out-
side the experiencer, the part of the arrow that reaches beyond the experi-
encer is drawn with a dashed line. In Figure 3b the trajector is elaborated,
but it does not specify the instigator. Instead it specifies an active zone (cf.
Langacker 1990, 1991) within the overall experiencer–location, i.e. the
active zone which is directly affected by the event.
A related subgroup of the 3rd SG NEUTR construction is exemplified in
sentence (4d), repeated below as (12a). In this subconstruction there is a
linking verb, most typically być ‘be’ or (z)robić się ‘grow, get, become’,
which is complemented with an adverb describing the state of a setting.21
Examples of such adverbs are: gorąco ‘hot’, ciepło ‘warm’, chłodno ‘cool’,
zimno ‘cold’, mroźno ‘freezing’, etc.; duszno ‘stuffy’, parno ‘muggy’,
wilgotno ‘humid’; ciemno ‘dark’, jasno ‘bright’; deszczowo ‘rainy’, bur-
zowo ‘stormy’, pochmurno ‘cloudy’; ładnie ‘nice’, brzydko ‘ugly’, ponuro
‘gloomy’. The construction is undoubtedly impersonal, because the trajec-
The ‘impersonal’ impersonal construction in Polish 273
The setting, however, is not the only event element that can be elabo-
rated in this construction. Instead of referring to the state of the setting, one
can refer to the state experienced by a sentient participant, as shown in
(13a). The construal with an experiencer comes close to that triggered by
(8), and like in (8) it is possible to specify the active zone, as is done in
(13b):
a b
c Exp d Exp
AZ
lm lm
cekinami.
sequin.PL.NON-VIR.INSTR
‘The dress was glittering with sequins.’
The ‘impersonal’ impersonal construction in Polish 275
po ogrodzie.
on/across garden.MASC.LOC
‘Bees were swarming in the garden.’
The sentences in (18) are impersonal. The verb does not show agree-
ment with any element in the sentence; it has the form of 3rd person singu-
lar neuter. The construal offered by those sentences is that of focusing on
the process alone, with the actual instigators of the process and the set-
ting/location made less prominent. The three construals, i.e. those associ-
ated with the instigator-subject variant, setting-subject variant and the im-
personal variant, are shown in Figure 5:
a tr tr setting b setting
tr tr tr
tr tr tr
tr tr tr
tr tr tr lm
lm tr
c setting
lm
The situation in (19) differs form that in (14a–b) and (18a–b) because
kawa ‘coffee’ is not an instigator in the sense that bees or sequins are. No-
tice that the presence of coffee is not necessary at all; it is the smell of
coffee that is crucial. This is revealed by the awkwardness of the sentence
in (20): there coffee is treated as an instigator, so the contribution of the
locative phrase would be to suggest that the instigator had been distributed
all over it. In other words it would be coffee as a substance distributed all
over the room not its smell itself. Such an interpretation is evidently feasi-
ble but much less likely.
To explain the motivation for the use of the 3rd person singular neuter
for of the verb in (21c–d) as compared to the verb forms showing agree-
ment in (21a–b), we need to look at the contexts in which the two negated
constructions are used. Sentence (21b) – the personal negated locative con-
struction is used when the absence is viewed from the perspective of the
participant, while (21c) – the impersonal negated locative construction –
when the absence is viewed from the perspective of the location. Thus
Maria nie była dzisiaj w pracy bo ma urlop ‘Mary.NOM hasn’t been at
work today because she has a day off’ is natural when spoken at home by a
member of Mary’s family, for instance, whereas Marii nie było dzisiaj w
pracy bo ma urlop ‘Mary.GEN hasn’t been at work today because she has a
day off’ is natural when spoken at work by one of her colleagues. Similarly
The ‘impersonal’ impersonal construction in Polish 279
Notice also that the preferred word order is the locative expression first.
Therefore the fact that the situation is viewed from the perspective of the
location motivates the use of the 3rd person singular neuter verb form.
The constructions analysed in this section show that the 3rd person sin-
gular neuter form of the verb is used when there is no participant elabo-
rated by a nominal in the nominative case and when the setting/location (if
elaborated) is elaborated by a prepositional phrase (or such forms as tutaj
‘here’, wszędzie ‘everywhere’, etc.). Thus the 3rd person singular neuter
form of the verb is evidently associated with the non-reified set-
ting/location of the event/state.
5. Conclusion
Notes
1. The data in the paper are in most cases made-up examples, based, however,
on real-life usage.
2. MM stands for “middle marker”. I follow here Kemmer, for whom “[a] middle
marker can be provisionally defined as a language specific morphosyntactic
marker that appears in the expression of some cluster of distinct situation
types (…) that are hypothesised to be semantically related to one another and
to fall within the semantic category of middle voice” (1993: 15). The Polish
middle marker się appears in the reflexive, reciprocal and related situation
types. By extension, it is also a marker of impersonality, as in (1b) (see Słoń
2001, 2003).
3. The term “instigator” used here refers to a causal factor that is not necessarily
human. For this reason the term is preferred over the term “agent”, especially
because the 3rd SG NEUTR construction is typically used to defocus non-human
instigators.
4. Attention-worthiness is related to newness of a particular item of information,
but it is not equivalent to it. While new information is clearly attention-
worthy, attention-worthiness may also refer to presenting an already men-
tioned item of information in a ‘new light’, so to say. In this interpretation,
topics are not attention-worthy as they constitute the background that contains
old information, necessary to keep the text coherent.
5. Pronouns can also be used for emphasis, which can be understood as an in-
stance of contrast with an element unexpressed linguistically.
6. Trajector in CG is defined as “the (primary) figure within a profiled relation”
(Langacker 1991: 555), while subject is “a nominal that elaborates the trajec-
tor of the process profiled at the clausal level of organisation. Its profile is
thus the primary clausal figure” (Langacker 1991: 554). It is necessary not to
confuse those two terms because the absence of subject does not imply the ab-
sence of trajector.
7. It must be noted that there is no sharp boundary between cases where the
identity of the trajector’s referent is fully recoverable despite the trajector’s
non-elaboration and cases where the identity of the trajector’s referent is so
diffuse that the trajector cannot really be elaborated at all. These two kinds of
situations constitute merely the endpoints of a continuum with a wide range of
The ‘impersonal’ impersonal construction in Polish 281
intermediate cases. Thus, the greater the degree of irrecoverability of the iden-
tity of the trajector’s referent and/or the degree of conventionalisation of sub-
ject omission, the greater the likelihood of classifying the construction/ occur-
rence as impersonal. The picture gets more complex when context is taken
into account, because what is an instance of an impersonal construction in one
context, may be treated as an instance of zero anaphora in another.
8. Gender distinctions in Polish occur in the past tense and in the periphrastic
future tense with the so-called l-forms of content verbs. They do not occur in
the present tense, in the simple future tense and in the periphrastic future tense
with content verbs in the infinitive. This is illustrated by the sentences below.
Past tense singular:
(i) Jan (na)pisał list.
John.MASC.NOM write.3SG.MASC.PAST.(PERF)/IMPERF letter.MASC.ACC
‘John wrote a letter.’
15. Doros (1975: 69) comments that this subconstruction is used in dialects or
literature rather than in everyday language and notes its lower frequency as
compared to Russian. In both languages the subconstruction is used with ref-
erence to forces of nature.
16. The identity of the instrument’s wielder can be predicted form the type of
process coded by the verb and the type of the instrument’s referent.
17. In this sense the 3rd person singular neuter marking of the verb may be consid-
ered to agree with the setting as Smith (1994) suggests for Russian imperson-
als.
18. Probably the original function of Polish się is a reflexive voice marker. In this
use it signals the co-reference of the direct object’s and subject’s referents, as
in (i):
References
Ariel, Mira
2000 The Development of Person Agreement Markers: From Pronoun to
Higher Accessibility Markers. In Usage-Based Models of Language,
Michael Barlow and Suzanne Kemmer (eds.), 197–260. Stanford:
CSLI Publications.
Bogusławski, Andrzej
1984 Polskie nieidentyfikacyjne wyrażenia osobowo-referencjalne [Polish
non-identifying personal-referential expressions]. Polonica 10: 49–
71.
Doros, Aleksander
1975 Werbalne konstrukcje bezosobowe w języku rosyjskim i polskim na
tle innych języków słowiańskich [Verbal impersonal constructions in
Russian and Polish as compared to other Slavic languages]. Prace
Komisji Słowianoznawstwa nr 32. Ossolineum.
Fauconnier, Gilles
1994 Mental Spaces: Aspects of Meaning Construction in Natural Lan-
guage. Cambridge: Cambridge University Press.
Grzegorczykowa, Renata
1996 Reguły deagentywizacji w polszczyźnie [Rules of deagentivisation in
Polish]. In Jazyk a jeho užívání, sborník k jubileu O. Uličneho [Lan-
guage and its use. Festschrift to O. Uličny], 58–65. Praha.
Jackiewicz, Alina
1992 Morphological Exponence of Impersonality with Special Reference
to Polish -no, -to Constructions and Their English and German
Equivalents. Linguistica Silesiana 14: 131–138.
Kemmer, Suzanne
1993 The Middle Voice. Amsterdam/Philadelphia: John Benjamins.
Kubiszyn-Mędrala, Zofia
1994 Problematyka konstrukcji bezmianownikowych w gramatyce genera-
tywno-transformacyjnej (na wybranych przykładach) [Problems of
nominativeless constructions in generative-transformative grammar
(on selected examples)]. Polonica 16: 163–179.
286 Anna Słoń
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal about
the Mind. Chicago: University of Chicago Press.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequi-
sites. Stanford: Stanford University Press.
1990 Concept, Image, and Symbol: The Cognitive Basis of Grammar
(Cognitive Linguistics Research 1). Berlin: Mouton de Gruyter.
1991 Foundations of Cognitive Grammar. Vol. 2: Descriptive Application.
Stanford: Stanford University Press.
1999 Grammar and Conceptualization. (Cognitive Linguistics Research
14.) Berlin/New York: Mouton de Gruyter.
Słoń, Anna
1999 A radial category of Polish się-constructions. Differences in con-
strual. In PASE Papers in Language Studies. Proceedings of the sev-
enth annual conference of the Polish Association for the Study of
English, Szczyrk, May 1998, Janusz Arabski (ed.), 151–166.
Katowice: Para.
2001 Konstrukcje nieosobowe a informacja pragmatyczna [Impersonal
construction and pragmatic information]. In Językoznawstwo
kognitywne II: Zjawiska pragmatyczne [Cognitive linguistics II:
pragmatic phenomena], Wojciech Kubiński and Danuta Stanulewicz
(eds.), 248–268. Gdańsk: Gdańsk University Press.
2002a Instigator Defocusing in Polish and English. A Cognitive Grammar
Analysis. In Proceedings of the 10th Annual Conference of the Pol-
ish Association for the Study of English. PASE Papers in Linguistics,
Translation and TEFL Methodology, Elżbieta Mańczak-Wohlfeld
(ed.), 81–93. Cracow: Jagiellonian University Press.
2002b Shifts in the relative prominence of participants, locations and set-
tings: Two negative constructions in Polish. In PASE Papers in Lan-
guage Studies: Proceedings of the Ninth Annual Conference of the
Polish Association for the Study of English. Gdańsk, 26–28 April
2000, Danuta Stanulewicz (ed.), 249–258. Gdańsk: Gdańsk Univer-
sity Press.
2003 Impersonal Constructions in English and Polish. A Cognitive
Grammar Approach. Ph.D. diss., Humanities Department, Maria Cu-
rie-Skłodowska University, Lublin.
Smith, Michael B.
1994 Agreement and Iconicity in Russian Impersonal Constructions. Cog-
nitive Linguistics 5 (1): 5–56.
The ‘impersonal’ impersonal construction in Polish 287
Tabakowska, Elżbieta
2004 On Grammatical (Cross)Categorisation: Nominal Elements in Polish
Verbs – paper presented at Perspectives on Slavistics conference,
Leuven, 17th–19th September 2004.
Węgiel, Maria
1994 Czy w gramatyce transformacyjnej potrzebna jest reguła S Æ VP?
(Kwestia tzw. Czasowników zeroargumentowych) [Is there a need
for the S Æ VP rule in transformative grammar? (The problem of the
so-called zero-argument verbs)]. Polonica 16: 155–162.
Wolińska, Olga
1978 Konstrukcje bezmianownikowe we współczesnej polszczyźnie
[Nominativeless constructions in Modern Polish]. Katowice: Univer-
sity of Silesia Press.
ȱȱ
ȱDZȱ¢ȱȱ
A Frame Semantic account of morphosemantic
change: the case of Old Czech věřící
Mirjam Fried
Abstract
1. Introduction*
The Slavic field has always maintained strong interest in the study of lan-
guage change, and Slavic linguists have devoted substantial energy to de-
292 Mirjam Fried
The -NT- stem (i.e. present active participle) is derived from the present
(as opposed to past/infinitival) stem by adding the suffix -c, while the pre-
A Frame Semantic account of morphosemantic change 293
sent stem is formed by adding a thematic vowel to the root. The OCz ver-
bal system consisted of six conjugation classes and the thematic vowel was
-ú everywhere except in the i-stems, such as věřiti, which were marked
with -ie, as shown in (1). In addition, there was a highly productive sub-
class (the type kup-uj- ‘buy’) that represents an innovation in that it is
based on explicit aspectual morphology, deriving imperfective stems from
inherently perfective roots by adding -uj before the present stem suffix.
These patterns are summarized in the left portion of Table 1. The last two
columns of Table 1 illustrate cases in which the PAs deviate from the regu-
lar morphological template; following the terminology in Marvan et al.
(1963), I refer to those forms as pseudo-PAs. Their irregularity consists in
containing stem extensions (-ú, -ujú/-ijí) that belong to different morpho-
logical classes and as we can see, the verb věřiti involves both types of
irregular formations: the -ú suffix in the thematic vowel slot (deriving
věřúc-í instead of the expected věřiec-í) as well as the sequence -uj-ú- (de-
riving věřijící). It is also worth pointing out that pseudo-PAs often (though
not always) had a distinctly shifted meaning from the regular PAs, some-
times involving metaphorical shifts but by no means limited to those (cf.
Marvan et al. 1963; Fried 2003, 2005b).2
sound change (first ú > í and later ie > í), the result of which was, in both
cases, spelled i and y in different texts. Consequently, we cannot be always
sure what the orthography in a given token represents etymologically (-ie-
vs. -ú-) after a certain point on the timeline.3 In contrast, the form věřijící is
a straightforward reflex of the u > i fronting (< věřujúcí).
The PAs offer valuable material that speaks to a number of questions of
general interest: the developmental potential of the hybrid form itself,
given the inherent tension between its internally verbal but externally
nominal character; the role of the communicative context in the PA’s inter-
pretation and function; the relative (in)stability in maintaining particular
form-meaning associations; and the role of collocations and fixed expres-
sions in semantic change. This case study focuses mainly on the first two
questions, by examining in detail the semantic relationship between the
polysemous verb věřiti and its corresponding PAs. However, the PAs’ se-
mantic development is also inextricably linked with their place and func-
tion in larger grammatical patterns and correlates with their textual distri-
bution as well. Only a systematic analysis of these interconnected
dimensions can lead to a comprehensive picture of all the factors that col-
lectively motivate the gradual shifts in the forms’ meaning and categorial
status, starting with a richly polysemous structure that spans several se-
mantic domains and three functional domains (referential, attributive,
predicative), and leads to the outcome we know from MCz: věřící as an
actor noun (‘practicing Christian’) and marginally as an adjective with the
same meaning. The analysis will demonstrate that the observed changes
can be best described in terms of Hopper and Thompson’s 1984 idea of the
promotion and demotion of particular aspects of meaning; in the case of
PAs these aspects have to do with the relative weight of their verbal and
nominal properties.
The nature of the data and the analytic goals – namely, capturing the
speakers’ knowledge of the meaning of a complex and functionally some-
what fluid morphological category – support the view that linguistic cate-
gories are best understood as functional prototypes (e.g. Daneš 1966; Hop-
per and Thompson 1985; Croft 1991, 2001) and call for an approach that
allows us (i) to represent meaning in terms of particular conventionalized
patterns of understanding, in which lexical meaning, syntactic function,
and communicative function form an integrated whole, and (ii) to structure
these patterns in a polysemy network that captures their mutual relatedness
and motivations for shifts. I will, therefore, appeal to the conceptual and
representational apparatus of the grammatical model known as Construc-
tion Grammar (esp. Fillmore 1989; Croft 2001; Fried and Östman 2004)
A Frame Semantic account of morphosemantic change 295
and its own theory of meaning, Frame Semantics (Fillmore 1982; Fillmore
and Atkins 1992; Atkins 1994; Atkins et al. 2003; Fillmore et al. 2003;
Fried and Östman 2003; Fried 2004, 2005a).4
In Frame Semantics, linguistically relevant semantic information is or-
ganized and structured in “interpretive frames” (Fillmore 1982: 122),
which represent the complete background scene associated with a given
linguistic expression: the scene’s participants, settings, and any other
unique semantic features (collectively referred to as “frame elements”) that
are necessary for speakers’ native understanding of what the lexical item
means and how it can be used in context. In the case of predicates, i.e. ar-
gument-taking lexemes, the frame also carries information about the con-
ventional expression of the syntactically relevant participants as they mani-
fest themselves in the syntactic organization of sentences. A single
linguistic expression may be (and often is) associated with multiple frames
and, conversely, a single frame may be shared by multiple expressions;
each such expression, then, represents a particular conceptualization of
certain parts of the larger background scene. Both aspects of frame sharing
are among the factors that have shaped the diachronic path of the PA
věřící.
In addition, the PA provides a good test case for one of the basic tenets
of Construction Grammar, which holds that the meaning or function of a
grammatical pattern is not just a sum of its parts, derivable compositionally
from the properties of its constituents. The PA has a particular meaning at
the word-level, as a symbolic sign that contributes semantic content to a
larger syntactic pattern it occurs in; at the same time, it is internally com-
plex, consisting of a sequence of inflectional morphemes attached to a
lexical root. This arrangement necessarily raises the question of how the
two dimensions are related: Is the overall meaning of the PA predictable
from its morphosemantic structure, or is it non-compositional, and in what
ways? The present analysis will show that the PA is indeed best treated as
a morphological construction as it is understood in Construction Grammar:
a conventionalized association between a complex form and its meaning.
Taking a diachronic perspective enhances our insight into how such a con-
ventionalization may arise.
The analysis is based on an extensive corpus of authentic data excerpted
manually from OCz texts that stretch across more than 300 years and pro-
vide a representative sample of genres (historical, biblical, administrative,
expository, poetry, popular entertainment, correspondence, drama, etc.).
My corpus contains 121 tokens of the PAs derived from věřiti and they
come from over 50 different texts. I excerpted about 74 different texts in
296 Mirjam Fried
their entirety, which has yielded more than 55% of the PA tokens in the
corpus. The remaining 45% come from about 120 additional texts (about
one quarter of them biblical) and were collected more or less at random
from the OCz archive at ÚJČ in Prague. The corpus contains more than
1200 tokens of PAs, which represent over 240 different verb roots, and
about one third of those roots show both a regular form and at least one
pseudo-PA. All three PA variants derived from the verb věřiti are amply
attested throughout the OCz period; this is not surprising, given the mean-
ings the verb expresses: as we shall see, they all have to do with salient
events of the speakers’ daily life. This fact together with the PAs’ rela-
tively high frequency across very diverse manuscripts is evidence that they
were well entrenched in the vernacular.
The paper is organized as follows. Section 2 summarizes the semantic
structure of the verb věřiti, establishing a basis for analyzing the meanings
of the PAs. Section 3 presents the details of the PA uses both as adnominal
modifiers and as syntactic nouns, incorporating them into the network of
frames associated with the verb root. Section 4 provides the overall dia-
chronic structure of the PA, including its irregular forms. Section 5 briefly
concludes the paper.
The verb věřiti was used in several senses, which I label as BELIEF, FAITH,
CREDIT, and TRUST. They all can be described as having to do with the con-
cept of belief (accepting something as truthful or reliable) but they repre-
sent four distinct conceptualizations, which are manifested by different
complementation patterns and are motivated by the contexts in which they
are used.
The conceptually simplest scenario (BELIEF) is illustrated in (2). It pre-
supposes a person who believes (Believer) and a person who receives the
Believer’s trust (Recipient of Trust) with respect to some Content. In OCz,
the conventional way of encoding this semantic structure was to mark the
Believer in the nominative case, the Recipient in the dative, and the Con-
tent, which may or may not be explicitly mentioned, in various preposi-
tional phrases, often a locative – ‘to believe somebody (in/about some mat-
ter)’.
A Frame Semantic account of morphosemantic change 297
belief are fused into a single entity, which is conventionally marked by the
(v +) Acc phrase. FAITH thus expresses a more restricted instance of the
concept of believing.
The examples in (4) illustrate the CREDIT sense, where the concept of be-
lieving somebody’s word is set in the context of a commercial transaction,
whereby one party (seller) provides a desired commodity to another
(buyer), in exchange for a payment. But the scene expressed by věřiti in the
commercial sense is more complex than that since the payment presup-
posed by a commercial transaction is necessarily missing here.
sense as well. The TRUST sense is exemplified in (5) and we note that it still
has three arguments (nominative-, dative-, and genitive-marked).
Here the Recipient of Trust plays the role of a caretaker who is en-
trusted with a valued object and the Believer is thereby cast in the role of a
Dependent: somebody who depends on the trustworthiness of the care-
taker. In this respect, the TRUST sense shares the semantic shift that is pre-
sent in the CREDIT sense in highlighting the belief as expectation for an
action, rather than relating simply to the content of somebody’s thoughts.
In the oldest examples, this difference in meaning between BELIEF and
TRUST is consistently signaled by different case marking (Nom-Dat-Loc vs.
Nom-Dat-Gen), and the subsequent development only strengthens the se-
mantic contrast. In the BELIEF sense, the formal alternatives for expressing
the Content role are gradually expanded to a content clause introduced by
že ‘that’ (6a), whereas the corresponding role in the TRUST sense (Valued
Entity) is attested in the form of a purpose clause introduced by aby ‘in
order to/so that’ (6b).
300 Mirjam Fried
CREDIT
FAITH
GEN
TRUST
aby-S
(ex. 3)
cat v lxm věřiti
cat v lxm věřiti
Frame TRUST
(belief as an expectation) Frame FAITH
FEs: #1 Dependent (content of thought)
#2 Caretaker FEs: #1 Believer
#3 Valued-Entity #2 Deity
val {#1 [Nom], #2 [Dat], #3 [Gen] } val {#1 [Nom], #2 [Acc(v) ] }
(ex. 5, 6b)
[aby-S] cat v lxm věřiti
Frame BELIEF (content of thought)
FEs: #1 Believer
#2 Recipient-of-Trust
(ex. 4) #3 Content
cat v lxm věřiti val {#1 [Nom], #2 [Dat], (#3 [Locv]) }
In the next section, we will examine how the verb meaning manifests it-
self in the PA form, focusing on the following questions: Which of the
meanings occur? Are there any correlations between the verb meanings and
the categorial status of the PA? To what extent can we detect semantic
shifts vis-à-vis a compositional interpretation of the PAs? And, finally,
does the regular vs. irregular morphology of the participial stem play a role
in any semantic distinctions?
A Frame Semantic account of morphosemantic change 303
3. PA functions
The PA occurs in all three syntactic environments we find with other OCz
PAs: in an adnominal position (7); as a syntactic noun, in (8) shown as a
dative-marked argument with a possessive modifier; and in a copular struc-
ture with být ‘be’(9). In the space of this paper, I can only address the
nominal and adnominal patterns, which, however, represent the vast major-
ity of attested uses (111 tokens out of the total of 121).
As already noted, the PAs combine properties that predispose them to-
ward both verbal and non-verbal behavior. Their verbal root creates the
potential for expressing verbal valence; the -NT- morphology explicitly
marks voice of the participial stem (encoding its inherently active orienta-
tion, cf. Haspelmath 1996) and relative tense, expressing an eventuality
that is temporally delimited by the main predicate (in the sense of contem-
poraneousness; cf. also Schultze-Berndt and Himmelmann 2004).10 At the
same time, the nominal inflection (CNG) makes the PAs externally non-
verbal entities: functionally they could be actor nouns (capitalizing on the
304 Mirjam Fried
potential of the PA as a ‘definite participle’, i.e. ‘the one V-ing’), and mor-
phologically they are unambiguously adjectives. By strengthening or sup-
pressing any of these a priori available features (nominal, adjectival, ver-
bal), a given PA can potentially develop any of the corresponding
functions: reference, modification, or predication. However, each function
displays a distinct preference with respect to a particular root meaning;
Table 2, as a preview of the subsequent discussion, summarizes the relative
frequencies. The number in parentheses refers to the purpose meaning,
shown in (12) below, which may or may not be an instance of PA, and the
copular uses (bottom row) are added for the sake of completeness only;
they require further commentary that is beyond the scope of this paper. It
must be stressed, though, that these numbers can only be used for their
orientational value. Since we are dealing with an incompletely attested
language statistical references are only useful as indicators of general ten-
dencies, not in any absolute terms.
Table 2. Relative frequencies of věřící in the corpus vis-à-vis the verb senses
BELIEF FAITH CREDIT TRUST
Adnominal PAs 6 (+ 26?) 13 -- --
PAs as syntactic nouns -- 46 20 --
PAs in (copular) predicates 6 4 -- --
The expected structure and the most transparent meaning of the adnominal
PA would be a combination of a noun expressing the ‘agent’ of the PA
(Believer) and the PA as an adnominal modifier, as shown in (7), here re-
peated as (10); for easier identification of the syntactic structure, the PA
with its complements, if present, will always be enclosed in brackets <>:
(10) lid <v buoh věřící > ‘(the) people believing in God’
The adnominal PAs occur only in the BELIEF and FAITH senses. In the BE-
LIEF meaning, the PA is always bare and appears to be older. Interestingly,
even the earliest example, given in (11), shows a clearly shifted meaning:
the modified entity is not the Believer, as would be expected, but the Con-
tent, resulting in a distinctly passive/modal reading: ‘believable’, rather
than ‘believing’; structurally, the PA in (11) is a resultative secondary
predicate. Note also that this token must be classified as a pseudo-PA, and
not just by virtue of its age (the change ie > í had not taken place yet): the
manuscript spelling (-úcí) sometimes indicates the irregularity directly, as
in (11).
This usage is significant for two reasons. First, it follows the pattern
found with many pseudo-PAs, namely, a striking semantic shift; in this
case the meaning reflects a reconfiguration of the semantic participants
contributed by the verbal stem. In particular, the reconfiguration can be
attributed to the loss of the verbal category of voice, whereby the PA be-
comes fully dissociated from its active orientation signaled by the -NT-
morphology: the noun modified by the PA no longer bears the expected
agent relation to the event expressed by the root. All this could be taken as
evidence that the sequence -ú-c-í found in this pseudo-PA and imported
from another conjugation class had become a (partially?) frozen unit with-
out a transparent internal structure (Gebauer 1958: 89–92) and was at-
306 Mirjam Fried
tached to the root věř- as a single (derivational) suffix -úcí (we will have to
qualify this conclusion somewhat in section 4).11
The second point of interest is the kind of nouns this pseudo-PA collo-
cates with: upravenie ‘testimony’ in (11), svědek ‘witness’, etc. The gen-
eral context of testifying is common to all the other adnominal examples of
věřící in the BELIEF sense, whether used in the modal meaning ‘believable’
or the purpose meaning to be discussed below. It suggests that, unlike the
verb věřiti, which occurred in all kinds of contexts in which believing the
truthfulness of something was at issue, the adnominal PA was used more
narrowly and in a more specialized sense, largely restricted to situations
which require or presuppose the act of witnessing as a precondition for the
belief.
The PAs in the BELIEF sense appear to be fairly marginal. The modal
meaning (3 tokens in my corpus) disappears by the mid- to late 1400s, but
despite the small number of examples in the corpus, we cannot dismiss
these pseudo-PAs simply as scribal errors or a quirk of an individual
author, since the modal (and other non-active) interpretations are amply
documented by PAs formed from other verbs (Fried 2005b, forthcoming-b)
and the form věřící thus fits a more general pattern. What is perhaps more
surprising is the fact that the active meaning, in the habitual sense of
‘(prone to) believing’, seems to be equally rare, attested from about 1400
on (3 tokens). It is completely marginal in MCz as well (no tokens in the
800,000-word spoken Czech corpus, PMK, and only 20 in the
100,000,000-word corpus of written Czech, SYN2000), although it is of
course possible to form the PA; its formation remains morphologically
fully productive, as expected for an inflectional form. I will return to its
low incidence in actual texts at the end of the paper.
The corpus also contains a relatively large number of examples (26 to-
kens) involving one particular collocation: with the noun list ‘letter’, such
as in (12); for reasons to be addressed in a moment, I present this PA in the
actual OCz spelling.
It is not clear, though, that the forms in (12) are necessarily PAs. Since
the spelling in the manuscripts does not indicate length, we can only guess
whether the form in question represents a PA (long í, etymologically either
< ú or < ie), or a specific deverbal adjective dedicated to expressing pur-
pose and derived from the infitival stem + cí (in which case the vowel pre-
ceding -c- would be short), as shown in (14b):
308 Mirjam Fried
Both analyses are plausible. What speaks in favor of (14b) is the pur-
pose meaning. On the other hand, the analytic uncertainty is peculiar to the
i-stems only since, without marking the length orthographically, we cannot
easily tell the difference between the present and infinitival stems (věří- vs.
věři-, respectively). With other stems, however, the distinction is unambi-
guous (cf. laj-ú-c-íPA vs. lá-cí from lá-ti ‘to reprimand’) and there is plenty
of evidence that the morphologically indisputable PAs were quite com-
monly used in the purpose reading as well (cf. Michálek 1963; Fried
2005b).
We can leave this question concerning věřící/věřicí open since nothing
in the overall account of the PA development hinges on resolving it. Let us
simply note that it is possible that the BELIEF sense of věřiti might have
been extended in the adnominally used PA also in the purpose direction, as
part of a generally available strategy (and one attested with other PAs) in
supressing the inherently active orientation of the PA.
Examples of the FAITH sense, shown in (10) above (lid v buoh věřící
‘people believing in God’), are first attested in the late 1300s and as is
apparent from Table 2, they are not very frequent either, especially in
comparison to the corresponding syntactic noun, to be discussed in section
3.2. The noun modified by the PA is typically lidi/lid ‘people’ or člověk
‘person’, and the generic meaning of these nouns also renders the colloca-
tion semantically equivalent to the syntactic nouns. In contrast to adnomi-
nal PAs in the BELIEF sense, the FAITH PAs sometimes express the non-
subject complement (Deity), especially in the oldest examples, as in (10).
This complement, however, does not provide a highly informative contri-
bution: the referent is always God/Christ, expressed either directly (buoh
‘god’, jezukristus ‘Christ’) or by a reflexive pronoun. This pattern suggests
pragmatic ‘emptying’ of the complement, whose presence thus serves
merely as a placeholder in the valence of the stem, not as an expression
identifying a pragmatically unpredictable, novel participant in the reported
event. As a result, the non-verbal potential of the PA is strengthened, de-
spite the presence of the root’s valence, and these PAs generally cast Chris-
tian faith as a property that characterizes a particular person or group of
people (those who can be classified as ‘having Christian faith’). Similarly,
lack of this property helps identify its opposite, as illustrated in (15).
A Frame Semantic account of morphosemantic change 309
In a further contrast to the BELIEF PAs, the FAITH examples are poten-
tially interesting with respect to the relative order of the PA and the noun it
modifies: while the tokens with complements, as in (10), show the linear
arrangement frequently found with the PAs in general ([NP – [comple-
ments – PA]]), the bare tokens, including the uses in the oldest layer (be-
fore 1380), tend to show the order in which the PA precedes the noun it
modifies, such as in (15)–(16). This can be taken as another indicator of a
gradual shift toward using the PA in a modifying function, for which the
Mod-N order was slowly establishing itself across all types of modifiers as
the only grammatical option (a more detailed discussion of the word order
and its role in the PA diachronic development can be found in Fried 2003
and forthcoming-b). Granted, there are also examples of the N-PA order,
but they do not seem to extend beyond mid-15th century and they all come
from biblical or religious texts, in which the N-PA order remained as a
genre-specific feature long after the original functional distinction between
contrastive and non-contrastive modifier became neutralized (cf. Kurz
1958). We can thus conclude that the syntactic patterning correlates with
the functional and semantic development. All these findings are arranged
chronologically in Figure 3. The lines with the bullets indicate the overall
presence of the PA in a given verb sense, the thin lines mark the presence
310 Mirjam Fried
of specific syntactic features and distinct meanings, and the dashed lines
indicate relatively low frequency in the corpus.
BELIEF
meaning (habitual)
'credible'
'authenticating' ?
FAITH
Mod-Head order
(habitual)
meaning
'devoted'
Figure 3. Chronological structure of PAs in adnominal patterns
Frame BELIEF
(content of thought)
FEs ... cat a lform věřící
sem ['devoted/faithful']
( ? collocation věřící list ) cat vNT-ppl lxm věřiti
'authenticating letter' (ex. 12)
Frame FAITH
cat a lform věřící (content of thought)
FEs: #1 Believer
val { #2 [Dat], (#3 [Loc]) }
#2 Deity
cat vNT-ppl lxm věřiti
val {#1 [Nom], #2 [Acc(v) ] }
Frame BELIEF
(content of thought) (ex. 16)
FEs: #1 Believer
#2 Recip. of Trust
#3 Content
val {#1 [Nom], #2 [Dat], (#3 [Loc]) }
The diagram also captures, at least roughly, the difference in the de-
grees to which the verbal component contributed by the root may fade in a
given PA class (indicated by the gray color). For example, the morphologi-
312 Mirjam Fried
cally regular BELIEF sense PA (lower left) remains a fully inflectional form
whose meaning corresponds directly to its morphemic structure, i.e., is
semantically fully compositional and hence, is not specified at the word
level. In contrast, the pseudo-PA (upper left) obliterates most of the inter-
nal verbal features; the only pieces of information that remain are the
meaning of the root (in the frame attribute) and reference to the participant
that is the referent of the NP modified by the PA (FE #3 Content). The
FAITH PA illustrates the gradual shift that is more typical of PAs in general:
in the early stage it retains the categorial tension between the NT-stem and
the external morphology, and its meaning and function are still more or
less predictable from the morphological structure, although the class of
referents for the Deity FE is fixed to ‘Christ’ (this restriction is indicated
by bold-face in the Frame specification), in a departure from the way the
finite verb is used (recall example (3a)). In subsequent development,
though, the interpretation of this PA extends further into a fully atemporal
classificatory meaning, suppressing any verbal/participial potential of the
form (in gray).
věříciemu
believe.PA.DAT.PL
[OtcB 19a; late 1300s; legend]
‘{Apolonius…taught him to aim for the eternal dwelling and
patiently seek God’s mercy…, telling him} that everything is
possible if one believes/for a believer’
This means that the syntactically nominal PA still could mark relative
tense and, crucially, only a particular context can determine which inter-
pretation is more accurate or is more likely intended by the speaker. It must
be noted that this is not a quirk of the form věřící, but follows the general
behavior of OCz PAs, both nominal and adnominal ones (Fried 2003 and
forthcoming-b).
The potential for ambiguity with respect to the relative strength of the
verbal character persists through the first part of the 15th century, but
overall, we observe a steady shift toward semantic nouns referring to enti-
ties that are characterized by certain properties, removed from any tempo-
ral or event-based interpretation that would be suggested by the participial
stem. ‘Having faith’ is not understood as an eventuality ascribed to an en-
tity (the restrictive relative clause interpretation) but as a classification of
an individual, independently of the event expressed by the main predicate.
This shift correlates with a gradual change in syntactic properties through-
out the 15th century. There is a noticeable erosion of the complement
structure: the attested occurrences are mostly bare (28 tokens out of 46), as
in (8), often accompanied by a universal quantifier (in 10 cases). If a com-
plement is present, it remains just as communicatively redundant as in the
oldest texts.
The low information value of the complement is nicely illustrated in
one particular text (VýklŠal, a religious tract from early 15th century, in-
terpreting the Song of Songs as a simile for proper Christian faith), which
consistently alternates between a bare PA věřící ‘believer’ and the phrase
věřící v Krista ‘believer in Christ’. The use of the latter suggests a fixed
collocation that simply stresses the Christian nature of the religious context
explicitly, but otherwise is not intended as a contributor of a novel event
participant. In fact, by mid-15th century we see the crystallization of the
PA’s meaning into denoting ‘a Christian’, i.e. as a label for a particular
social and religious group identity, or a class of people, rather than the
more general ‘one who believes in God/Christ’. Put differently, the
compositional meaning provided by the morphosemantic structure evolves
into a more specialized meaning that is not directly predictable from the
A Frame Semantic account of morphosemantic change 315
FAITH sense
compl. (v ACC)
modifiers (Quant.)
'believer'
meaning
'Christian'
'church member'
CREDIT sense
compl. - none
modifiers (Possess.)
meaning 'creditor'
(ex. 20)
COMMERCIAL
TRANSACTION cat n lform věřící
sem ['creditor']
TRANSFER lxm věřiti
cat vNT-ppl
Frame CREDIT
(belief as an expectation)
Frame BELIEF FEs: #1 Creditor
(content of thought)
#2 Debtor
FEs ... #3 Valued-Entity
val {#1 [Nom], #2 [Dat], #3 [Acc] }
(exs. 8, 17, 18)
cat n lform věřící
sem ['Christian']
val { ( #2 [Acc (v) ] ) }
V
BELIEF
'credible' (habitual)
A
věřúcí 'authenticating'
věřící list
V
CREDIT
'creditor'
N
věřící
V
FAITH
'believer' 'Christian'
N
'church member'
věřúcí 'Christian'
věřijící
'believing/Christian'
A
věřúcí 'devoted'
věřijící
TRUST V
The low incidence of the BELIEF PA in its active reading (labeled ‘habit-
ual’) can be perhaps motivated on pragmatic grounds. The CNG suffix pre-
disposes the PA toward developing readings in which the meaning of the
stem is interpreted as marking salient attributes of an entity or identifying
referents with such attributes, but this may be incompatible with what it
means to believe that something is true. We tend to perceive belief as a
fleeting state of affairs, not something that characterizes a person in a
noteworthy way (for that, Czech has a dedicated adjective, důvěřivý ‘trust-
ing/gullible’, which marks the habitualness and salience of the attribute by
the suffix -iv).
It is also significant that the PAs do not follow the same developmental
path for every sense. This fact is independent of the verb věřiti or any of its
senses, since the same unpredictable preferences for functional develop-
ment are commonly attested with other PAs. The diachronic patterning
only reveals something about this particular morphological form: while the
A Frame Semantic account of morphosemantic change 321
form itself carries the inherent potential to develop any (and all) of the
three syntactic functions (referential, attributive, predicative), we cannot
make a prediction about what direction of change we should expect. For
now I have to leave open the question of whether the diachronic tendencies
are truly unpredictable for every verb or whether there might be some very
general factors that predispose certain verbs or verb classes to a particular
functional outcome. But even if verb semantics (or other factors) turn out
to be relevant in this respect, such a result will not invalidate the observa-
tion that this is not a one-size-fits-all issue. The only generalization avail-
able at this point is the following: the external properties, not surprisingly,
have a stronger claim to dominance than the internal ones, hence a much
higher proportion of shifts toward reference and modification, as compared
to only marginal persistence of verbal behavior. But the competition be-
tween the formal (adjectival) and semantic (referential) potential of the
external morphology seems much less clear-cut. In the case of věřící, the
modificational uses cover a greater semantic territory within the semantic
network, but the referential ones turned out to be the most robust.
The oldest attestations are all within the prototypical (BELIEF) sense, but
they are all morphologically irregular – věřúcí (there is no example of a
clearly regular form indicated by the spelling věřiecí, even though the true
participle is attested quite commonly in the regular form věřiec ‘[while]
believing’). These pseudo-forms show a dramatic shift in meaning (‘be-
lievable’) in the adnominal usage, thus supporting the general hypothesis
that the irregular forms tend to be associated with some irregularity in
meaning or function. The corresponding active participial usage, ‘(the one)
believing’, does not appear until the point when the difference between the
regular and irregular formation is phonetically fully neutralized, as re-
flected in the spelling, at the end of the 14th century. From all this, we
could infer a diachronic development in which a pseudo-form starts out as
an independent form and is gradually pulled into the network of regular
PAs, potentially adjusting to their regular internal structure. However, as
noted in section 3.2, the same pseudo-form used in the nominal slots seems
to function the same way as the regular PAs, suggesting that the sequence
-ú-c-í must have retained some degree of transparency all along, at least
when used with this particular verb root and in this particular syntactic
function. It will require additional research to establish firmly that the dia-
chronic path of the pseudo-PAs vis-à-vis their regular counterparts sug-
gested above indeed is the correct one; and if it is, it also raises interesting
questions for the grammaticalization theory, minimally with respect to the
possibility that a frozen form may disintegrate back into its component
322 Mirjam Fried
5. Conclusions
Notes
3. I assume the commonly accepted cut-off point at the end of 14th century (e.g.
Vintr 1992): written i/y represents í < ú before that date, but is potentially am-
biguous after that.
4. As is well known, Frame Semantics and the version of Construction Grammar
presupposed in this paper incorporate the notion of prototype as a crucial or-
ganizational principle in representing speakers’ linguistic knowledge. Parts of
the analysis will thus be directly compatible with prototype-based research in
lexical semantics, especially as laid out in Geeraerts’ ground-breaking work in
cognitively oriented historical semantics. However, the present paper is not so
much concerned with the (re)organization of the conceptual structure that un-
derlies the meaning of lexical items (Geeraerts’ explicitly stated focus). I am
more concerned with changes in the lexico-grammatical organization associ-
ated with categorial changes in a particular inflectional word-form, for which
the apparatus of Construction Grammar, as a cognitively based model of lin-
guistic structure, is particularly suitable.
5. Glosses: NOM ‘nominative’, GEN ‘genitive’, DAT ‘dative’, ACC ‘accusa-
tive’, LOC ‘locative’, INS ‘instrumental’, M/F/N ‘masculine/feminine/ neu-
ter’, SG/PL’singular/plural’, NEG ‘negative’, PRES ‘present’, PST ‘past’,
FUT ‘future’, PPL ‘past participle’, PASS ‘passive participle’, IMP ‘impera-
tive’, INF ‘infinitive’, AUX ‘auxiliary’, RF ‘reflexive’, COND ‘conditional’,
PP ‘pragmatic particle’.
6. A note on presenting the examples: when additional context is helpful for
clearer understanding, it will be enclosed in curly brackets {} and left without
interlinear glossing, as in (3b). If the context in the original is too elaborate, I
will add an explanatory summary only in the English translation, enclosed in
parentheses (), as in (2a).
7. The cited texts and each example’s exact location in a text are identified by
the abbreviations and citing conventions established by the Old Czech Dic-
tionary (Staročeský slovník 1968).
8. The genitive in (5b) could be analyzed simply as a genitive of negation, but it
does not call into question the genitive as part of the conventional comple-
mentation pattern of this sense of věřiti. Notice that (5a) also contains the
genitive, even though there is no negation in the sentence.
9. While many of the manuscripts in the corpus can be dated quite precisely,
many others can only be placed within an estimated time period (a decade or
more). For the present purposes, it is sufficient to follow the practice estab-
lished by the Old Czech Dictionary (Staročeský slovník 1968), which recog-
nizes six chronological layers, each identified by reference to its upper bound-
ary ([up until] 1300, [up until] 1350, and so on).
10. In this respect it may also be worth noting that the copular use, sporadic but
persistent throughout the OCz period, as often as not functions as a predicate,
expressing simply a present-tense active event, as in (9) above. Moreover,
A Frame Semantic account of morphosemantic change 325
References
Atkins, B.T.S.
1994 Analyzing the verbs of seeing: A frame semantics approach to corpus
lexicography. In Berkeley Linguistic Society 20: 42–56.
Atkins, Sue, Charles J. Fillmore, and Christopher R. Johnson
2003 Lexicographic relevance: Selecting information from corpus evi-
dence. International Journal of Lexicography 16: 251–280.
Bisang, Walter, Nikolaus P. Himmelmann, and Björn Wiemer (eds.)
2004 What Makes Grammaticalization? A Look from Its Fringes and Its
Components. Berlin: Mouton de Gruyter.
Bybee, Joan, Revere Perkins, and William Pagliuca
1994 The Evolution of Grammar: Tense, Aspect, and Modality in the Lan-
guages of the World. Chicago: Chicago University Press.
Croft, William
1991 Syntactic Categories and Grammatical Relations: The Cognitive
Organization of Information. Chicago: Chicago University Press.
2001 Radical Construction Grammar. Oxford: Oxford University Press.
Daneš, František
1966 The relation of centre and periphery as a language universal. Travaux
linguistiques de Prague 2: 9–21.
Fillmore, Charles J.
1982 Frame Semantics. In Linguistics in the Morning Calm, The Linguis-
tic Society of Korea (ed.), 111–137. Seoul: Hanshin Publishing Com-
pany.
326 Mirjam Fried
Stephen M. Dickey
Abstract*
1. Preliminaries
In line with its assumption that all linguistic units are equally symbolic and
its rejection of a qualitative distinction between lexical and grammatical
units, cognitive linguistics has approached the analysis of the semantic
categories expressed by affixes and grammatical endings in essentially the
same way that it approaches the meanings of lexical units. Lehrer (2003)
argues that derivational affixes exhibit in principle the same kind of
polysemy that lexical elements do, even if to a lesser extent, given the fun-
damentally abstract nature of the meanings derivational affixes express.
Nikiforidou (1991) documents a structured polysemy for the genitive case
in Greek (and also cross-linguistically), which appears to reflect its seman-
tic development as a category.
Though these studies do not really consider issues of prototype theory,
the cognitive approach to meaning predicts that the categories expressed by
derivational and grammatical morphemes will exhibit prototype effects and
structure in a fashion comparable to lexical units. Tabakowska (1999) ar-
gues that the meanings of the Polish prefix po- form a structured network
containing extensions from three related prototypes, and Tabakowska
(2003) analyzes za- in a similar manner. Tabakowska’s analyses are con-
cerned only with synchrony, but the network structure based on extensions
from prototypes that she hypothesizes for Polish prefixes should have con-
sequences for analyses of diachronic development as well. Two recent
studies of prototype theory in its diachronic application are relevant in this
regard. Kemmer (1992) discusses the implications of (cross-linguistic)
prototypicality for the development of middle markers, and Geeraerts
(1997) discusses in detail the prototype effects (in a broad sense of the
term) affecting the semantic development of lexical units. A cognitive ap-
proach predicts that the semantic development of Slavic perfectivizing
prefixes (as derivational affixes) will display prototype effects similar to
those outlined by Kemmer and Geeraerts.
This paper applies principles of prototype theory to a diachronic analy-
sis of the meanings expressed by the Russian verbal prefix po-. Russian, as
a Slavic language, disposes of an aspect system consisting largely of pairs
of derivationally related imperfective (impf) and perfective (pf) verbs (for
a brief description of Slavic derivational aspect systems, see Dickey 2000:
7–12). Many such pairs are created by adding a perfectivizing prefix to a
simplex impf verb (e.g., stroit' (impf) – postroit' (pf) ‘build’), and among
the Russian perfectivizing prefixes po- has generally been recognized as
the most productive (cf., e.g., Čertkova 1996: 123–124 and Tixonov 1998:
A prototype account of the development of delimitative po- in Russian 331
If we recall that, as Hopper (1979) points out, one of the main functions
of pf verbs cross-linguistically is to sequence events in narratives, then it
becomes clear that by expressing a perfective construal of atelic activities,
delimitatives in fact perform a crucial systemic function in Russian – the
extension of the aspect opposition to a whole class of atelic verbs, i.e.,
Vendler’s activities.2 However, despite this crucial function, delimitative
verbs have not traditionally been considered pf “partner” verbs of their
source verbs on a par with ordinary telic pf verbs. Thus, while napisat'
‘write to completion’ is considered to be the pf “partner” verb of pisat'
‘write’, delimitative popisat' ‘write for a while’ has not traditionally been
considered to be paired with impf pisat' ‘write’ when the latter is construed
as an open-ended activity. Rather, delimitative popisat' ‘write for a while’
is considered to be an unpaired pf procedural verb, i.e., a special pf verb
which profiles some specific temporal configuration of a situation, but
which does not express the canonical, completed version of the situation.
And yet it is clear that delimitatives in po- are more important to the
system of Russian aspect than most other kinds of procedural verbs. For
this reason, recent treatments have recognized that delimitatives have a
special status among the types of Russian procedurals. Mehlig (2006) ob-
serves that po- delimitatives are the type of pf procedural that is most pro-
ductively used to perfectivize atelic impf verbs. Some treatments have even
allowed that they are “pf partner verbs” of their source verbs when the
action expressed by the source verb is construed as atelic (cf. in this regard
Čertkova 1996, Petruxina 2000: 187, Dickey and Hutcheson 2003, and
Dickey 2006). Given the current importance of po- delimitatives for the
Russian aspectual system, solving the puzzle of the development of delimi-
tative po- takes on a particular significance in a historical analysis of the
Russian aspectual system.
The view taken here is that the meanings of the prefix po- form (and
have formed) a family resemblance category (cf. Taylor 1995 and Geer-
aerts 1997), and that the semantic development of po- evident from histori-
cal sources indicates that there has been a shift in its prototypical center. In
particular, I argue that the category expressed by po- has shifted from a
cluster of meanings centered around a prototype of resultativity (based on
its spatial PATH/SURFACE-CONTACT meaning – see section 3) to a proto-
typical cluster centered around indefinite temporal delimitation. While this
initial point may be argued fairly easily on a descriptive level, the details of
the shift in the network prototype(s) are more complex than it would ap-
pear at first glance. This paper discusses the shift to delimitativity as the
A prototype account of the development of delimitative po- in Russian 333
Let us take as a point of departure the statistical data on the semantic de-
velopment of po- given by Dmitrieva (1991: 71), presented here in figures
1 and 2. Dmitrieva, adopting Isačenko’s (1960: 209–309) classification,
classifies po- verbs into the following types: (1) resultatives, e.g., postroit'
‘build’, (2) delimitatives, e.g., postojat' ‘stand for a while’, (3) ingressives,
e.g., poletet' ‘fly/take to flight’, (4) distributives, e.g., pobrosat' ‘throw [all
of]’, (5) attenuatives, e.g., poostyt' ‘cool off somewhat’, (6) intermittent-
attenuative verbs, e.g., pobrasyvat' ‘throw from time to time/on and off’.
The resultatives, delimitatives, ingressives, distributives and attenuatives
(types 1–5) are all pf verbs; the intermittent-attenuative verbs (type 6) are
imperfectives, as they also contain the imperfectivizing -yva- suffix. As the
intermittent-attenuative verbs gain their meaning from the combination of
attenuative (or delimitative) po- with the impf suffix -yva-, they do not
represent a distinct sense of po- and are not considered in this analysis
(though I have left them in figures 1 and 2 to give an idea of their produc-
tivity). This leaves us with five distinct senses of po-: resultativity, delimi-
tativity, ingressivity (restricted primarily to determinate motion verbs3),
distributivity (either object- or subject-oriented), and attenuativity. Before
discussing the elements of each sense relevant to the analysis, let us first
examine figures 1 and 2 (taken from Dmitrieva 1991: 71) to get a basic
picture of the semantic development of po- since the eleventh century.
Dmitrieva’s data are based on the largest Old Russian (ORus) and Mod-
ern Russian (MRus)4 dictionaries available at the time of her research, the
MSDJa and SSRLJa (respectively). Though the SSRLJa contains many
more entries than the MSDJa, we may consider a comparison of the respec-
tive percentages of the different senses of po- in each dictionary to be a
fairly accurate reflection of the changes in the semasiological structure of
the prefix since the eleventh century. A comparison of Dmitrieva’s data
reveals that po- has undergone a significant shift in the relative frequencies
of its senses from ORus to MRus: in ORus, 73.5% of all po- verbs were
resultatives, whereas in MRus the resultatives are only 26% of all po-
verbs. On the other hand, in ORus delimitatives comprised only 3.8%,
compared to 31.8% in MRus. Regarding the statistics for delimitative verbs
in MRus, it must be emphasized that, as Isačenko (1962: 391–392) points
334 Stephen M. Dickey
out, delimitative po- “is so productive that even the most comprehensive
dictionaries register only a small fraction of the delimitatives that actually
occur.” Thus, the 31.8% percentage for po- delimitatives in MRus should
be considered fairly conservative.
Intermittent-
Attenuative: 8.9%
Attenuative: 0.5%
Distributive: 9.1%
Ingressive: 4.9%
Delimitative: 3.8%
Resultative: 73.5%
Figure 1. Types of verbs prefixed with PO- in Old Russian (Dmitrieva 1991: 71)
Intermittent-
Attenuative: 9.2%
Resultative: 26%
Attenuative: 11.9%
Distributive: 19%
Figure 2. Types of verbs prefixed with PO- in Modern Russian (Dmitrieva 1991:
71)
A prototype account of the development of delimitative po- in Russian 335
meaning, e.g., pozveniti ‘begin to ring’). This left two primary spatial tra-
jector-landmark configurations: the GOAL configuration (i.e., movement
toward a landmark; cf., e.g., ORus postignuti ‘reach’ and povyknuti ‘get
used to’ – note that in the latter the motion is metaphorical), and, more
importantly, the PATH configuration (i.e., movement or location along a
path or surface; cf., e.g., ORus poplavati ‘roam/wander [an area]’ and po-
voditi ‘lead about’). The PATH configuration was clearly dominant in
ORus, and was closely associated with mere contact with some surface, cf.
Shull’s (2003: 43, 160–161) remarks on the redundancy between these two
notions (as well as that of motion through space); by historical times, PATH
had given rise to a productive SURFACE-CONTACT meaning that was very
resultative in nature, cf., e.g., ORus posmoliti ‘cover with resin’. It should
probably be assumed that in the original situation PATH and GOAL were
distinct local prototypes in the network of po-. However, the
PATH/SURFACE-CONTACT meaning was a very telic one, i.e., it tended to
profile the complete affectedness of the surface in question, so that pop-
lavati meant ‘roam all over [an area]’, and posmoliti meant ‘cover [com-
pletely] with resin’. Thus, PATH/SURFACE-CONTACT could produce resulta-
tive verbs by metaphor on a par with the GOAL configuration. Inasmuch as
GOAL was still a distinct meaning of ORus po-, it was already weakening,
which is evidenced by the parallel demise of the GOAL preposition in his-
torical times: po in the spatial meaning of ‘up to’ was replaced by do, and
po in the goal-oriented meaning of ‘for/to get’ was replaced by za. Note
that allative verbs such as postignuti ‘reach’ easily allow an alternative
construal as goal-oriented PATH situations, inasmuch as reaching necessar-
ily involves motion along some path during the approach (it should also be
pointed out that MRus has also replaced the prefix po- in the spatial GOAL
meaning, cf., e.g., dostič' ‘reach)’. I therefore assume that the prototype of
po- in ORus was the PATH/SURFACE-CONTACT meaning, which by meta-
phorical extension had produced a large class of ordinary resultative verbs,
e.g., ORus postaviti ‘erect [on a surface]’, as well as a smaller number of
specifically distributive verbs, e.g., ORus povoziti ‘transport [all of]’. The
po- distributives appear to have originally been a direct metaphorical ex-
tension of the SURFACE-CONTACT meaning: the meaning of covering a full
space is transferred to the quantitative domain whereby the objects (or
subjects) are conceptualized as points along a surface which are covered or
encompassed by the predicate in question (cf. Dmitrieva 2000).7 Though
the central, prototypical meaning of po- was PATH/SURFACE-CONTACT, this
meaning was compatible with and linked to the other two meanings, GOAL
and (the remnants of) SOURCE, by a shared abstract schema (in terms of
A prototype account of the development of delimitative po- in Russian 337
Examples such as (2a), (2b) and (2c) are much different from (2d) and
(2e), in that the extension of the predicate over the relatively long duration
in each example represents, as Sigalov (1975: 152) suggests, a resultative
sense of these po- delimitatives. Though the point is hard to argue conclu-
sively, in my view examples such as (2a) and (2c) represent telic predi-
cates, in the same way that walk a mile is a telic accomplishment predicate
as opposed to the atelic activity walk. Accordingly, (2a) and (2c) contain
telic predicates in the sense that the end of the specified intervals of time
(28 years and 13 years) are the endpoints of the respective predicates. In
any case, the predicates in (2a), (2b) and (2c) have the resultative feel men-
tioned by Sigalov (loc. cit.), due to the fact that in each case the duration of
the predicate reaches the maximum possible in the context, i.e., based on
the narrator’s knowledge of the world the predicates could not have con-
tinued longer than the specified period of time. It is for this reason that
ORus požiti ‘live for some time’ is used so often to sum up the duration of
rulers’ or saints’ lives preceding mention of their death, as in (2a) and (2b),
so that a major meaning of ORus požiti is something akin to ‘live the rest
of one’s days’ (this is probably why Sreznevskij’s MSDJa 2: 1082 defines
it first as MRus prožit' ‘live for a specified, long period of time’ and only
A prototype account of the development of delimitative po- in Russian 339
subsequently as požit' ‘live for some time’). Note that posědě 13 lět" ‘sat
for 13 years’ in (2c) is the exact same kind of usage, summing up the re-
mainder of the prince’s life before his death. Such usage is in clear contrast
with (2d) and (2e), where the delimitation of the predicate is clearly arbi-
trary, i.e., the predicates in (2d) and (2e) could have easily continued
longer than ‘just a little while’. It is because of this quality of arbitrarily
short delimitation that examples such as (2d) and (2e) have a much less
resultative feel (again, per Sigalov loc. cit.), and accordingly can hardly be
considered telic.10
The analysis that follows relies on the distinction between these two
senses of ORus delimitatives. Let us call the resultative/telic delimitativity
occurring in cases of a relatively long period of time ABSOLUTE DURATION,
and the delimitativity in cases of a relatively short and often indefinite pe-
riod of time RELATIVE DELIMITATION. ABSOLUTE DURATION usually in-
volves a precisely specified interval of time, such as ‘28 years’ in example
(2a), and, depending on the predicate, the interval may be longer than the
situation typically lasts; in any case, ABSOLUTE DURATION represents either
a relatively very long period of time and/or the maximum duration for
which the predicate could continue in the context. RELATIVE DELIMITA-
TION involves a relatively short interval of time for the predicate in ques-
tion, or one that is left indefinite, and perhaps even unspecified in which
case it is by default an interval of time typical for an episode of the situa-
tion in question.11 Again, note that delimitatives expressing ABSOLUTE
DURATION as in (2a) in fact functioned as resultative (telic) verbs (cf. Siga-
lov loc. cit.); indeed, (2a) occurs at the end of a narrative and sums up the
remainder of Vladimir’s life before his death. Not only was Sigalov (1975)
the first to distinguish these two distinct senses of ORus delimitative po-,
but he also recognized that, whereas ORus delimitatives could express both
ABSOLUTE DURATION and RELATIVE DELIMITATION, MRus delimitatives
have lost their ability to express ABSOLUTE DURATION, as shown by the
MRus translations of (2a), (2b) and (2c):
TR
Figure 3. Shull’s (2003: 153) schema for Russian PO- with motion verbs
In (5a) the full relevant trajectory is passing by another man along the
path; notice that an ablative/ingressive interpretation is completely inap-
propriate here. Perhaps a more illustrative example, albeit with another
lexical verb, poplyti ‘swim’, is given with its MRus translation in (6):
For ease of reference, Shull’s hypothesized schema for MRus po- with
motion verbs shall be termed INGRESSIVE-PARTIAL TRAJECTORY, and the
meaning expressed by MRus pro- with motion verbs shall be termed FULL
TRAJECTORY.
The situation regarding PATH/SURFACE-CONTACT po- with determinate
motion verbs thus shows an interesting parallel with the older delimita-
tives, which should not be overlooked. Just as ORus stative delimitatives in
po- could express two kinds of extension in time, ABSOLUTE DURATION
and RELATIVE DELIMITATION, ORus determinate motion verbs in po- could
express both FULL TRAJECTORY and INGRESSIVE-PARTIAL TRAJECTORY.
The only noteworthy difference lies in the fact that with delimitatives the
sense of ABSOLUTE DURATION was just as frequent as RELATIVE DELIMITA-
TION, whereas poiti most often expressed INGRESSIVE-PARTIAL TRAJEC-
TORY (which is expected, given that the default profile of go is the source
of motion), and only occasionally expressed FULL TRAJECTORY. The se-
mantic parallel between ORus determinate motion verbs in po- and delimi-
tatives in po- is summarized in table 1.
Let us sum up what we have established so far: ORus po- represented a
network originally consisting of a salient PATH/SURFACE-CONTACT mean-
ing, which was the central prototype of the prefix prior to the latter half of
the seventeenth century. The PATH/SURFACE-CONTACT sense of po- had as
metaphorical extensions salient meanings of resultativity (e.g., postaviti
‘build’) and distributivity (e.g., pometati ‘throw [all of]’). Another meta-
phorical extension evident in a small group of stative activity verbs was the
delimitative meaning (e.g., posěděti ‘sit for some time’); delimitativity had
two variants, ABSOLUTE DURATION and RELATIVE DELIMITATION. Lastly,
due to their particular spatial default meaning, determinate motion verbs
prefixed with PATH/SURFACE-CONTACT po- had an ingressive meaning, i.e.,
INGRESSIVE-PARTIAL TRAJECTORY, which can nevertheless be considered
an extension of the central PATH/SURFACE-CONTACT meaning; such verbs
could also express FULL TRAJECTORY.
A prototype account of the development of delimitative po- in Russian 345
Table 1. The semantic parallel between ORus determinate motion verbs and delimi-
tatives in PO-
PO- LIMITED MAXIMAL CONCEPTUAL
VERB TYPE TRAJECTORY TRAJECTORY DOMAIN
Determinate Ingressive- Full Trajectory Motion in Space
Motion Partial Trajectory
Verbs
Delimitatives Relative Absolute Duration Duration in Time
Delimitation
delimitatives on some basic descriptive level, we are left without any ap-
parent motivation for the shift in the prototype. This problem is com-
pounded by the fact that the new prototype, delimitativity, also inexplica-
bly loses the ability to express one of its senses, ABSOLUTE DURATION,
leaving MRus delimitatives able only to express RELATIVE DELIMITATION.
It is important to point out that this is an unexpected situation: as Dickey
and Hutcheson (2003) observe, in some other Slavic languages, e.g.,
Czech, the existing delimitatives have retained their ability to express AB-
SOLUTE DURATION, which is in clear contrast with MRus, as shown in (7a)
and (7b), taken from Petruxina (2000: 164).
Thus, the core problem is really this: how and why did the ORus delimi-
tatives lose the ability to express ABSOLUTE DURATION around the seven-
teenth century?
But for this problem, we can easily analyze the spread of the delimita-
tive meaning of po- as a case of semantic change from a subset, as outlined
by Geeraerts (1997: 68–79). For example, English meat originally meant
‘food’, with the meaning of ‘meat’ as a subset; gradually the meaning
‘meat’ develops into the sole meaning of the word. Similarly, Dutch winkel
has come to mean ‘shop’ in the following way. Originally winkel meant
‘corner’, which includes ‘street corner’ as a subset. From the subset ‘street
corner’ a metonymical extension produces the meaning ‘building located
on a street corner’, which in turn includes ‘shop located on a street corner’
as a subset. The older meaning of ‘corner’ continues to exist for some time
alongside the new meaning ‘shop located on a street corner’. A process of
semantic generalization from this second subset produces the simple mean-
ing ‘shop’. We may consider the delimitative meaning of ORus po- to be
ABSOLUTE DURATION, which includes RELATIVE DELIMITATION as a subset,
348 Stephen M. Dickey
which then develops into an independent meaning. But while in the case of
English meat the salience of meat as a prized source of nourishment plau-
sibly motivates the change, and in the case of Dutch winkel the change is
motivated by the communicative need for clarity in matters of commerce
coupled with the salience of shops in our cultural awareness, it is almost
impossible to imagine an analogous motivation in the abstract domain of
time that would give ORus posěděti ‘sit [RELATIVE DELIMITATION]’ the
upper hand against posěděti ‘sit [ABSOLUTE DURATION]’ to the point where
the latter disappears (along with a whole class of resultative verbs in po-)
as the former becomes a salient token in the development of a productive
new derivational pattern.
This is where ORus poiti ‘go’ with its default meaning of INGRESSIVE-
PARTIAL TRAJECTORY becomes relevant. In various languages of the world,
the motion verb go has undergone various metaphorical extensions, the
most common of which involve the nearly ubiquitous TIME IS SPACE meta-
phor, which leads to its grammaticalization as a future auxiliary, cf., e.g.,
the French aller-future and the English gonna-future (cf. Hopper and
Traugott 2003: 23, 87–90), as well as the Hausa zaa-future (cf. Abdoulaye
2001; zaa is an ingressive motion verb). I suggest that ORus poiti (along
with the other determinate motion verbs prefixed in po-, e.g., poěxati
‘ride’, poplyti ‘swim’, etc.), with its default meaning of INGRESSIVE-
PARTIAL TRAJECTORY, served as the catalyst for the development of RELA-
TIVE DELIMITATION as the prototype of po-.
The INGRESSIVE-PARTIAL TRAJECTORY expressed by ORus poiti bears a
striking resemblance to the RELATIVE DELIMITATION meaning expressed by
delimitative verbs. For example, both the INGRESSIVE-PARTIAL TRAJEC-
TORY expressed by determinate motion verbs in po- and the RELATIVE DE-
LIMITATION expressed by ORus delimitatives in po- (as well as their MRus
counterparts) profile some indefinite quantity of an action that is less than
a canonical, completed event. In other words, the aspectual meaning ex-
pressed by ORus poiti and that expressed by a verb such as ORus pobesě-
dovati ‘discuss’ (in its RELATIVE DELIMITATION meaning) are parallel: in
the case of poiti the profile is less than the complete trajectory of the path,
and in the case of pobesědovati the profile is some amount of discussion
less than a complete discussion that produces a result. As suggested in
figure 4, the difference is primarily one of cognitive domain, i.e. determi-
nate motion in space versus communicative interaction in time.
When and how would ORus poiti come to serve as a token for the de-
velopment of a new class of delimitatives? In order to answer this question,
we must examine poiti as a perfective verb. Recall that the aspectual mean-
A prototype account of the development of delimitative po- in Russian 349
ing of poiti, i.e. its status as a perfective of iti ‘go’, was the result of the
redundancy of the PATH/SURFACE-CONTACT meaning of po- in combination
with iti ‘go’, which had a default focus on the source (inception) of some
trajectory (which always occurs on some path). It appears that ORus iti –
poiti became linked as a kind of aspectual pair relatively late in time.
Though the process appears to have been completed by the seventeenth
century (cf. Mayo 1984: 32), in earlier centuries ORus unprefixed iti ‘go’
regularly occurred in narrative sequences of events – the canonical context
for the pf in the modern language. A representative example is (8а), cf. the
MRus translation in (8b), which requires the pf pojti:
Strekalova (1968: 49) observes on the basis of similar data that Polish
iść ‘go’ was not integrated into the Polish aspectual system as an impf verb
until around the sixteenth century (which involved it becoming paired with
pf poiść as its default pf). The establishment of ORus iti ‘go’ as an impf
verb must have occurred at approximately the same time, and likewise
depended on the development of poiti as its pf correlate (partner verb).
Accordingly, it is most likely that the ORus aspect pair iti – poiti became
established around the sixteenth century, and no earlier than the fifteenth
century. Thus, poiti became a pf correlate of iti not long before the advent
of the new delimitatives. If we assume that ORus pf poiti was a catalyst for
the rise of the new delimitatives in the seventeenth century, the chronology
makes sense: poiti as a pf correlate of iti was available to serve as model
for the creation of a new class of atelic verbs, the delimitatives, which be-
350 Stephen M. Dickey
gan in the seventeenth century; however, poiti did not exist in this capacity
much earlier, which makes it unnecessary to explain why delimitatives did
not begin their productivity earlier.
The hypothesis advocated here is that innovative pf poiti, in which po-
expressed an emerging atelic meaning of INGRESSIVE-PARTIAL TRAJEC-
TORY in its capacity as a perfectivizing prefix, played a crucial role in the
development of the new delimitatives. This hypothesis accords well with
the mechanisms of prototypical change outlined above, in that a salient
token of a peripheral meaning in a category network can only facilitate the
development or strengthening of that meaning into a local prototype or
even the central meaning of the category. Moreover, poiti, as a form of the
highly frequent default motion verb ‘go’, is about the only verb in the lin-
guistic system of Russian that would have enough salience to contribute
significantly to such a reorganization of the semantic nature of po- in this
way. The relatively late emergence of poiti as the pf correlate to iti in-
volved a new kind of meaning expressed by po-, INGRESSIVE-PARTIAL TRA-
JECTORY, which as such would be incorporated into its network in one way
or another. I suggest that the new INGRESSIVE-PARTIAL TRAJECTORY mean-
ing assumed a relatively salient position in the network of po- given its
association with the highly salient motion verb poiti. The close resem-
blance between INGRESSIVE-PARTIAL TRAJECTORY and RELATIVE DELIMI-
TATION was the point of contact in the evolving network of po- that re-
sulted in the development of the new delimitatives: INGRESSIVE-PARTIAL
TRAJECTORY and RELATIVE DELIMITATION were reanalyzed together as the
core of a nascent cluster of atelic meanings of the prefix.15 With the addi-
tional conceptual and systemic salience of poiti, it seems much more plau-
sible to analyze the development of RELATIVE DELIMITATION as a case of
Geeraerts’ semantic change from a subset. Further semantic developments
shifted the prototypical center of the prefix away from the old telic cluster
of PATH/SURFACE-CONTACT and resultative meanings to the new atelic
cluster centered around INGRESSIVE-PARTIAL TRAJECTORY. The shift pro-
ceeded through two processes: (1) the swift expansion of the predicate
types producing new delimitatives, and (2) the realignment of some older
types of po- verbs with the new prototypical center.
Sigalov (1975) documents the spread of delimitative po- at the end of
the ORus period from the small class of stative delimitatives inherited from
Common Slavic. According to Sigalov (ibid., 171), the delimitative mean-
ing spread initially to po- derivatives of indeterminate motion verbs, e.g.,
ORus poběgati ‘run for a while’, poxoditi ‘walk for a while’, and “verbs of
psychological activity”, e.g., ORus pomolitisja ‘pray for while’ and
A prototype account of the development of delimitative po- in Russian 351
poveselitisja ‘be cheerful for a while’, in the sixteenth and especially sev-
enteenth centuries. Afterwards, delimitatives were derived from derivatives
of verbs of speech (e.g., pogovoriti ‘talk for a while’), verbs of sound (e.g.,
poguděti ‘drone for a while’) and verbs of physical activity (e.g., pokopati
‘dig for a while’), in the seventeenth and eighteenth centuries. Dickey
(2005: 45–47) briefly analyzes the spread of delimitativity among the vari-
ous lexical classes of predicates as a process of retextualization (cf. Nich-
ols and Timberlake 1991), based on the salience of the token poiti (as well
as the other determinate motion verbs in po-) in the development of the
new class of delimitatives, though without explaining the precise relation-
ship between the two predicate types – and as a consequence, without pro-
viding a plausible semantic motivation for the development. The prototype
analysis offered here provides the plausible motivation for the change: the
coalescence of the two atelic meanings expressed by po-, the RELATIVE
DELIMITATION of the old delimitatives and the newer INGRESSIVE-PARTIAL
TRAJECTORY of poiti, combined with the salience of poiti ‘go’ as a token of
the new category of atelic perfectivity created a situation which was ripe
for the spread of the derivational model of atelic po- perfectives (i.e., de-
limitatives) to various predicate types by analogy. Once iti – poiti was es-
tablished as an aspectual pair in Russian, it is reasonable to assume that
verbs of lesser salience, i.e. activity verbs of various kinds, would begin
systematically to derive atelic pf partner verbs by analogy (or metaphorical
extension). As suggested at the beginning of this paper, it is my view that
the development of the highly productive class of atelic delimitatives con-
tributed to the grammaticalization of Russian aspect in a major way by
extending the impf : pf aspect opposition to a large class of atelic activity
predicates.16
The assumption that poiti played a role in the rise of the new po- de-
limitatives in Russian allows us to make sense of the fact that the indeter-
minate motion verbs (e.g., ORus xoditi ‘walk’, ězditi ‘ride’, etc.) produced
delimitatives in po- relatively early. It is in fact unclear to what extent in
Late Common Slavic the so-called indeterminate motion verbs really com-
prised a distinct class with specific aspectual properties, as opposed to a
loose set of relatively stative manner of motion verbs (i.e., verbs expressing
a manner of motion, e.g., walking, as an activity without focusing on any
concomitant change of position, hence the term “stative”). Vaillant (1939:
294) expresses doubts on this issue, observing that (Common) Slavic pairs
of verbs such as nesti and nositi ‘carry’ are “just as independent of one
another as [French] aller [‘go’] and marcher [‘walk’]”. Thus, it is possible
(and in my view most accurate) to view indeterminate verbs in older stages
352 Stephen M. Dickey
why would the prefix lose this meaning when the corresponding preposi-
tion po has clearly maintained PATH/SURFACE-CONTACT as its spatial
meaning? This puzzling development is most easily explained by the re-
dundancy of a path prefix with a verb of motion (see section 3), as all mo-
tion occurs redundantly on some path. The salience of go as the highly
frequent default motion verb lent a corresponding salience to the new non-
spatial meaning of the prefix, INGRESSIVE-PARTIAL TRAJECTORY, which
arose by the subsumption of the spatial meaning of the prefix in the pre-
fixed motion verb. The subsequent reanalysis of INGRESSIVE-PARTIAL TRA-
JECTORY and RELATIVE DELIMITATION as a new central meaning of the
prefix laid the foundation for the complete loss of its original spatial mean-
ing.20 It should be stressed that the spatial meaning of the corresponding
preposition po would not be subject to subsumption by a verbal stem, and
so would be unlikely to be affected by developments in the verbal system.
The ability of the hypothesis presented here to explain this difference be-
tween the semantic properties of the prefix po- and the preposition po in
MRus, which to my knowledge has never been properly considered, let
alone satisfactorily explained, is an important point in its favor. (Note that
Tabakowska 1999 does examine the differences between the preposition po
and the prefix po- in Polish, but in my view her analysis does not carry
over to Russian.)
Finally, one last descriptive issue must be addressed: if RELATIVE DE-
LIMITATION as the meaning of the productive new class of Russian delimi-
tatives resulted in large part due to its conceptual proximity to INGRESSIVE-
PARTIAL TRAJECTORY, why is it that MRus pojti and other determinate
motion verbs in po- cannot combine with a facultative adverbial time
phrase, unlike MRus delimitative verbs, cf., e.g., *pojti neskol'ko minut ‘go
for a few minutes’ vs. posidet' neskol'ko minut ‘sit for a few minutes’?
First, it is worth pointing out that just because ORus poiti was the source of
the meaning of INGRESSIVE-PARTIAL TRAJECTORY that produced the new
delimitatives does not mean that this verb itself was (is) a member of that
specific class of verbs. Further, the contrasting combinability with adver-
bial time phrases is the expected consequence of the natures of the respec-
tive predicate types. Despite its particular source-oriented trajectory-
landmark profile, MRus pojti ‘go’ as a determinate verb of motion is never-
theless most often used in a goal-oriented sense, which is demonstrated by
the fact that pojti combines freely with GOAL prepositional phrases as in
pojti v magazin ‘go to the store’ (cf. in this respect Shull’s 2003 views on
the effect of the goal-oriented nature of language on Slavic prefixation).
The notion of linear progression towards a goal is so central to the meaning
A prototype account of the development of delimitative po- in Russian 357
While such examples are fairly marginal, especially (9a), they neverthe-
less indicate a certain potential to express the duration of a motion event
inherent in determinate motion verbs in po-. This seems particularly true of
(9b), in which pojdet ‘PO-goes’ is reduplicated to produce a meaning akin
to ‘keeps going’ or ‘goes for a while’. Such reduplication is not very diffi-
cult to find, even in the past tense, cf. the following examples:
Again, though examples (9) and (10) are admittedly marginal, such us-
age of determinate motion verbs in po- nevertheless does occur, and the
usage in (10) produces the same kind of durative meaning as reduplicated
po- delimitatives. It is important to point out that ordinary telic pf verbs are
to my knowledge unattested in such usage. Thus, despite the fact that de-
terminate motion verbs in po- are unable to profile the duration of the ac-
tion directly, they nevertheless do display a certain resemblance to po-
delimitatives when reduplicated. Sémon (1986: 614) attributes the usage in
(9) to the innovative productivity of delimitative po-, but it is much more
likely that this usage is in fact a relic of older patterns of usage of determi-
nate motion verbs in po-.
With regard to this last point, it is very important to understand that de-
terminate motion verbs prefixed with po- were not always unable to com-
bine directly with an adverbial time phrase. Aitzetmüller (1991: 172) ob-
serves that determinate motion verbs in po- sometimes had delimitative
meaning in Old Church Slavic, and gives the following example:
360 Stephen M. Dickey
ORus poiti is also attested with such time phrases as late as the fifteenth
century, as shown in (13):
As far as I am aware, no such usage has been attested since the fifteenth
century, i.e., more or less a century before poiti began to function as the pf
correlate of iti ‘go’ (in any case, such usage would have disappeared once
indeterminate motion verbs in po- took on a delimitative meaning).21 In
view of this fact, I consider such usage to be a case of po- expressing AB-
SOLUTE DURATION, i.e., as a metaphorical transfer of FULL TRAJECTORY,
which was in turn derived from the old PATH/SURFACE-CONTACT meaning
of the prefix. Thus, such older usage does not affect the hypothesis advo-
cated here. In support of this view, it should be pointed out that attestations
of poiti expressing ABSOLUTE DURATION before the sixteenth century also
accord with the documented fact that during the sixteenth and seventeenth
centuries pro- (‘through’) took over the expression of PATH/SURFACE-
CONTACT and thus FULL TRAJECTORY from po-. In this respect, pro-, as the
new PATH/SURFACE-CONTACT prefix, played an important role in the shift
to INGRESSIVE-PARTIAL TRAJECTORY as the prototypical meaning of po-.
One may only speculate as to why ORus would develop a new expres-
sion of PATH/SURFACE-CONTACT. It is likely that the new atelic INGRES-
SIVE-PARTIAL TRAJECTORY created an unacceptable or unstable polysemy
A prototype account of the development of delimitative po- in Russian 361
5. Concluding remarks
This is why the otherwise highly valuable findings and analyses pre-
sented by Bybee, Perkins and Pagliuca (1994) will probably never be of
more than secondary use to scholars attempting to figure out how Slavic
aspect developed into a grammaticalized category. It is due to the lexical
nature of the prefixes involved that the semantic pathways of the develop-
ment of Slavic perfectivizing prefixes as prototypical categories bear more
resemblance to the lexical cases discussed by Geeraerts (1997) than to the
demonstrated paths of purely grammatical units. (The considerably “lexi-
cal” nature of the grammatical category of Slavic aspect presents no diffi-
culties for a cognitive analysis, which assumes that there is a continuum
between lexical and grammatical units and moreover that grammatical and
lexical units can and often do share the same semantic organization.)
Though the importance of universal, recurrent mechanisms of change (e.g.,
metaphor and metonymy, etc.) remains beyond doubt, the watershed se-
mantic developments in the history of the Slavic perfectivizing prefixes in
fact appear to have been of an accidental nature: Dickey (2005) concludes
that the highly productive Slavic resultative prefix s-/z- was primarily the
accidental result of a sound change, and the analysis presented here argues
that the development of Russian perfectivizing po- was in large part a con-
sequence of the singular combination of the prefix with a particularly sali-
ent predicate type (determinate motion verbs such as iti ‘go’). In this latter
case as well, the development of Slavic perfectivizing prefixes conforms to
Nichols’ and Timberlake’s (1991: 129) suggestion that processes of gram-
maticalization are “less straightforward and obvious than is usually as-
sumed.”
It is worth pointing out that the account of the development of the prefix
po- given here comports with recent views on the timing of the develop-
ment of Russian aspect. Bermel (1997) demonstrates that Russian aspect
was not grammaticalized early, but rather developed more or less continu-
ously throughout the Old Russian period and into the Modern Russian pe-
riod. Nørgård-Sørensen (1997) concludes that Russian aspect must have
emerged as a grammaticalized category in Russian in the seventeenth cen-
tury. The evidence of a significant shift in the semantic network of po-
beginning in the sixteenth century, as well as the establishment of delimita-
tives as a productive class in the seventeenth and eighteenth centuries, fits
in well with their conclusions, especially those of Nørgård-Sørensen on the
importance of the seventeenth century in the development of Russian as-
pect. Given the coincidence in timing, I suggest that the development of
po- was another important change that occurred in the Russian language at
this time, and that the significance of the shift in the meaning of po- (as
A prototype account of the development of delimitative po- in Russian 363
manifested in the rise of delimitative verbs) for the Russian aspect system
is more or less comparable to that of the rise of imperfective suffixation.
As Hopper and Traugott (2003: 71) observe, causal explanations are
impossible in investigations into grammaticalization, and it makes sense
only to speak in terms of motivations. Similarly, Nichols and Timberlake
(1991: 131) completely dispense with any pretense (!), conceding that “the
task of investigating historical semantics of grammatical morphemes is in
principle impossible” and hoping only “to sketch what seems a plausible
path of development.” In a case as messy as Russian po-, these are surely
the only reasonable approaches. The one thing that does seem to be clear
about po- is that it has been characterized by a frustrating degree of
polysemy. For this reason, a prototype approach suggests itself naturally,
given its demonstrated suitability to handle complex cases of polysemy.
Though the analysis presented above cannot claim to have solved the entire
puzzle (and the chronology must remain at the level of century-scale ap-
proximations), at the very least it has the advantage of utilizing independ-
ently motivated principles of prototype theory as a tool for organizing our
knowledge of the semantic development of Russian po- in order to hy-
pothesize plausible paths of development. Moreover, it again demonstrates
the value of prototype theory and its ability to handle the semantic devel-
opment of Slavic perfectivizing prefixes, which straddle the distinction
between lexicon and grammar and each of which must be understood fully
in their diachronic development if we are to solve the problem of the
grammaticalization of Slavic aspect.
Notes
* This article was supported in part by a New Faculty General Research Fund
Grant awarded by the University of Kansas. This support is acknowledged
with gratitude. I would like to sincerely thank Alina Israeli and Irina Six for
their frequent insights as well as some Modern Russian translations of Old
Russian sentences, and Erin Moulton for providing some of the Old Russian
data contained here. I am also grateful for the comments of two anonymous
reviewers that helped improve this article. Any errors or inaccuracies are natu-
rally mine alone.
1. There is no universal consensus on the telic vs. atelic status of po- delimita-
tives. However, it should be pointed out that the majority of Slavic aspectolo-
gists consider them to be atelic. Bulygina (1982: 60) makes the crucial point
that if one can describe a situation as zanimalsja (impf) ‘occupied oneself
364 Stephen M. Dickey
with/did’, then no matter how short the duration of the activity, it may be de-
scribed perfectively with pozanimalsja (pf) ‘PO-occupied oneself with/PO-did’;
Kučera (1984) makes the same point. This fact proves beyond doubt the at-
elicity of po- delimitatives in Russian. Mehlig (e.g., 1994, 2006) uses the term
netransformativnyj, and Janda (2006) labels atelic verbs uncompletable. Both
these terms are equivalent to atelic for purposes of this analysis.
2. Note that not all Slavic languages have developed delimitatives as a produc-
tive class of verbs in order to extend the aspect opposition to activity predi-
cates; for details, see Dickey and Hutcheson (2003).
3. Russian has divided motion verbs into two classes, DETERMINATE (e.g., idti
‘go’) profiling a one-way, goal-oriented trajectory, and INDETERMINATE (e.g.,
xodit' ‘walk/go’) profiling motion without an obviously linear trajectory (ha-
bitual, aimless, etc.). The exact nature of the latter class in older stages of Rus-
sian is in fact less than clear (see section 4).
4. No single periodization of the Russian language is universally accepted, but
we may generally divide its history into the following periods: Old East Slavic
(from the eleventh to the fourteenth century), Middle Russian (from the late
fourteenth to the seventeenth century), and Modern Russian (from the eight-
eenth century to date). Spanning the end of the Middle Russian period and the
first century or so of the Modern Russian period was an important transitional
period that saw the formation of the Russian national language (from the late
seventeenth to the early nineteenth century). As the distinction that is primarily
relevant for this discussion is that between pre-modern Russian and modern
Russian after the transitional period (i.e., from the beginning of the nineteenth
century), these two periods are hereinafter labeled Old Russian and Modern
Russian respectively.
5. There are many interesting issues involved with this change that are not im-
mediately relevant for the purpose at hand. One that deserves a brief comment
is the healthy increase in distributive verbs. While it is true that the percentage
of po- distributives increased considerably, distributive po- has not become
nearly as productive as delimitative po-: pere- became the primary distributive
prefix in MRus (cf., e.g., perebrosat' ‘throw [all of one after another]’), largely
taking over this function from po- except in the derivation of distributives
from verbs already containing a prefix, e.g., po-vy-brasyvat' ‘throw out [all
of]’ (from vy-brasyvat' ‘throw out’). The replacement of distributive po- by
distributive pere- should in fact be viewed as part of the overall process that
led to po- being a primarily delimitative suffix.
6. Regarding Dmitrieva’s and Sigalov’s claims, one reviewer wonders whether
po- delimitatives are absent in ORus texts because they are characteristic of
colloquial style. I believe that it is better to take the texts at face value in this
case, as po- delimitatives do occur in ORus texts, but only for a small set of
stative activity predicates, e.g., poležati ‘lie for a while’, posěděti ‘sit for a
while’ and postojati ‘stand for a while’ – the very same predicates for which
A prototype account of the development of delimitative po- in Russian 365
12. Note that Veyrenc (1968) characterizes the distinction between MRus po- and
pro- as that between “extension relative” and “extension absolue du procès”,
which I take to be an independent confirmation of the relevance of RELATIVE
DELIMITATION and ABSOLUTE DURATION for these prefixes in (the history of)
Russian.
13. I assume that this situation was partially the result of the incorporation of an
etymologically distinct SOURCE meaning into the network of the prototypical
PATH meaning, but this issue is not immediately relevant for the present dis-
cussion.
14. Note the stipulation prototypical. It is hard to find detailed commentary on the
actionality (and perfectivity) of MRus pojti. As an ingressive verb, it is ordi-
narily assumed to reach its telos “immediately, when the movement starts,” as
one anonymous reviewer phrases it. However, Zaliznjak and Šmelev (2000:
109) observe that “the ingressive meaning of such verbs is attenuated,” as
phrases such as pojti v kino ‘go to the movies’ or poexat' v Pariž ‘drive to
Paris’ do not mean ‘begin to go’ but refer to the actions “as such”. Zaliznjak
and Šmelev attribute this effect to a kind of metonymy whereby the beginning
of the action represents the entire action. While this could be the case,
Zaliznjak’s and Šmelev’s view comports with Shull’s (2003) hypothesis that
spatial po- indicates the initiation of the trajectory plus some unspecified
amount of the background trajectory: if the very beginning of a trajectory can
metonymically represent an entire trajectory, then the beginning of a trajectory
plus some additional non-zero amount of its traversal must certainly also be
able to represent the entire trajectory metonymically. Moreover, there are
several reasons to believe that spatial po- with determinate motion verbs
expresses more than merely the beginning of the action. First, as Shull (2003:
149–150) points out, pojti need not refer to the initiation of motion, but can
refer to the initiation of a new trajectory within a larger motion event, cf., e.g.,
her example Sobaka podbežala k kovru, pobežala po kovru i ubežala dal'še
‘The dog ran up to the carpet, ran [PO-ran] on the carpet and ran on.’ In this
respect, as Shull (ibid., 155–157) observes, spatial po- differs markedly from
the main MRus inceptive prefix za-, as the latter must express the transition
from the lack of an activity to the activity, e.g., zabegat' ‘start running
[transition from the state of not running to the state of running]’. Second, when
pojti is reduplicated it expresses duration in a fashion resembling true
delimitatives in po- (see examples 10–11), which is very hard to explain if
spatial po- does not express something more than the absolute beginning of the
action; note that other inceptive verbs such as zabegat' ‘begin to run’ and
momentary verbs such as skočit' ‘jump’ are not reduplicated to express
duration.
15. Note that Aitzetmüller (1991: 171) has also suggested a general connection
between the spatial and temporal meanings of po- in the evolution of the
A prototype account of the development of delimitative po- in Russian 367
Slavic aspectual system, cf. his example in (12) below. I am grateful to Tho-
mas Daiber for making me aware of Aitzetmüller’s view.
16. The development of this new class of pf verbs was also a contributing factor in
a significant change in the semantics of the pf aspect in Russian, which unfor-
tunately cannot be considered here; for discussion, see Dickey and Hutcheson
(2003) and Dickey (2005).
17. This view is based on numerous considerations that cannot be taken up here.
Briefly, we may say that, as Stern (2002) and Greenberg and Dickey (2006)
observe, in OCS and ORus the putative “indeterminate” motion verbs did in
fact occur in determinate contexts (e.g., i sam" si nosę kr'st' izide v" nari-
caemoe kranievo město, eže glagolet" sę evreisky gol"gafa ‘and himself bear-
ing his cross he went out to the place called the place of the skull, which in
Hebrew is called Golgotha’). Such usage is much easier to explain if such
verbs were manner of motion verbs, as opposed to strictly “indeterminate”. It
is also worth noting that Veyrenc (1966) suggests a similar situation in MRus,
arguing that “indeterminate” motion verbs such as xodit' ‘walk, go’ represent a
kind of activity verb (verbes du fonction) as opposed to genuine motion verbs
(verbes du déplacement).
18. These forms resemble conjugated forms of pf determinate motion verbs pre-
fixed in po- in the East Slavic languages (e.g., Rus pf pojdu ‘I will go’ < pojti
‘go’), but are future tense forms of the Cz unprefixed impf verbs. Note that,
with the exception of jít ‘go’, there are synonymous compound future tense
forms for these verbs (e.g., budu jet alongside pojedu ‘I will drive/be driv-
ing’); cf. Kopečný (1962: 47) on the “absolute” synonymy of the Czech impf
compound and po- future tense forms. It should be pointed out that though
such po- futures are undoubtedly impf, they do not necessarily emphasize an
ongoing process on a par with English progressive forms (e.g., ‘s/he will be
going’); rather, they are the default future tenses of Czech impf motion verbs,
and thus have been translated with English simple-tense forms.
19. Shull (2003: 160–164) argues that her schema for po- given in figure 3 also
accounts for resultatives, in that po- contributes merely the notion that some
amount of the trajectory denoted by the verb is completed, and the construal
expressed by the source verb itself determines the telicity or atelicity of the
po- derivative (i.e., stroit' ‘build’ is a telic notion, whereas sidet' ‘sit’ is atelic).
I think that this is a viable approach, but cannot consider the issue in the nec-
essary detail here. Suffice it to say that it seems possible to integrate the sur-
viving resultative verbs in the hypothesized network for MRus po- as marginal
cases, but cases that are nonetheless incorporated in the network in a princi-
pled way.
20. In this respect, it should be pointed out that po- has lost its spatial meaning in
Ukrainian (cf. Šerex 1951: 292) and Bulgarian (cf. Ivanova 1966: 124), as
well as in Polish (cf. Śmiech 1986: 18). Thus, it appears that in the individual
Slavic languages there is a three-way correlation between the prefixation of
368 Stephen M. Dickey
determinate motion verbs with po-, the productivity of delimitative po-, and
the lack of a spatial meaning for po-. If this is true, it cannot be a mere coinci-
dence.
21. Another remnant of this older situation seems to be Bulgarian poida ‘go for a
short time’, given in the RSBKE 2: 599. The verb is tagged ‘rare’, and judged
to be dialectal by Bulgarian informants.
References
Abdoulaye, Mahamane L.
2001 The grammaticalization of Hausa zaa ‘be going’ to future. Journal of
African Languages and Linguistics 22 (1): 1–32.
Aitzetmüller, Rudolf
1991 Altbulgarische Grammatik als Einführung in die slavische
Sprachwissenschaft. Freiburg: U. W. Weiher.
Bermel, Neil
1997 Context and the Lexicon in the Development of Russian Aspect.
Berkeley: University of California Press.
Bondarko, A. V.
1961 K voprosu o glagolax dviženija v češskom jazyke (formy tipa
ponesu, poletím) [On the question of verbs of motion in Czech
(forms of the type ponesu, poletím)]. Slavia 30: 527–547.
Bulygina, T. V.
1982 K postroeniju tipologii predikatov v russkom jazyke [Toward con-
structing a typology of predicates in Russian]. In Semantičeskie tipy
predikatov [Semantic Types of Predicates], O. N. Seliverstova (ed.),
7–85. Moscow: Nauka.
Bybee, Joan, Revere Perkins, and William Pagliuca
1994 The Evolution of Grammar. Tense, Aspect and Modality in the Lan-
guages of the World. Chicago: University of Chicago Press.
Camus, Remi
1998 Quelques considérations sur le préverbe po- en russe contemporain.
Revue des Études Slaves LXX (1): 101–112.
Čertkova, Marina
1996 Grammatičeskaja kategorija vida v sovremennom russkom jazyke
[The Grammatical Category of Aspect in Contemporary Russian].
Moscow: Izdatel'stvo Moskovskogo universiteta.
Dahl, Östen
1985 Tense and Aspect Systems. Oxford: Blackwell.
Dickey, Stephen M.
2000 Parameters of Slavic Aspect: A Cognitive Approach. Stanford: CSLI.
A prototype account of the development of delimitative po- in Russian 369
Shull, Sarah
2003 The Experience of Space: The Privileged Role of Spatial Prefixa-
tion in Czech and Russian. Munich: Otto Sagner.
Sigalov, Pavel S.
1975 Istorija russkix ograničitel´nyx glagolov [The history of Russian
delimitative verbs]. Trudy po russkoj i slavjanskoj filologii: Serija
lingvističeskaja [Studies in Russian and Slavic Philology] 24: 141–
181.
1977 Istorija russkix perdurativnyx glagolov [The history of Russian per-
durative verbs]. Trudy po russkoj i slavjanskoj filologii: Serija ling-
vističeskaja [Studies in Russian and Slavic Philology] 23: 101–118.
Śmiech, Witold
1986 Derywacja prefiksalna czasowników polskich [The Prefixal
Derivation of Polish Verbs]. Wrocław: Zakład narodowy im.
Ossolińskich.
Stern, Ariann Nicole
2002 The Verbs of Motion in Old Russian Texts: A Comparative Gram-
matical Analysis of a Nascent Verb Class. Ph.D. diss., University of
California at Los Angeles.
Strekalova, Zoja Nikolaevna
1968 Iz istorii pol'skogo glagol'nogo vida [From the History of Polish
Verbal Aspect]. Moscow: Nauka.
Šerex, Jurij
1951 Narys sučasnoji ukrajins'skoji literaturnoji movy [A Sketch of Con-
temporary Literary Ukrainian]. Munich: Naukove tovarystvo im.
Ševčenka.
Tabakowska, Elżbieta
1999 ‘Pobłądzić po malowniczych zaułkach Starego Miasta’: Semantyka
polskiego przyimka po i przedrostka po- [‘To roam around the pic-
turesque lanes of Stare Miasto’: The semantics of the preposition po
and the prefix po-]. In Collectanea Linguistica in Honorem Casimiri
Polański, Maria Brzezina and Halina Kurek (eds.), 269–278.
Kraków: Księgarnia Akademicka.
2003 Space and time in Polish: The preposition za and the verbal prefix
za-. In Motivation in Language. Studies in Honor of Günter Radden,
Hubert Cuyckens, Thomas Berg, René Dirven, and Klaus-Uwe Pan-
ther (eds.), 153–177. Amsterdam: John Benjamins.
Taylor, John R.
1995 Linguistic Categorization. Prototypes in Linguistic Theory. Oxford:
Clarendon Press.
A prototype account of the development of delimitative po- in Russian 373
Tixonov, Aleksandr
1998 Russkij glagol: Problemy teorii i leksikografirovanija [The Russian
Verb: Theoretical and Lexicographical Problems]. Moscow: Akade-
mia.
Veyrenc, Jacques
1966 Russe idti et xodit': mouvement de déplacement et mouvement de
fonction. Bulletin de la Société de Linguistique de Paris 61 (1): 21–
41.
1968 Expansion syntaxique et classement des préverbes. Slavica 8: 245–
249.
Voloxina, G. A., and Z. D. Popova
1997 Kategorija glagol'nogo vida v svete semantičeskogo ustrojstva glago-
l'nyx pristavok [The category of aspect in light of the semantic orga-
nization of verbal prefixes]. In Trudy aspektologičeskogo seminara
Filologičeskogo fakul'teta MGU im. M. V. Lomonosova [Studies of
the Aspectological Seminar of the Philological Faculty of Moscow
State University]. Vol. 3, M. Ju. Čertkova (ed.), 34–41. Moscow: Iz-
datel'stvo Moskovskogo universiteta.
Zaliznjak, Anna A., and Aleksej D. Šmelev
2000 Vvedenie v russkuju aspektologiju [An Introduction to Russian As-
pectology]. Moscow: Jazyki russkoj kul'tury.
Sources
Eleni Bužarovska
Abstract
The article focuses on the semantic change of the indefinite pronoun nešto
into an epistemic mitigation modal in Macedonian, as part of a wider Balkan
Slavic context. The presence of the secondary nešto in other South Slavic
languages suggests that the development of the pronominal nešto into a
mitigation marker represents a motivated change based on some universal
conceptual mechanisms. The author explores the possible paths of this
change and suggests that strengthening of invited inferences and subjectifi-
cation were the two cognitive mechanisms that played a major role in this
metonymically-based process. Due to contextual inferences, the schematic
referential nešto was subject to change in colloquial speech. The result of
this semasiological change was an epistemic approximative modal nešto that
functions as a modal of epistemic stance to encode the speaker’s slight de-
crease of the assertoric force of the proposition. The theoretical framework
of the analysis of the semantic change of nešto is based on the Taylor’s
model of schema-instances relations (Taylor 2002) and Traugott and
Dasher’s (2002) model of invited inferencing theory of semantic change
(IITSC). The former is used to account for the development of the polyse-
mous, secondary nešto, whereas the latter explains the rise of several epis-
temic meanings from the secondary nešto in specific contexts. On this view,
the development of the secondary nešto represents a case of coded inferen-
tial change that resulted in the creation of two pragmatic utterance-type
meanings of the secondary nešto: approximative quantification and epis-
temic modal. Via metonymic extension the utterance-type meanings have
further developed into nine interrelated utterance-token, context-dependent
meanings.
Keywords: Balkan Slavic, mitigation marker, inferencing, subjectification
376 Eleni Bužarovska
1. Introduction
This article investigates the change of the Balkan Slavic indefinite pronoun
nešto into an epistemic pragmatic marker. In colloquial usage nešto has
undergone a major semantic shift into a non-assertoric epistemic particle
that marks approximation and mitigation as the following Serbian (1), Bul-
garian (2) and Macedonian (3) examples illustrate:1
The comparison between examples (1), (2) and (3) and the correspond-
ing minimal pairs without nešto (1’), (2’), and (3’) shows that speakers,
with the help of nešto, “soften” their assertions. Thus, the mitigation sense
of nešto in examples (1), (2), and (3) is rendered in English by three differ-
ent lexemes: ‘somewhat’, ‘might’ and ‘sort of’.
The pronouns nekoj ‘someone’ and nešto ‘something’ are the central expo-
nents of the category of reference in standard Macedonian. Out of two
existing universal ontological categories, humans and objects, nešto desig-
nates objects. In contrast to nekoj, nešto exhibits a higher degree of ab-
stractness due to the lack of human reference. Nekoj ‘someone’ is marked
for both gender and number; its function in an NP is to carry information
The rise of an epistemic pragmatic marker in Balkan Slavic 379
about a human referent that the speaker is unable to identify. The indefinite
pronoun nešto, marked for neuter gender, denotes things and belongs to the
category of NPs.6 The NPs that contain the pronominal nešto carry infor-
mation that the topic of conversation is unidentified both to the speaker and
to the other participants of the speech situation. On the syntactic plane,
nešto functions as a phrasal head:
The reference of nešto is not limited only to things, but also extends to
any indeterminate entity such as events and objects of thoughts. The pro-
nominal nešto pertains to a physical object in (5), to an event in (6) and to
a proposition in (7).
In examples (5), (6), and (7), the pronominal nešto invokes the notion of
‘an indeterminate entity’ as its underspecified lexical meaning makes it
schematic both for things and events.7 The full interpretation of nešto de-
pends on the context and knowledge of the speech event circumstances.
Because the pronominal nešto lacks specific meaning and has an extremely
wide scope of reference, in certain contexts it can create specific pragmatic
inferences of the speaker’s slight doubt in the truth of the assertion. In such
contexts, the speaker’s avoidance to provide any additional characteriza-
tion of the thing/event s/he is talking about (coded by nešto) might imply
that s/he epistemically evaluates the event. For instance, when the speaker
utters sentences like I see something, He knows something or Something
happened the hearer infers that the speaker may imply “I think so, but I do
not claim it”. Namely, the “failure” of the pronominal nešto to express the
specifications of the profiled event produces pragmatic inferences of a
weak epistemic force.8 Presumably the speaker’s ignorance or unwilling-
380 Eleni Bužarovska
According to Croft and Cruse (2004: 20), the dictionary view of linguis-
tic meaning generally describes only the concept’s profile. Due to the cen-
tral position of profile in determining a word’s meaning, any shift in pro-
file has truth-conditional consequences.
In the case of secondary nešto the profile of indeterminate entity has
been shifted to a different semantic frame, i.e. of events in the mental
world of the speaker. In other words, nešto underwent semantic change due
to profile shifting from denoting an entity (in the real world) that the
speaker conceives as indeterminate to the very relation of indeterminacy
The rise of an epistemic pragmatic marker in Balkan Slavic 381
between a conceived entity and the speaker. The result of this semantic
change was the rise of two secondary polysemous nešto, syntactically and
semantically different. The first nešto signifies a small approximative
degree, whereas the meaning of the second nešto may be described as epis-
temic approximation because it expresses slight uncertainty of the speaker
with respect to the epistemic status of the proposition. Subsequently, these
two schemas in different contexts created a network of several related
meanings that will be analyzed in the following sections.
The analysis of the collected examples has helped identify four quantifica-
tion functions of nešto. They occur in different contexts and all four are
characterized by the inferential meaning of ‘smallness’ or ‘paucity’ and
approximative quantification. The reason for the contextual extensions of
nešto lies in its simple semantic matrix. Langacker claims that “[t]he ma-
trix for most predicates is complex, requiring specifications in numerous
domains both basic and abstract” (1987: 163). Since nešto does not have
specifications in any of the domains, it is prone to semantic shift in con-
texts where the profiled entity lacks quantitative specification. In addition,
the contextually dependent quantification meaning of nešto is enriched
with a “paucal” approximation component.
Below follows the classification of the approximative nešto according
to its function in different phrases. The quantifier nešto has developed four
contextual meanings depending on its syntagmatic co-occurrence. As an
indefinite quantifier of the head in a comparative AdjP, nešto designates an
approximative small degree of difference in a shared property; with NPs,
nešto denotes an approximative small quantity of an object or a substance;
with temporal PPs, nešto approximates the time specification; and with
VPs, it quantifies a small temporal scope over an event.
entities that possess the same property in different degrees. The approxima-
tive nešto codes a small indeterminate difference between the degrees of
the shared property in the two entities.
In (11) the request without nešto is more direct, while the omission of
nešto in (12) conveys the meaning that the speaker’s expectations of find-
ing food in the fridge are slightly higher.
The rise of an epistemic pragmatic marker in Balkan Slavic 383
3.2.4. Inferences
The secondary nešto comprises a series of modal meanings that denote the
speaker’s decreased assertoric force. These meanings arise when the
speaker makes motivated semantic extensions from the conventionalized
secondary meaning in specific contexts. They should be viewed as contex-
tual variants because they all express the speaker’s “weakened” epistemic
force but at the same time they include “particulars of the speech situation
that are not linguistically coded” (Langacker 1987: 157).12 They differ with
respect to the salience of inferential meanings that the epistemic nešto cre-
ates in different contexts due to these “particulars”. Given the similarity of
these meanings, it is important to explore their hierarchical organization
and the mechanisms that triggered the change of one meaning into another.
The classification of the compiled examples produced five senses of the
epistemic nešto. These senses were established on the basis of semantic-
pragmatic criteria and the type of contexts in which the epistemic nešto
occurs.
The rise of an epistemic pragmatic marker in Balkan Slavic 387
b. Nešto se zamerija pa ne se
something REFL anger.1PL.AOR and not REFL
družat.
socialize
‘They seem to be angry with each other, so they don’t socialize.’
The target meaning of mitigation again results from the semantic exten-
sion of the source meaning: approximative quantification and insufficient
circumstantial knowledge. The speaker implies that s/he does not know
why the event characterized by some property possesses this property to
the extent (or degree) “quantified” by the adverbial. In addition to the “cir-
cumstantial” inference “I don’t know why p is true”, the invited inference
‘it seems to me’ arises in this context. The English translation of nešto with
the verb seem is also indicative of the “appearance” inference. The blend-
ing of these two inferences into “I don’t know why p is true but it seems to
me that p is true” produces the mitigation sense of nešto.13
With second person subjects, nešto serves to soften the disapproval of
the manner in which the addressee performs the activity. In such contexts
nešto often co-occurs with the dative short pronoun mi ‘to me’, the so
390 Eleni Bužarovska
The sentence initial nešto conveys the speaker’s decreased assertoric force
about situations involving his/her physical or mental state.14 The inferential
meaning ‘it seems to me’ arises in propositions that refer to the mental or
emotional world of the speaker. In contexts with first person subjects nešto
acquires the “appearance” inference ‘it seems to me’. The new inference,
in turn, fuses with the source meaning ‘insufficient causal or circumstantial
knowledge’ inherited from the schema of the epistemic “circumstantial”
nešto. Like in the previous type, this contextual meaning also consists of
two inferences: “circumstantial” and subjective appearance. However, it
differs from the mitigating nešto in two respects: the quantification infer-
ence is missing and the “circumstantial” inference is stronger than that of
“appearance”. Due to the strengthening of the ‘it seems to me’ inference,
nešto conveys the meaning ‘for some reason I think that p is true, but I
don’t assert that p’. Therefore the non-assertoric epistemic function of
nešto is similar to the function of a pragmatic assertoric mitigator, but it
has not fully become one.
This highly subjective, mitigating nešto has derived from the attitudinal
epistemic nešto in the new context involving events in the real world via
the strengthening of the “appearance” (‘it seems to me’) inference. The
difference from the attitudinal nešto lies in the reverse salience of the two
inferences that constitute its meaning: “appearance” and “circumstantial”
indeterminacy. The latter has faded in this context at the expense of the ‘it
392 Eleni Bužarovska
seems to me’ inference. In (39) nešto signals that the proposition that refers
to the hearer or another participant of a speech situation expresses the
speaker’s viewpoint and hence does not have to be true. In (40) this mean-
ing is enriched by the inference ‘he worries me’ coded by the dative mi.
signals both the message force and the content. They belong to aspects of
meaning subjectively construed as they linguistically encode “clues which
signal the speaker’s potential communicative intentions.” (Fraser 1996:
168). The aspect of meaning is subjectively construed in all five epistemic
senses of nešto, but to a different degree. Thus, all contextual instances of
nešto except the “circumstantial” one carry the inferential meaning ‘it
seems to me’, which is more strongly felt in the last two due to the higher
degree of subjectification.
5. Semantic change
It was assumed earlier that the rise of the epistemic nešto was motivated by
the semantics of the pronominal nešto and R-heuristic principle. According
to the containment theory (Willet 1998), the semantic structure of the
source element contains the meaning of the target element. In the process
of the derivation of some lexical unit the basic meaning is preserved in the
semantics of the target element. In modal nešto the dominant meaning of
indeterminacy and the inference of “small” quantification blends with the
meaning of subjective impression. The rise of inferential meaning of small
quantification was already explained in section 2.5. By using nešto the
speaker signals that the small degree of uncertainty in the truth of the
proposition comes from his/her own impression. While the pronominal
nešto is characterized by referential indeterminacy in the real world, the
epistemic nešto expresses “subjective” indeterminacy which involves the
mental world of the speaker. Accordingly, this semantic change triggers the
conversion of nešto from an indefinite pronoun to an epistemic degree
adverbial.
Although the semantic change of nešto involves mapping between enti-
ties in two different domains, and therefore is an example of how a mem-
ber of one category is represented in terms of another, there are grounds to
The rise of an epistemic pragmatic marker in Balkan Slavic 395
5.3. Frequency
Np (prototype) Ns (extension)
indeterminate indeterminate
(small) thing relation
N schema
indeterminacy
approximation
Nq (quantifier) Nm (modal)
N1 compar.
quantifier N3 temporal N4 event N5 epistemic of
quantifier quantifier indeterminate cause
N7 attitudinal N6 mitigator
N2 nominal epistemic of modifier
quantifier
N9 hedge N8 mitigator
Figure 2 can be further elaborated in two tables that show the change of
meaning of approximative and epistemic senses of nešto, a change trig-
gered by the presence of context-dependent invited inferences.
The approximative nešto results from inferences of approximation
and/or small degree/quantity that arise in particular contexts: with com-
parative AdjP/AdvP, nešto becomes an approximative small degree quanti-
400 Eleni Bužarovska
7. Conclusion
Notes
the rare exception of example 4). The absence of written historic evidence has
restricted our treatment of nešto to synchronic analysis. The corpus of sen-
tences with the secondary nešto consists of 300 examples collected from vari-
ous chat rooms on the internet. It also includes examples taken from conversa-
tions or interviews that the author has conducted with students and friends.
3. Traugott and Dasher’s (2002: 16–49) IITSC model is based on semanticiza-
tion of pragmatic implicatures of a lexeme in new specific contexts. On this
view, semantic changes arise from an usage-based process in which speakers
exploit invited inferences of a lexeme and “re-analyze” context-dependent
implicatures to the point where the lexeme acquires a new meaning. The se-
mantic change from the source meanings to new coded meanings goes through
two intermediate stages (a) utterance-token meanings (or invited inferences,
IINs) and (b) utterance-type meanings (generalized invited inferences, GIINs).
Invited inferences that arise in context become generalized invited inferences
(akin to generalized conversational implicatures) after they undergo strength-
ening in specific contexts. The GIINs represent the preferred, pragmatically
polysemous meanings that have not been crystallized into new coded mean-
ings. Pragmatic strengthening is achieved via associations and metonymy
rather than analogy and metaphor.
4. The term ‘invited inferences’, borrowed from Geis and Zwicky (1971), is used
here to denote deductive processes of thought – inferences – that arise in spe-
cific contexts. They are “invited” in the sense that they are suggested by the
context. Traugott and Dasher (2002: 5) point out that they do not restrict the
term “invited inferences” to generalized implicatures. It has a broader inter-
pretation in their IITSC model as “[i]t is meant to elide the complexities of
communication in which the speaker/writer evokes implicatures and invites
the addressee/reader to infer them.”
5. Taylor’s model is based on Langacker’s network model of knowledge struc-
tures (1987: 162) according to which a knowledge system can be described as
a network of nodes and arcs. The nodes correspond to conceived entities,
while the arcs correspond to the conceived relationships between the entities.
6. Topolińska distinguishes pronominal and nominal nešto; the latter has nomi-
nal categorial properties and can be pluralized: Mnogu nešta go grizea nego-
voto srce ‘A lot of things tormented his heart’ (1974: 168). The process oppo-
site to the conversion of a pronoun to a nominal category is noted by Heine
and Kuteva (2001: 196). They underline the close relation between indefinite
pronouns and nouns denoting ‘thing’ cross-linguistically and argue that “[t]his
grammaticalization appears to be a more general process whereby generic
nouns give rise to pronominal categories.”
7. Van Kemenede (1999: 1001) finds that “[t]hose elements that have under-
specified lexical meaning are more prone to grammaticalization than others.”
8. The term ‘epistemic force’ is used here to denote the degree of belief with
which one asserts the content of an assertive speech act (cf. Sweetser 1991:
The rise of an epistemic pragmatic marker in Balkan Slavic 403
84). For indirect-quote constructions Givón (1990: 531) uses the term ‘epis-
temic commitment’.
9. Relations are conceptually dependent because they cannot be conceptualized-
without also conceptualizing the entities they interconnect. Langacker (1987:
215) maintains that a nominal predication profiles a thing, i.e. a region in
some domain, whereas a relational predication puts interconnections in pro-
file, rather than presupposing them as part of the base.
10. The symbols (N1), (N2) etc stand for the nine secondary meanings of nešto
(N), represented in Figure 2.
11. I would like to thank the editors for this insightful suggestion.
12. According to Langacker (1987: 157) the contextual meaning of an expression
represents: “[t]he richly detailed conceptualization that constitutes our full
understanding of the expression and includes all relevant aspects of the con-
ceived situation.”
13. The view that inferences, as part of the inherent meaning of a lexeme, can be
coded into a new meaning was expressed by Bolinger in the early 1970s: “It is
probably a fact, that in the course of time, inferences become references”
(1971: 522). Traugott and Dasher (2002: 5) by using the term ‘invited infer-
ences’ emphasize the active role of the speaker in rhetorical strategizing of the
communicative act. Therefore, in the treatment of nešto the term ‘invited in-
ferences’ or ‘inferences’ is preferred to ‘context-induced inferences’.
14. The Russian indefinite pronoun čto-to ‘something’ may also assume the func-
tion of an attitudinal epistemic especially in negative assertions to “soften” the
force of personal feelings or attitudes:
15. Examples (36–38) without the dative mi ‘to me’ are characterized by the pres-
ence of two inferences: speaker’s impression ‘it seems to me’ and mode of
knowledge ‘I don’t know why’. The clitic mi enriches the implicational mean-
ing of examples (36–38) with the “empathy” inference: ‘I feel sorry for you’
(36), ‘that worries me’ (37), and ‘I feel happy for you’ (38).
16. In Macedonian and Serbian corpus examples, the dative mi typically co-
occurs with statal predicates such as upset, familiar, suspicious, tired, pale,
confused, etc. In Macedonian example (i) mi conveys the speaker’s compas-
sion or involvement; mi can reinforce the irony of the whole utterance as illus-
trated in Serbian example (ii). It is worth noting that there were no sentences
with “appearance” nešto and mi among Bulgarian examples.
17. Schwenter and Traugott (2000: 16) note that any lexeme recruited into an
epistemic modal domain acquires scalar values. On how non-scalar expres-
sions acquire scalar values, see König’s (1991) treatment of the particle even.
18. Heine et al. (1991) consider metaphorical extension as the main mechanism of
semantic change. Similarly, Sweetser (1991: 19) underlines the role of meta-
phor in the creation of polysemy patterns based on metaphorical intra-domain
connections. Arguing that “[w]e cannot escape using the general concept of
multiple domain-structures in our analysis of linguistic meaning” (1991: 147),
she distinguishes three different domains represented by sentences: real world
content, epistemic and speech acts. The latter two domains are structured in
terms of the basic content domain (1991: 146). However, Traugott and Dasher
argue (2002) that metaphorisation often is an outcome of metonymical
change. They note that “[n]either conceptual metaphorisation nor conceptual
metonymisation in principle exclude each other: easily comprehended meta-
phors are consistent with typical associations; both exploit pragmatic mean-
ing; both enrich meaning” (Traugott and Dasher 2002: 29). Similarly, the
metaphorisation of the pronominal nešto into an epistemic modal may have
metonymical origins in inferences arising in particular contexts: the link
between the notion of referential indeterminacy and the notion of epistemic
uncertainty may become metaphorical through the process of decontex-
tualization and conventionalization.
19. The use of dative in highly subjective contexts is not accidental. Manoliu
(2001: 304) claims that “[t]he dative case carries a special pragmatic connota-
tion, since it points to the most salient constituent after the Agent.”
20. The change of nešto qualifies as grammaticalization rather than lexicalization.
Both processes are historical changes that result in the production of new
forms: grammaticalization in functional and lexicalization in lexical forms
(see Brinton and Traugott 2005: 96–97). Lexicalization usually involves fu-
sion of a compound or a syntactic construction into a new lexical, contentful
form. The inputs of lexicalization are semantically highly specified items, a
criterion that excludes nešto from being considered an eligible candidate for
lexicalization change.
21. According to Brinton (1996: 38), “[t]he interpersonal mode is the expression
of the speaker’s attitudes, evaluations, judgments, expectations and demands,
The rise of an epistemic pragmatic marker in Balkan Slavic 405
as well as of the nature of the social exchange, the role of the speaker and the
role assigned to the hearer.”
References
Bolinger, Dwight
1971 Semantic overloading: a study of the verb ‘remind’. Language 47:
522–547.
Brinton, Laurel J.
1996 Pragmatic Markers in English. Berlin/New York: Mouton de
Greyter.
Brinton, Laurel J., and Elizabeth C. Traugott
2005 Lexicalization and Language Change. Cambridge: Cambridge Uni-
versity Press.
Bybee, Joan, William Pagliuca, and Revere D. Perkins
1991 Back to Future. In Approaches to Grammaticalization. Vol. 2,
Traugott and Heine (eds.), 17–58. Amsterdam: John Benjamins.
Bybee, Joan, and Sandra A. Thompson
2000 Three Frequency Effects in Syntax. Berkeley Linguistic Society 23:
65–85.
Bybee, Joan
2003 Mechanisms of Change in Grammaticalization: The Role of Fre-
quency. In The Handbook of Historical Linguistics, B. D. Joseph and
J. Janda (eds.), 602–623. Oxford: Blackwell.
Croft, William, and D. Alan Cruse
2004 Cognitive Linguistics. Cambridge: Cambridge University Press.
Fraser, Bruce
1996 Pragmatic Markers. Pragmatics 6: 167–190.
Fischer, Olga, and Anette Rosenbach
2000 Introduction. In Pathways of Change. Grammaticalization in Eng-
lish, Olga Fischer, Anette Rosenbach, and Dieter Stein (eds.). Am-
sterdam: John Benjamins, 1–37.
Geis, Michael L., and Arnold M. Zwicky
1971 On invited inferences. Linguistic Inquiry 2: 561–566.
Givón, Talmy
1990 Syntax. A Functional-Typological Introduction. Vol. 2. Amsterdam:
John Benjamins.
Haspelmath, Martin
2003 The geometry of grammatical meaning: semantic maps and cross-
linguistic comparison. In The New Psychology of Language. Vol. 2,
406 Eleni Bužarovska
Elżbieta Tabakowska
What is uppermost in mind comes up first
(Dwight Bolinger)
Abstract
The paper deals with the ordering of multiple (mainly double) adjectival
modifiers within Polish nominal phrases. Linear ordering of constituents
within those structures, generally considered haphazard and erratic, is shown
to be motivated. Two basic assumptions are made: (i) although in the dia-
chronic development of language the motivation can be bleached and the
particular structure sanctioned by convention, linear arrangement of ele-
ments within an expression is always motivated, and (ii) the motivation is
basically iconic. The structure of NPs is shown to depend on the traditional
division of adjectives into two categories: the characterising (attributive)
and the specifying (restrictive). The borderline between the two, however, is
fuzzy, with the allotment of an adjective to one or the other category de-
pending on communicative needs, which are often discourse-sensitive. In the
speaker’s choice of a construal, iconic principles conspire (or are overrid-
den) for optimum effect.
1. Introduction
lation of principles (rather than rules sensu stricto) that are basically ex-
pressive (or iconic) in nature. The latter are considered to be more univer-
sal because of the nature of mechanisms that motivate them (cf. Haiman
1985, ch. 6). In spite of its search for language universals, highly formal-
ised contemporary mainstream linguistics, firmly rooted in Saussurean and
American structuralism, has been traditionally focusing on rules of the first
type. This general approach applies also to what constitutes the topic of the
present paper, that is, the order of constituents within the nominal phrase.
The area of interest is fairly narrow: the following discussion will focus on
the ordering of multiple (mainly double) adjectival modifiers within Polish
nominal phrases. It could indeed be hard to justify this limitation, unless it
can be claimed that it might have some more general implications. And this
is precisely the claim that the present paper makes: the case, limited as it is,
clearly demonstrates that there are linguistic phenomena which – seen from
the traditional objectivist point of view – seem haphazard and erratic, or at
best “extremely troublesome” (cf. Łuczyński 1993), but when they are
analysed in a different, wider perspective, obvious regularities can be re-
vealed. In the case under discussion, these regularities stem from iconic
motivation.
In recent Polish works that deal with word order phenomena in general,
and with linear ordering of modifiers within the nominal phrase in particu-
lar, two main trends may be observed: analyses focus either on syntax,
aiming at formulating grammatical sequencing rules, or on the rhematic-
thematic (topic-comment) sentence structure (for a discussion of these two
general directions, see Zakrzewska 2001: 661). Within the first group, one
of the most comprehensive descriptions was offered by Topolińska (1984:
381 ff.), whose basic assumption is that sequencing of elements within the
NP reflects the “derivative history” of the phrase (cf. also Gębka-Wolak
2000: 139). However, the source of direct inspiration for the present paper
was a recent study of the problem, contained in chapter 3 of the monograph
written by a Polish linguist of generativist persuasion, Dorota Szumska (cf.
2006: 166 – 176).
Like Topolińska, Szumska gives a detailed and insightful description of
what things are like, without, however, answering the question why they
are as they are. Numerous works by Polish authors – older and more recent
alike (cf. e.g. Śliwiński 1990, Łuczyński 1993, Gębka-Wolak 2000) –
explicitly postulate the need for integrating the syntactic, the semantic and
the pragmatic perspectives as a necessary prerequisite to any adequate de-
scription of word order phenomena. Moreover, careful reading of these
works reveals that many authors do in fact refer, albeit only implicite, to
Iconicity and linear ordering of constituents within Polish NPs 413
what is defined as principles of iconicity. There are also some who are
quite explicit: in Zakrzewska’s discussion of functional explications of
linear ordering within Polish grammatical structures, iconicity appears as
one of “the four most important principles that govern word order in Pol-
ish” (2001: 668; translation – E.T.). As will be demonstrated by the follow-
ing discussion, the other three factors can be indeed classified as various
manifestations of iconicity.
The increasing bulk of evidence from linguistics, psychology and cogni-
tive science has been showing that iconicity, defined as “form miming
meaning”, or the perceived similarity between the shape of a sign and the
object that this sign stands for, is one of the basic mechanisms that underlie
cognitive processes taking place in the human mind and that it manifests
itself in linguistic expressions as a reflection of those processes in human
language. Iconic motivation is both universal and ubiquitous in natural
languages, and thus it is only natural to expect that its workings will be-
come apparent also in word order phenomena.
Iconic motivation of linguistic structure – and of linear arrangement of
constituents in particular – has recently attracted the attention of linguists
of the cognitive persuasion. In general, discussions that deal more or less
explicitly with iconic motivation as a mechanism governing word order
focus on the word order of main sentence constituents in SVO vs. SOV
languages (cf. e.g. Hopper and Traugott 2003, Kubiński 1999). As far as
our present area of interest is concerned, the noun-adjective ordering was
described for Old English by a Dutch historical linguist Olga Fischer
(2000), and for Modern English, briefly but insightfully, by the German
cognitive linguist Guenter Radden (1991). In the same cognitivist vein, the
ordering of modifiers within Modern Polish NPs was dealt with by Taba-
kowska (2001a); for Modern Italian, a similar study (though not openly
“cognitive” in the theoretical approach) was carried by Magajewska
(2006).
In spite of numerous attempts made within various theoretical frame-
works, principles governing the ordering of multiple (double) adjectival
modifiers within Polish NPs have not been systematically described, and
the question remains basically unresolved. The present paper is yet another
attempt at providing an answer. Underlying the analysis are two fundamen-
tal assumptions: (i) linear arrangement of elements within an expression is
always motivated (although in the diachronic development of language the
motivation can be bleached and the particular structure sanctioned by con-
vention), and (ii) the motivation is basically iconic. Before we proceed to
the actual analysis one important proviso should be made: “[the] act of
414 Elżbieta Tabakowska
interpretation must always proceed from meaning to form and not the other
way round” (Nänny and Fischer 1997: 2): motivation that underlies the
form of an expression can only be searched for after the meaning of the
expression had been established.
2. Discussion
The generally accepted semantic rule governing the linear ordering of ad-
jectival modifiers within Polish NPs, whereby characterising (attributive)
modifiers occur in preposition and specifying (restrictive) modifiers in
postposition, is believed to reflect mental processes of categorisation. The
assumption is that categories are arranged in a hierarchical system, with the
logical relation of inclusion holding between categories of higher and
lower levels. This model, going back to the Aristotelian system of categori-
sation, is fundamental for generative semantics. Applied to the traditional
classification of Polish adjectives, it also underlies the basic dichotomous
division of adjectives into those that “characterise” objects to which they
refer and those which perform the function of “specification”. The former
are used to describe ad hoc categories, which are created by the human
mind to express occasional and pragmatically conditioned relations be-
tween objects and their attributes; the latter serve the purpose of referring
to “objective” categories: such that correspond to well-established relations
perceived in the surrounding reality. In other words, while the latter corre-
spond to the Aristotelian differentiae specificae, the former reflect “free
choice” on the part of the speaker (for a discussion, see Tabakowska
2001a: 579). Hence the claim – made also by linguists describing lan-
guages other than Polish – that while the “specifying” adjectives have “lit-
eral, proper meaning, relevant for the contents of the noun”, the “character-
ising” ones “have meanings that are less precise, general, stylistically
marked, often figurative” (Magajewska 2006: 9).
Although the difference, traditionally described as “syntactico-
semantic”, is found to condition syntactic rules of pre- and postposition (in
such languages which allow both positions; but consider also the descrip-
tion of Old English NP in Fischer 2000), the same claim is made, mutatis
mutandis, in reference to the ordering of multiple elements within phrases
exemplifying any of the two positions: in either pre- or postposition, in
“natural” (unmarked) word order “characterising“ adjectives are found to
precede the “specifying” ones.
On the other hand, in nominal phrases containing multiple “specifying”
modifiers, the ordering reflects the rule (which is not language-specific)
pertaining to the scope of modification – the closer an attribute is to the
noun, the narrower becomes the scope of modification:
416 Elżbieta Tabakowska
As has been said, iconicity is, more or less explicitly, taken to be one of the
“most fundamental principles governing word order” (Zakrzewska 2001:
63). In her discussion on word order phenomena in modern Polish (which
includes a section on the ordering of adjectives within the NP),
Zakrzewska (2001: 664 – 665) enumerates three types of iconicity: sali-
ence, scope of modification and phrasal integrity. Other factors – logical
consequence, increasing complexity and rhematic-thematic structure – are
ascribed to mechanisms other than iconicity (Zakrzewska 2001: 668). The
remainder of the present paper aims at justifying the claim that linear ar-
rangement of adjectival modifiers within Polish NPs is basically pragmatic,
with iconicity principles underlying all word order varieties. It is those
principles (rather than rules) of iconicity (which will be classified as dia-
grammatic rather than imagic; for a discussion of this typological distinc-
tion, see Tabakowska 2001b) that govern “communicative intentions” of
speakers, and the same principles – petrified as “generalized intentions” in
the diachronic development of language – ultimately give rise to grammati-
cal conventions.
the most inherent property of the object (sofa ‘sofa’) is the material it is
made of, and thus the adjective takes the position closest to the noun. The
colour (brązowa ‘brown’), also inherently connected with the object, fills
the next closest slot, while the adjective referring to size – and thus less
objective and prone to reflect the speaker’s individual relative assessment –
is situated in the position most distant from the noun. None of the three
adjectives is a good candidate to grasp a pragmatically justified stable dif-
ferentia specifica: all are “characterising” rather than “specifying”. In con-
sequence, any change in the ordering of the three adjectives yields a con-
struction which would be judged unacceptable. Now consider
deed reveals its iconic motivation. Integrated phrases, however, may only
arise as the result of convention, or common consensus as to the way in
which a given fragment of the world should be perceived and conceptual-
ized.
It would thus not be unjustified to hypothesise that the change in status
of an adjective from characterising to specifying in languages like Polish,
where compounds and blends are not readily created, would result in shift-
ing the adjective from pre- to postposition, thus turning the structure into a
(potential) basic level term. In this way prepositioned slots would become
free for potential “characterising” modification. Vacillation between the
two possibilities would then mark either a transitory stage in diachronic
development or an incidental “one-shot” pragmatic motivation:
the makeup is strong (even) for the stage makeup as a specific category,
and iconicity imposes a default reading, with characterising modifiers pre-
ceding specifying ones.
Unlike in mental images, in linguistic renderings of images linear struc-
ture of language imposes ordering of individual elements of descriptions;
users of language try to override this limitation, an example being lexical
blends or portmanteau words. Whenever two or more modifiers are treated
“on a par”, that is, should indeed be expressed simultaneously as reflecting
simultaneity of perception, less innovative constructions allow for alternate
orderings:
But as shown by the English equivalents, even then the variants are not
“free”, as one or the other property of ‘the restaurant’ still “comes first”.
And sequentiality itself – apart from the resulting notion of distance – has
its own important role to play.
The principle of sequentiality refers to what has a long story as the concept
of ordo naturalis; primarily, it is manifested as a diagram that reflects tem-
poral or spatial succession, but – by metonymic and metaphoric extensions
– can also signify many abstract notions, such as continuity or change. It is
well known, for instance, that the default reading of a narrative imposes
upon the reader a chronology of events corresponding to the order in which
they had been described.
Adjectives express atemporal relations, therefore their linear ordering
within NPs reflects what cognitive grammar defines as a variety of abstract
Iconicity and linear ordering of constituents within Polish NPs 421
where the linear ordering of adjectives reflects the actual “zooming” order
in which the attributes were (or can be) perceived in reality, and the com-
mas correspond to pauses, marking consecutive stages of the conceptual-
izer's zooming abstract motion. The same effect can be observed in Szum-
ska's example
where the streets are perceived first (i.e. from a larger distance) as ‘nar-
row’ and only later (at a shorter distance) as ‘stone-paved’.
On a higher level of abstraction, linear sequentiality can be taken to re-
flect the cause-result relationship, with the spatiotemporal distance iconi-
cally (metaphorically) extended to stand for causal relationship, the natural
experiential sequence being “the cause before the result”. Such can indeed
be the explanation of Szumska’s further examples. In
“The last principle of iconicity, that of quantity, accounts for the tendency
to maintain the relation of direct proportion between the amount of form
and the amount of meaning: more form tends to carry more meaning while
less form is usually associated with less meaning” (Tabakowska 2001b: 8).
This type of iconicity is found on all levels of linguistic structure; as far as
word order phenomena are concerned, it finds a reflection in the well-
known phenomenon of the thematic-rhematic (or topic-comment) structur-
ing of expressions, whereby the new – or most relevant – information is
placed where it attracts most attention. In Polish syntax, there are two such
positions, marked by the increased amount of stress and phonological ma-
terial: at the end and at the beginning of expressions. Iconicity will natu-
rally favour the first of these possibilities, mainly due to the working of the
simple mechanism that makes human beings put “first things first”; what
speakers usually consider most important is what is closest to them, as
assessed from their particular point of view, or “what is at the moment
uppermost in the speaker’s mind tends to be the first expressed” (Hopper
and Traugott 2003: 61). A quarter of a century earlier Dwight Bolinger said
much the same thing (cf. the motto of this paper).
Like in other instances of verbal communication, speakers may simply
agree, by convention, as to what does, or should become uppermost in
their minds and therefore first in the relevant expression in a situation of a
given type, and gradually a linguistic rule emerges that reflects the conven-
Iconicity and linear ordering of constituents within Polish NPs 423
As is the case with flouting the maxims of conversation rather than just
following them, it is overriding the principles that carries extra informative
value. A possible exception is, probably, modification of iconicity-
governed linear ordering resulting from purely phonetic, or phonological,
considerations: because of their phonetic properties, some sequences just
“sound better” than others. It would be, however, fairly difficult to find
instances where the ordering would be absolutely free from other prag-
matic considerations, and, anyway, the result can never be identical as far
as the meaning is concerned.
In (12) the parallelism of structure suggest sameness of interpretation,
while in fact the last two adjectives are both characterising, allowing for
alternative ordering (biała aksamitna tunika vs. aksamitna biała tunika; cf.
also example (4) above). Yet it is only the “knowledge of the world” that
excludes the specifying interpretation.
where the adjective ‘cheap’ is shifted to the first position (against the prin-
ciple of perceptual linearity) as the most salient from the point of view of
the dealer, trying to attract the prospective buyer. Similarly, in
the reversal of the conventional order signifies that what the observer no-
ticed fist was the cheap material the lamp was made of, the epitome of the
cheap quality of the entire interior of which the lamp is an element (cf. also
(3) above). Such indeed is the function of (15) in its wider context (the
narrative in Anna Krzemieniecka’s novel Dziewczyna z Buenos). One
could say that in (15) the conventional iconicity principle is overridden by
“one-shot” discourse iconicity, which subdues conventional (intersubjec-
tive) perceptual sequentiality to the subjective sequentiality of an individ-
ual act of perception and conceptualization.
426 Elżbieta Tabakowska
(20), quoted after Pawica (2006: 135), comes from a literary text (A.
Szczypiorski’s Początek) and its professional, published English transla-
tion reveals the translator’s interpretation of the meaning: the possesive
adjective her preposes the noun (as it does in Polish) and reflects the lack
of (notional) distance between the possessor and the possessed. Large, in
closest postposition, but not specifying (with the dash marking the non-
specificity in the English version), comes as the most salient element of
(subjective) perception. Dark (which canonically should come closer to the
noun, i.e. precede large), reflects the iconic perceptual “zoom”. Finally,
washed clean... is the narrator’s interpretation rather than actual percep-
tion, and as such it follows the adjectives opening the “perceptual linear-
ity” chain. The ultimate effect is what is traditionally – and imprecisely –
called “literary style”.
3. Conclusion
References
Bolinger, Dwight
1952 Linear modification. Language 67: 1117–1144.
1977 Meaning and Form. London: Longman.
Fischer, Olga
2000 Position of the adjective from an iconic perspective. In The Moti-
vated Sign. Iconicity in Language and Literature 2, Olga Fischer and
Max Nänny (eds.), 249–276. Amsterdam/Philadelphia: John Benja-
mins.
Gębka-Wolak, Małgorzata
2000 Związki linearne między składnikami grupy nominalnej we
współczesnym języku polskim [Linear relations involving the compo-
nents of the nominal group in current Polish]. Toruń: Wydawnictwo
Uniwersytetu Mikołaja Kopernika.
Haiman, John
1985 Natural Syntax. Cambridge: Cambridge University Press.
Hopper, Paul J., and Elizabeth Closs Traugott
2003 Reprint. Grammaticalization. Cambridge: Cambridge University
Press, 1993.
Kubiński, Wojciech
1999 Word Order in English and Polish. On the Statement of Lineariza-
tion Patterns in Cognitive Grammar. Gdańsk: Wydawnictwo
Uniwersytetu Gdańskiego.
Łuczyński, Edward
1993 O interpunkcji w wyrażeniach typu samotny biały żagiel i
niespokojne, wzburzone morze [On punctuation in expressions
analogous to samotny biały żagiel and niespokojne, wzburzone
morze]. Język polski [The Polish language] LXXIII (4–5): 334–339.
Magajewska, Mirosława
2006 La posizione e la funzione dell'aggettivo qualificativo in Italiano.
Summary of an unpublished Ph.D. diss., Jagiellonian University,
Kraków, Poland.
Nänny, Max, and Olga Fischer
1997 Iconicity as a creative force in language and literature. Leaflet an-
nouncing the symposium “Iconicity in Language and Literature”,
University of Zűrich, 20–22 March 1997.
430 Elżbieta Tabakowska
Pawica, Michał
2006 The English equivalents of the Polish instrumental case. Ph.D. diss.,
Université Laval, Québec.
Radden, Guenter
1991 The Cognitive Approach to Natural Language. Series A. Paper no.
300. Duisburg: L.A.U.D.
Szumska, Dorota
2006 Przymiotnik jako przyłączone wyrażenie predykatywne [Adjective as
an attached predicative expression]. Kraków: Universitas.
Śliwiński, Władysław
1990 Łączliwość składniowo-semantyczna przymiotników z rzeczownikami
we współczesnym języku polskim [Syntactic and semantic adjective-
noun combinability in current Polish]. Kraków: Jagiellonian Univer-
sity Press.
Tabakowska, Elżbieta
2001a O szyku wyrazów w obrębie wyrażeń argumentowych przedmi-
otowych [On word order in object argument expressions]. Prace
filologiczne [Philological works] XLVI: 661–671. Warsaw: Institute
of the Polish Language, Warsaw University.
2001b Iconicity. In Handbook of Pragmatics, 1–17. Amster-
dam/Philadelphia, John Benjamins.
Topolińska, Zuzanna
1981 Remarks on the Slavic Noun Phrase. Wrocław: Ossolineum.
1984 Składnia grupy imiennej [The syntax of the nominal group]. In
Gramatyka współczesnego języka polskiego. Składnia [The grammar
of current Polish. Syntax], Zuzanna Topolińska, (ed.), 301–389.
Warszawa: Państwowe Wydawnictwo Naukowe.
Zakrzewska, Ewa D.
2001 Funkcjonalna eksplikacja szyku wyrazów w polszczyźnie [A
functional explanation of word order in Polish]. Prace filologiczne
[Philological works] XLVI: 661–671. Warsaw: Institute of the Polish
Language, Warsaw University.
Discourse-aspectual markers in Czech sound
symbolic expressions: Towards a systematic analysis
of sound symbolism
Masako U. Fidler
Abstract*
This article deals with variation among sound symbolic expressions (SSEs)
(uninflected, largely onomatopoeic expressions) in Czech: repetition,
lengthening, and suffix-like formations. By showing consistent correlation
between these three types of variation and functions related to aspectuality
and discourse, I demonstrate that SSEs do not merely imitate sounds and
motions, but accentuate specific aspects of events, set the locus of perspec-
tive, make implicit contrast between possible worlds or texts, foreground
events, or sum up a stretch of text. The major goal of this article is to ex-
plore the possible relationship between SSEs and grammar. Contrary to de
Saussure’s view that SSEs are “not an organic part of the language,” I argue
that there are properties that bridge SSEs and more conventionalized part of
the language. I show that the discourse-aspectual functions of variants
among SSEs can help explain some processes of word derivation and varia-
tion in language, e.g. cases of SSE-based aspectual derivation and nominali-
zation, as well as vowel length fluctuations in Colloquial Czech. Some of
the suffix-like formations among SSEs parallel quantification and diminu-
tion expressed by grammatical morphemes. By showing systematic func-
tions of components of SSEs, I argue that a SSE can be seen as a composite
structure that consists of the basic form and of a discourse-aspectual marker,
which parallels the model presented in Langacker (2003:67) for
[FLING+ER]; the basic SSE describes the properties of sound and/or mo-
tion, while the discourse-aspectual marker specifies functions related to as-
pectuality and discourse.
Keywords: Sound symbolism, onomatopoeia, Czech, aspect
432 Masako U. Fidler
1. Introduction
(a) sounds that reflect human psychological states and physical symptoms
(e.g. sounds indicating displeasure, exclamation of surprise, cough, hic-
cups);
(1) Kubát vzal cep a BÁC, BÁC! Zloděj se skácel. (Werich 1960: 61)
‘Kubát took the flail and [baɕɕts baɕɕts]!2 The thief was knocked
down.’
(2) Čert ... dělal hercum-percum a brlbrlbrl .... (Kubátová 1994: 62)
‘The devil ... did (made) [hertsum pertsum a brlbrlbrl] ...’
434 Masako U. Fidler
The indirect quotation in (4a) describes what she said: wanting to buy
an ant. The direct quotation in (4b) depicts how she said it. Its focus is
more on the acoustic aspects of how this message was delivered than on
the content of the message. SSEs are similar to direct quotation in this re-
spect, as shown below:
(5) Vážený pane, bla bla bla… videokač, bla bla bla… proto vás zveme…
(Disney 1991: 3/6)
‘Sir, [bla bla bla]… Videoduck, [bla bla bla]… therefore we invite
you...’
The SSEs in (5) above are used to illustrate the somewhat cumbersome
manner of speaking rather than to convey the entire utterance.
Czech sound symbolic expressions 435
In (7a) the simplex SSE is used instead of a verb. It illustrates the act of
hanging (by reporting the cracking of the neck), which constitutes one of
the sequentially ordered events. Example (7b) is different from (7a) in that
it represents more than one occurrence of the cracking sound. The differ-
ence, however, is not merely related to sound imitation. Unlike (7a), the
SSE in (7b) presents how the snow crunches as the character walks. The
point of perspective is internal to the event, giving the impression as
though the reader were seeing the snow and hearing the sounds through the
character’s eyes.
Events viewed from the internal vs. external perspective can render sen-
sations other than sound and motion as in (8a) and (8b):
(8a) presents one of the points in the story (packing up, playing dice,
having lunch where the speakers had a good mushroom gulash, and getting
on a train). The SSE notes the good quality of the meal and the speed with
which the food was eaten up (because it was good). In this regard, the sim-
plex variant can be viewed as part of the sequence of events. It is very
close to sentences such as: byl velmi dobrý and/or hned jsme ho snědli ‘it
[the goulash] was very good, we ate it up right away.’
The repeated mňam, mňam in (8b) is part of a text in which the speaker
recalls the pleasure of savoring the chocolates. Co-occurring with the im-
perfective verb chrupaly ‘they crunched,’ the multiplex variant stretches or
zooms in on the scene of savoring the chocolate, as if the speaker were
reliving the experience.
Functional approximations of (7a, 8a) and of (7b, 8b) are given in Fig-
ures 1 and 2 respectively. Here, t is the temporal axis, E0 refers to the event
associated with the SSE (E-1 and E+1 refer to events preceding and follow-
ing E0), and the thick lines represent the profiled parts of the event. The
line representing the event in Figure 2 is profiled in dots to show its indi-
vidual components.
E-1 E0 E+1
t
E0
t
Figure 2. Functional approximation of (7b) and (8b): Multiplex variants
Czech sound symbolic expressions 439
The simplex variant in (9a) represents the most dramatic event within
the stretch of text. The event of the fall is not only part of a sequence of
440 Masako U. Fidler
events, it is likewise the most foregrounded part of the storyline. The mul-
tiplex variant in (9b) can be considered a discourse correlate of (7b, 8b). It
represents a sound that resonates simultaneously as the character talks. The
point of perspective in this example is internal to the entire speech situa-
tion.
Figures 3 and 4 are functional approximations of (9a) and (9b). T0 is the
current segment of text. T-1 is the text segment that precedes T0 and T+1 the
text segment that follows it. The arrow in Figure 3 represents text progres-
sion. In Figure 3 the text segment T0 is viewed as more important and pro-
filed in relation to the other text segments that belong to the same episode,
T-1 and T+1. The perspective point is outside T0. In Figure 4, the perspective
point is internal to the entire stretch of text, creating the impression that the
reader is inside the speech event, hearing the utterance and the sound si-
multaneously. The entire T0 is profiled. The SSE is represented as a dotted
line as a part of this text; it is dotted to reflect focus on details (including
quantity) of the sound.
text progression
T-1 T0 T+1
Figure 3. Functional approximation of (9a): A simplex variant
T0
(10) “Ten muž musí být šílený.” “Ani ne šílený, jako spíš...”
Uch juch juch uch
(www.idnes.cz/IdsKosile/MIMORADNE/KOMIKS/komiks.asp?x
=include/komiks/_moravicus/2001/0519/0519_1.htm, accessed
5/1/07)
Czech sound symbolic expressions 441
‘[in the left dialog bubble] “This man must be crazy!” [in the right
dialog bubble] “Not really crazy, but rather…” [in the background,
hand-drawn] [ux jux jux ux]10’
The SSEs represent sounds that are continuously present throughout the
conversation between the two men.
In as much as simplex forms are associated with closed events, there is
a tendency for them to take the focus away from the internal properties of
the sound or motion (i.e. how something sounds or moves). Consequently,
they are more likely to develop non-sound symbolic meanings, in particu-
lar, to become verbs reporting what happened (rather than how something
happened), reporting cause-effect relationship between events. This is ex-
emplified by the SSE paf. Paf is a SSE that represents a gun shot when it
occurs in multiplex forms.
(12) … Taková akční bomba!!! Hrál jsem i tu hru a byl jsem z toho paf!!
(www.kfilmu.net/uzivatele.php?akce=vicenazoru&koho=3230,
accessed 9/7/06)‘
‘Such a cool action [movie] [lit. such an action bomb]!!! I played the
game, too, and was floored by it!!’
However, the difference between (13a) and (13b) is not limited to sound
iconicity, but also concerns aspectuality. The short variant is used to tell
the addressee to stop talking immediately; the long variant is used to urge
the addressee to calm down without a sense of urgency. In other words, the
former is associated with an immediate and abrupt change of state, the
latter with a process. The main goal of the speaker in (13b) is to achieve a
state where the addressee is calm, but this change is seen as gradual transi-
tion from an upset state to calmness.
Czech sound symbolic expressions 443
The contrast between long and short variants can be roughly repre-
sented by Figures 5 and 6.
S ~S
t
Figure 5. Functional approximation of (13a): A short variant
t
Figure 6. Functional approximation of (13b): A long variant
T ~T
text progression
Figure 7. Functional approximation of (14a): A short variant
SSE
(15) a. “Né, né, né” volá maminka, “mé, mé, mé” piští děti. (Knoblo
chová 2000: 4)
‘“No, no, no,” the mother calls, [meɕɕ meɕɕ meɕɕ],” squeak the
kids.’
In (15a) the long vowel [meɕ] represents the neutral bleating sound of
little goats. In comparison to this SSE, the SSE [meɕɕ] in (15b) is a form
with added length. Thus (15a) should be considered the short variant and
(15b) the long variant for this particular SSE. The discourse functions of
these examples are consistent with the short-long contrast discussed above.
(15a) marks a text boundary: the text of the mother goat and the text of the
kids who reject the mother’s text. In (15b) the SSE sums up the major emo-
tion (a threat), which is elaborated by the other parts of the utterance.
When discussing the contrast between long and short vowels, it is worth
considering the two different directions in which the semantics of SSEs
prask and prásk seem to develop. According to the 1989 dictionary both
prask and prásk render a cracking, bursting or tearing sound; the long
prásk represents a more resonating and louder sound than prask. The ex-
amples elsewhere are more or less consistent with the 1989 dictionary
definitions, as illustrated in (16a) and (16b) below.
The examples above all refer to a change of state on the lexical level. In
(17a) the exterior of the ulcer breaks open, allowing its contents to flow
out. In (17b) the money bursts out of the place where it was kept. In (17c)
the speaker receives a failing grade, thereby being demoted out of a group
of students where he had belonged. None of these examples represents an
audible sound.13
In contrast, prásk occurs more frequently than prask as an indeclinable
SSE (9 for prask as opposed to 114 in Syn2000).14 L-participles, which are
homonymous with prásk, nonetheless do exist, but they have different
properties from those homonymous with prask:
(19) Nedovedou si představit, jakou jsem měl radost! Sedmnáct let jsem ji
neviděl!
‘They can’t imagine how glad I was! I hadn’t seen her in 17 years!’
Ale pak zas: Prásk! Maďarsko! A mamince ten pas nedali.
(Syn2000)
‘But then again: a big surprise! Hungary! And they didn’t give mom
the passport.’
(20) Měla jsem za ten dlouhý život chvíle, kdy jsem si připadala
‘During this long life I had moments when I felt’
jako ta nejšťastnější a prásk kozu do vozu,
‘like the happiest [person], and a big surprise [lit. [praɕɕsk] the goat
into the cart],
za tři neděle bylo všechno jinak. (Syn2000)
‘after three weeks everything was different.’
In the examples above, prásk does not represent a loud physical sound,
but reports a state: ‘All of a sudden there was a big surprise like an explo-
sion.’ Likewise, the second example prásk, which is accompanied by a
complement-like component in the accusative, compares an unhappy pe-
riod to the sound of slamming a goat into a cart. In each sample prásk
represents the existence of a set of interrelated states and situations that
constitute a crisis or a misfortune.
Even less sound-symbolic meaning of prásk can be found in a syntactic
function that resembles a nominal:
448 Masako U. Fidler
The form can be considered a direct object of the verb mít ‘to have’.
Here, prásk is a group of properties that are more important than technique
for the athlete.15
Examples (17–21) show that the discourse-aspectual properties of prask
and prásk are at the base of how they develop less sound symbolic mean-
ings. The designation of transition is present both in the SSE prask and in
the verb in which prask is the root. The summary function is present in
both in the SSE prásk and in the SSEs that depart from sound representa-
tion. The discourse-aspectual functions of these SSEs parallel Langacker’s
two cognitive processes of summary and sequential scanning (1987: 247–
248).
Sequential scanning involves the processing of the component states in
a sequence. As this type of cognitive processing is said to be prototypical
of the meaning a verb, we can anticipate that prask is predominantly incor-
porated as part of a verb. On the other hand, summary scanning designates
all facets of the complex scene as simultaneously available and it is proto-
typical of the meaning of a noun (Langacker 1987: 199–200, 247). This
explains the noun-like behavior and the stative-predicate-like behavior in
(19–20) with prásk.
Sporadic occurrences of vowel length and shortening in the spoken Pra-
gue Czech is also consistent with the correlation between the two types of
cognitive processing and short-long vowel contrast. In spoken Prague
Czech, which is becoming the interdialect for Czech (Kučera 1961: 16–
20), variation in vowel length seems to be correlated with parts of speech.
According to the examples in Townsend (1990: 40–43), the most wide-
spread shortening is in verb forms e.g. in půjčit ‘lend’ > pučit and půjdu ‘I
will go’ > pudu. In contrast, vowel lengthening is much more frequent in
nominal stems and nominal-based adverbs than in verbs: dveře ‘door’ >
dvéře, nahoru/nahoře ‘up(stairs)’ (originally a preposition na + horu/hoře
> nahóru/nahóře.
Lengthening occurs also in familial words16 such as bratr ‘brother’ >
brácha, sestra ‘sister’ > ségra, tatínek ‘father’ > táta. The referential terms
Czech sound symbolic expressions 449
such as ségra, brácha and táta suggest the speaker’s personal relationship
to the referents. For example, ségra is not merely anyone’s sister, but the
speaker’s sister or the sister of someone whom the speaker knows well.
The term can therefore be viewed as emphasizing the large amount of the
speaker’s knowledge of the referent.17
The discourse-aspectual functions of long and short variants of SSEs
are correlated with two types of cognitive processing. This correlation pro-
vides a potential clue to motivating the lengthening and shortening in more
conventionalized parts of the language.
A simple repetition káp, káp would render rain drops falling one by one
with significant intervals; it would also help the reader imagine the number
of drops, as though s/he were witnessing them (Figure 2). The form with -y
represents indefinite multiple drops in a more compact form; -y thus essen-
tially marks plurality. Furthermore, the rain drops in this context constitute
the necessary condition for the merry-making to take place; the frogs
would not quack so loudly and evoke the image of a big party without the
rain.
The following example represents the sound and the motion of letter-
writing played out in a film:
450 Masako U. Fidler
w
x
~w
~x
w1 w2
x1 2 x2
T1 1 T
~w ~x1 ~w2 ~x2
Æ
5. Conclusions
Notes
* I want to express my gratitude to Laura Janda, Tore Nesset, the editors of this
volume as well as the two anonymous referees for their feedback. Obviously
all remaining errors and inconsistencies are entirely my own.
1. For examples of SSEs that represent motion, see e.g. Hamano 1998:2.
2. Sound symbolic expressions in Czech will not be translated, but will be repre-
sented in IPA. Since this study focuses not on grammatical categories, but on
discourse-aspectual functions, examples are not glossed, but are translated
side-by-side with the original.
3. In this article the term “aspectual” is an adjective connected with aspectuality.
Aspectuality is a semantic property, which overlaps but is not identical to ver-
bal aspect, which is a grammatical category.
4. Nuckolls (1996: 78) also notes similarities between aspectual functions of
SSEs on the one hand and cinematique techniques and gesture on the other.
This suggests that visual representation is highly relevant for the description of
SSEs, hence the use of diagrams in this study.
5. My approach belongs to a group of studies that use the general concept of
multiple domain structure in analyzing linguistic meaning (discussed in Sweet-
ser 1990: 147). The current study, however, differs from the existing literature
in dealing with those phenomena which have thus far been treated inconsis-
tently in lexicography and grammar and whose meanings are said to be situ-
ated in the phonological space rather than in the semantic space (Langacker
1987: 79–80). This article also finds that apparently random sound properties
in SSEs are connected to some of the very basic cognitive operations such as
placement of perspective and scanning processes.
6. The observations in this article are based on those onomatopoeic expressions
for which variation in length, repetition, and suffix-like formations are attested
in a database extracted from the 1989 8-volume dictionary. 406 indeclinable
expressions that are associated with sound and/or motion were identified. The
criteria used for selection were (1) indeclinability (often labelled as interjec-
tions) and (2) the definition of the lexicon that indicates connection to sound
and/or motion (in the presence of key words such as označuje/vyjadřuje zvuk,
pohyb ‘signifies/ expresses sound, motion’). More contexts for these variants
were sought in other sources (Syn2000, Syn2005, internet, literature, and com-
ics) and analyzed in detail in near-minimal pairs. In this article I use near-
minimal pairs rather than quantitative data for discussion for two reasons.
First, onomatopoeia is one area that is challenging to automatic tagging, which
is otherwise an indispensable tool for linguists. For example, bac in bac! bac!
napodobil ránu ... ‘[bac]! [bac]! he imitated a shooting sound ...’ is tagged as
a verb bacit in Syn2000, although it is more probably an indeclinable SSE. An
identical set of grammatical attributes is given not only to SSEs but to other
difficult-to-tag words such as abbreviations; e.g. the attributes for the ono-
Czech sound symbolic expressions 455
matopoeic br in Br… moje hlava!! ‘[br]… my head!’ and the chemical sym-
bol Br in Význam některých prvků La, Ce, As, Br, Pb, Sb pro organismy ‘The
significance of some elements La, Ce, As, Br, Pb, Sb for organisms’
(Syn2005). Both samples are labeled as nouns. Second, quantifying the occur-
rences of SSEs is problematic. The National Corpus automatically counts each
instance of an SSE as one token even when it is repeated in the same sentence.
Thus in a single context where the same onomatopoeic expression occurs three
times in a string (e.g. bum, bum, bum), the Corpus enters them as three tokens.
7. Although partial repetition is also possible (e.g. pif paf), it is not included in
the discussion here.
8. This statement is made in relation to the multiplex variant.
9. Examples in Chung and Timberlake (1985) concern Russian verbal aspect, but
this difference between the imperfective and perfective aspect is sufficiently
fundamental to be applied to Czech verbal aspect as well.
10. Juch juch reports the speaker’s state of euphoria.
11. This summarizing function is actually present in (7b) as well. The SSE is the
most important speech act and the other utterance of the speaker (everything
will be fine) corroborates this speech act.
12. I use only the masculine 3rd person singular forms as examples to show that
the form occurs only as a homonymous l-participle verb form rather than as an
indeclinable SSE. The presence of this form, of course, assumes that there are
other l-participle forms of the verb prasknout such as prasknul, praskla,
prasknula, which contain the root prask.
13. All the 9 occurrences of prask in Syn2000 are masculine singular l-participle
forms, but are automatically tagged as “X” (unknown).
14. The actual number of tokens for prásk is 99 after eliminating those that count
each occurrence of prásk in a single sentence. All the tokens are automatically
tagged as interjections, although some of them show different properties as
shown in the examples here.
15. Prásk (and not Prask) is also the name of a TV talk show where the host
interviews celebrities for recent gossip; this use can be considered as prásk
representing a set of properties and situations (i.e.. a collection of up-to-date
gossip).
16. Implied solidarity in dimunitives is compatible with the notion of decomposa-
bility in that it imposes a set of interrelated properties shared between the
speaker and the referent (and/or the addressee) Here, too, lengthening is pos-
sible (oheň ‘fire’ > dim. ohýnek, květ ‘flower’ > dim. kvítek, sud ‘barrel’ >
dim. soudek). Conditions for lengthening, however, may be influenced by
other factors, which are beyond the scope of this article. I thank an anonymous
referee for this comment.
17. Length in conventional words has similar functions, as in the following exam-
ple: (a) Vždy když jsem vylezl, křičel tatíí, tatíí. Pozná mě i po hlase.
(Syn2005) ‘Whenever I crawled out, I shouted daddyyy, daddyy. He would
456 Masako U. Fidler
recognive me even by the voice.’ (b) “... prosím, prosím, prosím, prosííím...”
To poslední prosím se vzneslo vysoko. Proniklo stropem a usadilo se na větvi
magnolie. (Syn2005) ‘“... please please please pleeease…” This last please
soared high. It penetrated the ceiling and settled on a branch of the magnolia
tree.’ In (a), not only the word tati ‘daddy’ but also the way it is delivered fa-
cilitates easy identification of the speaker. Similarly, (b) reports a distinct way
in which the word prosím resonates and travels. In each instance length im-
plicitly refers to a set of distinct properties of way in which the word was de-
livered.
18. This example was taken from a song. The influence of rhyming on the choice
of the suffix formation in this case, however, seems minimal, as káp, only the
second part of the SSE, rhymes with čáp: ...déšť do blatouchů si dělá “kápy-
káp”,… /že jim ulít' čáp.
19. In fact, the forms –y and –ity contain phonological elements that resemble with
the most prevalent plural suffix –y in nominal declension, as pointed out by an
anonymous referee.
References
Sweetser, Eve
1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects
of Semantic Structure. New York/Port Chester/Melbourne/Sydney:
Cambridge University Press.
Ústav Českého národního korpusu FF UK
2000, 2005
Český národní korpus [Czech National Corpus] – SYN2000,
SYN2005. Prague. Accessible at http://ucnk.ff.cuni.cz.
Townsend, Charles Edward
1990 A Description of Spoken Prague Czech. Columbus, OH: Slavica.
Werich, Jan
1960 Finfárum [Finfarum, fairy stories]. Prague: Československý spisova-
tel.
Subject index
absolute duration, 329, 339, 340, 156, 157, 158, 159, 160, 161,
344, 345, 347, 348, 353, 360, 162, 164, 165, 166, 167, 168,
361, 366 169, 171, 172, 173, 174, 175,
action 176, 177, 178, 184, 187, 188,
– completable action, 84, 105 189, 192, 194, 195, 196, 197,
– non-completable action, 84, 93, 201, 202, 211, 224, 236, 251,
105 252, 263, 281, 293, 330, 331,
– non-singularizable action, 93 332, 333, 349, 351, 352, 353,
– singularizable action, 93 355, 363, 367, 438, 439, 441,
active zone, 257, 269, 272, 273, 274, 453, 455
280 – canonical imperfective, 125,
actual/actuality, 120, 126, 130, 131, 140
132, 133, 134, 136, 137, 138, – non-canonical imperfective,
140, 142, 143, 146, 159, 161, 125, 127, 133, 134, 139,
163, 164, 175, 177, 178, 179 141
adjective/adjectival – perfective, 1, 6, 12, 13, 83, 84,
– adjectival predicate, 21, 23, 51 85, 86, 87, 88, 89, 90, 91, 92,
– attributive adjective, 311, 414 93, 94, 95, 97, 98, 102, 105,
aspect, 1, 5, 6, 10, 11, 13, 83, 84, 85, 106, 111, 112, 113, 114, 115,
86, 87, 88, 89, 90, 91, 92, 93, 116, 117, 118, 122, 125, 126,
105, 116, 117, 118, 119, 122, 127, 128, 130, 133, 134, 135,
123, 124, 143, 145, 150, 152, 136, 139, 141, 142, 145, 149,
184, 185, 187, 191, 192, 193, 150, 151, 152, 153, 156, 157,
197, 205, 208, 211, 234, 236, 158, 159, 160, 161, 162, 164,
244, 330, 331, 332, 351, 353, 165, 166, 167, 168, 169, 171,
357, 362, 363, 367, 435, 439, 172, 173, 174, 175, 176, 177,
441, 454, 455 178, 184, 185, 187, 188, 189,
– aspect markers, 12, 121 191, 194, 195, 196, 199, 201,
– aspect opposition, 351, 364 202, 203, 204, 205, 208, 209,
– aspect pair, 341, 349 211, 224, 227, 236, 293, 330,
– cluster model, 83, 92, 95, 96 331, 332, 348, 349, 439, 453,
– discourse-aspectual marker, 14, 455
431, 436, 453 – perfectivizing prefix, 14, 85,
– imperfective, 1, 6, 12, 13, 83, 106, 118, 329, 330, 350, 355,
84, 85, 86, 88, 89, 90, 91, 92, 361, 362, 363
93, 94, 95, 96, 97, 98, 99, 101, assertoric mitigation, 377
102, 104, 105, 106, 111, 112,
113, 114, 115, 116, 117, 118, binding, 233, 242
122, 123, 124, 125, 126, 127, – binding scale, 13, 221, 222
130, 131, 134, 136, 140, 145, – degree of binding, 244
146, 149, 150, 151, 152, 153,
460 Subject index
semantic change, 14, 129, 291, 294, usage-based, 10, 11, 291, 402
346, 347, 350, 375, 377, 378,
380, 381, 394, 395, 396, 398, verb
401, 402, 404 – bi-aspectual, 12, 13, 83, 84, 85,
semantic shift, 144, 228, 291, 299, 87, 88, 89, 90, 91, 92, 94, 95,
302, 305, 316, 317, 319, 323, 96, 97, 98, 99, 100, 101, 102,
335, 376, 381, 383, 398 103, 104, 105, 106
sound symbolism, 11, 431, 432, 433 – delimitative verb, 329, 331,
– sound symbolic expression, 5, 332, 333, 337, 345, 348, 352,
14, 431, 432, 436, 450, 454 353, 356, 357, 363
Subject index 463