. 1
( 17)


This page intentionally left blank
A Reference Grammar of Russian
A Reference Grammar of Russian describes and systematizes all aspects of the
grammar of Russian: the patterns of orthography, sounds, in¬‚ection, syntax,
tense-aspect-mood, word order, and intonation. It is especially concerned with the
meaning of combinations of words (constructions). The core concept is that of the
predicate history: a record of the states of entities through time and across
possibilities. Using predicate histories, the book presents an integrated account of
the semantics of verbs, nouns, case, and aspect. More attention is paid to syntax
than in any other grammars of Russian written in English or in other languages
of Western Europe. Alan Timberlake refers to the literature on variation and
trends in development, and makes use of contemporary data from the internet.
This book will appeal to students, scholars, and language professionals interested
in Russian.

a l a n t i m b e r l a k e is Professor of Slavic Linguistics at the Department of Slavic
Languages and Literatures, University of California at Berkeley. He is the author of
The Nominative Object in Slavic, Baltic, and West Finnic (1974) and editor of The Scope of
Slavic Aspect (with M. S. Flier, 1985), American Contributions to the Eleventh International
Congress of Slavists (with Robert A. Maguire, 1993), and American Contributions to the
Twelfth International Congress of Slavists (with Robert Maguire, 1998).
A Reference Grammar of Russian
University of California at Berkeley
cambridge university press
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press
The Edinburgh Building, Cambridge cb2 2ru, UK
Published in the United States of America by Cambridge University Press, New York
Information on this title: www.cambridge.org/9780521772921

© Alan Timberlake 2004

This publication is in copyright. Subject to statutory exception and to the provision of
relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.

First published in print format 2004

isbn-13 978-0-511-16446-0 eBook (EBL)
isbn-10 0-511-16446-7 eBook (EBL)

isbn-13 978-0-521-77292-1 hardback
isbn-10 0-521-77292-3 hardback

Cambridge University Press has no responsibility for the persistence or accuracy of urls
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.

1 Russian 1

2 Sounds 28

3 In¬‚ectional morphology 92

4 Arguments 159

5 Predicates and arguments 270

6 Mood, tense, and aspect 371

7 The presentation of information 444

Bibliography 473
Index 493


1.1 The Russian language

1.1.1 Russian then and now
The present study is a comprehensive description of all aspects (except word
derivation) of modern standard Russian: its sounds, spelling, grammar, and
Russian has resulted from a long evolution that can be traced back to the ¬rst
millennium of our era. From the ¬fth century on, speakers of Slavic established
settlements over a vast area of Central and Eastern Europe, from the Danube in
the south to the Elbe in the northwest. In the east, they moved north from the
Dnepr valley to the Gulf of Finland and the Upper Volga, gradually displacing
or assimilating the previous Baltic and Finnic inhabitants.1 Russian developed
from the dialects of Slavic spoken in the north of this East Slavic territory. In
the ninth century, the East Slavic area came under control of Scandinavian
merchant-warriors. The Christianization of this land in 988 was followed by
subjugation to “the Mongol yoke” from the thirteenth century into the ¬fteenth
century. As the favored agent of the Golden Horde, the once small principality
of Moscow brought ever more land under its control. By the end of the ¬fteenth
century, when the Mongol yoke was de¬nitively removed, Moscow had become
the political and ecclesiastical center of the East Slavic lands, and the center of
the Russian language area.
Russian is not only a spoken language, but a written language used for all
cultural purposes. The modern form of Russian took shape over the course of
the eighteenth century. The morphology and phonology is based on the dialect
of Moscow. In its vocabulary, syntax, and rhetoric, Russian, while relying on
native Slavic elements, has a long history of adapting and internalizing foreign --
Byzantine, French, and most recently English -- models.
Parenthetically, it could be noted that the modern word h©ccrbq ˜Russian™ is an
adjective deriving from the noun H©cm ˜Rus™. According to a venerable etymology,

1 See Sedov 1982 on the complex archeological record of the East Slavic area.

2 A Reference Grammar of Russian

H©cm was a descriptive name for Scandinavians that is based on the Germanic et-
ymon ˜to row™, the Scandinavians being above all oarsmen.2 In East Slavic lands,
H©cm was used initially for the Scandinavian overlords and their principality
of Kiev. Over time it was extended to all East Slavic lands. Muscovy appropri-
ated the name for its political identity, culture, and language as it consolidated
Russian is the ¬rst language of approximately 150 million people. According
to an estimate for 2002 the Russian Federation had a total population of 145 mil-
lion people, among whom 81.5 percent, or 118 million, were ethnic Russians.3 In
the mid-nineties, there were an additional 25 million Russians in the newly in-
dependent countries that emerged from the breakup of the Soviet Union (Novaia
Rossiia 1994). Together that would make 143 million ethnic Russians. To that
¬gure could be added a substantial though indeterminate percentage of the
remaining 27 million members of other nationalities residing in the Russian
Federation. According to recent statistics, the rate of population growth in the
Russian Federation is negative (’0.33%), from which it would follow that the
number of speakers of Russian will not increase in the foreseeable future.

1.1.2 Levels of language
Russian is a spoken language and a written language. In its written form Rus-
sian has long been highly codi¬ed: grammars, dictionaries, and manuals de¬ne
standards for usage that are enforced in the educational system and through
editorial practices in publication. Although the Russian tradition is quite clear
about what usage counts as standard, it does acknowledge the existence of a
range of varieties, or registers, from archaic to bookish to standard (normative)
to colloquial (hfpujdjhyfz htxm) to substandard and uncultured (ghjcnjhtxbt).
The grammar recorded here is the normative grammar of standard, written
Russian, which is the culturally privileged, and also the most accessible, form
of Russian. Occasionally, there are asides on usage in less-than-standard or oral
language, but this study cannot treat colloquial Russian with the same attention
as the works of E. A. Zemskaia and colleagues,4 which have documented the sig-
ni¬cant differences between spontaneous spoken Russian and formal, written

2 Possible candidates are Roþer, Roþin, former names for Sweden™s Uppland region, and roþs- ˜oar™,
the genitive form used in compounding (Thomsen 1879:99--104, also Vasmer 1986--87:s.v. Hecm, de
Vries 1962: s.v. rj°r, Schenker 1995:57--60). A form of this etymon was adopted into West Finnic
languages (Finnish ruotsi ˜Sweden™) and into Slavic, and then found its way into Greek ( < ‚ ) and
Arabic (r¯s) sources from the ninth and tenth centuries.
3 At: http://www.odci.gov/cia/publications/factbook/geos/rs.html#People.
4 Zemskaia 1973, 1978, 1983; Zemskaia and Shmelev 1984; see also Timroth 1986.
Russian 3

Russian has undergone some change since the political and economic tur-
moil of the late eighties and early nineties, but it is dif¬cult to assess how
much. Most tangibly, there have been changes in vocabulary.5 Borrowing and
native derivational processes have produced many new words and word combi-
nations, leading to macaronic texts: ytqk-fhn ˜nail-art™, WEB-lbpfqy ˜WEB-design™,
Htrbq ,bhvbyutvcrbq lb-l;tq gj bvtyb Graham Mack lb-l;tbk ct,t, lb-l;tbk,
lf nfr b ljlbl;tbkcz, xnj c hflbj eitk ˜A certain Birmingham DJ, named
Graham Mack, DJ-ed, DJ-ed, and so DJ-ed out, that he had to leave the radio
station™. This internationalized vocabulary now dominates the linguistic land-
scape, just as Soviet-speak used to dominate language a half century ago. Along
with these changes in vocabulary has come a less quanti¬able but still palpable
change in the mores of language. Unedited, informal texts of written Russian
of a type that would never have become public during the days of active So-
viet censorship are now available in print and especially electronic form. And
yet, despite political changes and a loosening of speech manners, contemporary
Russian in its grammatical structure remains Russian.

1.2 Describing Russian grammar

1.2.1 Conventions of notation
The notational conventions employed here are those of Table 1.1.
In the body of the text, Cyrillic words and phrases will be given in italics,
and English translations in single quotation marks. Stress is marked in citation
forms of words or short phrases; stress is not marked on vowels in fragments of
text cited in the text or in set-off numbered examples. In numbered examples,
italics and quotations are not used.

1.2.2 Abbreviations
The abbreviations used in this study are listed in Table 1.2.

1.2.3 Dictionaries and grammars
The de¬nitive dictionary of Russian in Russian is the Slovar sovremennogo russkogo
literaturnogo iazyka, a seventeen-volume dictionary published over 1950--65. Self-
evidently it does not include the numerous new words from the last sev-
eral decades. Shorter Russian-language dictionaries are fully useful, notably
Ozhegov™s one-volume classic, which conveniently lists grammatical forms with
stress. More than adequate bilingual dictionaries are the Oxford dictionary (both
directions) and now the Novyi Slovar (Russian to English), the most up-to-date

5 Zemskaia 2000.
4 A Reference Grammar of Russian

Table 1.1 Conventions used

notation interpretation

grammatical gloss and Russian word
nom sg ntnh’lm
alternative grammatical gloss of Russian word
ntnh’lm<nom sg>
grammatical form conditioned by another word (preposition or
spelling of letter (or word) in Cyrillic, when spelling is at issue
[’] sound (from narrow phonetic through broad phonetic to
{’} or {’ : ø : ™} vowel series, or set of stressed and unstressed vowels related
by etymology and/or synchronic alternation
{-ej} or -tq morphological unit
{X : Y} any relation of elements, notably two stems of verbs,
{CVC-a-<pst/inf> : CVC-aj-|e|-}<prs> }
X∼Y two forms potentially available in the same context
aspect pair: perfective and secondary (derived) imperfective
aspect pair: simplex imperfective and semelfactive perfective
(gj)ghjc«nm or aspect pair: simplex imperfective and pre¬xed perfective

/±/?/— hierarchy of acceptability judgments: neutral, acceptable,
frequent / less preferred option / restricted, marginal /
dubious, ungrammatical

dictionary available. A selection of dictionaries -- Russian only and bilingual -- is
available on the web.
Russian dictionaries, unlike many dictionaries of English, do not give infor-
mation about etymology, for which one should consult the dictionary of Max
Vasmer (in its original German edition of 1953 or the Russian edition of 1986--87
revised by O. N. Trubachev), nor about earlier usage, for which one should use
Srevnevskii™s “materials” for a dictionary of Old Russian from 1893--1912 (and
later reprints), Slovar russkogo iazyka XI--XVII vv., or Slovar russkogo iazyka XVIII
veka. Lubensky (1995) should be consulted for Russian idioms.
For grammatical information, the “grammatical dictionary” of A. A. Zalizniak
(1977[a]), with 100,000 entries arranged in reverse alphabetical order, is de¬ni-
tive. Entries of the dictionary are indexed with paradigm numbers; excep-
tions are marked. The 142 introductory pages list paradigms with accentual
A variety of grammars is available, including two compact grammars in En-
glish (Unbegaun 1957, Wade 1992), which, however, do not treat syntax exten-
sively, as well as the multiple generations of “academy grammars” (for example,
Russian 5

Table 1.2 Abbreviations used

abbreviation interpretation
set of consonants / obstruents / sonorants / {[v v˛ ]}
C/C /R/W
C / C0
¸ set of palatalized consonants / set of non-palatalized consonants
V / V! / V * set of vowels / stressed vowels / unstressed vowels
P/T/K/ˇ S consonant articulations: labial / dental / velar / alveo-palatal
C / C / Ci/ Ci / Ci / Ci consonant grades (§2.5.2)

[z˛ ] / [r] / [r]
ü3 palatalized [z] / voiceless [r] / voiced [r]
[´] / [´] / [´]
5 5 55 [a] fronted in initial transition / ¬nal transition / both transitions
rr ü articulation in which one feature changes over duration of
nominative / accusative / genitive / dative / locative / instrumental
nom / acc / gen / dat /
loc / ins
primary / secondary genitive // primary / secondary locative
gen1 / gen2 // loc1 / loc2
syncretism of nominative and accusative (“inanimate accusative”) /
nom=acc / acc=gen
syncretism of accusative and genitive (“animate accusative”)
singular / plural / dual
sg / pl / du
masculine / feminine / neuter
msc / fem / nt
animate / inanimate
an / in
predicative (= “short”) adjective
noun / quanti¬er / adjective / possessive
nn / qu / adj / pss
Declension<I> ¬rst declension: Declension<Ia> and Declension<Ib>
¬rst declension (masculine type with nom sg {-… }: ,j,)
Declension<Ia> ´
¬rst declension (neuter type with nom sg {-o -e}: cn’lj)
Declension<II> second declension
Declension<III> third declension
third declension (feminine with nom sg {-…}: gkjoflm)
Declension<IIIa> ´
Declension<IIIb> third declension (neuter with nom sg -z: dh†vz)
third declension (masculine with nom sg {-…}: g©nm )
R / E /A /F /T /M stress paradigms -- stress on: root / ending / classi¬catory suf¬x
(verbs) / antethematic syllable / thematic syllable / mobile stress
present / past / future / in¬nitive / imperative / irrealis / realis /
prs / pst / fut / inf /
participle / adverbial participle (lttghbxfcnbt) / passive participle
imv / irr / rls / pcl /
dee / psv
imperfective / perfective // determinate (imperfective) /
if / pf // dt / id
1sg / 1pl / 2sg / 2pl / 3sg / ¬rst-person singular / ¬rst-person plural / second-person singular /
3pl second-person plural / third-person singular / third-person plural
/B/ ” / address by ns / address by ds / mutual address by ns / mutual
B ”B / ”B address by ds / asymmetric address, one speaker using ns, the
other ds /
У /˜ /О /¤ diminutive name / ¬rst name / patronymic / surname
word order: subject verb object domain manner
jyf v dpzkf o vtyz d yf
,fpfh m ,kfujgjkexyj
6 A Reference Grammar of Russian

RG 1980). The four-volume “functional grammar” is superb (Bondarko 1991--96).
Good grammars exist in other European languages (for example, Garde 1980 in
French, Isaˇenko 1975 in German). The discussion below, though it is informed
by this tradition of grammatical analysis, does not cite them in the interests of
avoiding a clutter of references.

1.2.4 Statistics and corpora
To characterize how likely some construction is, it is often useful to cite statistics
of usage. At the same time, it is important to acknowledge the limitations on
statistical statements. The likelihood of using some or another morphological
form or syntactic construction is really the likelihood of using the context in
which the form or construction is appropriate; statistics ultimately measure how
likely people are to say a whole context. For example, if we ¬nd that the com-
bination e ytq is less frequent than e ytt, what we have really found is that the
contexts in which e ytq is appropriate occur less frequently than those in which
e ytt is appropriate. Any statistical statement, even one that appears to deal with
morphological variants, is a measure of the frequency of the contexts in which
these variants are appropriate. When the discussion below cites statistical obser-
vations, it is usually to say, informally and without pretense of scienti¬c rigor,
that a certain construction occurs surprisingly often or not particularly often,
relative to what one might expect. The limitations on what statistical statements
mean should always be kept in mind.
As a corpus for making statistical observations, I initially used the “Uppsala
Corpus.” The corpus, assembled by the Slavic Institute of Uppsala University and
mounted on the web by the University of T¨ bingen,6 offers a balanced selection
of styles of texts through the 1980s; it has its own search. As time went on, I
made use of the broader resources of the web. The address “http://www.lib.ru/”
has a vastly larger number of (belletristic) texts. By using a powerful search en-
gine (such as Google, Zndex, or Rambler), it is possible to search this site or
the whole web for words or phrases, and produce quantities of Russian larger
by orders of magnitude than the Uppsala Corpus. For example, in the Uppsala
Corpus, the target ins sg nsczxtq produced no tokens, the target e ytq ¬ve
tokens. In contrast, a search of http://www.lib.ru/ (with Google, <20.X.02>) pro-
duced 233 hits for nsczxtq and 796 for e ytq; and on the whole web (with Google,
<20.X.02>), there were 8,790 hits for nsczxtq and 25,900 for e ytq. The new elec-
tronic resources, then, offer the possibility of vast quantities of Russian, most
of it very contemporary.

6 At: http://www.sfb441.uni-tuebingen.de/b1/korpora.html. The description (<http://www.slaviska.uu.
se/korpdesc.htm>) states that the corpus is based on 600 Russian texts, one million running words,
of informative (late 1980s) and literary texts (1960--88).
Russian 7

There are, however, some negatives, which grow in proportion to the size of
the corpus and the frequency of the target word or phrase. Unlike the Uppsala
Corpus, which was designed to serve as a corpus and has a balanced selection
of genres of texts, the web was not designed to serve as a corpus for linguistic
investigation. The web has properties that make it less than ideal as a corpus:
(a) the relative weight of genres -- www.libr.ru is heavy on literary texts and trans-
lations (if one has hesitations about translations), while the web as a whole has
a random mix of commercial writing, personal travelogues, detailed histories
of the repair records of automobiles, journalism, and religious texts; (b) the
quality of Russian, which includes translations, sites from outside Russia, and
informal personal writing and commercial writing that is no longer subjected
to the same editing as was Russian printed in the Soviet era; (c) the fact that
many of the texts show up on more than one site, undercutting the value of
statistical observations; (d) instability -- the sites are not stable over time, im-
peding replication and veri¬ability; (e) the number of positive hits, which can
be so large that the ¬nite amount of time it takes to evaluate any token makes
it dif¬cult to examine all the data. The enormous volume of Russian available
now is a mixed blessing.7
Allow me to cite cautionary tales. With respect to repetition: the phrase e;t
jnrhsdfk<pst.if> jryj ˜[he] already opened the window™ -- a familiar phrase in
aspectology -- gave a modest forty hits on the whole web (<20.XII.01>). But every
one of them was the same sentence from a text by A. Tolstoy. With respect to
stability, I searched the web for the expressions hfymit ytuj ˜earlier than him™
and d jnyjitybb ytuj ˜in relation to it™, and came up with 1,590 and 5,490 to-
kens, respectively (<20.XII.01>). The same search nine months later (<15.IX.02>)
yielded 2,080 and 7,190 tokens -- an increase of 17 percent. With respect to quan-
tity: I searched the web (<20.X.02>) for tokens of nsczxtq -- 8,790 hits -- and
nsczxm/ -- 10,800 hits -- with the goal of ¬nding out in crude terms the relative
frequency of these two forms of the instrumental case of nsczxf. It would take
perhaps eighty hours to evaluate all that data, if a modest ¬fteen seconds were
devoted to each token. In short, the investigator has no control over the web and
no way of determining what its properties as a corpus really are. The Uppsala
Corpus, though smaller, offers a more balanced corpus.
In light of such dif¬culties, it is important to emphasize the limitations on
citations from the web. All statistical statements made on the basis of the web
should be taken for what they are: informal characterizations of frequency over
unstable, often repetitive, collections of Russian assembled for other (commer-
cial, etc.) purposes than to serve as a corpus for linguistic investigation. The
corpus is not stable and one cannot control for repetition.
7 Browne 2001 explores the problems of using the web as a corpus.
8 A Reference Grammar of Russian

In the same vein, it is also important to register the disclaimer that there is
no guarantee that speci¬c websites, referred to occasionally below, will remain

1.2.5 Strategies of describing Russian grammar
The discussion of Russian below follows an unsurprising sequence: after these
preliminaries, ending with the writing of Russian, the discussion goes from
sound to morphology (grammar in the traditional sense) to syntax -- ¬rst argu-
ments, then predicates, then predicates in context (tense, aspect, modality) --
and ¬nally, selected discourse operations that apply to the presentation of in-
formation. Obviously there are many topics that belong in two places -- tense in
participles is a question of morphology and of predicate semantics in context;
the second genitive is a question of morphology, of arguments, and of predicates
(since the use of the second genitive depends on the syntactic context) -- and it
was necessary to make decisions about where to put discussion. Cross-references
are provided.
A word about the philosophy of grammar invoked here. Modern linguistics has
prided itself on identifying basic, primitive elements (phonemes, morphemes,
constituents of sentences) and their rules of combination. For some researchers,
the ultimate goal is to characterize which sentences are possible, which impos-
sible, and to state the rules of combination. My experience in assembling this
grammar has led in a different direction. Repeatedly I found that what was
signi¬cant was the construction -- the pattern, the con¬guration, the template
(nhfafhtn8 ). Patterns include all manner of linguistic knowledge: constituent
elements; typical lexical items that participate; strategies of interpreting the
meaning, or value, of the pattern in discourse; stylistic value -- in short, pat-
terns include all kinds of linguistic knowledge. The semantic, pragmatic, and
stylistic values of a construction are not entirely predictable from its primitive
elements and rules of combination, and though any construction certainly con-
tains smaller entities, it is not always possible (or important) to identify the
primitive elements. It becomes more important to say in what contexts, and
with what meaning, a construction can be used. The whole is often greater
than its parts. For example, the free (dative) in¬nitive construction (yfv<dat> yt
vbyjdfnm<inf> ub,tkb ˜it is not for us to avoid disaster™, ,tp htdjk/wbb yfv<dat>
yt lj,bnmcz<inf> cjdthitycndf ˜without a revolution it is not for us to achieve
perfection™) has recognizable parts: an in¬nitive, a dative that would be the sub-
ject if the in¬nitive were a ¬nite verb, and the other argument phrases governed
by the verb. There is no overt ¬nite verb; no form of ,ßnm ˜be™ is used in the

8 Zhivov and Timberlake 1997.
Russian 9

present tense. The meaning of this construction -- it makes a prediction about the
possibility of an imagined event -- cannot be computed just from its constituent
parts, the dative and the in¬nitive. Moreover, the construction has different
variants, each of which has a speci¬c stylistic value. The variant just illustrated
is folksy, apodictic. Another variant of the construction used in content ques-
tions is neutral and productive, as in, Rfr gjgfcnm d yfxfkj cgbcrf yfqltyys[
cfqnjd yf gjbcrjds[ vfibyf[? ˜How [is it possible] to get to the beginning of
the list of sites in search engines?™ Indeed, the initial portion of this question,
Rfr gjgfcnm . . . ˜how [is it possible] to reach . . .™, produced 18,900 hits on
the whole web (<20.X.02>). In general, then, the presentation of Russian gram-
mar below emphasizes whole combinations and their value (including stylistic),
downplaying the task of identifying primitive elements or articulating notations
for encoding rules of combination.
When there are two closely related constructions that differ by one linguistic
form -- for example, relatives made with rnj vs. rjnj ´hsq, genitive vs. accusative
with negated verbs, etc. -- it is an interesting question how speakers choose be-
tween the variants. In a notational approach to grammar, one can always create
different structures that will produce different cases (for example). But because
the structures will be distinct, there is no way of comparing the properties that
distinguish them -- the properties of the noun phrases, the discourse import --
and such an approach says nothing about how speakers make choices. As an
alternative, one can look for as many tangible variables as possible -- variables
such as the number of a noun, its position relative to the verb, the aspect of
the verb -- and measure their statistical contribution. But the result of a variable
rule is only a probability, which does not explain how a speaker works with a
half dozen to a dozen factors and makes a choice that is binary -- to use one
construction or another. In the following, I assume that speakers operate with
templates (constructions) that have multiple properties -- lexical to syntactic to
discourse. In any instance, speakers ask which template a given utterance better
matches. This is a holistic decision: in the genitive of negation, perhaps, speak-
ers evaluate a context as being concerned with absence of a situation (genitive)
as opposed to reporting an entity™s properties (accusative). To get to this holistic
judgment, speakers ask which template better ¬ts the context. And to answer
that question, speakers probably have to select one feature to pay attention to,
while others are ignored. In practical terms, this means it is dif¬cult, for many
constructions, to give watertight rules about usage (there are too many variables;
speakers have some freedom in how they rank and evaluate variables). What can
be done is to point out the general, holistic value of a construction, and, often,
some tangible linguistic features that are consistent with that holistic value that
will in¬‚uence choices.
10 A Reference Grammar of Russian

1.2.6 Two fundamental concepts of (Russian) grammar
While each construction, each problem of grammar, requires its own descrip-
tion, some general, recurrent ideas emerged. Two can be mentioned.
One is modality and the related concept of quanti¬cation. Every statement is
understood against alternatives. Sometimes there is just a contrast of the mere
fact that some x having one salient property exists at all, in contrast to the
possibility that x might not hold, or that a certain situation holds in contrast
to the possibility that might not exist (existential or essential quanti¬cation).
Sometimes a speci¬c individual x or property is contrasted with other possible
x™s or ™s (individuated quanti¬cation). Modality -- consideration of alternatives
by an authority -- pervades grammar.
The other is directionality, dialogicity. An utterance does not exist or have
meaning in isolation, but is manipulated by speakers and addressees in a three-
step process. The speaker invites the addressee to construct a background of
information, taken as given and known (¬rst step). Against this background
the speaker formulates, and the addressee evaluates, the current assertion (sec-
ond step). On the basis of that comparison, the speaker and addressee then
project further conclusions or anticipate further events (third step). Thus the
speaker invites the addressee to engage in a directional process of manipulating
These concepts -- modality (and quanti¬cation) and directionality -- pervade
the grammar of Russian and, no doubt, other languages.

1.3 Writing Russian

1.3.1 The Russian Cyrillic alphabet
Russian is written not in the Latin letters used for English and Western Euro-
pean languages but in an alphabet called Cyrillic (Russian rbhbkkbwf). Cyrillic,
with small differences, is also used for other languages -- Ukrainian, Serbian,
Bulgarian. Cyrillic will be used to write Russian throughout the discussion be-
low, with certain obvious exceptions: in the discussion of sounds and the inter-
nal structure of words, in glosses of Russian words or phrases, and in citations
of scholarly literature. For reference, the version of the Cyrillic alphabet used for
modern Russian is given in Table 1.3. In Column 1 the alphabet is presented in
the lower- and uppercase forms used in printing. Column 2 gives the italic vari-
ants. Column 3 gives longhand forms of lowercase and then uppercase letters as
used in connected, cursive writing (unusual uppercase letters are omitted); the
subsequent discussion, however, will not treat handwriting.9 The contemporary

9 With thanks to Victoria Somoff for the handwriting sample.
Russian 11

name of the letter is given in Column 4. These names are mostly transparent.
The names of consonant letters have a vowel added to the sound of the conso-
nant. Four unusual letters are referred to by descriptive phrases. For reference,
Column 5 gives the older names of the letters. Column 6 states approximate
sound values of individual Cyrillic letters in English, although there are obvi-
ous dif¬culties in attempting to state the sound of Cyrillic letters in terms of
English sounds: the closest English sound is not always particularly close; in-
dividual Cyrillic letters do not represent just a single sound (consonants can
be palatalized or not; vowel letters have different value depending on whether
or not they follow consonant letters). The statements of sound value are quite
Because Cyrillic is an alphabet, by establishing correspondences between each
individual Cyrillic letter and one or more Latin letters, it is possible to rewrite, or
transliterate, Cyrillic into Latin letters. Column 7 is the table of equivalences
established by the Library of Congress as used in slightly simpli¬ed form in this
study. (Other systems are discussed later: §1.3.7.) The ¬nal column gives sources of
the Cyrillic letters. The alphabet given in Table 1.3 is the contemporary alphabet.
The civil alphabet used until the reform of the October Revolution included two
additional letters: ¤®≥ “b ltcznthbxyjt” (alphabetized between ¤b≥ and ¤r≥) and
¤˜≥ “znm” (between ¤m≥ and ¤э≥). Additional letters are found in Russian Church
From various people, one often hears that Russian must be a dif¬cult lan-
guage because its alphabet is so dif¬cult. Nothing could be further from the
truth. Whatever the dif¬culties of Russian, they cannot be blamed on the al-
phabet, which anyone with a modicum of ability in language systems and a
vague acquaintance with the Greek alphabet can learn in half an hour, as will
be demonstrated after a brief introduction to the history of the alphabet.

1.3.2 A brief history of the Cyrillic alphabet
The beginning of writing in Slavic is a fascinating tale that deserves to be told
in brief.11 The story can be picked up at the end of the eighth century, around
796, when tribes of Slavs from the region of Moravia (in the south of the con-
temporary Czech Republic, along the Morava River) helped Charlemagne rid
Central Europe of the last remnants of the Avars, a confederation of Eastern ma-
rauders. This venture marked the beginning of more active relations between
Moravian Slavs and the West, both with secular political authorities (Charle-
magne until his death in 917, his descendants thereafter) and with ecclesiastical

Library of Congress Romanization: ¤®≥ > ¤¯≥, ¤˜≥ > ¤ˇ Russian Church Slavic used also ¤º≥ >
10 ± ±e≥.
¤™ ¤v≥ > ¤™≥
f≥, y
11 Dvornik 1970, Vlasto 1970.
Table 1.3 The Russian Cyrillic alphabet

Cyrillic Cyrillic Cyrillic contemporary archaic letter English equivalent Library of Congress etymology
(print) (italic) (longhand) letter name name (very approximate) Romanization of letter

f/F f/F f fp a Gk A
Masha | all
,/< ,/< ,э ,erb bother b Gk B
d/D d/D dэ dtlb Volga v Gk B
u/U u/U uэ ukfujk guard g Gk
l/L l/L lэ lj,hj do d Gk
t/T t/T tcnm
t [åt] e Gk E
Pierre | yell
=/+ =/+ = [åj] -- e Cyr t/T
Fyodor | yoyo
;/: ;/: ;э ;bdbnt azure, French je zh Gl ¦
p/P p/P pэ ptvkz zoo z Gk Z
b/B b/B b b i Gk H
beat | eat
q/¦ q/Q b rhfnrjt -- boy i Cyr b/B
r/R r/R rf rfrj car k Gk K
k/K k/K k/lb
эkm Leeds l Gk
v/V v/V vsckbnt
эv Masha m Gk M
y/Y y/Y yfi
эy no n Gk N
j/J j/J j jy o Gk O
go | only
g/G g/G gэ gjrjq pot p Gk
h/H h/H hws
эh rot r Gk P

c/C c/C ckjdj
эc sew s Gk
Table 1.3 (cont.)

Cyrillic Cyrillic Cyrillic contemporary archaic letter English equivalent Library of Congress etymology
(print) (italic) (longhand) letter name name (very approximate) Romanization of letter

n/N n/N nэ ndthlj toe t Gk T

e/E e/E e er u Gk Y
do | oops
a/A a/A athn
эa far f Gk
[/{ [/{ [f [th German ach kh Gk {

w/W w/W wэ ws tsetse, prints ts Gl –
x/X x/X xt xthdm church ch Gl ·
i/I i/I if if shallow, fish sh Gl ˜
o/O o/O of inf Josh should, fish shop shch Gl ù
( ( th
ndthlsq pyfr (th∞ ) [boundary marker] Gl ú
s/S s/S s, ths∞ ths pituitary y Gl »
m m thm
vzurbq pyfr (thm∞ ) [consonant palatalized] Gl ü
э/э э/э э j,jhjnyjt -- e Cyr t/T
best | Evan
//? //? / / iu Gl
cute | yule
z/Z z/Z z z b
ia Cyr
Diaghilev | Yalta

x | y = pronunciation after consonant letter | pronunciation not after consonant letter
Gk = Greek
Gl = Glagolitic
Cyr = (earlier) Cyrillic
x∞ = older name still used (SRIa 1.152)
14 A Reference Grammar of Russian

authorities. As part of this interaction, missionaries were sent to the Mora-
vians from the Franks (from the relatively new bishoprics of Regensburg, Passau,
Salzburg) and from the Italians (from the bishopric of Aquileia). The conversion
of Prince Mojmír of Moravia (r. 818--46) in 822 was followed by a general baptism
in 831. In this period of missionary activity, churches -- some in stone -- were
constructed at sites in Moravia such as Mikulˇice.
In 846, Mojmír™s nephew Rostislav took control and began to act with greater
autonomy. After the bishopric of Salzburg had its charter renewed in 860, Ros-
tislav took steps to avoid further ecclesiastical interference from the Franks. In
862, after having been put off by the Pope, he approached the Byzantine Emperor
Michael III with a famous request:

Though our people have rejected paganism and observe Christian law, we have not
a teacher who would explain to us in our language the true Christian faith, so that
other countries which look to us might emulate us. Therefore, O lord, send us such
a bishop and teacher. (Kantor and White 1976:45)

Emperor Michael and Patriarch Photius responded by sending Constantine
(canonized as St. Cyril) and Methodius, two brothers educated in Greek who
spoke a Slavic language, to Moravia to train disciples and translate the liturgy
and the Bible into Slavic. In order to write in Slavic, they devised an alphabet
which is now called Glagolitic. The letters of Glagolitic are stylized combinations
of strokes and loops; for example, the chapter title for Luke 11 (Marianus) reads
in Glagolitic, ˜on the catching
of ¬sh™).12 It is still an open question what sources Constantine and Methodius
used for this new alphabet. It has long been assumed that the model was Greek
minuscule,13 but it may have been cursive of a Latin (speci¬cally Carolingian)
type.14 Whatever the source of the alphabet, writing in Slavic has its origins in
the “Moravian mission” of Constantine (St. Cyril) and Methodius.
The Moravian mission began auspiciously. It was given papal approval when
the brothers traveled with their disciples to Rome (867). After Constantine died
in Rome (869), Methodius was appointed bishop of a large missionary area in-
cluding Moravia and Pannonia. In the long run, however, the mission proved vul-
nerable. It was resented by the Frankish bishops, who went so far as to imprison
12 Jagi´ 1883 (interleaf 110--11, 186).
13 Beginning with Taylor (1880, 1883), who exhibited apparent similarities between individual
Glagolitic letters and Greek minuscule letters.
14 Lettenbauer 1953 (summarizing an inaccessible study, Hocij 1940) cites intriguing pairs of
Glagolitic and Carolingian letters. For example, the Carolingian ¤j≥ is a vertical arc open on
the left, with loops both on the top and at the bottom, hence very similar to the double loop
of Glagolitic ¤®≥; Taylor™s Greek cursive omicron has no loops. Taylor™s Greek cursive ¤l≥ looks
like a modern English cursive ¤l≥, with an internal loop (that is, ¤ ≥), very unlike the Glagolitic
double-looped ¤¤≥, which looks like the Carolingian.
Russian 15

Methodius until the Pope secured his release. Rostislav, the Moravian prince who
originally sponsored the mission, was blinded and exiled. When Methodius died
in 885, a hostile bishop (Wiching of Nitra) chased out the troublemakers and
reinstalled the Latin rite. Disciples of Constantine and Methodius were fortunate
to make it to Ohrid and Bulgaria.
In Bulgaria, Tsar Boris, who had initially converted to Christianity in 863,
held a council in Preslav in 893, at which he abdicated, turned over power
to his pro-Christian son Symeon, and appointed Clement, one of the original
Moravian disciples, as bishop. Around this time, conceivably at this council,15
the practice was established of writing religious texts in Slavic in letters that
were modeled to the extent possible on Greek majuscule letters.16 (For Slavic
sounds that had no equivalents in Greek, letters were adapted from Glagolitic.)
This neophyte Christian culture, with sacred texts written in Slavic in this Greek-
like alphabet, ¬‚ourished in Bulgaria in the ¬rst half of the tenth century, until
the time (in 971) when Byzantium defeated Boris II and absorbed the Bulgarian
patriarchate. This tradition of writing was brought to Rus as a consequence of
the conversion to Christianity in 988. The alphabet that was imported was the
direct ancestor of the alphabet in which modern Russian is written, the alphabet
we call “Cyrillic.” As this brief sketch shows, Cyril himself did not invent the
Cyrillic alphabet. But he and his brother did invent the alphabet in which Slavic
was written systematically for the ¬rst time, and the alphabet they constructed
did provide the model for Cyrillic.
After having been brought into East Slavic territory, this alphabet was used
in the oldest principalities of Kiev, Novgorod, and Vladimir-Rostov-Suzdal from
the eleventh century on, and then subsequently in Moscow, the principality
that emerged as dominant as the “Mongol yoke” was loosened. This alphabet
has continued to be used with only modest changes until the present day. Peter
the Great attempted to reform the orthography in 1708--10. His new civil al-
phabet (uhf;lfyrf) had letters of a cleaner, less ornate (more Western) shape.
Peter also proposed that, in instances where more than one letter had the same
sound value, only one letter be preserved, the ¬rst of the sets for the
sound [i], for [z], for [o], ¤y/U≥ for [u], ¤a/f≥ for [f]; some other
were also to be eliminated.17 Al-
letters with quite speci¬c functions
though all of Peter™s proposals did not catch on, his initiatives led to modernizing
the graphic shape of the alphabet and set in motion the process of rationaliz-
ing the inventory of letters. While the general trend has been to simplify the

15 Dvornik 1970:250--52; Vlasto 1970:168--76.
16 The similarity is quite striking between early Cyrillic writing and contemporary Greek Gospels,
for example Lord Zouche™s gospel text from 980 (Plate IV, Gardtgauzen 1911).
17 Zhivov 1996:73--77.
16 A Reference Grammar of Russian

inventory of letters, ¤q э =≥ were introduced in the course of the eighteenth
Russian Cyrillic took its contemporary form in a reform of October 1918, which
built on the results of earlier commissions (most immediately, the commission
of 1917). The notable changes were that remaining duplicate letters were elim-
inated (¤b≥ in place of ¤®≥, ¤t≥ for ¤˜≥, ¤a≥ for ¤f≥) and the “hard sign” ¤(≥
was eliminated from the ends of words after consonant letters, where it had
previously been required. For example, nineteenth-century ¤,˜c(≥ ˜demon™ be-
came ¤,tc≥. Other changes concerned the spelling of speci¬c morphemes (for
example, adjectival msc sg ¤juj≥ in place of ¤fuj≥).
The principles established in 1918 were canonized by the publication of Rules
of Russian Orthography (= Pravila) in 1956. The principles and detailed rules have
largely been stable, despite occasional discussions of possible further reforms
of some annoying -- but in the larger scheme of things, insigni¬cant -- incon-
sistencies (for example, in 1964).18 There was uncertainty, and continues to be
uncertainty, with respect to the vexed question of how much to use ¤=≥. Other
unresolved questions include: use of the hard sign ¤(≥ as mark of separation;
spelling of ¤b≥ or ¤s≥ after ¤w≥; spelling of ¤t(=)≥ or ¤j≥ after ¤; i x o≥;
spelling of ¤mj≥ and ¤qj≥ in borrowings; use of ¤э≥ after consonants; use of
double letters in borrowings. At this moment, there is a renewed impetus to
address certain details of writing, notably those involving compounds.19

1.3.3 Etymology of letters
As noted, most Cyrillic letters were based on Greek upper case (majuscule) let-
ters. Many of the contemporary Cyrillic letters look like Greek letters, and as a
¬rst approximation they can be read as one might expect on that basis. Among
Cyrillic letters for consonants, we observe the following similarities (Greek ma-
juscule prototypes are written in parentheses; the approximate sound value is
recorded in Table 1.3): ¤u/U≥ (Greek ); ¤l/L≥ ( ); ¤p/P≥ (Z); ¤r/R≥ (K); ¤k/K≥ ( );
¤v/V≥ (M); ¤y/H≥ (N); ¤g/G≥ ( ); ¤h/H≥ (P); ¤c/C≥ ( / ); ¤n/T≥ (T); ¤a/A≥ ( );
¤[/X≥ (X). From the single Greek /B, Cyrillic has ¤,/<≥ (a bilabial stop [b]) and
¤d/D≥ (a labio-dental fricative [v]).
The consonant sounds of Slavic that did not have obvious correspondences
have unique symbols without any obvious source in the Greek or Latin alpha-
bets; they apparently derive from Glagolitic, which did have distinct letters for
these sounds: ¤;/:≥, ¤w/W≥, ¤x/X≥, ¤i/I≥, ¤o/O≥. Though these letters are
unfamiliar, sounds somewhat similar to those represented by these letters occur
18 Comrie, Stone, and Polinsky 1996 (ch. 8) gives a comprehensive survey from 1917 forward (see also
Chernyshev 1947). For the eighteenth and nineteenth centuries, see Grot 1873.
Proposals and rejected changes were accessible on www.gramota.ru/ <01.XII.01>.
Russian 17

in European languages. The most exotic is the sound spelled as ¤o/O≥, a con-
sonant of double length; it can be approximated by combining two tokens of
the sound written in English as ¤sh≥ in two words: Josh should, fish shop.
Vowel letters are largely based on Greek prototypes. As discussed below, there
are two parallel sets of vowel letters. In the ¬rst set (hard-vowel letters) we
¬nd: ¤f/F≥ (Greek /A), ¤э/Э≥ (an innovation, based on older Cyrillic ¤¦≥), ¤j/J≥
(Greek o/O), ¤e/E≥ (Greek /Y), ¤s≥ (derived from a combination of two letters,
the uniquely Slavic letter ¤(≥ and the Slavic adaptation of Greek ¬/I). The sound
corresponding to ¤s≥ is perhaps the single most dif¬cult for non-natives to
pronounce. Some Russians use this sound as a substitute for the vowel of pit
or hip in speaking English. A closer approximation would be a vowel that
changes from an [u]-like vowel to an [i]-like vowel, something like pituitary or
phooey, but pronounced as one syllable, not two. In the other set of vowel
letters (soft-vowel letters), two derive from Greek: ¤t/T≥ (from Greek µ), pro-
nounced as [e], and ¤b/B≥ (Greek /H), pronounced as [i]. One has a source in
Glagolitic (¤//?≥ = the sound [u]) and two others arose in the history of Russian
Cyrillic writing (¤z/Z≥ = the sound [a]; ¤=/+≥, derived from Cyrillic ¤t/T≥ = the
sound [o]).
Identifying the etymology of letters does not, of course, explain how the Cyril-
lic alphabet works. But it should make it clear that the majority of the letters, in
their graphic shape and (approximate) sound value, are familiar from a cursory
acquaintance with the Greek alphabet.

1.3.4 How the Cyrillic alphabet works (basics)
The Cyrillic alphabet is a good guide to pronunciation. It is generally clear how a
sequence of letters should be pronounced. One complication is that in every word
in Russian one vowel is strongly stressed, and the remaining unstressed vowels
are pronounced less clearly than the one stressed vowel (unstressed vowels are
“reduced”). Once one knows which syllable is stressed, phonetic reduction is not
dif¬cult for speakers of English. Unstressed vowels are commonly the indistinct
“schwa” vowel; Russian V’if is pronounced with [™] in the second syllable, thus
[m’ˇ™], much as the ¬nal vowel of the English version of this name, Masha, is
spoken. However, most writing does not indicate which vowel is stressed. In this
respect, spelling does not give complete information about pronunciation.
To understand how the Russian Cyrillic alphabet works, it is necessary to
mention one fact about consonant sounds. Most consonants can be pronounced
in two signi¬cantly different ways: not palatalized, when they are somewhat
similar to consonants in English, or palatalized, when the tongue is raised to-
wards the front and top of the mouth, towards the area behind the teeth. The
effect of palatalization is similar to the beginning of English few, pew, or, in one
18 A Reference Grammar of Russian

pronunciation, tutor, duke, with the difference that in English, there is a distinct
segment between the consonant and the vowel, while in Russian, this raising
of the tongue extends over the duration of the consonant. In Western sources,
there are many ways of representing palatalization in consonants. It is common
to write a superscripted letter (¤i y j≥) after the consonant to indicate that there
is a brief transition to the following vowel similar to a vowel [i]; thus the famil-
iar word ytn ˜no™, in which the “n” sound is palatalized, might be written as
[ni et] or [ny et] or [nj et]. An alternative is to write an apostrophe or acute accent
above or after the consonant letter, [n™et] or [n„et] or [n et] In this study, palatal-
ization will be written as a cedilla, [n˛ et], for the reason that palatalization is
generally pronounced throughout the duration of the consonant; it is not just
a transition to the following vowel. (When it is important to emphasize that a
consonant or group of consonants is not palatalized, the degree sign is placed
after the consonant letter: “Co ”.) The m u t a b l e consonants -- those that can be
either palatalized or not -- are the consonants spelled by the letters ¤g , d a v
n l c p y r u [ h k≥. The remaining consonants, those spelled by the letters
¤x o i ; w≥, are immutable: they are either intrinsically palatalized (the
sounds [c˛ ‹ s¦] spelled by ¤x o≥) or intrinsically not palatalized (the sounds [s ‹ z ‹ c]
spelled by ¤i ; w≥, respectively). Informally in the Russian tradition, conso-
nants that are not palatalized are called “hard,” palatalized consonants “soft.”
This convenient informal characterization is often used in the following.
The most important fact about Russian orthography is that it is organized
around the question of how to spell palatalization in consonants. As noted above,
there are two sets of vowel letters. Vowel letters indicate not only what vowel
is to be pronounced (as might be expected), but they also indicate what sounds
come before the vowel. In particular, letters of the soft set ¤b t z = /≥ indicate
that the preceding consonant is palatalized when they follow a consonant letter
from the set of mutable consonants ¤g , d a v n l c p y r u [ h k≥. Thus:
¤z≥ = the sound [a] plus palatalization of the consonant, as in ¤Lzubktd≥
˜Diaghilev™, pronounced [d˛ ’]; ¤/≥ = the vowel [u] plus palatalization in the pre-
ceding consonant, as in ¤h/vrf≥ ˜wineglass™ pronounced [rúmk™]; ¤t≥ = the
sound [e] plus palatalization, as in ¤ytn≥, pronounced [n˛ ©t]; ¤=≥ = the sound
[o] plus palatalization, as in the name ¤A=ljh≥, pronounced [f˛j ´d™r]; and ¤b≥,
as in the name ¤Lbvf≥, pronounced [d˛´m™]. If no consonant letter precedes --
at the beginning of a word, after another vowel, or after the boundary signs
¤m (≥ (discussed separately below) -- a soft-vowel letter as a rule indicates that
the glide sound [j] precedes the vowel. Thus, at the beginning of the word, the
soft-vowel letter ¤z≥ is pronounced with [j] before the [a] sound, as in ¤Zknf≥ --
that is, [jalt™], whence the common English form Yalta (in Library of Congress
transliteration, Ialta); the soft-vowel letter ¤/≥ begins with [ju], as in ¤?hbq≥,
Russian 19

whence English Yuri (Library of Congress Iurii); after a vowel, the soft-vowel letter
¤t≥ is automatically pronounced with [je], as in ¤Ljcnjtdcrbq≥, as is indicated
by one of the possible English spellings, Dostoyevsky.
Letters from the set of “hard-” vowel letters ¤s э f j e≥ indicate which vowel is
pronounced and, when they follow a consonant letter from the set of mutable
consonants ¤g , d a v n l c p y r u [ h k≥, they indicate that the preceding
consonant is not palatalized: ¤Vfif≥ ˜Masha™ indicates that [m] is followed by
[a], and the [m] is not palatalized; ¤Genby≥ ˜Putin™ indicates that unpalatalized
[p] is followed by [u]. When no consonant letter precedes -- at the beginning of a
word or after another vowel letter -- a vowel from this set indicates that there is
no [j] before the vowel: ¤fkmn≥ ˜viola™ [al˛t] begins with [a], not [ja]; ¤enrf≥ ˜duck™
[útk™] begins with [u], not [ju].
After the consonant sounds spelled by the letters ¤x o i ; w≥, which are
pronounced the same regardless of the following vowel, a mixed set of vowels
is used (§1.3.5).
When no vowel letter follows directly after the consonant letter, palataliza-
tion is marked by a special symbol ¤m≥, called the “soft sign” (vzurbq pyfr). For
example, the ¤m≥ at the end of ¤vfnm≥ ˜mother™ tells us that the sound of ¤n≥
is palatalized [ t˛], and ¤m≥ tells us that the initial consonant sound of ¤nmvf≥
˜darkness™ is palatalized [t˛].
The principles of Russian orthography can be presented as a set of branching
decisions involving combinations of vowel letters and contexts, as in [1].

[1] Algorithms of Russian spelling
if a consonant is spelled by ¤x o i ; w≥, it is pronounced the same in all
it can be spelled at the end of words or before another consonant letter; a following
vowel letter is one of the set ¤b t f j e≥
if a consonant is spelled by ¤g , d a v n l c p y r u [ h k≥, it is pronounced as
palatalized (soft) if
it is followed by ¤m≥ at the end of a word or before another consonant letter; or, a
following vowel letter is one of the set ¤b t z = /≥;
if a consonant is spelled by ¤g , d a v n l c p y r u [ h k≥, it is pronounced as
non-palatalized (hard) if
it occurs at the end of a word or before another consonant letter; or, a following
vowel letter is one of the set ¤s э f j e≥.

1.3.5 How the Cyrillic alphabet works (refinements)
In each of the two sets -- hard-vowel letters ¤s э f j e≥ and soft-vowel letters ¤b t
z = /≥ -- the letters behave in a similar fashion up to a point, but there are some
idiosyncrasies. The basic properties of vowel graphemes and the operational
20 A Reference Grammar of Russian

Table 1.4 Distribution and values of vowel letters

context ¤f≥ ¤e≥ ¤s≥ ¤j≥ ¤э≥ ¤z≥ ¤/≥ ¤b≥ ¤=≥ ¤t≥

Co Co Co Co Co C (Co )
/¤C≥ C¸ C¸ C¸ C¸ ¸

/# V V V V jV jV V jV jV

/¤V≥ V V V V jV jV V jV jV
— — — — —
/¤m (≥ jV jV jV jV jV
√ √ √ √ √
— — — — —
/¤x i ; o≥ (lexical) √ √ √ √ √
— — — — —
/¤x i ; o≥ (grammatical) √ √ √ √ √
— — — — —
/¤w≥ (lexical) √ √ √ √ — — — — — —
/¤w≥ (grammatical)

√ = beginning of word
= combination occurs
— = combination never (rarely) occurs
C = palatalized consonant
Co = consonant not palatalized

graphemes ¤m (≥ are given in Table 1.4. Shading indicates a cell that differs
from nearby cells.
Asymmetries and irregularities include the following. The pair ¤s≥ vs. ¤b≥
is similar to other pairs at least to the extent that ¤b≥, as a soft-vowel letter,
marks a mutable consonant as palatalized (thus ¤nb≥ implies [t˛i]), while ¤s≥
marks a consonant as not palatalized (thus ¤ns≥ implies [tÈ]). In this respect the
pair ¤s/b≥ is parallel to the pairs ¤f/z≥, ¤e//≥. However, there is one important
respect in which ¤b≥ does not behave the same as other soft-vowel letters. When
¤b≥ is used in initial position or after a vowel, it does not imply a preceding
[ j]. Thus no [j] occurs initially in ¤Bujhm≥, English Igor, which is pronounced
[íg™r˛], not — [jíg™r˛]; and no [j] occurs between the vowels of ¤Hfbcf≥, English Raisa
(pronounced [r ísa], not — [r jísa]) or in ¤vj«≥ nom pl ˜my™ (pronounced [m í],
not — [m jí]).
Cyrillic ¤э≥, until recently, was used sparingly, for historical reasons. Any orig-
inal — e in initial position or after a vowel acquired a prothetic [j], the only ex-
ceptions being native demonstrative stems (¤эnjn≥ ˜this™, ¤эnfrbq≥ ˜such a™) and
borrowings (¤эnf;≥ ˜¬‚oor™, ¤э[j≥ ˜echo™, ¤gjэn≥ ˜poet™). Further, consonants were
palatalized before original — e. Thus ¤э≥ is spelled only in acronymic formations
like ¤YЭG≥ (from yjdfz эrjyjvbxtcrfz gjkbnbrf ˜New Economic Policy, NEP™).
It used to be standard practice to spell any foreign “e” vowel with Cyrillic ¤t≥,
even when the preceding consonant was pronounced as hard; in a borrowing,
a spelling of ¤lt≥, in certain words, might be pronounced not as soft ([d˛ ]) but
as hard [do ]: ¤ltnfkm≥ [do ] ˜detail, part™, ¤,typby≥ [bo ] ˜fuel™. Recently, however,
¤э≥ is being used more often, after hard consonants (¤cэqk≥ ˜sale™, ¤Lэyyb lt
Dbnj≥ but ¤,tcn-ctkkth≥) and even after vowels (¤rhbэnjh≥ ˜creator™).
Russian 21

Cyrillic ¤=≥ is more of a diacritic modi¬cation of ¤t≥ than a separate let-
ter. It is not given a distinct position in alphabetical ordering in dictionaries;
thus, ¤t;tkb≥ ˜if™ is alphabetized between ¤=;≥ ˜hedgehog™ and its diminutive
¤=;br≥. ¤+≥ indicates that the vowel is stressed [ó]. In addition, after a con-
sonant letter, it indicates that a preceding mutable consonant is palatalized:
¤A=ljh≥ [f˛ód™r]. When there is no preceding consonant letter, the vowel is
preceded by [j]: ¤=;br≥ [jóˇïk]. Thus when it is used, then, ¤=≥ has a function
parallel to that of ¤z≥ or ¤/≥. But in fact ¤=≥ is not used in all texts or styles of
writing. If stress is marked generally -- it usually is not, but it can be, for example,
in dictionaries or pedagogical texts for foreigners -- then ¤=≥ is certainly used.
Apart from such aids, the more explicit ¤=≥ may be used in certain genres of
texts intended for mass audiences: encyclopedias, schoolbooks, publicistic texts.
In many other genres of text -- ¬ction, journalism -- ¤=≥ is generally not used, and
ordinary ¤t≥ is used instead. This letter is used in some of the recent postings
on the web (for example, in the catalogue of the Russian State Library20 ), but
not in the majority; no pattern is yet clear.
Individual borrowings that might be expected to have ¤=≥ do not necessarily
use that letter. Neither ¤=≥ nor ¤t≥ is used to indicate the sequence of palatalized
consonant followed by [o] in such borrowings as ¤cbymjh≥ ˜se˜ or™ or ¤,ekmjy≥
˜bouillon™. The sequence ¤qj≥ is used internally after vowels (¤hfqjy≥ ˜region™)
and is generally used in borrowings to represent [jo] initially: ¤qjl≥ ˜iodine™,
¤qjuf≥ ˜yoga™, ¤Qjhr≥ ˜York™ (though Japanese names do use ¤=≥: ¤+cfyj≥
˜Yosano™). The grapheme ¤=≥ is also used, lexically and locally, as an aid to the
pronunciation or identi¬cation of individual words, notably to distinguish the
neuter singular pronoun ¤dc=≥ from the plural ¤dct≥ ˜everyone, all things™: -- F
ns dc= эnj jgbib ˜you just describe all that™; Dfkthbq b z dc= ikb ˜Valery and I
kept on walking™. In discussions of spelling below, ¤=≥ is characterized as explicit
writing style, ¤t≥ as neutral style.
In compounds, soft-vowel letters indicate that [j] precedes the vowel, even
after a previous consonant letter: ¤djty/hbcn≥ ˜military lawyer™ [no j], ¤ltnzckb≥
˜children™s daycare™ [to j]. Remarkably, in borrowings ¤q≥ can be followed by soft-
vowel letters: ¤ajqt≥ ˜foyer™, ¤gfgfqz≥ ˜papaya™, ¤gfhfyjqz≥ ˜paranoia™, ¤Fqz-
Cjabz≥˜Hagia Sophia™, ¤(hfcnen) ctrdjqb≥ ˜sequoias (grow)™.
Consonant letters designating immutable sounds (¤x o i ; w≥) have unusual
properties, and are followed by a mixed set of vowel letters, normally ¤f≥, ¤e≥
(very exceptionally ¤/≥ in borrowings: ¤,hji/hf≥ ˜brochure™, ¤;/hb≥ ˜jury™),
¤t≥, and ¤b≥. Spelling of stressed [ó] after these letters is complicated. Native
roots use ¤=≥ in explicit style, or, in neutral orthographic style, ¤t≥: explicit
¤o=rb≥ ˜cheeks™, ¤;=knsq≥ ˜yellow™, nom pl ¤;=ys≥ ˜wives™, ¤vjkjlj;=ys≥
Hjccbqcrfz Ujcelfhdcndtyyfz <b,kbjntrf http://www.rsl.ru/ <10.X.02>. The site does not use

¤=≥ on its home page.
22 A Reference Grammar of Russian

˜newlyweds™, neutral ¤otrb≥, ¤;tknsq≥, ¤;tys≥, ¤vjkjlj;tys≥. ¤J≥ is used
in derivation when the vowel is stressed, as in diminutives: ¤yj;jr≥ ˜knife™,
¤,jhojr≥ ˜soup™, ¤nf,fxjr≥ ˜tobacco™, ¤vtijr≥ ˜bag™. In grammatical endings
¤j≥ is used when the vowel is stressed, as in: ins sg<Ia> ¤yj;jv≥ ˜knife™,
¤regwjv≥ ˜merchant™, ¤,jufxjv≥ ˜rich man™, ins sg<II> ¤leijq≥ ˜soul™, ¤jdwjq≥
˜sheep™, ¤cdtxjq≥ ˜candle™. Not under stress, derivatives and grammatical end-
ings are spelled with ¤t≥: gen pl dim ¤ryb;tr≥ ˜books™, ¤hextr≥ ˜handles™,
ins sg<Ia> ¤gkfxtv≥ ˜cry™, ¤ytvwtv≥ ˜German™, ¤gkz;tv≥ ˜beach™, ins sg<II>
¤elfxtq≥ ˜good fortune™, ¤uheitq≥ ˜pear™. While ¤t≥ (explicit ¤=≥) is usual in
roots, ¤j≥ is used under stress in certain lexical items: ¤ijhj[≥ ˜rustling™,
¤ijd≥ ˜seam™, ¤ghj;jh≥ ˜glutton™, ¤j;ju≥ ˜burn™, ¤;jkj,≥ ˜chute™, now usu-
ally ¤;=kj,≥ (¤;tkj,≥). Until the orthographic reform in 1918, ¤j≥ was used
in other native roots (¤;jknsq≥, pl ¤ojrb≥). In borrowings ¤j≥ is normal:
¤Ijgty≥ ˜Chopin™, ¤ijr≥ ˜shock™, ¤;jrtq≥ ˜jockey™. The principle, simpli¬ed
somewhat, is that after ¤x o i ; w≥, ¤j≥ is used for a stressed vowel in mor-
phological environments and internally in borrowings, ¤t≥ is used elsewhere
(lexical environments, unstressed vowel).
Another complication is that both ¤wb≥ and ¤ws≥ are used; ¤ws≥ occurs in old
lexemes (¤wsufyt≥ ˜Gypsies™, ¤wsgk=yjr≥ ˜chick™), ¤wb≥ in modern borrowings
(¤wbrk≥ ˜cycle™, ¤wbdbkbpfwbz≥ ˜civilization™). In grammatical endings ¤s≥ is
used (nom pl ¤jnws≥ ˜fathers™).
The “hard sign” ¤(≥ and the “soft sign” ¤m≥ do not represent any sound directly.
Rather, they are operational graphemes that indicate how adjacent graphemes
are to be understood. The “hard sign” ¤(≥, after being eliminated from the end
of words in the orthographic reform of 1918, has limited functions. It is used
after pre¬xes before a soft-vowel letter (¤j,(zcyzk≥ ˜explained™, ¤c(tcnm≥ ˜eat up™)
and in some borrowings (¤j,(trn≥ ˜object™, ¤rjy(/yrnehf≥ ˜con¬guration™). It
is a boundary grapheme, indicating that the following soft-vowel letter is to be
read as if it began a word -- that is, ¬rst comes the consonant (which may or
may not be pronounced as palatalized), then [j], then the vowel: ¤jn(tpl≥ [ t˛j†st]
∼ [ to j†st].
The “soft sign” ¤m≥ has greater utility. When no vowel letter follows, ¤m≥ in-
dicates that a preceding mutable consonant is palatalized. When a vowel letter
follows, ¤m≥ (like ¤(≥) indicates that the vowel letter is to be interpreted as if it
were in initial position, hence preceded by [j]; the preceding consonant is palatal-
ized if it is mutable: compare palatalized ¤,m/≥ ˜I beat™ [b˛ ju], but unpalatalized
¤im/≥ ˜I sew™ [ˇju]. When the symbol ¤m≥ is not followed by a vowel letter, it
indicates that the preceding consonant is palatalized. Thus the ¤m≥ indicates
that the lateral consonant is palatalized in gen sg ¤kmlf≥ ˜ice™, ¤njkmrj≥ ˜just™,
¤cnjkm≥ ˜so much™. After ¤x i ; o≥, which designate immutable consonants,
Russian 23

¤m≥ cannot mark palatalization, yet it still occurs in speci¬c morphological en-
vironments: in nouns of Declension<III> (¤yjxm≥ ˜night™, ¤djim≥ ˜louse™, ¤hj;m≥
˜rye™, ¤gjvjom≥ ˜aid, help™), in in¬nitives of velar-stems (¤gtxm≥ ˜to bake™), in the
imperative (¤gkfxm≥ ˜cry!™, ¤ckmim≥ ˜listen!™, ¤ht;m≥ ˜cut!™),21 and in the second
singular of the present tense (¤xbnftim≥ ˜you read™).

1.3.6 How the Cyrillic alphabet works (lexical idiosyncrasies)
In general, Russian writing can be converted automatically to a phonological rep-
resentation when it is supplemented by information about stress. There is only
a limited number of idiosyncratic instances in which spelling and phonology do
not match.
Orthographic ¤u≥ is pronounced as [v] in the genitive singular of masculine
and neuter adjectives -- for example, in ¤njuj≥ [t vj ˜that™, ¤gjcnjhjyytuj≥
[ . . . n˛ ¬v™] ˜outsider™. The same pronunciation occurs in the lexicalized geni-
tives ¤ctujlyz≥˜today™ and ¤bnjuj≥ ˜thus™. Historically this pronunciation goes
back to a sound change in which — g became [ ] in the southern half of the Rus-
sian language area, and was then reinterpreted as [v] in these words in central
dialects. Despite the spelling ¤cz≥, palatalization is now rare in the re¬‚exive
particle in the present tense and the masculine past (hd=ncz [rv˛ ots™], ,h’kcz
˜undertook™ [br’ls™]).
Some other peculiarities derive from the tension between [ ] and [ ] as the
pronunciation of ¤u≥. In individual lexical items with a sacral connotation, the
pronunciation of ¤u≥ as [ ] was maintained. The fricative is still possible in
interjections ¤ujcgjlb≥ ˜Lord™, ¤tq <jue≥ ˜oh God™, ¤fuf≥ ˜aha™, and was earlier
possible in the declension of the nouns ¤<ju≥ ˜God™ and in ¤,kfusq≥ ˜hon-
orable™ and ¤,jufnsq≥ ˜rich™. A fricative pronunciation is recommended in
¤,e[ufkmnth≥ ˜bookkeeper™, where it has a different source.
The fact that ¤u≥ was once widely pronounced as [ ] is indirectly responsi-
ble for another peculiarity of spelling. Foreign [h] was for a long time spelled
with Russian ¤u≥, because these foreign sounds were perceived to be similar to
[ ]. This convention was maintained long after ¤u≥ ceased to be pronounced
as [ ], and has carried over into modern borrowings, when it is pronounced
as [ ], not [ ]: ¤uevfybcn≥ ˜humanist™, ¤ujnntynjns≥ ˜Hottentots™, ¤ujyjhfh≥
˜honorarium™, ¤Ubnkth≥ ˜Hitler™. In recent years there is a tendency to use ¤[≥,
unless the spelling with ¤u≥ is already established: one discussion of Shake-
speare refers to ¤Ufvktn≥ ˜Hamlet™ and ¤{jncgeh≥ ˜Hotspur™. Note also ¤{tkmuf≥
˜Helga™, ¤{tkmcbyrb≥ ˜Helsinki™, ¤ntktcrjg bvtyb {f,,kf≥ ˜the Hubbell tele-
scope™, ¤Ejh[jkk≥˜Warhol™, ¤{fhktq-Lfdblcjy≥ ˜Harley-Davidson™.
21 The imperatives of verb stems ending in ¤o≥ take a vowel -- hßcrfnm, hs´ob -- suggesting ¤o≥
counts as a cluster.
24 A Reference Grammar of Russian

1.3.7 Transliteration
It is possible to convert words or whole texts written in Cyrillic into a Latin script
by transliterating: each Cyrillic letter is assigned to one or more Latin letters, and
the rules of conversion are applied blindly.22 For example, each time ¤u≥ occurs
in a Cyrillic text, the letter ¤g≥ is used in the Latin text; thus ¤Djkujuhfl≥ is
transliterated as ¤Volgograd≥, ¤Ufvktn≥ as ¤Gamlet≥ (though we know him by
another name), ¤njuj≥ as ¤togo≥ (though the ¤u≥ is pronounced as [v]). When
possible, the Latin equivalent is chosen so that its sound value corresponds to
the sound value of the Cyrillic letter.
A number of systems for transliteration are in use. They are quite similar,
and they are more or less equally adequate. There are also more informal, less
rigorous, strategies of Anglicizing isolated Russian words, used, for example, in
The linguistic system uses diacritics in preference to diagraphs for unusual
consonant letters, for example ¤x≥ is transliterated as ¤ˇ≥, using the Czech h’ˇek.
c c
The soft-vowel letters ¤z≥ and ¤/≥ are rendered as ¤ja≥ and ¤ju≥ in all positions,
whether they serve to mark a previous consonant as palatalized or to indicate
the presence of [j]. Cyrillic ¤q≥ is ¤j≥. In this system, Latin ¤j≥ has multiple
values: it occurs after a consonant in ¤djadja Vanja≥ (¤lzlz Dfyz≥), implying
[d˛ ], before a vowel in ¤Jalta≥ (¤Zknf≥), implying [j], and after vowels in ¤geroj≥
(¤uthjq≥). Thus in order to know what Latin ¤j≥ means, one has to know the
principles of Cyrillic writing. Cyrillic ¤э≥ is distinguished from Cyrillic ¤t≥ by
a diacritic, as ¤è≥ or ¤™≥ (continental). The linguistic system of transliteration
is rigorous in representing ¤=≥ when it is used in the source, and rigorous in
transliterating ¤m≥ and ¤(≥. The linguistic system is commonly adapted to serve
as a phonetic alphabet, a practice adopted here, though other sources prefer the
International Phonetic Alphabet.
All other systems avoid diacritics and use digraphs instead: ¤x≥ is ¤ch≥, ¤o≥
is ¤shch≥, and ¤w≥ is ¤ts≥. Differences concern how vowels and ¤q≥ are translit-
erated. One widely used system is that of the Library of Congress. The soft-vowel
letters ¤z≥ and ¤/≥ are rendered as ¤ia≥ and ¤iu≥, and Cyrillic ¤q≥ is also ¤i≥.
Thus the Latin transliterated letter ¤i≥ derives from multiple sources -- from
Cyrillic ¤b≥, obviously, but also from ¤q≥ and the soft-vowel letters ¤z≥ and
¤/≥. As a consequence, sequences such as ¤ii≥, ¤oi≥, ¤ei≥ are ambiguous. A fur-
ther dif¬culty is that spellings such as ¤Ialta≥ or ¤diadia Vania≥ or ¤Svetloiar≥
(¤Cdtnkjzh≥) seem not to be enlightening guides to English pronunciation.
The Library of Congress system, in its most rigorous formulation sanctioned
by the Library,23 uses ligatures (¤/≥ > ¤±u≥) and some diacritics (¤э≥ > ¤™≥), e

22 23
Neisweinder 1962, Shaw 1967/79, Hart 1983. Barry 1991.
Russian 25

Table 1.5 Romanizations of Russian Cyrillic

British Library of American Geographic Uppsala
Cyrillic linguistic System Congress Society popular Corpus

a a a a a a
b b b b b b
v v v v v v
g g g g g g
d d d d d d
e | ye†
e e e e (ye) e
¨ (e‡ )
e e e yo e (yo) oh
z zh zh zh zh zh
z z z z z z
i i i i i i
* (i‡ )
j i ¸ y y j
k k k k k k
l l l l l l
m m m m m m
n n n n n n
o o o o o o
p p p p p p
r r r r r r
s s s s s s
t t t t t t
u u u u u u
f f f f f f
x ∼ cho kh kh kh kh x
ts (ts‡ )
c ts ts ts c
c ch ch ch ch ch
s sh sh sh sh sh
sc shch shch shch shch w
(”‡ )
” ” qh
¯ (y‡ ) / ui§
y y y y y y
(™‡ )
™ ™ q
™ (e‡ )
è / ™o
э e © e e e eh
iu (iu‡ )
ju yu yu yu ju
±a (ia‡ )
ja ya ya ya ja

¤e≥ after consonant letter, ¤ye≥ elsewhere

less rigorous variant often used in practice
Continental variant
British Library in particular
26 A Reference Grammar of Russian

but these diacritics usually disappear in informal practice outside of the Library
itself. Similarly, prime and double prime, de¬ned as the Romanization of the soft
sign and the hard sign, are often replaced by a single or double closed quotation
mark, or omitted altogether. (Here they are maintained in transliterating names
of scholars, but not in Russian names in glosses.) Moreover, search programs in
electronic library catalogues ignore them.
The British System (British Standard 2979, 1958) renders consonant letters in
the same way, but has different equivalents for vowel letters: ¤z≥ =¤ya≥, ¤/≥
= ¤yu≥; ¤i≥ is used consistently for ¤q≥. The results in this system -- ¤Yalta≥,
¤dyadya Vanya≥, and ¤geroi≥ -- seem a more congenial guide to pronunciation
for English speakers. But there is a problem with ¤s≥, rendered in other systems
as ¤y≥. Hart™s Rules for Compositors (various editions, e.g., 1983) recommends ¤¯≥
for ¤s≥, but the diacritic disappears in practice, with the result that Roman ¤y≥
is used for two very different purposes. The British Library, whose practice is
re¬‚ected, for example, in the catalogues of books acquired (for example, British
Library 1974, 1979--87, 1986), uses ¤ui≥ for ¤s≥.
In the British System, the ending of proper names is simpli¬ed to ¤y≥, as in
¤Evgeny≥, ¤Klimenty≥, ¤Zlatopolsky≥. This sensible practice of simplifying and
domesticating proper names is becoming widespread.
In brief, each system has an advantage and a correlated disadvantage. The
British System has a more congenial way of rendering ¤z≥ and ¤/≥ than
the Library of Congress system, but does not have a good solution to ¤s≥.
The Library of Congress handles ¤s≥, but creates off-putting sequences such as
The US Board on Geographic Names of The American Geographic Society of
the Smithsonian Institute, like the British System, uses ¤y≥ in rendering ¤z≥ and
¤/≥. It even uses ¤ye≥ to render Cyrillic ¤t≥ in the position not after consonants --
in absolute initial position, after vowels, and after ¤m≥ and ¤(≥: ¤Dostoyevsky≥,
¤Yeltsin≥. This is roughly the strategy used in journalism to render Russian words
or names, though popular practice is less consistent than the transliteration
algorithms. Popular practice sometimes also transliterates Cyrillic ¤t≥ as ¤ye≥
even after consonants, leading to a profusion of ¤y≥: ¤Nye byt voynye≥24 (for yt
,snm djqyt! ˜there™ll be war no more™).
Computerization pulls in opposite directions. It has become easy to manipu-
late Cyrillic on computers. The letters of the Cyrillic alphabet are assigned to a
designated range of characters. These are not the ordinary characters, but ones
belonging to an enriched character table, and, with software, keystrokes are reas-
signed to that range. A mapping commonly used on the web is “KIO8,” for “rjl

24 Josef Skvorecky, The Engineer of Human Souls (New York, 1985), 357.
Russian 27

j,vtyf byajhvfwbtq, 8 ,bn” (Code for Information Exchange, 8-bit), or now
the speci¬cally Russian version “KIO8-R,” which assigns ASCII 192 through 255
(plus 179) to the Russian Cyrillic alphabet, which can then be typed, read, and
printed with the appropriate software.25 Microsoft devotes the interval from 0410
through 0451 of Unicode to Russian Cyrillic. Thus, on the one hand, because of
practical developments in computers, it has become increasingly natural sim-
ply to use Cyrillic without any transliteration, in discourse where acquaintance
with Russian can be presumed. On the other hand, there are many Cyrillic fonts
and mapping systems in use, and so far there is no standard for manipulating
Cyrillic. Accordingly, there is a pressure to simplify.26
The Library of Congress system and the British Standard have one prominent
ambiguity: transliterated ¤ii≥ can represent either gen sg ¤bcnjhbb≥ ˜history™
or gen pl ¤bcnjhbq≥; ¤oi≥ can represent either ¤jb≥ or ¤jq≥. Computerized
corpora develop strategies to avoid such ambiguities. The system of the Uppsala
Corpus, for example, is representative of the new mode of unambiguous Ro-
manization. The Uppsala Corpus uses digraphs with ¤h≥ for the unusual Cyrillic
consonant letters -- for example, ¤;≥ becomes ¤zh≥; it uses ¤j≥ for ¤q≥ and as
the operational graph in vowel letters -- for example, ¤z≥ becomes ¤ja≥. By using
¤j≥ consistently, ¤jq≥ and ¤jb≥ are distinguished in transliteration (as ¤oj≥ and
¤oi≥, respectively). This strategy may gain ground.
In e-mail communication with Russians (in the format of plain text in a Latin
alphabet), there is no standardized procedure. Not uncommon is a strategy like
that of computerized corpora, in which the unusual Cyrillic consonant letters
are spelled with digraphs with ¤h≥ as in most transliteration systems, while ¤j≥
is used for ¤q≥ and as the operational graph in vowel letters, for example ¤z≥
becomes ¤ja≥. Some Russians use ¤je≥ for ¤t≥ after vowels.
The various systems for Romanizing Cyrillic are similar and about equally
adequate. They face con¬‚icting demands. On the one hand, any transliteration is
supposed to be automatic and rigorous, and retain all the information contained
in the original, so that it is possible to reconstruct the original Cyrillic from the
Romanization. On the other, a transliteration is more congenial if it indicates
how Russian words might be pronounced and does not overwhelm the reader
with its foreignness. The two expectations inevitably con¬‚ict at certain points:
in the transliteration of ¤q≥, ¤s≥, and the soft-vowel letters, which have a dual
function in Russian, and also in the transliteration of ¤э≥, ¤=≥, ¤m≥, and ¤(≥.
25 Discussed on various sites, for example, http://koi8.pp.ru/.
26 One could note, for example, that of library catalogues accessible by the internet, Cambridge
University™s maintains ¤™≥, while Oxford™s has dispensed with it.

2.1 Sounds
Sounds are pronounced in different ways -- in one context as opposed to an-
other, from one occasion to the next, from one speaker to another. From these
different pronunciations in the ¬‚ow of speech, over the occasions of speech,
and across speakers, regular gestures and regular acoustic patterns can be ab-
stracted. The units derived by idealizing in this way will be written here in
square brackets.1 In Russian as in other languages, sounds can be classi¬ed into
vowels (stressed [’], unstressed [´], etc.) and consonants, which include obstru-
ents -- sounds made with a signi¬cant obstruction of the air ¬‚ow (such as [t], [z˛ ]) --
and sonorants (such as the nasal [n], the liquid [l˛], the glide [j]).
Russian phonology revolves primarily around two concerns: stress in vowels
and palatalization in consonants.2
Palatalization is an articulation of a consonant in which the blade of the
tongue moves toward the hard palate. For example, when the non-palatalized
“l” sound of w†k (w†ksq) ˜whole™ is pronounced, the tip of the tongue touches
near the teeth, while the middle of the tongue lies low in the mouth. In contrast,
when the palatalized “l” sound of w†km ˜goal™ is pronounced, the tip of the tongue
touches behind the upper teeth, and the blade and the middle of the tongue are
raised towards the hard palate. Most consonant articulations in Russian come
1 The discussion here, which is oriented around the level of phonology sometimes termed “broad
phonetic,” downplays questions of phonemics: non-linguists ¬nd the concept of phoneme unen-
lightening; variable rules respond to phonetic conditions; problematic cases of phonemic analysis
(in Russian, unstressed vowels; palatalized velars; [È]; [s˛ ‹ ], the sound corresponding to the letter
¤o≥) cannot be resolved without extensive discussion about the actual properties of the sounds,
rendering binary decisions about what is or is not phonemic uninformative.
Relationships among related sounds are viewed here as sets. Variants of stressed vowels are
grouped together as a set, with the most basic variant standing for the set. For example, [ó] stands
for the set including the sound [ó] that occurs between hard consonants and other variants, such
as the [o] that occurs between soft consonants, as in n=nz ˜aunt™. To discuss unstressed vowels in
55 „
relation to stressed vowels, the concept of a series of vowels is introduced.
2 Avanesov™s manual (1972) is informative about variation in phonology, if one corrects for its con-
servative standard. Panov 1990 is enriched by a valuable historical perspective. Matusevich 1976
and Bondarko 1977 have proven useful. Halle 1959 and Jones and Ward 1969 are good descriptions
in English. The research on variation (Panov 1968, Krysin 1974) is summarized and interpreted in
Comrie and Stone 1978 and Comrie, Stone, and Polinsky 1996.

Sounds 29

in two forms, with or without palatalization, like “l” sounds. It is convenient,
following the Russian tradition, to refer informally to non-palatalized conso-
nants as hard and palatalized consonants as soft. Non-palatalized consonants
are written by a symbol with no additional mark; [z] is the non-palatalized voiced
dental fricative that is the ¬rst consonant in, for example, p’k ˜hall™. The set of
non-palatalized consonants can be written as {C o }, or just C o , with a degree sign
to emphasize absence of palatalization. Palatalization is indicated by adding a
diacritic to the symbol used to represent the consonant. Various diacritics are
used: an acute accent ([z ] or [z]„ ), an apostrophe or single quote ([z™]), or -- and this
is the practice adopted here -- a cedilla; thus [z˛ ] represents the palatalized “z”
in dpz ˜he took™. The set of palatalized consonants can be represented {C}, or, ¸
more simply, as C. Palatalization, though a property of consonants, affects how
vowels are pronounced. Palatalization is also relevant to morphology.
Stress functions on many levels. Phonetically, stressed vowels differ from un-
stressed vowels ¬rst and foremost by being longer. As a consequence, stressed
vowels are more distinct in their pronunciation than unstressed vowels. Stress
is relevant to the lexicon and to morphology. Each lexical word -- noun, verb,
adjective, adverb -- has one syllable that is stressed. Accordingly, the number of
stresses in an utterance is the number of major words in the utterance. (This
excludes prepositions and particles such as ;t, which are written with spaces
as separate orthographic words, but do not have a stressed vowel.) Stress is not
assigned automatically to the same syllable in all words, such as the ¬rst sylla-
ble in Czech. Rather, different words can have stress on different syllables: v©rf
˜torment™ but ver’ ˜¬‚our™. Further, the place of stress can fall on different sylla-
bles in different in¬‚ectional and derivational forms of a word or word nest: thus,
gen sg cnjhjyß ˜side™, nom=acc pl cnj ´hjys; 1sg cvjnh/ ˜I see™, 2sg cvj
´ ´nhbim;
1sg jnjhd© ˜I will rip off™, pst pl jnjhd’kb, pst fem jnjhdfk’, psv msc sg jnj ´hdfy;
or nom sg ujkjd’, acc sg uj ´drf. Stress is then an ancillary marker
´kjde, dim ujkj
of morphology (in verbs: §3.2, in nouns: §3.6).
Stress plays a crucial role in the prosody of phrases. Shifts in intonation con-
tours occur on or around the stressed syllable (§7.2). To emphasize one word as
opposed to others, the stressed syllable is made louder, more prominent (some-
times termed sentence stress). Thus operations that deal with the prosody of
phrases are focused on stressed syllables.

2.2 Vowels

2.2.1 Stressed vowels
A word is organized in its phonetics around the one vowel that is stressed.
That stressed vowel is normally longer than other vowels. Vowels far away from
30 A Reference Grammar of Russian

the stressed vowel are very short. Vowels of the syllable immediately before the
stressed syllable are intermediate in duration; they are shorter than stressed
vowels, longer than other unstressed vowels. By virtue of being longer, stressed
vowels have more extreme articulations; the tongue has the time to reach further
to the perimeters of the vocal tract -- to be pronounced higher and further
front, or higher and further back, or lower down. Unstressed vowels, in contrast,
spend most of their modest duration in the transition away from a preceding
consonant and the transition to the following consonant; they do not reach the
same extremes of articulation (high or low, front or back) as stressed vowels.
If stressed vowels can be located on the perimeters of the vowel space shaped
like an inverted trapezoid, unstressed vowels form a smaller ¬gure inside the
space of stressed vowels. There are, evidently, ¬ve stressed vowel units in Russian
capable of distinguishing meanings of words, and a smaller number of distinct
unstressed vowels.
Vowels (and other sounds) can be classi¬ed both in terms of the articulatory
gestures used to produce them and the acoustic signals produced by these ges-
tures. To review the essentials of articulatory phonetics, vowels are produced by
allowing the air to ¬‚ow relatively freely through the oral cavity. The oral passage
can be given different shapes primarily by changing the position of the tongue
(and also by different positions of the lips and of the mandible), and different
vowel sounds result, which can be classi¬ed as front vs. back, high vs. mid vs. low,
and rounded (labialized) vs. unrounded (non-labialized). To review the essentials
of acoustic phonetics, the irregular shape of the vocal tract leads to a myriad
of harmonics of the fundamental frequency, F0 . The harmonics tend to cluster
within recognizable bands, or formants, measured at their centers in cycles per
second, or Hertz (= Hz). The ¬rst formant (clustering of harmonics), or F1 , is
proportional to aperture. Thus [’], the vowel produced with the widest aperture
and lowest position of the tongue, has the highest value of F1 , as high as 800 or

. 1
( 17)