Professional Documents
Culture Documents
Series Editors
Editorial Board
VOLUME 14
Βy
Thomas Olander
LEIDEN | BOSTON
Cover illustration: Novgorod birchbark letter no. 200 (c. 1240–1260). Drawing by 6- or 7-year old boy Onfim,
representing a horseman, the beginning of the Cyrillic alphabet and the boy’s name (Онфиме).
PG46.O73 2015
491.8—dc23
2015002197
This publication has been typeset in the multilingual “Brill” typeface. With over 5,100 characters covering
Latin, ipa, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities.
For more information, please see www.brill.com/brill-typeface.
issn 1875-6328
isbn 978-90-04-27049-7 (hardback)
isbn 978-90-04-270503 (e-book)
Acknowledgements vii
Phonetic Developments from PIE to Slavic viii
Abbreviations, Symbols and Conventions x
1 Introduction 1
1.1 Aim 1
1.2 Scope 3
1.3 Justification and Perspectives 5
1.4 Methodological Considerations 8
1.5 Terminology and Definitions 15
1.6 Structure of the Study 31
2 Phonological Background 39
2.1 Introductory Remarks 39
2.2 Reconstructed Language Stages 39
2.3 Phonological Developments from PIE to Slavic 46
3 Nominal Inflection 68
3.1 Introductory Remarks 68
3.2 Paradigm Types 69
3.3 Nominative(–Accusative) Singular 80
3.4 Accusative Singular 113
3.5 Genitive Singular 123
3.6 Dative Singular 144
3.7 Instrumental Singular 157
3.8 Locative Singular 168
3.9 Vocative Singular 180
3.10 Nominative–Accusative Dual 186
3.11 Genitive–Locative Dual 205
3.12 Dative–Instrumental Dual 213
3.13 Nominative(–Accusative) Plural 221
3.14 Accusative Plural 240
3.15 Genitive Plural 255
3.16 Dative Plural 267
3.17 Instrumental Plural 278
3.18 Locative Plural 286
vi contents
4 Verbal Inflection 296
4.1 Introductory Remarks 296
4.2 Paradigm Types 298
4.3 Present First-Person Singular 303
4.4 Preterite First-Person Singular 309
4.5 (Thematic) Imperative First-Person Singular 311
4.6 Present Second-Person Singular 312
4.7 Preterite Second-Person Singular 318
4.8 Imperative Second-Person Singular 320
4.9 Present Third-Person Singular 324
4.10 Preterite Third-Person Singular 330
4.11 Imperative Third-Person Singular 335
4.12 First-Person Dual 338
4.13 Second-Person Dual 342
4.14 Third-Person Dual 346
4.15 First-Person Plural 348
4.16 Second-Person Plural 355
4.17 Present Third-Person Plural 360
4.18 Preterite Third-Person Plural 363
5 Concluding Remarks 366
Bibliography 369
Index 408
Acknowledgements
This study was for the most part written between October 2008 and December
2013 while I was employed at the research project “Roots of Europe—Language,
Culture, and Migrations” at the Department of Nordic Studies and Linguistics,
University of Copenhagen. I am much indebted to my “Roots of Europe” col-
leagues, with whom I had many fruitful discussions: Adam Hyllested, Anders
Richardt Jørgensen, Benedicte Nielsen Whitehead, Birgit Anette Olsen, Bjarne
Simmelkjær Sandgaard Hansen, Guus Kroonen, Jenny Helena Larsson, Jens
Elmegård Rasmussen and Oliver B. Simkin.
In May 2013 Jens Elmegård Rasmussen unexpectedly passed away. His influ-
ence will be perceived on most pages of this study.
I would like to express my gratitude to Oliver B. Simkin, who reviewed part
of the manuscript, and to Guus Kroonen, who reviewed the whole thing. Over
the years I have also benefitted greatly from daily discussions about linguistic
and non-linguistic matters with my office mate Benedicte Nielsen Whitehead.
The members of the advisory board of the “Roots of Europe” project—
Andreas Willi, Brent Vine, Douglas Q. Adams, Joshua T. Katz, Michael Janda
and Rosemarie Lühr—were very helpful when I took their function literally
and asked for their advice during the yearly board meetings.
I owe a special thank you to Henning Andersen for his willingness to discuss
whatever I asked him about, including a number of topics that were relevant
to this study.
I am grateful to the Department of Nordic Studies and Linguistics for sup-
porting the entire endeavour, both morally and financially. Their financial sup-
port allowed for language revision which was meticulously carried out by Seán
Vrieland.
I am also indebted to the series editors of Brill’s Studies in Indo-European
Languages & Linguistics, Craig Melchert and Olav Hackstein, for accepting the
book in the series, and to Stephanie Paalvast at Brill for her patience. An anon-
ymous reviewer provided several useful comments and suggestions, many of
which have been incorporated in the text.
Finally, I wish to thank several other people who have contributed with
all kinds of useful input: Aigars Kalniņš, Aleksandar Loma, Andreas Øiestad,
Andrej Zaliznjak, Frederik Kortlandt, Holger Juul, Lars Brink, Marek Majer,
Martin Joachim Kümmel, Mate Kapović, Michaël Peyrot, Orsat Ligorio, Paweł
Nowak, Sofie Kluge, Stefan Jacobsson Schulstad, Thomas Hvid Kromann,
Tijmen Pronk, Tobias Mosbæk Søborg, Ursula Olander, Willem Vermeer and
Yoko Yamazaki.
Phonetic Developments from PIE to Slavic
Abbreviations
abl. ablative
acc. accusative
adess. adessive
adj. adjective
adv. adverb(ial)
Anat. Anatolian
aor. aorist
a.p. accent paradigm
Arm. Armenian
athem. athematic
Att. Attic
Bg. Bulgarian
BRu. Belarusian
BS Balto-Slavic
Čak. Čakavian
Cat. Catechism
com. common gender
CS Common Slavic
CS Cl “Classic” Common Slavic (i.e. excluding the Old Novgorod dialect)
CS N North Common Slavic (i.e. West and East Slavic)
CS S South Common Slavic
Cz. Czech
dat. dative
dial. dialect(al)
Dor. Doric Greek
du. dual
Elb. vocab. Elbing vocabulary
encl. enclitic
ESl. East Slavic
fem. feminine
gen. genitive
Gk. Greek
Gmc. Germanic
Go. Gothic
Hi. Hittite
Abbreviations, Symbols And Conventions xi
Symbols
C consonant
Cn n or more consonants
D voiced stop
h consonantal laryngeal
R sonorant consonant (liquid, nasal or semivowel)
T unvoiced stop
V (short or long) vowel
V̆ short vowel
V̄ long vowel
V̰ acute (glottalised) vowel
*X reconstructed item, including transpositions
**X hypothetical result of a development that is not supposed to have
actually taken place
μ mora
ˈX high tone
ˌX low tone (redundantly written before phonologically unaccented
word-forms)
# word boundary
. syllable boundary
+ attested from this point onwards
> < phonetically regular change
→ ← morphological reshaping or replacement
<X> orthographic representation
xiv abbreviations, symbols and conventions
Conventions
– when not preceded and followed by spaces, the slash (/) and the isogloss
mark (||) apply to the immediately preceding and following character only
– cross references of the type “§ 1.2.3” refer to chapter 1, section 2, sub-
section 3
– the references “see above” and “see below” refer to text within a section or
subsection
– bold numbers in square brackets, e.g. [12], refer to the phonetic develop-
ments listed in §§ 2.3.2–2.3.4
– “general” and “final” in §§ 2.3.2–2.3.4 refer to sound changes that take place
in all positions of the word and such that apply only in final syllables,
respectively
– a Proto-Slavic reconstruction is usually accompanied by the corresponding
Common Slavic reconstruction in brackets, e.g. PS ā-stem nom. sg. *naˈgā
(CS *nogà); see § 1.5.5
– for conventions in the notation of attested and reconstructed forms see
§§ 1.5.3–1.5.5
Chapter 1
Introduction
1.1 Aim
1 The point of departure is Proto-Slavic in the strict sense, not the more
loosely defined linguistic system Common Slavic on which similar stud-
ies are based (see §§ 1.5.5 and 2.2.3);
2 the entire system of inflectional endings in Proto-Slavic is taken into
account, not only a subset of it;
3 the relative chronological order of all relevant sound changes from Proto-
Indo-European to Proto-Slavic is made explicit and referred to through-
out the study (see § 2.3).
1.2 Scope
Since the focus of the study is on the reconstructed Proto-Slavic language and
its prehistory, the post-Proto-Slavic history of Slavic is treated less systemati-
cally; readers primarily interested in the more recent history of the develop-
ment of the Slavic inflectional endings should consult the rich literature on
that subject. Among the works that treat the history of the inflectional systems
of all the Slavic languages are Bräuer 1969a and 1969b, Vaillant 1958 and 1966,
and Vondrák 1908/1928. There are also numerous studies of the inflectional
systems of the individual Slavic languages, as well as an abundant literature
treating special problems within one or more languages. As the main goal of
most of those studies is not to reconstruct Proto-Slavic and its prehistory, they
are not included in the systematic references to literature in chapters 3 and 4.
In keeping with the focus on Slavic in this study, endings and categories that
existed in Proto-Indo-European and Proto-Balto-Slavic are only included if
they are relevant for the discussion of Slavic.
As the centre of our attention is the reconstruction of the Proto-Slavic inflec-
tional endings and their prehistory, inflectional categories that have plausibly
arisen at later stages of Slavic are only taken into account insofar as they pro-
vide indispensable information about the Slavic proto-language. Thus a reader
looking for a discussion of topics like the spread of the u-stem ending PS *‑au̯ u
(CS *‑ovъ) to the o-stems in the genitive plural in a number of Slavic dialects, or
the inflection of the definite forms of the adjective, will return empty-handed.
The same limitation applies to the phonological level, for instance the rise of
new quantitative distinctions in final syllables in Slavic (see § 2.2.3). These
problems, in my opinion, belong to the separate histories of the individual
Slavic languages and are therefore outside the scope of this study.
The Proto-Slavic inflectional accent may be considered the result of the
prosodic properties of the morphemes (cf. Dybo 1981: 261–262 and passim, on
Proto-Balto-Slavic; Zaliznjak 1985: 121–125 and passim, on early Old Russian).
Accordingly, I specify the accentual property of an ending in Proto-Slavic
and Proto-Balto-Slavic reconstructions—that is, its accentedness or unac-
centedness in the mobile accent paradigm—whenever the material allows
it. Contrary to Dybo, Zamjatina and Nikolaev (1990: 107–108, on Proto-Balto-
Slavic), I do not think that the “prosodic properties” have direct phonetic cor-
respondences, such as high or low tones on each morpheme of a word-form;
the properties are the outcome of the linguist’s analysis of the position of the
accent in inflection and derivation. Since I have already analysed the Slavic
and Baltic systems of paradigmatic accentuation and presented my opinion
4 chapter 1
*‑ā (CS *‑a) matches the Proto-Indo-European ablative singular *‑ah₂ad per-
fectly, and this fact is sufficient for comparing the two forms. The fact that the
Proto-Slavic form has genitive function, whereas the Proto-Indo-European
form is an ablative, is not given much attention in the present study.
On a more practical note, I have considered it unnecessary to compile an
index of words for this study. The study is not about lexemes but about end-
ings, and the endings are presented systematically in the relevant chapters.
It may be relevant to ask what justifies a study like the present one. Of course,
those with a specific interest in Proto-Slavic inflectional morphology will not
question its relevance. Similarly, the subject of the study will be relevant to
most Indo-Europeanists, as well as to Slavists and general linguists interested in
historical morphology and phonology. But it is also worth considering whether
the study may be of interest to a broader audience. I think it may.
Like other historical disciplines—archaeology, human genetics, history of
religions etc.—historical linguistics provides us with insights about our past.
Elements of the history of a language may reveal something about the his-
tory of the speakers of that language and of the society in which they lived.
Correspondences between two or more languages reveal something about the
historical relationship between these languages and, accordingly, among the
speakers of the languages.
This is perhaps most obvious when we are dealing with the history of lex-
emes. Knowing, for instance, that most Indo-European languages (though not
Anatolian) share specific genealogically related words for ‘wheel’, ‘axle’ and
similar concepts, we may conclude that the speakers of the ancestor language
had wheeled vehicles (see e.g. Anthony 2007: 59–82; 2013: 3). Similarly, loan-
words may tell us something quite specific about the society of the speakers of
both the recipient and the donor language.
If it is clear that lexical reconstruction, or etymology, contributes to our
extra-linguistic knowledge of the past, then what about historical morphology,
i.e., in our case, the reconstruction of the inflectional endings of Proto-Slavic
and their prehistory? The extra-linguistic relevance of historical morphology
is less obvious than in the case of etymology, but this does not mean that his-
torical morphology is not of interest to non-linguists. The fact that the mor-
phological system of a given language, as opposed to its lexicon, is not directly
linked to real-world phenomena means that it preserves another, supplemen-
tary, kind of information about the past. The methods of lexical reconstruction
6 chapter 1
are distinguished only in Anatolian (e.g. Mallory & Adams 2006: 55). This
model easily explains certain correspondences betwen non-Anatolian Indo-
European and Anatolian. However, the model has more entities and is thus
less simple. Today most scholars assume that the proto-language had three
laryngeals, which may be seen as a kind of compromise between simplicity
and explanatory power. A related problem, namely the possible effect of the
accent on the development of segments in the prehistory of Proto-Slavic, is
discussed in § 1.4.2, where it is argued that there probably was no such effect.
In the domain of inflectional morphology some scholars assume that Proto-
Indo-European possessed a paradigmatic allative case, preserved in Anatolian
(see Fortson 2004/2010: 113, 117). Again, this assumption allows for an easy
explanation of the Anatolian material, but at the expense of adding a new cat-
egory to the Proto-Indo-European inflectional system.
Similarly, the Leiden School reconstructs the Proto-Indo-European ā-stems
with an ablauting eh₂-suffix, showing forms like nom. sg. *‑h₂, acc. *‑eh₂‑m, gen.
*‑h₂‑os, gen. pl. *‑h₂‑om (e.g. Beekes 1995/2011: 200), whereas most other schol-
ars reconstruct a non-ablauting suffix *‑eh₂‑ throughout the paradigm. While
the alternations of the ā-stems as reconstructed by the Leiden School follow
the pattern of other hysterokinetic stems and in this sense can be regarded as
regular, the allomorphs provided by this system make it easy to explain many
apparent irregularities in the individual attested languages. For instance, the
unexpected short vowel of La. nom. sg. ‑a (not *‑ā) may directly reflect PIE
*‑h₂, whereas most other languages have introduced reflexes of the full-grade
allomorph *‑eh₂ in the nominative singular. Similarly, CS gen. pl. *‑ъ may
directly reflect PIE *‑h₂om, in contrast to the endings found in most other lan-
guages which point to *‑eh₂om or *‑āom, remade on the basis of the full-grade
variant *‑eh₂‑. As it is methodologically problematic to posit reconstructions
that can hardly be falsified, this type of reconstruction should be supported
by unambiguous evidence in order to be acceptable. In the case of the sup-
posed hysterokinetic paradigm of the ā-stems I do not think this is the case.
For methodological considerations of the same kind see Olander 2009: 212.
On a related note, Andersen (2009a: 7–8) observes that traditional inter-
pretations of morphological changes with reference to analogical relations in
the system are unsatisfactory, first because they are post hoc, second because
they do not pay sufficient attention to the relations between expression and
content. I feel that this point of criticism is also relevant to the present study.
In order to meet the criticism all morphological changes posited in the study
would have to be analysed from a different perspective. This would almost
undoubtedly lead to new insights into the more general lines of morphological
change relevant for the development of Proto-Indo-European into Proto-Slavic
10 chapter 1
and change our views on some of the morphological changes that are posited
in this study. Such an endeavour would also constitute a whole study in itself
and therefore cannot be undertaken here.
An important point where different methodological approaches may lead
to different analyses is the value ascribed to direct vs. indirect evidence. For
example, to some scholars an important argument in favour of the idea that
the regular reflex of PIE *‑os is CS *‑o (and not *‑ъ, as the competing view holds)
is that in i̯o-stems the inherited vocative singular form has been replaced with
the i̯u-stem form *‑ju. This replacement is easier to understand if the inherited
vocative form *‑je and the nominative ending *‑je < *‑jo < PIE *‑i̯os had merged,
leading to an inconvenient homophony between these forms (Vermeer 1991/
2009: 10). While I do understand the reasoning, I find it too indirect to be com-
pelling, and such arguments do not play any significant role in this study.
Although the main subject of this study is Slavic, I have tried to present and
briefly account for the most important problems pertaining to the inflectional
endings of the non-Slavic Indo-European languages. By doing so relatively sys-
tematically throughout the study, not leaving too many problems of the non-
Slavic languages aside, I hope to have avoided at least some potential cases of
Teeter’s law (“the language of the family you know best always turns out to be
the most archaic”, Watkins 1976/1994: 310).
A special subfield of historical phonology covers the Auslautgesetze, or laws
of final syllables. It is well known that a given sound or sequence of sounds
may have different developments in final syllables than in non-final syllables.
Although laws of final syllables do not constitute the centre of interest of this
study, it is evident that they play a significant role here as inflection takes place
at the end of the word in the languages dealt with in this study. I think it is
fair to say that laws of final syllables are, in most cases, more controversial
than general sound laws. As Pedersen commented more than a century ago,
it is “nicht verlockend, das gebiet der auslautsgesetze zu betreten, wo man
immer viel geschrei und wenig wolle findet” (1905a: 320–321; he did so anyway,
though).
There are at least three reasons why laws of final syllables are more difficult
to deal with than general sound laws (see also Holzer 1980: 7–9): (1) being by
definition restricted to a certain environment, laws of final syllables are rela-
tively specific already from the outset; (2) also because of the restricted envi-
ronment, the set of relevant instances is often small, in many cases limited to
a few ones; (3) since the majority of the instances of laws of final syllables con-
cern inflectional endings in the languages treated here, there is almost always
a potential alternative explanation in terms of analogy.
introduction 11
(cf. Gk. nom. pl. ἀγροί) (see e.g. Bräuer 1961: 104). While it cannot be entirely
excluded that this is the historically correct explanation, for the reasons just
mentioned it should only be accepted if there are no viable alternative expla-
nations that do not refer to alleged tonal characteristics of the elements in
question (for such alternatives see Olander 2009: 90 with references and an
explanation I no longer maintain; for my current view see Olander 2012: 332
and [17]).
to a proto-form *‑āi̯. Applying our knowledge that ā-stems show a suffix *‑ah₂‑
in Proto-Indo-European and that the dative singular marker is *‑ei̯ in other par-
adigms, we may, by using internal reconstruction, reconstruct an older stage
of that ending as *‑ah₂ai̯ (with colouring of *e to *a near *h₂). Whereas the
comparative method allows us to establish both regular and irregular forms in
reconstructed systems, internal reconstruction is suited to finding underlying
regularities.
There are cases where it is difficult to find the balance between the two
methods used in linguistic reconstruction. The following two examples dem-
onstrate how more emphasis put on either of the two methods may lead to
different reconstructions.
In the i-stem dative singular the Indo-Iranian material, e.g. Ved. ‑aye, points
to a Proto-Indo-European ending *‑ei̯ei̯ (see § 3.6.2). This form matches the
expectations we have on the basis of internal reconstruction, as it consists of
the i-stem suffix *‑ei̯‑ followed by the dative singular marker *‑ei̯. However, both
Italic and Balto-Slavic point to a shorter ending *‑ei̯. Thus internal reconstruc-
tion, supported by Indo-Iranian evidence, clearly points to *‑ei̯ei̯, yet it may be
argued that the actual proto-form is more likely to be *‑ei̯. The reason is that a
form *‑ei̯ei̯ may easily have replaced original *‑ei̯ in a pre-stage of Indo-Iranian
because it would fit better into the system. On the other hand, it is difficult to
imagine that *‑ei̯ei̯ was replaced with *‑ei̯ not only once, but twice, in branches
that do not otherwise have much in common, viz. Italic and Balto-Slavic.
In this case the result arrived at by internal reconstruction is less likely to be
correct precisely because of its internal transparency. This way of reasoning,
recognised already by Rasmus Rask in 1818 (see Morpurgo Davies 1998: 128),
is referred to as the principle of the forma difficilior. While there is disagree-
ment about the proto-form, most scholars would agree that the ending *‑ei̯ei̯
did exist at least at some pre-stage of the proto-language; the problem is a
chronological one. Here I should like to emphasise that the prehistory of the
Indo-European proto-language as such lies outside the scope of this study; the
only purpose of applying internal reconstruction to the proto-language is to
throw light on the inflectional endings of Proto-Indo-European itself, i.e. the
last common pre-stage of the Indo-European languages.
Another interesting example of incongruity between the results obtained
through the comparative method and through internal reconstruction is the
Proto-Indo-European thematic present second-person singular (see § 4.6). The
traditional reconstruction of the ending, PIE *‑esi, directly leads to the Indo-
Iranian, Italic and Germanic endings and is supported by internal reconstruc-
tion. The Balto-Slavic forms, however, are difficult to derive from this ending;
introduction 15
the Greek and Celtic evidence is ambiguous. This has led some scholars to
posit a Proto-Indo-European ending *‑ei̯ or *‑ehi, not supported by internal
reconstruction but directly yielding the Balto-Slavic endings. In this scenario
Indo-Iranian, Italic and Germanic would have normalised the ending on the
analogy of the athematic ending PIE *‑si.
Dilemmas like these are commonly encountered in the reconstruction of
inflectional endings and it is often difficult to find the right balance between
likely innovations and systematic reconstructions. The application of the prin-
ciple of the forma difficilior may sometimes be difficult to distinguish from the
fallacy of obscurum per obscurius.
The method that is most relevant for establishing Proto-Slavic is the com-
parative method. The Slavic proto-language is reconstructed primarily by
applying the comparative method to the attested Slavic languages and by tak-
ing into consideration loanword relations with neighbouring languages and
dialects. Evidence from related languages, first of all Baltic, may also contrib-
ute to clarifying the reconstruction of a Proto-Slavic form. For the reconstruc-
tion of pre-stages of Proto-Slavic, internal reconstruction only plays a minor
role because of the abundant comparative material available from Baltic and
the other Indo-European languages.
ing forms of the paradigm, e.g. nom. sg. *dukˈtī (*dъt’ì), with an oblique stem
*dukter‑ (*dъt’er‑).
The picture is quite different when we apply the same terminology to the
nominal system of the Indo-European proto-language. A word-form like PIE
�̥
nom. pl. *u̯ ĺkʷ‑o‑es �̥
consists of a stem *u̯ ĺkʷo‑, formed with a stem-forming suf-
fix *‑o‑ and an ending *‑es, which is also the case–number marker. Very early
in the prehistory of Slavic, however, the stem-forming suffixes and the endings
began a phonetically conditioned merging process in the o- and ā‑ stems and,
less pervasively, in the i- and u-stems. In the consonant stems, by contrast, the
consonant prevented a similar process, and one allomorph of the stem-form-
ing suffix was generalised in all forms.
The personal pronouns are characterised by stem suppletion between the
nominative and oblique forms of each number, and among the singular, dual
and plural forms. Alternatively, it is possible to regard the singular, dual and
plural forms as distinct lexemes, in which case there is a suppletive relation
only between the nominative and the oblique (see also § 3.2.7). In any case,
the singular, dual and plural forms have singular, dual and plural inflection,
respectively. The reflexive pronoun inflectionally groups with the singular of
the personal pronoun.
In the verbal system the situation is somewhat more complicated. Two sets
of endings are distinguished in the present, the preterite and the imperative
systems; these sets of endings are referred to as the thematic and athematic
endings. Most thematic endings are segmentable into an inflectional suffix
plus a person–number marker, e.g. PS prs. 3sg. *‑e‑ti; in these cases the person–
number marker is identical with the corresponding athematic ending, and the
two endings are treated together.
Verbal lexemes may combine a thematic present with either a thematic or
an athematic aorist, whereas an athematic present is always combined with
an athematic aorist, thus resulting in three verbal macroparadigms. I have
used third-plural forms to illustrate the macroparadigms, since this category
expresses the distinction between thematic and athematic endings both in the
present and in the aorist:
1 Thematic present with thematic aorist, e.g. PS prs. 3pl. *maˈganti (CS
*mògǫtь), aor. 3pl. *maˈgan (*mogǫ̀ );
2 thematic present with athematic aorist, e.g. PS prs. 3pl. *u̯ edanˈti (CS
*vedǫ́ tь), aor. 3pl. *u̯ ēsin (*věsę);
3 athematic present with athematic aorist, e.g. PS prs. 3pl. *dāˈde/inti (CS
*dádętь), aor. 3pl. *dāˈxin (*dašę̀).
18 chapter 1
The present and aorist stems are in some cases identical, as in PS *mag‑ (CS
*mog‑), but they may also be distinct, as in prs. *u̯ ed‑ (*ved‑) vs. aor. *u̯ ēs‑ (*věs‑).
The so-called ox-aorist, e.g. OCS 3pl. vedoxъ, is a post-Proto-Slavic innovation.
An interesting problem connected to the concept of a paradigm is the sta-
tus of the vocative form of nouns and adjectives. As recently emphasised by
Andersen (2012, esp. pp. 139–147) on the example of Russian, the vocative form
is not a declensional word-form but a derived form. Thus the vocative should
not, strictly speaking, be included in a treatment of inflectional morphol-
ogy like the present one. On the other hand, since the vocative has tradition-
ally been included in similar surveys and fits well into the general structure
of this study, it is treated on par with the proper inflectional forms. However,
the insight that the vocative form represents a derivation allows us to better
understand certain apparent irregularities related to it.
It is well known that vocative forms in Slavic and other languages have a
special prosodic behaviour. For instance, all forms of Čakavian (Novī) ženȁ are
accented on the second syllable, except the vocative forms sg. žȅno, pl. žȅne,
which have a falling tone on the first syllable. Data from the remaining Slavic
languages confirm that this pattern goes back to Proto-Slavic: PS *geˈnā (CS
*ženà) was accented on the second syllable in all forms except the vocative,
which was unaccented, e.g. voc. sg. *ˌgena (*žȅno) (§ 3.9.3). This puzzling pro-
sodic alternation becomes much easier to understand when it is realised that
the vocative was actually not part of the inflectional paradigm.
this study. In the following paragraphs I shall go into more detail with my view
on the concept of a proto-language. For the distinction between “Proto-Slavic”
and “Common Slavic” see § 1.5.5.
According to Antonsen (1994: 58), “a proto-language is by definition a
wholly reconstructed, unrecorded parent stage of a family of languages”
(see also Nielsen 2000: 32; Marchand 1973: 649–650). In my view nothing is
gained by not allowing, by definition, attested linguistic material to represent a
proto-language. A text that corresponds exactly to our expectations of a proto-
language may be regarded as a text in that proto-language.
It is useful to make a distinction between two types of proto-languages.
Some proto-languages, such as Proto-Indo-European, do not have any close
relatives and thus, in most respects, constitute the ending point of a proce-
dure of application of the comparative method; we may refer to such proto-
languages as primary proto-languages (Penzl 1972: 31, who also uses the term
“super-proto-languages”). Here the hypothesis of a genealogical relationship
between Indo-European and the Uralic languages should not be ignored (see
e.g. Pedersen 1933; 1931/1962: 336–338; Collinder 1934/1964: 3–4, 80–82 and pas-
sim; Schindler 1964, who proposes seven Indo-European–Uralic lexical cor-
respondences, albeit “[o]hne auf die umstrittene Frage des ‘Indo-Uralischen’
in extenso einzugehen” [p. 171], Cowgill 1974/2006: 32; Kortlandt 1990/2010a:
6; 2010b/2010a: 37; Comrie 2002: 412; Rasmussen 2005; Klingenschmitt 2005:
114–116; Kloekhorst 2008b; Hyllested 2009: 112–113; Kümmel forthc.: 8–9; cf.
the negative appraisal by Kessler & Lehtonen 2006: 39–40; Ringe & Eska 2013:
266, 276). But even if confirmed the relationship would be too distant, and the
reconstruction of the Indo-Uralic proto-language would be too fragmentary to
be relevant here; in practice, Proto-Indo-European is a primary proto-language.
Other proto-languages, such as Proto-Slavic, have close relatives and are
intermediate proto-languages between their descendant languages—in this
case, the Slavic languages—and the proto-language from which they them-
selves descend—namely Proto-Balto-Slavic, itself an intermediate proto-
language (see e.g. Holzer 1995a: 55; for the terminology cf. W. Hock 2006:
2–3 with n. 2). Another example of an intermediate proto-language is non-
Anatolian Indo-European (see § 1.5.3), which is a stage between between Proto-
Indo-European and Proto-Balto-Slavic, Proto-Indo-Iranian, Proto-Greek etc.
When analysing a group of related languages retrospectively, it is, to some
extent at least, possible to determine which developments have taken place in
the period between the disintegration of the common ancestor of those lan-
guages and each individual language, and to order those changes chronologi-
cally. The first innovation not shared by all descendant languages marks the
dissolution of the linguistic system we are interested in, i.e. the proto-language.
20 chapter 1
The proto-language itself was, by all probability, only one dialect among others
which have disappeared (cf. Holzer 1995a: 56–57).
It is obvious that a proto-language reconstructed by means of the com-
parative method does not provide a complete picture of the actually spoken
proto-language—just as even a comprehensive description of Upper Sorbian
or modern English only represents a fraction of the actual linguistic system
of Upper Sorbian or modern English. On the other hand, there can be no
doubt that historical linguistics does allow us to establish facts about proto-
languages and the world of their speakers (cf., however, the more sceptical
attitudes towards the reality of proto-languages expressed in Clackson 2007: 16
[criticised in Anthony 2013: 5] and Pulgram 1959, 1961). We should furthermore
be aware that in linguistic reconstruction it is often difficult, or impossible, to
determine the relative chronology between various innovations, for instance
between two phonological innovations, or between a phonological and a mor-
phological innovation.
A corollary of the definition of a proto-language given above—the last stage
before the first innovation not shared by all descendants—is that there is no
variation in a proto-language, i.e. that it has no dialects, sociolects, stylistic vari-
ation etc. (see e.g. Schlerath 1981: 180: “Es ist das Wesen jeder Rekonstruktion,
daß in ihr notwendigerweise alle diachronischen oder dialektischen oder sonst
irgendwie bedingten Unterschiede aufgehoben sind”; Eichner 1988: 11–20 with
references and discussion; Fox 1995: 133–136; Holzer 1995a: 56–57, 73, accord-
ing to whom variation in categories that are not reconstructed is irrelevant;
Weiss 2015: 138). In practice, however, the concept of a proto-language becomes
more useful, and closer to real languages, if we loosen that definition and
accept some variation in proto-languages (see e.g. Penzl 1972: 30; Rasmussen
forthc. b). It often makes sense to reconstruct two or more phonologically dis-
tinct variants of a morpheme, instead of choosing one of them as the original
one and the other one as an innovation, thereby projecting the origin of the
variation back to an earlier stage. The variants may, for instance, have been
stylistic variants that existed side by side in the proto-language.
Thus a reconstructed language stage may be considered a proto-language
even if it displays variation, although, strictly speaking, variation at one stage
of a language is the result of linguistic developments at earlier stages of that
language. The usefulness of allowing for variation in the definition of a proto-
language becomes especially relevant when we are dealing with intermedi-
ate proto-languages. It is sometimes the case that variation in what seems to
be a clear instance of an intermediate proto-language goes back to variation
already present in a higher-level proto-language. If a strict definition of a proto-
language is applied, this would mean that the intermediate proto-language in
introduction 21
question never existed, which is hardly a desirable situation for the historical
linguist. To mention a relevant example, certain dialectal differences among
the Slavic languages may go all the way back to Proto-Balto-Slavic (see § 1.5.4);
under a strict definition of a proto-language, with no variation, there would
have been no Proto-Slavic.
Accordingly, some variation—sporadic variation in specific morphemes—
should be permitted in the definition of a proto-language. When we are facing
more systematic variation, it is usually better to speak of a common language
and to place the proto-language further back in the relative chronology, to a
stage when the variation has not yet arisen. For instance, in the i̯ā-stem accu-
sative plural ending we find reflexes of *‑jě in West and East Slavic, but of *‑ję
in South Slavic. These endings should not be regarded as variants in a proto-
language, but we should establish one common proto-form since the variants
are not the result of sporadic developments; in this specific case, I reconstruct
PS *‑i̯ə̄n (see § 3.14.5).
In the following subsections I shall examine how the terminology presented
in this subsection is applied to the specific reconstructed language stages
dealt with in this study, viz. Proto-Indo-European, Proto-Balto-Slavic and
Proto-Slavic.
1.5.3 Proto-Indo-European
In accordance with the definitions given in § 1.5.2, Proto-Indo-European refers
to the last stage before the first innovation that did not affect all known descen-
dant languages. Because of the special position of the Anatolian branch it has
become increasingly clear that the traditional understanding of Proto-Indo-
European, as exposed in e.g. Brugmann’s classical Grundriss (second edition
published between 1897 and 1916), must be modified.
Most Indo-Europeanists currently agree with Sturtevant (e.g. 1962) that
Anatolian was the first branch to split off from the remaining Indo-European
languages (see Ringe 2006a: 5; for a notable recent exception cf. the polemic
statement by Tichy 2006: 16: “A current suggestion that Anatolian was the
first and Tocharian the second language to branch off, as compared with the
Neogrammarian tradition, partly has provoked and partly may be influenced
by Teeter’s Law”). The status of the relationship between Anatolian and the
non-Anatolian Indo-European branches is a matter of much debate, especially
when it comes to the degree of retention and innovation in the two branches
(see below). The Slavic language group belongs to the non-Anatolian branch
of Indo-European, variously referred to as “Core Indo-European” (thus e.g.
Melchert 2014: 257 n. 1 and passim), “IE Proper” (Cowgill 1985a/2006: 108),
“North Indo-European” (Ringe 2006a: 5–6), “non-Anatolian Indo-European”
22 chapter 1
conservative Indo-European languages. When the branches that are not sys-
tematically represented here, viz. Celtic, Armenian, Albanian and Tocharian,
offer relevant material for the reconstruction of a Proto-Indo-European end-
ing, material from those branches is also presented and discussed.
1.5.4 Proto-Balto-Slavic
The Baltic branch of Indo-European comprises the two East Baltic languages,
Lithuanian and Latvian, and the extinct West Baltic language Old Prussian. For
the question of the reality of a Baltic branch see below in this subsection.
In the discussion of the subgrouping of the Indo-European language branches
the so-called Balto-Slavic question is one of the key issues: can the attested
Baltic and Slavic languages be traced back to a common post-Proto-Indo-
European—or, more precisely, post-non-Anatolian Indo-European—ances-
tor or not? I think the answer is that they can, and should, at least within the
framework of the present study. By tracing back the identical developments in
the two branches to a common ancestor we obtain the simplest model of the
relationship between Baltic and Slavic, without a notable loss of explanatory
power. In order to avoid misunderstandings I shall briefly elaborate on my con-
ception of the relationship between Baltic and Slavic (see also Olander 2009:
9–10; a very useful history of research is presented in W. Hock 2006; see also
Young 1994; Holzer 1995c: 305).
I assume that after the disintegration of non-Anatolian Indo-European
there was a period of development common to the later Baltic and Slavic
branches. The dialects of pre-Proto-Balto-Slavic were similar enough to each
other to carry out identical innovations shared by all dialects, for instance the
introduction of paradigmatic accentual mobility in vowel stems [4] and other
prosodic developments (see Olander 2009: 163 and passim; but cf. Andersen
2009b: 19: the loss of the ictus from a final high mora is “a contact innovation
[. . .] technically [it] is not a shared Slavic and Baltic innovation”). The traces
of pre-Proto-Balto-Slavic dialectal variation that are left in Proto-Balto-Slavic
are, as far as we can see, only sporadic. A number of innovations shared by all
Balto-Slavic languages may easily be regarded as older than the oldest non-
sporadic innovation not shared by all languages.
It seems clear, though, that there are cases of variation that cannot be
avoided in a reconstructed Balto-Slavic proto-language, such as the existence
of different lexemes for the same notion, or the existence of variants with ini-
tial *a or *e in the same lexeme in different areas (Andersen 1996: 206 and
passim). But these cases do not have a systematic character and are not signifi-
cant enough to outweigh the analytical advantages of positing a Balto-Slavic
introduction 25
1.5.5 Proto-Slavic
The Slavic languages are traditionally divided into a South Slavic, an East
Slavic and a West Slavic group. The South Slavic group includes Slovene, Serbo-
Croatian, Bulgarian and Macedonian and is represented here by Old Church
Slavonic. East Slavic comprises Russian, Belarusian and Ukrainian and is rep-
resented by Old Russian and the Old Novgorod dialect. West Slavic includes
Czech, Slovak, Upper and Lower Sorbian, Polish, Kashubian and the extinct
Polabian language and is represented by Old Czech. It is sometimes useful to
have a distinction between North Slavic, comprising West and East Slavic, and
South Slavic. Furthermore, in this study a distinction is made between the Old
Novgorod dialect on the one hand and Classic Common Slavic, comprising
26 chapter 1
the remaining Slavic dialects, on the other; for the term “Common Slavic”
see below.
The Old Novgorod dialect is attested in birchbark letters from the Novgorod–
Pskov area, the earliest of which date back to the eleventh century (see, also
for the following remarks, Zaliznjak 1988: 164). The birchbark letters represent
some of the oldest written attestations of Slavic. The Old Novgorod dialect is
particularly important because it differs from the remaining Slavic dialects by
exhibiting phonological archaisms not preserved elsewhere (Zaliznjak 1991;
but cf. Bjørnflaten 1990). Moreover, the texts represent the vernacular of the
time and thus give a welcome counterweight to other early Slavic texts, which
are mostly translations with religious content. Although the first birchbark let-
ters were discovered in Novgorod in the early 1950s, the significance of the evi-
dence from this dialect has only been fully acknowledged in the last decades.
The material has not yet been taken systematically into account in compre-
hensive treatments of Slavic inflectional morphology, although various parts of
the system have been handled in separate studies. This study aims at attaching
due weight to the evidence from the Old Novgorod dialect.
The terminology regarding reconstructed stages of Slavic is heterogeneous
(see Andersen 1985; Mareš 1986; Lunt 1997; Feeney 2003: 42–43). The different
use of the terms makes it difficult to compare various researchers’ conceptions
of the prehistoric development of Slavic. Since Proto-Slavic constitutes the
point of reference in this study, a precise understanding of the use of the term
will be appropriate.
In accordance with the general definitions given in § 1.5.2, Proto-Slavic is
defined in terms of the relative chronology of linguistic changes: it is the last
stage of Slavic before the earliest innovation that is not shared by all Slavic
dialects (similarly e.g. Lindstedt 1991: 112; Holzer 2004: 50: “the last common
ancestral form of all Slavic languages and dialects”); cf. the alternative defini-
tions of Proto-Slavic, e.g. “the earliest specifically Slavic systems” (Lunt 1981:
41), “the initial stage of Slavic” (Andersen 1985: 81) or “the earliest stage of
Slavic that can be reached by means of the retrospective methods we use in
comparative and internal reconstruction” (Andersen 1996: 183; similarly Lunt
1981: 41; Igartua 2005a: 89–93). While Holzer’s Proto-Slavic linguistic system is
approximately the same as the one referred to in the present study, he consid-
ers it impossible to establish the relative chronology between the oldest iso-
gloss separating Slavic from the remaining Indo-European languages and the
oldest isogloss within Slavic, but he adduces external evidence—from histori-
cal sources, dialect geography and loanword relations—for a uniform Slavic
proto-language (Holzer 1995a: 58–73). Like any proto-language by the defini-
tion of § 1.5.2, Proto-Slavic denotes a specific linguistic stage here, not a period
introduction 27
of development (thus also e.g. Holzer 2007: 15; but cf. e.g. Trubeckoj 1922/1988:
218: “La période protoslave, durant laquelle le slave commun n’était qu’un dia-
lecte indo-européen conservant encore avec les dialects voisins des relations
assez intimes” [emphasis as in original]). In contrast to the question of the
existence of a Balto-Slavic or even a Baltic proto-language (see § 1.5.4), hardly
anyone doubts there to have been a Slavic proto-language (see e.g. Stang 1966:
12; Holzer 1995a).
There was some sporadic variation in the Slavic proto-language, e.g. in the
initial vocalism (Andersen 1996: 9–19) and, as we shall see in this study, in the
inflectional endings. Holzer, to whom variation in proto-languages is only rel-
evant in so far as it concerns the reconstructed categories (see § 1.5.2), aims at
reconstructing a uniform Proto-Slavic (1995a: 73). This may be practicable in
some respects, but it is difficult to do consistently when dealing with inflec-
tional morphology.
In order to establish which chronological stage in the development of Slavic
should be labelled Proto-Slavic, the first step is to identify the oldest innova-
tion that was not common to all Slavic dialects, in accordance with the defi-
nition of a proto-language given in § 1.5.2 (see also Holzer 1995b: 247: “Die
Einheitlichkeit des Urslavischen ist mit der ersten Innovation, die nur von
einem Teil des Slavischen durchgeführt wurde, zerfallen”). It is not as easy as
it sounds to pin down this innovation, however (cf. Holzer 1995a: 59: “welche
die älteste Isoglosse [ist], die es zwischen slavischen Sprachen gibt [. . .], wissen
wir nicht”). In the following paragraphs I shall examine the evidence for the
chronology of the changes that took place in the critical phase of the disinte-
gration of the Slavic proto-language.
A candidate for the first non-shared innovation is the second palatalisation
[23], which undoubtedly has partly different scopes and outcomes in the Slavic
dialects. If the second palatalisation is the first post-Proto-Slavic innovation, it
follows that Proto-Slavic is the language stage immediately after the monoph-
thongisation of diphthongs [22], which, in turn, followed the first palatalisa-
tion [21]. However, another candidate comes in as a possible and, I believe,
more appropriate marker of the beginning of the post-Proto-Slavic period: the
fronting of non-front vowels after palatal consonants [20].
The fronting of non-front vowels [20] differs from the other changes relevant
to the discussion of the dissolution of Proto-Slavic by apparently not being a
single event in the history of Slavic, but rather a phonotactic rule that was valid
for an extended period of time, during which other changes took place (Meillet
1900: 8–9; Vermeer 2008b: 519–520 with further references). Vowel fronting
was active when oral diphthongs were monophthongised [22], since the out-
comes of PS *ē and *ai̯ are distinct after *i̯ in Slavic. On the other hand, forms
28 chapter 1
like PS o-stem neut. nom.–acc. sg. *ˌsirdika > *ˌsirdica [23] > *ˌsirdice [20] (CS
*sь̑ rdьce [29]) show that vowel fronting also operated after the second pala-
talisation [23], which must have been posterior to the monophthongisation
of dipthongs. Since at least the end of the process of vowel fronting belongs to
the post-Proto-Slavic period (see also the external evidence adduced in Holzer
1998a: 61), it seems justified, at least from a practical perspective (see below),
to operate with a Proto-Slavic system where the process had not started yet.
As for the place of the first palatalisation [21] in the relative chronology rel-
evant for the identification of the first post-Proto-Slavic innovation, the cru-
cial point is that the distinction between the non-palatalised and palatalised
variants of */k g x/ was not phonemicised until *ā̆i ̯ merged with *ē [22], i.e. a
stage when vowel fronting [20] was already active. Consequently, if we accept
that vowel fronting was a post-Proto-Slavic process, the first palatalisation also
belongs to post-Proto-Slavic (cf. the discussion in Vermeer 2014: 186–187).
While the relative chronology of Slavic sound changes makes it clear that
the monophthongisation of oral diphthongs [22] cannot have preceded
vowel fronting [20] and the first palatalisation [21], the external evidence for
such a chronology is less conclusive. Certain Slavic loanwords in Baltic have
been mentioned as evidence in favour of the view that the monophthongisa-
tion of *i̯- and *u̯ -diphthongs is post-Proto-Slavic, e.g. Lv. krìevs ‘Russian’, cf.
the East Slavic tribe name Kríviči, with original *ei̯ (see Holzer 1998a: 39–41;
1998b: 61–62). I do not find this argument compelling, however, since Lv. ie may
reflect Slavic monophthongised *ẹ̄ as well as older *ei̯; besides, the borrowing
need not be from Proto-Slavic, but may be from a pre-stage of Proto-Slavic.
Similarly, the Russian river name Luga, a borrowing from Finnic (cf. Finnish
Laukaanjoki), does not necessarily show that the borrowing took place before
the monophthongisation of diphthongs (thus Holzer 2006: 135). A foreign
diphthong *au̯ may have been borrowed into Slavic as *ō or *ū (corresponding
to CS *u) long after the monophthongisation ceased to be active because a syl-
lable-final *u̯ was incompatible with Slavic phonotactics (thus also Moszyński
1972: 59, according to whom the monophthongisation of *au̯ , *eu̯ belongs to
pre-Proto-Slavic, whereas that of *ai̯, *ei̯ was a later process; Stang 1969a: 26–30
assumes that all diphthongs ending in a semivowel had been monophthon-
gised in Proto-Slavic).
It should be acknowledged that the Slavic evidence could also fit with a
phonetically more advanced proto-language where the fronting of non-front
vowels [20], the first palatalisation [21] and the monophthongisation of oral
diphthongs [22] had already been carried out, if it is accepted that the front-
ing of non-front vowels applied again in post-Proto-Slavic, after the second
introduction 29
palatalisation [23]. However, as it is simpler and, as far as I can see, it does not
create problems to regard the fronting of non-front vowels as a post-Proto-
Slavic change, I find it reasonable to use the term Proto-Slavic to refer to a
relatively conservative stage in the development of Slavic.
A further argument for identifying a conservative phonological stage as
Proto-Slavic is that positing a phonologically more advanced proto-language
carries a higher risk of being incorrect. For instance, a Proto-Slavic system
that included the monophthongisation of oral diphthongs would be histori-
cally incorrect if new or overlooked evidence turned out to prove that the
monophthongisation was in fact a post-Proto-Slavic change (such a hypoth-
esis has indeed been advanced; see [22]). But if, conversely, evidence that was
neglected here would show that the monophthongisation was actually a pre-
Proto-Slavic development, a more conservative reconstruction that did not
include the monophthongisation would still be an actual pre-stage of all Slavic
languages, even if not strictly Proto-Slavic. Similarly, if the view should turn
out to be correct according to which the (earlier) “third” and (later) “second”
palatalisations were separated by the fronting of non-front vowels and the
monophthongisation of oral diphthongs, a conservative Proto-Slavic phono-
logical system would need no or less adjustment than a more advanced recon-
struction (for references to literature supporting this chronology see [20]).
On a more practical level, regarding the fronting of non-front vowels as a
post-Proto-Slavic process has the advantage that the o- and i̯o-stems and the
ā‑ and i̯ā-stems have not yet split into distinct inflectional paradigms. At this
stage the inflectional system is considerably simpler than after the fronting of
non-front vowels.
Common Slavic refers to the Slavic dialect continuum during the period after
the dissolution of the Slavic proto-language (c. 600 AD) until the loss of the
weak jers (c. 1200 AD) (Trubeckoj 1922/1988: 218; Andersen 1985: 75; Lindstedt
1991: 112; Holzer 1995a: 57). Unlike Proto-Slavic, which is a specific language
stage, Common Slavic subsumes a number of developments that affected all,
or most, Slavic dialects after the dissolution of Proto-Slavic. The evidence of
relative chronology and of loanword relations between Slavic and non-Slavic
languages makes evident that the Common Slavic reconstructions are anach-
ronistic in a number of respects (see Lindstedt 1991; Andersen 1998a: 421–422;
Holzer 1995a: 79; Marchand 1973: 653).
For instance, reconstructed word-forms of the type *stornà, *bȇrgъ, usually
so presented in handbooks of Slavic as the ancestral form of OCS strana, brěgъ,
Ru. storoná, béreg, Po. strona, brzeg etc., do not belong to the Slavic proto-lan-
guage proper. The South Slavic forms are better explained if we assume that
30 chapter 1
*a had not yet become *o at the time of the metathesis of liquids: thus OCS
strana, brěgъ derive from *starˈnā, *ˌbergə through a stage *strāˈnā, *ˌbrēgə with
retention of the original two morae of the root syllable (Lindstedt 1991: 114–115;
Holzer 1995a: 79; 1995b: 250). Among the many loanwords from Slavic into non-
Slavic languages, Holzer (1995a: 79; 1995b: 250) knows of no instance pointing
to *CoRC, i.e. with o-vocalism before the metathesis. Now if South Slavic and
loanword relations point to original *ar, there is no reason to assume Proto-
Slavic ever had *or. This makes it likely that PBS *a was still *a in all positions
in Proto-Slavic, although attested Slavic dialects show o in most positions.
All this being said, as long as one is aware of the theoretical and practical
problems related to the reconstructions referred to as Common Slavic, the
traditional notation is a useful tool in the historical study of the Slavic lan-
guages and their prehistory for at least two reasons: (1) Common Slavic recon-
structions are graphically closer to the attested Slavic languages, making their
identification easier; (2) they constitute the point of reference in almost all
handbooks (see also § 2.2.3). Therefore, most Slavic forms discussed in this
study are presented both in their Proto-Slavic shape (e.g. *starˈnā, *ˌbergə) and
in Common Slavic notation (*stornà, *bȇrgъ).
In the majority of cases it is possible to transpose a Proto-Slavic inflectional
ending into Common Slavic notation (see also Lunt 1985: 191 n. 21). In a few
instances, however, it is not possible to provide a uniform Common Slavic end-
ing due to phonological developments (cf. Krajčovič 1974: 99–101; Le Feuvre
2007: 17). An example is the i̯ā-stem accusative plural ending, mentioned in
§ 1.5.2, where South Slavic has *‑ę while North Slavic points to *‑ě; neither form
can be derived from the other. Likewise, the Old Novgorod dialect has an end-
ing ‑e in the o-stem nominative singular, whereas Classical Common Slavic has
*‑ъ; again, there is no chance of deriving one of the endings from the other (at
least phonetically). Such cases of Common Slavic dialectal differentiation are
presented as CS S *‑ę || CS N *‑ě in the i̯ā-stem accusative plural (see § 3.14.4),
and as CS Cl *‑ъ || ONovg. ‑e (see § 3.3.14) in the masculine o-stem nominative
singular.
On the morphological level there are cases where a Proto-Slavic ending can-
not be inferred by comparing the Slavic languages only, but where evidence
from Balto-Slavic or other Indo-European languages is decisive. For instance,
in the o-stem instrumental singular South Slavic has *‑omь and North Slavic
*‑ъmь. It is often stated that the former ending is the original one and that North
Slavic has introduced the u-stem ending secondarily. It is more likely, however,
that Proto-Slavic had preserved the expected ending *‑ā, reconstructed on the
basis e.g. of Li. ‑u, from PBS *‑ō̰, and that South and North Slavic replaced the
introduction 31
inherited ending with different innovations. Since there are almost no traces
of the original ending, the Common Slavic form is presented as CS S *‑omь, CS N
*‑ъmь (§ 3.7.5).
In accordance with the definitions given in § 1.5.2, speakers of the Common
Slavic dialects were able to carry out similar or identical innovations; since
these innovations were carried out after the first change that set the Slavic
dialects apart, they are post-Proto-Slavic. An example of a Common Slavic
phonetic innovation is the change of *a to *o, part of a radical reshaping of
the vowel system in prehistoric Slavic [29]; another is the loss of jers in weak
position. On the morphological level, all attested Slavic dialects have replaced
the thematic vowel PS *‑a‑, from PIE *‑o‑, with ‑e‑ under the influence of the
i̯e-verbs, in which *‑i̯a‑ became *‑i̯e‑ by the fronting of non-front vowels [20],
a process that, in this study, is assigned to the Common Slavic period (see the
discussion above).
Loanword relations show that the Slavic language spoken at the time of the
Slavic expansions in the early centuries AD was comparatively uniform from
a phonological point of view (see Lunt 1985: 203 with n. 45; Kortlandt 1990/
2010a: 3; Holzer 1995a; 2002: 552–553). The evidence from loanword relations
indicates that the language spoken at the time of the maximum expansion of
Slavic-speaking tribes around 600 AD was very similar to the last stage of Slavic
before the first innovation that was not shared by all dialects, i.e. Proto-Slavic.
The problem of the original location of the speakers of Slavic before the
expansions is not relevant to the present study (for discussions see Mallory
1989: 76–81; Nichols 1993; Holzer 1995a: 59; 2002: 551–553; 2007: 15; Andersen
1998a: 415–417, with maps).
1.6.1 Overview
The present study consists of five chapters:
1.6.2 Endings
In chapters 3 and 4 of the study each ending is treated according to a fixed
pattern consisting of a table with the linguistic material, followed by a survey
of the literature and a discussion of the ending. The table contains a recon-
structed Proto-Slavic word-form containing the ending under discussion,
together with the corresponding Common Slavic word-form; the Proto-Balto-
Slavic and Proto-Indo-European reconstructions of the ending; and attested
Slavic, Baltic and non-Balto-Slavic Indo-European word-forms containing the
ending. In the Proto-Slavic and Proto-Balto-Slavic reconstructions the accen-
tual properties of the ending are noted: endings that are accented in mobile
paradigms are marked with “ˈ”, e.g. PS ā-stem nom. sg. *‑ˈā, whereas unaccented
34 chapter 1
endings are marked with “ˌ”, e.g. PS ā-stem acc. sg. *ˌ‑ām. Whenever possible, I
have chosen Proto-Slavic example words with a mobile accent paradigm.
As already mentioned, in certain cases two or more endings are sufficiently
transparent morphologically that they may be treated together, e.g. the pres-
ent third-person singular endings athematic PS *‑ti (CS *‑tь) and thematic *‑eti
(*‑etь).
The material presented in the tables is not exhaustive, but is meant to rep-
resent the forms that are most relevant to our purposes. This means that more
conservative forms are given precedence over innovations, although the most
frequently used form of a paradigm is generally also presented, even if an inno-
vation. If there are no formal traces of an ending in a language, either because
the category has been lost or the paradigm no longer exists, I write a dash (“–”),
e.g. in the Lithuanian ū-stem nominative singular (§ 3.3.1). Since the primary
and secondary endings (see § 4.1) have merged in the Baltic languages and the
preterite has been remade, Baltic verbal endings are only given in the present
and imperative systems. The symbol “→” before an ending or a word-form indi-
cates a morphological reshaping or replacement of the original ending, not
taking into account prosodic factors, e.g. the analogical introduction of final
accentuation in Li. athem. prs. 2sg. esì (§ 4.6). Material from languages other
than the ones presented systematically, e.g. Čakavian, Oscan or Old Norse, is
given in cases where they provide additional information of relevance for the
reconstructions.
The linguistic data are taken from standard handbooks, including Diels
1932–1934/1963 for Old Church Slavonic, Gebauer 1896 and 1898 and, occa-
sionally, Trávníček 1935 for Old Czech. Data from the Old Novgorod dialect are
quoted from Zaliznjak 1995/2004; late attestations from this dialect are marked
as such. Old Lithuanian forms are from Mikalojus Daukša’s writings unless oth-
erwise noted. Hittite forms are generally quoted in broad transcription from
Hoffner & Melchert 2008 (see there p. 11). Forms attested in the Old Hittite
period are marked with “OS”, for “Old Script”; unmarked forms are only attested
in later stages of the language.
Each table is followed by an overview of the assessment of the ending in
selected publications, mostly handbooks, dealing with the relevant language
groups. After the overview of the literature I present my analysis of the devel-
opment of the ending from Proto-Indo-European via Proto-Balto-Slavic to
Proto-Slavic and further into the attested Slavic languages. In the analysis of
the ending I refer systematically to the phonetic changes leading from one lan-
guage stage to the next. This yields the following scheme:
introduction 35
– Vaillant 1966: Slavic verbs; includes the development from Common Slavic
to the individual Slavic languages
– Bräuer 1969a, 1969b: Slavic nouns; includes the development from Common
Slavic to the individual Slavic languages
– Arumaa 1985: Slavic nouns, pronouns, verbs
– Igartua 2005a: Slavic nouns
– Aitzetmüller 1978/1991: Old Church Slavonic nouns, pronouns, verbs
In Stang 1942 the Baltic and Slavic verbal systems are treated separately;
accordingly, I have regarded that monograph as two independent works, on
Baltic and Slavic, respectively.
I silently normalise attestations and reconstructions from other publica-
tions (except in direct quotations). The most important changes are: Proto-
Indo-European H₁, H₂, H₃, ə̯₁ ə̯₂ ə̯₃ and ʔ ʕ ʕʷ > h₁ h₂ h₃; H and hₓ > h; y w > i̯ u̯ ; ḱ ǵ
ǵʰ and k̂ ĝ ĝʰ > k̑ g̑ g̑ ʰ; ku > kʷ; bh dh gh g̑ h > bʰ dʰ gʰ g̑ ʰ; Slavic ĭ ŭ > ь ъ; Old Norse
unaccented e o > i u. In quotations from Olander 2009, V̄ ˀ has been changed
to V̰̄ . Obvious spelling mistakes in forms quoted are corrected. Breves above
short vowels are removed. Asterisks are added to reconstructed forms. I do not
(re)syllabify quoted reconstructions, nor is laryngeal colouring applied in
reconstructions that are not mine. I leave out hyphens denoting morpheme
boundaries unless they are important for the discussion. When appropriate,
I have adapted the terminology found in other publications to that used in this
book.
Chapter 2
Phonological Background
as the issue has not been settled, and also for the sake of clarity, it seems most
reasonable to mark the syllabicity of sonorants in Proto-Indo-European recon-
structions (see also Kümmel 1997: 118; 2007: 16–20; cf. Mayrhofer 1986: 158–160
[liquids and nasals], 160–163 [semivowels]; cf. also the practice of the Leiden
School with no phonologically relevant distinction between syllabic and non-
syllabic sonorants, explicitly stated in e.g. Beekes 1995/2011: 120 and Kortlandt
2010c/2010a: 62, with criticism of the standard system). Accordingly, I follow
the tradition of writing PIE *i u vs. *i̯ u̯ , and *r l m n vs. *r̥ l ̥ m̥ n̥ .
The syllabification of the laryngeals seems to have taken place according
to different rules in the individual daughter languages; for instance, a word-
initial laryngeal followed by a consonant was vocalised in Greek but not in
Indo-Iranian (cf., however, Rasmussen 1994: 434: “the syllabification rules
themselves must have operated in the protolanguage already”).
*smordós with lengthening by Winter’s law [6]; and Li. málti, Lv. mal̃t reflect
PBS *ˈma̰ ltēi̯ from *mo̰ ltēi̯ < PIE *molhtēi̯ with an acute syllable caused by the
laryngeal [1]. That we have a-timbre in these cases is not simply because the
delabialisation of *o to *a [7] took place before Winter’s law [6]; this is shown
by the PBS *ō-vocalism of examples like Li. núogas (a.p. 3), Lv. nuôgs from PBS
*ˌnō̰gas < PIE *nogʷós [4|6|7].
Proto-Balto-Slavic had a free accent, the accented syllable probably being
marked by high pitch. Certain lexical word-forms only contained syllables with
low pitch; they had become phonologically unaccented through an accent loss
in certain final syllables [4] (Olander 2009: 144–145).
from vowel quantity. The retention of length probably depended on the num-
ber of syllables in the word, with various subsequent levellings in the Slavic
dialects (see Kapović forthc.; I am grateful to Mate Kapović for sharing his
thoughts on this problem with me). Likewise, the distinction between the long
final vowel of OPo. i̯ā-stem nom. sg. wolå and the short final vowel of ā-stem
woda is of post-Proto-Slavic date (see Fecht 2010 and Andersen 2014: 79–99
for two recent treatments of this problem). An analysis of the origin of these
quantitative distinctions is outside the scope of this study.
As mentioned in § 1.5.5, the Proto-Slavic phonological system of this study
differs significantly from the traditional system. The present system is based
on the chronology suggested by internal evidence from Slavic and by external
relations between Slavic and non-Slavic languages. The external evidence for
the Proto-Slavic phonological system assumed here comes from the languages
that Slavic was in contact with at the end of the Slavic expansions, culminat-
ing around 600 AD: Finnic, Baltic, Germanic, Romance, Greek, Albanian and
Hungarian. The relative chronology is complicated and there is substantial dis-
agreement on the evaluation of the material.
It should be noted that some scholars have been sceptical about the evi-
dence provided by external relations. Trubeckoj (1922/1988: 219–221) maintains
that Proto-Slavic did not have *a but rather *o in a word like *bobъ, despite the
fact that it is rendered papu in Finnish (similarly Vasmer 1907: 157–164, on the
Greek evidence for the early Slavic vowel system). This point of view is diffi-
cult to reconcile with the facts (see Vermeer 2008a: 538–539; Mareš 1969/1999:
26–27).
Language-external relations only rarely provide information on endings
(Holzer 1980: 9). An interesting positive example is the phrase Tagazino ‘his
son’ attested in an Austrian charter from 827, written in Latin. The phrase
approximately represents *[taga sɨːnu] from PS *taga sūnu (CS *togo synъ)
(Holzer 1998b: 64; 2004: 54, 59). This shows that the delabialisation of PS *ū to
*[ɨ̄] was earlier than the development of PS *a to *o. In general, however, it is
clear that what loanwords relations allow us to establish for roots and suffixes
is also likely to be relevant for endings. If, for instance, it can be shown that
the root-vowel of OCS neut. nom.–acc. sg. kolo reflects PS *a (not *o), we may
assume that the vowel of the ending also reflects PS *a.
The notational system referred to as Common Slavic in this study (see
§ 1.5.5) is, in most practical respects, identical to Old Church Slavonic, except
for the palatalisations of dentals [26], the metathesis of liquids and a few other
modifications (Lunt 1985: 190–191; Lindstedt 1991: 113; Holzer 2002: 554). To take
an example from a standard handbook, the chapter on “Proto-Slavonic” in
The Slavonic languages (Schenker 1993/2002) refers to a “Late Proto-Slavonic”
stage, defined as “the period encompassing the beginning of dialect differ-
Phonological Background 45
entiation within Slavonic” (p. 61), i.e. corresponding roughly to the Common
Slavic period of the present study. The “Late Proto-Slavonic” phonological sys-
tem (p. 82) comprises eleven short vowels, *o a e ě ь i u ъ y ę ǫ, and nine long
vowels, *ō ā ē ě̄ ī ū ȳ ę̄ ǭ. The consonant system includes eight stops, *p b t d t´
d´ k g, eight fricatives, *v s z š ž s´ z´ x, three affricates, *c ʒ č, three nasals, *m
n n´, and four liquids, *r l r´ l´. While the author acknowledges that this lan-
guage stage is not ante-dialectal (in the sense of our “Proto-Slavic”), the stage is
used as the point of reference in the following sections on (inflectional) mor-
phology, syntax and lexicon (although, for some reason, without indication of
vowel quantity). Other studies have systems that differ in the details, but the
basic appearance is the same; for examples see the literature mentioned under
“Proto-Slavic and the Slavic languages” in § 1.6.3.
The reason why most treatments of the Slavic inflectional system, its prehis-
tory and its further development have referred to a phonological system close
to that of Old Church Slavonic is obvious: a large part of the linguistic devel-
opments that took place after the break-up of Proto-Slavic, including those
that gave the Slavic dialects their new, Common Slavic look, took place in the
entire Slavic-speaking territory with only few or no dialectal differences. At
first glance, this system does seem to be Proto-Slavic.
The traditional system works fine for a description of the derivational mor-
phology, syntax and lexicon of Slavic, where phonological details often do not
play a decisive role. But when it comes to a precise understanding of the Proto-
Slavic inflectional system, where phonological accuracy matters more than
elsewhere, it is a problem that the traditional system is far from the one recon-
structed on the basis of loanword relations between Slavic and neighbouring
non-Slavic languages. In the traditional framework, analyses of the inflectional
system may become imprecise, for instance by blurring an actual distinction
between two endings. An imprecise description of the synchronic system, in
turn, has direct bearings on the diachronic interpretation of the system, both
backwards (towards Proto-Balto-Slavic and Proto-Indo-European) and for-
wards (towards the attested Slavic languages). A well-founded interpretation
of the diachronic relationship between two (or more) stages of development of
a language presupposes precise descriptions of the synchronic stages involved.
Impreciseness on a limited scale would have little significance, and there are
no reconstructed systems for which absolutely precision can be claimed. But in
the case of Proto-Slavic we know that so many things happened in the few cen-
turies between the break-up of the Slavic proto-language and the first direct
written attestations that analyses may yield quite different results depending
on which system they are based on.
46 chapter 2
the developments presented in this section, however, rest on safer ground than
this example.
An interesting general observation that can be made on the prehistoric pho-
netic development of Slavic is that when word-final consonants are lost, they
tend to leave a trace on the preceding vowel. Final stops are apparently lost
without a trace [9] in pre-Proto-Balto-Slavic, but at a later stage the loss of
final fricatives leads to centralisation of a preceding vowel [17], the loss of final
dentals leads to raising of a preceding long vowel [18], and the loss of final *‑m
causes rounding and raising of a preceding *a to *u [19].
The overview presented in the following subsections is not the first attempt
at establishing the relative chronology of sound changes from Proto-Indo-
European to Slavic. Most studies are specialised and only treat a subset of the
relevant phonetic changes; a typical example is Olander 2010, treating only the
development of endings containing a word-final nasal. Some studies, however,
present a more or less systematic treatment of the relative chronology of the
major part of the sound laws relevant for the development of the inflectional
system from Proto-Indo-European to Slavic (e.g. Trubeckoj 1922/1988; Jakobson
1929/1971; Milewski 1932; van Wijk 1950; Ebeling 1963; Shevelov 1964, with an
overview of sound changes pp. 633–634; Mareš 1969/1999; Holzer 1980 and,
quite differently, 1998b and 2001; Kortlandt 2011c [originally published in 1989];
Matasović 2005; Carrasquer Vidal 2011).
In the following overview all changes are marked as either “general” (apply-
ing in all positions of the word) or “final” (applying only in final syllables). A
comma between two or more items in curly brackets indicates parallel behav-
iour of the items; for instance “_{C, #}” means “before a consonant or word
boundary”.
h>∅/ { } _V
i̯_C₀#
Vh > V̰̄ / _{C, #}
VRh > V̰ R / _{C, #}
with original long vowels. Sequences of vowel, sonorant and laryngeal are
also marked prosodically, contrasting with sequences of vowel and sonorant
with no following laryngeal; for the resulting vowel quantity in original *VRH
sequences see § 2.2.2. The most likely phonetic candidate for the prosodic
marking is glottalisation (see § 2.2.2 with references).
A special case is the loss of laryngeals after *i̯ in final syllables (cf. the dif-
ferent analysis in Hoffmann 1976a: 615 n. 12, followed by Mayrhofer 1986: 131).
This rule accounts for the short vowel of the second- and third-person singular
optative in Proto-Balto-Slavic, viz. PIE 2sg. *‑oi̯h₁s, 3sg. *‑oi̯h₁d > PBS *‑ˈai̯s, *‑ˈai̯
[1|5|7|9]. The second- and third- person forms yielded PS ipv. *‑ˈəi̯ [12|17] (CS *‑i
[22|29]) and Li. permissive 3sg. ‑iẽ, respectively.
When accented, the new long vowels are realised with high pitch on the first
mora. The same applies to original long vowels, as in PIE *dʰugh₂tḗr. The pitch
contour becomes phonologically distinctive when the vowel contractions [3]
take place, yielding long vowels with high pitch on the second mora.
The loss of laryngeals is probably earlier than the vowel contractions [3],
which presuppose hiatuses created by the loss of intervocalic laryngeals. It is
also earlier than the delabialisation of *o to *a [7] since PIE *‑oh yields *‑ō̰.
Apart from the loss of laryngeals after *i̯, which is limited to final syllables, the
development applies in all positions of the word.
V[–high]V[–high] > V̄
Two non-high vowels in hiatus contract to a plain long vowel. If one of the
vowels is accented, the new long vowel receives high pitch on the second mora,
i.e. a rising tone. This rising tone contrasts with the falling tone of original long
vowels and long vowels from laryngeal contraction [1]. Note that these tones
are unrelated to the opposition between acute and circumflex syllables in
Baltic and Slavic, which are conditioned by the original presence or absence
of a laryngeal. The prosodic situation at this stage is comparable to that of
Attic Greek, where the distinction between rising and falling tones is largely
restricted to the last syllable of the word. When the mobility law [4] eliminates
rising tone from final syllables, the tonal distinction disappears.
These vowel contractions take place after the loss of laryngeals [1] and
before the mobility law [4]. Also, the fact that PIE o-stem dat. sg. *‑oei̯ con-
tracted to PBS *‑ōi̯, not *‑āi̯, as indicated by PS *ˌ‑āu̯ [14|15] (CS *‑u [22|29]) and
Li. ‑ui, shows that the vowel contractions took place before the delabialisation
of *o to *a [7]; PBS *‑āi̯ yielded *‑āi̯ in Proto-Slavic and ‑ai in Lithuanian, e.g. PS
ā-stem dat. sg. *ˌ‑āi (CS *‑ě [22|29]), Li. gálvai.
It is possible that in sequences of a high vowel followed by a non-high vowel
(after the loss of intervocalic laryngeals [1]) a glide was inserted, e.g. PIE i-stem
loc. du. *‑ihou̯ > *‑iou̯ [1] > *‑ii̯ou̯ [3] > PBS gen.–loc. du. *‑ii̯au̯ [7] > PS *‑ii̯au̯
(CS *‑ьju [22|29]).
μ́ > μ / _C₀#
Olander 2006 (see also 2007b; for a critical assessment and elaboration of
this first version of the sound law see Andersen 2009b; see also Holzer 2009:
154; and the rejection in Kortlandt 2009/2009). In my monograph Balto-Slavic
accentual mobility from 2009 I proposed a revised version of the sound law,
identical to the one presented here (for the reception of the revised version
see the reviews: Fecht 2009; Kim 2010; Oslon 2010; Rinkevičius 2010; Petit 2011,
2013; Bichlmeier 2013; D. Birnbaum 2013; and the extensive review-like article
Kortlandt 2010c/2010a).
The mobility law took place after the vowel contractions [3] (and, conse-
quently, after the loss of laryngeals [1]). Apart from that, the loss of accent is
difficult to place in the relative chronology, except that it is very likely to belong
to the pre-Proto-Balto-Slavic period (see Olander 2009: 163; but cf. Andersen
2009b: 19).
D > T /_#
Word-final obstruents are devoiced. This change takes place before Winter’s
law [6]; otherwise we would have a long vowel in Li. 1sg. pron. nom. àš (OLi.,
dial. eš), Lv. es from PIE *h₁ég̑ , and in PIE pron. o-stem neut. nom.–acc. sg.
*‑od > PBS *ˌ‑a [4|5|7|9] > PS *ˌa (CS *‑o [29]) (see Matasović 2005: 151; and cf.
Kapović 2009: 64–65; in Kortlandt 2013: 6, however, the Baltic forms are traced
back to *eś with “secondary shortening”). If all word-final stops were voiced in
Proto-Indo-European by the rule mentioned in § 2.2.1, forms like PS aor. 3sg.
*ˌu̯ ede (not **ˌu̯ edē) from PIE *‑ed would confirm this chronology.
1 V > V̰̄ / _D
VR > V̰ R / _D
2 Dʰ > D
Vowels are lengthened and glottalised before a voiced unaspirated stop, also
with an intervening sonorant. Subsequently voiced aspirated stops lose their
aspiration and merge with plain voiced stops. There probably were some fur-
ther conditions restricting the domain of Winter’s law, but despite several
attempts no entirely satisfactory formulation has been found yet (see Olander
2009: 150–151 with references).
Phonological Background 51
7 Delabialisation of *o to *a (general)
o>a
k ̑ g̑ > ś ź
The reflexes of the Proto-Indo-European palatal stops *k ̑ g̑ ⁽ʰ⁾ are assibilated to
*ś ź, perhaps through an intermediate stage as affricates (*ć ʒ́). In pre-Proto-
Slavic, after the ruki change [12] has stopped working, *ś becomes *s, merging
with original *s; in Lithuanian, *ś is reflected as š, merging with the ruki variant
of *s. The voiced counterpart *ź becomes z in Slavic and ž in Lithuanian. The
assibilation may have taken place after Winter’s law [6] (where reservations
concerning the relative chronology are presented) and before the loss of word-
final stops [9].
52 chapter 2
As pointed out to me by Guus Kroonen (pers. comm., 2013), it has been sug-
gested that words like Po. dzwon, Ukr. dzvín, Mac. dzvonec indicate that Proto-
Indo-European palatal stops (or perhaps rather affricates) were preserved into
“Early Slavic” (Kortlandt 2008/2009: 43). This question does not seem to be rel-
evant for the present study.
T > ∅ /_#
Tautosyllabic *åēu̯ becomes PBS *i̯åāu̯ (see Pedersen 1935; Stang 1966: 74; Kortlandt
1979a/2009: 33; Holzer 2001: 37–38). For instance, the Proto-Indo-European
root *leu̯ dʰ‑ > *leu̯ d‑ [6] becomes PBS, PS *li̯au̯ d‑ [10] (CS *l’ud‑ [22|26|29]), Li.
Phonological Background 53
liáudis (a.p. 1); and PIE *kéu̯ d‑ > *ˈkēu̯ d‑ [6] becomes PBS, PS *ˈki̯āu̯ d‑ [10] (CS
*čűd‑ [21|22|26|29]).
In endings the effects of this change were reverted by analogy with endings
which had regularly preserved *åēu̯ (i.e. before a vowel) or its later reflex *åāu̯
[11]. This applies to PBS u-stem gen. sg. *‑au̯ s, loc. sg. *‑āu̯ and voc. sg. *‑au̯
(not *‑i̯au̯ s, *‑i̯āu̯ and *‑i̯au̯ as expected). For the latter two forms we could
assume that the change did not take place in absolute final position; however,
as the forms can easily be analogical, I prefer the more general formulation
of the change according to which any tautosyllabic PIE *åēu̯ is reflected as
PBS *i̯åāu̯ .
The change is earlier than the more general change of *eu̯ to *au̯ [11].
e > a / _u̯
Short *e is backed to *a before *u̯ , e.g. PIE u-stem nom. pl. *‑eu̯ es > PBS *‑au̯ es
[11] > PS *‑au̯ e [17] (CS *‑ove [29]) (Mikkola 1913: 44; Hamp 1976; Kortlandt
1979a/2009: 33; Kapović 2006b: 116–118 with a useful treatment of the problem).
According to a widespread view, the change did not take place before front
vowels (e.g., for Baltic, Endzelīns 1911/1974: 82–84; 1971: 32; Stang 1966: 32–33,
73–74, following Endzelīns; for Balto-Slavic, Vaillant 1950: 110; Bräuer 1961: 73;
Holzer 2001: 38). In that case PIE u-stem dat. sg. *‑eu̯ ei̯ and nom. pl. *‑eu̯ es >
*‑eu̯ e cannot regularly have yielded PS *‑au̯ ei̯ (CS *‑ovi [22|29]) and *‑au̯ e (CS
*‑ove [29]). I find the solution more likely to be that *eu̯ becomes *au̯ before
any vowel, as well as before consonants and in word-final position.
The position of this change relative to other sound changes is difficult to
determine, except that it postdates the change of tautosyllabic *åēu̯ to *i̯åāu̯
[10]; the change may have taken place in pre-Proto-Balto-Slavic (thus Kapović
2006b: 116), or it may be a later, parallel change in Slavic and Baltic.
In Proto-Balto-Slavic the phoneme */s/ had two main allophones, *[ʂ] after *ī � ̆ i̯
åū u̯ r k, and *[s] elsewhere. In pre-Proto-Slavic the ruki variant is phonemicised
54 chapter 2
i̯ > u̯ / ō_C₀#
Proto-Balto-Slavic *ōi̯ becomes *ōu̯ in final syllables (see also Pedersen 1905a:
323–325; Meillet 1915: 5; 1924/1934: 153–154; 1963: 31; Kortlandt 1979/2011a: 103;
1983/2011a: 175). I am unaware of any examples or counterexamples in non-
final syllables. The clearest example of this development is PIE o-stem dat. sg.
*‑oei̯ > PBS *ˌ‑ōi̯ [3|4] > *ˌ‑ōu̯ [14] > PS *ˌ‑āu̯ [15] (CS *‑u [22|29]). The develop-
ment also takes place in PIE o-stem instr. pl. *‑ōi̯s > PBS *‑ˈōi̯s > *‑ˈōi̯x [12] >
*‑ˈōu̯ x [14] > *‑ˈāu̯ x [15] > PS *‑ˈū [17] (CS *‑y [22|29]).
This development is later than the contraction of vowels [3] and, since it
does not affect Lithuanian and, apparently, Old Prussian (Stang 1966: 71–72,
181–182), it belongs to the post-Proto-Balto-Slavic period. It obviously predates
the delabialisation of *ō to *ā [15].
15 Delabialisation of *ō to *ā (general)
ō>ā
The long vowel *ō becomes *ā, e.g. PBS o-stem masc. nom.–acc. du. *ˌ‑ō̰ > PS
*ˌ‑ā [13|15] (CS *‑a [29]), merging with PBS ā-stem nom. sg. *‑ˈā̰ > PS *‑ˈā [13]
(CS *‑a [29]).
The merger of *ō and *ā is post-Proto-Balto-Slavic as the two vowels remain
distinct in Baltic; cf. the corresponding forms Li. lángu, def. adj. mažúoju, and
galvà, def. adj. mažóji. The delabialisation of *ō is later than the change of *ōi̯
to *ōu̯ [14].
In word-final sequences of a high vowel (*i or *u) plus *ns (in the relevant cases
reflecting PIE *‑ms; see § 3.14.1), the *n is lost with compensatory lengthening
of the vowel (thus also Bräuer 1961: 106; similarly Kortlandt 1979/2011a: 103). The
only two examples of this development are PIE i-stem acc. pl. *‑ims > *‑ins >
*ˌ‑īs [4|16] > PS *ˌ‑ī [17] (CS *‑i [29]) and PIE u-stem acc. pl. *‑ums > *‑uns > *ˌ‑ūs
[4|16] > PS *ˌ‑ū [17] (CS *‑y [29]).
Since word-final *‑s is a conditioning factor, the development must have
preceded the loss of word-final fricatives [17].
56 chapter 2
1 ā̆ > ə̄ ̆ / _(R)S#
2 S > ∅ / _#
sponds to PS *ˌ‑au̯ (CS *‑u [22|29]). Expected pre-PS *‑əu̯ was probably replaced
with PS *ˌ‑au̯ (CS *‑u [22|29]) by analogy with the case endings that contained
*au̯ . Note that the problem does not arise if we assume that the backing of
*eu̯ to *au̯ [11] was later than the centralisation of *a to *ə, which, however,
would separate the former change from the similar development in Baltic. The
analogical replacement assumed here is similar to the one mentioned under
the diphthongisation of tautosyllabic *åēu̯ to *i̯åāu̯ in pre-Proto-Balto-Slavic [10],
which affected the same ending.
At an early stage pre-PS *-āu̯ s > *ə̄u̯ was apparently assimilated to PS *ū,
as in PBS instr. pl. *‑ˈōi̯s > *‑ˈāu̯ s [14|15] > *‑ˈə̄u̯ > PS *‑ˈū [17] (CS *‑y [29]). This
minor sound law is required to generate the correct output in the i̯o-stem
instrumental plural ending PIE *‑ˈi̯ōi̯s > *‑ˈi̯āu̯ s [14|15] > *‑ˈi̯ə̄u̯ > PS *‑ˈi̯ū [17] (CS
*‑ji [20|29]). In my article on the development of *‑os and *‑ās in Slavic I had
overlooked this problem (2012: 333, where “*‑i̯ə̄i”̯ should be corrected to “*‑i̯əi̯”).
18 Loss of word-final dentals after long vowels, with raising of the vowel
(final)
The sequences *‑ēr and *‑ōn become *‑ī and *‑ū word-finally. The sound law
has an ad hoc flavour to it, since it only affects two endings, viz. PIE r-stem
nom. sg. *‑ēr > PS *‑ˈī [18] (CS *‑i [29]) and PIE n-stem nom. sg. *‑ō → pre-PS
*‑ōn > PS *‑ū [18] (CS *‑y [29]). The change does not affect word-final *‑V̄ m, as
seen in PBS ā-stem acc. sg. *‑ām > PS *‑ān [19] (CS *‑ǫ [28]; *‑jǫ [20|27|28]); and
by PBS 1sg. pron. acc. *mēm, 2sg. *tēm, refl. *sēm > PS *mēn, *tēn, *sēn [19] (CS
*mę, *tę, *sę [28]).
As shown by the corresponding Lithuanian endings ‑ė̃ and ‑uõ, the raising
does not apply in Baltic and is therefore post-Proto-Balto-Slavic. If the formu-
lation given here is correct, this change is later than the loss of word-final *‑s
[17], or at least later than the change of *‑s to *‑x or *‑h, if that was an interme-
diate stage in the process (cf. Olander 2012: 335–336). The change precedes the
merger of word-final *‑m and *‑n [19].
According to Kortlandt (1983/2011a: 126–127 and passim), any word-final
sonorant caused raising. While this view accounts nicely for OCS mati and
kamy, it meets difficulties in the accusative singular forms of the personal and
reflexive pronoun mentioned above. These forms may be saved by the further
condition that raising only took place in non-initial syllables (Kortlandt, pers.
comm., 2014).
58 chapter 2
1 a > u / _m#
2 m > ∅ / V̆ _#
u-stems (*‑is : *‑im and *‑us : *‑um). Note in this context that the Latvian pro-
nominal form tùo cannot reflect PBS *‑um but must go back to *‑am, *‑ām or
*‑ōm (I am grateful to Aigars Kalniņš for drawing my attention to the relevance
of this form). If I understand it correctly, Kortlandt’s view implies that the ana-
logical replacement of *u with *a in the accusative singular has taken place
independently in East and West Baltic, as Kortlandt does not operate with a
Baltic proto-language after Proto-Balto-Slavic (see Kortlandt 1977/2009: 5).
An argument in favour of a pre-PBS change *‑om > *‑um is, in Kortlandt’s
view, the genitive plural ending, which he reconstructs as PIE *‑om and consid-
ers to be the ancestor not only of Slavic ‑ъ but also of the Baltic forms, viz. Li.
‑ų, Lv. ‑u and OPr. pron. ‑on. Kortlandt does not seem to mention the Latvian
pronominal genitive plural form tùo (see Endzelīns 1923: 295; Stang 1966: 184),
which, like the homophonous accusative singular form mentioned above,
points to PBS *‑ōm or *‑am (or *‑ām), not *‑um (thus also Jasanoff 1983a: 142 n.
9; cf. the discussion in Hill 2013: 166–167). Since it is likely that an ending *‑ōm
(from PIE o-stem *‑o‑om, ā-stem *‑ah₂‑om) existed in pre-Proto-Balto-Slavic
and that it would have regularly yielded Li. ‑ų, Lv. ‑u, OPr. ‑on anyway (Stang
1966: 184), the Baltic genitive plural endings hardly constitute a convincing
argument for Kortlandt’s hypothesis.
The most promising material for Kortlandt’s idea of an early narrowing of
*‑om to *‑uN is constituted by Old Prussian accusative singular forms. Here we
find ‑on in certain cases where most scholars expect ‑an from PIE *‑om, e.g. in
the passive participle accusative singular. According to Kortlandt, OPr. ‑on is
the regular outcome of PIE *‑om in these cases. However, I find it more attrac-
tive to assume that ‑on is the result of an analogical introduction of u-stem
endings in original o-stem paradigms. Consequently, the evidence from Baltic
pointing to PBS *‑am, not *‑um, as the outcome of PIE *‑om should be taken at
face value (see also Olander 2010: 90–91).
between the beginning of vowel fronting [20] and the first palatalisation [21],
thereby indicating that the former sound change precedes the latter. It is also
thinkable, though, that the whole stem was replaced, i.e. that *maže- replaced
*maga‑; in that case the analogical substitution may have taken place after the
first palatalisation [21].
The fronting of non-front vowels was an active process before the monoph-
thongisation of oral diphthongs [22], as shown by the fact that PS *ē and *ā̆i ̯
are distinguished after palatal consonants, e.g. PS inf. *staˈi̯ētēi̯ (CS *stoja̋ ti
[22|27|29]) vs. PS i̯o-stem loc. pl. *‑i̯ai̯xu > *‑i̯ei̯xu > *‑i̯īxu [22] (CS *‑jixъ [26|29])
(see e.g. Holzer 1998b: 61; 2001: 39). On the other hand, non-front vowels were
also fronted after consonants fronted by the second palatalisation [23]. It thus
seems that for an extended period of time there existed a phonotactic rule
according to which palatal consonants could only be followed by front vowels.
The process was no longer active when *ē was backed to *ā after palatal con-
sonants [27].
The results of the fronting of PS *‑ə and *‑ə̄ are less secure than in the case
of the other vowels (see Olander 2012: 333–334). On the basis of the available
evidence it seems that the different outcomes of non-fronted *‑ə ‑ə̄ in Classic
Common Slavic (*‑ъ ‑y) and the Old Novgorod dialect (‑e ‑ě) are not paralleled
by different outcomes of fronted *‑ə ‑ə̄, which apparently merge with *‑i ‑ē,
yielding *‑ь ‑ě in the entire Slavic dialect continuum. In the case of fronted *‑ə,
however, the material does not allow us to establish the only relevant ending,
the masculine i̯o-stem nominative singular, with certainty in the Old Novgorod
dialect (see Olander 2012: 328, with references in n. 62). Due to the replace-
ment of the endings of the i̯ā-stem genitive singular and nominative plural in
South Slavic with that of the accusative plural ending PS *‑i̯ə̄n̆ (CS S *‑ję || CS N
*‑jě; see § 3.5.4), the regular result of fronted *‑ə̄ is unknown here, but it is rea-
sonable to assume that it was *‑ě as in the remaining dialects.
Since the attested reflexes of PS *‑i̯ə̄n̆ (CS S *‑ję || CS N *‑jě [20|28|29], e.g. OCS
‑ję, ORu. ‑jě) differ from those of PS *‑i̯an, *‑i̯en (CS *‑ję [20|28], e.g. OCS ‑ję,
ORu. ‑ja) and PS *‑i̯ān (CS *‑jǫ [20|27|28], e.g. OCS ‑jǫ, ORu. ‑ju), I assume that
palatalised *ə ə̄ did not directly merge with *i ē but passed through an inter-
mediate stage *ɪ ɪ̄. It is possible that the unfronted and fronted reflexes of PS *ə̄ ̆
at this point were realised phonetically as *[ʊ̄̆ ] and *[ɪ̄]̆ , respectively (see [17]),
but for the sake of clarity I write the former variant as *ə̄.̆ The merger of *ɪ ɪ̄ and
*i ē [29] took place after the monophthongisation of nasal diphthongs [28].
This vowel fronting is later than the change of *a to *u before word-final *‑m
[19], as shown by e.g. i̯o-stem masc. acc. sg. PIE *‑i̯om > *‑i̯am [7] > PS *‑i̯u [19]
> *‑i̯ü [20] (CS *‑jь [29]).
62 chapter 2
The velars *k g x are palatalised to *č ž š before front vowels and *i̯. It is often
assumed that *g is palatalised to *ǯ (parallel to *k > *č) before being lenited to
*ž, but it is also possible that palatalised *ǵ was first lenited to *γ́ , which is then
palatalised to *ž (Andersen 1969: 553).
The first palatalisation is later than the ruki change [12] since only ruki *x,
not non-ruki *s, is palatalised before front vowels. Phonetically, the process of
palatalisation may have started already in pre-Proto-Slavic, but it is only when
original *ē merges with *ē from *ā̆i ̯ [22] (Bidwell 1961: 107 n. 4), when *i̯ is lost
after consonants [26] (Schenker 1993/2002: 69) or when *ē and *ā merge after
palatal consonants [27] (van Wijk 1950: 304; Stieber 1969–1973/1989: 66–67;
Kortlandt 2011c/1989: 164) that the distinctions that had arisen as a result of
the first palatalisation become phonologically relevant. In this study these pro-
cesses, including the first palatalisation, are relegated to the post-Proto-Slavic
period (see §§ 1.5.5 and 2.2.3). Thus in Proto-Slavic, *[č ž š] were allophones of
*/k g x/ (thus also Holzer 2003: 32–33 with n. 39; 2008: 201–202; for the sugges-
tion in Holzer 1998b: 65 n. 20 that a split into */k/ and */č/ was introduced in
pre-Proto-Slavic with loanwords, see Holzer 2006). The first palatalisation is
probably later than the fronting of vowels after palatal consonants ([20], with
discussion). It is earlier than the metathesis of liquids, since PS *kerˈdā yields
*čerˈdā [21] > *črēdā > OCS črěda (Holzer 2001: 39).
{åāi̯ > ē; åāu̯ , åēu̯ > ō; åēi̯ > ẹ̄; əi̯, ɪi̯ > ī } / _.
The oral diphthongs *åāi̯ åāu̯ åēu̯ åēi̯ əi̯ ɪi̯ are monophthongised to *ē ō ō ẹ̄ ī ī in a
process labeled “the First CS Vowel Shift” by Andersen (1985: 73; 1998b: 239;
Phonological Background 63
1998a: 429). The diphthong *ə̄u̯ had been eliminated early on [17], and *əu̯
and *ə̄i ̯ never existed; original *eu̯ had become *au̯ [11], but a new diphthong
*eu̯ had arisen from the fronting of non-front vowels [20]. The diphthongs *əi̯
and *ɪi̯ had played marginal roles in the system, being only represented in the
o- and i̯o-stem nominative plural (§ 3.13.7) and in the imperative second- and
third-person singular forms of the e- and i̯e-verbs (§§ 4.8.2 and 4.11.2).
This monophthongisation is later than the beginning of the fronting of
non-front vowels [20]. It is earlier than the second palatalisation [23] which
requires *ā̆i ̯ to have become a front vowel. Since monophthongised *åēi̯, as
opposed to original *ī, apparently does not trigger palatalisation of a following
velar [23], the diphthong probably went through a stage *ẹ̄ before becoming
*ī [29] (Kortlandt 2011c/1989: 165; Kallio 2006: 155, on Karelian viehkuŕi from
a pre-stage of ORu. vixъrь). For the intermediate stage *ō (CS *u [29]) as the
outcome of PS *åāu̯ åēu̯ , see Holzer 2001: 39–40 (where n. 16 mentions Holzer’s
earlier view that *ō was the Proto-Slavic stage, advanced in e.g. Holzer 1998b:
62–65).
According to a hypothesis put forward by Vermeer (1986/2013, 2000: 17–22),
in the Novgorod–Pskov area the monophthongisation of oral diphthongs
is later than the second palatalisation (the idea is supported by Kortlandt
1989/2011a: 188; see also Zaliznjak 1995/2004: 45; for criticism see the references
in Vermeer 2000: 20; and Dombrowski 2006 § 3).
{
{ k > c; g > ʒ; x > s||š } / _(u̯ )V[+front]
å ī(n)_V[–high]
}
The velar consonants *k g x are palatalised to *c ʒ s||š in two environments:
(1) before front vowels, possibly preceded by *u̯ , and (2) between *ī,� ̆ possibly
followed by *n, and *åā or *ə̄ ̆ (the only non-high vowels that occurred after
velars at that point). The outcome of the palatalisation of *x is *s in South and
East Slavic, but *š in West Slavic, e.g. PS ā-stem dat. sg. PS *ˈmāu̯ xāi̯ (CS *műsě
|| *műšě) > OCS musě, ORu. musě vs. Cz. mouše, Po. musze.
The second palatalisation is later than the monophthongisation of oral
diphthongs [22]. The results of the second palatalisation become phonologi-
cally relevant either when *ī�n̆ merges with *åēn [28], or when *ü and *ǖ, *åēi̯
merge with *i and *ī, respectively [29] (cf. Andersen 1998c: 591).
Opinions are divided as to whether we are dealing with two distinct pro-
cesses, viz. “second palatalisation” and “third palatalisation”, taking place at
separate times in different environments (e.g. Jagić 1901: 128; Pedersen 1905a:
385; Holzer 2001: 39–40), or they are two sides of the same process (see above
64 chapter 2
all Andersen 1998c: 588–589; also e.g. Andersen 1969: 554 n. 5; Vaillant 1950:
55; Kortlandt 1979/2011a: 266; 2011c/1989: 165; Vermeer 1986/2013, 2000, 2003b,
2006, 2008a). As stressed by Andersen (1998c: 588), since the outcomes of the
progressive and regressive changes are identical in the individual dialects, it
is considerably more attractive to assume that we are dealing with one single
change. A corollary of this view is that vowel fronting [20] was an active pro-
cess for an extended period of time (see the discussion in § 2.2.3). The history
of research on the second palatalisation of velars in Slavic is documented by
Vermeer in the publications just mentioned.
The second palatalisation has its maximal extension in South Slavic and the
major part of East Slavic. In West Slavic and some East Slavic areas, especially
in the Novgorod–Pskov region, there are certain restrictions (see also Andersen
1998c; Zaliznjak 1995/2004: 42–47):
e > i / _i̯
The short vowel *e is raised to *i before *i̯, e.g. PIE, PBS i-stem masc. nom. pl.
*‑ei̯es > PS *‑ei̯e [17] > *‑ii̯e [24] (CS *‑ьje [29]; cf. [25]). The nature and chronol-
ogy of the change is disputed, as is the relationship of the Slavic change with
the similar change in Baltic (see e.g. W. Hock 1995: 76–80 with references). Ras
mussen (1993: 476–477) suggests that the change is regular only in the begin-
ning of the word, whereas in other positions *ei̯e yields CS *i. According to
Andersen (2014), this change is the first in a series of sound changes that even-
tually led to vowel contraction (“monosyllabication”) in post-Proto-Slavic [25].
Phonological Background 65
1 j > ∅ / V_V
2 VV > V̄
In a series of sound changes beginning with the loss of intervocalic *j two adj
acent vowels contract. For a thorough investigation of the Common Slavic
vowel contractions see Andersen 2014.
The process remains active over an extended period of time, with partly
different results in the Slavic dialects. The chronology of the changes involved
also depends on factors such as the presence or absence of morpheme bound-
aries between the vowels.
Ci̯ > Cʲ
ē > ā / C[+front]_
After palatalised consonants, *ē becomes *ā, e.g. PS *ˈgēli > *ˈžēli [21] > *ˈžāli
[27] (CS *ža̋ lь [29]).
This change takes place after the first palatalisation [21]; otherwise *ē would
have been retained in the example just mentioned. Since the back vowel *ā
now occurs after palatalised consonants, the fronting of non-front vowels after
palatalised consonants [20] is no longer an active process at this point. The
change is later than the monophthongisation of oral diphthongs since e.g. PS
i̯ā-stem loc. sg. *‑i̯āi̯ > *‑i̯ēi̯ [20] yields *‑i̯ẹ̄ [22] (CS *‑ji [29]), unaffected by the
backing of *ē to *ā. Had the backing been earlier than the monophthongisa-
tion, *‑i̯āi̯ > *‑i̯ēi̯ [20] would have yielded **‑i̯āi̯ [27] > **‑i̯ē [22] (CS *‑jě [29]). As
Proto-Slavic probably also had non-final diphthongs with a long first part (see
§ 2.2.3), the change must be posterior to the metathesis of liquids, e.g. PS inf.
*gērˈtēi̯ > *žērˈtēi̯ [21] > *žērˈtẹ̄ [22] (CS *žertì [29]) > OCS žrěti (not **žrati). The
change is earlier than the monophthongisation of nasal diphthongs because of
PS i̯e-prs. 1sg. *‑i̯ān > *‑i̯ēn [20] > *‑i̯ān [27] (CS *‑jǫ [28]).
{ ī ̆n, åēn > ę; åān, åūn > ǫ; ə̄n̆ > ū||ē; ɪ̄n̆ > ę||ē } / _.
The nasal diphthongs *ī�n̆ åēn and *åān åūn are monophthongised to *ę and *ǫ in
all Slavic dialects. The reflex of *ə̄n̆ is *ū > y [29] everywhere except in the Old
Novgorod dialect, which has *ē > ě [29], e.g. PS ā-stem acc. pl. *‑ə̄n > OCS ‑y,
ONovg. ‑ě. Another isogloss is delineated by the outcome of *ɪ̄n̆ , which is ę in
South Slavic and *ē > ě [29] in West and East Slavic, e.g. PS i̯ā-stem acc. pl. *‑i̯ə̄n
> *‑i̯ɪn̄ [20] > OCS ‑ję, ORu. ‑jě. The cases that involve PS *ə̄ ̆ are treated in more
detail in Olander 2012.
This change is later than the backing of *ē to *ā after palatalised consonants
[27]. It is earlier than the merger of *ɪ ɪ̄ with *i ē into *ь ě [29].
By the “Second CS Vowel Shift” (Andersen 1985: 73; 1998b: 239; 1998a: 429) the
old quantitative vowel distinctions are reinterpreted as qualitative distinc-
Phonological Background 67
tions. The short vowels *i ü ɪ merge as *ь, and *e a u become *e o ъ in all Slavic
dialects. As for the long vowels, *ī ẹ̄ ǖ merge as *i; *ē ɪ̄ merge as ě; and *ā ō
ū become *a u y. The development of the central vowels marks an isogloss
that separates the Old Novgorod dialect, where *ə ə̄ merge with *e ē as e ě,
from the remaining Slavic dialect area, where *ə ə̄ merge with *u ū as ъ y (see
Olander 2012). After a labial the development of *ə ə̄ to ъ y seems to be pan-
Slavic, as in PBS dat. pl. *‑mas > PS *‑mə [17] (CS *‑mъ [29]) > OCS, ONovg. ‑mъ
and PBS 2pl. pron. encl. acc.–dat. *u̯ ōs > PS *u̯ ə̄ [15|17] (CS *vy [29]) > OCS vy,
ONovg. vy (Olander 2012: 335, where the pronominal forms are not taken into
consideration).
The reinterpretation of the vowel system is, in many dialects, accompanied
by the rise of new long vowels as the results of contractions, compensatory
lengthening etc. In some instances it seems that originally long vowels pre-
serve their quantity in Slavic dialects (see § 2.2.3).
This change includes three mergers: that of *i, *ü and *ɪ into *ь; that of *ī,
*ẹ̄ and *ǖ into *i; and that of *ē and *ɪ̄ into *ě. The rounded front vowels *ü ǖ
and the closed mid vowel *ẹ̄ were posited in order to explain why there is no
second palatalisation after PS *i̯åū and *åēi̯ [23], and the centralised front vow-
els *ɪ ɪ̄ were required to explain the distinct reflexes of PS *i̯ə̄n̆ in the Slavic
dialects [20|28|29]. Accordingly, the reshaping of the vowel system postdates
these changes. It should be noted that in the actual course of events, both the
remaking of the vowel system and the various mergers have probably taken
place in parallel and at different times in different Slavic dialects.
Chapter 3
Nominal Inflection
singular the boundary between the consonantal stem-suffix and the case–
number marker is still clearly seen in Proto-Slavic consonant stems, as men-
tioned in § 3.1. Accordingly, apart from the different nominative(–accusative)
formations, only one set of endings is treated in this study.
Like the i- and u-stems, the Proto-Indo-European consonant stems showed
an alternation between long grade, full grade and zero grade in the stem-form-
ing suffix, e.g. non-neut. r-stem nom. sg. *‑ēr, acc. sg. *‑er‑m̥ , gen. sg. *‑r‑os. The
suffix had e-timbre in some paradigm types and o-timbre in others. In Slavic
it is generally the e-timbre that survives; in the nominative singular ending of
n-stems, however, the o-timbre is preserved. The vocative form had full grade
of the stem, e.g. PIE *dʰugh₂ter > Ved. duhitar, Gk. ϑύγατερ. In Slavic consonant
stems the nominative singular was used as the vocative form (by conversion;
see Andersen 2012: 147–148).
The nominative(–accusative) singular of the various consonant stems has
the following forms in Proto-Slavic (and Common Slavic):
In the oldest attested stages of the Slavic dialects consonant stems constitute a
separate declension, but they tend to merge with other stems, often i-stems, a
tendency which, in the case of many lexemes, was probably completed already
in pre-Proto-Slavic. A similar tendency is seen in the Baltic languages (Stang
1966: 219), and it is probable that the beginnings of the process go back to the
common ancestor of the two branches.
The formation of consonant stems in Slavic is treated by Vondrák 1906/1924:
658–669; Vaillant 1958: 165–181; Arumaa 1985: 13–49, 63–68. For Proto-Indo-
European see Olsen forthc. chapters 2, 5 and 7–11.
3.2.2 i-Stems
The Proto-Slavic i-stems probably did not have separate masculine and femi-
nine subparadigms yet, but in Common Slavic the two genders received
distinct forms in the instrumental singular and the nominative plural. The
Proto-Indo-European neuter subparadigm of the i-stems had almost com-
pletely disappeared, the only clear remnant of a neuter form being the nom-
inative-accusative plural ending found in the numeral PS *ˌtrī (CS *trȋ) from
PIE *tríh₂.
Two forms of the inanimate interrogative pronoun are sometimes regarded
as i-stems, viz. PS nom.–acc. *ˌki (CS *čь̏ ) (§ 3.3.16) and gen. *kesa (*česo, *čьso)
(§ 3.5.7). From a Slavic viewpoint these forms do not belong to the i-stem
paradigm, but since they may historically be related to the i-stems, they are
included in the table below.
72 Chapter 3
3.2.3 u-stems
The u-stem declension still constitutes a separate paradigm in the oldest Slavic
languages, but tends to merge with the o-stems in most dialects. Slavic u-stems
are always masculine. The Indo-European proto-language also had feminine
u-stems, traces of which may be preserved in Baltic, and neuter u-stems; the
latter are preserved in Old Prussian, e.g. pecku, meddo, but have become mas-
culine in East Baltic and Slavic (see Stang 1966: 213–214; Vaillant 1958: 113–114;
Arumaa 1985: 58–59). Since adjectives preserve the old inflection more faith-
fully than nouns in Greek, the adjective ἡδύς is used as the example word in
the tables.
Nominal Inflection 73
nom. *ˌsūnu (*sy̑ nъ) *ˌsūnū (*sy̑ ny) *ˌsūnau̯ e (*sy̑ nove)
acc. *ˌsūnu (*sy̑ nъ) *ˌsūnū (*sy̑ ny) *ˌsūnū (*sy̑ ny)
gen. *ˌsūnau̯ (*sy̑ nu) *sūnau̯ au̯ (*synovu) *sūnaˈu̯ u (*synòvъ)
dat. *sūnaˈu̯ ei̯ (*synovì) *sūnumā (*synъma) *sūnuˈmə (*synъ̀mъ)
instr. *sūnuˈmi (*synъ̀mь) *sūnumā (*synъma) *sūnuˈmī (*synъmì)
loc. *sūˈnāu̯ (*synù) *sūnau̯ au̯ (*synovu) *sūnuˈxu (*synъ̀xъ)
voc. *ˌsūnau̯ (*sy̑ nu) – –
The Proto-Indo-European u-stems, like the i-stems (see § 3.2.2), had a suffix
in the full grade (*-eu̯ -) or the zero grade (*-u-) followed by the case–number
marker in the zero or full grade.
The Proto-Slavic u-stems are inflected as shown in table 4.
The Slavic languages show traces of i̯u-stems, e.g. OCS mǫžь from PS *ˌmangi̯u
(CS *mǫ̑žь), although they tend to be incorporated into the i̯o-stem declension
from an early stage (Meillet 1918a; 1924/1934: 415; Igartua 2005a: 276–280 with
references). To the extent that i̯u-stems exist as a distinct type in Proto-Slavic,
their inflection is identical to that of u-stems until the fronting of non-front
vowels [20] in post-Proto-Slavic. The i̯u-stems are therefore not analysed sepa-
rately in this chapter.
For the formation of (i̯)u-stems in Slavic see Vondrák 1906/1924: 656–658;
Vaillant 1958: 117–121; Arumaa 1985: 56–63. Proto-Indo-European u-stems are
treated in Olsen forthc. chapter 4.
3.2.4 ā-stems
The Proto-Slavic ā-stems have three subparadigms: (1) nouns and adjectives
with a nominative singular in PS *‑ā (CS *‑a), (2) nouns with a nominative
singular in *‑ī (CS *‑i) and (3) pronouns. Type (2) characterises a small group of
nouns, the *baˈgūni̯ī (CS *bogy̋ n’i) type. In Proto-Indo-European this type, the
so-called devī� ́ type, constituted a separate paradigm, but in Proto-Slavic the
type is inflected as a (i̯)ā-stem with the exception of the nominative singular.
Pronominal forms are treated individually only when they are substantially
different from the corresponding noun endings, not when they differ only by
the presence of the interfix PS *‑ai̯‑ (see § 3.2.6). The Slavic ā-stem nouns are
74 Chapter 3
mostly feminine, but masculine ā-stems designating male persons occur, e.g.
OCS sluga ‘servant’ (type 1), sǫdьji ‘judge’ (type 2).
According to the standard view, the Proto-Indo-European ā-stems con-
tained a suffix *‑ah₂‑ to which the consonant-stem case–number markers were
attached (see e.g. Clackson 2007: 96; cf. Beekes 1995/2011: 199–201 for a different
view). The only exception is the vocative singular, which seems to end in *‑a
with no laryngeal.
Although the ā-stems seem to have been productive in most Indo-European
branches except Anatolian, it is difficult to find ā-stem cognates that indis-
putably go back to the proto-language. According to one view, ā-stems with a
non-ablauting suffix *‑ah₂‑ never existed in the common ancestor of Anatolian
and the remaining Indo-European languages (cf. e.g. Beekes 1995/2011: 199–
201; and the discussions of feminine stems in *‑h₂ in Clackson 2007: 104–112;
Melchert 2014: 259; forthc. § 3.3.1.1). What matters for our purposes is that we
can quite safely reconstruct an ā-stem paradigm at least for non-Anatolian
Indo-European (see § 1.5.3).
Table 5 presents the Proto-Slavic ā-stem paradigm.
nom. *naˈgā (*nogà; *dušà); *ˌnagāi̯ (*nȍʒě; *dȗši) *ˌnagə̄ (*nȍgy||ě; *dȗšě)
*baˈgūni̯ī (*bogy̋ n’i)
acc. *ˌnagān (*nȍgǫ; *dȗšǫ) *ˌnagāi̯ (*nȍʒě; *dȗši) *ˌnagə̄n (*nȍgy||ě;
*dȗšę||ě)
gen. *naˈgə̄ (*nogỳ||ě̀; *dušě̀) *nagāu̯ (*nogu; *dušu) *naˈgu (*nògъ; *dúšь)
dat. *ˌnagāi̯ (*nȍʒě; *dȗši) *naˈgāmā (*noga̋ ma; *naˈgāmə (*noga̋ mъ;
*duša̋ ma) *duša̋ mъ)
instr. noun *nagān (*nogǫ, *naˈgāmā (*noga̋ ma; *naˈgāmī (*noga̋ mi;
→*nogojǫ̀ ; *dušǫ, →*dušejǫ̀ ); *duša̋ ma) *duša̋ mi)
pron. *taˈi̯ān (*tojǫ̀ ; *jejǫ̀ )
loc. *naˈgāi̯ (*noʒě̀; *dušì) *nagāu̯ (*nogu; *dušu) *naˈgāxu (*noga̋ xъ;
*duša̋ xъ)
voc. *ˌnaga (*nȍgo; *dȗše) – –
Nominal Inflection 75
3.2.5 o-stems
The Proto-Slavic o-stem declension comprises four subparadigms: (1) mas-
culine nouns and adjectives, (2) neuter nouns and adjectives, (3) masculine
non-personal pronouns and (4) neuter non-personal pronouns. The subpara-
digms only differ in a few forms and may conveniently be grouped together
in one paradigm. Regarding the selection criteria for pronominal endings see
§ 3.2.6. Whereas Proto-Indo-European had a few feminine o-stem nouns such
as *snusó‑ ‘daughter-in-law’, Proto-Slavic o-stems are invariably masculine; for
practical reasons I generally refer to Proto-Indo-European non-neuter o-stems
as masculine.
In Proto-Indo-European the o-stems, or thematic stems, contained the
thematic suffix *e/o—almost invariably *o—followed by the case–number
marker. The timbre of the thematic vowel was apparently determined by the
voicedness of the following segment in pre-Proto-Indo-European: *o before
voiced segments, *e elsewhere (Rasmussen 1989b: 139). In nouns there was a
tendency to replace the e-timbre with the more common o-timbre. Most case–
number markers of the o-stems were identical to those of the consonant stems,
but in some forms they differed, e.g. o-stem abl. sg. *‑h₂ad vs. C-stem gen.–abl.
sg. *‑e/os. The o-stem paradigm is relatively well preserved in Proto-Slavic.
Table 6 shows the Proto-Slavic o-stem paradigm.
From an inflectional point of view there is no i̯o-stem paradigm distinct
from the o-stem paradigm in Proto-Slavic; all endings of the o- and i̯o-stems
are identical. With the fronting of non-front vowels [20] in the Common Slavic
period do most endings develop two allomorphs, depending on the absence or
presence of a stem-final *‑i̯‑. The masculine i̯o-stem vocative singular form in
*‑i̯e is replaced with the reflex of the i̯u-stem form PS *‑i̯au̯ (CS *‑ju) in Common
Slavic.
76 Chapter 3
masc. nom. *ˌtakə (*tȍkъ||e; *gȍjь) *ˌtakā (*tȍka; *gȍja) *ˌtakəi̯ (*tȍk||ci; *gȍji)
neut.
nom.–acc. *ˌsuta (*sъ̏to; *pȍl’e) *ˌsutai̯ (*sъ̏tě; *pȍl’i)
*sutˈā (*sъtà; *pol’à)
masc. acc. *ˌtaku (*tȍkъ; *gȍjь) *ˌtakā (*tȍka; *gȍja)
*ˌtakə̄n (*tȍky||ě;
*gȍję||ě)
gen. noun *ˌtakā (*tȍka; *gȍja); *takāu̯ (*toku; *goju) noun *taˈku (*tòkъ;
pron. *taˈga (*togò; *jegò) *gòjь); pron. *tai̯ˈxu
(*tě́xъ; *jíxъ)
dat. *ˌtakāu̯ (*tȍku; *gȍju) *takamā (*tokoma; *takaˈmə (*tokòmъ;
*gojema) *gojèmъ)
instr. *taˈkā (→*tȍkomь, *takamā (*tokoma; *taˈkū (*tokỳ; *gojì)
→*tȍkъmь; →*gȍjemь, *gojema)
→*gȍjьmь)
loc. noun *ˌtakai̯ (*tȍk||cě; *takāu̯ (*toku; *goju) *takai̯ˈxu (*tok||cě́xъ;
*gȍji); pron. *taˈmi (*tòmь; *gojíxъ)
*jèmь)
voc. *ˌtake (*tȍče; →*gȍju) – –
Arumaa 1985: 173–174 with references). It is unclear if the loss of the *s in Slavic
is phonetically or morphologically conditioned (Matzinger 2001: 196–198;
a morphological solution is preferred by e.g. Brugmann 1909–1911: 360, 362;
Vaillant 1958: 369–370).
In this study specifically pronominal endings are only analysed when they
differ from the corresponding noun endings by more than the addition of an
interfix; for instance, PS masc.-neut. gen. sg. *‑aga (CS *‑ogo; *‑jego) and masc.–
neut. loc. sg. *‑ami (*‑omь; *‑jemь) are included. Pronominal endings contain-
ing an interfix PS *‑ai̯‑ or *‑am‑ plus a noun ending or case–number marker are
not treated; for examples see above.
*mъn‑). Those of the second-person singular are based on *t‑, *teb‑, *tab‑ (CS
*t‑, *teb‑, *tob‑); a parallel situation is found in the reflexive pronoun, which
has the stems *s‑, *seb‑, *sab‑ (CS *s‑, *seb‑, *sob‑). There is much less allomor-
phy in the dual and plural forms: the oblique first-person dual and plural forms
are based on a stem PS *n‑ (CS *n‑), while the second-person dual and plural
are based on *u̯ ‑ (CS *v‑).
The inflection of the first- and second-person pronouns and the reflexive
pronoun in Proto-Slavic is presented in tables 7–9.
(no number)
nom. –
acc. *sēn (*sę)
gen. *ˌsebe (*sȅbe)
dat. tonic *se/aˈbāi̯ (*se/obě̀̀); encl. *sei̯ (*si)
instr. *saˈbai̯ān / *sabaˈi̯ān (*sobòjǫ / *sobojǫ̀ )
loc. *se/aˈbāi̯ (*se/obě̀)
The stem variants have mixed origins. Some of them are inherited from Proto-
Indo-European, others have arisen by sound changes in the prehistory of Slavic,
and still others are the results of analogical developments. While a treatment
of the stem variants as such is outside the scope of this study, the forms of
the personal pronoun that present diachronic interest for the analysis of the
inflectional system are included here. For the variation between *men‑ and
*mun‑ and between *teb‑, *seb‑ and *tab‑, *sab‑ see §§ 3.5.8 and 3.6.6.
As for the notoriously unstable quantity of the vowel of the (usually mono-
syllabic) nominative forms of the personal pronoun in Indo-European, the
Baltic languages do not seem to show a clear pattern. Lithuanian points mostly
to originally short forms: 1sg. àš, 2sg. tù (NWŽem. tọ̀), 1du. Žem. vẽdọ, 2du. jùdu,
1pl. mẽs (dial. mès, mė̃s); but long 2pl. jū̃s. A similar pattern is observable in
Latvian: es, tu, mẽs (dial. mes), jũs. Old Prussian has as / es, tu / tū / toū, mes,
ioūs, with long vowels in the second-person forms. Proto-Slavic, on the other
hand, invariably has a long vowel: 1sg. *ˈēzu (CS *ja̋ zъ, *já), 2sg. *ˌtū (*ty̑ ), 1du.
*u̯ ū (*vy), 2du. *u̯ ē (*vě), 1pl. *ˌu̯ ū (*vy̑ ), 2pl. *ˌmū (*my̑ ).
There is no general agreement as to whether the long vowels reflect Proto-
Indo-European long vowels or short vowels followed by a laryngeal. Taken at
face value, however, the acute tone of the Latvian and Old Prussian long forms
point to the earlier presence of a laryngeal (see also the discussion in Kortlandt
2013). The forms with short vowels attested in Baltic and the remaining Indo-
European languages may result from a loss of the final laryngeal in pausa (cf.
Rix 1976/1992: 75, 178), a hypothesis which is supported by the fact that in the
one form that clearly has not been shortened in Baltic, viz. Li. 2pl. jū̃s, Lv. jũs,
OPr. ioūs, the laryngeal was in a closed syllable. It is also very likely that at least
some of the short forms found in Indo-European languages are the results of
80 Chapter 3
PIE Brugmann 1909–1911: 130: PIE *‑ūs | Beekes 1995/ 2011: 201: PIE
*‑uh | Rasmussen forthc. a: – | Debrunner & Wackernagel 1930/
1975: 187 | Rix 1976/1992: 71: Gk. ‑ῦς goes back to PIE *‑uhs | Sihler
1995: 327–328: Gk. ‑ῡ�ς́ is probably from PIE *‑uh₂s | Weiss 2009/
2011: 253: Ved. śvaśrū́‑, OCS svekry reflect PIE *‑uh₂s; La. socrus
has been reshaped on the model of the u-stems | Krahe 1942/
1967: – | Boutkan 1995: – | Kloekhorst 2008a: –
PBS Olander 2009: 168 | Endzelīns 1971: – | Stang 1966: – | Otrębski
1956: – | Endzelīns 1923: –
PS Vondrák 1908/ 1928: 2 (1906/ 1924: 658–659) | Hujer 1910: 9–12:
Slavic ‑y reflects PIE *‑ūs with acute tone | Meillet 1924/ 1934:
429 | Vaillant 1958: 263: PIE *‑ūs | Bräuer 1969a: 179: CS *‑y goes
back to PIE *‑ūs | Arumaa 1985: 63–65 | Igartua 2005a: 303–304:
PIE *‑uh₂s | Aitzetmüller 1978/1991: 102: OCS ‑y reflects PIE *‑ūs
Nominal Inflection 81
PBS PIE *‑uhs regularly yielded PBS *‑ˈṵ̄s [1|4]. The ending is not preserved
in Baltic.
PS PBS *‑ˈṵ̄s is regularly reflected as PS *‑ˈū [12|13|17] (CS *‑y [29]), which is
relatively well attested in the old Slavic dialects. In most cases the ending was
replaced with other formations, as in Old Czech svekrev which has the accusa-
tive ending PS *‑uu̯ i (CS *‑ъvь [29]) (but cf. Kortlandt 1997a/2009: 132–133, who
assumes that the original accusative singular ending in Slavic is ‑ovь from PIE
*‑eu̯ hm̥ ). The ū-stems had a certain productivity in loanwords at a late stage
of Common Slavic, e.g. CS *bűky ‘letter’ from PGmc. *bōk‑ (see Pronk-Tiethoff
2013: 80–81).
PIE Brugmann 1909–1911: 125–127: PIE *‑ḗr, *‑ē ̃ | Beekes 1995/2011: 195:
PIE (hysterodynamic) *‑ḗr | Rasmussen forthc. a § 9: PIE (hysterody-
namic) *‑ēr | Debrunner & Wackernagel 1930/1975: 203 (Wackernagel
1896: 108): PIE *‑ē (perhaps originally preconsonantal variant) and
*‑ēr | Rix 1976/1992: 150: PIE *‑ē(r) | Sihler 1995: 290–291: PIE *‑ēr;
*‑r was deleted in IIr. and BS by analogy with n-stem nom. sg. | Weiss
2009/2011: 198–200: PIE *‑ēr < *‑ers | Krahe 1942/1967: 38–39: PIE
*‑ḗr | Boutkan 1995: 269–273: PIE *‑ēr | Kloekhorst 2008a: 108,
326: Hi. ḫašterza probably represents PIE *h₂stḗr plus *‑s
82 Chapter 3
PS Some authors (e.g. Vaillant loc. cit., Igartua loc. cit.) consider Slavic *‑ī for
expected *‑ē to be analogical from the type CS *bogy̋ n’i (§ 3.3.13). In an early
study Meillet explained CS *‑i as the result of *‑ēs, consisting of original *‑ē
plus an analogical *‑s (1897: 105). Because of the morphologically isolated
status of the form, however, it is difficult to find a reasonable analogical
model and to imagine a motivation for a replacement of the ending with a
form from another paradigm. Therefore I find it more reasonable to assume
the form is phonetically regular. The most straightforward explanation of
PS *‑ˈī (CS *‑i [29]) is to assume *ē was raised to *ī before final *‑r, which then
disappeared [18]. The ending is preserved in the old Slavic dialects.
Nominal Inflection 83
PIE Brugmann 1909–1911: 125–126 (1897: 392): OCS ‑y < *‑ū < *‑ų̄ < *‑ūn < PIE
*‑ṓn, alongside *‑ō̃ | Beekes 1995/2011: 194: PIE *‑ōn | Rasmussen
forthc. a § 8: Slavic ‑y reflects PIE *‑ō̃ or *‑ons | Debrunner &
Wackernagel 1930/1975: 270–271: PIE form did not end in *‑n; *‑n was ana-
logically reintroduced in various branches | Rix 1976/1992: 145: PIE
*‑ō(n) | Sihler 1995: 290: PIE *‑ō < pre-PIE *‑ons; Gk. added ‑ν by anal-
ogy with r-stem nom. sg. | Weiss 2009/2011: – | Krahe 1942/1967:
44: ONor. hani and probably also Go. hana are from PIE *‑ēn, alongside
*‑ẽ; WGmc. forms reflect PIE *‑õ, alongside *‑ōn | Boutkan 1995: 278–
279, 281–282: PIE, PGmc. *‑ōn is preserved only in Go. | Kloekhorst
2008a: 108, 302, 589: Hi. ḫāraš is from PIE *‑ōn plus *‑s
PBS Olander 2009: 168 | Endzelīns 1971: 161–162 | Stang 1966: 219–220:
Li. ‑uõ reflects PIE *‑ō, not *‑ōn | Otrębski 1956: 56: original ending
was *‑ōn | Endzelīns 1923: 321–322
PS Vondrák 1908/1928: 2 | Hujer 1910: 36–41: Slavic ‑y is from *‑ṓn; Li.
‑uõ is from *‑ō̃ | Meillet 1924/1934: 425–426: Slavic ‑y is from *‑ō plus
*‑s | Vaillant 1958: 204–206: Slavic *‑y does not reflect PIE *‑ō but rep-
resents analogical *‑ōns or *‑ōn | Bräuer 1969b: 8 (1961: 105): CS *‑y <
*‑ūn reflects PIE acute *‑ōn | Arumaa 1985: 103–106: development of
*‑ō to *‑ū > CS *‑y is unclear | Igartua 2005a: 322–329: PIE *‑ōn; CS *‑y
reflects *‑ōn plus secondary nom. sg. marker *‑s | Aitzetmüller 1978/
1991: 93–94: OCS ‑y, Li. ‑uõ are from PIE *‑ōn
the *n from the remaining forms was added to the nominative singular on the
pattern of the r-stems; the same restitution of *‑n apparently took place in
Slavic (see below).
PBS PIE *‑ō regularly yielded PBS *‑ˈō, preserved in Li. ‑uo. In Latvian the
n-stems have joined the i-stem paradigm and the ending has been normalised
as ‑ns, as in Old Prussian; the original vocalism of the nominative singular is
preserved in OLv. akmuons.
PS By analogy with the relationship in the r-stems between nom. sg.
*‑ēr and the stem *‑er‑, PBS *‑ˈō was remade to pre-PS *‑ōn > PS *‑ū [18] (CS
*‑y [29]) with *‑n from the stem *‑en‑. The ending has disappeared in most of
the attested Slavic dialects, but is preserved in e.g. OCS (Codex Suprasliensis)
kamy, Polab. komoi, Slnc. ką̃ mä, ką̃ m (see Lorentz 1903: 245; for further traces of
the type in Slavic see Pronk 2009: 107 with n. 14). The length of Cz. kámen possi-
bly preserves a trace of an originally disyllabic form, viz. *kámy (Trávníček 1935:
250). ORu. kamy, plamy belong to the literary language (Gorškova & Xaburgaev
1981: 155). In Old Russian, Old Czech and other dialects, the original form was
substituted with one based on the stem PS *ˈkāmen‑ (CS *ka̋ men‑) etc. with
addition of the i-stem ending PS *‑i (CS *‑ь); cf. the similar process in Latvian.
It is often assumed that korę, attested as a nominative-accusative form in
Old Church Slavonic and Russian Church Slavonic, reflects the Proto-Indo-
European n-stem type with a nominative singular in *‑ēn represented by Gk.
ποιμήν (Hujer 1910: 40; Meillet 1924/1934: 149, 425–426; Mikkola 1950: 47; Vaillant
1950: 216; 1958: 195–196; Aitzetmüller 1978/1991: 93–94; Igartua 2005a: 327–328;
for discussions of the form see also Vaillant 1932b: 32–36; Arumaa 1985: 103–
106; Havlová & Erhart 1996: 338–339; Pronk 2009: 109–110 with n. 19). While this
type certainly did exist in Proto-Indo-European (although *‑ēn for *‑ē is prob-
ably a later analogical creation), I do not consider it to be directly reflected in
Slavic korę since the combined evidence of PS *dukˈtī (CS *dъt’ì) from pre-PS
*‑ēr and PS *ˈkāmū (CS *ka̋ my) from pre-PS *‑ōn points to a general loss of
word-final dentals after the mid vowels *ē ō in pre-Proto-Slavic [18]. Thus pre-
PS *‑ēn would probably have yielded PS **‑ī (CS **‑i).
The form korę may be an innovation, replacing kory (also attested in OCS)
by analogy with the nominative singular of the neuter n-stems (§ 3.3.4; see
Bräuer 1969b; Lunt 1955/2001: 226 n. 48 considers korę to be a “local innova-
tion”; Derksen 2008: 237 reconstructs CS *kory; but cf. Havlová & Erhart 1996:
339; Trubačёv 1985: 65). The instance of korę in an Old Church Slavonic man-
uscript found in 1975 is regarded by Mareš (1993: 129) as a neuter, occurring
alongside masc. kory (for the manuscript see also H. Birnbaum & Schaeken
1997: 32–33, 147).
Nominal Inflection 85
PIE Brugmann 1909–1911: 145–146 (1897: 392): PIE *‑n̥ , *‑ḗn, *‑ē̃; OCS ‑ę is from
*‑ḗn, not from *‑n̥ | Beekes 1995/2011: 205: PIE *‑n̥ | Rasmussen
forthc. a § 8: PIE *‑n̥ ; Slavic ‑ę < *‑en is analogical from inflected
forms | Debrunner & Wackernagel 1930/1975: 271–273: Ved. ‑a reflects
PIE *‑n̥ ; Ved. variant ‑ā is metrically lengthened and does not corre-
spond to Go. ‑o, and probably not to OCS ‑ę | Rix 1976/1992: 144,
152: PIE *‑n̥ | Sihler 1995: 297: PIE *‑n̥ | Weiss 2009/2011: 313: PIE
*‑n̥ | Krahe 1942/1967: 46: Go. ‑o points to PIE *‑õ; other Gmc. lan-
guages point to PIE *‑ōn; perhaps the form was originally a collective
with lengthened grade, like Gk. ὕδωρ | Boutkan 1995: 285–286: PIE
*‑n̥ > PGmc. *‑uN was replaced with forms based on nom.–acc. pl. in the
Gmc. dialects | Kloekhorst 2008a: 108, 518: Hi. ‑an reflects PIE *‑n
PBS Olander 2009: – | Endzelīns 1971: 162 | Stang 1966: 225: OPr.
‑en and OCS ‑ę reflect *‑en, not *‑n̥ | Otrębski 1956: – | Endzelīns
1923: –
PS Vondrák 1908/1928: 2 (1906/1924: 660–661): Slavic ‑ę is from *‑n̥ t | Hujer
1910: 49–50: Slavic ‑ę reflects PIE *‑ḗn, alongside *‑n̥ | Meillet 1924/
1934: 426: PIE *‑n̥ ; Slavic ‑ę is from *‑ēn, perhaps reflecting neut. nt-stem
ending *‑ēnt | Vaillant 1958: 204–205: CS *‑ę is not from PIE *‑n̥ , but
from new ending *‑ēn, analogical from neut. nt-stem ending | Bräuer
1969b: 25–26: CS *‑ę is not from PIE *‑en, but perhaps from PIE *‑ent
or *‑n̥ t | Arumaa 1985: 107: CS *‑ę does not reflect PIE *‑n̥ , but
goes back to *‑en introduced from oblique cases; there is no evidence
for acute *‑ḗn | Igartua 2005a: 331–334: PIE *‑n̥ could not yield CS
*‑ę | Aitzetmüller 1978/1991: PIE *‑n̥ ; OCS ‑ę is from secondary *‑ēn
PIE The form, which ends in the suffix in the zero grade with a zero case–
number marker, is preserved in most branches of Indo-European. In Germanic
86 Chapter 3
it was remade with the long vowel of the nominative-accusative plural (Lane
1963: 163–164; Boutkan loc. cit.).
PBS PIE *‑n̥ regularly yielded PBS *‑in [2]. Since PBS *‑im is reflected as OPr.
‑in in the accusative singular of the i-stems, it seems natural to assume that
PBS *‑in would also yield OPr. ‑in. In that case, ‑en in semen would have to
be secondary, probably taken from the word-internal variant of the suffix, not
attested in Old Prussian but reconstructable as PBS *‑en‑ (see Olander 2010: 93
with n. 21). It cannot be excluded, though, that ‑en is the phonetically regular
reflex of PBS *‑in and that i-stem acc. sg. ‑in has its vowel from other forms of
the paradigm.
PIE
Brugmann 1909–1911: 147 | Beekes 1995/ 2011: 197: PIE
*‑nt | Rasmussen forthc. a: – | Debrunner & Wackernagel 1930/
1975: 262: PIE *‑n̥ t | Rix 1976/1992: 144: ending of Gk. ἑκόν replaces *‑α
< PIE *‑n̥ t | Sihler 1995: 615 | Weiss 2009/2011: – | Krahe 1942/
1967: – | Boutkan 1995: – | Kloekhorst 2008a: 183–184: Hi. ‑an goes
back to *‑ant
Nominal Inflection 87
PBS PIE *‑n̥ d regularly yielded PBS *‑in [2|5|9]. A neuter ending is not pre-
served in the Baltic languages, but it is possible that its masculine counterpart
is attested in OPr. smunents, which seems to point to full grade of the suffix.
PS The ending is preserved as PS *‑in (CS *‑ę [28]), with regular reflexes in
the old Slavic dialects.
88 Chapter 3
OCS masc.–neut. bery; def. beryi, Li. suką̃ s Ved. bháran, ‑aṃs (t‑)
→grędęi, →živǫi; pišę Lv. →vȩr̂duošs; OLv. OAv. saošiiąs; YAv.
ORu. masc.–neut. →veda, →ved’a; →piša; sargus saošiiąs, barō
ONovg. masc. →priěd’a; →promyšlaja OPr. skellānts Gk. →φέρων
OCz. masc.–neut. →nesa; tešě La. →ferēns
Go. bairands
Hi. kunanza
PIE Brugmann 1909–1911: 130–131 (1897: 392): OCS ‑y, ‑ję reflect PIE
*‑(i̯)onts; Gk. ‑ων is analogical from n-stems | Beekes 1995/2011: 197:
PIE *‑nt(s) | Rasmussen forthc. a: – | Debrunner & Wackernagel
1930/1975: 261: Ved. ‑an, Av. ‑ąs are from PIE *‑Vnts; Gk. ‑ων perhaps reflects
PIE s-less ending | Rix 1976/1992: 150, 234: PIE *‑onts is preserved in
ὀδούς; Gk. ‑ων is perhaps from n-stem nom. sg. | Sihler 1995: 616: PIE
*‑ōn (< pre-PIE *‑onts) is only preserved in Gk.; other languages, includ-
ing OCS, reflect recomposed form *‑onts | Weiss 2009/2011: 436: PIE
*‑onts | Krahe 1942/1967: 79 | Boutkan 1995: 366 | Kloekhorst
2008a: 109, 183–184: Hi. ‑anza is from *‑ent(s), remade from PIE *‑nt(s)
PBS Olander 2009: – | Endzelīns 1971: 253 | Stang 1966: 113, 263–265: Li.
‑ą̃ s goes back to PIE *‑onts | Otrębski 1956: 250: Li. ‑ąs, OCS ‑y reflect
*‑onts | Endzelīns 1923: 719–721: OLv. ‑us, Li. ‑ą̃ s reflect PIE *‑onts
PS Vondrák 1908/1928: 2, 56–57, 138–139: OCS ‑y, ‑ję reflect *‑on(t)s, *‑i̯on(t)s;
Ru. nesa is analogical based on soft kryja | Hujer 1910: 42–46 (follow-
ing Zubatý): OCS ‑y, ‑ję reflect PIE *‑(i̯)ṓn, also in Gk. ‑ων; ORu. ‑a, OCz.
‑a reflect PIE variant *‑ō̃ | Meillet 1924/1934: 152, 430: ‑y, ‑ję are from
PIE *‑(i̯)onts | Vaillant 1958: 544–545, 548–549: OCS ‑y, ‑ję, Lv. ‑us
reflect PBS *‑ans < PIE *‑ō̆nts; Li. ‑ą̃ s is from *‑ants with reintroduced *‑t‑;
ORu. nesa, nes’a have ‑( j)a from i̯e-prs. piša, i-prs. nos’a, where reflexes
of *‑ę for regular *‑ě are due to influence from oblique cases; similarly
in NSl. | Bräuer 1969a, 1969b:—(1961: 107): OCS ‑y reflects *‑on(t)s;
OCS ‑ję, Ru. ‑ja go back to CS *‑ję < *‑i̯on(t)s | Arumaa 1985: 107 (1964:
117): Slavic ‑y reflects *‑on(t)s or *‑ṓn; soft ‑ję is from *‑(i̯)onts; NSl. forms
in ‑a perhaps show early loss of *n before *‑s in sandhi | Igartua
Nominal Inflection 89
2005a: 359–362: OCS ‑y is from PIE masc. *‑ons < *‑onss < *‑onts; OCS ‑ǫ
perhaps reflects PIE neut. *‑ont; ORu. ‑a, OCz. ‑a are from soft ending
*‑ję | Aitzetmüller 1978/1991: 236–237: OCS ‑y is from *‑on(t)s or *‑ōn;
soft ‑ję is from *‑(i̯)onts
PBS It is unclear if PIE *‑onts yielded PBS *‑ants [7] or *‑ans [7|9] (see the
discussion in [9]). The latter possibility is indicated by OPr. ‑ānts, but the *t
may easily have been reintroduced under the influence of the oblique stem
*‑ant‑ (Stang 1966: 265; Zucha 1986: 134). Alternatively, the Old Prussian ending
may reflect *‑antis, a new form created on the analogy of the i-stems (Leskien
1876: 21; Endzelīns 1923: 719; 1944: 98). Latvian ‑uošs reflects *‑anti̯as, consisting
of the suffix *‑anti̯‑ and the o-stem nominative ending *‑as. OLv. ‑us preserves
the regular outcome of PBS *‑ants.
Slavonic the normal reflexes are ‑y in e-presents and ‑ę in i̯e-presents. The defi-
nite form appears not only as ‑yi as expected, but also as ‑ęi and, less frequently,
as ‑ǫi. In some manuscripts the ę of ‑ęi is written with a special sign in this form
(see e.g. Diels 1932–1934/1963: 232–233; Vaillant 1948/1964: 30, 127; Ferrell 1971:
88–89).
In Old Russian we find ‑a in the e-present and ‑( j)a in the i̯e-present. From
the eleventh century e-present forms like nes’a become increasingly frequent
(Ferrell 1965a: 15–16 with n. 3). In the Novgorod area participles are not inflected
for case; here we only find forms in ‑ja, even in the oldest texts (Zaliznjak 1995/
2004: 134).
In Old Czech we find ‑a in the e-present and ‑ě in the i̯e-present. Old Polish
has forms like biorø in the e-present and piszø in the i̯e-present, ø representing
the merger of the Common Slavic nasal vowels *ę and *ǫ in early Polish. In an
Old Polish manuscript, the Kazania Świętokrzyskie from the fourteenth cen-
tury, there are five instances of rzeka (written <reca>), apparently functioning
as a particle, which seems to contain the present active participle ending. In
this manuscript the form rzeka is attested alongside one occurrence of rzekø,
which is also found in other Old Polish texts.
In the oblique stem of the active participle of the i̯e-present we expect
PIE *‑i̯onti̯‑ > PS *‑i̯anti̯‑ [7] to be reflected as CS *‑jęt’‑ [20|26|28], but the old
Slavic dialects point to *‑jǫt’‑ with generalisation of the suffix of the e-present
(Vaillant 1950: 192).
As for the diachronic interpretation of the material, the Old Church
Slavonic forms ‑y and ‑ję are in line with what we expect from PIE *‑onts and
*‑i̯onts, since ‑y and ‑ję are also found in the phonologically similar structures
in the accusative plural of the ā- and o-stems. There we see that PIE *‑(i̯)ahm̥ s >
*‑(i̯)ahn̥ s merged with *‑(i̯)ōms > *‑(i̯)ōns into pre-PS *‑(i̯)āns (through [1|2|3]
and [15], respectively) > PS *‑(i̯)ə̄n [17] (CS Cl *‑y || ONovg. ‑ě [28|29]; CS S *‑ję
|| CS N *‑jě [20|28|29]). If the o-stem ending had an original short *o (see the
discussion in § 3.14.5), the structure would be even closer to that of the parti-
ciple, viz. PIE *‑(i̯)oms > *‑(i̯)ons > pre-PS *‑(i̯)ans [7] > PS *‑(i̯)ən [17] (CS Cl *‑y
|| ONovg. ‑ě [28|29]; CS S *‑ję || CS N *‑jě [20|28|29]). The problematic forms are
the North Slavic ones in *‑( j)a.
Some authors derive these forms directly from PIE *‑ō, a variant of the
ending *‑ōn seen in Gk. ‑ων (Zubatý 1893: 504; Lamprecht 1987: 95; Hujer
loc. cit.; Trávníček 1935: 342). The fact that an ending containing a nasal is
needed anyway for the form in ‑ę in the OCS i̯e-verbs, combined with the mea-
gerness of the evidence for this Proto-Indo-European shape (see above), ren-
ders this view unattractive (see also Ferrell 1971: 86).
Nominal Inflection 91
It is also tempting to connect the North Slavic reflexes of PIE *‑(i̯)onts with
the *t of the Proto-Indo-European form, especially in view of the fact that the *t
may have been preserved in Baltic, indicating that it was still present in Proto-
Balto-Slavic (such a stance is taken by Kortlandt 1979/2011a: 105; 1983/2011a:
128–129; cf. also van Wijk 1925: 283–284). However, since I find it difficult to see
how the presence of *t phonetically could lead to a merger of pre-PS *‑ants and
*‑ā in CSN *‑a, I regard this idea as unlikely (see also Ferrell 1967b: 71).
In the following paragraphs I shall present my own interpretation of the
attested forms. It takes as its point of departure the hypothesis that pre-PS *ā̆
was centralised to PS *ə̄̆ in final syllables closed by a fricative [17] (see also
Olander 2012: 333–334), although I must admit that this hypothesis does not
shed significantly more light on the particular problem of the participle than
already existing explanations do. My interpretation of the material is in line
with those of Torbiörnsson 1921, 1923a, 1923b and Ferrell 1965a, 1967b, 1971;
Ferrell also treats the motivational factors behind the morphological changes
in the Slavic dialects.
It is, in my opinion, a priori most reasonable to assume that the phonetically
regular reflex of PIE *‑onts > PBS *‑an(t)s [7|9] is PS *‑ən [17] (CSCl *‑y [28|29] ||
ONovg. *‑ě [28|29]; CS S *‑ję [20|28] || CS N *‑jě ([20|28|29]), parallel to the accu-
sative plural forms mentioned above. The Old Church Slavonic forms ‑y, ‑ję
are the direct reflexes of PS *‑ən, *‑i̯ən. The ending ‑ę(i) has been transferred
from the ei̯e-present (thus also e.g. Stang 1942: 95). The variant ‑ǫ(i) may either
represent the original neuter form PS *‑an (CS *‑ǫ [28]) < PIE *‑ond ([5|7|9])
(Igartua 2005a: 362–364; but cf. p. 360) or it may be analogical based on the
oblique stem ‑ǫšt‑ < PS *‑anti̯‑ (CS *‑ǫt’‑ [26|28]) < PIE *‑onti̯‑ ([7]) (Vaillant
1948/1964: 127).
In Old Russian and Old Czech expected CS *‑y in hard stems and CS N *‑jě in
soft stems have been replaced with the reflexes of *‑a or *‑ę in both hard and
soft stems. It would theoretically be possible to postulate a phonetic change of
(short) PS *‑(i̯)ən to CS N *‑( j)ę > ‑( j)a, since there are no counterexamples, at
least if the masculine o-stem accusative plural had an original long *ō. Since
such a development is not supported by other pieces of evidence, it is purely
ad hoc. More importantly, a change of non-palatalised *‑ən to CS N *‑ę > ‑( j)a
does not seem to fit into the general development of Slavic final syllables. It
seems more reasonable to look for an analogical explanation.
The ending ‑a of Old Russian and Czech hard stems is often explained as
analogically introduced from the soft stems, where it is regarded as regular (e.g.
Torbiörnsson 1923b: 123). However, as mentioned above, the masculine ending
PIE *‑i̯onts would most likely yield PS *‑i̯ən > CS N *‑jě, which is incompatible
92 Chapter 3
with ORu. ‑( j)a. Accordingly, ‑( j)a cannot be phonetically regular in the mas-
culine nominative singular of the i̯e-present active participle.
The form in ‑( j)a may have originated in the ei̯e-verbs, where ‑( j)a is prob-
ably the regular outcome of both PS masc. *‑ei̯ən (CS *‑ę [20|25|28]) and neut.
*‑ei̯an (CS *‑ę [20|25|28]); or in the original neuter form of the i̯e-present, viz.
PS *‑i̯an (CS *‑ję [20|28]) from PIE *‑i̯ond (via [5|7|9]). In Czech, ‑( j)ě in the
i̯e-present may be the regular reflex of PS *‑i̯ən (CS N *‑jě [20|28|29]) from PIE
*‑i̯onts (via [7|9|17]). In order to explain the Czech ending ‑a in the e-present,
we may assume that *‑ę had become *‑a at the stage when the ending was
introduced here from the i̯e- and ei̯e-verbs. Subsequently *‑a became ‑ě after
palatalised consonants in the i̯e- and ei̯e-verbs, but remained ‑a after non-
palatalised consonants in the e-verbs.
OPo. rzeka is too sparsely attested to build any hypotheses on; it may be
a dialectal form (Ferrell 1971) or a borrowing from Old Czech (see Šaxmatov
1915/2002: 111; Stang 1969b: 160), or it may have lost the nasal through an allegro
development (Jakobson 1929/1971: 32 n. 21). The more widespread Old Polish
ending ‑ø, a nasal vowel reflecting the merger of *‑ę and *‑ǫ, probably does not
reflect PS *‑ən or *‑i̯ən, which would more likely have yielded *‑y, *‑je as shown
by the development of the parallel structures in the masculine o- and i̯o-stem
accusative plural, e.g. zęby and męže. Instead, OPo. ‑ø may represent the mas-
culine and neuter ending of the ei̯e-verbs or the neuter ending of the e- and i̯e-
presents, viz. PS *‑(i̯)an (CS *‑( j)ę), or it may have been taken from the oblique
stem of the e- or i̯e-present, viz. PS *‑(i̯)anti̯‑ (CS *‑( j)ǫt’‑).
OCS masc.–neut. nom. →vedy; def. Li. nom.–acc. sg., Ved. →bhárat
→vedyi, →grędęi, živǫi; pišę masc.(-fem.) nom. pl. suką̃ OAv. yasō.x́ iiə̄n;
ORu. masc.–neut. nom. →veda, Lv. dial. masc. nom. pl., YAv. ?
→ved’a; piša; ONovg. masc. masc.–fem. nom. sg. ȩsu Gk. φέρον
nom. sg. →priěd’a; masc. nom. OPr. enterpon(́?) La. →ferēns
sg. promyšlaja Go. →bairando
OCz. masc.–neut. nom. →nesa; tešě Hi. ḫūman (OS)
1975: 262: Ved. ‑at, Av. ‑at̰ reflect PIE *‑n̥ t; Gk. ‑ον is from *‑ont | Rix 1976/
1992: – | Sihler 1995: 615: PIE *‑n̥ t; Gk. ‑ον is from *‑ont with analogical
o-grade | Weiss 2009/2011: – | Krahe 1942/1967: 79 | Boutkan
1995: 366 | Kloekhorst 2008a: 183–184: Hi. ‑an reflects *‑ant
PBS Olander 2009: – | Endzelīns 1971: 253 | Stang 1966: 263–265: Li. ‑ą̃
is from PIE neut. nom.–acc. sg. form ending in *‑nt | Otrębski 1956:
250: Li. ‑ą reflects nom.–acc. sg. *‑ont | Endzelīns 1923: 720–721: Lv. ‑u,
Li. ‑ą̃ are probably from PIE *‑ont
PS Vondrák 1908/1928: 56–57, 138–139 | Hujer 1910: 47–49: OCS ‑y, ORu.,
OCz. ‑a are from masc.; OCS ‑ję is from PIE *‑i̯n̥t or *‑i̯ont | Meillet
1924/1934: 431: Slavic ‑y is from masculines | Vaillant 1958: 545: PIE
*‑n̥ t or *‑ont; masc. ending ‑y was transferred to neuters in Slavic by anal-
ogy with phonetically regular merger of masc. and neut. in i̯e-prs. (‑ję
< *‑i̯ons, *‑i̯ont) and i-prs. (‑ę < *‑īn(t)s, *‑īnt) | Bräuer 1969b: – |
Arumaa 1985: 108: Li. ‑ą̃ is from *‑ont; Slavic ‑y is probably masc. end-
ing | Igartua 2005a: 362–364: OCS ‑y is from masc.; OCS ‑ǫ is perhaps
from neut. *‑ont; OCS ‑ę reflects *‑n̥ t | Aitzetmüller 1978/1991: 236–
237: OCS ‑y is not the regular result of *‑ont; soft ‑ję reflects *‑i̯ont
PBS PIE *‑ond > PBS *‑an [5|7|9] is preserved in Lithuanian suką̃ and in the
Latvian dialect form ȩsu.
PS The expected reflex of PBS *‑an is PS *‑an (CS *‑ǫ [28]; *‑ję [20|28]). In
e- and i̯e-presents the neuter form was distinct from the masculine form in
Proto-Slavic; in ei̯e-verbs, on the other hand, PBS *‑ei̯an yielded PS *‑ei̯an (CS
*‑ę [20|25]), merging with masc. *‑ei̯ən (CS *‑ę [20|25|28]) at an early stage of
Common Slavic. This fact, together with other changes such as the merger of
PS neut. *‑i̯an and masc. *‑i̯ən in *‑ję in South Slavic, led to the eventual disap-
pearance of the distinction between neuter and masculine forms in the short
participle, which has distinct nominative and accusative singular forms even
in the neuter. The earliest attested stages of the Slavic dialects do not seem to
distinguish the two genders formally from each other.
94 Chapter 3
PS masc.–neut. →*u̯edu (CS *vedъ) PBS *‑u̯ōs(?) PIE masc. *‑u̯ ōts
PIE Brugmann 1909–1911: 128, 149 (1906: 563, 571–574): PIE had masc. *‑u̯ ōs
and *‑us, the latter preserved in Ved. ‑úḥ, Av. ‑uš, PBS *‑us > OCS ‑ъ;
PIE neut. form was *‑u̯ os, perhaps also *‑us > OCS ‑ъ | Beekes 1995/
2011: 198, 279: PIE masc. *‑uōs | Rasmussen forthc. a § 10: PIE masc.
*‑u̯ ōts, neut. *‑u̯ ot; weak stem *‑us‑ was generalised in BS; Gk. neut. ‑ός is
based on analogy with masc. | Debrunner & Wackernagel 1930/1975:
300: Ved. ‑vān is “nasale Erweiterung” of PIIr. *‑vās < PIE *‑u̯ ōs; alleged
masc. nom. sg. in ‑úḥ is “nicht gesichert” | Rix 1976/1992: 235: PIE
masc. *‑u̯ ōs, neut. *‑u̯ os | Sihler 1995: 619–621: PIE masc. *‑u̯ ōs, neut.
*‑u̯ os | Weiss 2009/ 2011: – | Krahe 1942/ 1967: – | Boutkan
1995: – | Kloekhorst 2008a: –
PBS Olander 2009: – | Endzelīns 1971: 256–257: Baltic masc. forms reflect
*‑u̯ ēs (also in IIr.), with secondary *n; Li. neut. ‑ę is analogical from
masc. | Stang 1966: 265–267: PIE masc. *‑u̯ ōs; OPr. ‑uns, ‑ons reflect
*‑u̯ ōns with nasal from prs. ptc.; Li. masc. ‑ęs, Lv. ‑is have nasal from prs.
Nominal Inflection 95
ptc. and e-grade from acc. sg. and perhaps loc. sg.; Li. neut. ‑ę is second-
ary | Otrębski 1956: 257–258: Li. ‑ęs, Lv. ‑is reflect *‑ēns, originally ptc.
of ē-prt., created on the pattern of prs. ptc. ‑ą̃ s; Li. pl. ‑ę is analogical to
prs. ptc. ‑ą; Slavic masc.–neut. ‑ъ reflects neut. ending *‑us | Endzelīns
1923: 727–728: Li. and Lv. masc. endings are from PBalt. *‑u̯ ēns with nasal
from prs. ptc.; earlier *‑u̯ ēs is perhaps identical to Av. ending
PS Vondrák 1908/1928: 2 (1906/1924: 668) | Hujer 1910: 52–54, 56: Slavic
masc. ‑ъ reflects *‑us, replacing *‑u̯ ōs; neut. ‑ъ is from PIE *‑us, along-
side *‑u̯ os > Gk. ‑ος | Meillet 1924/1934: 430–431: Slavic ‑ъ reflects *‑us
with generalised zero grade | Vaillant 1958: 554–556: PIE *‑u̯ ōs was
replaced with PBS *‑us (neut. nom.–acc. sg. ending and oblique stem)
> CS *‑ъ | Bräuer 1969b: – | Arumaa 1985: 106, 108: Slavic masc.
‑ъ represents generalised zero grade *‑us; Slavic neut. ‑ъ is zero grade
*‑us | Igartua 2005a: – | Aitzetmüller 1978/1991: 241: OCS ‑ъ has
generalised zero-grade form *‑us from fem.
PS At a pre-stage of Slavic the zero-grade ending *‑us > PS *‑u [17] (CS *‑ъ
[29]) was also introduced in the masculine form, leading to a merger of the
masculine and neuter forms in Proto-Slavic. The ending was retained in the
old Slavic dialects.
96 Chapter 3
PIE PIE *‑os, consisting of the o-grade of the suffix and a zero ending, is pre-
served directly in Greek and Latin. In Indo-Iranian the different outcomes of
the velar consonant in OAv. nom.–acc. aogō vs. instr. sg. aojaŋhā also point
to an o-vowel in the nominative-accusative singular (see e.g. Martínez & de
Vaan 2014: 29). In Gothic, PGmc. nom.–acc. *‑az from PIE *‑os was replaced
with ‑iz under the influence of the weak stem; the same process took place in
West Germanic, e.g. OEng. lemb from PGmc. *lambiz. ONor. røkkr from PGmc.
*rekʷaz preserves the old form, but the word is declined as an o-stem. Expected
Nominal Inflection 97
PBS PIE *‑os regularly yielded PBS *‑as [7]. The type has not been preserved
in Baltic; Li. debesìs, Lv. debess from PBS *nebas are now i-stems, although the
words preserve traces of consonant-stem inflection.
PS The expected outcome of PBS *‑as is PS *‑ə [17] (CS Cl *‑ъ || ONovg. ‑e [29];
CS *‑jь [20|29]), as in the masculine o-stem nominative singular. The actual
ending PS *‑a (CS *‑o [29]) ultimately stems from the pronominal inflection,
where it is the regular outcome of PBS *‑a < PIE *‑od [7|9] (see e.g. Fortunatov
1888: 572 n. 1; 1897: 164 n. 1; Gălăbov 1973: 8–9; Matzinger 2001: 191; Olander 2012:
322). It is likely that the pronominal ending PIE *‑od > PBS *‑a first spread to
adjectives, then to o-stem nouns and finally to s-stems.
The replacement of pre-PS *‑ə with *‑a may seem trivial at first sight, but—
as Tijmen Pronk reminds me (pers. comm., 2012)—a proportional analogy
of the type *lēt‑ : *lēta :: *nebes‑ : X would have led to X = PS **nebesa (CS
**nebeso), and a generalisation of the suffixal vowel would have led to **nebe.
Such creations are indeed attested in more recent times, e.g. Cz. koleso and
nebe. In order to understand why the neuter o-stem ending *‑a was introduced
in the s-stems one should bear in mind the special status of the consonant-
stem nominative(–accusative) singular in Proto-Slavic. Unlike the other forms
of the paradigm (e.g. gen. sg. *slau̯ ese), this form was not analysable in terms of
a stem (*slau̯ es‑) plus an ending (*‑e). In my view it was the isolated status of
the form that cleared the way for the introduction from the o-stems of the most
common neuter ending *‑a, possibly by the process referred to as “syntagmatic
assimilation” by Andersen (1980: 16–18 and pers. comm., 2013).
PBS PIE *‑is yielded PBS *ˌ‑is [4]. The ending is preserved in the Baltic lan-
guages; the final accentuation in Lithuanian is analogical from the ā‑, ē‑ and
consonant stems (Olander loc. cit.).
PS PBS *ˌ‑is is reflected as PS *ˌ‑i [12|17] (CS *‑ь [29]), preserved in the old
Slavic dialects.
PIE The original ending PIE *‑us, consisting of the zero grade of the u-stem
suffix *‑u‑ followed by the nominative singular marker *‑s, is preserved in all
old Indo-European languages.
PBS PIE *‑us is retained as PBS *ˌ‑us [4], preserved in Baltic with secondary
non-initial accentuation in Lithuanian by analogy with the nominative singu-
lar forms of other stems (Olander loc. cit.).
PS The regular reflex of PBS *ˌ‑us is PS *ˌ‑u [12|17] (CS *‑ъ [29]), preserved in
the old Slavic dialects.
PBS PIE *‑ah₂ regularly yielded PBS *‑ˈā̰ [1]. In Lithuanian the ending was
shortened by Leskien’s law; the original long vowel is seen in the definite adjec-
tive mažóji. In Latvian the short ending is a result of the general shortening of
final long vowels.
PS The regular reflex of PBS *‑ˈā̰ is PS *‑ˈā [13] (CS *‑a [29]; *‑ja [20|27|29]),
preserved in the old Slavic dialects.
Nominal Inflection 101
PIE Brugmann 1909–1911: 124–125, 139–140: PIE acute *‑ī�;́ Gk. ‑ια is analogical
based on acc. sg. ‑ιαν < PIE *‑(i)i̯m̥ | Beekes 1995/2011: 201, 204: OCS
ladii is from PIE *‑ih; OCS ‑ǫšti is from *‑ih₂ | Rasmussen forthc. a
§ 5: PIE *‑ih₂ | Debrunner & Wackernagel 1930/1975: 167 | Rix
1976/1992: 131: Gk. ‑ια reflects PIE *‑ih₂‑∅ | Sihler 1995: 275–278:
PIE *‑ih₂ | Weiss 2009/2011: 305–306 | Krahe 1942/1967: 23–25:
Go. bandi is from PIE *‑i̯ə or *‑ī | Boutkan 1995: 231–235: PGmc. *‑ja
reflects *‑i̯h̥ from PIE *‑ih₂ | Kloekhorst 2008a: –
PBS Olander 2009: 167 | Endzelīns 1971: 142 | Stang 1966: 197 |
Otrębski 1956: 31 | Endzelīns 1923: 396
PS Vondrák 1908/1928: 2 (1906/1924: 513): PIE *‑ī | Hujer 1910: 6–9: PIE
́
acute *‑ī� | Meillet 1924/1934: 400 | Vaillant 1958: 96–99 | Bräuer
1969a: 139–140 | Arumaa 1985: 80–86, 146–148 | Igartua 2005a:
194–196 | Aitzetmüller 1978/1991: 89
PS The regularly reflex of PBS *‑ˈī�̰ is PS *‑ī [13] (CS *‑i [29]). Correspondences
such as OCS prs. act. ptc. fem. nom. sg. vedǫšti, ORu. nesuči and OCz. nesúci
show that *‑ī was replaced with *‑ˈi̯ī under the influence of the remaning
forms of the paradigm, which was inflected as a *i̯ā-stem paradigm. Since all
Slavic languages point to *‑i̯ī (CS *‑i), I assume the replacement took place in
pre-Proto-Slavic.
‑ъ (e.g. o-stem masc. nom. sg. vlьkъ); the former reflex is found in cases
where the pronunciation was “relativement nette et lente, [. . .] dans les
mots principaux de la phrase”, whereas the latter reflex is found in cases
where the pronunciation was “trouble et rapide, [. . .] dans des mots plus
ou moins accessoires”; one variant or the other was generalised in dif-
ferent forms of the paradigm | Vaillant 1958: 29 (1950: 210): OCS ‑ъ
is either analogical based on the acc. sg., imitating the pattern of i- and
u-stems, which had identical nom. and acc. sg.; or from PIE *‑os “avec
une réduction spéciale de la finale”, via PBS *‑as > Slavic *‑əs > *‑ъ; the
merger of the reflexes of *ə and *u is later than beginning of second
palatalisation because *ə does not prevent palatalisation of PBS *‑ikas
to Slavic ‑ьcь | Bräuer 1969a: 20–21, 73 (1961: 103): CS *‑ъ is either
directly from *‑ъ(s) < *‑əs or *‑us < PIE *‑os, or it is due to influence from
u-stems | Arumaa 1985: 130–131: pre-Slavic *‑o is from PIE *‑os; CS *‑ъ
is u-stem nom. sg. | Igartua 2005a: 100–110: PIE *‑os yielded pre-Slavic
*‑o, substituted with CS *‑ъ from acc. sg. by analogy with identical nom.
and acc. sg. in u- and i-stems; ONovg. ‑e was introduced from i̯o-stems,
which had PIE *‑i̯os > *‑je (following Krys’ko, Vermeer); ending *‑je is per-
haps preserved in some ONovg. i̯o-stems | Aitzetmüller 1978/1991: 76,
83: OCS ‑ъ < *‑us has *u for expected *o from acc. sg. *‑um < *‑om by anal-
ogy with identical vowels in i-and u-stem nom. sg. *‑is, *‑us, acc. sg. *‑im,
*‑um; i̯o-stem ‑jь < *‑i̯is < *‑i̯us has replaced regular *‑i̯os
PBS PIE *‑os yielded PBS *ˌ‑as [4|7], preserved in the Baltic languages.
A number of scholars assume that the regular reflex of PIE *‑os is CS *‑o
(see e.g. Vermeer 1991/2009). The most important arguments in favour of this
view are constituted by certain Slavic personal names ending in ‑o, such as
ORu. Ivanъko, and the neuter s-stem nominative-accusative singular ending
PIE *‑os, which seems to have yielded CS *‑o, e.g. in OCS slovo. These forms
allegedly preserve the phonetically regular reflex of PIE *‑os.
An alternative view holds that PIE *‑os yielded CS *‑ъ. This view is mainly
based on the form under consideration here. It is argued that while it is pos-
sible to set up a formally perfect proportion leading to the replacement of the
inherited o-stem ending *‑o with the u-stem ending *‑ъ by analogy with the
accusative forms *‑ъ (from *‑om) and *‑ъ (from *‑um), it is unlikely that such
an analogy would ever be set up because of the essential role of the opposition
between the nominative and accusative singular forms in Slavic. Accordingly,
the phonetically regular ending must be *‑ъ.
Both views are challenged by the Old Novgorod ending ‑e, which clearly
cannot reflect CS *‑o, nor *‑ъ (for the characteristics of the ending see Zaliznjak
1995: 99–104; Le Feuvre 1998b: 213–284). Proponents of the view that PIE *‑os
yielded CS *‑o have assumed that ONovg. ‑e is the vocative form (e.g. Kwon
2009), or that it represents the original i̯o-stem ending *‑je from *‑jo < PIE *‑i̯os
(e.g. Vermeer 1991/2009; Krys’ko 2007: 99–114).
If it is accepted that the general reflex of PIE *‑os in Slavic is *‑ъ, however,
another possibility appears, namely that PIE *‑os had two reflexes in Slavic:
‑e in the Old Novgorod dialect and *‑ъ everywhere else (Zaliznjak, Dybo and
Nikolaev apud Zaliznjak 1988: 170; Olander 2012: 330–331). Apart from giving
a straightforward explanation of the masculine o-stem nominative singular
forms, this hypothesis also receives support from the genitive singular and
nominative plural of the ā-stems, where PIE *‑ah₂s and *‑ah₂as merge in pre-
PS *‑ās and show up as ‑ě in the Old Novgorod dialect and ‑y in the other Slavic
dialects. The development of pre-PS *‑ās to ‑ě in the Old Novgorod dialect
(merging with PS *‑ē) and to ‑y in other dialects (merging with PS *‑ū) is an
exact parallel to the situation in the masculine nominative singular, where PIE
*‑os > pre-PS *‑as [7] yields ‑e in Novgorod (merging with PS *‑e) and ‑ъ (merg-
ing with PS *‑u) elsewhere. In both cases it is likely that the loss of word-final
*‑s in pre-Proto-Slavic caused centralisation of a preceding *ā or *a to PS *ə̄ or
*ə [17].
As for the apparent counterexamples to the hypothesis that PIE *‑os is
reflected as CS Cl *‑ъ, ONovg. ‑e, the Common Slavic neuter s-stem nominative-
accusative singular *‑o is treated in § 3.3.9. The ‑o of personal names of the
type Ivanъko is slightly more difficult to explain; as I see it, ‑o is most likely to
be the neuter o-stem ending, originally used with hypocoristic function (see
Olander 2012: 322–323, with references in fn. 21).
Nominal Inflection 105
PS *ˌsuta (CS *sъ̏to; *pȍl’e) PBS pron. *ˌ‑a (noun *ˌ‑am) PIE pron. *‑od (noun *‑om)
OCS město; pol’e Li. adj. šálta Ved. pron. tát (noun yugám)
ORu. lěto; pole; ONovg. Lv. adv. maz OAv., YAv. pron. tat̰ (noun
selo; vъzgolovьje OPr. pron. sta xšaϑrəm)
OCz. město; moře (noun asseran) Gk. pron. τό (noun ζυγόν)
La. pron. istud (noun iugum)
Go. pron. →þata (noun
barn); ONor. pron. þat (noun
barn)
Hi. pron. apāt (noun pedan;
adj. dannattan, dannatta)
PIE Brugmann 1909–1911: 149–150 (1897: 391) (noun), 358 (pron.): OCS ‑o
reflects accented PIE *‑óm | Beekes 1995/ 2011: 212 (noun), 226
(pron.) | Rasmussen forthc. a § 4: Slavic ‑o, Li. ‑a reflect PIE pron.
*‑od | Debrunner & Wackernagel 1930/ 1975: 90 (noun), 496–497
(pron.) | Rix 1976/ 1992: 138 (noun), 184 (pron.) | Sihler 1995:
258 (noun), 385 (pron.) | Weiss 2009/ 2011: 200, 225 (noun), 335
(pron.) | Krahe 1942/1967: 12 (noun), 62 (pron.): PGmc. pron. was
*þat, preserved in ONor. and WGmc., from PIE *tod | Boutkan 1995:
174–175 (noun), 300 (pron.) | Kloekhorst 2008a: 104, 172 (noun), 799
(pron.)
PBS Olander 2009: 167 | Endzelīns 1971: 134 (noun), 192 (pron.), 260
(adv.) | Stang 1966: 187–188 (noun), 241–242 (pron.): Li. adj. ‑a, OCS to
reflect PIE *tod/t; OPr. sta, wissa, Li. vìsa, kìta reflect PIE *‑od/t; tataĩ pre-
serves *‑t before ptcl. ‑ai | Otrębski 1956: 13 (noun), 123–125 (pron.): Li.
‑a < PIE *‑od | Endzelīns 1923: 464–465 (adv.)
PS Vondrák 1908/1928: 2: Slavic ‑o is from pronouns | Hujer 1910: 34–36:
Slavic ‑o is from PIE pron. *‑od and neut. s-stem *‑os; Slavic ‑je is the
regular outcome of PIE *‑i̯om | Meillet 1924/1934: PIE *‑os, *‑om
yielded Slavic ‑o, ‑ъ; ‑o became neut. ending under the influence of pron.
*tot | Vaillant 1958: 43–44: Slavic ‑o is from PIE pron. *‑od | Bräuer
1969a: 63, 93: Slavic ‑o is from PIE pron. *‑od | Arumaa 1985: 131–132:
Slavic ‑o is from PIE *‑o, variant of *‑om | Igartua 2005a: 116–123:
Slavic ‑o reflects archaic PIE o-stem neut. ending *‑o; Slavic masc. ‑ъ
partly reflects recent PIE o-stem neut. ending *‑om (dvorъ type) |
106 Chapter 3
Aitzetmüller 1978/1991: 84–85: OCS ‑o, ‑je are taken from PIE pron. *‑od, to
a lesser extent from neut. s-stem *‑os
PBS The Proto-Indo-European noun ending *‑om > PBS *ˌ‑am [4|7] is pre-
served in Old Prussian, especially in the Elbing vocabulary, e.g. asseran, dalp-
tan. In East Baltic the neuter gender has been lost, but Lithuanian predicative
adjectives, e.g. šálta, and pronouns like kìta, vìsa preserve the PBS pronominal
ending *ˌ‑a < PIE *‑od [4|5|7|9]. A useful discussion of the development of the
neuter o-stems in Baltic and Slavic is found in Derksen 2011. Petit 2010: 140–204
contains an overview of the fate of the neuter gender in the Baltic languages.
In my opinion, the fact that neuters with root-accent often turn up as mas-
culine words in Slavic does not show that accented and unaccented *‑om had
different reflexes in Slavic. It may simply be a tendency that evolved on the
basis of formal and/or semantic similarities between masculine nouns and
existing neuters that happened to be root-accented. There are alternative
sources for the attested differences in gender, such as independent themati-
sations of athematic nouns in the individual branches of Indo-European, or
simply occasional changes in gender triggered by the identical masculine and
neuter endings in all forms except the nominative singular and the nominative
and accusative dual and plural.
OCS čьto, ničьže Li. adv. daũg; OLi. adv. daugi Ved. →kím; →idám; ptcl. ‑cit
ORu. čьto; ONovg. čьto Lv. adv. daũdz OAv. īt̰; YAv. cit̰
OCz. ‑č OPr. – Gk. τί
La. quid
Go. →ita, OEng. hit
Hi. kuit (OS)
PIE Formally, the Slavic form may correspond to Ved. interr. pron. kím, to the
neuter noun i-stem ending PIE *‑i and to the ending *‑id found in the inter-
rogative pronoun in most ancient Indo-European languages except Vedic. It is
most likely that it is the latter form that is continued in Slavic. PIE *‑id contains
the pronominal neuter nominative-accusative singular marker *‑d.
PBS PIE *‑id yielded PBS *ˌ‑i [4|5|9]. The ending is probably preserved in
Baltic in the adverb OLi. daugi ‘much’, Li. daũg ‘much’, Lv. daũdz ‘much’. It is
also possible that the Lithuanian interjection dial. šìtai ‘look!’ preserves the
ending including the final stop (cf. Stang loc. cit.).
PS PBS *ˌ‑i is only preserved in the inanimate form of the interrogative pro-
noun, PS *ˌki (CS *čь̏ [29]) (for the lack of a gender distinction in this pronoun
see H. Birnbaum & Schaeken 1997: 75–76). The pronoun was extended with
an article to CS *čь̏ ‑to, preserved in Old Church Slavonic and East Slavic. The
unextended version is seen in e.g. OCz. pro‑č, where it is preceded by a preposi-
tion (Gebauer 1896: 464–465).
PIE Brugmann 1909–1911: 382, 409–410: Slavic azъ has perhaps arisen through
contraction of conjunction a and pron. *ezъ | Beekes 1995/2011: 232:
PIE *h₁eg̑ , with *ō or *‑e/oh added in some languages, *‑(h)om in others; Hi.
u-vocalism is from acc.–dat. ammuk; OCS azъ for *jazъ reflects *ēg(h)om
with Winter’s lengthening (“Winter–Kortlandt’s law”) | Rasmussen
forthc. a § 13: PIE *ég̑ ; Hi. ūk has u from *tu; PBS *eź; Slavic azъ per-
haps consists of conjunction a plus pron. *ez plus paragogic vowel
‑ъ | Schmidt 1978: 21–47, 109: PIE *eg̑ , *eg̑ hom, *eg̑ hō(m), perhaps also
Nominal Inflection 109
*eg̑ h; Go. ik, ONor. ek, OEn. ic may reflect PIE *eg̑ hom or *eg̑ ; Runic ek is not
from PIE *eg̑ hom, perhaps from *eg̑ ; Hi. uk has u from 2ps.; Slavic jazъ, ja
are from PIE *eg̑ (h)om, *eg̑ ; length in Slavic ja < *( j)ē is analogical after ty
< PIE *tū | Debrunner & Wackernagel 1930/1975: 453–455: Ved. ahám
goes back to PIE | Rix 1976/1992: PIE *ég̑oh₂ is reflected in Gk. ἐγώ, La.
egō̆; PIE *eg̑ h₂‑óm is seen in Gk. Hom., Dor. ἐγών (with ō from ἐγώ), Ved.
ahám etc. | Sihler 1995: 369–372, 375: PIE *eg̑oh, but reconstruction is
difficult | Weiss 2009/2011: 326: OCS azъ, IIr. and Gmc. forms are from
PIE *h₁eg̑ h₂óm; other forms point to *h₁ég̑oh₂, *h₁ég̑ h₂; Hi. has u from 2sg.
*tu; OLi. eš, Lv. es are problematic | Krahe 1942/1967: 50: Baltic forms
are from PIE *eg̑ ; IIr. and Gmc. forms go back to PIE *eg̑ (ʰ)om; Gk. and La.
forms are from PIE *eg̑ō | Seebold 1984: 20–24: early East Nordic ek/ik
reflects PIE simple *eg̑ (ʰ); early East Nordic ‑(e)ka is from PIE conjunctive
*eg̑ (ʰ)om; OHG ihha perhaps reflect PIE emphatic *eg̑ō | Kloekhorst
2008a: 111–115, 912: PIE *h₁eg̑ h; PAnat. *ʔúǵ has *u from oblique forms, not
from ‘you’
PBS Kapović 2006a: 16–37, 159: CS *ja̋ goes back to PIE *ég̑ ; CS *jázъ is from PIE
*eg̑ hóm; Li. àš, OLi. eš represent PIE sandhi variant *ék̑ | Endzelīns
1971: 186: Common Baltic form was probably *ež, but Li. dial. õš may have
different origin | Stang 1966: 247: Baltic forms point to *ež, possibly
originally followed by one more syllable | Otrębski 1956: 135: Li. àš,
OLi. eš go back to *ež | Endzelīns 1923: 372
PS Vondrák 1908/1928: 70: Slavic jazъ perhaps reflects *i̯ēzъ with compensa-
tory lengthening; loss of *j‑ in OCS may be due to sentence-initial posi-
tion | Meillet 1924/1934: 84, 105, 452: Slavic form is “énigmatique”,
seems to reflect *a‑ > ja‑; forms without ‑z(ъ) are secondary; ‑ъ may cor-
respond to Ved. ‑ám | Vaillant 1958: 443 (1950: 183): PBS *ež reflects
*eg; Slavic *( j)azъ contains paragogic vowel; Slavic long vowel may
have arisen in variant where *‑ž was lost; or Slavic ja‑ is perhaps from
*ē‑ with “allongement secondaire expressif de *e‑ [. . .] et sur le modèle
de ty” | Arumaa 1985: 158–159: PIE *eg̑; it is possible that both *( j)
azъ and *( j)az existed in PS | Aitzetmüller 1978/1991: 107–108: OCS
azъ (with paragogic vowel and loss of *‑i̯) and ja found in other Slavic
languages reflect *i̯āz, corresponding, apart from quantity, to Baltic *ež
PIE The word for ‘I’ was *h₁ég̑ (or *ég̑ ) in the proto-language (for the question
of an initial *h₁ see Katz 1998: 22–23; Kapović 2006a: 159). The original form is
presumably preserved in the Old Avestan hapax as-cīt̰ (Mayrhofer 1986–2001,
1: 155), in Baltic and perhaps in Germanic. The remaining languages, including
Slavic, show *h₁ég̑ followed by some additional element. The Indo-Iranian
forms probably reflect *h₁eg̑‑hóm, where *hom (or *h₁em, as in La. idem; cf.
110 Chapter 3
Meiser 1998/2006: 161; Weiss 2009/2011: 342) is the emphasising particle also
found in other pronominal forms, e.g. Ved. 1sg. pron. nom. tuvám, dem. pron.
masc. nom. sg. ayám, neut. nom.–acc. sg. idám, 1sg. pron. dat. máhyam, 2sg.
túbhyam. Examples such as PIE neut. nom.–acc. sg. *még̑ h₂ > Ved. máhi, Gk.
μέγα show that *g̑ h merged with *g̑ ʰ in Indic, yielding Vedic h, at least in the
case of *h₂ (cf. Hoenigswald 1965: 95). This laryngealist solution seems more
economical to me than the widespread assumption of influence from the
dative of the first-person singular pronoun máhyam (e.g. Meillet 1894: 282;
cf. the discussions in Mayrhofer 1986–2001, 1: 155 and Schmidt 1978: 21–22).
The *‑ō found in Greek and Italic probably represents the ending of the first-
person singular of thematic verbs, which was added to the pronoun in the two
branches; it is less likely that it reflects a deictic element similar in function
to *hom. Hom. ἐγών (used before vowels) may contain the particle *hom and
have secondary ω from ἐγώ (Rix loc. cit.). In Germanic, Early Runic tonic ek may
preserve PIE *h₁ég̑ (thus e.g. Seebold loc. cit.; cf. § 3.3.15), whereas encl. ‑eka
points to PIE *h₁ég̑ plus *hom (similarly e.g. Kroonen 2013: 116, on the evidence
of the breaking in East Nordic, e.g. Old Swedish iak). Due to the reduction of
final syllables in Germanic, Go. ik, ONor. ek, OEng. ic etc. are ambiguous (see
e.g. Ernout & Meillet 1932/1951: 343; Schmidt 1978: 24–25; Sihler 1995: 369). The
u-vocalism of Anatolian was probably imported from the nominative of the
second-person singular pronoun; the accusative-dative form ammuk may also
have played a role.
PBS PIE *h₁ég̑ regularly yielded PBS *ˈeś [5|8], preserved in the Baltic lan-
guages. As shown by the Slavic evidence, Proto-Balto-Slavic must also have
known a variant *ˈḛ̄źam from *h₁eg̑ hóm [1|6|7|8], consisting of PIE *h₁ég̑ and
the particle *hom that was also appended to this pronoun in Indo-Iranian and
Germanic. Much of the discussion of the Slavic form has concerned the length
of the initial vowel, but this problem was solved with the discovery of Winter’s
law, as seen by Winter himself (1978: 433; see also Kortlandt 1997b/2009: 29;
Kapović 2009). The existence of a monosyllabic form without lengthening in
Baltic and a disyllabic form with lengthening in Slavic suggests that the devoic-
ing of word-final obstruents [5] took place before Winter’s law [6] in pre-Proto-
Balto-Slavic. The Slavic form with lengthening also suggests that PIE *Dh and
*Dʰ did not merge in Balto-Slavic—unless, of course, the univerbation of *h₁ég̑
and *hom happened after the loss of laryngeals [1], but earlier than the devoic-
ing of word-final *g̑ [5] seen in Baltic, which is likely anyway.
PS In my view the best diachronic explanation of the attested Slavic forms
assumes that PBS *ˈḛ̄źam, consisting of the same elements as the Indo-Iranian
forms and Early Runic encl. ‑(e)ka, yielded PS *ˈēzu [13|19]; the older variant
Nominal Inflection 111
PBS *ˈeś < PIE *ég̑ disappeared in Slavic. With an automatic prothetic *i̯ PS
*ˈēzu became *ˈi̯ēzu (CS *ja̋ zъ [27|29]). The reflexes of CS *ja‑ found in most
Slavic dialects are the regular outcome of PS *ē‑, but the a‑ of Old Church
Slavonic (and Bulgarian) is unexpected. Instead of assuming a PIE form with
o-grade (Ernout & Meillet 1932/1951: 343; Meillet 1903/1973: 333: Slavic forms
require PIE *ō [or *ā]; Andersen 1996: 148–149; cf. also Sobolevskij 1908: 204:
OCS azъ cannot contain *ē‑), unattested outside Slavic, I am inclined to agree
with the authors who assume that post-PS *ˈi̯ēzu > *ˈi̯āzu [27] became *ˈāzu >
OCS azъ by a sporadic loss of *i̯ in certain East South Slavic dialects (Kapović
2006a: 24–27; see also H. Birnbaum & Schaeken 1997: 73).
The Common Slavic form *já, widely reflected in the Slavic dialects, is, in
my opinion, most economically seen as a shortened version of PS *ˈēzu (CS
*ja̋ zъ), perhaps under the influence of the monosyllabicity of the nominative
form of all other personal pronouns, viz. PS 2sg. *ˌtū, 1du. *u̯ ē, 2du. *u̯ ū, 1pl.
*ˌmū, 2pl. *ˌu̯ ū (CS *ty̑ , *vě, *vy, *my̑ , *vy̑ ). The rough correlation of reflexes of
a disyllabic form with acute accentuation (e.g. Sln. jàz) on the one hand and
reflexes of a monosyllabic form with neoacute accentuation (e.g. Slnc. jǻu̯) on
the other indicates that the reflexes of neoacute accentuation found in various
Slavic dialects may be due to this shortening (cf. Kortlandt 2009/2009: 108; and,
against this view, Kapović 2006a: 34–37, with an overview of the forms; 2009).
If the acute accentuation is old in Slavic, this points to root-accented PBS
*ˈḛ̄źam. The alternative view, namely that the Slavic neoacute is old, would
mean that the word had final accentuation in Proto-Balto-Slavic, i.e. *ḛ̄ˈźam.
The accentual implications for the Proto-Indo-European reconstruction are
unclear as we do not know at which stage the univerbation of *ég̑ and *hom
took place (see above), and which prosodic effects it had. All we can say is that
if the form with neoacute is old (which it is probably not), the form *ḛ̄ˈźam
with final accent must have arisen after the mobility law [4]; otherwise the
form would have become unaccented.
84; Katz 1998: 23 n. 35; a different explanation of the Anatolian form as innova-
tory is given in Rieken 2009: 40–41 with further references).
PBS In Balto-Slavic PIE *túh yielded *ˈtṵ̄ [1], preserved in Slavic and in OPr.
toū. Taken at face value the Old Prussian variant tu points to a Proto-Balto-
Slavic short variant, but we may also be dealing with a later syntactically con-
ditioned shortening (Endzelīns 1944: 129–130 suggests that the omission of
the macron is a graphical matter; Stang’s formulation [loc. cit.] also seems to
indicate that he assumes the Old Prussian forms reflect an original long form).
Short variants are found in Latvian tu and in NWŽem. tọ̀ (Zinkevičius 1966:
298; Stang loc. cit.); standard Li. tù is ambiguous as PBS *tṵ̄ would have been
shortened by Leskien’s law.
PS The long variant PBS *ˈtṵ̄ yielded PS *ˌtū [13] (CS *ty̑ [29]), preserved in all
Slavic languages.
PIE The ending was the pure case–number marker, *‑m̥ , the reflexes of
which are preserved in most Indo-European languages. PIIr. *‑am is usually
thought to contain a nasal analogically transferred from the vowel stems (e.g.
Wackernagel loc. cit.), but some authors assume that it is the regular outcome
of PIE *‑m̥ (Sihler loc. cit.). The latter view allows us to view both this form and
the secondary athematic ending of the first-person singular of the verbs, Ved.
‑am and OAv. ‑əm from PIE *‑m̥ , as phonetically regular. However, the cardi-
nal numbers PIIr. *saptá ‘seven’ (Ved. saptá, YAv. hapta), from PIE *septḿ̥, and
*dáća ‘ten’ (Ved. dáśa, OAv. dasā, YAv. dasa), from *dék̑m̥, contradict this view;
and while *dáća may perhaps be remade from *dáćat from PIE *dék̑m̥t (Sihler
1995: 416–417), this explanation does not work for *saptá (not discussed in this
context by Sihler).
Hi. ‑an is probably the o-stem ending, as PIE *‑m̥ would have yielded *‑un
(Melchert 1994: 181; Kloekhorst loc. cit.).
PS PBS *ˌ‑im yielded PS *ˌ‑i [19] (CS *‑ь [29]). The ending is preserved in the
old Slavic dialects.
PIE PIE *‑im, consisting of the stem suffix *‑i‑ and the accusative singular
marker *‑m, is preserved in the old Indo-European languages. For the pho-
notactically unexpected syllabification of the sequence *‑im see § 2.2.1. It is
most likely that La. ‑im reflects the Proto-Indo-European i-stem ending *‑im,
whereas ‑em originates in the consonant-stem ending *‑m̥ (cf., however, Sihler
loc. cit.).
PBS PIE *‑im is preserved as PBS *ˌ‑im [4], with the expected reflexes in the
Baltic languages.
PS PBS *ˌ‑im yields PS *ˌ‑i [19] (CS *‑ь [29]), preserved in the old Slavic
dialects.
PIE The ending contained the zero grade of the stem suffix, *‑u‑, followed
by the accusative singular marker *‑m (for the vocalisation of *‑um see § 2.2.1).
The ending is preserved in the old Indo-European languages. The form was
also used for deverbal formations, e.g. Ved. inf. ‑tum and La. supine ‑tum.
PBS PIE *‑um became PBS *ˌ‑um [4], preserved in the Baltic languages.
Outside the u-stem paradigm the form is preserved in the East Baltic and Slavic
supine, e.g. Li. vèstų, Lv. dial. ȩ̂stu, PS *ˌnestu (CS *nȅstъ), and in Old Prussian
infinitives, e.g. boūton.
PS PBS *ˌ‑um regularly yielded PS *ˌ‑u [19] (CS *‑ъ [29]), preserved in the old
Slavic dialects.
not *‑eh₂m̥ , for which there is no evidence in any IE language; La. ‑am was
regularly shortened from PItal. *‑ām with secondarily reintroduced long
vowel | Weiss 2009/2011: 229, 232–233: pre-PIE *‑eh₂m yielded PIE
*‑ām, not *‑eh₂m̥ ; shortening in La. ‑am is regular | Krahe 1942/1967:
21: PIE *‑ām | Boutkan 1995: 226: PIE *‑ehm | Kloekhorst 2008a:
107: Hi. ḫ-stem arḫan has *‑om, remade from PIE *‑éh₂m
PBS Olander 2009: 169: BS unaccented form points to PIE disyllabic *‑ah₂m̥ ;
Gk. acute tone points to sandhi variant *‑ah₂m | Endzelīns 1971:
143–144: Baltic forms reflect ‑an < *‑ān (also in OCS ‑ǫ) from PIE *‑ām;
PIE acute tone, preserved in Gk., was replaced with circumflex tone in
Baltic and Slavic, probably by analogy with o‑, i- and u-stems; shorten-
ing of *‑ān to *‑an in Baltic is shown by OPr. rānkan | Stang 1966: 39,
199: Baltic tone does not correspond to Gk. tone, perhaps due to different
development of *‑VhN sequences; in OPr., stems ending in a velar show
rounding, pointing to *‑ān, in Cat. I and II; in Cat. III the stem has been
restored | Otrębski 1956: 25: Li. ‑ą < *‑an reflects *‑ān | Endzelīns
1923: 305–306
PS Vondrák 1908/1928: 3 (1906/1924: 150): Slavic and Baltic endings do not
match Gk. ending [with respect to tone]; Slavic ‑ǫ is from originally
acute *‑ām, with circumflex tone from i- and u-stems | Hujer 1910:
92–94 | Meillet 1924/ 1934: 397–398 | Vaillant 1958: 80, 87: BS
and Gk. tones do not match; i̯ā-stem ending ‑jǫ is analogical based on
ā-stems | Bräuer 1969a: 104, 127: CS *‑ǫ reflects PIE *‑ām; CS *‑jǫ
reflects PIE *‑i̯ām, alongside *‑i̯əm, *‑īm | Arumaa 1985: 148: PIE
acute *‑ām; BS circumflex is perhaps analogical from other vocalic
stems | Igartua 2005a: 197–199: CS *‑ǫ < PIE *‑ām < *‑eh₂m; CS *‑jǫ
is analogical for *‑ję | Aitzetmüller 1978/1991: 87, 90: OCS ‑ǫ reflects
*‑ām; i̯ā-stem ending ‑jǫ has replaced regular *‑ję by analogy with hard
stems
PIE PIE *‑ah₂m̥ consists of the ā-stem suffix *‑ah₂‑ and the accusative singu-
lar marker *‑m. Balto-Slavic points to a disyllabic ending *‑ah₂m̥ , first because
the ending does not trigger Saussure’s and Leskien’s laws in Lithuanian, second
because the form became unaccented in pre-Proto-Balto-Slavic by the mobil-
ity law [4]. By contrast, the Greek acute tone points to a monosyllabic ending
PIE *‑ah₂m. The most likely explanation of this discrepancy is that Greek has
generalised the antevocalic sandhi variant. In Latin ‑am the vowel was short,
as in all other word-final sequences of a vowel plus *‑m (Allen 1965/1978: 74).
PBS PIE *‑ah₂m̥ is reflected as PBS *ˌ‑ām [1|2|3|4], with the expected reflexes
in East Baltic. The normal ending in Old Prussian is ‑an. In the Third Catechism
118 Chapter 3
this ending is also found after velars, e.g. rānkan, which seems to point to short
a since long ā is rounded after velars in Old Prussian (Hill 2013: 170–171). In the
First and Second Catechisms, however, we find <‑uan>, <‑wan>, <‑un>, <‑on>
after velars, which are more likely to reflect long PBalt. *‑ān than short *‑an.
Forms like rānkan have been normalised under the influence of other forms
of the paradigm (Stang 1930/1970: 121–124; loc. cit.; Kortlandt 1988/2009: 192).
PS PBS *ˌ‑ām yielded PS *ˌ‑ān [19] (CS *‑ǫ [29]; *‑jǫ [20|27|28]), preserved in
the old Slavic dialects. Since I consider the shortening of long diphthongs to be
a relatively late process in Slavic, I consider CS *‑jǫ to be the regular correspon-
dence of PS *‑i̯ān, whereas PS *‑i̯an became *‑ję [20|28] (cf., however, Vaillant
loc. cit.; Aitzetmüller loc. cit.; Igartua loc. cit.).
PIE Brugmann 1909–1911: 138–139 (1897: 391): OCS ‑ъ is from PIE unac-
cented *‑om | Beekes 1995/2011: 212: Slavic ‑ъ reflects PBS *‑um < PIE
*‑om; expected Baltic *‑ų has been replaced with ‑ą | Rasmussen
forthc. a § 4 | Debrunner & Wackernagel 1930/1975: 90: PIE *‑om
or *‑on | Rix 1976/ 1992: 138 | Sihler 1995: 258 | Weiss
2009/2011: 201, 221 | Krahe 1942/ 1967: 10 | Boutkan 1995: 174–
175 | Kloekhorst 2008a: 104
PBS Olander 2009: 168–169 | Endzelīns 1971: 134: Li. ‑ą, Lv. ‑u, OPr. ‑an
reflect PIE *‑om | Stang 1966: 182: Li. ‑ą, Lv. ‑u, OPr. ‑an reflect PBalt.
*‑an | Otrębski 1956: 14: Li. ‑ą, OPr. ‑an go back to *‑om | Endzelīns
1923: 292, 387: Lv. noun ending ‑u is from *‑uo, corresponding to Li. ‑ą,
OPr. ‑an, Slavic ‑ъ etc.; Lv. pron. masc.–fem. tùo corresponds to Li. tą̃
PS Vondrák 1908/1928: 3: OCS ‑ъ, Li. ‑ą reflect PIE *‑om | Hujer 1910: 92:
Slavic ‑ъ, Li. ‑ą reflect PIE *‑om | Meillet 1924/1934: 404–405: Slavic ‑ъ is
Nominal Inflection 119
PIE The ending *‑om consisted of the thematic vowel *o, as expected before
a voiced segment (§ 3.2.5), and the accusative marker *‑m. The ending is
preserved with the regular phonetic changes in the ancient Indo-European
languages.
PBS It is generally assumed, correctly I believe, that PIE *‑om yielded PBS
*ˌ‑am [4|7], which was preserved in the Baltic languages with the regular pho-
netic changes (cf. [19] for an alternative view).
PS All Slavic dialects point to an ending *ˌ‑u (CS *‑ъ [29]; *‑jь [20|29]), the
expected reflex of PBS *ˌ‑am [19]. For a discussion of the chronology of
the raising of PIE *o to *u before a word-final nasal see [19]. For the view
that the regular reflexes of PIE *‑om, *‑i̯om are CS *‑ǫ, *‑ję, which were replaced
with the (i̯)u-stem endings *‑ъ, *‑jь (Orr 2000: 134–137; for the i̯o-stem ending
also Igartua loc. cit.), see § 1.4.1.
OCS tę, →tebe Li. →tavè; dial. →tavì, Ved. →tvā́m; encl. →tvā
ORu. →tebe; encl. t’a; ONovg. →tavę̃ OAv., YAv. →ϑβąm; encl. OAv.,
(late) →tebe; encl. t’a Lv. →tevi YAv. →ϑβā
OCz. tě; OPo. cię, →cie OPr. tien; encl. tin Gk. (Att., Ion.) σέ; Dor. τέ; encl.
Att. →σε
La. →tē; OLa. →tēd
Go. →þuk, ONor. →þik
Hi. acc.–dat. →tuk (OS); encl.
→‑ta (OS)
PIE Brugmann 1909–1911: 383, 413–415: OCS mę, tę, sę reflect PIE *‑ēm, along-
side *‑ē, *‑e | Beekes 1995/2011: 232–234: PIE 1sg. *h₁mé, *h₁me; Go.
mik, Hi. ammuk (‑u‑ from 2sg. acc. sg. tuk) have ‑k from nom. sg.; Gk. 2sg.
σέ is from PIE *tué | Rasmussen forthc. a § 13: PIE tonic *mé (*mḗ?),
*t(u̯ )é (*t(u̯ )ḗ?), *s(u̯ )é, encl. *me, *t(u̯ )e, *s(u̯ )e; PBS *mēn, *tēn, *sēn have
been normalised from PIE *mē, *tē, *sē; PGmc. *mek has *‑k from nom.
*‑ek; PGmc. *þek is analogical after *mek; /‑ug/ of Hi. ammuk is perhaps
from 1sg. nom. and 2sg. acc. | Schmidt 1978: 48–64, 109–112, 120–127,
143–152: PIE 1sg. *(e)me, *mē; OCS mę, OPr. mien reflect *mēm, consist-
ing of *mē plus acc. marker *‑m; PIE 2sg. *tu is represented by Gk. (Dor.,
Nominal Inflection 121
Phocian) nom.–acc. τύ, Go. þuk, Hi. tuk, OIr. suffixed pron. ‑t; encl. form
was *te; *tu̯ e, *tu̯ ē, *te are post-PIE forms; OCS tę, OPr. tien reflect *tu̯ ē
(with analogical loss of *u̯ ) plus acc. marker *‑m; PIE refl. *ti is preserved
in Anat. | Debrunner & Wackernagel 1930/1975: 457–458, 475–476:
Ved. mā́m, tvā́m, corresponding to OAv., YAv. ϑβąm and OCS mę, tę,
perhaps consist of form without final nasal plus pron. ‑am; Ved. encl.
mā, tvā may reflect originally tonic forms | Rix 1976/1992: 177–178:
PIE tonic *mḗ, *tḗ or *méh₁, *téh₁, encl. *me, *te; Gk. ἐμέ (with *e‑ from
nom.), σέ (from *tu̯ ‑ with *u̯ from nom. *tuh₂) reflect PIE encl. forms *me,
*te | Sihler 1995: 372–374, 378–379: PIE tonic *mé, *tu̯ é, *su̯ é, encl. *me,
*te, *se; Ved mā́m, tvā́m are from *ma‑ám, *tva‑ám; La. mē, tē, sē < mēd,
tēd, sēd perhaps contain some ptcl. | Weiss 2009/2011: 326–328, 333:
PIE tonic *mé, encl. *me; initial vowel of Gk. ἐμέ, Hi. ammuk may stem
from nom. sg.; OLa. mēd is from *me plus *(‑)Vd, probably a ptcl.; IIr. and
BS have added the acc. marker of nouns; 2sg. forms pointing to *tu̯ e may
have *u̯ from nom. | Krahe 1942/1967: 51–54: Gmc. forms reflect PIE
*me, *te with ‑k from nom. | Seebold 1984: 34–37, 57–58: Gmc. *mik,
*þik go back to PIE *me‑g̑ (e), *te‑g̑ (e); *g̑e is also seen in Gk. ἐμέγε, Hi.
ammuk | Kloekhorst 2008a: 111–115, 1019: non-Anat. IE languages have
*h₁mé, *mē(?), *tué, *tuē(?) from PIE obl. *h₁men‑/*‑h₁mn‑, *teu‑/*tu‑; Hi.
ammuk reflects PIE *h₁mn‑ plus *u from acc. *tu‑; Hi. ammuk, tuk have
been extended by an element *‑ge or by *‑g̑ from 1sg. nom.; Hi. refl. ptcl.
‑z reflects PAnat. *‑ti
PBS Kapović 2006a: 81–82, 94–96, 113, 115, 149, 158, 160–161: CS *mę̑, *tę̑, Li.
manè, tavè, OPr. mien, tien point to PBS *mḗn, *tḗn; OPo. mie, cie are not
archaisms but reflect CS *mę, *tę with denasalisation by analogy with
gen.; Li. manè reflects *manḗn, a contamination of gen. (used as acc.)
*mane or stem *man‑ and acc. *mḗn; PIE had tonic *mé, *tu̯ é and, with
monosyllabic lengthening, *mḗ, *tu̯ ḗ; PBS encl. forms were perhaps *mei̯,
*tei̯, replacing PIE *mē̆, *tē̆ | Endzelīns 1971: 185–187: Li. manè, tavè,
savè reflect PIE *mene, *teu̯ e, *seu̯ e, with a from dat.; Li. manę̃, tavę̃, savę̃,
Lv. mani, tevi, sevi have been extended by acc. ending *‑n; OPr. mien, tien
or tin, sien or ‑sin (to be read mīn, tīn, sīn) and OCS mę, tę, sę reflect *mēn,
*tēn, *sēn | Stang 1966: 248–250: OPr. mien, tien, sien reflect *mēn, tēn,
*sēn, also in OCS mę, tę, sę; the forms have perhaps been extended by acc.
marker *‑m; Li. manè, tavè, Lv. mani, tevi perhaps reflect ‑ę̃ with irregular
shortening | Otrębski 1956: 137: Li. manè, tavè, savè are old gen. forms;
OLi. acc. manę, tawę́, sawę́ reflect *‑en, consisting of gen. *‑e plus acc.
*‑n; Li. dial. manì, tavì, savì, Lv. mani, tevi, sevi are perhaps from *manį
etc. with i-stem ending | Endzelīns 1923: 373, 376: Lv. mani, tevi, sevi
122 Chapter 3
reflect either *‑en (gen. *mene plus acc. ending *‑n), *‑ēn (contamination
of e.g. gen. *teve and *tēn, also in OCS tę, probably in OPr. tien), or *‑in
(from i-stems)
PS Vondrák 1908/1928: 71–72: OCS mę, tę reflect *mēm, *tu̯ ēm; OPo. mie, cie,
sie go back to mię, cię, się | Meillet 1924/1934: 453: CS *mę etc. (origi-
nally tonic) correspond to Ved. mā́m etc.; CS *me etc. (originally atonic),
reflected in OPo., Kash., Slnc. and Polab., correspond to Gk. με etc.; nasal of
*mę etc. is an old ptcl. | Vaillant 1958: 444–445: OPo. mie, cie, sie may
be analogical from tonic acc.–gen. mnie etc.; PIE had tonic and atonic
*me, *t(u̯ )e, *s(u̯ )e; the tonic variants tended to be lengthened to *mē etc.
and to be extended by the acc. marker, as in tonic PBS *mēn etc. from
*mēm etc. | Arumaa 1985: 159–160, 163, 165: OCS mę, tę, sę from *mēm,
*tēm, *sēm probably reflect PIE *mē, *tu̯ ē, *sē (with variants in *‑e) plus
acc. ending *‑m from nouns and non-pers. pronouns | Aitzetmüller
1978/1991: 108, 111, 113: Li. manè consists of stem man‑ plus the same end-
ing as OPr. mien; OCS mę, tę, sę reflect *mēm, *tēm, *sēm, consisting of
encl. *mē, *tē, *sē plus acc. *‑m; loss of nasal in OPo., Po. dial. mie is due to
analogy with mnie
PBS Old Prussian and Slavic point to PBS *mēm, *tēm, *sēm. Since the latter
two forms do not contain *u̯ , the Balto-Slavic forms seem to be based on the
original enclitic forms PIE *me, *te, *se. The long vowel, shown by the diph-
thongisation in Old Prussian and by the preservation of the nasal in Slavic
(cf. [19]), may have arisen by a (sporadic or regular) vowel lengthening in
Nominal Inflection 123
PS PBS *‑ēm yielded PS *‑ēn [19] (CS *‑ę [28]). In Old Church Slavonic the
personal pronouns mę and tę still occur in accented position, but in the early
history of the Slavic languages the reflexes of *mēn, *tēn, *sēn came to be used
as enclitics, and the original genitive forms *mene, *tebe, *sebe took over the
role of the tonic accusative. The accentuation of the forms in Slavic is uncer-
tain (Kapović 2006a: 81–82 reconstructs CS *mę̑, *tę̑; but cf. Vaillant 1958: 444).
According to some authors, Common Slavic had preserved an enclitic vari-
ant *me alongside tonic *mę (Meillet loc. cit.; cf. also Havránek 1928: 47–79;
Andersen 1998a: 441). Since the reflexes of *me have a limited distribution in
the Slavic languages and are easy to explain as the results of influence from the
original genitive forms, it is more likely that the forms are secondary (see e.g.
Vaillant loc. cit.).
PS PBS *ˌ‑es regularly yielded PS *‑e [17] (CS *‑e [29]). The ending is pre-
served in the old Slavic dialects, but there is a tendency everywhere to replace
it with the corresponding endings of the vocalic stems.
PBS Olander 2009: 171–172: PIE *‑éi̯s yielded unaccented PBS *ˌ‑ei̯s seen in Slavic;
Li. accented ‑iẽs is based on analogy with ā- and ē-stems | Endzelīns
1971: 152: original ending was preserved in OLv., Lv. dial. ‑is; Lv. ‑s is from
C-stems | Stang 1966: 206–207: Go. ‑ais points to o-grade; Lv. ‑s is from
C-stems, but ‑is is preserved in OLv., Lv. dial. | Otrębski 1956: 40: Li.
‑iẽs reflects *‑eis, also in Osc. ‑eis, OCS ‑i | Endzelīns 1923: 314–315: Li.
‑ies, OLv., Lv. dial. ‑is reflect PIE *‑eis; Lv. ‑s is C-stem ending
PS Vondrák 1908/1928: 2: PIE *‑oĩs or *‑eĩs (?) | Hujer 1910: 68, 118: Go. ‑ais
is from PIE *‑ois; Slavic ‑i reflects *‑eis or, more likely, *‑oĩs | Meillet
1924/1934: 418: Slavic ‑i is from *‑ois or *‑eis | Vaillant 1958: 133: PIE
*‑ei̯s is preserved in WGmc.; Go. ‑ais has analogical a from dat. sg. ‑ai,
which is taken from the ā-stems | Bräuer 1969a: 155: CS *‑i reflects PIE
*‑ei̯s, alongside *‑oi̯s attested in Go. ‑ais | Arumaa 1985: 124: CS *‑i, Li.
‑iẽs reflects *‑eis; Go ‑ais goes back to *‑ois | Igartua 2005a: 250–251:
CS *‑i, Li. ‑iẽs reflect PIE *‑ei̯s; Gmc. points to PIE *‑oi̯s | Aitzetmüller
1978/1991: 73
PBS PIE *‑ei̯s is reflected as PBS *ˌ‑ei̯s [4], yielding Li. ‑iẽs and OLv., Lv. dial.
‑is. Standard Latvian ‑s has been imported from the consonant stems. In Old
Prussian there are no certain attestations of this ending.
PS PBS *ˌ‑ei̯s became PS *ˌ‑ei̯ [12|17] (CS *‑i [22|29]), preserved in the old
Slavic dialects.
157 | Stang 1966: 214: Li. ‑aũs, Lv. ‑us from PIE *‑ous; OPr. soūnas is
remade on the model of the o-stems | Otrębski 1956: 47: Li. ‑aus
reflects PIE *‑eus or *‑ous | Endzelīns 1923: 326: Lv. ‑us is from PBalt.
*‑aus
PS Vondrák 1908/1928: 2: OCS ‑u, Li. ‑aus, Go. ‑aus reflect PIE *‑oũs | Hujer
1910: 118: Li. ‑aus, Go. ‑aus, Slavic ‑u point to PIE *‑oũs rather than
*‑eus | Meillet 1924/1934: 413: Slavic ‑u is from *‑ous | Vaillant
1958: 109: Slavic ‑u, Li. ‑aus, Lv. ‑u are from PIE *‑ous | Bräuer 1969a:
146: CS *‑u reflects PIE *‑ou̯ s | Arumaa 1985: 124 | Igartua 2005a:
283–284: Slavic ‑u reflects PIE *‑o/eus | Aitzetmüller 1978/1991: 70:
OCS ‑u is from PIE *‑ous rather than *‑eus
PBS PIE *‑eu̯ s yielded pre-PBS *ˌ‑i̯au̯ s [4|10]. By analogy with other forms of
the paradigm the *i̯ was eliminated, resulting in PBS *ˌ‑au̯ s. The ending is pre-
served in East Baltic, whereas in Old Prussian the o-stem ending is used. While
the form remained unaccented in Slavic, in Lithuanian it received final accen-
tuation in the mobile paradigms by analogy with the ā- and ė-stems.
PS PBS *ˌ‑au̯ s would probably yield PS *ˌ‑əu̯ [12|17] (CS *‑ju [22|29]), although
the sequence is only found here. The attested Slavic languages point to PS *ˌ‑au̯
(CS *‑u [22|29]). The ending, with its aberrant *ə, was normalised to *‑au̯ under
the influence of the other forms of the paradigm and the corresponding i-stem
ending PS *‑ei̯.
Nominal Inflection 129
PS *naˈgə̄ (CS Cl *nogỳ || ONovg. ‑ě; CS *dušě)̀ PBS *‑ˈā̰s PIE gen.–abl. *‑ah₂s
PIE Brugmann 1909–1911: 152–155: PIE *‑ā̃s; OCS ‑y, ‑ję represent n-stem gen.
sg. *‑ons; WSl. and ESl. ‑( j)ě is unexplained, perhaps based on ī-stem end-
ing *‑i̯e(s) | Beekes 1995/2011: 200: PIE *‑h₂os; OCS ‑y may be from
nom. pl. | Rasmussen forthc. a § 5: PIE *‑ah₂os; OCS ‑y is from acc.
pl. | Debrunner & Wackernagel 1930/1975: 38–39, 119–121: old ending
*‑ās is preserved in Ved. gnā́s‑páti‑, jā́s‑pati‑; PIE circumflex tone, seen in
Gk. and Li., probably results from contraction of *‑āes or *‑āos; IIr. ‑āy‑
was probably introduced on the pattern of devī-stems | Rix 1976/1992:
132: PIE *‑eh₂s; Gk. circumflex tone is taken from dat. sg., gen. pl. and dat.
pl.; Li. circumflex tone is obligatory in a long final syllable; a proto-form
PIE *‑eh₂es with two full grades is unlikely | Sihler 1995: 269–270: Gk.
ending is “easiest to explain” from PIE *‑éh₂os/es; La. ending is remade
after o-stem ending | Weiss 2009/2011: 229, 234: PIE *‑eh₂es; La. ‑ae
< ‑āī is based on analogy with o-stem ending | Krahe 1942/1967: 21:
PGmc. *‑ōz goes back to PIE *‑ãs | Boutkan 1995: 225–227: PGmc. *‑ōs
< *‑ās is remade from PIE *‑h₂os | Kloekhorst 2008a: 107: Hi. ḫ-stem
arḫaš reflects PIE *‑h₂ós
PBS Olander 2009: 170–171: PBS *‑ˈā̰s is the regular reflex of PIE *‑áh₂s; Gk.
and Li. circumflex tones are independent analogical innovations |
Endzelīns 1971: 142–143: PIE *‑āe/os; OPr. ‑as after g or w points to short
a, which is analogical from o-stems | Stang 1966: 197–198: Gk. and
Li. circumflex tones point to *‑eh₂e/os; OPr. ‑as for *‑ūs after labials and
velars, not attested in Cat. I and II, is based on analogy with acc. sg. |
Otrębski 1956: 25: PIE *‑ās | Endzelīns 1923: 304
130 Chapter 3
In Indo-Iranian the ending has been extended by *‑āy‑ as in the dative and
locative endings.
PBS PIE *‑ah₂s yielded PBS *‑ˈā̰s [1]. The regular outcomes of this ending in
Lithuanian would have been *‑às when accented and *‑as when unaccented,
by Leskien’s law. The ending would also have attracted the accent in a.p. 2, e.g.
*rankàs, by Saussure’s law. Since the ending does not show the effects of any of
these processes, it must have been non-acute in pre-Lithuanian. I assume that
at some point in the prehistory of Lithuanian, inherited PBS *‑ā̰s was replaced
with non-acute *‑ās under the combined influence of the dative singular
of the ā‑stems and the genitive singular of the remaining paradigms, above
all the ė‑stems, where the circumflex tone of Li. ‑ė̃s probably arose as a result
of the contraction of *‑iˈi̯ā̰s. Since neither the other Baltic languages nor Slavic
allows us to see whether this ending was acute or not in Proto-Balto-Slavic, the
best indication of the chronology of the change is the assumption that it was
facilitated by the circumflex tone of the ė-stem ending. As the contraction of
*‑iˈi̯ā̰s to Li. ‑ė̃s with circumflex tone is probably an internal Baltic process, it is
likely that the replacement of *‑ā̰s with *‑ās took place after the dissolution of
Proto-Balto-Slavic.
Old Prussian forms like ālgas, with ‑as following a velar or a labial, cannot
regularly reflect *‑ās. Instead, they have probably been remade after forms like
the accusative singular (Stang loc. cit.)
PS PBS *‑ˈā̰s regularly yielded PS *‑ˈə̄ [13|17] (CS Cl *‑y || ONovg. ‑ě [29]; CS *‑ě
[20|29]). The hard ending is preserved in most old Slavic dialects. In Serbo-
Croatian the soft ending seems to have replaced the hard ending already in the
earliest documents (Svane 1958: 64) and it prevails in most modern dialects;
the length of the ending in many dialects is often explained as the result of the
influence from the pronominal form, CS *‑oję (e.g. Bräuer 1969a: 111; Vaillant
1958: 93; for an alternative interpretation of the length see Kapović forthc. with
an overview of the quantity of the ending in various dialects). The soft ending
has also replaced the hard ending in Slovene.
In the soft stems reflexes of PS *‑i̯ə̄ (CS *‑jě) appear as expected in West and
East Slavic. In South Slavic we find reflexes of *‑ję. The most reasonable expla-
nation of the South Slavic ending is that it ultimately stems from the ā-stem
accusative plural, where *‑ję is the regular reflex of PS *‑i̯ə̄n [20|28] < PBS *‑i̯āns
[17]. We are dealing with a proportional analogy on the pattern of the ā- and
i-stems. The fact that the genitive singular and nominative plural were (seg-
mentally) identical to the accusative plural in the hard ā-stems (merging in
‑y) and in the i-stems (merging in ‑i) provided a catalyst for a transfer of the
132 Chapter 3
i̯ā-stem accusative plural form to the genitive singular and nominative plural
(similarly Tedesco 1951: 173). Because of this analogy in the genitive singular
and nominative plural of the i̯ā-stems, the phonetically regular reflex of PS *‑i̯ə̄
is not preserved in Old Church Slavonic, but it is likely that it would have been
*‑jě as in the remaining Slavic dialects.
In the Old Novgorod dialect the ending ‑ě is regular in both hard and soft
stems in the earliest birchbark letters; the ending ‑y is rare. It is clear from the
attestations that ‑y does not originally belong to the inflectional system of the
Old Novgorod dialect, but has intruded from standard Old Russian (Zaliznjak
1995/2004: 97). The most plausible explanation of this situation is that ‑( j)ě is
simply the phonetically regular merger of PS *‑ə̄ and *‑i̯ə̄ from pre-PS *‑ās and
*‑i̯ās in this dialect (see Olander 2012: 334–335).
Together with the ā-stem nominative plural, the ā-stem genitive singular
ending is the main argument in favour of the old, but never particularly popu-
lar view that CS *‑y is the regular outcome of pre-PS *‑ās (see Olander 2012:
331–332, with references in n. 85). Scholars who are reluctant to accept a pho-
netic development of *‑ās to *‑y in Slavic have put forward alternative expla-
nations of the genitive singular and nominative plural of the ā-stems. In the
following paragraphs I shall present some of these explanations and the objec-
tions that may be raised against them (see also the overviews in Hujer loc. cit.;
Igartua loc. cit.; 2005b: 282–290).
According to one view, quite popular in the older literature, OCS ‑y and ‑ję
reflect an n-stem ending PIE *‑ō̆ns also found in Germanic, e.g. Go. ōn-stem
gen. sg. qinons (e.g. Mikkola 1897: 249–250; Brugmann loc. cit.; Hujer loc. cit.;
V. Kiparsky 1967: 83). As noted by Igartua (2005a: 201–202; 2005b: 284, both
with references), this view is implausible as there are no other traces of this
inflection in Slavic; the ōn-stems are very likely to be a Germanic innovation.
Besides, as Guus Kroonen reminds me (pers. comm., 2014), PIE *‑ōns would
yield **‑os in Gothic, not ‑ons.
Other authors believe that ‑y and ‑ję in the ā- and i̯ā-stems reflect the origi-
nal accusative plural ending, which was transferred to the nominative plural
and from there to the genitive singular by analogy with the i-stems, where the
three endings had merged by way of phonetic developments (Ljapunov 1905:
36; Vaillant loc. cit.; Mareš 1962b: 19; 1966: 168; Bräuer loc. cit.; Kortlandt 1975/
2011a: 45; Rasmussen loc. cit.; Igartua loc. cit.; 2005b; see also already Scherer
1868: 291, 474–475).
This explanation is similar to the one I have put forward above in order to
account for the introduction of the accusative plural ending in the nominative
plural and genitive singular of the South Slavic i̯ā-stems. There are, however,
two significant differences between the two explanations. First, I assume that
the i̯ā-stem genitive singular and accusative plural were influenced primarily
Nominal Inflection 133
by the closely related ā-stems, which had the phonetically regular endings,
and only possibly by the i-stems as well. Second, I assume that the process
took place at a relatively late stage, affecting only South Slavic. By contrast,
the explanation of the genitive singular and accusative plural endings of both
the ā- and i̯ā-stems as analogical presupposes that the decisive influence came
from the more remotely related i-stems; and that the process took place at an
early stage, since it affected all Slavic dialects.
In the first place I find the idea unlikely that the ā- and i̯ā-stems, which
played a prominent role in Slavic morphology, should have used the more
marginal i-stems as a model here for such a significant remodelling of the
paradigms. Accordingly, even if it were possible to set up a perfect propor-
tional analogy for the transfer of the ending of the accusative plural endings
of the ā- and i̯ā-stems to the genitive singular and nominative plural on the
model of the i-stems, it is improbable that such an analogy would ever be
operationalised.
Furthermore, a closer inspection reveals that the necessary proportional
analogy is not even as good as it seems. In order for the analogy to work, the
three i-stem forms involved must be identical. The genitive singular reflects
PIE *‑ei̯s and the accusative plural reflects *‑ins; the origin of the feminine
nominative plural is less obvious, but it is most likely that the ending has been
transferred from the accusative plural and thus also reflects *‑ins (see § 3.13.3).
Thus the forms only merged when *ei̯ (from *‑ei̯s) merges with *ī (from *‑ins).
Since original *ī triggers the progressive part of the second palatalisation while
*ei̯ does not, this means that the merger is later than the second palatalisation.
As we have seen in the chronology of Slavic sound changes (§ 2.3), the second
palatalisation [23] is a rather late, post-Proto-Slavic, process, with partly dif-
ferent results in the Slavic languages. Thus the prerequisite for an analogical
substitution of the ending of the ā-stem genitive singular with that of the accu-
sative plural is in place at such a late point in time that the Slavic dialects were
already beginning to diversify. It is therefore surprising that all Slavic dialects
seem to have carried out this—to put it mildly—non-trivial innovation in a
completely uniform manner; we would expect to find at least some traces of
different strategies for the replacement of the expected endings CS **‑a, **‑ja
from pre-PS *‑ās.
Those who divide the second palatalisation into two processes, an (earlier)
progressive and a (later) regressive one, may assume that the merger of *ei̯ and
*ī was completed already in pre-Proto-Slavic. This allows for a scenario where
the analogical replacement of the genitive singular with the accusative plural
was also a pre-Proto-Slavic development. However, there is reason to believe
that the second palatalisation was one single process (see [23]), which rules
out this scenario.
134 Chapter 3
PIE Brugmann 1909–1911: 163–167 (1897: 152): PIE *‑ēd̃ /t or *‑ōd̃ /t; Li. ‑o, OCS
‑a are from word-final *‑ōd | Beekes 1995/2011: 212–213: *‑ād replaces
*‑ōd < *‑oh₁ed in BS | Rasmussen forthc. a §§ 2, 4: PBS *‑ā̃ probably
goes back to PIE *‑oat; OPr. ‑as is probably abl. ending plus *‑s from other
paradigms | Debrunner & Wackernagel 1930/1975: 37, 94–95: Ved. ‑āt,
Gk. dial. adv. ϝοικω, Li. ‑o, Lv. ‑a, Slavic ‑a reflect PIE *‑ōd, *‑ēd | Rix
1976/1992: 170: Gk. adv. τῶ is from *‑ō̃t < PIE *‑oet, also in OLa. ‑ōd, Ved.
‑āt | Sihler 1995: 250–251, 258–259, 269: OLa. ‑ōd, Gk. Dor. adv. ὄπω
reflect PIE *‑ōt; Li. ‑o points to *‑āt; *‑t is probably related to Hi. instr. ‑at,
abl. ‑az(a) | Weiss 2009/2011: 202, 222: perhaps PIE *‑oh₂Vd (in OLa.
‑ōd), *‑eh₂Vd (in PBS *‑ād > Li. ‑o) | Krahe 1942/1967: 83–84: Go. galeiko
is from PIE abl. *‑õd; OEng. wīde is from variant PIE *‑ẽd | Boutkan
1995: 181–182, 378–381: Go. adv. jaiþro goes back to PIE *‑ōt; there was no
PIE abl. ending *‑ēt | Kloekhorst 2008a: 104, 231–232: Hi. ‑az is from
PIE *‑oti
PBS Olander 2009: 169–170: PBS *ˌ‑ā is from PIE *‑ó(h)at with regular loss of
accent | Endzelīns 1971: 134: East Baltic and Slavic endings reflect
*‑āt < abl. *‑o‑at; OPr. ‑as is probably related to OEng. gen. dōmæs [i.e.
from PIE gen. *‑e/oso] | Stang 1966: 181: Li. ‑o, Lv. ‑a, OCS ‑a reflect
PIE *‑oat or *‑oh₂et; OPr. ‑as from PIE *‑os(i̯)o | Otrębski 1956: 13–14:
Li. ‑o, Lv. ‑a, Slavic ‑a represent unclear innovation *‑ā, not found in
OPr. | Endzelīns 1923: 292: Lv. ‑a, Li. ‑o, Slavic ‑a may reflect old abl.
ending in *‑t or *‑d
PS Vondrák 1908/1928: 3: OCS ‑a, Li. ‑o, Ved. ‑ād, OLa. extrād, Go. ƕaþro
reflect PIE *‑ād | Hujer 1910: 114–117: Slavic ‑a, Li. ‑o (from *ō̃ in an
open final syllable), La. ‑ō(d), Gk. adv. ‑ῶ(ς) are from PIE *‑ō̃d/t, along-
side *‑ēd̃ /t | Meillet 1924/1934: 389 | Vaillant 1958: 30, 48 (1950:
Nominal Inflection 135
112): Slavic ‑a, Li. ‑o, Lv. ‑a reflect PIE *‑ōd/t; OPr. ‑as is from *‑ā plus *‑s
from other paradigms; Li. ‑o, Lv. ‑a point to *‑ād/t, with a-vocalism from
other forms; Slavic ‑ja for expected ‑jě in soft stems, attested in Glagolitic
OCS and OCz., is based on analogy with hard stems | Bräuer 1969a:
21–22, 73: gen. case is recent, perhaps not even PIE; CS *‑a < PIE abl.
*‑ōd < *‑oad | Arumaa 1985: 134–136: Slavic ‑a is probably from
PIE *‑ō̃d/t | Igartua 2005a: 123–133: Slavic ‑a, Li. ‑o, Lv. ‑a reflect
PIE *‑ō/ād; Slavic ‑ja for expected *‑jě is based on analogy with hard
stems | Aitzetmüller 1978/1991: 76–78, 83: OCS ‑a results from merger
of PIE abl. *‑ō/ād and loc. *‑ō; Slavic ‑ja in soft stems is based on analogy
with hard stems
1975: 460). Against the presence of *h₂ speaks the hypothesis that the thematic
vowel was *o only before voiced segments. However, the *o may have been
introduced secondarily, either in pre-Proto-Indo-European or, more likely, at a
pre-stage of Latin and, perhaps, Greek. Thus all in all the combined evidence
of Balto-Slavic and internal reconstruction of Proto-Indo-European speaks
slightly in favour of a proto-form *‑ah₂ad.
PS PBS *ˌ‑ā was retained as PS *ˌ‑ā (CS *‑a [29]; *‑ja [20|27|29]), preserved in
the old Slavic dialects.
PS →*taˈga (CS *togò; *jegò) PBS *‑as(i ̯)a (*‑es(i ̯)a) PIE noun, pron. *‑os(i ̯)o (*‑es(i ̯)0)
PIE Brugmann 1909–1911: 358–360: Ved. tásya, OAv. kahiiā, cahiiā, YAv.
aētahe, Hom. τοῖο, ἐμεῖο, OPr. stesse reflect PIE pron. *‑e/osi̯o, alongside
*‑e/oso | Beekes 1995/2011: 226, 228, 230–231: PIE *(to)si̯o, interr. adj.
*kʷoso; OCS form is unexplained; Go. ‑is from *‑eso is an original pron.
ending | Rasmussen forthc. a § 14: PIE *kʷósi̯o; Slavic *koγo is perhaps
from *koso by imitation of interr. pron. *[tʃeso], where the second syllable
Nominal Inflection 137
PIE The ancient Indo-European languages point to PIE *‑osi̯o and *‑oso; these
forms are obviously related to *‑esi̯o and *‑eso, the latter of which is reflected
in the inanimate interrogative pronoun in Slavic, OCS česo (see § 3.5.7). The
historical relationship between the four variants, found both in nouns and pro-
nouns in the various languages, is difficult to determine; I shall not speculate
about it here (see e.g. Rasmussen 1987a/1999: 122–124 for an overview). While
*‑o/es(i̯)o is usually regarded as having a pronominal origin, it became a noun
138 Chapter 3
PBS Baltic shows no traces of forms in *‑os(i̯)o and *‑eso. For OPr. stesse see
§ 3.5.7. The Slavic forms may point to the existence of PBS *‑as(i̯)a from PIE
*‑os(i̯)o [7].
PS The Proto-Slavic ending is *‑aga (CS *‑ogo [29]). It it still to a large extent
true what Leskien wrote about the ending: “Der Slavische Genitiv sing. togo ist
einer der schwierigsten Casus der indogermanischen Declination und daher
den mannichfachsten Erklärungsversuchen ausgesetzt gewesen, die nach
meiner Meinung alle zu nichts geführt haben oder sehr unsicher sind” (1876:
109 [emphasised in original]). Indeed, the form is difficult; see the useful over-
view of the literature provided by Matzinger (2001: 200–206). Matzinger him-
self puts forward a hypothesis according to which Slavic *‑ogo and *‑ovo reflect
uninflected adjective formations, *to‑u̯ o(‑d) (in Russian and Kashubian) and
*to‑gʰo(‑d) (elsewhere) (see also Andersen 1998a: 441); but this solution is not
very attractive. According to a more popular hypothesis, *‑ogo contains a par-
ticle, an idea which goes back to Miklosich and Maretić (see Berneker 1904: 374;
cf. Pedersen 1905b: 223 and the elaborate scenarios proposed by Meillet loc. cit.
and Vaillant loc. cit.). While this view perhaps cannot be dismissed, it requires
a number of additional hypothetical steps and is thus rather uncertain.
Since we find an unambiguous reflex of an original pronominal genitive
ending in PS *kesa (CS *česo), it is to begin with more likely that PS *‑aga (CS
*‑ogo) also somehow reflects a pronominal genitive ending. I am therefore
more inclined to think that PS *‑aga reflects an irregular phonetic develop-
ment of PBS *‑asa, perhaps caused by the fact that the ending contains two
identical vowels (cf. the different scenario assumed by Rasmussen 1987a/1999:
122–124). This development is similar to the Russian development of g to v in
the ending ‑ogo, with a parallel in Slovincian, which is due to a late weaken-
ing of [ɡ] to [γ] to [v] between two identical vowels (Stang 1969a: 75–76). It is
evident, however, that the explanation by a sporadic sound change is purely
ad hoc.
The ending was preserved in the old Slavic dialects.
Nominal Inflection 139
PS *kesa (CS *česo, *čьso) PBS *‑es(i ̯)a PIE *‑es(i ̯)o (*‑os(i ̯)0)
PIE Brugmann 1909–1911: 358–360: Hom. τέο, Att. τοῦ, Go. ƕis, OCS česo reflect
PIE pron. *kʷeso, alongside *‑e/osi̯o; OCS čьso represents *‑so transferred
to the i-stems | Beekes 1995/2011: 230–231: OCS and Gk. point to PIE
*kʷeso, alongside adj. PIE *kʷoso | Rasmussen forthc. a § 14: PIE
masc.–neut. *tési̯o; OCS česo < *keso is from *kʷesi̯o, perhaps with *s from
gen. pl. | Debrunner & Wackernagel 1930/1975: 502: OCS česo, Gk. τέο,
Go. ‑is contain PIE *‑so, alongside *‑si̯o | Rix 1976/1992: 187: inanim.
interr. PIE *kʷéso (vs. animate *kʷóso) is preserved in Gk. τοῦ, Hom. τέο,
OAv. cahiiā (with *‑si̯o from dem. pron.), Go. ƕis, OCS česo | Sihler
1995: 259–260, 387–388, 397–399: OAv. cahiiā, Hom. τέο, Att. τοῦ are from
PIE interr. pron. *kʷesi̯o, also in Go. ƕis (which is not from *‑eso) and
OCS česo (which may have resulted from contamination with kogo,
togo) | Weiss 2009/2011: [see o-stem pron., § 3.5.6] | Krahe 1942/
1967: [see o-stem pron., § 3.5.6] | Boutkan 1995: 175–181: Hi. points
to PIE noun ending *‑os, while other languages point to *‑osi̯o; Gk. τέο,
OCS česo, Go. ‑is, OHG ‑es reflect pron. *‑eso; OEng. þæs shows secondary
introduction of *a | Kloekhorst 2008a: [see o-stem pron., § 3.5.6]
PBS Olander 2009: – | Endzelīns 1971: 192, 195–196: Li. kõ, Lv. kà have noun
endings | Stang 1966: [see o-stem pron., § 3.5.6] | Otrębski 1956:
141 | Endzelīns 1923: 386, 398
PS Vondrák 1908/1928: 90–91 (1906/1924: 62): OCS česo, Go. þis, OHG des
have *‑so for *‑si̯o; ь in OCS čьso arose in unaccented position in e.g.
ničьso | Hujer 1910: 112–113: PIE *‑si̯o, secondarily *‑so; OCS česo is per-
haps from *kʷoso | Meillet 1924/1934: 443–444: Hom. τέο, OHG hwes
reflect PIE *kʷeso; Slavic česo was reduced to čьso, facilitated by nom.–
acc. čьto | Vaillant 1958: 372, 405: OCS česo reflects PBS *‑esa from
PIE *‑eso; OCS čьso is modelled on čьto | Bräuer 1969a, 1969b: – |
140 Chapter 3
Arumaa 1985: 175, 183–184: OCS česo, čьso preserve old pron. end-
ing | Igartua 2005a: 126: Slavic česo, čьso reflect *kʲeso, perhaps dis-
similated from *kʲesi̯o | Aitzetmüller 1978/1991: 123: OCS česo is from
*kʷeso; *‑eso is also found in Go. þis, OHG des; OCS čьso has ь from nom.–
acc. čьto
PIE The ancestor of PS *‑esa (CS *‑eso) is PIE *‑eso (or *‑esi̯o), a variant of
which is probably also reflected in PS *‑aga (CS *‑ogo); for problems connected
with these variants and their reflexes in the other Indo-European languages
see § 3.5.6.
PBS Lithuanian and Latvian employ the ending of the noun, which ulti-
mately goes back to the Proto-Indo-European ablative form. In Old Prussian
the genitive singular of the demonstrative pronoun, definite article and third-
person pronoun is stesse (alongside stessei and other variants), which may
reflect PIE *‑esi̯o, but this is uncertain (cf. the discussion in Stang loc. cit. with
Endzelīns 1944: 111–114). On the basis of the Old Prussian form and PS *‑esa we
may reconstruct PBS *‑es(i̯)a from PIE *‑es(i̯)o [7].
PIE Brugmann 1909–1911: 383, 415–416: PIE *mene, *teu̯ e, *seu̯ e; OCS tebe, sebe
are remade after dat.–loc. tebě, sebě | Beekes 1995/2011: 232–235: PIE
*h₁méne, *teue, *seue; OCS has ‑b‑ from dat. | Rasmussen forthc. a
§ 13: PIE *méne, *téu̯ e, *séu̯ e; OCS mene, tebe, sebe are from PBS *mené,
*tåu̯ é, *såu̯ é, with ‑eb‑ for *‑ov‑ in 2sg. and refl. by analogy with dat.; Li.
manę̃s etc., Lv. manis etc. are acc. forms plus gen. ending ‑s | Schmidt
1978: 87–93, 109–110, 136–144, 154–156, 165–166: 1sg. PIE *(e)menē̆ (origi-
nally gen.–abl.); 2sg. gen.–dat. *teu̯ (originally loc.), with *‑e from 1sg. in
Slavic and IIr.; refl. gen.–abl. *sē̆nē̆; Ved. máma has arisen by assimilation
from PIIr. *mana; Slavic tebe, sebe have ‑b‑ from dat. | Debrunner &
Wackernagel 1930/1975: 461: Ved. máma is from *áma (> Gk. ἐμ‑, Arm. im),
142 Chapter 3
with m‑ from other case forms; Ved. táva goes back to PIE *teu̯ e | Rix
1976/1992: 177–178, 180: PIE tonic *méne, *téu̯ e; attested Gk. forms are
based on acc. stem | Sihler 1995: 372–373, 375–377: PIE tonic *méme
(with dissimilation in Av. and BS), *téu̯ e, *séu̯ e; Gk. 1sg. forms are based
on *éme (based on PIE *méme, altered under the influence of the nom.
form); Gk. 2sg. and refl. forms are based on acc.; La. meī, tuī, suī are masc.
gen. sg. of poss. pron. meus, tuus, suus | Weiss 2009/2011: 327, 329: PIE
*méne, *teu̯ e; Ved. máma has arisen by assimilation; La. meī is gen. sg. of
poss. pron. meus | Krahe 1942/1967: 50, 52–54: Go. meina, þeina, seina
are forms of the possessive pron., perhaps neut. acc. pl. | Seebold
1984: 47–51: Gmc. *mīnē < *meinē is based on PIE *mene under the influ-
ence of loc. / possessive stem *mei‑; 2sg. Gmc. *þīnē and refl. *sīnē are
based on 1sg. | Kloekhorst 2008a: 240: Hi. ‑ēl is unclear
PBS Kapović 2006a: passim; 80, 94–96, 100, 113–114, 158–159: PIE *méne, *téu̯ e;
PBS *mène, *tàu̯ e; CS *mȅne, *tȅbe; East Baltic forms are based on the
acc. plus gen. marker ‑s; circumflex Li. ‑ę̃s is analogical from gen. of
nouns | Endzelīns 1971: 186–187: PIE *mene (indicated by OCS mene,
YAv. mana), *teu̯ e (indicated by Ved. táva, YAv. tauua; OCS has b for *v by
analogy with dat. tebě), *seu̯ e; Li. dial. manè, tavè, savè reflect PIE forms,
with a from dat.; the new acc. forms Li. manę̃, tavę̃, savę̃ were used with
gen. function and received an ‑s; Lv. dial. mani, tavi, sevi may correspond
to Li. forms in ‑ę̃; Lv. ‑is may correspond to Li. ‑ę̃s | Stang 1966: 250–
251: PBalt. *mene, *teve, *seve were altered to East Baltic *manes, *teves,
*seves with *‑s from nouns; forms in *‑ę̃s have nasal from acc.; forms in
‑ę̃ are original acc. forms | Otrębski 1956: 135–136: forms ended origi-
nally in *‑e, as in Li. dial. manè, tavè, savè; in Li. manę̃s, tavę̃s, savę̃s, *‑e
was first replaced with *‑es from C-stems, then the ending was nasalised
under the influence of the acc.; *man‑ has a for *e under the influence
of poss. pron. mãnas | Endzelīns 1923: 372–374: original *mene, *teu̯ e,
*seu̯ e are preserved in Li. dial. manè, tavè, savè (with a from dat.), Lv. dial.
man, tev, sev; Li. dial. manę̃, tavę̃, savę̃ have nasal from acc.; Li. dial. manès,
tavès, savès and Li. manę̃s, tavę̃s, savę̃s, Lv. manis, tevis, sevis are based on
*mane etc. and *manen etc., respectively, plus gen. marker *‑s
PS Vondrák 1908/1928: 70–71: OCS mene is inherited; OCS tebe, sebe reflect
*teu̯ e, *seu̯ e with b from dat. | Meillet 1924/1934: 454: Slavic mene cor-
responds to YAv. mana; Slavic tebe, sebe reflect *teve, *seve with b from
dat.–loc. tebě, sebě | Vaillant 1958: 445–446: PIE *mene, *teu̯ e, *seu̯ e;
Slavic has b from stem te/ob‑, se/ob‑ in other oblique cases | Arumaa
Nominal Inflection 143
1985: 160–161, 163: Slavic mene is from PIE *mene; Slavic tebe, sebe are from
*teu̯ e, *seu̯ e, with *b from dat. | Aitzetmüller 1978/1991: 108, 111, 113:
PIE *mene, *teu̯ e, *seu̯ e; OCS tebe, sebe have b from dat.
PS PBS *mene, *tebe, *sebe yielded Proto-Slavic *ˌmene, *ˌtebe, *ˌsebe (CS
*mȅne, *tȅbe, *ˌsebe), preserved in South and East Slavic. In Old Czech and
generally in West Slavic, the stem *men‑ has been replaced with mn‑ from the
dative, instrumental and locative singular.
144 Chapter 3
PIE Brugmann 1909–1911: 172–174: OCS ‑i is from PIE *‑ai or from possible
i-stem ending *‑ī | Beekes 1995/2011: 186–187, 194–195: PIE *‑(e)i |
Rasmussen forthc. a §§ 2, 8: OCS ‑i is from PIE non-ablauting *‑ei̯ |
Debrunner & Wackernagel 1930/1975: 36–37: Ved. ‑e goes back to PIE *‑ai
or *‑ei | Rix 1976/1992: 154: Gk. ‑ι is loc. sg.; PIE *‑ei̯ is preserved in Myc.
po-me-ne, Gk. dial. Διει, OLa. ‑ei | Sihler 1995: 284–285: La. ‑ī, OLa. ‑ei,
Osc. ‑eí, Myc. po‑de are from PIE *‑ei̯ | Weiss 2009/2011: 199, 201–202:
La. ‑ī, OLa. ‑ei, Myc. ma-te-re reflect PIE *‑ei | Krahe 1942/1967: [see
loc. sg., § 3.8.1] | Boutkan 1995: [see loc. sg., § 3.8.1] | Kloekhorst
2008a: 376: Hi. ‑i is the regular reflex, after certain consonants, of PIE dat.
sg. *‑i (alongside *‑ei, not preserved in Hi.)
PBS Olander 2009: – | Endzelīns 1971: 163: Li. dial. dùkter(i) is from
*‑ie | Stang 1966: 208, 220: PIE dat. *‑i is preserved in Li. ‑i, Gk. ποδί;
variant *‑ei is reflected in Li. dial. ‑ei, OPr. ‑ei, Ved. ‑e, OLa. ‑ei | Otrębski
1956: 53, 57: OLi., Li. dial. ‑i may correspond to Gk. μητέρι; OLi., Li. dial. ‑iẽ
is from PIE *‑ei, also in OLa. ‑ei | Endzelīns 1923: 322: Lv. akmenim,
akmeńam have i̯o-stem ending; old ending is preserved in Li. dial.
ãkmen(i)
PS Vondrák 1908/1928: 3: OCS ‑i is from PIE *‑aí | Hujer 1910: 124–127:
Li. ‑i, Gk. ‑ι are from loc. *‑i; Slavic ‑i is from PIE acute *‑ai | Meillet
1924/1934: 390, 422: PIE *‑e/oi | Vaillant 1958: 185: Slavic ‑i, OLi., Li.
dial. ‑i go back to PIE *‑ei, also in OLa. ‑ei, Osc. ‑ei, Ved. ‑e | Bräuer
1969b: 8, 53, 71, 83: CS *‑i is from PIE *‑ei̯ or *‑ai̯ | Arumaa 1985: 110–111:
Slavic ‑i is from PIE *‑ei | Igartua 2005a: 336–337: Slavic ‑i is from PIE
*‑ei | Aitzetmüller 1978/1991: 94, 98, 100, 102: OCS ‑i is from PIE *‑ei
Nominal Inflection 145
PBS The outcome of PIE *‑ei̯ was PBS *‑ei̯, preserved in Old Lithuanian forms
like akmenie and OPr. giwāntei. As in the Greek and Germanic endings dis-
cussed above, Lithuanian forms in ‑i probably reflect the locative singular end-
ing PIE *‑i. Lv. akmenim and akmeńam are the i̯o-stem endings.
PS PBS *‑ei̯ yielded PS *‑ei̯ (CS *‑i [22|29]), preserved in the old Slavic dialects.
PIE Brugmann 1909–1911: 170–172: PIE had *‑(i)i̯ai, *‑ei̯ai, perhaps also *‑ī > OCS
‑i; or OCS ‑i is from C-stems | Beekes 1995/2011: 203: PIE *‑eii [implicit
from u-stems] | Rasmussen forthc. a § 6: PBS *‑ei̯ei̯ | Debrunner
& Wackernagel 1930/1975: 149–150 | Rix 1976/1992: 154: Gk. endings
reflect PIE loc. sg. | Sihler 1995: 314, 316: PIE *‑ei̯ei̯ yielded La. ‑ī and
perhaps Hom. ‑εϊ; [see also loc. sg., § 3.8.2] | Weiss 2009/2011: 242,
244–245: PIE *‑ei̯ei > *‑ei; the haplology took place “perhaps already in
Italo-Celtic” | Krahe 1942/1967: 30 [see loc. sg., § 3.8.2] | Boutkan
1995: 246–248 [see loc. sg., § 3.8.2] | Kloekhorst 2008a: 105, 376: Hi. ‑ai
is from PIE [loc.] *‑eii
PBS Olander 2009: 173–174: PBS *ˌ‑ei̯ from PIE *‑éi̯ has regular loss of
accent | Endzelīns 1971: 152: Lv. ‑ij is based on analogy with ā- and
ē-stems | Stang 1966: 136, 207–208: Li. dial. ‑ie, Lv. dial. ‑i, OCS ‑i, La. ‑ī
reflect *‑ei̯, from *‑ei̯ei̯ through haplological shortening; OLi. ‑i represents
PIE *‑i, the zero grade of dat. sg. *‑ei̯ | Otrębski 1956: 40: Li. ‑ie, OCS ‑i
are from *‑ei̯ | Endzelīns 1923: 315–317: OLv. ‑i is from *‑ei̯, also in Li.
‑ie, OPr. ‑ei, OCS ‑i; Lv. ‑ij perhaps replaced ‑i on the analogy of ā-stem ‑ai,
ē-stem ‑ei
PS Vondrák 1908/1928: 3 | Hujer 1910: 127–128: Slavic ‑i goes back to
C-stem *‑ai or to i-stem loc. sg., imitating identical ā-stem dat. and loc. sg.
forms | Meillet 1924/1934: monosyllabic ending of Slavic, Osc., La., Li.
(Žem.) may point to PIE haplology | Vaillant 1958: 133–134: PIE *‑i̯ei
was replaced with athem. *‑ei̯in Slavic and other IE languages | Bräuer
1969a: 155: CS *‑i reflects *‑ei̯, perhaps the result of haplology | Arumaa
1985: 124–125: OCS ‑i < *‑ei reflects *‑ei̯ei̯, partly under the influence of
C-stems, partly by haplology; shortening process has started already in
PIE | Igartua 2005a: 251–256: Slavic ‑i is from PIE *‑ei̯ (pure stem);
forms pointing to PIE *‑ei̯ei̯ are secondary | Aitzetmüller 1978/1991:
73–74: OCS ‑i is from *‑ei, which is either the C-stem ending or based on
analogy with the loc. sg.
PIE We expect an ending *‑ei̯ei̯ or *‑i̯ei̯, consisting of the full or zero grade
of the i-stem suffix and the dative singular marker *‑ei̯. The expected ending is
found in Indo-Iranian. The Balto-Slavic endings apparently reflect *‑ei̯, which
is most likely the result of a haplological shortening of *‑ei̯ei̯. Judging from the
fact that Italic also points to *‑ei̯ we may surmise that the haplology had already
taken place in pre-Proto-Indo-European and that the Indo-Iranian ending is a
more recent analogical creation (similarly Meillet 1913b).
The Greek forms are most likely innovations based on loc. sg. PIE *‑ēi̯ (but
cf. Sihler loc. cit.). La. ‑ī reflects PIE *‑ei̯ rather than *‑ei̯ei̯ (Meiser 1998/2006:
Nominal Inflection 147
138; but cf. again Sihler loc. cit. and Weiss loc. cit.). The Gothic masculine end-
ing ‑a is imported from the o-stems, while the feminine ending ‑ai probably
reflects the i-stem locative singular ending *‑ēi̯. According to the hypothesis
put forward by Hansen (2014: 160–168), Go. ‑ai may also represent the phonetic
outcome of PIE *‑ei̯, though he finds it more likely that Go. ‑ai goes back to PIE
loc. sg. *‑ēi̯. In Hittite, the adjective ending ‑ai reflects the locative ending *‑ēi̯.
The fact that the Balto-Slavic ending is unaccented in the mobile paradigm
indicates that *‑ei̯ei̯ had already been haplologically shortened to *‑ei̯ when
the mobility law took place in pre-Proto-Balto-Slavic [4]. The i-stem dative sin-
gular thus coincided with the corresponding consonant-stem ending, consti-
tuting one of the points of intersection between the two stem-types, together
with the accusative singular and plural endings.
PBS PIE *‑ei̯ yielded PBS *ˌ‑ei̯ [4]. In Lithuanian the original ending was
replaced with the ā-stem ending in feminine i-stems (miñčiai) and by the i̯o-
stem ending in masculines (žvė�riui). In Old Lithuanian and in dialects the
original ending is preserved as ‑ie (see Stang loc. cit.; Zinkevičius 1966: 230). A
similar situation is found in Latvian, where the standard ending ‑ij is remade
after the ā- and ē-stem endings ‑ai and ‑ei (Endzelīns loc. cit.; Stang loc. cit.;
Forssman 2001: 129), but the original ending is preserved in dialectal forms in
‑i. Old Prussian also preserves the old ending.
According to Klingenschmitt (1992: 105–107), the Baltic and Slavic end-
ings cannot reflect PIE *‑ei̯, which would yield acute *‑i in Lithuanian and
final accent in Slavic (see also W. Hock 1995: 79). This view depends on
Klingenschmitt’s ideas about Balto-Slavic accentology, which I have criticised
in Olander 2009: 43–45.
PS The regular reflex of PBS *ˌ‑ei̯ is PS *ˌ‑ei̯ (CS *‑i [22|29]), preserved in the
old Slavic dialects.
PIE Brugmann 1909–1911: 170–172: PIE *‑(u)u̯ ai, *‑eu̯ ai; OCS ‑i is from *‑ai or
from possible i-stem ending *‑ī | Beekes 1995/2011: 203: PIE *‑eui; CS
*‑ov‑ is from *‑eu‑ | Rasmussen forthc. a § 6: OCS ‑ovi reflects PBS *‑au̯ ei̯
< PIE *‑eu̯ ei̯ | Debrunner & Wackernagel 1930/1975: 148–150: Ved. ‑ave,
OCS ‑ovi are from PIE *‑eu̯ a/ei̯ | Rix 1976/1992: 154: Gk. endings reflect
PIE loc. sg. | Sihler 1995: 323: Ved. ‑ave, OCS ‑ovi, OLa. ‑uei, La. ‑uī go
back to PIE *‑eu̯ ei̯, probably also in Myc. i‑je‑we /hujjewei/ | Weiss
2009/2011: 249, 251: La. ‑uī reflects PIE *‑eu̯ ei̯ | Krahe 1942/1967: 32:
Gmc. uses loc. ending | Boutkan 1995: 256–257: Gmc. dat. is histori-
cally a loc. | Kloekhorst 2008a: 105, 376: Hi. dat.–loc. ‑awi reflects PIE
*‑eui
PBS Olander 2009: 173–174: PBS *‑ˈau̯ ei̯ is from PIE *‑éu̯ ei̯; expected desinential
accentuation is perhaps preserved in Ru. adv. domój, dolój, Ukr. domív,
dolív | Endzelīns 1971: 157: Lv. ‑um has ‑m from o-stems | Stang
1966: 214–215: Ved. ‑ave, OCS ‑ovi reflect PIE *‑e/ou̯ ei; Li. ‑ui is o-stem end-
ing; Lv. ‑um is based on analogy with o-stem ending ‑am | Otrębski
1956: 47: Li. ‑ui is from o-stems | Endzelīns 1923: 326: Lv. ‑um is based
on analogy with masc. i-stem ‑im
PS Vondrák 1908/1928: 3: OCS ‑ovi, Ved. ‑ave go back to PIE *‑eu̯ ai | Hujer
1910: 126–127: Slavic ‑ovi, Ved. ‑ave, La. ‑uī are from PIE *‑eu̯ ai | Meillet
1924/1934: 414 | Vaillant 1958: 110: Slavic ‑ovi is from PIE *‑eu̯ ei (with
secondary ‑o‑) or *‑ou̯ ei | Bräuer 1969a: 146: PIE *‑eu̯ ai̯ | Arumaa
1985: 124–125: OCS ‑ovi goes back to *‑eu̯ ei; ORu. adv. dolovь, domovi,
domovь are old dat. forms, not loc. | Igartua 2005a: 284–286: Slavic
‑ovi reflects *‑e/ou̯ ei, perhaps consisting of loc. ending *‑ou plus dat.
marker *‑ei̯ | Aitzetmüller 1978/1991: 70
PIE PIE *‑eu̯ ei̯ consists of the full grade of the u-stem suffix followed by the
dative ending *‑ei̯. The ending is preserved in Indo-Iranian and Italic. In Greek,
Gothic and Hittite it was replaced with the locative ending.
PBS PIE *‑eu̯ ei is continued as PBS *‑ˈau̯ ei̯ [11]. The ending is not represented
in Baltic, where Lithuanian uses the o-stem ending, and the Latvian ending
is based on analogy with the o-stem ending ‑am, which originates in the pro-
nouns. The ending is not attested in Old Prussian.
PS PBS *‑ˈau̯ ei̯ is retained as PS *‑aˈu̯ ei̯ [13] (CS *‑ovi [22|29]), preserved in the
old Slavic dialects except in the Old Novgorod area. Here the o-stem ending
was generalised, a tendency also seen elsewhere.
Nominal Inflection 149
PBS PIE *‑ah₂ai̯ regularly yielded PBS *ˌ‑āi̯ [1|3|4]. Lithuanian and Old
Prussian preserve the original ending, whereas Latvian ‑ai represents a gen-
eralisation of the diphthong from monosyllabic pronouns such as tài, šài. The
expected Latvian ending ‑i is found in Old Latvian and in dialects, as well as in
adverbial expressions like pa rùoki ‘convenient’.
PS The ending is retained as PS *ˌ‑āi̯ (CS *‑ě [22|29]; *‑ji [20|22|29]). The end-
ing is preserved intact in the old Slavic dialects except in the Novgorod–Pskov
area, where the soft ending has been replaced with the hard one. The non-desi-
nential accentuation has been analogically replaced with desinential accen-
tuation in most of the Slavic-speaking area; traces of the original accentuation
are preserved in various Slavic dialects (see Olander loc. cit. with references).
PBS Olander 2009: 173: PBS *ˌ‑ōi̯ from PIE *‑óei̯ is regular | Endzelīns 1971:
134: Li. ‑ui is from *‑ōi̯ < *‑oei; Lv. ‑am is from pronouns | Stang 1966:
181–182: Li. ‑ui, dial. ‑ai reflect *‑uoi < PBalt. *‑ōi, also in OPr. ‑u (after velars
and labials), ‑ai (elsewhere); Lv. ‑am is from pronouns | Otrębski 1956:
14: Li. ‑ui < *‑uoi is from PIE *‑ōi, also in Gk. ‑ῳ | Endzelīns 1923: 292:
Lv. ‑am is from pronouns
PS Vondrák 1908/ 1928: 3: Li. ‑ui, Gk. ‑ῳ reflect PIE *‑ōĩ; OCS ‑u is
unclear | Hujer 1910: 129–135: Slavic ‑u is certainly not from PIE dat.
*‑ō̃i, but perhaps from o-stem loc. *‑oũ | Meillet 1924/1934: 407–408:
Slavic ‑u goes back to PIE *‑ōi, also in Li. ‑ui, Gk. ‑ῳ | Vaillant 1958: 31:
Slavic ‑u, Li. ‑ui reflect PIE circumflex *‑ōi | Bräuer 1969a: 22, 73: PIE
*‑ōei̯ > *‑ōi̯ is perhaps regularly reflected in CS *‑u, which is not from stem
*‑o plus loc. ptcl. *‑u | Arumaa 1985: 136–137: Slavic ‑u, ‑ju are not from
PIE *‑ōi, but more likely from CS u-stem loc. *‑u; accent of u-stem loc.
does not fit | Igartua 2005a: 133–140: Slavic ‑u is probably from u-stem
dat.–loc. *‑ō̆u | Aitzetmüller 1978/1991: 78–80, 83: OCS ‑( j)u does not
reflect PIE *‑ōi, but a u-stem loc. form
PBS PIE *‑oei̯ contracted to PBS *ˌ‑ōi̯ [3|4], with regular loss of accent. The
ending is preserved as ‑ui in Lithuanian. In Old Prussian it became ‑u after
velars and labials and ‑ai elsewhere, but the former ending spread at the
expense of the latter. In Latvian, *‑ōi̯ was replaced with the pronominal ending.
PS PBS *ˌ‑ōi̯ developed into *ˌ‑ōu̯ [14] > PS *ˌ‑āu̯ [15] (CS *‑u [22|29]; *‑ju
[20|22|29]), preserved in the old Slavic dialects.
152 Chapter 3
OCS tebě Li. →táu; OLi. táwi; dial. tàvī Ved. →túbhya(m)
ORu. tobě; ONovg. tobě Lv. →tev OAv. →taⁱbiiā, →taⁱbiiō; YAv. ?
OCz. tobě OPr. →tebbei Gk. →σοί, →τοί
La. →tibi; OLa. →tibī, →TIBEI
Go. →þus
Hi. dat.–loc. →tuk (OS)
PIE The original forms of the dative singular of the personal and reflexive
pronouns were reshaped in all branches of Indo-European, but the evidence of
Indo-Iranian, Latin and Balto-Slavic allows us to reconstruct PIE *még̑ ʰi, *tébʰi,
*sébʰi. In Indo-Iranian the forms were extended by a Proto-Indo-Iranian *‑a of
unclear origin, perhaps identical to the one found in o-stem dat. sg. ‑āya; the
forms in PIIr. *‑i̯a were preserved in Avestan and to some extent in Indic, where
they competed with the more widespread forms in ‑yam, probably containing
the pronominal element PIE *hom (§ 3.3.17). In Latin, as in Old Prussian and
dialectally in Lithuanian, the ending *‑ei̯ of the consonant stems and i-stems
was imported. The Germanic forms contain an unclear *‑s. There probably
were no locative forms of the personal and reflexive pronouns in the proto-
language (see e.g. Kapović 2006a: 161).
PBS OPr. mennei, tebbei, sebbei and Li. dial. mùnī, tàvī, sàvī (from *‑ie) con-
tain the reflex of the i- and consonant-stem ending PBS *‑ei̯. The Lithuanian
and Latvian standard forms have PBS *‑i, which is usually not regarded as old,
but identified with the dialectal dative singular ending ‑i, which probably
reflects the original consonant-stem locative singular ending (§ 3.6.1).
In pre-Proto-Balto-Slavic the first-person singular form was remade to
*meni by analogy with the genitive singular; some East Baltic dialects point to
a variant *mun‑ also found in Slavic, with an unclear vowel. In East Baltic the
consonant *u̯ of the genitive singular was introduced in the dative forms of the
second-person singular and reflexive pronouns. Moreover, the stem-vowel *a,
phonetically regular in PBS gen. sg. *tau̯ e, *sau̯ e from PIE *téu̯ e, *séu̯ e [11], was
introduced in the first-person singular form in East Baltic, and in some dialects
also in the second-person singular and the reflexive pronoun.
PS The Slavic languages point to PS *‑ai̯ or *‑āi̯; PS *‑ē would also be pho-
netically possible, but is not very feasible from a morphological point of view.
While PS *‑ai̯ could represent PIE *‑oi̯, found in the o-stem locative singular
and perhaps in the enclitic dative of the personal and reflexive pronouns, PS
*‑āi̯ could represent the merger of ā-stem dat. sg. *‑ah₂ai̯ and loc. sg. *‑ah₂i.
Since the instrumental forms of the personal and reflexive pronouns are based
on the corresponding ā-stem form, it seems reasonable to assume that this
applies to the dative-locative singular as well.
Nominal Inflection 155
OCS si Li. refl. ptcl. ‑s; OLi. Ved. –; Prākrit anaphoric pron. se
ORu. si; ONovg. ? acc.–dat. ‑si OAv. dem. pron. hōi; YAv. hē, šē
OCz. si Lv. refl. ptcl. ‑s Gk. dat. οἱ
OPr. acc. ‑si(?), →‑sin La. –
Go. –
Hi. –
PIE Brugmann 1909–1911: 407–408: PIE *moi, *toi, *soi and *mei, *tei,
*sei | Rasmussen forthc. a § 13: PIE *moi̯, *t(u̯ )oi̯, *s(u̯ )oi̯; OCS mi, ti, si
have *-ei̯ with *e from loc. | Beekes 1995/2011: 233–235: PIE gen. and
dat. 1sg. *h₁moi, 2sg. *toi; but refl. gen. *sei (in OCS si and Go. poss. pron.
seina), dat. *soi | Schmidt 1978: 73–79, 109, 135, 144, 164: OCS 1sg. mi
may reflect PIE *mei̯ and (acute) *moi̯; Gk. shows that the latter form is
original; 2sg. *toi̯ has replaced PIE *teu̯ by analogy with 1sg.; refl. *se/oi̯ is
an old analogical formation to 1sg. *moi̯ | Debrunner & Wackernagel
1930/1975: 473–475, 484–485: Prākrit se corresponds to Gk. οἱ | Rix
1976/1992: 177–178: PIE dat. *moi̯, *toi̯, gen. *mei̯, *tei̯ | Sihler 1995:
372, 376–377: PIE 1sg. mei̯, moi̯(?), 2sg. *tei̯, *toi̯, refl. *soi̯ | Weiss 2009/
2011: 327, 329: La. voc. sg. mī < OLa. MEI is from *moi̯ (or *mei̯); OLa.
tīs is from PIE *toi̯ plus gen. sg. marker *‑s | Krahe 1942/1967: – |
Seebold 1984: – | Kloekhorst 2008a: 584–585, 802–803: Hi. encl. 1ps.
acc.–dat. -mu is from PIE encl. dat. *moi, acc. *mē̆; -u is from 2ps. encl.
acc.–dat. -ttu; Hi. 2ps. acc.–dat. ‑tta, ‑ttu is related to PIE encl. dat. *toi,
acc. *tuē̆
PBS Kapović 2006a: 91, 113, 157: PIE *me/oi̯, *te/oi̯; PBS *mei̯, *tei̯; CS *mi,
*ti | Endzelīns 1971: 187: Li. ‑mi, ‑ti, ‑si, Lv. ‑si‑ go back to *mie, *tie, *sie,
reflecting *mei, *tei, *sei | Stang 1966: 253–254: Li. ‑mi, ‑ti, ‑si reflect
*‑mie, *‑tie, *‑sie from PIE *moi, *toi, *soi or *mei, *tei, *sei; Li. forms in ‑m,
‑t, ‑s may partly reflect acc. *‑me, *‑te, *‑se; Li. ‑i may reflect acute and per-
haps also circumflex *‑ie | Otrębski 1956: 136–137: OLi., Li. dial. ‑m(i),
‑t(i), Li., OLi. ‑s(i) correspond to OCS mi, ti, si, reflecting PIE *moi, *toi,
*soi or *mei, *tei, *sei | Endzelīns 1923: 705–708: Lv. and Li. reflexive
particles reflect *sie < PIE *sei, also in OCS si; OPr. -si is from *sē
PS Vondrák 1908/ 1928: 71: Slavic mi, si, ti go back to PIE *moi, *soi,
*toi | Meillet 1924/1934: 452 | Vaillant 1958: 446–448: OCS mi, si,
Nominal Inflection 157
ti reflect PIE *mei, *sei, *tei | Arumaa 1985: 162, 164, 165: OCS mi, ti, si
reflect *moi, *toi (*tu̯ ei/oi), *soi | Aitzetmüller 1978/1991: 108, 111, 113:
OCS mi does not reflect *moi; OCS si is from *sei
PIE The forms were *moi̯, *toi̯, *soi̯ or *mei̯, *tei̯, *sei̯ in the proto-language.
Greek points to the former reconstructions, while Slavic points to the latter;
PIE *‑oi̯ would have yielded PS **‑ai̯ [7] (CS **‑ě [22|29]). The evidence of the
remaining languages is inconclusive. The Latin possessive pronoun vocative
form mī (OLa. PATER MEI) may reflect PIE *moi̯ or *mei̯. The Old Latin second-
person pronoun genitive form tīs goes back to PIE *toi̯ or *tei̯ plus the geni-
tive singular marker *‑s (for an alleged parallel first-person form mīs see Weiss
2009/2011: 327 n. 7).
PBS The forms found in Balto-Slavic may all reflect PIE *‑ei̯. The Old
Lithuanian forms ‑mi, ‑si, ‑ti for expected *‑mie, *‑sie, *‑tie have probably arisen
as the result of a syntactically conditioned reduction (similarly Stang loc. cit.).
PS PBS *mei̯, *tei̯, *sei̯ were retained as PS *mei̯, *tei̯, *sei̯ (CS *mi, *ti, *si
[22|29]), preserved in the old Slavic dialects.
PIE Brugmann 1909–1911: 186–188: Li. ‑imi, OCS ‑ьmь are from
i-stems | Beekes 1995/2011: 187: PIE *‑eh₁ | Rasmussen forthc. a
§ 2: PIE *‑V́ h₁, *‑h₁ | Debrunner & Wackernagel 1930/1975: 34–36, 205:
158 Chapter 3
harb from PIE *ph₂tr̥bʰi; the final vowel *‑i is indicated by the adverb iwik‘ ‘in
some way’, where the vowel was preserved before the particle ‑k‘ (Schmitt 1981/
2007: 73, 112–113). Greek ‑φι from *bʰi, with instrumental, ablatival and locatival
functions (Schwyzer 1939/1968: 551), is attested in e.g. ὄρεσφι ‘on the mountains’.
PS The expected outcome of PBS *‑mi is PS *‑mi; forms where the ending is
attached directly to the stem are not attested in Slavic. In masculine and neuter
nouns the ending was replaced with the i-stem ending PS *‑imi (CS *‑ьmь [29]),
whereas in feminines the new feminine i-stem ending based on the ā-stems, PS
*‑ii̯ān (CS *‑ьjǫ [20|27|28|29]), was introduced. It is unclear if the introduction
of new gender-based endings found in the attested dialects had already been
carried out in Proto-Slavic, or if Proto-Slavic still had the original ending *‑mi.
PS *gastiˈmi (CS masc. *gostь̀ mь; PBS →*‑ˈimi PIE *‑ih₁, *‑i ̯eh₁
fem. →*kostьjǫ̀)
OCS masc. gostьmь, (→)gostemь; Li. fem. →mintimì, Ved. śúcyā; fem. śúcī; masc.–
fem. →kostьjǫ masc. →žvė�riu neut. →śúcinā
ORu. masc. putьmь; fem. Lv. →avi OAv. aṣ̌ī; YAv. paⁱti, ārmaⁱti(‑ca),
→kostьju; ONovg. masc. OPr. – haš́a
putemъ; fem. →vьrьšьju Gk. –
OCz. masc. hostem; fem. →kosťú, La. abl. →turrī, →turre; OLa.
→‑iú →OPID; Osc. →slaagid
Go. –; OEng. masc. (→)wini;
OSax. masc. (→)hugi
Hi. sg.–pl. →ḫalkit
160 Chapter 3
PIE Brugmann 1909–1911: 187–188, 191–193: PIE *‑ī; OCS ‑mь reflects *‑mi,
alongside *‑bʰi; OCS fem. noštьjǫ is probably based on *nošti on the
model of ā-stem ‑ojǫ | Beekes 1995/2011: 203: PIE *‑ih₁ [implicit
from u-stems] | Rasmussen forthc. a § 6: PIE *‑i̯eh₁, *‑ih₁; PBS
*‑imi | Debrunner & Wackernagel 1930/1975: 145–148: Ved. ‑ī is the old
ending | Rix 1976/1992: – | Sihler 1995: 311–312, 317: PIE *‑i̯eh₁, *‑ih₁;
La. ‑ī < OLa. ‑īd is based on analogy with o-stem abl. sg. | Weiss 2009/
2011: 242, 244: PIE *‑ih₁, not preserved in La. | Krahe 1942/1967: – |
Boutkan 1995: 236, 240, 248: PIE *‑ih₁ may be preserved in OEng. masc. ‑i,
OSax. masc. ‑i | Kloekhorst 2008a: [see C-stems, § 3.7.1]
PBS Olander 2009: 176: PIE *‑íh₁, *‑i̯éh₁ were replaced with *‑ˈimi in pre-
PBS | Endzelīns 1971: 153: Lv., OLi., Li. dial. ‑i may correspond to
Ved. ‑ī, but may also have arisen analogically | Stang 1966: 209: Li.
‑imi from *‑imī�;́ Lv. ‑i is taken from the acc. sg. by analogy with the o‑,
ā‑ and ē-stems | Otrębski 1956: 40–41: Li. ‑imi corresponds to OCS
‑ьmь | Endzelīns 1923: 317–318: Lv., Li. dial. ‑i may correspond to Ved.
‑ī, but they may also be the result of the introduction of the acc. sg. end-
ing by analogy with the o‑, ā‑ and ē-stems
PS Vondrák 1908/1928: 5: Slavic and Li. forms reflect *‑mi | Hujer 1910: 150:
fem. ‑ьjǫ is from ā-stems | Meillet 1924/1934: 419: fem. ‑ьjǫ is modelled
on ā-stem ‑ojǫ | Vaillant 1958: 134–135: PIE *‑ī is preserved in OLi. and
Li. dial. ‑ì, Lv. ‑i, ORu. conj. či; Slavic ‑jǫ is from i̯ā-stems; Slavic ‑mь is the
characteristic masc.–neut. ending | Bräuer 1969a: 155: CS fem. *‑ьjǫ
is based on analogy with ā-stems | Arumaa 1985: 125–126: form with
*‑mi is the older one in Slavic and Baltic | Igartua 2005a: 256–258: PIE
*‑ī from *‑ih₁ may be preserved in OCS adv. radi, ORu. ptcl. / conj. či; Slavic
pron. *‑ijǫ replaces *‑ьmь in fem. nouns | Aitzetmüller 1978/1991: 74:
OCS fem. ‑ьjǫ is from (i̯)ā-stems
§ 3.7.1). The same element seems to have taken over the function of the instru-
mental singular in Balto-Slavic and is also attested in Greek (see § 3.7.1).
PIE Brugmann 1909–1911: 187–188, 191–193: PIE *‑ū; OCS ‑mь is from *‑mi,
alongside *‑bʰi; OCS ‑omь for *‑ъmь is o-stem ending | Beekes 1995/
2011: 203: PIE *‑uh₁ | Rasmussen forthc. a § 6: PIE *‑u̯ eh₁, *‑uh₁; PBS
*‑umi | Debrunner & Wackernagel 1930/1975: 145–148: old ending is
Ved. ‑ū, not attested as a paradigmatic case ending | Rix 1976/1992: – |
Sihler 1995: 321, 323–324: PIE *‑u̯ eh₁(?), *‑uh₁(?); La. ‑ū may reflect PIE
*‑uh₁, remade *‑ūd or loc. *‑ēu̯ , *‑eu̯ i | Weiss 2009/2011: 249, 252:
162 Chapter 3
PIE Like in the i-stems there were two different endings in the Proto-Indo-
European u-stems, originally belonging to different paradigm types. We find
reflexes of PIE *‑uh₁, with zero grade of both stem-suffix and case–number
marker, as well as PIE *‑u̯ eh₁, with full grade of the marker. The latter ending is
represented in both Vedic and Avestan, whereas the former is not preserved in
Vedic (Boutkan loc. cit. leaves open the possibility that the Old English variant
‑u also goes directly back to PIE *‑uh₁). La. ‑ū may continue PIE *‑uh₁, but the
situation in other stems suggests that the ending rather reflects the merger of
loc. sg. *‑ēu̯ and early La. ‑ūd, created analogically on the model of the o-stem
ablative singular (cf. Meiser 1998/2006: 146; Sihler loc. cit.). In Greek there is
no trace of the original locative singular form. Armenian has an instrumental
singular in ‑u, e.g. zgestu, possibly from *‑ubʰi, similar to the ending found in
Balto-Slavic.
PS PBS *‑ˈumi yields PS *‑uˈmi [13] (CS *‑ъmь [29]), regularly preserved in
East and West Slavic. In the Old Church Slavonic manuscripts the u-stem end-
ing has been replaced with the new o-stem ending ‑omь (see § 3.7.5).
Nominal Inflection 163
OCS glavǫ, →glavojǫ; dušǫ, →dušejǫ Li. gálva; def. adj. Ved. →jihvayā́; adv. doṣā́
ORu. →ženoju; →zemleju; ONovg. (sъ) mažą́ ja; East Li. dial. OAv. daēnā, →daēnaiiā;
→Lukoju; (sъ) →bratьjeju runkù YAv. daēna, →daēnaiia
OCz. →rybú, →‑au, →‑ou; →dušú, →‑iú Lv. gal̂vu; def. adj. Gk. adv. κρυφῆ
mazuõ La. abl. →uiā; OLa.
OPr. – →PRAIDAD
Go. –; ONor. dat. skǫr
Hi. –
PS *taˈi ̯ān (CS *tojǫ̀; *jejǫ̀) PBS →*‑ai ̯ā̰N PIE pron. *‑oi ̯ah₂(a)h₁
PIE Brugmann 1909–1911: 190–191, 366: PIE *‑ā, possibly also *‑ām reflected
in BS; OCS ‑ǫ corresponds to Li. ‑ą, dial. ‑u, Lv. ‑u; OCS ‑ojǫ is pron. end-
ing | Beekes 1995/2011: 200, 228: PIE noun ending is *‑h₂eh₁; PIE pron.
ending is *‑oi(?); OCS ‑ojǫ is from pronouns | Rasmussen forthc. a
§ 5: Li. ‑à reflects PBS *‑ā́n, with reduced form of *‑mi from PIE *‑ah₂h₁
(following Vaillant); OCS ‑ojǫ is pron. ending; old noun ending is perhaps
preserved in def. adj. ‑ǫjǫ | Debrunner & Wackernagel 1930/1975:
115–119, 504: Ved. ‑ā is original noun ending, from PIE *‑ā, probably from
164 Chapter 3
stem-suffix *‑ā‑ plus ending *‑V̄ ; ‑ayā is pron., inherited from PIE; Slavic
‑ojǫ is pron. ending; ‑ǫ is noun ending but is strongly in retreat | Rix
1976/1992: – | Sihler 1995: 268: proto-form is unclear | Weiss 2009/
2011: 229: PIE *‑eh₂eh₁ | Krahe 1942/1967: 21: PIE instr. *‑ā is reflected
in ONor. dat. skǫr < *‑u and WGmc. ‑u | Boutkan 1995: 225, 228:
PIE *‑h₂eh₁ yielded PGmc. *‑ō, preserved as dat. ending in NGmc. and
WGmc. | Kloekhorst 2008a: –
PBS Olander 2009: 175–176: PIE *‑ah₂(a)h₁; Gk. κρυφῆ may point to disyl-
labic ending; BS has nasal of uncertain origin; PS *‑ai̯ān is of pron. ori-
gin | Endzelīns 1971: 144, 193: Li. ‑a, East Li. ‑u, Lv. ‑u reflect *‑ā́n; it
is unclear whether OCS rǫkǫ is contracted from rǫkojǫ | Stang 1966:
199, 245: Li. and Lv. forms go back to *‑ā́n; the nasal is characteristic of
BS | Otrębski 1956: 25, 151: Li. ‑a, East Li. ‑u reflect *‑ā́n/m; reflexes of
older pron. form tą́ are found in dialects | Endzelīns 1923: 306: Lv. ‑u
is from PBalt. *‑ān
PS Vondrák 1908/1928: 5: OCS rǫkǫ may correspond to Ved. ‑ā (with loss
of ‑n), but rǫkojǫ is more frequent and required by other Slavic lan-
guages | Hujer 1910: 157–160: IIr. and Gmc. point to *‑ā́; Baltic forms
reflect *‑ā́m; Slavic proto-form in nouns was a pronominal form *‑ojǫ, not
*‑ǫ | Meillet 1924/1934: 398, 435: Slavic noun ending ‑ǫ is an archaism,
corresponding to Baltic endings; pron. ending ‑ojǫ is old | Vaillant
1958: 82, 373–374: PIE *‑ā is preserved in Ved. and Gk., where ‑ῆ has cir-
cumflex tone from dat. sg.; PBS *‑ān is a reduced form of *‑āmi; pron. ‑ojǫ
has replaced older ending ‑ǫ in nouns; Slavic pron. ‑ojǫ, with ‑ǫ replacing
*‑ā, corresponds to Ved. ‑ayā | Bräuer 1969a: 104–105, 127: PIE ending-
less form was replaced with *‑ā́m in PBS; this ending is generally replaced
with pron. *‑ojǫ in Slavic, but preserved in Slavic def. adj., e.g. OCS dobrǫjǫ;
forms like OCS rǫkǫ are probably shortened from *‑ojǫ | Arumaa
1985: 152, 177: original ending is contracted from *‑ā‑ and *‑e; Slavic and
Baltic endings reflect *‑ā́n with unclear nasal; CS *‑ojǫ is pron. end-
ing | Igartua 2005a: 215–220: PIE *‑ā; Slavic ‑ǫ < *‑ām; Slavic ‑ojǫ is
pron. ending | Aitzetmüller 1978/1991: 87, 90, 115: OCS noun ending ‑ǫ
is probably contracted from ‑ojǫ, which has been introduced due to influ-
ence from pron. ending; Baltic preserves original ending *‑ām
alongside more frequent ‑aiiā̆, which, like its Vedic counterpart ‑ayā, has arisen
by analogy with the devī-inflection and the pronominal ending (see Lühr 1991
with references). In Vedic, ‑ā is found in adverbs and occasionally as a para-
digmatic case ending. Certain Greek adverbs seem to be petrified instrumen-
tal forms, e.g. κρυφῆ ‘secretly’. In Latin and Germanic there are no forms that
directly continue the old instrumental singular ending.
If original, the circumflex of Gk. adv. κρυφῆ points to an earlier disyllabic
noun ending, which would most likely reflect PIE *‑ah₂ah₁, with the full grade
of the marker imported from paradigms where it was stressed. If we depart
from a Proto-Indo-European disyllabic ending, a Balto-Slavic unaccented
form, as in Li. gálva, is expected via the mobility law [4]. Since the ending was
remade in Balto-Slavic, however, the accentual evidence from this branch is
not cogent.
PBS PIE *‑ah₂(a)h₁ yielded pre-PBS *‑ā̰ [1|3], to which a nasal of unclear ori-
gin was added, yielding PBS *ˌ‑ā̰N. Some authors (e.g. Vaillant loc. cit.) explain
the nasal as a reduced form of *‑mi, which marks the instrumental singular in
the Balto-Slavic consonant, i- and u-stems. It is indeed possible to assume a
regular loss of word-final *‑i after syllables containing a long vowel, but there
would be no other positive examples of the development. It is more likely, I
believe, that PS prs. 1sg. *‑ān (CS *‑ǫ) does not reflect pre-PBS *‑ō̰mi, corre-
sponding to Ved. ‑āmi, since Li. ‑u shows that Proto-Balto-Slavic had inherited
the unextended ending *‑ō̰ from the proto-language (see § 4.3.3). In the pro-
nouns PIE *‑oi̯ah₂(a)h₁ yielded PBS *‑ai̯ā̰ [1|3|7].
The noun ending was preserved in East Baltic, yielding Li. ‑à (def. adj.
mažą́ ja, East Li. ‑ù) and Lv. ‑u (def. adj. mazuõ). The pronominal ending was
replaced with the noun ending. In Old Prussian the instrumental case was lost.
PS The expected outcome of the noun ending PBS *‑ā̰N is PS *‑ān [13] (CS
*‑ǫ [28]; *‑jǫ [20|27|28]). In the pronouns PBS *‑ai̯ā̰N yielded PS *‑ai̯ān [7|13]
(CS *‑ojǫ [20|27|28|29]; *‑jejǫ [20|27|28|29]). The noun ending may be directly
attested in certain Old Church Slavonic manuscripts, where ‑ǫ, ‑jo are less fre-
quent variants of ‑ojǫ, ‑ejǫ. Many authors, however, assume that ‑( j)ǫ represents
a shortened version of the pronominal ending (e.g. Hujer loc. cit.; Tedesco 1951:
173). Outside Old Church Slavonic, positive evidence for the original noun end-
ing seems to be lacking, although the material is not entirely clear (for the gen-
eral situation in the Slavic languages see Hujer loc. cit.; for Old Church Slavonic
see van Wijk 1931: 179–180; Diels 1932–1934/1963: 175; Vaillant 1948/1964: 98;
for Czech and Slovak see Trávníček 1935: 66–67; according to Holzer 1997: 98,
166 Chapter 3
following Popović 1960: 39–40, 353, reflexes of *‑ǫ are found in western dialects
[Czech, Polish, Slovene, Kajkavian, West Čakavian], while eastern dialects [Old
Church Slavonic, Russian, East Čakavian, Old Serbian] reflect *‑ojǫ).
It is probable that at least some of the instances of ‑jǫ in Old Church Slavonic,
as well as some of the corresponding forms in the other Slavic dialects, are
direct reflexes of the old noun ending PS *‑ān. While this does not necessarily
mean that Proto-Slavic still distinguished a noun ending *‑ān from a pronoun
ending *‑ai̯ān, it is not implausible that this was the case.
OCS →gradomь; →kon’emь Li. lángu; def. adj. Ved. yajñā́, →devéna
ORu. →stolъmъ; →konьmь; ONovg. mažúoju OAv. yasnā; YAv. yasna
(sъ) →Asafъmъ; →bьzumьjemь Lv. tȩ̃vu; def. adj. Gk. –; Dor. (Thera) adv. τεδε
OCz. →chlapem; →oráčem mazuõ La. –
OPr. – Go. dat. daga, ƕammeh; adv.
ƕe; OSax. instr. dagu; OHG
instr. tagu
Hi. sg.–pl. →antuḫšet
tagu; divergence between SSl. ‑omь vs. WSl., ESl. ‑ъmь may point to origi-
nal Slavic *‑a from *‑ō | Otrębski 1956: 14: Li. ‑u is from acute *‑uo <
*‑ō, also in Ved. ‑ā | Endzelīns 1923: 293: Lv. ‑u, Li. ‑u reflect *‑uo < PIE
*‑ṓ, also in Ved. ‑ā
PS Vondrák 1908/1928: 5: Li. ‑ù, ‑úo‑ point to PIE *‑ṓ, perhaps also in CS
*vьčerà; Slavic ‑mь is based on analogy with u-stems | Hujer 1910:
150–151: original ending *‑omь is preserved in SSl., replaced with u-stem
ending *‑ъmь in NSl. | Meillet 1924/1934: 391, 408: PIE *‑ō (as in Li.
‑ù) or *‑ē (as in Ved. paścā́ and perhaps OCS adv. dobrě) | Vaillant
1958: 31–33; 121–122: PIE *‑ē (as in YAv. pasča) or *‑ō; SSl. ‑omь, NSl. ‑ъmь
replace PBS *‑ō, preserved in Li. ‑ù, ‑úo‑, Lv. ‑u and perhaps in OCS adv.
vьčera | Bräuer 1969a: 23, 74: PBS *‑mi is an innovation compared to
instr. endings with *‑bʰi or vowel lengthening; original ending *‑ō per-
haps preserved in OCS adv. vьčera | Arumaa 1985: 137–138: PIE *‑ō,
*‑ē | Igartua 2005a: 140–143: PIE *‑ō/ē from *‑o/eh₁ was replaced with
‑omь in SSl. and by ‑ъmь in WSl. and ESl. | Aitzetmüller 1978/1991: 80,
83: Li. ‑ù, Ved. ‑ā reflect PIE *‑ō, replaced in Slavic with ‑omь, ‑ъmь
PBS PIE *‑oh₁ or, less likely, *‑oeh₁ yielded PBS *‑ō̰ [1|3], preserved in East
Baltic. Lithuanian points to an unaccented ending, which could only be the
regular result of the mobility law [4] if the point of departure was PIE *‑oeh₁;
168 Chapter 3
otherwise it must be secondary. In Slavic the relic form *u̯ ikeˈrā (CS *vьčerà)
points to final accentuation (Vasmer 1924: 179–180), which is the expected out-
come of the most plausible proto-form, PIE *‑oh₁. Stang (1964/1970: 111), how-
ever, regards forms like Ru. górodom, SCr. grȃdom as indications of earlier initial
accentuation in the o-stem instrumental singular in Slavic, as in Lithuanian.
PS PBS *‑ō̰ yielded PS *‑ˈā [13|15], which was lost as a paradigmatic ending in
the attested Slavic languages. In South Slavic a new ending *‑omь, soft *‑jemь,
was created on the basis of the analogy of the u- and i-stem endings *‑ъmь and
*‑ьmь. In West and East Slavic the u-stem ending *‑ъmь, soft *‑jьmь, directly
replaced the original ending.
Some scholars assume that the OCS ‑omь represents the Proto-Slavic ending
and that West and East Slavic have replaced this ending with *‑ъmь (e.g. Hujer
loc. cit.; Holzer 1997: 98). I find it more likely that the inherited ending PS *‑ā
was replaced directly with *‑omь in the south and by *‑ъmь in the north (thus
also Karstien 1936: 110–111; Vaillant loc. cit.; Stang 1964/1970: 110; loc. cit.; 1969b:
28–29; Kortlandt 1982/2011a: 148; Igartua loc. cit.). As suggested by Vasmer (1924:
179–180), the original instrumental singular is probably preserved in the adverb
*u̯ ikeˈrā ‘yesterday’ (CS *vьčerà [21|29]) (thus also e.g. Karstien 1936, with addi-
tional possible examples; Andersen 1998a: 437; but cf. Jakobson 1971: 631, who
supports the old view that Ru. včerá is an original genitive form). If the general
substitution of PS *‑ā with *‑omь and *‑ъmь was a post-Proto-Slavic process,
though, it is somewhat surprising that there are so few traces of the original
ending left.
PIE Brugmann 1909–1911: 177–179, 183–185: OCS kamene reflects PIE zero end-
ing followed by the same postposition *‑e as in Li. rañkoje; OCS ‑i is from
i-stems | Beekes 1995/2011: 187: PIE loc. *‑i and dat. *‑éi originally
reflect one case; PIE also had a loc. with full or long grade of the stem-
suffix and no ending | Rasmussen forthc. a §§ 2, 8: PIE loc. form had
no ending but an accent shift to the final stem-vowel; PBS loc. consisted
of the stem plus a ptcl. | Debrunner & Wackernagel 1930/1975: 41–44,
207, 273–275: PIE *‑i, *‑∅; endingless form is preserved in Ved., Av. n-stem
forms and OCS n-stem ‑ene, where ‑e is a postposition | Rix 1976/1992:
154 | Sihler 1995: 248, 253, 285: PIE *‑i, *‑∅; La. ‑e is from PIE loc. *‑i,
perhaps also from instr. *‑(e)h₁; OCS ‑e may be gen. ending | Weiss
2009/2011: 199, 202, 204: La. ‑e reflects PIE *‑i, alongside *‑∅ | Krahe
1942/1967: 35, 39, 45: ONor. feðr, OEng. mēder point to PIE ending
*‑i | Boutkan 1995: 258, 261–262, 265, 268–271, 274, 278: PIE, PGmc. *‑i
is seen in ONor. feðr | Kloekhorst 2008a: 860: Hi. tagān reflects PIE
endingless *dʰg̑ ʰṓm; [see also dat. sg., § 3.6.1]
PBS Olander 2009: – | Endzelīns 1971: 163–164: Lv. ‑ī, Li. ‑yjè are i-stem
forms | Stang 1966: 221–222: PIE loc. sg. *‑i is preserved in Li. dial.
šùnip | Otrębski 1956: 53–54: Li. ‑yjè is i-stem ending | Endzelīns
1923: 323: Lv. forms are modelled on the i-stems
PS Vondrák 1908/1928: 4 | Hujer 1910: 135, 145–146: Slavic kamene, slovese
are remade from *kame, *slove, going back to endingless forms *kamen,
*sloves, with introduction of *‑en‑, *‑es‑ from other forms | Meillet
1924/1934: 125: Slavic form reflects endingless loc. plus postposition
‑e | Vaillant 1958: 184–185: BS *‑e replaces inherited *‑i by analogy
with gen. sg. *‑es on the pattern of the relationship between i- and u-stem
loc. sg. *‑ēi, *‑ōu and gen. sg. *‑eis, *‑ous | Bräuer 1969b: 9, 54, 72, 83:
Slavic ‑e is probably an endingless loc. form with a postposition; other IE
languages point to *‑i | Arumaa 1985: 112–113: Slavic ‑e did not arise
in *slove from endingless *sloves | Igartua 2005a: 338–341: PIE *‑i
was replaced with gen. ending *‑e in Slavic | Aitzetmüller 1978/1991:
94–95: OCS ‑e either replaced *‑i or was added to an endingless form; ‑e
may be imported from gen. sg. by analogy with identical gen. and loc.
endings in i‑ and u-stems
Germanic, where the dative historically represents the locative, the umlaut in
ONor. feðr, OEng. mēder indicates the former presence of a front vowel.
PS PBS *‑i would have yielded PS *‑i (CS *‑ь [29]), merging with the non-neu-
ter accusative singular. This ending is sometimes assumed to have survived into
attested Slavic dialects, e.g. ORu. nočь ‘by night’ (see Ferrell 1967a: 659–661).
The paradigmatic locative singular forms have the rather unclear ending CS
*‑e or the ending CS *‑i, which seems to be the i-stem locative singular ending
(cf. below for an alternative view). The ending *‑e is often considered to reflect
a postposition *‑en, perhaps also seen in the Baltic locative marker reflected
as Li. ‑e in o-stem langè, ā-stem galvojè etc. However, the Baltic marker has an
acute tone and thus seems to reflect PBS *‑ḛ̄n, which would yield PS *‑ēn [13]
(CS *‑ę [28]). Even if we assume a Proto-Balto-Slavic short-vowel ending *‑en,
the result would be PS *‑en (CS *‑ę [28]), providing only PBS *m, not *n, was
lost after short vowels [19]. Moreover, the development of a postposition into a
case marker in Slavic, in contrast to Baltic, would be unique; in all other stems
the locative singular directly reflects the Indo-European proto-form (similarly
Ferrell 1967a: 658; Orr 1984: 201–202).
A more plausible alternative is that the genitive singular ending CS *‑e <
pre-PS *‑es replaced the original locative singular ending by analogy with the
identical genitive and locative endings of the i‑ and u-stems, CS *‑i and *‑u,
respectively (Vaillant loc. cit.; Ferrell 1967a: 658–659; Kortlandt 1983/2011a: 127;
Aitzetmüller loc. cit.; Orr 1984; Igartua loc. cit. [where “Desinencia protoesl.: *‑e
(< *‑en)” on p. 339 must be a lapsus]; Le Feuvre 2011: 346–347). This replace-
ment was possible only after the monophthongisation of oral diphthongs [22]
or, in other frameworks, after the merger of short and long diphthongs (but cf.
Vaillant loc. cit.); that is, it took place in post-Proto-Slavic. This may explain the
variety of allomorphs found in the Slavic dialects; note the contrast with the
ā-stem genitive singular ending CS *‑y, for which the limited variation makes
it hard to believe that it is the result of a post-Proto-Slavic analogical change
(see § 3.5.4).
As for the alternative locative singular ending CS *‑i, some scholars con-
sider it to be the result of an analogical development similar to the one just
mentioned but based on another proportion, namely the identical forms of
the dative and locative singular in the ā-stems (CS *‑ě) and in the i-stems (CS
*‑i), which led to the introduction of the dative singular ending (CS *‑i) in the
Nominal Inflection 171
locative singular of the consonant stems as well (see Orr 1984: 205–206).
However, in view of the general tendency of the consonant stems to take over
endings from the i-stems, I find it more likely that we are simply dealing with a
simple transfer of the i-stem ending CS *‑i to the consonant stems.
PIE Brugmann 1909–1911: 175–176, 182: OCS ‑i reflects PIE *‑ēi, *‑ē or *‑ei,
alongside *‑ei̯i | Beekes 1995/ 2011: 203: PIE *‑ēi [implicit from
u-stems] | Rasmussen forthc. a § 6: PBS *‑ei̯ | Debrunner &
Wackernagel 1930/1975: 152–157: PIE *‑ēi perhaps preserved in Go. ‑ai,
Ved. Agnāy‑ī ‘Gattin des Agni’; sandhi variant PIE *‑ē represented by
Ved. ‑ā; Ved. ‑au is taken from u-stems, perhaps already in PIIr., but not
in PIE | Rix 1976/1992: 154: Gk. ‑ει for Old Att. ‑ηι has /e/ from gen. sg.
‑εως | Sihler 1995: 311, 313–314: Hom. dat. ‑ηϊ, Att. inscr. ‑ηι reflect PGk.
*‑ēi̯i, consisting of PIE *‑ēi̯ (< *‑ei̯i) and analogical *‑i; Osc. dat. fuutreí con-
tinues old loc. form | Weiss 2009/2011: 242, 245: PIE *‑ēi is not preserved
in La. | Krahe 1942/1967: 30: Go. ‑ai is from PIE loc. *‑ēi | Boutkan
1995: 246–248: Go. dat. ‑ai is from loc. *‑ēi | Kloekhorst 2008a: 105,
376: Hi. adj. ‑ai is from PIE *‑eii
PBS Olander 2009: 177–178: PBS *‑ˈēi̯ goes back to PIE *‑ḗi̯ | Endzelīns 1971:
153: Lv., Li. endings are based on analogy with loc. pl.; Lv. dial. forms in ‑ie
contain PIE *‑ēi with *‑je from ā- and ē-stem loc. | Stang 1966: 195–197,
209–211: Li. ‑yje, Lv. ‑ĩ from *‑īję́ are analogical creations; Žem. ‑ie, OLi. ‑eie,
Lv. dial. ‑ē, ‑ei reflect *‑ēi̯en, containing PIE ending *‑ēi, also in OCS kosti,
plus postposition *en | Otrębski 1956: 41: OLi., Li. dial. forms reflect-
ing ‑ėje contain *‑ēi (also in OCS ‑i) plus postposition *‑ę | Endzelīns
172 Chapter 3
1923: 303, 318: Lv. ‑ī, Li. ‑yje have replaced *‑ie < *‑ē(i̯); Lv. ‑ē, ‑ei are either
based on analogy with ē-stems or reflect old *‑ē or *‑ēi̯
PS Vondrák 1908/1928: 4: OCS ‑i is from *‑ēi | Hujer 1910: 142: Slavic
‑i reflects PIE *‑ēi, alongside *‑ē̃ | Meillet 1924/1934: Slavic ‑i goes
back to *‑ēi | Vaillant 1958: 133: PIE *‑ēi | Bräuer 1969a: 156: PIE
*‑ēi̯ | Arumaa 1985: 126–127: OCS ‑i reflects *‑ēi | Igartua 2005a:
258–260: Slavic ‑i is from PIE *‑ēi | Aitzetmüller 1978/1991: 74: OCS ‑i is
from *‑ēi
PIE In place of the expected ending PIE *‑ei̯i, consisting of the full grade
of the i-stem suffix plus the locative singular marker *‑i, we find *‑ēi̯, which is
probably simply the result of a pre-Proto-Indo-European sound change *‑ei̯i >
*‑ēi̯ (see e.g. Szemerényi 1996/1999: 118). PIE *‑ēi̯ is preserved in Germanic, e.g.
Go. fem. dat. anstai, and in Balto-Slavic.
PIIr. *‑ā represents a sandhi variant of PIE *‑ēi̯. Vedic ‑au is not original; it
was probably copied from the u-stems (see Debrunner & Wackernagel loc. cit.).
In Greek, where the locative function was taken over by the dative, inherited
*‑ēi̯ was recharacterised by the locative marker *‑i, yielding *‑ēi̯i, attested in
Hom. πόληϊ (see Chantraine 1945/1984: 88). Some scholars assume that PIE *‑ēi̯
was preserved in Latin in adverbial expressions like rūrī ‘in the country’ and
temperī ‘early’ (e.g. Meiser 1998/2006: 139); but since most of the examples with
‑ī are old consonant stems, not i-stems, they hardly preserve an archaism, and
it is more likely that the ‑ī was imported from the o-stems (cf. also Leumann
1926–1928/1977: 426–427, 431; Weiss 2009/2011: 214). The paradigmatic ablative
endings OLa. ‑īd, La. ‑ī are based on analogy with the corresponding o-stem
ending ‑ōd (see § 3.5.5). On the other hand, PIE *‑ēi̯ is preserved in the Gothic
feminine dative ending ‑ai. The Hittite adj. dative-locative ending ‑ai is traced
back to PIE loc. *‑ōi by some (e.g. Weitenberg 1984: 381; Kimball 1999: 229 with
references) and to *‑eii by others (Kloekhorst loc. cit.).
PS PBS *‑ˈēi̯ is preserved as PS *‑ˈēi̯ (CS *‑i [22|29]). The ending is retained in
the old Slavic dialects, also in the infinitive marker PS *‑tēi̯ (CS *‑ti).
Nominal Inflection 173
PIE Brugmann 1909–1911: 176–177, 182: OCS ‑u goes back to PIE *‑ēu or *‑eu, or
perhaps *‑eu̯ i, which may be preserved in ORu. adv. domovь | Beekes
1995/2011: 203: PIE *‑ēu | Rasmussen forthc. a § 6: PBS *‑au̯ , PIE
*‑eu̯ (i) | Debrunner & Wackernagel 1930/1975: 152–157: Ved. ‑au, La.
abl. noctū, Go. ‑au, OCS ‑u probably reflect PIE *‑ēu; unclear if Ved. ‑avi
reflects PIE *‑eu̯ i or is an Indic innovation | Rix 1976/1992: 154: Gk.
βασιλεῖ for Hom. βασιλῆϊ has /e/ from gen. sg. ‑έως | Sihler 1995: 321,
324: La. ‑ū is perhaps from PIE *‑eu̯ i, possibly also in Ved. ‑avi and Gk.
‑ει, or from *‑ēu̯ ; [see also instr. sg., § 3.7.3] | Weiss 2009/2011: 249,
252: Ved. ‑au, Go. ‑au reflect PIE *‑ēu; La. diū is from *‑eu̯ (i) | Krahe
1942/1967: 32: Go. ‑au, ONor. ‑i go back to PIE *‑ēu; OEng. ‑a, OSax. ‑0 per-
haps reflect PIE variant *‑ōu | Boutkan 1995: 250, 253, 256–257: Go.
‑au, ONor. ‑i, OEng. ‑a, OSax. ‑o reflect PGmc., PIE *‑ēu | Kloekhorst
2008a: 105: Hi. ‑awi is from PIE *‑eu̯ i
PBS Olander 2009: 177–178: PBS *‑ˈāu̯ reflects PIE *‑ḗu̯ , alongside
*‑éu̯ i | Endzelīns 1971: 158: Li. ‑ūje, Lv. ‑ū, Lv. dial. ‑ui are modelled
on i-stems; Li. dial. ‑uo( je), Lv. dial. ‑uo reflect old loc. sg. followed by *j
and a postposition | Stang 1966: 215–216: Slavic ‑u, Ved. ‑au are from
PIE *‑ō(u); this ending may be preserved in Žem. ‑ọu, ‑ū, which perhaps
also reflects *‑uoję, like Lv. dial. ‑uo; Li. ‑uj(e) < *‑uję is based on analogy
with i-stems | Otrębski 1956: 47: Li. ‑uje contains *‑ę; [see also i-stems,
§ 3.8.2] | Endzelīns 1923: 327: Ved. ‑au, OCS ‑u are from PIE *‑ō(u̯ ); Lv.
‑ū, Lv. dial. ‑ui, Li. ‑ūje are innovations on the model of the i-stem loc. sg.;
an older stage is preserved in Lv. dial. ‑uo, Li. dial. ‑uo( je)
PS Vondrák 1908/1928: 4: OCS ‑u, OLi. ‑uo, Ved. ‑au reflect *‑ōu, probably
also in OCS doma | Hujer 1910: 138–141: Slavic ‑u may reflect PIE
174 Chapter 3
*‑ē̆u (with *‑u for *‑ju due to influence of other forms of the paradigm)
or *‑ō̆u; Slavic adv. doma reflects variant PIE *‑ō̃u; Slavic adv. domovь,
dolovь reflect variant PIE *‑eu̯ i | Meillet 1924/1934: 414: accentua-
tion of Ru. sadú etc. shows that Slavic ‑u is from long *‑ōu; Slavic doma
is from variant *‑ō | Vaillant 1958: 109–110: Slavic ‑u reflects PBS, PIE
*‑ōu | Bräuer 1969a: 146: CS *‑u reflects PIE acute *‑ṓu̯ , alongside PIE
*‑ēu̯ attested in OHG suniu | Arumaa 1985: 126–127: OCS ‑u reflects
*‑ēu or *‑ō[u] | Igartua 2005a: 288–289: Slavic ‑u is from PIE *‑ōu or
*‑ēu | Aitzetmüller 1978/1991: 70–71: OCS ‑u is from *‑āu, with long-
grade vowel shown by the accentuation of e.g. Ru. na beregú
PIE The expected ending PIE *‑eu̯ i was formed by the full grade of the u-stem
suffix followed by the locative singular marker *‑i; this ending is reflected in
Ved. ‑avi, Gk. ‑εϊ, ‑ει and Hi. ‑awi. Another ending PIE *‑ēu̯ is represented by
the Vedic variant ‑au, by Avestan ‑āu, by Germanic forms like Go. ‑au and by
Balto-Slavic. The variant *‑ēu̯ was probably created already in pre-Proto-Indo-
European by analogy with the corresponding i-stem ending *‑ēi̯ < *‑ei̯i. The
Vedic neuter variant in ‑uni is a more recent creation based on analogy with
the n-stems. Latin ‑ū is ambiguous; it may represent the merger of the inno-
vated ablative ending OLa. ‑ūd and the old locative ending *‑ēu̯ or *‑eu̯ i (Sihler
loc. cit.).
PBS The regular reflex of PIE *‑ēu̯ would be *‑ˈi̯āu̯ [10] in Balto-Slavic. The
ending was altered to *‑ˈāu̯ by analogy with the forms of the paradigm that
contained PIE *eu̯ > PBS *au̯ [11]. It is more economical to assume that this
analogy took place in pre-Proto-Balto-Slavic than individually in Baltic and
Slavic, but the latter scenario cannot be excluded. In East Baltic the ending was
modified by the addition of a postposition (see § 3.8.1); it is likely that Žem. ‑ộu
also contains the postposition (cf. Stang loc. cit.). The ending is not preserved
in Old Prussian.
PS PBS *‑ˈāu̯ is retained as PS *‑ˈāu̯ (CS *‑u [22|29]), preserved in the old
Slavic dialects. The Common Slavic adverbs *domovь ‘homeward’, *dolovь
‘downward’, which are sometimes taken to reflect Proto-Indo-European u-stem
locative singular forms in *‑eu̯ i (e.g. Hujer 1909, loc. cit. with discussion and
references), reflect shortened dative singular forms in PS *‑au̯ ei (CS *‑ovi) (see
e.g. Vasmer 1964/1986: 525, 528 with references).
Nominal Inflection 175
PBS In the Baltic languages the original locative ending is not preserved
directly. In the East Baltic paradigmatic locative form a postposition (see
§ 3.8.1) has been added to the ending, yielding Li. ‑ojè, Lv. ‑ã. The Lithuanian
ending ‑ojè, with its long o, points to PBS *‑ˈa̰ i ̯ from PIE *‑ah₂i̯ [1], rather than
to PBS *‑ai̯ from PIE *‑ah₂i [1] (see Seržant 2004: 59–60). Traces of the old loca-
tive ending are also found in Old Lithuanian adessive forms like żmonáip. The
adessive ending ‑aip may likely go back to either *‑ah₂i̯ or *‑ah₂i.
PS PBS *‑ˈa̰ i ̯ regularly yielded PS *‑ˈāi̯ [13] (CS *‑ě [22|29]; *‑ji [20|22|29]), pre-
served in the old Slavic dialects. In the Old Novgorod dialect there are only
few attestations of the original soft ending ‑i, which has in most cases been
replaced with the hard ending ‑ě (Zaliznjak 1995/2004: 98).
PIE Brugmann 1909–1911: 180–181: PIE *‑eĩ and *‑oĩ; OCS ‑ě reflects PIE
*‑oi̯ | Beekes 1995/2011: 212–213: PIE *‑oi; Celtiberian, Osc. *‑ei may be
from pronouns | Rasmussen forthc. a § 4: OCS ‑ě, Li. adv. namiẽ, OPr.
bītai reflect PBS *‑ai from PIE noun ending *‑oi̯ with “heavy prosodics”,
alongside pron. *‑ei̯; Li. ‑e perhaps consists of loc. *‑ẹ̄ plus postposition
*en | Debrunner & Wackernagel 1930/1975: 96: Ved. ‑e is from PIE
*‑oi or *‑ei, which were probably originally the unaccented and accented
forms, respectively | Rix 1976/1992: 139: Gk. dial. (Arcadian) dat. εργοι,
Gk. adv. οἴκοι, Ἰσθμοῖ, Ved. ‑e, OCS ‑ě go back to PIE *‑oi̯; Gk. circumflex
tone is secondary | Sihler 1995: 260–261: Italic forms reflect PIE *‑ei̯ or
*‑oi̯ | Weiss 2009/2011: 223: La. ‑ī is from OLa. ‑ei, reflecting PIE *‑ei̯ or
*‑oi̯ | Krahe 1942/1967: 10: OEng. ‑e probably reflects *‑ī < PIE *‑ei, also
in Gk. adv. ἐκεῖ, Osc. comenei | Boutkan 1995: 184–186, 381–382: PIE
*‑oi is preserved in Gmc., e.g. Go. adv. uta; PIE variant *‑ei is preserved in
e.g. OEng. (on) bergi | Kloekhorst 2008a: 376: Hi. ‑i is C-stem dat. sg.
PBS Olander 2009: 177: PS *ˌ‑āi (CS *‑ě) goes back to PBS *ˌ‑ai̯ (the regular out-
come of PIE *‑ói̯), with analogical length | Endzelīns 1971: 134–135:
OPr. adv. bītai is perhaps from PIE *‑oi, also in Gk. adv. οἴκοι, OCS ‑ě; Li.
namiẽ goes back to *‑ei, also in Gk. dial. adv. οἴκει; Li. ‑e is from acute
*‑en | Stang 1966: 182–183: OPr. adv. bītai is perhaps old o-stem loc. sg.;
Li. adv. namiẽ etc. preserves PIE *‑oi, also in Slavic ‑ě, Gk. adv. οἴκοι, Ved.
‑e, La. adv. domī; Li. ‑e is from *‑ę́ (with obscure acute tone), containing
the old loc. ending plus a postposition *en | Otrębski 1956: 14: Li. adv.
namiẽ reflects PIE *‑oi or *‑ei, also in Gk. adv. οἴκοι, οἴκει; Li. ‑e contains
*‑en | Endzelīns 1923: 293: Lv. ‑ā is from ā-stems; older form seen in
Lv. dial. adv. vakar with *‑i < *‑ie corresponding to Li. ‑ie or ‑ę̀
PS Vondrák 1908/1928: 4: OCS ‑ě, ‑ji, Li. adv. namiẽ, Gk. οἴκοι etc. reflect PIE
circumflex *‑oi | Hujer 1910: 73–74, 135: Slavic ‑ě, ‑ji go back to PIE cir-
cumflex *‑oĩ | Meillet 1924/1934: 408: OCS ‑ě, ‑ji reflect PIE *‑oi, along-
side forms with *e-vocalism | Vaillant 1958: 30–31: Slavic ‑ě, ‑ji are
from PIE *‑oi, also in OPr. adv. schai, Li. adv. namiẽ, OLi., Li. dial. dieviẽpi,
and in adverbs of the type OPr. labbai, Li. labaĩ, OCS dobrě | Bräuer
1969a: 23–24, 74: CS *‑ě reflects PIE circumflex *‑õi̯ | Arumaa 1985:
138–139: CS *‑ě is from PIE *‑oi̯, alongside *‑ei | Igartua 2005a: 143–145:
Slavic ‑ě, ‑ji reflect PIE *‑oi, alongside *‑ei | Aitzetmüller 1978/1991: 81,
84: OCS ‑ě, ‑ji are from *‑ai̯ < loc. *‑oi, perhaps a regional innovation of IIr.,
Slavic and probably Baltic
178 Chapter 3
PBS PIE *‑oi̯ regularly yielded PBS *ˌ‑ai̯ [4|7], which is only fragmentarily
attested in Baltic. In Lithuanian the ending was replaced with a new, etymo-
logically unclear ending ‑è, also in OLi. dangujęjis, probably from *‑ḛ̄n. This
ending may consist of the inherited ending *‑ẹ̄ from PBS *‑ai̯, followed by a
postposition (see § 3.8.1). The original ending is preserved in adverbs such as
namiẽ ‘at home’ and dialectal vãkarie ‘in the evening’, oriẽ ‘in the air’ (see Stang
loc. cit.; Zinkevičius 1966: 403). The Latvian ending ‑ã is taken from the ā-stems,
where it is of analogical origin (see § 3.8.4); it is possible that the adverb vakar
‘yesterday’ preserves the original locative singular form. OPr. adv. bītai ‘at night’
shows the expected form (see e.g. Endzelīns 1944: 86), but it is uncertain if the
word is an o-stem (Stang loc. cit.).
PS PBS *ˌ‑ai̯ was regularly retained as PS *ˌ‑ai̯ (CS *‑ě [22|29]; *‑ji [20|22|29]).
Note that I no longer find it necessary to assume that the vowel was analogi-
cally lengthened in Slavic (in contrast to Olander loc. cit.), since I now consider
any word-final PBS *‑ai̯ to be reflected as CS *‑ě. The ending causes the second
palatalisation [23] of a preceding velar outside the Old Novgorod dialect, e.g.
PS *ˌtakai (CS *tȍk||cě) > OCS tocě. The ending is preserved in the old Slavic
dialects. In the otьcь type with c resulting from the second palatalisation [23]
Nominal Inflection 179
the locative form otьci is analogical for expected *otьcě (see Vermeer 2006: 230
with references in n. 4).
PIE Brugmann 1909–1911: 363: OCS tomь is probably related to Li. dial. tami,
tami‑pe, reflecting *‑(s)mi or *‑(s)min | Beekes 1995/2011: 228–230,
294: PIE *tosmi, *h₁ésmi; Umbr. esmik is from PIE *h₁ésmi plus deictic
*k̑e | Rasmussen forthc. a § 14: PIE *tésmi, *tésmei̯; Ved. tásmin is from
PIE *tésmi plus ptcl. *‑n; Umbr. esme goes back to *ésmei̯; OCS tomь is from
PBS *‑åsmi | Debrunner & Wackernagel 1930/1975: 501: Ved. ‑smin
may reflect original *‑smi plus *n, zero grade of the PIE prep. *en; Ved.
metre may point to earlier *‑smi | Rix 1976/1992: – | Sihler 1995:
377–378, 386–387: PIE *tosmi | Weiss 2009/2011: 337–338, 340: PIE
*tó/ésmi | Krahe 1942/1967: – | Boutkan 1995: – | Kloekhorst
2008a: 426
PBS Olander 2009: – | Endzelīns 1971: 193 | Stang 1966: 241, 246–257:
Li. tamè probably reflects *tami plus *en and corresponds to OCS tomь,
YAv. ahmi; Lv. tài is from fem.; other Lv. forms are also secondary; OLi.
adess. tamp(i) probably reflects old loc. ending *‑ami plus pi; OLi. adess.
tamip(i) may contain old form *tami, but may also be modelled on pers.
pron. manip, tavip, savip | Otrębski 1956: 142–143, 147, 151: Li. jame is
from *jami, preserved in OLi. adess. jam(i)p(i) and corresponding to OCS
jemь | Endzelīns 1923: 387–389: Lv. tai is originally fem. form; Lv. dial.
tajā, dial. tanī are the results of contamination; Lv. dial. tamī and other
forms reflect *tami with ‑ī from the nouns; original form ended in *‑in, as in
Ved. tásmin, OLi. iamimp(i), or in *‑i, as in YAv. aētahmi, Gk. dial. οτιμι, OCS
tomь (which may also go back to *‑in), OLi. iamip (which may be secondary)
180 Chapter 3
PS PBS *‑e/asm‑i yielded PS *‑e/am‑i (CS *‑e/om‑i [29]; *‑jem‑i [20|29]), with
the usual Slavic loss of *s in the interfix. Since PIE *‑Vn sequences seem to have
yielded a nasal vowel in Slavic [19], PS *‑sm‑i cannot correspond directly to
Ved. ‑sm‑in (but cf. Ivšić 1970: 223; Holzer 1980: 10).
PIE Brugmann 1909–1911: 135–136, 289: PIE *‑i, alongside *‑oi or *‑ei | Beekes
1995/2011: 203: PIE *‑ei [implicit from u-stems] | Rasmussen forthc. a
§ 6: PBS, PIE *‑ei̯ | Debrunner & Wackernagel 1930/1975: 157–158: Ved.
‑e, YAv. ‑e, Li. ‑iẽ, OCS ‑i probably reflect PIE *‑oi; Gk. ‑ι, Go. ‑i may have
been modelled on nom. | Rix 1976/1992: 152: Gk. ‑ι has replaced PIE
*‑oi̯ | Sihler 1995: 311, 315–316: PIE *‑e/oi̯; Gk. ‑ι is analogical, but it is
not clear from where; in La. the nom. is used | Weiss 2009/2011: 242,
244: PIE *‑ei is not preserved in La. | Krahe 1942/1967: 27: Go. gast,
Gk. μάντι go back to PIE *‑i | Boutkan 1995: 236, 240, 243–244: per-
haps PIE *‑(e)i, PGmc. *‑ei; Go. ‑∅ for expected *‑i may be o-stem end-
ing | Kloekhorst 2008a: 376–377 [see o-stems, § 3.9.4]
PBS Olander 2009: – | Endzelīns 1971: 153: Li. ‑ie is from *‑ei | Stang
1966: 211: Li. ‑iẽ, OCS ‑i, Ved. ‑e reflect PIE *‑ei | Otrębski 1956: 41: Li. ‑ie,
OCS ‑i reflect *‑ei | Endzelīns 1923: 314: old voc. is not preserved in Lv.,
except perhaps in asini ‘blood’, possibly with ‑i < *‑ie corresponding to Li.
‑iẽ, OCS ‑i, Ved. ‑e
PS Vondrák 1908/ 1928: 4: OCS ‑i, Li. ‑iẽ, Ved. ‑e, ‑i reflect PIE *‑oi or
*‑ei | Hujer 1910: 90: Slavic ‑i, Li. ‑ie reflect PIE *‑ei; forms pointing
to PIE *‑i are innovations | Meillet 1924/1934: 418 | Vaillant 1958:
135: PIE *‑ei; Gk. ‑ι points to a variant *‑i | Bräuer 1969a: 156: PIE
*‑ei̯ | Arumaa 1985: 124: OCS ‑i goes back to *‑ei | Igartua 2005a:
247–248: Slavic ‑i is from PIE *‑ei | Aitzetmüller 1978/1991: 74
PBS PIE *‑ei̯ is preserved as *ˌ‑ei̯ in PBS, yielding Li. ‑iẽ. Latvian has no special
vocative form in the i-stems. The form is not attested in Old Prussian.
PS PBS *ˌ‑ei̯ is continued as PS *ˌ‑ei̯ (CS *‑i [22|29]), preserved in the old Slavic
dialects.
182 Chapter 3
PIE It is likely that the u-stem vocative form ended in *‑eu̯ , parallel to the
i-stem form in *‑ei̯. The comparative evidence alone allows a reconstruc-
tion *‑eu̯ or *‑ou̯ . Go. ‑au, usually taken to point unambiguously to *‑ou̯ , may
Nominal Inflection 183
perhaps also reflect PIE *‑eu̯ (Hansen 2014). Taken at face value the Balto-Slavic
material points to PIE *‑ou̯ rather than *‑eu̯ , but it may easily be explained ana-
logically (see below).
The Gothic variant in ‑u, mostly attested in Greek names, is not original
(Boutkan loc. cit.; Hansen 2014: 159 n. 16). Gk. ‑υ is an innovation based on the
nominative singular. In Latin the nominative form is used. The Hittite form in
‑u is ambiguous; the form in ‑ui may have be remodelled on the o-stems (see
Eichner 1974a: 40–41).
PBS PIE *‑eu̯ would probably have yielded pre-PBS *ˌ‑i̯au̯ [10]. The attested
Baltic and Slavic languages point to PBS *ˌ‑au̯ , which may have arisen in the
individual branches or in pre-Proto-Balto-Slavic by analogy with the forms of
the u-stems where *eu̯ yielded *au̯ [11] (similarly Kortlandt 1983/2011a: 128). The
form in *‑au̯ is regularly continued as Li. ‑au and probably Lv. ‑u. Old Prussian
does not show attestations of u-stem vocative forms.
PS PBS *ˌ‑au̯ is retained as PS *ˌ‑au̯ (CS *‑u [22|29]), preserved in the old
Slavic dialects.
PS PBS *ˌ‑a yielded PS *ˌ‑a (CS *‑o [29]; *‑je [20|29]), preserved in the old
Slavic dialects.
Nominal Inflection 185
PIE Brugmann 1909–1911: 134: OCS ‑e reflects PIE *‑e; in i̯o-stems, OCS
‑ju, Li. ‑jau, Lv. ‑’u are from u-stems | Beekes 1995/2011: 212: PIE
*‑e | Rasmussen forthc. a § 4: OCS ‑e is from PIE *‑e | Debrunner
& Wackernagel 1930/1975: 96–97: Ved. ‑a, Gk. ‑ε, La. ‑e, OCS ‑e go back to
PIE *‑e | Rix 1976/1992: 138: Gk. ‑ε is from PIE *‑e | Sihler 1995: 256,
258: Gk. ‑ε, La. ‑e, BS endings reflect PIE *‑e | Weiss 2009/2011: 200, 221:
La. ‑e is from PIE *‑e | Krahe 1942/1967: 10: Gk. ‑ε, La. ‑e, Go. dag reflect
PIE *‑e | Boutkan 1995: 174: Go. þiudan is from PIE *‑e | Kloekhorst
2008a: 376–377: Hi. ‑i is from accented pre-Hi. *‑é < PIE *‑e
PBS Olander 2009: – | Endzelīns 1971: 135: Lv. tȩ̀v may be from *tēve |
Stang 1966: 183: OPr., Li. ‑e, Lv. tȩ̀v are from PIE *‑e | Otrębski 1956:
14–15: Li. ‑ai is an old emphatic ptcl. | Endzelīns 1923: 293–295: Lv.
cìlvȩ̃k probably reflects *‑e, also in OPr. ‑e, OCS ‑e and most likely Li. ‑e
PS Vondrák 1908/1928: 4: OCS ‑e is from PIE *‑e; OCS ‑ju is from i̯u-stems |
Hujer 1910: 88–89: Slavic ‑e is from PIE *‑e; Li. ‑e, apparently with acute
tone, is puzzling | Meillet 1924/1934: 404 | Vaillant 1958: 33: BS
*‑e is from PIE *‑e | Bräuer 1969a: 24, 74: PIE *‑e; CS i̯o-stem end-
ing *‑ju is from i̯u-stems | Arumaa 1985: 132: PIE *‑e; CS *‑ju is from
u-stems | Igartua 2005a: 111–113: Slavic ‑e reflects PIE *‑e; Slavic ‑ju is
from i̯u-stems | Aitzetmüller 1978/1991: 80, 83: OCS ‑e is from *‑e; OCS
‑ju has ‑u from u-stems
PIE The masculine o-stem vocative singular form ended in *‑e in the proto-
language, which is the e-grade of the thematic vowel. The form is well pre-
served in the ancient Indo-European languages.
186 Chapter 3
As for the Hittite form in ‑i (see Hoffner & Melchert 2008: 75 for attesta-
tions), I find it more attractive to assume that the preservation of PIE *‑e as
Hi. ‑i is due to the special prosody of the vocative form than to derive ‑i from
umlaut in forms like atti(me) or from an analogical process (Eichner 1974a: 39,
72 n. 21), or from accented pre-Hi. *‑é (Kloekhorst loc. cit.). The special develop-
ment in a vocative form is reminiscent of the situation in modern Bulgarian,
where only vocative forms are excepted from the general reduction of unac-
cented /o/ towards [ʊ], e.g. žéno [ˈʒeno] (see e.g. Maslov 1982: 52, 156). For the
special status of the vocative form in general see §§ 1.5.1 and 3.1.
PBS The old vocative is regularly reflected as PBS *ˌ‑e, which yields Li. ‑e and
OPr. ‑e and is probably also reflected in Lv. tȩ̀v. The Lithuanian form in ‑ai is a
more recent formation.
PS PBS *ˌ‑e is continued as PS *ˌ‑e (CS *‑e [29]); the ending causes first pala-
talisation of a preceding velar consonant [21], e.g. PS *ˌbage (CS *bȍže). The
i̯o-stem ending PS *‑i̯au̯ (CS *‑ju [20|22|29]) originates in the i̯u-stems (see
§ 3.9.2). As observed by Meillet (1918a), OCS mǫžь, used as the example of a i̯o-
stem in many expositions, still preserves traces of earlier i̯u-stem inflection in
Old Church Slavonic texts. An adjective i̯o-stem form buje from bujь ‘foolish’ is
attested in the Codex Zographensis (see Diels 1932–1934/1963: 190).
PBS As shown by the Old Lithuanian ending ‑e, PIE *‑h₁e became *ˌ‑e in Proto-
Balto-Slavic. Later the consonant-stem ending was replaced with the i̯o-stem
ending in masculines and the (i̯)ā-stem ending in feminines in Lithuanian. The
ending is not preserved in Latvian and Old Prussian, where the dual category
was lost.
Nominal Inflection 189
PS Slavic does not show any traces of the original ending *ˌ‑e, which was
replaced with the i-stem ending PS *‑ī (CS *‑i [29]). This ending is preserved in
the old Slavic dialects.
PBS PIE *‑ih₁ yielded PBS *‑ī�̰ [1]. The Lithuanian dual forms akì, ausì, syn-
chronically i-stems, may reflect PIE *h₃okʷih₁, *h₂au̯ sih₁ directly, and it is pos-
sible that the same ending is preserved in Lv. abi aci found in old texts (see
Mühlenbach 1903: 20). Since the non-neuter dual i-stem ending was also *‑ih₁
in Proto-Indo-European, however, the question of which ending is reflected in
Baltic is perhaps only of academic interest.
As for the accentual evidence, Lithuanian is ambiguous since both PBS
*aˈkī�̰ and *ˌakī�̰ would result in akì, the former form directly, the latter by
Saussure’s law. In Slavic the form is unaccented, but this accentuation cannot
be original since it requires an accented ending with hiatus in Proto-Indo-
European (see [4]).
Nominal Inflection 191
PS PBS *‑ī�̰ regularly yielded PS *‑ī [13] (CS *‑i [29]). This ending is attested in
the old Slavic dialects alongside PS *‑ai̯ (CS *‑ě), which is the o-stem ending.
According to Hirt (1893: 357) the OCS ‑i may reflect unaccented *‑ě, but I con-
sider this solution unattractive (see § 1.4.2).
The ending *‑ī is also preserved in the dual forms of PS *ˌaka ‘eye’ and *ˌau̯ xa
‘ear’ (CS *ȍko, *ȗxo), viz. *ˌakī, *ˌauxī (CS *ȍči, *ȗši). These nouns are neuter in
the singular but feminine in the dual in Old Church Slavonic (Vaillant 1948/
1964: 112). At an older stage, perhaps still in Proto-Slavic, they were neuter. The
evidence from non-Slavic languages suggests that the words should be regarded
as consonant stems, at least from a historical point of view (see the discussion
in Hujer loc. cit. and, for the Proto-Indo-European situation, Forssman 1969:
42; but cf. Aitzetmüller 1978: 76 n. 107, who considers the words to be neuter
i-stems originally).
́
PIE Brugmann 1909–1911: 200: PIE *‑ī� | Beekes 1995/2011: 216–217: PIE *‑ih₁
[implicit from u-stems] | Rasmussen forthc. a § 6: PBS *‑ī is PIE neut.
ending | Debrunner & Wackernagel 1930/1975: 49–51: Ved. ‑ī etc. prob-
ably result from pre-PIE contraction of stem-suffix *‑i‑ and du. marker
*‑e | Rix 1976/1992: 159: Gk. ‑ει is from *‑ei̯e < PIE *‑ei̯h₁; Ved. ‑ī, Li. ‑i,
OCS ‑i reflect PIE *‑ih₁ | Sihler 1995: 256, 314: Gk. ‑ει is from *‑εε, with
‑ε‑ for *‑ι‑ | Weiss 2009/2011: – | Krahe 1942/1967: – | Boutkan
1995: – | Kloekhorst 2008a: –
PBS Olander 2009: 180: accentuation of PBS *ˌ‑ī�̰ < PIE *‑ih₁ is based on analogy
with ā-stem du. and nom. and acc. pl. of all stems | Endzelīns 1971:
154–155: original ending is preserved in Li. ‑ì, Lv. abi ausi, perhaps also
́
OPr. nozy | Stang 1966: 211: Li. ‑ì is from *‑ī� | Otrębski 1956: 83–84:
192 Chapter 3
́
Li. ‑i is from *‑ī� | Endzelīns 1923: 333: traces of du. ending ‑i are pre-
served in Lv. (abi) ausi, (pìeci) asi
́
PS Vondrák 1908/1928: 5: OCS ‑i, Li. ‑i, Ved. ‑ī are from *‑ī� | Hujer 1910:
84–85: Slavic ‑i, Li. ‑i, Ved. ‑ī, Av. ‑i are from PIE *‑ī�,́ which has arisen by
analogy with o-stem ending | Meillet 1924/1934: 420 | Vaillant
1958: 137: PIE *‑ī | Bräuer 1969a: 156: PIE *‑ī | Arumaa 1985: 129–
130: OCS ‑i goes back to acute *‑ī�,́ either from *‑i‑ plus laryngeal or from
*‑i‑ plus du. ending *‑e | Igartua 2005a: 260: OCS ‑i, Li. ‑i, Ved. ‑ī, Av. ‑i
reflect PIE *‑ih₁ | Aitzetmüller 1978/1991: 76: OCS ‑i is from *‑ī
PIE PIE *‑ih₁ consisted of the zero grade of the stem-suffix followed by the
zero grade of the case–number marker (see § 3.10.1 for the problems related to
the marker). The ending is preserved in Indo-Iranian and Balto-Slavic. Greek
‑ει cannot reflect *‑ih₁, but may possibly continue *‑ei̯e, a recomposed form
containing the full-grade variant of the stem-suffix, *‑ei̯‑, followed by *‑e from
the consonant stems.
PBS PIE *‑ih₁ is reflected as PBS *ˌ‑ī�̰ [1]. In the mobile paradigm the form
became unaccented by analogy with the corresponding ā-stem and neuter
o-stem form and with the nominative and accusative plural forms of all stems.
The ending is preserved in Lithuanian. In Latvian it can be seen after quantifi-
ers in old texts, e.g. devińi nakti ‘nine nights’ (Mühlenbach 1903: 9). It is possible
that OPr. nozy represents an i-stem dual form (Endzelīns 1971: 154–155), but
other interpretations are possible (Mažiulis 2004: 54 considers it an ī/i̯ā-stem).
PS PBS *ˌ‑ī�̰ yielded PS *ˌ‑ī [13] (CS *‑i [29]), preserved in the old Slavic dialects.
PBS PIE *‑uh₁ becomes PBS *ˌ‑ṵ̄ [1], with loss of accent by analogy with the
ā-stems and neuter o-stem forms and with the nominative and accusative plu-
ral forms of all paradigms. Among the Baltic languages the ending is preserved
only in Lithuanian.
PS PBS *ˌ‑ṵ̄ is continued regularly as PS *ˌ‑ū [13] (CS *‑y [29]), preserved in
the old Slavic dialects.
194 Chapter 3
PIE Brugmann 1909–1911: 198–199: PIE acute *‑aí (or *‑oí) | Beekes 1995/
2011: 216–217: PIE *‑eh₂ih₁ | Rasmussen forthc. a § 5: OCS ‑ě, ‑ji, Li. ‑ì,
‑íe‑ are from PBS *‑ā́i ̯ < PIE *‑ah₂ih₁, also in Ved. ‑e | Debrunner &
Wackernagel 1930/1975: 51: Ved. ‑e, OAv. ‑ē, YAv. ‑e, Li. ‑ì, ‑íe‑, OCS ‑ě, OIr.
i-mutation reflect PIE (acute) *‑ai, probably with the same *‑ī as the neu-
ters | Rix 1976/1992: 135: Ved. ‑e, OCS ‑ě go back to PIE *‑eh₂ih₁; Gk.
‑ō is from o-stems, and later ‑ā is based on analogy with nom. pl.; Myc.
‑a‑e is perhaps from *‑eh₂‑ih₁ | Sihler 1995: 272–273, 275: expected
PIE ending is *‑eh₂h₁ or *‑eh₂(h₂)e, but perhaps the ā-stems had no dis-
tinctive du.; Hom. masc. ‑ᾱ is from *‑αε, preserved in Myc. | Weiss
2009/2011: 229 (366): La. ‑ae phonologically or analogically continues PIE
*‑eh₂ih₁ | Krahe 1942/1967: – | Boutkan 1995: – | Kloekhorst
2008a: –
PBS Olander 2009: 179–180: unaccented PBS *‑ā̰i ̯ is the regular reflex of
PIE *‑áh₂ih₁ | Endzelīns 1971: 145–146: Li. ‑i, ‑ie‑, Lv. ‑i reflect *‑ei,
different from Slavic ‑ě | Stang 1966: 199: Li. ‑i, ‑íe‑ reflect PIE
*‑eh₂ī | Otrębski 1956: 84: Li. ‑i probably reflects *‑ai, corresponding
to OCS ‑ě, Ved. ‑e | Endzelīns 1923: 332–333: Lv. ‑i, Li. ‑i, ‑ie‑ are from
*‑ei, also in Welsh dwy; OCS ‑ě is from *‑ai or *‑oi
PS Vondrák 1908/1928: 5: OCS ‑ě, Li. ‑i, ‑ie‑, Ved. ‑e are from *‑ai | Hujer
1910: 82–84: Slavic ‑ě, ‑ji go back to circumflex *‑ai < *‑əi; acute tone of Li.
‑ì, ‑íe‑ is analogical from nom.–acc. du. of other paradigms | Meillet
1924/1934: 399 | Vaillant 1958: 84–85: acute tone of Li. ‑ì, ‑íe‑ is perhaps
analogical from i- and C-stems; IIr. points to *‑ai; long vowel of PBS *‑āi is
Nominal Inflection 195
PBS The reflex of PIE *‑ah₂ih₁ was PBS *ˌ‑a̰ i ̯ [1|3|4]. Lithuanian shows the
glottalisation in forms like rankì (a.p. 2) from *ˈranka̰ i ̯ and in the definite adjec-
tive mažíeji. Endzelīns (1923: 332) reports that Latvian has retained the old dual
form in expressions like abi kãji atspêries ‘mit beiden Füßen sich stemmend’.
PS PBS *ˌ‑a̰ i ̯ is reflected as PS *ˌ‑āi̯ [13] (CS *‑ě [22|29]; *‑ji [20|22|29]), pre-
served in the old Slavic dialects.
OCS grada; kon’a Li. lángu; def. adj. mažúoju Ved. devā́, deváu
ORu. stola; kon’a; ONovg. OLv. acc. divu dārzu OAv. spādā; YAv. zasta
(g)ospodina; kon’a OPr. – Gk. ἀγρώ
OCz. oráčě, ‑e La. num. pl. duo, ambō
Go. –
Hi. –
196 Chapter 3
PBS PIE *‑oh₁ is reflected as PBS *ˌ‑ō̰ [1]. According to the mobility law [4],
a Proto-Indo-European ending consisting of an accented vowel followed by a
laryngeal would retain the accent in Proto-Balto-Slavic. The unaccented form
in Lithuanian and Slavic may be due to the influence of the corresponding
ā-stem and neuter o-stem forms and of the nominative and accusative plural
of all stems. In Baltic the dual is preserved as a paradigmatic category only in
Lithuanian dialects, which show ‑ù, def. adj. ‑úoju. Traces of old dual forms are
found in Latvian dialects as well, especially after the numerals ‘two’ to ‘nine’
e.g. acc. divu dārzu (see Endzelīns loc. cit.).
PS PBS *ˌ‑ō̰ yielded PS *ˌ‑ā [13|15] (CS *‑a [29]; *‑ja [20|27|29]). The ending
was preserved in the old Slavic dialects.
PIE Brugmann 1909–1911: 201: Ved. ‑e, YAv. ‑e, OCS ‑ě, ‑ji reflect PIE *‑oi, which
was probably acute; *‑oi is perhaps also preserved in OEng. num. twā, Gk.
δύο from sandhi variant *‑o[i̯], La. uīgintī | Beekes 1995/2011: 217: PIE
*‑oih₁ | Rasmussen forthc. a §§ 2, 4: Ved., YAv. ‑e, OCS ‑ě go back to PIE
*‑oi̯h₁ | Debrunner & Wackernagel 1930/1975: 51–53: Ved. ‑e consists
of *‑a‑ plus *‑ī | Rix 1976/1992: 141: in Gk. the masc. ending replaces
*‑oe < *‑oi̯e < PIE *‑oi̯h₁, preserved in Ved. yugé, OCS iʒ[ě] | Sihler
1995: 265: Ved. ‑e, OCS ‑ě are from PIE *‑oi̯h₁; Gk. ‑ω is probably masc. end-
ing | Weiss 2009/2011: 209: PIE *‑oih₁ | Krahe 1942/1967: 86: OHG
neut. zwei may reflect PIE *du̯ ai, also in Ved. dvé, OCS dъvě | Boutkan
1995: 306: OHG zwei contains *‑oi | Kloekhorst 2008a: –
PBS Olander 2009: 180: PS *ˌ‑āi̯ is from PIE *‑ói̯h₁, with analogical loss of
accent | Endzelīns 1971: – | Stang 1966: – | Otrębski 1956:
– | Endzelīns 1923: 332, 357–358: Lv. divi dar̃bi may reflect neut.
o-stem ending PBalt. *‑ei or *‑ai; Lv. dial. num. dui̯ may reflect fem. and
neut. form *duvi
PS Vondrák 1908/1928: 5: OCS ‑ě, Ved. ‑e are from PIE *‑oi | Hujer
1910: 81–82: Slavic ‑ě, Ved. ‑e, YAv. ‑e reflect PIE *‑oi, consisting of
*‑o‑ plus *‑ī | Meillet 1924/1934: 410: Slavic ‑ě, ‑ji are from *‑oi or
*‑ai | Vaillant 1958: 45 | Bräuer 1969a: 63, 93: CS *‑ě reflects PIE
*‑oi̯ < *‑o‑ plus *‑ī | Arumaa 1985: 146: OCS ‑ě is from PIE *‑oi < *‑o
plus *‑ī | Igartua 2005a: 147–148: Slavic ‑ě, OPr. num. dwai contain PIE
*‑oī < *‑oih₁ | Aitzetmüller 1978/1991: 85: OCS ‑ě, ‑ji are from PIE *‑oi or
from ā-stems
PBS PIE *‑oi̯h₁ regularly yielded PBS *ˌ‑ai̯ [1|4|7], preserved in Slavic. Note
that the chronology given here, with early loss of word-final laryngeals after *i̯,
produces a result that is in harmony with the mobility law, in contrast to what
I believed earlier (Olander loc. cit.). The ending has not survived in Baltic. It is
sometimes maintained that OPr. num. masc. acc. pl. dwai is an original neuter
dual (Igartua 2005a: 148), but it is more likely that it is simply the nominative
Nominal Inflection 199
plural ending (cf. Stang 1966: 277; for alleged traces of the neuter nominative-
accusative dual ending in East Baltic see Fraenkel 1928: 20).
PS The regular reflex of PBS *ˌ‑ai̯ is PS *ˌ‑ai̯ (CS *‑ě [22|29]; *‑ji [20|22|29]).
This ending has regular reflexes in the old Slavic dialects. In the Old
Novgorod area the neuter ending has been replaced with the masculine end-
ing (see Zaliznjak 1995/2004: 111), a tendency observable in other dialects
as well.
PIE Brugmann 1909–1911: 383–384, 412: PIE *u̯ ē, *u̯ e | Beekes 1995/2011: – |
Rasmussen forthc. a § 13: OCS vě, Li. dial. vèdu from PIE *u̯ ḗ, along-
side *u̯ é | Schmidt 1978: 168–176, 205–206: Go. wit, Žem. vẽdọ point
to PIE *u̯ e; OCS vě points to *u̯ ē, which may have been lengthened by
analogy with 2du. *i̯uh; Ved. vā́m is from PIIr. *vā̆‑am | Debrunner
& Wackernagel 1930/1975: 462–464 | Rix 1976/1992: [see encl. 1du.
and 2du. acc.–dat., § 3.10.10] | Sihler 1995: 373, 375, 381–382: OCS vě is
from PIE *u̯ éh₁; Ved. vā́m is from *vā̆‑ám; Ved. āvám has ā‑ from oblique
stem; [see also encl. 1du. and 2du. acc.–dat., § 3.10.10] | Weiss 2009/
2011: 332: PIE *u̯ eh₁ | Krahe 1942/1967: 51: Go. wit etc. reflect PIE
*u̯ e‑d‑ | Seebold 1984: 25–26: Gmc. forms may point to *wit from PIE
*u̯ e‑de, alongside *u̯ e | Kloekhorst 2008a: –
PBS Kapović 2006a: 91, 113, 148: CS *vě ̑, Av. vā point to PIE *u̯ ḗ; Go. wit points to
PIE *u̯ é | Endzelīns 1971: 189: Žem. vèdu preserves the original form;
Li. mùdu, Lv. dial. mũdui̯ have been remodelled on pl. forms | Stang
1966: 257: Žem. vẽdọ points to PIE *u̯ e; OCS vě points to PIE *u̯ ē; Li. mu‑
is remade based on 2du. ju‑ | Otrębski 1956: 139: ve‑ in Li. dial. masc.
200 Chapter 3
vẽdu, fem. vẽdvi corresponds to OCS vě | Endzelīns 1923: 381: Lv. dial.
mũdui̯ corresponds to Li. mùdu
PS Vondrák 1908/1928: 73–74: Slavic vě is from *u̯ ē̆, also in Li. vèdu, Go.
wit | Meillet 1924/ 1934: 454–455 | Vaillant 1958: 454–455: PBS
acute *u̯ ē goes back to PIE *u̯ ē̆ | Arumaa 1985: 169–170: PIE *u̯ ē or
*u̯ e | Aitzetmüller 1978/1991: 110: OCS vě points to *u̯ ē‑; Go. wit points
to *u̯ e‑; Li. is ambiguous
PBS PIE *u̯ éh₁ is reflected as PBS *u̯ ḛ̄ [1], preserved in Slavic. A shortened vari-
ant PBS *u̯ e is found in Žem. masc. vẽdu, fem. vẽdvi. The standard Lithuanian
forms mùdu, mùdvi have mu‑ from the oblique first-person plural stem; the
second-person dual form jù‑ may also have played a role. In Lithuanian the
numeral ‘two’ has been attached to the forms.
PS PBS *u̯ ḛ̄ yields PS *u̯ ē [13] (CS *vě [29]), preserved in the old Slavic dialects.
PIE Brugmann 1909–1911: 384–385, 409, 412: PIE *i̯ū, *i̯u; Gk. νώ, σφώ, OCS va
are PIE encl. acc. (and gen.–dat.) *u̯ ō used as nom. | Beekes 1995/2011:
– | Rasmussen forthc. a § 13: OCS vy, Li. jùdu are from PBS, PIE *i̯ū́,
alongside *i̯ú | Schmidt 1978: 207–212, 245–246: Ved. yuvám reflects
PIE *i̯uh plus PIIr. ptcl. *‑am; Gmc. forms with *i for *u modelled on
1du.; OCS vy reflects *i̯ū, with v‑ from oblique forms | Debrunner &
Wackernagel 1930/1975: 462–464 | Rix 1976/1992: [see encl. 1du. and
2du. acc.–dat., § 3.10.10] | Sihler 1995: 373, 381–382: PIE *yuh₁, *yūh₁(?);
[see also encl. 1du. and 2du. acc.–dat., § 3.10.10] | Weiss 2009/2011:
332: PIE *i̯uhh₁ | Krahe 1942/1967: 53: Gmc. *jit reflects *jut with *i
from 1du. *wit | Seebold 1984: 26: Gmc. *jit is perhaps from PIE *i̯u‑de,
with *i for *u by analogy with 1du. *wit | Kloekhorst 2008a: –
PBS Kapović 2006a: 77–78, 91, 98–99, 113, 148–149, passim: CS *vy̑ reflects
pre-PS *i̯ū, with *u̯ ‑ from oblique forms; PBS *i̯ū́, *i̯ù(?) go back to PIE
*i̯ū́, *i̯ú | Endzelīns 1971: 189: Li. jùdu, Lv. dial. jũdui̯ have been
remodelled on pl. forms | Stang 1966: 257: Li. jùdu, jùdvi continue
*i̯ū́ | Otrębski 1956: 139 | Endzelīns 1923: 381: Lv. dial. jũdui̯ corre-
sponds to Li. jùdu
PS Vondrák 1908/1928: 73–74: Glagolitic manuscripts use pl. vy for va, origi-
nally an acc. form | Meillet 1924/1934: 455: Slavic does not have
reflexes of *i̯ū, but uses the acc. form va | Vaillant 1958: 454–455:
Slavic vy reflects PIE *i̯ū, with v‑ for *j‑ as in nom. pl.; nom. function of va
is secondary | Arumaa 1985: 170 | Aitzetmüller 1978/1991: 112: Li.
jùdu reflects *i̯ū́; in Slavic *i̯‑ was replaced with *u̯ ‑
PBS PIE *i̯úh yielded PBS *i̯ṵ̄ [1]. It cannot be established whether Li.
masc. jùdu, fem. jùdvi go directly back to PBS *i̯ṵ̄ or if they reflect a shortened
variant *i̯u.
PS PBS *i̯ṵ̄ is reflected as PS *u̯ ū [13] (CS *vy [29]), with *u̯ ‑ from the oblique
forms of the paradigm (Vaillant loc. cit., Schmidt loc. cit., Kapović loc. cit.). I find
it less plausible that vy does not continue the dual form, but represents the
plural form, which replaced older va in certain dialects (thus e.g. Meillet
1918b; loc. cit.; Vondrák loc. cit.; van Wijk 1931: 198; Vaillant 1948/1964: 148–149).
202 Chapter 3
In some Old Church Slavonic traditions the inherited form vy was replaced
with acc. va in order to reestablish the distinction between the nominative
forms of the dual and plural.
It is sometimes stated that Old Russian had va in both the nominative and
accusative (e.g. Černyx 1952/1954: 203; Matthews 1960: 118). However, it seems
that vy is the normal form for the nominative and accusative in Old Russian;
the few instances of va, always in accusative function, occur only in religious
texts (Gadolina 1963: 112–113; see also the interesting observations on the use of
this pronoun in old East Slavic documents by Zaliznjak 1995/2004: 131).
PS *nā (CS *na) PBS *nō̰ PIE encl. obl. *noh₃ (tonic acc. *n̥h₃u̯é)
OCS acc. na, →ny; Li. – (masc. nom.– Ved. encl. acc.–gen.–dat. →nau (tonic
encl. dat. ChSl. na acc. →mùdu; fem. Brāhmaṇa+ →āvā́m)
ORu. acc. na, →ny; →mùdvi) OAv. encl. gen. nā (tonic ?); YAv. encl. ?
ONovg. ? Lv. – (tonic ?)
OCz. acc. →ny OPr. – Gk.—(nom.–acc. →νώ; Hom. →νῶϊ)
La. –
Go.—(acc.–dat. →ugkis)
Hi. –
PS *u̯ā (CS *va) PBS *u̯ō̰ PIE encl. obl. *u̯oh₃ (tonic acc. *uh₃u̯é)
OCS acc. va, →vy; Li.–(masc. nom.– Ved. encl. acc.–gen.–dat. →vām (tonic
encl. dat. va acc. →jùdu; fem. →yuvā́m)
ORu. acc. va, →vy; →jùdvi) OAv. encl. ? (tonic ?); YAv. encl. ? (tonic ?)
ONovg. acc. ?; Lv. – Gk.—(nom.–acc. →σφώ; Hom. →σφῶϊ)
encl. dat. va OPr. – La. –
OCz. acc. →vy Go.—(acc.–dat. →igqis)
Hi. –
Nominal Inflection 203
PIE Brugmann 1909–1911: 383–386, 409, 412: Gk. νώ, σφώ, OCS na, va
reflect PIE encl. acc. (and gen.–dat.) *nō, *u̯ ō | Beekes 1995/2011:
– | Rasmussen forthc. a § 13: OCS na, va reflect PBS *nō, *u̯ ō < PIE
acc.–gen.–dat. *noh₃, *u̯ oh₃ | Schmidt 1978: 182–189, 205–206, 225–
233, 245–246: 1du. PIE acc. *noh, gen., loc. *nohhu, encl. dat. *noh; 2du. PIE
acc. *sgʷʰes, post-PIE *u̯ oh (> PIIr. *vā, OCS va), gen.(‑loc.?) *i̯uheh₂qʷhu,
encl. dat. *u̯ oh | Debrunner & Wackernagel 1930/1975: 477: Ved. vām
has ‑m from forms like tonic acc. du. yuvā́m; expected vā seems to be
attested once | Rix 1976/1992: 179: Gk. σφώ is isolated | Sihler 1995:
373, 381–382: Gk. 1du. νώ, Ved. nāu, OAv. nā continue PIE encl. acc. *nō̆h₁;
2du. PIE encl. acc. *u̯ ō̆h₁; Gk. forms with σφ‑ are obscure | Weiss 2009/
2011: 332: PIE acc. *noh₁, *u̯ oh₁ | Krahe 1942/1967: 51, 53: Gmc. 1du.
*un‑ is from PIE *n̥ ‑, also in Gk. νώ, Ved. nau; Gmc. *k is from acc. sg.; Go.
ugkis, ONor. okr have taken the ending of 1sg. mis, mér; Go. 2du. igqis is
based on 1du., with *i‑ from nom. du. *jit, acc.–dat. pl. Go. izwis etc.; *‑iz
is from acc.–dat. pl. *(u̯ )izu̯ iz | Seebold 1984: 37–39: PIE probably had
encl. 1du. *nō, 2du. *u̯ ō; PGmc. *ink‑, *unk‑ are from PIE *n̥ əu̯ ‑, *iməu̯ ‑ (by
dissimilation from *iu̯ əu̯ ‑) | Kloekhorst 2008a: –
PBS Kapović 2006a: 91, 99, 115, 130, 132, 150: CS *nȃ, *vȃ reflect PIE encl. acc.–
gen.–dat. *noh₁, *u̯ oh₁ | Endzelīns 1971: 189: Li. dial. nuodu may be
compared with OCS na, Gk. νώ | Stang 1966: 257: Li. dial. nuodu attests
stem *nō‑ | Otrębski 1956: 139: Li. dial. nuodu contains nuo‑ corre-
sponding to OCS na, Gk. νώ, Ved. nau | Endzelīns 1923: [see 1du. nom.,
§ 3.10.8; 2du. nom., § 3.10.9]
PS Vondrák 1908/1928: 73: Slavic acc., encl. dat. na reflects *nō | Meillet
1924/1934: 455 | Vaillant 1958: 455: PIE, PBS encl. 1du. *nō, 2du.
*u̯ ō | Arumaa 1985: 170: Slavic va is from *u̯ ō | Aitzetmüller 1978/
1991: 110, 112: OCS 1du. na may be compared to νώ, Av. nā, Ved. nau; OCS
2du. va, which may have replaced vy, is based on na or num. dъva
PIE The Slavic accusative dual forms of the personal pronoun reflect Proto-
Indo-European enclitic oblique forms, 1du. *noh₃ and 2du. *u̯ oh₃. According to
Katz (1998: 38–39, 40–41, 198 n. 9 and passim), the Proto-Indo-European oblique
pronominal forms functioned not only as accusatives, but also as other non-
nominative cases, at least in the dual and plural. The reconstruction PIE *‑oh₃
instead of *‑ō or *‑oh₁/h₂ is based not on external comparison, but on internal
structural arguments, since *noh₃, *u̯ oh₃ can then be connected directly with
the tonic oblique forms *n̥ h₃u̯ é, *uh₃u̯ é (cf. e.g. Cowgill apud Klein 1988: 267
n. 14; Katz 1998: 65–66). The Greek and Germanic forms most likely represent
204 Chapter 3
the old tonic forms (see e.g. Katz 1998: 206–224; for Germanic cf. also Kroonen
2013: 276, 589).
The enclitic pronouns were preserved directly only in Slavic and, for the
second-person dual, probably in Avestan nā (see, however, Katz 1998: 66 with
n. 40). The Vedic forms 1du. nau, 2du. vām clearly contain the same material
as the Slavic and Avestan forms, but the ‑u and the ‑m are not immediately
comprehensible; they may be interpreted as na‑u and *u̯ ā-u (> vām by dissimi-
lation), containing a dual marker *u (Katz 1998: 198 with n. 7).
In two Old Avestan passages, some manuscripts have a variant vā where
most manuscripts have vā̊. The variant vā is sometimes thought to be an accu-
sative dual and to reflect PIE *u̯ oh₃ directly (e.g. Meillet 1920: 209; Schmidt
1978: 228). It is likely, however, that the correct reading is vā̊, an accusative plu-
ral form (Klein 1988: 267 n. 14; Katz 1998: 198 n. 8).
PBS PIE *noh₃, *u̯ oh₃ yielded PBS *nō̰, *u̯ ō̰ [1]. The forms are not preserved
in Baltic.
A Lithuanian dialect form acc. nuodu is often adduced as evidence of the
existence of the stem *nō‑ in Lithuanian (e.g. Otrębski 1956: 139; Stang loc. cit.;
Zinkevičius 1966: 306; Endzelīns loc. cit.). As Sabaliauskas (1976) reports, how-
ever, the form is a typographical error for mudu (see also Kapović 2006: 99).
PS The regular reflexes of PBS *nō̰, *u̯ ō̰ are PS *nā, *u̯ ā [13|15] (CS *na, *va
[29]). The use of na, va as enclitic dative forms is attested in Church Slavonic
texts; that of va also in the Codex Suprasliensis and in the Old Novgorod dialect
(see Dolobko 1925; Vaillant loc. cit.; Kapović 2006: 79; and, for the Old Novgorod
material, Zaliznjak 1995/2004: 722). In some Old Church Slavonic manuscripts,
na and va are used as accusative forms. More commonly, however, we find
ny and vy in this function; ny is originally the first-person plural accusative,
while vy represents the merger of the second-person dual nominative and first-
person plural nominative and accusative. Old Russian shows examples of 1du.
na and ny in accusative function, but in the second-person dual, vy has almost
completely replaced va, which is found only in texts with religious content
(Gadolina 1963: 112–113). In some Slavic dialects, e.g. Old Czech, only the new
forms ny, vy are found.
Nominal Inflection 205
PS *dukterau̯ (CS *dъt’eru) PBS gen.–loc. *‑au̯ PIE loc. *‑hou̯ (gen. *‑hohs)
PIE Brugmann 1909–1911: 206–210, 294, passim: PIE gen. *‑ous, loc.
*‑ou | Beekes 1995/ 2011: 216–217: PIE had gen. *‑h₁e/ohs(?), loc.
*‑h₁ou | Rasmussen forthc. a § 2: PIE gen. *‑oh₃s, loc. *‑oh₃u are
original o-stem endings that have been generalised | Debrunner &
Wackernagel 1930/1975: 55–58: Ved. ‑oḥ, OCS ‑u may reflect PIE gen.–loc.
du. *‑ous; Av. ‑ō, Li. ‑au may point to PIE s-less form; it is unclear if there
were different endings in gen. and loc. | Rix 1976/1992: 160: Gk. *‑oi̯ii̯ n
is taken from o-stems | Sihler 1995: [see o-stems, § 3.11.5] | Weiss
2009/2011: 209–210: PIIr. gen.–loc. or loc. marker *‑hau(s) matches PBS *‑ou
in OCS ‑u, Li. adv. ‑au | Krahe 1942/1967: – | Boutkan 1995: – |
Kloekhorst 2008a: –
PBS Olander 2009: – | Endzelīns 1971: – | Stang 1966: – | Otrębski
1956: [see ā-stems, § 3.11.4] | Endzelīns 1923: [see ā-stems, § 3.11.4]
PS Vondrák 1908/1928: 5–6: the original gen. was perhaps *‑ous, the loc.
was *‑ou | Hujer 1910: 119–120: Slavic ‑u represents merger of PIE
gen. *‑ous (> Li. adv. dvíejaus, Ved. gen.–loc. ‑oḥ) and loc. *‑ou (> Li. adv.
dvíejau, Av. loc. ‑ō; lost in Ved.); Av. gen. ‑ā̊ < *‑ōs is remade based on
nom. | Meillet 1924/1934: 396–397, 399: unclear whether an *‑s has
been lost in Slavic ‑u | Vaillant 1958: 190 | Bräuer 1969b: 9, 54, 72,
83: CS *‑u goes back to PIE *‑ou̯ (s); CS *‑ьju is from i-stems | Arumaa
1985: 119: Baltic-Slavic *‑ou(s) may be an adverbial formation | Igartua
2005a: 343: PIE gen. *‑o(u)s, loc. *‑ou | Aitzetmüller 1978/1991: 96, 98,
101, 104: OCS ‑u is original or from o-stems; [see also u-stems, § 3.11.3]
206 Chapter 3
PIE While Vedic and Slavic have only one case–number marker for the geni-
tive and locative dual, Avestan presents two different markers, e.g. u-stem gen.
du. ahuuā̊ vs. loc. du. aŋhuuō. Most scholars assume that the Avestan situa-
tion is more original than that of the other languages (cf., however, Fritz 2011:
100–101, 200, who reconstructs PIIr. gen.–loc. *‑h‑au from PIE *‑h₁‑ou̯ ). The
Avestan genitive marker ‑ā̊ points to PIIr. *‑ās, while loc. ‑ō points to PIIr. *‑au̯ .
Ved. gen.–loc. ‑oḥ, on the other hand, reflects PIIr. *‑au̯ š. If the hypothesis is
accepted that a final *s centralises a preceding *ā̆ in pre-Proto-Slavic, then CS
gen.–loc. *‑u from PS *‑ā̆u̯ is most straightforwardly explained as the reflex of
an ending that does not end in *‑s; PBS *‑āu̯ s and *‑au̯ s would yield PS **‑ū
and **‑əu̯ [12|17] (CS **‑y and, probably, **‑y [22|29]). Li. pusiaũ, pusiáu and
Lv. pušu (original i̯ā-stem forms) support an s-less reconstruction PBS *‑ā̆u̯; as
pointed out by Vaillant (1958: 38–39; see also Forssman 2001: 146), Li. dvíejaus, a
variant of dvíejau, may have required its final ‑s by analogy with other adverbs
ending in ‑aus. The marker has been remade in Greek (see e.g. Rix 1976/1992:
141; Eichner 1982: 132–138).
With all the caveats that surround the reconstruction of an oblique dual
ending in Proto-Indo-European, we may assume, mainly on the basis of the
Avestan evidence, that the Proto-Indo-European case–number marker was
*‑hohs in the genitive dual and *‑hou̯ in the locative dual (similar reconstruc-
tions are given by Eichner 1982: 10–12, 41–42; Beekes loc. cit.; Rasmussen loc.
cit.). A reconstruction *‑(h)ohu with a laryngeal between *‑o‑ and *‑u (Eichner
1982: 42; Rasmussen loc. cit.) seems to be excluded by PS *‑au̯ (CS *‑u) and PIIr.
*‑au̯ , not PS **‑au̯ u [1|3|7] (**‑ovъ), PIIr. **‑au̯ u. The reflexes of PIIr. *‑ās and
*‑au̯ are preserved in Avestan, whereas in Indic the markers were contami-
nated to *‑au̯ š > Ved. ‑oḥ (Bartholomae 1882: 83; 1895: 218 n. 4; Mayrhofer 1989:
18; Martínez & de Vaan 2014: 44; Rasmussen forthc. a § 4). Balto-Slavic pre-
served the original locative marker only, which took over the functions of the
genitive.
It is possible that the Proto-Indo-European case–number markers are the
original o-stem endings that had spread to the other paradigms already in the
proto-language (e.g. Eichner 1982: 42; Rasmussen loc. cit.). An argument against
this view is the fact that the i-, u- and consonant-stem endings have disyllabic
scansion in Vedic (Debrunner & Wackernagel 1930/1975: 56–57). This is more
easily understandable if we assume that the case–number markers had an ini-
tial laryngeal, i.e. PIE i-stem *‑i‑hohs, loc. *‑i‑hou̯ (see Hoffmann 1976b: 561 n. 2;
Schindler 1989: 25; Weiss 2009/2011: 209–210).
PS *gastii ̯au̯ (CS *gostьju) PBS gen.–loc. *‑ii ̯au̯ PIE loc. *‑ihou̯ (gen. *‑ihohs)
PIE Brugmann 1909–1911: 206–210, 288: PIE gen. *‑(i)i̯ous, loc. *‑(i)i̯ou |
Beekes 1995/2011: [see C-stems, § 3.11.1] | Rasmussen forthc. a: [see
C-stems, § 3.11.1] | Debrunner & Wackernagel 1930/ 1975: 56–57:
disyllabic scansion of Ved. i-stem ‑yoḥ, u-stem ‑voḥ is older than mono-
syllabic form, which is modelled on gen. pl.; Avestan monosyllabic writ-
ing may be “defektive Schreibweise”; OCS ‑ьju, ‑ovu do not prove PIE
strong stem; [see also C-stems, § 3.11.1] | Rix 1976/1992: [see C-stems,
§ 3.11.1] | Sihler 1995: [see o-stems] | Weiss 2009/2011: [see C-stems,
§ 3.11.1] | Krahe 1942/1967: – | Boutkan 1995: – | Kloekhorst
2008a: –
PBS Olander 2009: – | Endzelīns 1971: – | Stang 1966: – | Otrębski
1956: [see ā-stems, § 3.11.4] | Endzelīns 1923: [see ā-stems, § 3.11.4]
PS Vondrák 1908/ 1928: [see C-stems, § 3.11.1] | Hujer 1910: 119–120:
Ved. ávyoḥ shows suffixal zero grade; Slavic ‑ьju has full grade from
gen. pl. ‑ьjь; [see also C-stems, § 3.11.1] | Meillet 1924/ 1934: [see
C-stems, § 3.11.1] | Vaillant 1958: 137 | Bräuer 1969a: 156: PIE
*‑(i)i̯ou̯ (s) | Arumaa 1985: 130 | Igartua 2005a: 260–261: if the dis-
tinction between PIE gen. *‑ious and loc. *‑iou is old, the two forms have
merged in Slavic, where ‑ьju seems to reflect *‑eijou(s) | Aitzetmüller
1978/1991: 76: OCS ‑ьju from *‑ei̯‑; [see also u-stems, § 3.11.3]
208 Chapter 3
PS PBS *‑ii̯au̯ is retained as PS *‑ii̯au̯ (CS *‑ьju [22|29]). The ending is pre-
served in the old Slavic dialects.
PS →*sūnau̯au̯ (CS *synovu) PBS gen.–loc. *‑uu̯au̯ PIE loc. *‑uhou̯ (gen. *‑uhohs)
PIE Brugmann 1909–1911: 206–210, 290: PIE gen. *‑(u)u̯ ous, loc. *‑(u)u̯ ou |
Beekes 1995/ 2011: [see C-stems, § 3.11.1] | Rasmussen forthc. a:
[see C-stems, § 3.11.1] | Debrunner & Wackernagel 1930/1975: [see
i-stems, § 3.11.2; C-stems, § 3.11.1] | Rix 1976/ 1992: [see C-stems,
§ 3.11.1] | Sihler 1995: [see o-stems, § 3.11.5] | Weiss 2009/ 2011:
[see C-stems, § 3.11.1] | Krahe 1942/1967: – | Boutkan 1995: – |
Kloekhorst 2008a: –
Nominal Inflection 209
PIE Like in the corresponding form of the other paradigms, the Avestan for-
mal distinction between a genitive and a locative dual suggests that the Indo-
European proto-language also possessed this distinction. The ending consisted
of the stem-forming suffix of the u-stems followed by the case–number marker
(for which see § 3.11.1). Indo-Iranian points to zero-grade of the suffix, while
Slavic points to full grade. Both ablaut grades are easily explained as analogical
from other endings of the paradigm. I assume that Indo-Iranian preserves the
original situation and Slavic replaced the aberrant allomorph *‑uu̯ ‑ < *‑uh‑(V)
with the allomorph *‑au̯ ‑ found in other forms of the u-stem paradigm.
PS In pre-Proto-Slavic the ending *‑uu̯ au̯ was remade to PS *‑au̯ au̯ (CS *‑ovu
[22|29]) by analogy with the forms of the paradigm that had original full grade.
The ending is preserved in Old Church Slavonic and Old Russian. In the Old
Novgorod dialect the ending is not attested, while in Old Czech the o-stem
form ‑ú is used.
210 Chapter 3
PS *nagāu̯ (CS *nogu; *dušu) PBS gen.–loc. *‑āu̯ PIE loc. *‑ah₂hou̯ (gen. *‑ah₂hohs)
OCS glavu; dušu Li. adv. pusiáu; dial. Ved. gen.–loc. →jihváyoḥ
ORu. ženu; zeml’u; ONovg. adv. pusiaũ OAv. loc. ? (gen. ?); YAv. loc. ?
(bьz dovu) nogutu (read Lv. adv. pušu (gen. →uruuaraiiā̊)
nogatu); ? OPr. – Gk. gen.–dat. →φυγαῖν; Arcadian
OCz. →rybú; →dušú, →‑iú →κραναιυν; Myc. dat. →wa‑na‑so‑i
/wanassojin/(?)
La. –
Go. –
Hi. –
PIE In the ā-stems we may reconstruct PIE gen. du. *‑ah₂hohs and loc. du.
*‑ah₂hou̯ , consisting of the suffix *‑ah₂‑ and the case–number marker (for
which see § 3.11.1). In Indo-Iranian the element *‑ai̯‑ was inserted before the
Nominal Inflection 211
PBS The original genitive ending was lost in Balto-Slavic and its functions
were taken over by the locative PIE *‑ah₂hou̯ > PBS *‑āu̯ (or perhaps *‑ā̰u̯) [1|3].
The Baltic languages do not have a separate paradigmatic genitive or locative
(or genitive-locative) form in the dual, but the original ā-stem locative dual
ending is preserved in the adverb Li. pusiáu, dial. pusiaũ, Lv. pušu, originally
belonging to the paradigms of Li. pusė, Lv. puse.
PS PBS *‑āu̯ was retained as *‑āu̯ in Proto-Slavic (CS *‑u [22|29]; *‑ju
[20|22|29]). The ending is preserved in the old Slavic dialects, with secondary
length in Old Czech (cf. § 3.11.1).
PS *takāu̯ (CS *toku; *goju) PBS gen.–loc. *‑āu̯ PIE loc. *‑ohou̯ (gen. *‑ohohs)
PIE Brugmann 1909–1911: 206–208: PIE gen. *‑ous (or *‑aus), loc. *‑ou (or
*‑au) | Beekes 1995/ 2011: [see C-stems, § 3.11.1] | Rasmussen
forthc. a § 4: PIE gen. *‑oh₃s, loc. *‑oh₃u are preserved in Av.; the two
cases have been contaminated to one in Ved. and Slavic; Slavic ‑u is
from PIE loc. *‑oh₃u; Li. dvíejaus is a contamination of the gen. and
loc. forms | Debrunner & Wackernagel 1930/ 1975: [see C-stems,
§ 3.11.1] | Rix 1976/1992: 141: Gk. *‑oi̯ii̯ n < *‑oi̯sin may be based on
dat.–loc. pl. *‑oi̯si and instr.(?) du. PIE *‑o(i)bʰim | Sihler 1995: 265:
212 Chapter 3
PIE The genitive and locative dual endings consisted of the o-grade of the
thematic vowel followed by the case–number marker, i.e. gen. *‑ohou̯ and loc.
*‑ohohs; for the reconstruction of the markers see § 3.11.1.
In Indo-Iranian the expected endings *‑āu̯ and *‑ās have been replaced with
*‑ai̯au̯ and *‑ai̯ās, containing an element *‑ei̯‑ or *‑oi̯‑ before the case–number
marker. This element may be of pronominal origin or it may stem from the
numeral ‘2’ (Debrunner & Wackernagel 1930/1975: 99–100 with references;
but cf. Sihler loc. cit.), or it may originate in the neuter nominative-accusative
dual (Brugmann loc. cit.). For the Greek forms see § 3.11.1. The Germanic forms
reflect original dual forms, but the endings have been remade (see e.g. Eichner
1982: 10–11, 29–30).
PS PBS *‑āu̯ was retained as PS *‑āu̯ (CS *‑u [22|29], *‑ju [20|22|29]), pre-
served in the old Slavic dialects except Old Czech, where it acquired analogical
length (cf. § 3.11.1).
Nominal Inflection 213
nasalisation in sandhi, the numeral dat. du. dibN points to the former pres-
ence of a final nasal (Lewis & Pedersen 1937/1989: 114). A nasal is also found
in Vedic, but it is excluded by the Avestan and Balto-Slavic forms. Slavic has
a long vowel, whereas Latvian points to a short vowel; the Indo-Iranian and
Lithuanian evidence is ambiguous in this respect. For the recurring problem
concerning *m (Balto-Slavic and Germanic) vs. *bʰi̯ (Indo-Iranian) vs. *bʰ
(elsewhere) see § 3.16.1.
Admittedly slightly biased by the Slavic perspective of this study, I tenta-
tively reconstruct the dative-ablative-instrumental marker as PIE *‑bʰi̯ah₂. The
reconstruction of a-timbre of the vowel is based only on the Baltic evidence,
which is very insecure (see below). The ending is preserved in Avestan and
Balto-Slavic, whereas in Vedic and Celtic it has been extended by a nasal; the *i
pointed to by the Celtic ending may be due to influence from the instrumental
plural.
PBS In Baltic and Slavic the original consonant-stem ending was replaced
with that of the i-stems. Whether this change took place already in Proto-
Balto-Slavic or in the separate histories of the two branches is difficult to say,
but here I assume that the event was pre-Proto-Balto-Slavic. Thus PIE *‑bʰi̯ah₂
yielded pre-PBS *‑mā̰ [1], which was reshaped to PBS *‑imā̰.
As a Proto-Balto-Slavic final *‑m was not preserved as a consonantal seg-
ment in Lithuanian, the Lithuanian case–number marker ‑m must have been
followed by a vowel at an earlier stage. And indeed, in old texts and dialects we
find the endings ‑ma, ‑mu and ‑mi, with plural function in systems where the
dual has been lost (see Zinkevičius 1966: 205–206, 233–234). What the origi-
nal vowel was cannot be clearly seen from the Lithuanian facts. Most authors
regard ‑ma as the oldest ending, as it may reflect *‑mā and correspond to CS
*‑ma (Arumaa 1933: 78; Vaillant 1958: 39; Zinkevičius 1966: 206; Kazlauskas 1968:
132, 172; 1970: 90; but cf. Mažiulis 1966: 44–45). In any case, a vowel that was sub-
ject to apocope in Lithuanian was probably originally short, either reflecting a
Proto-Balto-Slavic short vowel or a long acute vowel. Only the latter option is
compatible with the Slavic evidence.
In Latvian the original dative–instrumental dual endings are used in the
corresponding plural forms (see e.g. Endzelīns 1923: 298–299). Taken at face
value, the Latvian case–number marker ‑m points to a Proto-Balto-Slavic final
short vowel other than *‑u, since long vowels are reflected as short vowels and
*u is preserved in final syllables (see e.g. Endzelīns 1923: 49–50). It is possible,
however, and in harmony with the Slavic evidence, that a Proto-Balto-Slavic
long vowel was first shortened according to the general Latvian sound laws
and then, at a later stage, underwent apocope, parallel to the apocope that
Nominal Inflection 215
PS PBS *‑imā̰ yielded PS *‑imā [13] (CS *‑ьma [29]), preserved in the old
Slavic dialects except Old Novgorod, where it is not attested. For PS *aˈkīmā
‘eyes’ and *au̯ ˈxīmā ‘ears’ (CS *oči̋ma, *uši̋ma) see under the i-stems (§ 3.12.2).
PS *aˈkīmā (CS *oči ̋ma) PBS *‑ˈīmā̰ PIE dat.–abl.–instr. *‑ihbʰi ̯ah₂
̃
OCS očima Li. dat. akìm; instr. akim̃ Ved. →akṣī ́bhyām
ORu. očima; ONovg. ? Lv. dat.–instr. pl. acīm; OAv. aṣ̌ibiiā; YAv. ušibiia
OCz. očima dial. →acim Gk. –
OPr. – La. –
Go. –
Hi. –
PIE Brugmann 1909–1911: 204–206: Ved. akṣī�bhyām,́ OCS očima are based
on nom.–acc. du. | Beekes 1995/ 2011: [see C-stems, § 3.12.1] |
Rasmussen forthc. a § 6: PIIr. *‑ibʰi̯a(‑am) from PIE *‑ibʰ . . . m |
Debrunner & Wackernagel 1930/1975: [see C-stems, § 3.12.1] | Rix
1976/1992: 160; [see also o-stems, § 3.12.5] | Sihler 1995: – | Weiss
2009/2011: [see C-stems, § 3.12.1] | Krahe 1942/1967: – | Boutkan
1995: – | Kloekhorst 2008a: –
PBS Olander 2009: – | Endzelīns 1971: 155 | Stang 1966: 211; [see also
o-stems, § 3.12.5; ā-stems, § 3.12.4] | Otrębski 1956: [see C-stems,
§ 3.12.1] | Endzelīns 1923: 319–320: Lv. dat.–instr. pl. ‑im is older than
‑īm, which has long ī from loc. pl. by analogy with ā- and ē-stems; ‑im is
probably originally a du. ending; [see also o-stems, § 3.12.5]
PS Vondrák 1908/ 1928: 6 | Hujer 1910: 153–154; [see also C-stems,
§ 3.12.1] | Meillet 1924/ 1934: 397, 420 | Vaillant 1958: 137; [see
also o-stems, § 3.12.5] | Bräuer 1969a: 156: PIE *‑imā | Arumaa
1985: 130; [see also C-stems, § 3.12.1] | Igartua 2005a: 261–262: PIE
*‑ibʰ(i̯)ām | Aitzetmüller 1978/1991: 76; [see also u-stems, § 3.12.3]
PBS PIE *‑ibʰi̯ah₂ yielded PBS *‑imā̰ [1], and *‑ihbʰi̯ah₂ yielded *‑ˈī�m ̰ ā̰ [1].
The endings underwent syncope in Lithuanian and Latvian (see § 3.12.1).
The ending with short *‑i‑ was preserved in Latvian dialects as the dative–
instrumental plural form. Li. ‑im may represent the merger of PBS *‑imā̰ and
*‑ˈī�m
̰ ā̰. Similarly, the standard Latvian ending ‑īm, usually regarded as based on
analogy with the ā- and ē-stems (Endzelīns 1923 loc. cit.), may in fact directly
reflect the ending *‑ī�m ̰ ā̰ of the words for ‘eye’ and ‘ear’. As the dual forms of
these words must have been prominent, it is not surprising that their ending
could be generalised.
PS PBS *‑imā̰ yielded PS *‑imā [13] (CS *‑ьma [29]), and *‑ˈī�m ̰ ā̰ yielded *‑ˈīmā
[13] (CS *‑ima [29]). The former ending is preserved in the old Slavic dialects as
the paradigmatic i-stem form, and the latter is continued in the synchronically
irregular words for ‘eye’ and ‘ear’, PS *aˈkīmā and *au̯ ˈxīmā (CS *oči̋ma, *uši̋ma).
PIE The ending *‑ubʰi̯ah₂ consisted of the zero grade of the stem-suffix fol-
lowed by the case marker (see § 3.12.1). The ending is preserved in Indo-Iranian
and Balto-Slavic.
PBS PIE *‑ubʰi̯ah₂ became PBS *‑umā̰ [1]. The ending was preserved in East
Baltic, with apocope of the final vowel (see § 3.12.1). In most Latvian dialects,
including the standard language, a long stem-vowel was introduced on the pat-
tern of the ā- and ē-stems, but the original short vowel is preserved in the dia-
lectal ending ‑um.
PS PBS *‑umā̰ yielded PS *‑umā [13] (CS *‑ъma [29]). The ending is preserved
in Old Church Slavonic and Old Russian. In Old Czech the o-stem form is used.
PBS PIE *‑ah₂bʰi̯ah₂ is regularly reflected as PBS *‑ā̰mā̰ [1]. The ending is pre-
served in East Baltic and Slavic.
PS The reflex of PBS *‑ā̰mā̰ is PS *‑āmā [13] (CS *‑ama [29]; *‑jama [20|27|29].
The ending is not attested in the Old Novgorod dialect, but is preserved in the
remaining old Slavic dialects.
PBS PIE *‑obʰi̯ah₂ yielded PBS *‑amā̰ [1|7]. The ending is preserved, with
apocope of the final vowel, in Lithuanian and, as the dative-instrumental plu-
ral ending, in Latvian. In the latter language the stem-vowel *‑a‑ was replaced
with ‑ie‑ from the pronominal inflection. See § 3.12.1 for details concerning the
development of the case–number marker.
PS PBS *‑amā̰ yielded PS *‑amā [13] (CS *‑oma [29]; CS *‑jema [20|29]). The
ending is not attested in the Old Novgorod dialect, but it is preserved in the
other old Slavic languages.
PBS PIE *‑es remained as PBS *ˌ‑es, preserved in old and dialectal Lithuanian.
In standard Lithuanian the ending was replaced with the i-stem ending ‑ys
(§ 3.13.3) or the ii̯o-stem ending ‑iai (see on the o-stems, § 3.13.7). In Latvian
and Old Prussian the i̯o-stem ending was introduced.
PS Proto-Slavic inherited PBS *ˌ‑es as *‑e [17] (CS *‑e [29]). The ending is pre-
served in the old Slavic dialects, but with a restricted distribution. In feminines
the accusative ending PS *‑ī (CS *‑i) was usually employed, and n-stems often
formed the plural by a collective in PS *‑ii̯e (CS *‑ьje [29]), e.g. OCS kamenьje,
OCz. kamenie.
PIE Brugmann 1909–1911: 235–238: OCS ‑a reflects *‑ā, replacing PIE *‑ə, along-
side other endings | Beekes 1995/2011: 186, 188: PIE endings *‑h₂, *‑∅
are originally sg., collective forms; *h₂ is the nom. sg. of a derived h₂-stem;
*‑∅ with lengthening of preceding vowel is the nom. sg. of an animate,
hysterodynamic variant of a neut., proterodynamic stem | Rasmussen
forthc. a § 2: PIE *‑h₂ (*‑ə₂ after consonants) lengthens preceding vowel
and is sometimes lost | Debrunner & Wackernagel 1930/1975: 61–66:
Ved. ‑i, Gk. ‑α go back to PIE *‑ə; Go. ‑a, OCS ‑a and probably La. ‑a con-
tinue o-stem ending; Ved. and Av. long stem-vowel goes back at least to
PIIr. | Rix 1976/1992: 156: Gk. ‑α is from PIE *‑h₂; La., Gmc., Slavic *‑ā
are from o-stem ending *‑eh₂; a PIE variant with long grade and zero end-
ing, e.g. Av. s-stem manā̊ < *‑ōs, found in s‑, r‑ and n-stems, was replaced
with *‑(e)h₂ in most IE languages | Sihler 1995: 286, 290: Ved. ‑i, Gk.
‑α are from PIE *‑h̥ ₂; La. ‑a < *‑ā continues o-stem ending; Ved. n-stem
nā́mā, OAv. nāmą[m] reflect PIE *‑ōn < pre-PIE *‑onh₂ | Weiss 2009/
2011: 199, 210: PIE ending was *‑h₂; after a sonorant *‑h₂ was lost with
lengthening of preceding vowel; vowel lengthening was generalised in
other stems | Krahe 1942/1967: 46: n-stem Gmc. *‑ōnō goes back to
PIE *‑ōnə, with *‑ō from o-stems | Boutkan 1995: 264, 268, 278, 284–
286: OHG s-stem lembir < *‑iru < *‑ezō reflects PIE *‑esh₂ plus new ending
PGmc. *‑ō; PGmc. n-stem *‑ōnō is from PIE *‑ō̄nh₂ plus new ending PGmc.
*‑ō < *‑eh₂ | Kloekhorst 2008a: 105, 161–162: Hi. ‑a is from o-stems
PBS Olander 2009: – | Endzelīns 1971: – | Stang 1966: – | Otrębski
1956: – | Endzelīns 1923: –
PS Vondrák 1908/1928: 6: OCS ‑a is o-stem ending *‑ā́ | Hujer 1910: 76–78:
IIr. ‑i reflects PIE *‑ə; Slavic ‑a is from o-stems | Meillet 1924/1934:
392–393 [see neuter o-stems, § 3.13.8] | Vaillant 1958: 44–45, 186–187:
original ending was replaced in Slavic or BS with *‑ī from i-stems, traces
of which are preserved in Slavic; Slavic ‑a is from o-stems | Bräuer
1969b: 26, 54–55, 83: CS *‑a reflects PIE o-stem ending *‑ā | Arumaa
1985: 113–114: Slavic ‑a is o-stem ending; PIE ending is preserved in
IIr. and Gk. | Igartua 2005a: 346–347: PIE *‑h₂; Slavic ‑a is from
o-stems | Aitzetmüller 1978/1991: 97–98: OCS ‑a is o-stem ending
*‑V̄ R (Ringe 2006a: 20–21, 42; Weiss loc. cit.), preserved in e.g. Hi. wid/tār, OAv.
dāmąn; this process is reminiscent of Szemerényi’s law (see § 3.3.2). When
the sonorant was vocalic, the laryngeal was retained and subsequently caused
long reflexes of the sonorant in the daughter languages, e.g. Ved. kármā < *‑n̥ h₂.
After other consonants the laryngeal had vocalic reflexes, becoming ‑i in Vedic,
‑α in Greek and, perhaps, ‑a in Latin.
The original distribution was largely lost in the daughter languages; for
instance, OAv., YAv. manā̊ < *‑ōs is based on analogy with the type in PIE *‑ōR <
pre-PIE *‑oRh₂. In Greek and Latin the form reflecting *‑h₂ was generalised and
also spread to the o-stems. In Germanic and Slavic, on the other hand, the orig-
inal consonant-stem formations were replaced with the o-stem ending *‑ah₂.
PS The old Slavic dialects show reflexes of the original neuter ending PS *‑ī
(CS *‑i [29]) from PBS *‑ī�̰ [13] in certain forms, viz. the active present participle,
the comparative and the numeral PS *keˈtūrī (CS *čety̋ ri) (see Vaillant loc. cit.).
This suggests that the consonant-stem ending was not replaced directly with
the neuter o-stem ending, but that in an intermediate period the neuter i-stem
ending was in use (see above). In consonant-stem nouns the neuter o-stem
ending *‑ˈā (CS *‑a [29]) from PBS *‑ˈā̰ [13] had probably already been gener-
alised in Proto-Slavic, as indicated by the unanimous evidence in the old Slavic
dialects.
PS *ˌgastei ̯e (CS masc. (→)*gȍstьje; PBS *ˌ‑ei ̯es PIE *‑ei ̯es
fem. →*kȍsti)
PIE Brugmann 1909–1911: 215–217: OCS ‑ьje goes back to PIE *‑ei̯es; OCS fem.
‑i is acc. ending | Beekes 1995/2011: 203: PIE *‑eies [implicit from
u-stems] | Rasmussen forthc. a § 6: PIE *‑ei̯es; PBS reconstruction is
insecure | Debrunner & Wackernagel 1930/1975: 158–159: Ved. ‑ayaḥ,
OCS ‑ьje, Go. ‑eis, Gk. ‑εες, ‑εις are from PIE *‑ei̯es | Rix 1976/1992: 155:
Gk. ‑εις, Ved. ‑ayaḥ, La. ‑ēs, Go. ‑eis reflect PIE *‑ei̯es | Sihler 1995: 314:
Hom. ‑ηες, Gk. dial. ‑εες are from remodeled *‑ēi̯es; Gk. dial. ‑εες may also
represent original ending *‑ei̯es; La. ‑ēs is from PIE *‑ei̯es | Weiss 2009/
2011: 242, 245: La. ‑ēs, Osc. trís go back to PIE *‑ei̯es | Krahe 1942/1967:
27: Go. ‑eis, ONor. ‑ir, OSax. ‑i, OHG ‑i are from PIE *‑ei̯es | Boutkan
1995: 236, 240, 248–249: Go. ‑eis, ONor. ‑ir, OEng. ‑i, ‑e, OSax. ‑i are from
PGmc., PIE *‑eies | Kloekhorst 2008a: 249: Hi. ‑eš is from PIE *‑eies;
plene writing does not point to a long vowel but denotes hiatus
PBS Olander 2009: 181–182: development of the form in BS is unclear; if
from PIE *‑éi̯es, PBS unaccented *ˌ‑ei̯es may be due to analogy with o-
and ā-stem forms | Endzelīns 1971: 153: Baltic forms reflect *‑ii̯es,
with analogical *i for *e | Stang 1966: 189–190, 211–212: Ved. ‑ayaḥ,
La. ‑ēs, Gk. ‑εις, OCS ‑ьje, Go. ‑eis are from PIE *‑ei̯es; Li. ‑ys, Lv. ‑is, OPr.
‑is have either been modelled on ā- and ė-stems or reflect *‑ii̯es with
analogical *i | Otrębski 1956: 41: Li. ‑ys is perhaps from *‑ii̯es, also
in OCS ‑ьje | Endzelīns 1923: 318: Lv. ‑is, Li. ‑ys and probably OPr. ‑īs
reflect PBalt. *‑ii̯es, corresponding to OCS ‑ьje, Gk. dial. ‑ιες; *‑ii̯es has
replaced original *‑ei̯es under the influence of ī-stems and other forms
containing ‑i‑
PS Vondrák 1908/1928: 6: OCS masc. ‑ьje is from *‑ei̯es; OCS fem. ‑i is acc. end-
ing | Hujer 1910: 62–63: Slavic masc. ‑ьje is from *‑ei̯es; fem. ‑i possibly
reflects *‑īs by analogy with acc. pl. *‑ins on the pattern of ā-stem nom.
pl. *‑ā̃s, acc. pl. *‑ans | Meillet 1924/1934: 419: Slavic masc. ‑ьje, Ved.
‑ayaḥ, Gk. ‑εις, La. ‑ēs, Arm. erek‘ go back to PIE *‑ei̯es; Slavic fem. ‑i is
the acc. pl. ending modelled after the ā-stems | Vaillant 1958: 135–136:
Slavic masc. ‑ьje reflects *‑ii̯es, with generalised *‑i‑ from PIE *‑ei̯es; fem.
‑i is acc. ending | Bräuer 1969a: 156–157: CS masc. *‑ьje is from PIE
*‑ei̯es; CS fem *‑i is acc. ending | Arumaa 1985: 127: CS *‑ьje goes back
to *‑ii̯es, replacing *‑ei̯es, with analogical *i from other forms | Igartua
2005a: 262–266: Slavic masc. ‑ьje is from *‑ii̯es < *‑e[i̯]es, with analogical
*i from other forms of the paradigm; perhaps *‑ei̯es is itself analogical for
*‑eis, preserved in fem. ending ‑i | Aitzetmüller 1978/1991: 74–75: OCS
masc. ‑ьje is from PIE *‑ei̯es; OCS fem. ‑i reflects acc. ending *‑īs
226 Chapter 3
PBS The ending was retained as PBS *ˌ‑ei̯es. The unaccentedness of the form
pointed to by Lithuanian and Slavic is not in accordance with the mobility law
[4]; it may analogical from the nominative plural of the o- and ā-stems. The
endings found in Baltic do not reflect PBS *‑ei̯es but imitate the relationship
between the nominative and accusative plural in the ā‑stems, e.g. pre-Li. acc.
*‑ā̰ns : nom. *‑ās :: acc. *‑ī�n̰ s : nom. X, where X = *‑īs > Li. ‑ys, Lv. ‑is, OPr. ‑is.
PIE Brugmann 1909–1911: 232, 234–235: OCS tri, pron. si, Ved. trī�,́ YAv. ϑrī, Alb.
fem. tri, La. trīgintā, OIr. trí reflect PIE *‑ī (probably acute) from *‑i plus *‑ə;
Gk. ‑α is from C-stems; Italic *‑ā, Gmc. forms are from o-stems; origin of Li.
trýlika is uncertain | Beekes 1995/2011: 237: PIE *trih₂ | Rasmussen
forthc. a § 16: OCS tri reflects PIE *trih₂ | Debrunner & Wackernagel
1930/1975: 63–64, 160–162, 346: Ved. trī�,́ YAv. ϑrī, La. trīgintā, OIr. trí,
OCS tri are inherited; Ved. ‑īni is based on analogy with o-stem ending
‑āni | Rix 1976/1992: 156: Gk. τρία, Ved. trī� ́ reflect PIE *trih₂; La. tria,
Go. þrija are from *‑eh₂ | Sihler 1995: 315, 317: Gk. τρία, Ved. trī�,́ OCS tri
reflect PIE *trih₂; La. tria is from *trīā, replacing expected *trī | Weiss
2009/2011: 210, 367–368: PIE *trih₂ is preserved in OCS tri, La. trī(gintā)
and probably Gk. τρία; La. tria replaces expected *trī | Krahe 1942/
1967: 29, 87: Go. þrija is perhaps from PIE *trii̯ā; OEng. speru, OSax.
aldarlagu have o-stem ending | Boutkan 1995: 239–240, 249: PGmc.
*‑iō; PIE *‑ieh₂ is not preserved in Gmc.; OEng. speru etc. use o-stem end-
ing | Kloekhorst 2008a: 162: PIE *‑ih₂ is perhaps preserved in Hi. pron.
forms in ‑e and Middle Hi. (MS) adj. par‑ku‑e
PBS Olander 2009: – | Endzelīns 1971: 182: Li. trýlika probably preserves
neut. form, also in OCS tri, Ved. trī�,́ La. trīgintā | Stang 1966: 280: first
part of Li. trýlika is identical with OCS tri, Ved. trī�,́ OAv. ϑrī | Otrębski
1956: 166: Li. trý‑ is not clear in all details | Endzelīns 1923: –
́
PS Vondrák 1908/1928: 6: OCS tri retains old ending, also in Ved. trī� | Hujer
1910: – | Meillet 1924/1934: 419, 432, 440: original ending is preserved
in Slavic in num. tri, in comparatives and in dem. pron. si | Vaillant
1958: 186–187, 545–546, 562, 627: traces of i-stem ending *‑ī are preserved
in Slavic prs. act. ptc., in comparative forms and in numerals | Bräuer
1969a: 156–157: Slavic tri, Ved. trī� ́ reflect PIE *‑ī | Arumaa 1985: 190:
OCS tri is from PIE *trī | Igartua 2005a: 244, 266: OCS tri is from PIE
*‑trih₂ | Aitzetmüller 1978/1991: 97, 136: traces of original ending are
preserved in OCS prs. act. ptc., in comparatives and in num. četyri; num.
tri “(mit demselben ‑i wie in oči, uši [. . .])” [but those are du. forms] cor-
responds to Li. trýlika, Ved. trī�,́ Av. ϑrī
PBS PIE *‑ih₂ yielded PBS *‑ī�̰ [1]. In Baltic the ending was only preserved as
the first part of Li. num. trýlika ‘13’.
PS PBS *‑ī�̰ is continued as PS *ˌ‑ī [13] (CS *‑i [29]) in the numeral PS *ˌtrī
(CS *trȋ), in the demonstrative pronoun *sī (CS *si [29]) and perhaps in a few
other formations (see Vaillant loc. cit.). The old Slavic dialects show the regular
reflexes of *ˌtrī.
PIE Brugmann 1909–1911: 215–217: OCS ‑ove reflects PIE *‑eu̯ es | Beekes
1995/2011: 203 | Rasmussen forthc. a § 6: OCS ‑ove is from PBS *‑au̯ es <
PIE *‑eu̯ es | Debrunner & Wackernagel 1930/1975: 158–159: Ved. ‑avaḥ,
OCS ‑ove, Go. ‑jus, Gk. ‑εες, ‑εις are from PIE *‑eu̯ es | Rix 1976/1992: 155:
Gk. ‑εις, Ved. ‑avaḥ, Go. ‑jus go back to PIE *‑eu̯ es | Sihler 1995: 321, 325–
326: Gk. dial. ‑εες, Myc. pa‑ke‑we /pakʰewes/, Ved. ‑avaḥ, OCS ‑ove, Go. ‑jus
are from PIE *‑eu̯ es; La. ‑ūs is either regular by early syncope of the final
vowel or imported from the acc. pl. by analogy with the C-stems | Weiss
2009/2011: 249, 252: La. ‑ūs reflects PIE *‑eu̯ es | Krahe 1942/1967: 33:
Ved. ‑avaḥ, OCS ‑ove, Gk. ‑εις, Go. ‑jus, ONor. ‑ir, OSax. ‑i, OHG ‑i reflect PIE
*‑eu̯ es; OEng. ‑a perhaps reflects PIE variant *‑ou̯ es | Boutkan 1995:
83–86, 250, 257: Go. ‑jus, ONor. ‑ir, OSax. ‑i, OHG. ‑i, OEng. ‑a go back to
PGmc. *‑ewes from PIE *‑eu̯ es | Kloekhorst 2008a: 249–250: Hi. ‑aweš
< *‑eu̯ ēs reflects PIE *‑eu̯ es, with long *ē from i-stems
PBS Olander 2009: 181–182: PBS *ˌ‑eu̯ es is from PIE *‑éu̯ es with analogical
unaccentedness from o- and ā-stems | Endzelīns 1971: 158: Li. ‑ūs, Lv.
‑us reflect Common Baltic *‑uu̯ es with analogical *u, replacing *‑au̯ es pre-
Nominal Inflection 229
served in Li. dial. ‑aus, ‑ous, OCS ‑ove; *a and *o in Baltic *‑au̯ es, Slavic ‑ove
may stem from gen. pl. | Stang 1966: 189–190, 216–217: Li. dial. ‑aus
reflects PIE *‑e/ou̯ es; Li. ‑ūs, Lv. ‑us are based on acc. pl. by analogy with
ā-stems; Lv. ‑i is o-stem ending | Otrębski 1956: 48: Li. ‑ūs may go back
to *‑uu̯ es; Li. dial. ‑aus is gen. sg. ending | Endzelīns 1923: 328–329: Lv.
‑us, Li. ‑ūs are modelled on i-stem ending or reflect ū-stem ending *‑uu̯ es,
also in CS *‑ъve, Ved. ‑uvaḥ, Gk. ‑υες; Li. dial. ‑ous preserves PBalt. *‑au̯ es,
identical with Slavic ‑ove
PS Vondrák 1908/1928: 6: OCS ‑ove, Go. ‑jus, Gk. ‑εες, ‑εις, Ved. ‑avaḥ continue
PIE *‑eu̯ es | Hujer 1910: 63: Slavic ‑ove corresponds to Ved. ‑avaḥ,
Gk. ‑εες, Go. ‑jus | Meillet 1924/1934: 412, 414: Slavic ‑ove, with o for
expected *e before a front vowel, is due to influence from other forms, e.g.
gen. pl. ‑ovъ | Vaillant 1958: 111: Slavic ‑ove goes back to PIE *‑eu̯ es or
*‑ou̯ es | Bräuer 1969a: 147: PIE *‑eu̯ es; Slavic ‑o‑ is analogical from gen.
pl. | Arumaa 1985: 127: CS *‑ove reflects *‑eu̯ es, with regular develop-
ment *eu > *ou in Slavic | Igartua 2005a: 291–293: Slavic ‑ove, Li. dial.
‑aus reflect *‑ou̯ es from PIE *‑eu̯ es, with *o from gen. pl.; Li. ‑ūs is from
*‑uu̯ es with *u from remaining pl. forms | Aitzetmüller 1978/1991: 71
PIE The ending consisted of the full grade of the stem-suffix, *‑eu̯ ‑, followed
by the nominative plural marker *‑es. The ending is preserved in Indo-Iranian,
Greek, Germanic and perhaps Latin (for Latin see Sihler loc. cit.; Meiser 1998/
2006: 146; Weiss loc. cit.). The Hittite ending was remodelled on the i-stems.
PS PBS *ˌ‑au̯ es yielded PS *ˌ‑au̯ e [17] (CS *‑ove [29]), preserved in the old
Slavic dialects. The long final vowel of Old Czech is taken from the i-stems, the
original short vowel being preserved in dialects (see Trávníček 1935: 297–298
with n. 84).
230 Chapter 3
PIE Brugmann 1909–1911: 213–214: PIE *‑ā̃s; OCS ‑y, ‑ję have acc. ending |
Beekes 1995/2011: 200: PIE *‑eh₂es; Gk. and La. forms are due to influence
of pron. o-stem ending *‑oi̯; Slavic ‑y is acc. form | Rasmussen forthc.
a § 5: La. ‑ae is modelled on pron. o-stem ending; PIE *‑ah₂as is not pre-
served in Slavic, where ‑y is from acc. pl. | Debrunner & Wackernagel
1930/1975: 123: PIE *‑ā̃s goes back to *‑ā plus *‑es; Ved. variant ‑āsaḥ is
from o-stems | Rix 1976/1992: 133: PIE *‑eh₂es; Gk. and La. endings are
from pron. | Sihler 1995: 271: PIE *‑eh₂es was replaced with ‑αι in Gk.
and with *‑āi̯ > ‑ae in La. | Weiss 2009/2011: 229, 235: PIE *‑eh₂es; La.
‑ae is modelled on o-stem ending ‑oi (> ‑ī) | Krahe 1942/1967: 21–22:
PIE *‑ãs | Boutkan 1995: 228–229: PIE *‑eh₂es yielded PGmc. *‑ōs; [see
also ā-stem acc. pl., § 3.14.4] | Kloekhorst 2008a: –
PBS Olander 2009: 182: loss of accent in PBS *ˌ‑ās < PIE *‑áh₂as is regu-
lar | Endzelīns 1971: 144: PIE *‑āes | Stang 1966: 200: PBalt. *‑ā̃s
reflects *‑eh₂es | Otrębski 1956: 25–26: PIE *‑ās | Endzelīns 1923:
307
PS Vondrák 1908/1928: 6: Slavic ending is probably acc. pl. | Hujer 1910:
74–76: PIE circumflex *‑ās; Slavic ‑y is acc. pl. ending, based on the pat-
tern of the i-stems | Meillet 1924/1934: 151, 398: Slavic ‑y is from *‑ū <
*‑ās | Vaillant 1958: 83, 87–88: PIE *‑ās; Slavic ‑y, SSl. ‑ję, NSl. ‑jě are
from acc. pl. | Bräuer 1969a: 106, 127: PIE *‑ās was replaced with acc.
pl. in Slavic; PIE soft *‑i̯ās or *‑i̯ēs is not preserved in Slavic; CS *‑ję, *‑jě
are taken from acc. pl. | Arumaa 1985: 153: PIE *‑ā̃s goes back to *‑ā‑
plus *‑es; OCS ‑y, ‑ę are probably from acc. | Igartua 2005a: 226–228:
Nominal Inflection 231
PIE *‑ās; in Slavic the acc. pl. ending ‑y was introduced by analogy with
fem. i-stems | Aitzetmüller 1978/1991: 88, 90: OCS ‑y is from acc. pl.,
not from PIE *‑ās; OCS ‑ję is from acc. pl.; ORu. ‑jě, OPo. ‑ie are from
PIE *‑i̯ās
PS PBS *ˌ‑ās yielded PS *ˌ‑ə̄ [17] (CS Cl *‑y || ONovg. ‑ě [29]; CS *‑jě). In hard
stems the phonetically regular reflex of PS *‑ə̄ is ‑y everywhere except in the Old
Novgorod dialect, where it is ‑ě (see Olander 2012: 331 and passim). Although
the evidence is less clear than in the case of the ā-stem genitive singular, the
ending ‑ě seems to be the original one in the Old Novgorod dialect, whereas ‑y
is a borrowing from standard Old Russian (see the material in Zaliznjak 1995/
2004: 98–99). In Serbo-Croatian and Slovene the replacement of the hard with
the soft ending can be observed already at the time of the earliest attestations
(Svane 1958: 66–67).
The regular outcome of PS *‑i̯ə̄ in the soft stems is *‑jě in all Slavic dialects.
Reflexes of this ending have been preserved in West and East Slavic. In South
Slavic soft stems the endings of the genitive singular and nominative plural
have been replaced with acc. pl. ‑ję on the pattern of the phonological merger
in ‑y of the genitive singular and the nominative and accusative plural forms of
the hard ā-stems and the i-stems (see § 3.5.4).
232 Chapter 3
PS →*ˌtakəi ̯ (CS *tȍk||ci; *gȍji) PBS noun *ˌ‑ai ̯ PIE pron. *‑oi ̯ (noun *‑oes)
PIE Brugmann 1909–1911: 211–213: PIE nom. *‑ō̃s; OCS ‑i is from PIE
pron. *‑oí | Beekes 1995/ 2011: 212–213: BS has pron. ending *‑oi;
PIE *‑ōs is from *‑oes; older variant *‑oses is preserved in IIr. and
Gmc. | Rasmussen forthc. a §§ 2, 4: PIE noun ending was *‑ōs < *‑oes,
replacing older *‑oi̯ preserved in pronouns; OCS ‑i, OPr. ‑ai, Lv. ‑i, Li. adj. ‑ì,
‑íe‑ are from pron. PIE *‑oi̯, with acute tone as in Gk. ‑οί | Debrunner
& Wackernagel 1930/1975: 100–102, 497–498: Ved. ‑āḥ, Osc. ‑ús, OIr. voc.
firu reflect PIE *‑ōs < *‑oes; Av. ‑ā̆ is not neut. pl. ending; PIIr. *‑āsas prob-
ably consists of old ending *‑ōs plus *‑es; Ved. pron. te corresponds to
OAv. tōi, YAv. te, Gk. τοί, Go. þai, Li. tiẽ, OCS ti | Rix 1976/1992: 140: Gk.
‑οι, La. ‑ī, OLa. ‑oe, TochB yakwi, OIr. fir, Li. ‑ai, OCS ‑i go back to PIE pron.
*‑oi; original noun ending PIE *‑ōs < *‑oes is preserved in Ved. ‑āḥ, Go. ‑os,
Osc. ‑ús, Hi. arunaš, OIr. voc. firu | Sihler 1995: 261: PIE noun ending
was *‑ōs, pron. ending was *‑oi̯ | Weiss 2009/2011: 205, 223–224: noun
ending was *‑ōs < *‑oes; La. ‑ī < ‑ẹ̄ < ‑ei reflects PIE pron. *‑oi | Krahe
1942/1967: 11, 63: Gmc. noun forms in *‑ōs and *‑ōz reflect PIE *‑ōs; Go.
pron. þai is from PIE *toi | Boutkan 1995: 187–191, 305–306: Ved. ‑ās
is from PIE *‑ōs, alongside *‑oses seen in IIr. and Gmc. | Kloekhorst
2008a: 249–250: Hi. ‑eš is from i-stem ending PIE *‑eies
PBS Olander 2009: 181: Slavic unaccented form from PIE *‑ói̯ is regu-
lar | Endzelīns 1971: 135–136: OPr., Li. ‑aĩ, OCS ‑i (< *‑oi) are modelled
on adjectives; Lv. ‑i is identical with Li. adj. ending ‑i | Stang 1966:
66–68, 184, 242: Li. ‑ai, adj. ‑ì, ‑íe‑, pron. ‑iẽ, Lv. ‑i, OPr. ‑ai are from PIE
*‑oi | Otrębski 1956: 3–4, 15: Li. adj. ‑ie‑, ‑i reflect PIE *‑oi; Li. ‑aĩ is based
Nominal Inflection 233
PIE The masculine o-stem nominative plural had different endings for nouns
and pronouns in Proto-Indo-European. Nouns ended in *‑oes, containing the
thematic vowel followed by the nominative plural marker *‑es. The compo-
nents of the pronominal ending *‑oi̯ are less clear; perhaps *oi̯ was an originally
pronominal plural morpheme (Rasmussen loc. cit.). The original distribution
of the endings was preserved in Indo-Iranian, Germanic and, in modified form,
in Anatolian. In other branches one variant or the other was generalised.
In Avestan the noun ending ‑ā̆ is probably a collective formation in PIE
*‑ah₂ (see e.g. Hoffmann 1958/1975: 70; Eichner 1985: 153–157; Mayrhofer 1989:
18; Kellens 1989: 46; Hoffmann & Forssman 1996: 120; but cf. Debrunner &
Wackernagel loc. cit.). The Hittite noun ending ‑eš is from the i-stems, going
back to PIE *‑ei̯es (Melchert 1984: 121–122; Kloekhorst loc. cit.).
PBS The Old Prussian and Latvian endings may reflect PBS *ˌ‑ai̯, the expected
outcome of PIE *‑oi̯ [4|7]. Lithuanian presents different variants of this ending:
nouns have ‑aĩ, adjectives have ‑ì (with the word-internal alternant ‑íe‑ in the
definite form), and pronouns have ‑iẽ. Only the pronominal form corresponds
unambiguously to our expectations. Since PBS *ai̯ in certain cases does seem
to yield Li. ai (see the discussion in Stang 1966: 52–68), the Lithuanian noun
ending ‑aĩ may go directly back to PBS *‑ai̯. However, the final accentuation of
the form in mobile paradigms is not compatible with a short diphthong; it is
234 Chapter 3
perhaps most likely that Li. ‑aĩ represents the old neuter ending PIE *‑ah₂, with
regular final accentuation in Lithuanian by the mobility law [4], later extended
by a somewhat obscure element *‑i (see e.g. Hujer loc. cit.; Olander loc. cit.; and
the overview in Eichner 1985: 157–161).
The acute tone of the adjective ending ‑ì, ‑íe‑ is supported by Lv. pron. tiẽ
and points to the former presence of a laryngeal. The source of this laryngeal is,
however, very difficult to determine; Kortlandt (1993/2009: 148) assumes that
the neuter form “*taʔ was disambiguated to nom. *taʔi and acc. *taʔns”.
PS The expected outcome of PBS *ˌ‑ai̯ in Proto-Slavic is *ˌ‑ai̯ (CS *‑ě [22|29]),
which is not attested anywhere in Slavic. Inherited *ˌ‑ai̯ was apparently altered
to *ˌ‑ai̯s in pre-Proto-Slavic due to the influence of the nominative plural forms
of the other paradigms, possibly including the reflex of the original o-stem
noun ending PIE *‑oes, although it is more likely that this ending had disap-
peared in pre-Proto-Balto-Slavic already. A similar development took place in
Old Norse pron. masc. nom. pl. þeir from PGmc. *þai plus *‑r from the other
paradigms (see e.g. Krahe loc. cit.; and cf. Kroonen 2010: 23, who traces ONor.
þeir back to PGmc. *þai + *‑iz). Similarly, an Old Latin nominative plural end-
ing ‑eis is attested in pronouns and in nouns modified by pronouns, e.g. heisce
magistreis (Weiss loc. cit.). The ending was probably formed by adding the
original nominative plural marker of nouns, *‑s, to the pronominal ending ‑ei
< *‑oi̯.
Pre-PS *ˌ‑ai̯s regularly yielded PS *ˌ‑əi̯ [12|17] (CS *‑i [22|29]; *‑ji [20|22|29)
(see Olander 2012: 332, with references to similar solutions in n. 89). The end-
ing is reflected regularly in the old Slavic dialects, where it invokes the second
palatalisation [23] of stems ending in a velar (except in the Old Novgorod dia-
lect), e.g. PS *ˌtakəi̯ (CS *tȍk||ci) > OCS toci etc.
PIE Most languages point to PIE *‑ah₂ from *‑eh₂, consisting of the e-grade
of the thematic vowel and the marker of the neuter nominative-accusative
plural, *h₂. In Greek the consonant-stem ending was generalised. Latin ‑a may
be the outcome of a regular shortening of *‑ā, preserved in other Italic lan-
guages (Sihler loc. cit.), or it may be the old consonant-stem ending. The Vedic
variant ‑āni has arisen by analogy with the n-stems.
236 Chapter 3
PBS PIE *‑ah₂ regularly yielded PBS *‑ˈā̰ [1]. The ending is probably attested
in OPr. slayo, which may represent the plural of slayan. In Lithuanian, numer-
als like keturiólika seem to preserve the old neuter ending in its word-internal
variant ‑ó‑ and its word-final variant ‑a, both from PBS *‑ˈā̰, the latter with
Leskien’s shortening.
PS The Slavic languages point to PS *‑ˈā (CS *‑a [29]; *‑ja [20|27|29]), the
regular outcome of PBS *‑ˈā̰ [13].
PIE Brugmann 1909–1911: 383, 386–387, 410–412: IIr. forms point to PIE *u̯ ei;
Li. mẽs, OPr. mes, OCS my have m‑ from 1sg. forms; OCS my has ‑y from
acc. ny | Beekes 1995/2011: 233–234: PIE *uei; OCS my has m‑ from ver-
bal 1pl. ending and ‑y from 2pl. vy | Rasmussen forthc. a § 13: OCS my,
with ‑y from ny, goes back to PBS *mḗs, which has *m‑ from 1sg., *‑ē‑ from
du. and *‑(e)s from pl.; PIE form was *u̯ éi̯ | Schmidt 1978: 167–176,
178–180, 204–206: PIE *u̯ ei̯, alongside older *mes; OCS my reflects acc.
form *mōns, also in Li. mùs, OPr. mans | Debrunner & Wackernagel
1930/1975: 466–467: Ved. vayám, Go. weis point to PIE *u̯ ei̯ | Rix 1976/
1992: 178–179: Go. weis, Hi. wēš point to PIE *u̯ éi̯s; Gk. ἡμεῖς and Aeol. ἄμμες
reflect *n̥ smees and *n̥ smes, based on the acc. form | Sihler 1995:
373, 380: PIE *u̯ ei; Gk. forms are based on acc.; La. nōs reflects the encl.
acc. form | Weiss 2009/2011: 330: Hi., Gmc. and IIr. point to PIE *u̯ ei̯‑,
whereas BS and Arm. point to a proto-form beginning with *m‑; La. nōs is
the acc. form | Krahe 1942/1967: 51: Go. weis goes back to PGmc. *u̯ īz,
PIE *u̯ ei̯es; PGmc. *u̯ iz in ONor. vér, OEng. wē̆ etc. has been shortened
in unaccented position | Seebold 1984: 27–30: IE languages point to
Nominal Inflection 237
*u̯ ei and a younger form *mes, probably from *u̯ es; Go. and Old Swedish
point to PGmc. *wīz or *weiz from PIE *u̯ ei plus *‑s, while West Nordic
and WGmc. point to PGmc. *wiz or *wez from PIE *u̯ es, alongside more
recent *mes; Slavic form has acc. ending | Kloekhorst 2008a: 115–116,
1004: Hi. wēš reflects PIE *uei(e)s, *uei
PBS Kapović 2006a: 56, 75, 87–90, 154–155, 158, 161: Armenian and BS point to
PIE *més, alongside *wéi̯, perhaps also *wéi̯(e)s; CS *my̑ is modelled on
2pl. *vy̑ | Endzelīns 1971: 188: long vowel of Lv. mẽs, Li. mẽs, dial. mė̃s
is analogical from 2pl.; original short vowel is preserved in OPr. mes, Lv.
dial. mes; *m‑ may have replaced original *u̯ ‑ by analogy with 1pl. ending
of verbs | Stang 1966: 254–255: OPr. mes, Li. mẽs, Lv. dial. mes point to
PBalt. short *e; Lv. mẽs with long *ē may be due to influence from 2pl., or
it may be the result of emphatic lengthening; PIE *mes is also found in
Arm. mek‘ and Slavic my, where ‑y is from 2pl. vy; Ved. and Gmc. point to
PIE *u̯ ei | Otrębski 1956: 137–138: Li. mẽs, dial. mė̃s, Lv. mẽs have long
vowel from 2pl. jū̃s; OPr. mes, Lv. dialects preserve short vowel, which is
modelled on C-stems | Endzelīns 1923: 377–378: Lv. mẽs, Li. mẽs have
long vowel from 2pl.; short vowel is preserved in OPr. mes, Lv. dial. mes;
Slavic my is remodelled on vy; Baltic, Slavic, Arm. *m‑ for *u̯ ‑ is probably
from 1pl. ending of verbs
PS Vondrák 1908/1928: 72–73: Slavic my is perhaps remade based on acc.
ny | Meillet 1924/1934: 454: Ved. vayám, Go. weis point to PIE *u̯ ei; Li.
mẽs, Arm. mek‘ point to *mes, also in Slavic my, with ‑y from 2pl. vy and
ā-stem nom. pl. | Vaillant 1958: 451–452: IIr., Hi., Go. point to PIE *u̯ ei‑;
Arm. *m‑ is from 1sg. pron. stem im < *(e)me; BS *mēs from *u̯ ēs by anal-
ogy with verbal 1pl. ending; *u̯ ēs is from *u̯ eis with *ē from 1du. pron. *u̯ ē;
Slavic my has ‑y from vy | Arumaa 1985: 166–167: PIE *u̯ ei; Slavic my
may reflect acc. pl. *mons (also in OPr. mans) from *nons with *m‑ from
pron. obl. sg.; *nons replaces older *nos | Aitzetmüller 1978/1991: 109:
Ved. vayám, Go. weis, Hi. wēš reflect PIE *u̯ ei‑; OCS my has ‑y from 2pl. vy;
OCS m‑, also in Baltic and Arm., may be due to influence from 1pl. ending
of verbs, or it may be old
PIE There are two main candidates for the original nominative form of the
second-person plural pronoun: *u̯ ei̯ and *mes. While the former seems to be
continued in Anatolian, Tocharian, Indo-Iranian and Germanic, the latter is
found in Balto-Slavic and Armenian. Since plausible secondary sources of
*m‑ do exist, it is likely that *u̯ ei̯ is the original form. In Hittite and Germanic
a pluralic *‑(e)s was added to the form, and in Indo-Iranian the particle
*hom was added. TochB nom.–acc. wes, TochA was from *u̯ os may have been
238 Chapter 3
influenced by the enclitic oblique form PIE *nos (see e.g. Adams 1988: 154;
Pinault 1989: 111; 2008: 535–536; Rasmussen loc. cit.). The Greek and Latin forms
are based on the tonic accusative and the enclitic oblique forms, respectively.
In Armenian the inital *m‑ may stem from the first-person plural ending of the
verbs or it may be due to influence from the oblique forms of the first-person
singular pronoun (see e.g. Matzinger 1997: 87–88, who seems to prefer the
latter solution).
PBS The initial *u̯ ‑ of the Proto-Indo-European form was replaced with *m‑
in Proto-Balto-Slavic, most likely due to the influence of the first-person plural
ending of the verbs. At a common pre-stage of the Baltic languages the final
part of the pronoun was replaced with *‑es by analogy with the consonant
stems. The replacement of *ˈmei̯ with *ˈmes may have taken place in pre-Proto-
Balto-Slavic or in pre-Proto-Baltic. The form *ˈmes was preserved in OPr. mes,
Li. mẽs, Li. dial. mès and Lv. dial. mes. In Li. dial. mė̃s and standard Lv. mẽs,
*ˈmes was lengthened by analogy with 2pl. pron. PBS *ˈi̯ṵ̄s (Endzelīns loc. cit.;
for other variants of the form in Lithuanian dialects see Zinkevičius 1966: 302
and passim).
PS In Slavic, the vocalism of the pronoun was changed to *‑ū by analogy with
2pl. pron. *u̯ ū, where *‑ū reflects PBS *‑ūs. PS *ˌmū (CS *my̑ [29]) is preserved
in all old Slavic dialects.
PIE Brugmann 1909–1911: 384–385, 410–412: PIE *i̯ūs; OCS vy has ‑y from
acc. vy | Beekes 1995/ 2011: 233–234: PIE *iuh | Rasmussen
forthc. a § 13: OCS vy regularly reflects PBS, PIE *i̯ū́s | Schmidt 1978:
207–212, 243, 245: PIE *i̯uhs; OCS vy reflects acc. *u̯ ō̆ns, with secondary
Nominal Inflection 239
PBS PIE *i̯úhs is reflected as PBS *ˈi̯ṵ̄s [1], directly preserved in Lv. jũs and
OPr. ioūs, with regular acute tone. Li. jū̃s may have circumflex tone from the
first-person pronoun (cf. Rasmussen 1992b/1999: 481 and 1992c/1999: 542, who
assumes Balto-Slavic métatonie douce in monosyllabic words).
PS The expected Proto-Slavic outcome of PBS *i̯ṵ̄s is *i̯ū [12|13]), but the *i̯‑
was replaced with *u̯ ‑ from the oblique forms in pre-Proto-Slavic, resulting in
PS *ˌu̯ ū (CS *vy̑ [29]).
Go. ‑uns reflect *‑n̥ s; Hi. ‑uš points to earlier *‑m̥ s | Krahe 1942/1967:
40, 46: r-stem Go. fadruns is from PIE *‑n̥ s; NGmc. and WGmc. r-stems
use nom. pl. form; n-stem forms Go. hanans, ONor. hana are perhaps by
haplology from PGmc. *‑anuns < PIE *‑onn̥ s | Boutkan 1995: 259, 271,
275, 278–281: r-stem forms Go. broþruns, OEng. ‑ru reflect PGmc. *‑runs
< PIE *‑rn̥ s; n-stem forms Go. ‑ans, ONor. ‑a are either analogical or have
resulted by haplology; elsewhere in Gmc. the nom. pl. form is widely
used | Kloekhorst 2008a: 928–929: Hi. ‑uš points to PIE *‑ms, not *‑ns
PBS Olander 2009: 193–194: accentuation of Li. dùkteris, ORu. dóčeri, dščéri
is not original | Endzelīns 1971: 165: Li. ‑is, OLv. ‑is, OPr. ‑ins, OCS ‑i
reflect PIE *‑n̥ s; Lv. ‑’us is modelled on (i)i̯o-stems | Stang 1966: 223:
Li. ‑is, Lv. ‑is reflect PBalt. *‑ins < PIE *‑n̥ s, also in Ved. ‑aḥ, Gk. ‑ας, La. ‑ēs,
OCS ‑i; Lv. ‑’us is from (i)i̯o-stems | Otrębski 1956: 53, 57: Li. ‑is, OCS ‑i
are from *‑ins < PIE *‑n̥ s | Endzelīns 1923: 323–324: OLv. ‑is reflects
PIE *‑n̥ s; Lv. ‑’us is i̯o-stem ending
PS Vondrák 1908/1928: 7: OCS ‑i reflects *‑ins < PIE *‑n̥ s | Hujer 1910:
109–110: Slavic ‑i most likely reflects PIE *‑n̥ s | Meillet 1924/1934: 422:
OCS ‑i is from PIE *‑n̥ s | Vaillant 1958: 186: Slavic ‑i, OPr. ‑ins, Li., OLv.
‑is are from PIE *‑n̥ s | Bräuer 1969b: 10, 72 | Arumaa 1985: 114–115:
Slavic ‑i is either from *‑n̥ s (probably from older *‑m̥ s) or is taken over
from i-stems | Igartua 2005a: 345: OCS ‑i, Li. ‑is go back to PIE *‑n̥ s <
*‑m̥ s | Aitzetmüller 1978/1991: 95, 101, 103: OCS ‑i is from PIE *‑n̥ s or
from i-stems
PIE *mē̆msó‑. Since the accusative plural marker is not *‑ms in Baltic and
Germanic, this might then be taken as evidence that the marker could not have
been *‑ms in the proto-language. However, it is not difficult to imagine a regu-
lar change of *‑ms to *‑ns in final syllables at some point in the development of
Proto-Indo-European to East Baltic and Germanic. Accordingly, the accusative
plural marker is reconstructed as *‑ms here, with the postconsonantal variant
*‑m̥ s appearing regularly in consonant stems (note that Eichner apud Griffith
2006: 44 n. 1 reconstructs PIE *‑m̥ s after consonants but *‑ns after vowels).
PBS PIE *‑m̥ s > *‑n̥ s regularly yielded PBS *ˌ‑ins [2|4], preserved as OPr.
‑ins, Li. ‑is and OLv. ‑is. The modern Latvian form akmeńus is taken from the
(i)i̯o-stems.
The acute tone of the Lithuanian accusative plural ending in all paradigms
constitutes a problem, as there is no indication of the presence of a laryngeal
in the accusative plural marker in Proto-Indo-European (see also the discus-
sion in § 3.14.5 on the o-stem ending). According to Kortlandt (1975/2011a:
44; 2005: 23; 2006/2009: 100), the Balto-Slavic languages point to an accusa-
tive plural in *‑hNs, where the acute tone was generalised from the ā-stems
(see also Pronk 2012: 213 fn. 4). In my view, however, *VhR sequences did not
become glottalised in Balto-Slavic unless followed by a vowel, cf. e.g. Li. ā-stem
acc. sg. ‑ą from PBS *‑ām < PIE *‑ah₂m̥ . I find it more likely that we are dealing
with a phonetic change in a pre-stage of East Baltic where a vowel was glot-
talised when followed by word-final *‑ns (cf. Olander loc. cit.; but see Kim 2012
§ 4.2).
The accentuation of monosyllabic stems like šunìs, by Saussure’s law
from pre-Li. *ˈšunī�s̰ , is regular, in contrast to that of polysyllabic stems like Li.
dùkteris for expected *dukterìs < pre-Li. *dukˈterī�s̰ from *‑érn̥ s < PIE *-érm̥ s
(Olander loc. cit.).
PS PBS *ˌ‑ins yielded PS *ˌ‑ī [16|17] (CS *‑i [29]). The ending is preserved in
Old Church Slavonic and Old Russian. In the Old Novgorod dialect, dni may also
preserve the original form, although the ending ‑y of the preceding demonstra-
tive pronoun ty points to a non-dialectal form (cf. Zaliznjak 1995/2004: 394). In
the form (pro) gorodišč’ane, ‑e may denote either the o-stem accusative plural
ending /‑ě/, or the consonant-stem nominative plural ending /‑e/ (Zaliznjak
1995/2004: 448). In Old Czech the original ending has been replaced with
the consonant-stem nominative plural ending in feminine words and by the
o-stem accusative plural in masculines.
Nominal Inflection 243
OCS gosti Li. mintìs Ved. masc. →śúcīn, →‑īm̐ ś (ca); fem.
ORu. gosti; ONovg. sani Lv. avis →śúcīḥ
OCz. hosti OPr. ackins OAv. aṣ̌īš; YAv. gaⁱrīš
Gk. →πόλεις; num. τρῖς; Hom. πόλῑς;
Cretan πολινς
La. →turrēs, turrīs
Go. gastins
Hi. →ḫalkiuš
PS Vondrák 1908/ 1928: 7; Vondrák 1906/ 1924: 151: OCS ‑i reflects PIE
*‑ins | Hujer 1910: 108–109: Slavic ‑i reflects PIE *‑ins | Meillet
1924/1934: 392, 419: OCS ‑i reflects *‑ins | Vaillant 1958: 136: Slavic ‑i is
from *‑ins; Li. acute tone is from o- and ā-stem acc. pl. | Bräuer 1969a:
157 | Arumaa 1985: 128: CS *‑i reflects PIE *‑ins (alongside *‑i̯n̥s) or
perhaps corresponds to IIr. *‑īs in fem. i-stems and ī-stems | Igartua
2005a: 266–268: Slavic ‑i is from PIE *‑ins | Aitzetmüller 1978/1991: 75:
OCS ‑i in fem. and probably also in masc. words is from fem. *‑īs, also in
Ved. fem. ‑īḥ, Li. ‑ìs
PIE PIE *‑ims consists of the zero grade of the i-stem suffix followed by the
accusative plural marker (see § 3.14.1). The ending is preserved in Cretan Greek
dialect forms like ‑ινς, in La. ‑īs, in Germanic and in Balto-Slavic. The Avestan
ending most likely also reflects PIE *‑ims (Debrunner & Wackernagel loc. cit.;
Hoffmann & Forssman 1996: 89); it cannot be excluded, however, that it is iden-
tical to Ved. fem. ‑īḥ, which is modelled on the ā-stem ending ‑āḥ. Ved. masc.
‑īn(s) is modelled on the o-stem ending ‑ān(s). La. ‑ēs is the consonant-stem
ending, while the Attic Greek form in ‑εις goes back to *‑ens, with *e imported
from other case forms. The Hittite ending ‑iuš has been recomposed by the
stem-vowel ‑i‑ and the ending ‑uš from the consonant stems and o-stems.
PBS PIE *‑ims > *‑ins regularly yielded PBS *ˌ‑ins [4]. The ending is preserved
unaltered in Old Prussian. At a pre-stage of Lithuanian and Latvian, *‑ins
became glottalised *‑ḭns (see § 3.14.1), eventually yielding ‑is in both languages.
PS PBS *ˌ‑ins yielded PS *ˌ‑ī [16|17] (CS *‑i [29]), preserved in the old Slavic
dialects.
PIE PIE *‑ums consisted of the zero grade of the u-stem suffix followed by
the accusative plural marker *‑ms (§ 3.14.1); Indo-Iranian also shows reflexes
of an ending with the expected vocalisation *‑u̯ m̥s (cf. § 2.2.1). The ending
*‑ums is preserved in Greek dialects, Latin, Germanic, Balto-Slavic, Anatolian
(Weitenberg 1984: 375; Kimball 1999: 251, 327) and probably Avestan (Debrunner
& Wackernagel loc. cit.; Hoffmann & Forssman 1996: 89). Ved. fem. ‑ūḥ is mod-
elled on the ā-stem ending ‑āḥ. The Vedic masculine ending ‑ūn(s) is based on
analogy with the o-stem ending ‑ān(s). Attic Gk. ‑εις < *‑ens has *e from other
forms of the paradigm.
PS PBS *ˌ‑uns is reflected as PS *ˌ‑ū [16|17] (CS [29]), preserved in the old
Slavic dialects.
PIE Brugmann 1909–1911: 225–226: PIE *‑ās; OCS ‑y, ‑ję reflect *‑ans (perhaps
from *‑āns), also found in other branches, which is probably modelled
on o-, i‑, u-stems | Beekes 1995/2011: 200: PIE *‑eh₂ns | Rasmussen
forthc. a § 5: Li ‑às, East Li. ‑às, Lv. ‑as reflect PBS *‑ā́s < PIE *‑ās; Li. def.
adj. ‑ą́ sias indicates that PBS def. adj. contained a nasal; OCS ‑y, ‑ję are
from *‑(i̯)āns | Debrunner & Wackernagel 1930/1975: 123–124: PIE
*‑ās is from earlier *‑āns; languages pointing to *‑āns, *‑ans have reintro-
duced the nasal | Rix 1976/1992: 133: PIE *‑ās < *‑eh₂n̥ s; nasal of PGk.
*‑āns is taken from o-stem acc. pl. | Sihler 1995: 254, 271: Gk. and Italic
forms reflect *‑ans < *‑āns < *‑eh₂ms, for expected *‑eh₂m̥ s; the loss of the
nasal after a long vowel may have happened independently in the various
branches | Weiss 2009/2011: 229, 235–236: La. ‑ās directly reflects PIE
*‑ās < pre-PIE *‑eh₂ns | Krahe 1942/1967: 22–23: PGmc. *‑ōz is from
PIE *‑ās < *‑āns | Boutkan 1995: 141–142, 225, 229: PGmc. *‑ōns reflects
PIE *‑eh₂ns (following Kortlandt); the difference between PGmc. nom.
pl. *‑ōs and acc. pl. *‑ōns is retained in OEng. nom.–acc. pl. doublets ‑e,
‑a; Go. ‑os regularly reflects PGmc. *‑ōns < PIE *‑eh₂ns | Kloekhorst
2008a: –
Nominal Inflection 247
PBS Olander 2009: 184: PBS *ˌ‑ās reflects PIE *‑áh₂m̥ s with regular loss of
accent | Endzelīns 1971: 145: it is doubtful if East Li. ‑às, Lv. ‑as are
inherited from PIE | Stang 1966: 200: East Li. ‑as, Lv. ‑as are from PIE
*‑ās; Li. def. adj. ‑ą́ s, OPr. ‑ans reflect new formation *‑āns | Otrębski
1956: 26: Li. points to *‑ās and (expected) *‑ans; *‑ās may have arisen
when a new loc. pl. was created on the basis of the acc. pl. | Endzelīns
1923: 308: Lv. ‑as, East Li. ‑as reflect PBalt. *‑ās, alongside *‑āns
PS Vondrák 1908/1928: 7: Slavic ‑y, ‑ję/ě go back to *‑(i̯)ons < *‑(i̯)ans, ana-
logically shortened from *‑(i̯)āns | Hujer 1910: 98–104: PIE *‑ās from
pre-PIE *‑āns; an ending *‑āns was recreated on the analogy of the other
stems in PIE or in the individual daughter languages; East Li. ‑as, Lv. ‑as
reflect *‑ās; Slavic ‑y, ‑ję are from *‑(i̯)āns, but ‑jě is from *‑i̯ās | Meillet
1924/1934: 398–399: PIE *‑ās; Slavic ‑y, ‑ję are from *‑ons < *‑āns with
reintroduced nasal | Vaillant 1958: 83–84, 87–88: Slavic ‑y is from
*‑āns; SSl. ‑ję, NSl. ‑jě reflect *‑i̯āns; Li., Lv. endings reflect *‑āns and *‑ās;
BS also had *‑āns and *‑ās | Bräuer 1969a: 106–107, 127: CS *‑y goes
back to PIE *‑āns, perhaps via shortened *‑ans; CS *‑ję and *‑ě go back
to *‑jens from *‑i̯ons or *‑i̯ans, corresponding to *‑i̯ās or *‑i̯ans, both
from *‑i̯āns, in other languages | Arumaa 1985: 153–154: PIE *‑āns
reflects *‑ā‑ or *‑eə₂‑, plus *‑ns; Slavic ‑y is not from *‑āns; ‑ję is hardly
from *‑i̯āns | Igartua 2005a: 228–235: Slavic ‑y, ‑ję reflect PIE *‑ā̆ns <
*‑e(h₂)ms | Aitzetmüller 1978/1991: 88, 91: PIE *‑āns, *‑ās would yield
PS *‑ā(s); OCS ‑y, ‑ję represent masc. (i̯)o-stem ending *‑(i̯)ons, which in
pronouns was masc. and fem.; ESl., WSl. ‑jě reflects *‑i̯ons
outcome of PIE *‑ah₂m̥ s, perhaps via *‑ōuns (Guus Kroonen, pers. comm., 2014;
cf., somewhat differently, Boutkan loc. cit.). Alternatively, they may represent
the nominative used for the accusative on the model of the identical nomina-
tive and accusative singular forms in ‑a (Rasmussen 1992a/1999: 507 n. 2). In
fact, the only evidence indisputably pointing to a proto-form without a nasal is
provided by Lithuanian dialects and Latvian, for which see below.
PS PBS *ˌ‑āns became *ˌ‑ə̄n in Proto-Slavic [17] (CS Cl *‑y || ONovg. ‑ě [28|29];
CS S *dȗšę || CS N *‑ě [20|28|29]). The ending is generally well preserved in the
old Slavic dialects; for the isogloss separating the Old Novgorod dialect, with
‑ě, from the remaining Slavic dialects, with ‑y, see Olander 2012: 333–335 and
passim. In Serbo-Croatian and Slovene the hard ending was replaced with the
soft ending. In South Slavic soft ā-stems, the accusative plural ending spread
to the genitive singular and nominative plural by imitation of the (segmental)
syncretism between the three endings in the hard ā-stems (see § 3.5.4).
OCS grady; kon’ę Li. lángus; def. adj. Ved. devā́n, devā́m̐ś (ca)
ORu. stoly; koně; ONovg. kolotokě, mažúosius; Žem. OAv. maṣ̌iiə̄ṇg, maṣ̌iiąs(‑cā);
(late) (pro) →sigy; koně def. adj. gerū́sius, YAv. haomą, haomąs(-ca)
OCz. chlapy; oráčě, ‑e gerúnsius Gk. ἀγρούς; Cretan ελευθερονς
Lv. tȩ̃vus; def. adj. La. lupōs
mazuõs Go. dagans
OPr. deiwans Hi. antuḫšuš
Nominal Inflection 249
PIE Brugmann 1909–1911: 221–225 (1897: 392): PIE *‑ons is more likely
than *‑ōns; OCS ‑y, ‑ję reflect *‑ǭs, *‑ę̄s < *‑ų̄ s, *‑ę̄s from “Urslav.” *‑ons,
*‑ens | Beekes 1995/2011: 212–213: Av. points to PIE *‑ons; long vowel
of Ved. ‑ān may stem from nom. pl.; Li. ‑ùs points to *‑ohns, perhaps
with *h from laryngeal stems | Rasmussen forthc. a § 4: Li. ‑ùs, ‑úos‑,
Lv. ‑us, OPr. ‑ans, ‑ons, OCS ‑y reflect PBS *‑ṓns from PIE *‑ōns, per-
haps from pre-PIE pron. *‑oi̯ms; Hi. ‑uš is probably from C-stem ending
*‑m̥ s | Debrunner & Wackernagel 1930/1975: 59–60, 102: PIE ending
was probably *‑ons; Ved. ‑ān has long vowel from nom. pl. ‑āḥ | Rix
1976/1992: 140: Gk. ‑ους, dial. ‑ονς reflect PIE *‑ons | Sihler 1995: 262–263:
Gk. ‑ους, dial. ‑ονς, Osc. ‑úss, La. ‑ōs are from PIE *‑ons < *‑oms | Weiss
2009/2011: 206, 224: OCS ‑y goes back to *‑ūns < *‑ōns < PIE *‑ons; some
languages point to *‑ōns | Krahe 1942/1967: 11–12: Go. ‑ans, ONor.,
OHG ‑a reflect PIE *‑ons | Boutkan 1995: 167, 170, 191–194: Go. ‑ans,
NWGmc. forms in ‑a reflect PGmc. *‑ans < PIE *‑ons | Kloekhorst
2008a: 928–929: Hi. ‑uš is from o-stem ending *‑oms, merging with C-stem
ending *‑ms
PBS Olander 2009: 183: PBS *ˌ‑ans is the regular outcome of PIE noun ending
*‑óns, alongside pron. *‑ōns, represented by Ved. ‑ān; East Baltic reflexes of
*‑V̰̄ (n)s are the results of a vowel lengthening before *‑ns | Endzelīns
1971: 136–137: Li. ‑us, Lv. ‑us are from *‑ōns | Stang 1966: 186: Li. ‑us,
‑úos‑, Lv. ‑us point to *‑úons < *‑ōns, corresponding to Ved. ‑ān; OPr. ‑ans,
Gk. dial. ‑ονς, Go. ‑ans reflect *‑ons, shortened from *‑ōns | Otrębski
1956: 15: Li. ‑us is from *‑ōns, corresponding to Ved. ‑ām̐ ś ca | Endzelīns
1923: 299: Li. ‑us, Lv. ‑us reflect *‑uons < PIE *‑ōns, perhaps also in Ved. ‑ān
PS Vondrák 1908/ 1928: 7 (1906/ 1924: 151–152): OCS ‑y is from PIE
*‑ons | Hujer 1910: 104–106: Slavic ‑y, ‑ję reflect PIE *‑ons; Baltic *‑ṓs
has arisen by analogy with nom. pl. *‑ō̃s on the model of ā-stem nom.
*‑ā̃s, acc. *‑ā́s; Ru., Po., Cz., Sorb. *‑jě is taken from i̯ā-stems | Meillet
1924/1934: 152, 392, 409: OCS ‑y is from PIE *‑ons | Vaillant 1958:
34–35, 48–49: Li. ‑us, Lv. ‑us is from *‑uons from remade ending *‑ōns;
OPr. ‑ans, Slavic ‑y, ‑ję || ‑jě reflect stem-vowel *a plus acc. pl. marker
*‑ns | Bräuer 1969a: 27, 74–75: CS *‑y reflects *‑ūs < *‑uns < PIE *‑ons;
South CS *‑ję, West and East CS *‑jě reflect PIE *‑i̯ons | Arumaa 1985:
141: PIE *‑ons; Baltic *‑ōns has long vowel from nom. pl. *‑ōs; [explanation
of CS *‑y, *‑ę is unclear] | Igartua 2005a: 158–163: CS *‑y, *‑ję reflect
PIE *‑o/ōns | Aitzetmüller 1978/1991: 82, 84: OCS ‑y, ‑ję reflect PIE
*‑ons, *‑i̯ons; Li. ‑us < *‑ōs is from *‑ons
(for the marker see § 3.14.1). While most languages are compatible with the
reconstruction of an ending containing a short vowel, and some even seem to
require it, certain pieces of evidence make it more likely that the ending was
in reality *‑ōms (see Kim 2012 for a clear and comprehensive overview of the
problem). According to Rasmussen (1989b: 139 with n. 21), the proto-language
had *‑ons in nouns and *‑ōns < pre-PIE *‑oi̯ms in pronouns.
While the long vowel of Ved. ‑ān, ‑ām̐ ś ca may have been introduced ana-
logically from the nominative plural, it is more straightforward to assume that
it is the direct reflex of a Proto-Indo-European long vowel. Note, incidentally,
that the proportion set up by Debrunner & Wackernagel (loc. cit.)—“‑āns zu
‑ās nach Sg. ‑am zu ‑as”—would rather lead to acc. pl. *‑ām than *‑āns. The
prehistory of the Gaulish and OIr. accusative plural forms is also less complex
when departing from *‑ōns with a long vowel (see Griffith 2006: 50–63; Kim
2012 § 4.3). A third argument in favour of an original long vowel is constituted
by Li. ‑ùs, ‑úos‑, which is difficult to reconcile with a proto-form containing
a short form (I am grateful to Tobias Mosbæk Søborg, pers. comm., 2013, for
directing my attention to this fact). I think the Lithuanian ending does point
to a long vowel, though somewhat more indirectly than it appears at first sight
(see below).
The branches that are more straightforwardly accounted for on the basis
of a proto-form containing a short rather than a long vowel are Avestan,
Tocharian and, for accentual reasons, Balto-Slavic. The Avestan forms have
been analysed as the direct reflexes of *‑āns (Kümmel 2013b; see also Kim 2012
§ 4.1), but it is also not difficult to imagine that the long vowel was replaced
with a short vowel under the influence of the accusative plural of the other
paradigms. The latter explanation may also account for TochB ‑eṃ, TochA adj.
‑es (Kim 2012 §§ 2.1 and 4 with references). In Balto-Slavic an ending contain-
ing a long vowel should not have lost its accent by the mobility law [4]; how-
ever, the form may relatively easily have become unaccented by analogy with
the accusative plural forms of all the other paradigms (note that this account
of the facts differs from the one given in Olander loc. cit.).
In Anatolian, Hi. ‑uš may reflect either PIE *‑ōms or *‑oms (see Kim 2012
§§ 3 and 4.4; both Melchert 1994: 185–186 [with references] and Kloekhorst loc.
cit. depart from short *‑oms).
PBS Li. ‑ùs, def. adj. ‑úos‑, Žem. gerū́sius, gerúnsius (Zinkevičius 1966: 212;
1980: 211–212), point to an ending containing a long acute vowel, whereas Lv.
‑us, ‑uõs may reflect either *‑ō̰(n)s or *‑a̰ ns. The Old Prussian ending ‑ans is
immediately derivable from *‑ans, but it cannot be ruled out that it may also
reflect *‑ō̰ns or *‑ōns (Stang loc. cit.). It is tempting to relate the Lithuanian long
Nominal Inflection 251
vowel to the Lithuanian and Latvian acute tone, and this is routinely done.
However, since Proto-Balto-Slavic acute diphthongs consisting of an original
o-vowel and a sonorant seem to be show up with a in the Baltic languages
(see § 2.2.2), it is probable that a pre-Proto-Balto-Slavic sequence *‑ō̰ns would
have been reflected as PBS **‑a̰ ns > Li. **‑às, **‑ą́ s‑. I therefore assume that
the development was PIE *‑ōms to (non-acute) *ˌ‑ōns in Proto-Balto-Slavic;
this is in accordance with the view, not generally accepted, that PIE plain long
vowels receive a non-acute tone in Proto-Balto-Slavic (see Olander 2009: 146–
148 with references; and cf. the discussion in Villanueva Svensson 2011 with
the responses by Kortlandt 2012b and Pronk 2012). When final sequences of
a nasal plus *s led to glottalisation of a preceding vowel at a pre-stage of East
Baltic—a development already required in order to explain the acute tone of
the Lithuanian accusative plural in the other paradigms (see § 3.14.1)—the
sequence PIE *‑ōms > PBS *‑ōns became *‑ō̰ns, yielding the attested Lithuanian
and Latvian outcomes.
As noted above, the derivation of the Balto-Slavic forms from PIE *‑ōms
with a long vowel implies that the unaccentedness of the form in Proto-Balto-
Slavic is secondary.
PS *nə̄ (CS *ny) PBS *nōs PIE encl. obl. *nō̆s (tonic acc. *n̥smé)
OCS ny Li. acc. →mùs Ved. encl. acc.–gen.–dat. naḥ (tonic →asmā́n)
ORu. ny; Lv. acc. →mũs OAv. encl. acc. nā̊; gen.–dat. nə̄ (tonic acc. ə̄hmā);
ONovg. →ny OPr. acc. →mans YAv. encl. acc.–gen.–dat. nō (tonic ahma)
OCz. ny Gk.—(→ἡμᾶς; Hom. →ἡμέας; Aeol. [Hom.,
Sappho] ἄμμε)
La. acc. nōs
Go.—(acc.–dat. →uns, →unsis)
Hi. encl. acc.–dat. ‑naš (OS) (tonic acc.–dat.–loc.
→anzāš)
252 Chapter 3
PS *u̯ə̄ (CS *vy) PBS *u̯ōs PIE encl. obl. *u̯ō̆s (tonic acc. *usmé)
OCS vy Li. acc. →jùs Ved. encl. acc.–gen.–dat. vaḥ (tonic →yuṣmā́n)
ORu. vy; Lv. acc. →jũs OAv. encl. acc. vā̊, gen.–dat. və̄ (tonic acc. ?);
ONovg. vy OPr. acc. →wans YAv. encl. acc.–gen.–dat. vō (tonic acc. ?)
OCz. vy Gk.—(→ῡ̔μᾶς; Hom. →ῡ̔μέας; Aeol. [Hom.,
Pindarus] ὔμμε)
La. acc. uōs
Go.—(acc.–dat. →izwis)
Hi. encl. acc.–dat. →‑šmaš (OS) (tonic acc.–dat.–
loc. →šumāš)
PIE Brugmann 1909–1911: 383–385, 408–409, 419–420: PIE tonic acc. *n̥ sme,
*usme; encl. and tonic acc.–gen.–dat. *nē̆s, *nō̆s, *u̯ ē̆s, *u̯ ō̆s; OPr. mans,
wans, Li. mùs, jùs, OCS ny, vy go back to *‑ns | Beekes 1995/2011: 233–
234: PIE encl. acc. *nōs, *uōs, gen. *nos, *uos, dat. *ns (> Go. uns, Hi. tonic
stem anz‑) | Rasmussen forthc. a § 13: PIE *nos, *u̯ os; long vowel of
OAv. nā̊, vā̊ is analogical from ā-stem acc. pl.; long vowel of La. nōs, uōs is
either the result of monosyllabic lengthening, analogy with o-stem nom.
and acc. pl., or influence from corresponding du. forms; PGmc. 1pl. *uns
has arisen by contamination of tonic PIE *n̥ smé and encl. *nos; PGmc.
2pl. *izwiz has arisen by assimilation from *uzwiz < *usu̯ és, a contamina-
tion of PIE *usmé and *u̯ os; OCS ny, vy reflect PBS *nṓns, *u̯ ṓns from PIE
*nos, *u̯ os, with introduction of o-stem ending | Schmidt 1978: 172–
173, 177–181, 204–205, 212–233, 243: PIE 1pl. tonic *mōs, encl. acc.–gen.–
dat. *m/nos; Slavic ny is from *nō̆ns, with secondary acc. marker *‑n‑; PIE
2pl. tonic *usme, later *u̯ ōs, encl. gen.–dat.–instr. *u̯ os(?) | Debrunner
& Wackernagel 1930/1975: 477–478: Ved. encl. naḥ, vaḥ, correspond-
ing to OAv. nə̄, və̄, are old in gen.–dat. function, but have replaced acc.
*nāḥ, *vāḥ, corresponding to OAv. nā̊ | Rix 1976/1992: 179: Ved. naḥ,
vaḥ reflect PIE encl. *nos, *u̯ os | Sihler 1995: 379, 381: Ved. nas, vas,
Hi. ‑naš go back to PIE encl. acc. *nos, u̯ os; La. nōs, uōs, OAv. nā̊, vā̊, OCS
ny, vy reflect alternative PIE forms *nōs, *u̯ ōs | Weiss 2009/2011:
329–331: La., BS, OAv. (encl. acc.) point to PIE *nōs; IIr., perhaps Hi. point
to *nos | Krahe 1942/1967: 52–53: Gmc. 1pl. forms are from PIE zero
Nominal Inflection 253
grade *n̥ s; Ved. naḥ is from full grade *nes; La. nōs is from long grade *nōs;
Go. izwis has arisen by dissimilation from *uiz‑u̯ iz, a reduplicated form
based on shortened *u̯ iz | Seebold 1984: 39–44: Go. uns goes back to
*unss from PIE *n̥ sés or *n̥ sós, also in Hi. anzāš; Go. izwis is perhaps from
*izg̑ ʰu̯ o/es | Kloekhorst 2008a: 115–116, 596, 770: Hi. encl. ‑naš is from
PIE encl. *nos; Hi. ‑šmaš is identical to dat.–loc. pl. of encl. pron. ‑a; the
final part ‑aš is probably the dat.–loc. ending ‑aš
PBS Kapović 2006a: 76, 91, 98, 101, 104, 113, 132, 158: CS encl. acc.–dat. *ny̑ , *vy̑ ,
OPr. mans (with m‑ from nom. or by dissimilation from *nans), wans
reflect *nōns, *u̯ ōns from PIE encl. acc.–gen.–dat. *nōs, *u̯ ōs (alongside
*nos, *u̯ os) with secondary acc. marker *‑n‑; Li. mùs, jùs, Lv. mũs, jũs reflect
*mū́ns, *i̯ū́ns, which have arisen either by analogy with u-stems or on the
basis of nom. pl. *i̯ū́s with secondary acc. marker *‑n‑ | Endzelīns
1971: 189: OPR. mans (< *nans), wans correspond to OCS ny, vy; Lv. mũs,
jũs, Li. mùs, jùs reflect *mū̆ns, *jū̆ns or, by analogy with u-stems, *mùns,
*jùns | Stang 1966: 255–256: OPR. wans is remade from *u̯ ō̆s on the
basis of acc. forms of nouns and pronouns; OPr. mans has m‑ for *n‑ from
nom. pl. mes; Li. mùs, jùs, Lv. mũs, jũs probably reflect *múns, *júns, based
on nom. 2pl. form | Otrębski 1956: 138: Li. mùs is modelled on jùs,
which has ‑us from u-stems | Endzelīns 1923: 380–381: Lv. mũs, jũs, Li.
mùs, jùs (with shortening in unaccented position) reflect *mū̆ns, *jū̆ns,
modelled on u-stems; OPr. mans, wans, OCS ny, vy are remade from *nōs,
*u̯ ōs on the model of *tōns etc.
PS Vondrák 1908/1928: 72–73: Slavic my, vy correspond to La. nōs, uōs or OPr.
mans, wans | Meillet 1924/1934: 455: Slavic ny, vy may reflect *nōs,
*u̯ ōs directly, or they may contain acc. marker *‑ns, as does OPr. mans,
wans | Vaillant 1958: 452–453: Slavic ny, vy, OPr. mans, wans reflect PIE
*nes, *u̯ es, with masc. and fem. endings -y, ‑ans | Arumaa 1985: 166–
169 | Aitzetmüller 1978/1991: 109, 111–112: OCS ny, vy, OPr. mans, wans
go back to *nōns, *u̯ ōns, replacing *nōs, *u̯ ōs, preserved in La. nōs, uōs
The reflexes found in the remaining languages are based on the tonic
forms, although the prehistory of the Germanic forms is not clear, especially
the second-person plural, e.g. Go. izwis (for discussions see Katz 1998: 107–133;
Kroonen 2008).
PBS The long variants PIE *nōs, *u̯ ōs were retained as PBS *nōs, *u̯ ōs. In pre-
Proto-Baltic a nasal was introduced on the pattern of the accusative plural
ending of nouns and non-personal pronouns, and the first-person plural pro-
noun received an *m‑ by analogy with the nominative form PBalt. *ˈmes. PBalt.
*mōns, *u̯ ōns are regularly reflected as OPr. mans, wans. At a pre-stage of East
Baltic the second-person plural pronoun also introduced the initial consonant
of the nominative form, resulting in Li. mùs, jùs, Lv. mũs, jũs. For the acute tone
of the East Baltic forms see § 3.14.5; if one assumes that Proto-Indo-European
plain long vowels regularly become acute in Proto-Balto-Slavic, the East Baltic
forms may directly reflect PIE *‑ōs.
PS The Slavic forms are usually traced back to *nō̆ns, *u̯ ō̆ns, with second-
ary adaption of the forms to the general accusative plural marker *‑ns. If this
reconstruction is correct, the Slavic material would correspond exactly to OPr.
mans, wans, except that Slavic preserves the original consonant in the first-
person plural. However, as pointed out by Meillet (1897: 96; see also Kapović
2006: 76; and cf. Rasmussen loc. cit.), the fact that OCS ny, vy etc. are not
restricted to the accusative but also serve as dative forms can hardly be seen as
anything but an archaism, corresponding to the use of naḥ in Vedic as a more
general oblique form. In that case I find it hard to the understand—contrary
to Kapović and Rasmussen, but in agreement with Meillet—why an accusa-
tive ending would have been introduced in a form that had (at least) accusa-
tive and dative function. The straightforward solution is to assume that PIE,
PBS *nōs, *u̯ ōs are directly reflected as PS *nə̄, *u̯ ə̄ [17] (CS Cl *ny; CS *vy [29])
(similarly Meillet 1897: 96; loc. cit.; Vondrák 1906: 108; 1908: 87; but cf. Hujer 1910:
75–76 n. 1; Arumaa 1985: 131; and the hesitation of Vondrák loc. cit.). This solu-
tion obviously only works if one accepts the view that PIE *ō > pre-PS *ā had
a special treatment before word-final *s in Slavic. The expected Old Novgorod
first-person form *ně was replaced with ny by analogy with 2pl. vy, where the
development PS *‑ə̄ > ‑y is regular after a labial (see [29]).
Another indication that the reflexes of PIE *nōs, *u̯ ōs did not yet have a nasal
in pre-Proto-Slavic is the fact that the remaining oblique forms are based on
the stems *nās‑, *u̯ ās‑ in Slavic, e.g. PS gen.–loc. *ˈnāsu (CS *na̋ sъ) from pre-PS
gen. *ˈnās‑am, loc. *ˈnās-su (Meillet 1897: 96). Note that the acute is probably
not regular here; it may have been introduced by analogy with the dual forms.
Nominal Inflection 255
PIE Brugmann 1909–1911: 244–247: OCS ‑ъ reflects PIE *‑ōm̃ | Beekes 1995/
2011: 188: PIE *‑om, originally an adjectival form, is preserved in Umbr.,
Celt. and BS; IIr. endings reflect remade *‑ahom or *‑oom | Rasmussen
forthc. a § 2: PIE *‑õm or *‑oom | Debrunner & Wackernagel 1930/
1975: 67–69, 71–72: Ved., Av., Gk., Li. have generalised original o-stem end-
ing PIE *‑ō̃m; short vowel of Slavic and Celtic points to generalisation of
pure case–number marker *‑om | Rix 1976/1992: 156–157: Gk. ending
reflects o-stem ending PIE *‑ōm, which largely replaced original *‑om,
preserved in OCS ‑ъ | Sihler 1995: 254–255: PIE marker was perhaps
*‑om, contracting with the stem-vowel to *‑ōm in o-stems, with differ-
ent generalisations in various languages | Weiss 2009/2011: 199, 208:
PIE *‑ohom; there is no evidence for *‑om | Krahe 1942/1967: 35, 40,
45: PIE *‑ōm | Boutkan 1995: 140, 259, 262, 268, 275: PIE reconstruc-
tion *‑om is supported by BS and probably also by Toch., Arm., Iranian,
Anat.; Go. ‑e is i-stem ending *‑eiom; expected Gmc. ending *‑aN is not
preserved | Kloekhorst 2008a: 105, 172: Hi. ‑an reflects PIE *‑om, also
in OCS ‑ъ, Li. ‑ų̃ , OIr. ferN; Ved. ‑ām, Gk. ‑ων are generalised o-stem endings
PBS Olander 2009: 185: PIE case–number marker was *‑om | Endzelīns
1971: 164–165: C-stem ending is preserved in certain Li. and Lv.
words | Stang 1966: 222–223: Li. ‑ų̃ , Lv. ‑u reflect PBalt. *‑ōn, cor-
responding to Ved. ‑ā́m, Gk. ‑ῶν, La. ‑um | Otrębski 1956: 56: Li. ‑ų <
*‑uon goes back to PIE *‑ōm | Endzelīns 1923: 323: Lv. akmeńu has
i-stem ending; old ending is preserved in Lv. dial. akmanu
PS Vondrák 1908/1928: 6 (1906/1924: 264–270): the form is based on the nom.
pl. stem; Slavic ‑ъ is from *‑ō̃m with shortening and compensatory length-
256 Chapter 3
ening | Hujer 1910: 120–124: all stems had PIE *‑ōm ̃ , regularly yield-
ing Slavic ‑ъ, ‑jь (whereas acute *‑ṓm/n yielded Slavic ‑y) | Meillet
1924/1934: 393–394, 422: PIE had a form with *ō and a form with *o; the
latter is required by Slavic, OPr., Umbr., Irish | Vaillant 1958: 187:
Slavic ‑ъ is from *‑on, but Li. ‑ų̃ , Lv. dial. ‑u reflect *‑ōn; [see also o-stem
nouns, § 3.15.5] | Bräuer 1969b: 10, 55, 72, 83: CS *‑ъ reflects PIE
*‑om, alongside PIE *‑ōm in other languages; [see also o-stem nouns,
§ 3.15.5] | Arumaa 1985: 115: IIr., Gk., Li. point to PIE *‑ō̃m; Slavic,
OIr. point to short ending; Slavic may also have late shortening of *‑ōn
to *‑on, which would explain metatony in e.g. Čak. nom. sg. krȁva, gen.
pl. krȃv | Igartua 2005a: 347–348: it seems preferable to assume
a short ending *‑om or *‑on, directly reflected in Slavic, Sabellic and
Celtic | Aitzetmüller 1978/1991: 95, 98, 101, 103: OCS ‑ъ from is *‑om;
[see also u-stems, § 3.15.3]
In Celtic, Old Irish has the reflex of a short vowel in the genitive plural, e.g.
o-stem ferN, pointing to *‑om (I am grateful to Anders Richardt Jørgensen for
his input concerning the Celtic material). The explanation of the Old Irish
form as the result of an early shortening of *‑ōm to *‑om is contradicted by
Celtiberian, where e.g. o-stem alizokum shows that *‑ōm became *‑ūm by
the pre-Proto-Celtic development of *ō to *ū; and an ending *‑ūm would
have resulted in OIr. **fiurN. This situation is most easy to understand if we
assume that Proto-Celtic had two endings, *‑om in consonant stems and
*‑ūm in o- and ā-stems, corresponding to the Proto-Balto-Slavic situation (see
Gorrochategui 1994: 326–327). In Old Irish the short ending was generalised,
whereas Celtiberian generalised the long ending. Alternative explanations are
proposed by McCone (1996: 57–58, 61), who reconstructs an intermediate stage
PCelt. *‑ọm, and by Eska (2006), according to whom the change of *ō to *ū took
place in Common Celtic after the shortening of *‑ōm to *‑om, which did not
affect Hispano-Celtic.
While it is true that most Indo-European branches point to a long case–
number marker in the genitive plural, the supposed shortening of *‑ōm to *‑ъ
in Slavic is quite implausible. I agree with Meillet’s (1922: 258) criticism of the
idea of such a shortening: “En slave, le ‑ŭ de vlĭkŭ, slovesŭ ne peut être rapporté
à une ancienne longue que par des hypothèses arbitraires et contraires à tout
le traitement des longues en slave”. As I see it, the most important point in
Meillet’s criticism is the latter part, namely the fact that the shortening would
go against the otherwise exceptionless preservation of the distinction between
Proto-Indo-European final syllables containing long and short vowels in Slavic
(see also Meillet 1915: 7). It only adds to the point that the assumed shortening
is ad hoc in a Slavic context (as admitted by Jasanoff 1983a: 144).
Since a case–number marker *‑ōm or *‑o(h)om has an atypical structure
for Proto-Indo–European, and since the generalisation of the ending of the o-
and ā-stems in most Indo-European branches is quite easily understandable,
I assume that the Proto-Indo-European genitive plural marker was *‑(h)om. In
Slavic the marker, still clearly seen in the i‑, u‑ and consonant stems in attested
Slavic dialects, spread analogically to the o- and ā-stems where it replaced the
inherited endings.
PBS PIE *‑(h)om is continued as PBS *‑am [1|7]. In East Baltic the ending was
replaced with the ō- and ā-stem ending PBS *‑ōm < PIE *‑o(h)om, *‑ah₂(h)om
(see §§ 3.15.4 and 3.15.5). At a later, partly historical stage, a number of conso-
nant stems took over the i-stem ending (§ 3.15.2), e.g. Li. dial. akmenių̃ . While
it is formally possible to derive Lv. ‑u from PBS *‑am, it is more likely that it
should be identified with Li. ‑ų from PBS *‑ōm. The genitive plural of conso-
nant stems is not preserved in Old Prussian.
258 Chapter 3
PS PBS *‑am is reflected as PS *‑u [19] (CS *‑ъ [29]), preserved in the old
Slavic dialects. The ending spread from the consonant stems and the i- and
u-stems to the o- and ā-stems.
PS *gasteˈi ̯u (CS *gostь̀ jь) PBS *‑ˈei ̯am, *ˌ‑i ̯am PIE *‑ei ̯(h)om, *‑i ̯om / *‑ihom
PIE Brugmann 1909–1911: 242–244: PIE *‑(i)i̯ōm; OCS ‑ьjь is from *‑ejъ |
Beekes 1995/2011: 203: PIE *‑eiom [implicit from u-stems; see also C-stems,
§ 3.15.1] | Rasmussen forthc. a § 6: PBS *‑i̯ōn | Debrunner &
Wackernagel 1930/1975: 71–72, 162–163: OCS ‑ьjь, Go. þrije point to PIE
*‑ei̯‑; Ved., Av. forms with ‑n‑ are modelled on o-stems; [see also C-stems,
§ 3.15.1] | Rix 1976/1992: 157: Att. ‑εων replaces ‑ιων (preserved in Ionic)
by analogy with gen. sg. ‑εως | Sihler 1995: 314–315, 318: PIE *‑i̯ō̆m is pre-
served in La. ‑ium | Weiss 2009/2011: 242, 246: La. ‑ium reflects *‑i̯ōm <
PIE *‑i̯ohom | Krahe 1942/1967: 27–28: Gk. dial. ‑ιων, La. ‑ium point to
PIE *‑i(i̯)ōm, on which OSax. gestio, OHG gestio are based | Boutkan
1995: 140, 249–250: Go. ‑e is from *‑eiom; WGmc. may show indirect traces
of hysterodynamic *‑iom | Kloekhorst 2008a: –
PBS Olander 2009: 185–186: CS *‑ь̀ jь is the regular reflex of PIE *‑éi̯om |
Endzelīns 1971: 153–154 | Stang 1966: 212–213: Slavic ‑ьjь points to
*‑ii̯ōn or *‑ei̯ōn; Baltic forms are unclear | Otrębski 1956: 42: Li. ‑’ų̃ is
from *‑i̯ōm | Endzelīns 1923: 318–319: Lv. ‑’u, Li. ‑’ų̃ are from PIE *‑i̯õm
or *‑ii̯õm, also in OCS ‑ьjь, Gk. dial. ‑ιων
PS Vondrák 1908/1928: 6: the form is based on nom. pl. stem; [see also
C-stems, § 3.15.1] | Hujer 1910: 123–124: Slavic ‑ьjь is from *‑ei̯ōm; [see
also C-stems, § 3.15.1] | Meillet 1924/1934: 419: Slavic ‑ьjь is from *‑ei̯‑
plus ending | Vaillant 1958: 136–137: PIE *‑(i)i̯ō̆n; Gk. ‑εων reflects
*‑ei̯ōn with analogical *‑ei̯‑ from nom. pl. | Bräuer 1969a: 157: CS *‑ьjь
is from PIE variant *‑ei̯om; [see also o-stem nouns, § 3.15.5] | Arumaa
Nominal Inflection 259
PS PBS *‑ˈei̯am yielded PS *‑eˈi̯u [13|19] (CS *‑ьjь [20|24|29]). The ending is
preserved in the old Slavic dialects.
PS *sūnaˈu̯u (CS *synòvъ) PBS *‑ˈau̯am, *ˌ‑u̯am PIE *‑eu̯(h)om, *‑u̯om / *‑uhom
PIE Brugmann 1909–1911: 242–244: PIE *‑(u)u̯ ōm; OCS ‑ovъ reflects
*‑evъ | Beekes 1995/ 2011: 203: PIE *‑euom; [see also C-stems,
§ 3.15.1] | Rasmussen forthc. a § 6: PBS *‑u̯ ōn / *‑au̯ ōn | Debrunner
& Wackernagel 1930/1975: 71–72, 162–163: OCS ‑ovъ, Go. ‑iwe, Gk. ‑έ(ϝ)ων
point to PIE *‑eu̯ ‑; Ved., Av. forms with ‑n‑ are modelled on o-stems; [see
also C-stems, § 3.15.1] | Rix 1976/1992: – | Sihler 1995: 327: PIE
form is uncertain, but probably contained *‑eu̯ ‑; Gk. ‑εων shows generali-
sation of ε | Weiss 2009/2011: 249, 252: PIE *‑u̯ ohom yielded La. ‑uum,
which may also continue PIE variant with e-grade of suffix | Krahe
1942/1967: 33: Gk. ‑εων is from PIE *‑eu̯ ōm; Go. ‑iwe replaces PGmc.
*‑iwō | Boutkan 1995: 258: Go. ‑iwe is from PIE *‑eu̯ om with second-
arily introduced i-stem ending ‑e | Kloekhorst 2008a: –
PBS Olander 2009: 185–186: CS *‑òvъ is the regular reflex PIE
*‑éu̯ om | Endzelīns 1971: 158 | Stang 1966: 217–218: OCS ‑ovъ,
Go. ‑iwe point to PIE *‑eu̯ ōm; Baltic endings are perhaps taken from
o-stems | Otrębski 1956: 48: Li. ‑ų̃ , Av. ‑uuąm go back to PIE *‑u̯ ōm;
Slavic ‑ovъ is from PIE *‑eu̯ ōm | Endzelīns 1923: 329–330: Lv. ‑u, Li. ‑ų̃
perhaps reflect PBalt. *‑u̯ õn
PS Vondrák 1908/1928: 6: the form is based on nom. pl. stem; [see also
C-stems, § 3.15.1] | Hujer 1910: 123–124: Slavic ‑ovъ is from *‑eu̯ ōm;
[see also C-stems, § 3.15.1] | Meillet 1924/ 1934: 414 | Vaillant
1958: 111–112: PIE *‑(u)u̯ ōn | Bräuer 1969a: 147: PIE *‑eu̯ om; [see also
o-stem nouns, § 3.15.5] | Arumaa 1985: 128–129: CS *‑ovъ goes back
to *‑ou̯ ōm | Igartua 2005a: 294–295: PIE *‑e/ou̯ ōm | Aitzetmüller
1978/1991: 71: OCS ‑ovъ is from full grade of u-stem suffix followed by ptcl.
*‑om, also found in OIr.; *‑ōm found in other languages reflects *‑oom,
*‑āom
*‑õm or *‑ãm; Ved. ‑ānām and Gmc. forms like OHG ‑ōno are analogi-
cal from fem. n-stems | Boutkan 1995: 140, 229: PGmc. *‑ōan is from
PIE *‑h₂om, with introduction of *‑ō‑ < *‑eh₂‑; Gmc. ‑(e)na is taken from
n-stems | Kloekhorst 2008a: –
PBS Olander 2009: 185–186: PBS *‑ˈōn reflects PIE *‑áh₂om, with analogical
final accent; CS *‑ъ represents generalisation of case–number marker
*‑om from i‑, u‑, C-stems | Endzelīns 1971: 144 | Stang 1966: 200: Li.
‑ų̃ , Lv. ‑u go back to PBalt. *‑ōn < PIE *‑ōm | Otrębski 1956: 26: Li. has
generalised ‑ų in all paradigms | Endzelīns 1923: 307: Lv. ‑u, Li. ‑ų̃ are
from PBalt. *‑õn
PS Vondrák 1908/1928: [see C-stems, § 3.15.1] | Hujer 1910: 122–123: Slavic
‑ъ, ‑jь probably go back to *‑ō̃m, less likely to *‑ā̃m; [see also C-stems,
§ 3.15.1] | Meillet 1924/1934: 399: Slavic ‑ъ is from zero grade of suf-
fix, *‑ə‑, followed by *‑on | Vaillant 1958: 84: PIE *‑ōn/m is from *‑ā‑
plus *‑ō̆n/m; [see also o-stem nouns, § 3.15.5] | Bräuer 1969a: 106, 127:
Slavic ‑ъ reflects unclear *‑om; other IE languages point to *‑ōm, perhaps
from *‑ā plus *‑ōm; CS *‑jь < *‑jъ reflects PIE *‑i̯om, alongside *‑i̯ōm or
*‑i̯ām in other languages [see also o-stem nouns, § 3.15.5] | Arumaa
1985: 154: CS *‑ъ is from *‑om, shortened from *‑ō̃m | Igartua 2005a:
235–238: PIE *‑ōm < *‑eh₂om; Slavic ‑ъ reflects *‑om found in the other
paradigms | Aitzetmüller 1978/1991: 88, 90: OCS ‑ъ, ‑jь reflect ptcl. *‑om
added directly to the root; [see also C-stems, § 3.15.1; u-stems, § 3.15.3]
PBS PIE *‑ah₂(h)om yielded pre-PBS *ˌ‑ōm [1|3|4], which was remade to PBS
*‑ˈōm with final accentuation by analogy with the i- and u-stems. The ending,
which merged with the o-stem ending *‑o(h)om, is preserved in East Baltic,
where it was extended to the genitive plural of all paradigms. In Old Prussian
PBS *‑ōm yielded ‑on in menschon (Cat. I).
PS In Slavic the expected ending PBS *‑ˈōm > PS *‑ˈān [15|19] was replaced
with the reflex of PBS *‑am from the i‑, u‑ and consonant stems, i.e. PS *ˈ‑u [19]
(CS *‑ъ [29]; *‑jь [20|29]). The endings of the hard and soft stems are preserved
in the old Slavic dialects. The Serbo-Croatian ending ‑ā may be based on anal-
ogy with a genitive plural ending ‑ī, which arose in the i-stems (Johnson 1972:
356; Kortlandt 1978/2009: 115–116 with references); subsequently ‑ā spread from
the ā-stems to the other stems.
Nominal Inflection 263
OCS gradъ; kon’ь Li. langų̃ ; def. adj. mažų̃ jų; Ved. devā́ñ (jánma),
ORu. stolъ; konь; ONovg. (u) pron. tų̃ →devānām
vežьnikъ; mužь Lv. tȩ̃vu; def. adj. mazuõ; OAv. →yasnanąm; YAv.
OCz. chlap; oráč pron. tùo →š́iiaoϑnanąm
OPr. grecon, grecun, Gk. ἀγρῶν
→swintan La. →lupōrum, deum; Osc.
Núvlanúm
Go. →dage; ONor. daga
Hi. →antuḫšaš
PIE Brugmann 1909–1911: 238–241 (1897: 391): IIr. *‑nām has *n from n-stems;
OCS ‑ъ reflects PIE circumflex *‑ōm̃ | Beekes 1995/2011: 212–213: PIE
*‑om is preserved in BS and Old Irish; other languages have innovated;
[see also C-stems, § 3.15.1] | Rasmussen forthc. a § 4: PIIr. *‑ānām
represents dissimilated *‑āmōm; Baltic forms, OCS ‑ъ are from PBS
*‑ō̃n < PIE *‑õm | Debrunner & Wackernagel 1930/ 1975: 70–71:
Ved. ‑ānām reflects a PIIr. innovation for PIE *‑ō̃m, perhaps by analogy
with ā-stems; [see also C-stems, § 3.15.1] | Rix 1976/1992: 140–141:
PIE *‑ōm < *‑oom; [see also C-stems, § 3.15.1] | Sihler 1995: 264–265:
PIE *‑ōm; La. ‑ōrum is modelled on ā-stem ‑ārum < pron. *‑ah₂sōm; [see
also C-stems, § 3.15.1] | Weiss 2009/2011: 208, 224: OLa. ‑um < *‑ōm <
PIE *‑ohom; La. ‑ōrum is modeled after ‑ārum, of pron. origin | Krahe
1942/1967: 11: PIE *‑ōm (< *‑o‑ plus *‑ōm) is preserved in ONor. and
WGmc.; Go. ‑e perhaps reflects *ō after *i̯ | Boutkan 1995: 194–196:
ONor., WGmc. endings do not reflect *‑ōm but ā-stem ending *‑ōan; Go.
‑e represents i-stem *‑eian | Kloekhorst 2008a: 104: Hi. ‑an goes back
to PIE *‑oom
PBS Olander 2009: 185–186: PBS *‑ˈōn reflects PIE *‑óom, with analogical final
accent; CS *‑ъ results from generalisation of case–number marker *‑om
from i‑, u‑, C-stems | Endzelīns 1971: 136: Li. ‑ų, Lv. ‑u go back to PIE
*‑ōm; OPr. ‑an, OCS ‑ъ probably reflect PIE *‑om | Stang 1966: 184–185:
Li. ‑ų̃ , Lv. ‑u reflect *‑uõn < *‑ōn; OPr. grecon, grekun, griquan are from
*‑ōn; swintan apparently reflects *‑an; Slavic ‑ъ reflects *‑ōn | Otrębski
1956: 15: Li. ‑ų̃ reflects *‑un < *‑uon < *‑ōn < PIE *‑ōm | Endzelīns 1923:
295: Lv. ‑u, pron. ‑uo, Li. ‑ų̃ are from PBalt. *‑ōn [i.e. *‑õn]
264 Chapter 3
PBS PIE *‑o(h)om yielded pre-PBS *ˌ‑ōm [1|3|4], but the form was altered to
PBS *ˈ‑ōm with final accentuation by analogy with the i- and u-stems. In East
Baltic PBS *‑ōm yielded Li. ‑ų, Lv. ‑u; an intermediate stage is preserved in Lv.
pron. tùo (see Endzelīns 1923: 295; Stang loc. cit.), which rules out a Balto-Slavic
proto-form *‑um as assumed by Kortlandt (e.g. 1978/2009: 116). The East Baltic
development PBS *‑ōm > *‑uon > *‑un > Li. ‑ų, Lv. ‑u is paralleled by that of
o-stem dat. sg. *‑ōi > *‑uoi > ‑ui (Kümmel 2013a: 199).
The o-stem ending, which merged with the corresponding ā-stem ending,
was generalised in East Baltic, replacing the reflex of PIE *‑om in the i‑, u‑ and
consonant stems. The Old Prussian material points to both *‑ōn and *‑an, the
former reflecting the Proto-Balto-Slavic o-stem ending *‑ōm < PIE *‑o(h)om
and the latter reflecting the consonant-stem ending *‑am < PIE *‑(h)om (simi-
larly Meillet 1922: 258; Endzelīns 1944: 87; see also Stang 1966: 184–185).
Thus while East Baltic witnessed a spread of the long ending PBS *‑ōm < PIE
*‑o(h)om (and *‑ah₂(h)om), Slavic generalised the short ending PBS *‑am < PIE
*‑om from the consonant stems; Old Prussian shows both endings.
Nominal Inflection 265
PS In Slavic the consonant-stem ending PBS *‑am > PS *‑ˈu [19] (CS *‑ъ [29];
*‑jь [20|29]), also appearing in the i- and u-stem endings *‑eˈi̯am, *‑aˈu̯ am, was
introduced in the o-stems, as well as in the ā-stems. The ending is preserved in
the old Slavic dialects, but at a later stage there was a tendency in masculine
nouns to replace the original ending PS *‑ˈu > ‑∅ with the reflexes of the cor-
responding u-stem ending PS *‑aˈu̯ u, in order to avoid a zero ending that was
homonymous with the nominative singular. For SCr. ‑ā see § 3.15.4.
PS →*tai ̯ˈxu (CS *tě ́xъ; *jíxъ) PBS *‑ˈai ̯s‑am, *‑ˈei ̯s‑am PIE *‑oi ̯s‑om, *‑ei ̯s‑om
PIE Brugmann 1909–1911: 369–370: OCS těxъ, jixъ, ONor. þeira, Ved. téṣām,
reflect PIE *‑oisōm, alongside *‑eisōm (*‑sōm̃ ) in Ved. eṣā́m, Osc. eisunk,
OPr. stēison | Beekes 1995/2011: 227: PIE form was perhaps *toisom,
but Go. þize points to *‑es‑; Hi. kēnzan reflects *‑ns‑om with unex-
plained *n | Rasmussen forthc. a § 14: PIE *‑oi̯sõm | Debrunner
& Wackernagel 1930/1975: 503: Ved. téṣām corresponds to OPr. steison,
OCS těxъ | Rix 1976/1992: 182–183, 185: Ved. téṣām, ONor. þeira go
back to PIE *toi̯sōm; Gk. τῶν has noun ending; Gk. adj. τοῖος is prob-
ably based on gen. pl. τοίων < PIE *tó-isōm | Sihler 1995: 389, 391:
PIE *toi̯sō̆m, *ei̯som | Weiss 2009/2011: 336–338, 340: PIE *toi‑s‑ohom
(*tóisōm), *(h₁)eisohom; diphthongal stem *toi‑ was originally a collec-
tive | Krahe 1942/1967: 63: ONor. þeira goes back to PGmc. *þaizō(m);
Go þize has ‑i‑ from gen. sg., and ‑e under the influence of noun ending
‑e | Boutkan 1995: – | Kloekhorst 2008a: 426–427: Hi. ‑nzan, Lyc.
gen. pl. ebẽhẽ reflect *‑nhsom, with *‑som as in Ved. téṣām, La. eōrum, OCS
těxъ
PBS Olander 2009: – | Endzelīns 1971: 193–194: OPr. steison is reminiscent
of OCS těxъ, Ved. téṣām, ONor. þeira | Stang 1966: 243: OPr. stēison
266 Chapter 3
has *ei in contrast to OCS, Gmc. | Otrębski 1956: 151: Li. tų̃ has noun
ending | Endzelīns 1923: 390: Lv. tùo corresponds to Li. tų̃ , Gk. τῶν
PS Vondrák 1908/1928: 79: OCS těxъ is from stem *toi̯‑ plus pron. ending
*‑sōm | Hujer 1910: – | Meillet 1924/1934: 436 | Vaillant 1958:
375: Slavic ‑ěxъ reflects PIE *‑oison, alongside *‑oisōn; OPr. ‑eis‑ results
from contamination of sg. ‑es‑ and pl. *‑ais‑; *‑ois‑ is also preserved in
ONor. ‑eir‑, OEng. ‑ār‑, Go. adj. gen. pl. blindaiz[e]; Go. þiz‑ has *‑es‑
from sg. | Bräuer 1969a, 1969b: – | Arumaa 1985: 176: Slavic
těxъ reflects *toisōm, also in Ved. téṣām; unclear if lack of separate
feminine form in Slavic is an archaism or an innovation | Igartua
2005a: – | Aitzetmüller 1978/1991: 116: OCS těxъ goes back to PIE
*toisō̆m, also in Ved. téṣām, ONor. þeira; lack of separate feminine form in
Slavic is probably an archaism
PIE Ved. téṣām, OAv., YAv. auuaēšąm, ONor. þeira and the Balto-Slavic evi-
dence point to an ending *‑oi̯som, with the usual substitution of the genitive
marker *‑om with *‑ōm in Indo-Iranian and Germanic (see § 3.15.1). It is pos-
sible that the paradigm of the Greek adjective τοῖος ‘such’ is based on the geni-
tive plural form τοίων, which could reflect the ending *‑oi̯sōm. From the stem
with e-grade Oscan has eisunk, reflecting PIE *ei̯som plus a particle.
The existence of a separate feminine form PIE *‑ah₂som has been
questioned on the basis of Slavic, which has one form for all three genders
(Aitzetmüller loc. cit.; cf. Vaillant 1958: 374; Arumaa loc. cit.). The problem is
difficult to solve.
PBS PIE *‑oi̯som yielded PBS *‑ˈai̯sam [7], preserved in Slavic. The e-grade
variant *‑ei̯som > PBS *‑ˈei̯sam [7] is continued in OPr. stēison.
PS *duktermə (CS *pol’amъ, →*dъt’erьmъ, PBS *ˌ‑mas PIE dat.–abl. *‑bʰi ̯os
→*‑omъ)
PIE Brugmann 1909–1911: 119–120, 257–262: PIE form is unclear; Li. ‑mus, OPr.
‑mans are perhaps based on analogy with acc. pl. | Beekes 1995/2011:
188–189: Slavic ‑mъ, OLi. ‑mus point to PIE *‑mus; other languages have
introduced *bʰ from instr. *‑bʰi | Rasmussen forthc. a §§ 2, 4, 5: PIE
*‑bʰi̯os; Gmc. and BS *m is perhaps from pronouns like *tésmõi̯; Li. ‑mus,
OPr. ‑mans may have been influenced by acc. pl. *‑ōns | Debrunner &
Wackernagel 1930/1975: 13, 66–67, 75, 209: relationship between *bʰ and
*m is uncertain | Rix 1976/1992: 117: PIE ending was *‑bʰos, perhaps
originally *‑mos, preserved in OCS ‑mъ, OLi. ‑mus; *‑bʰos may have been
remade based on instr. *‑bʰi; IIr. *i̯ is from instr. pl. *‑bʰi; [see also loc. pl.,
§ 3.18.1] | Sihler 1995: 248, 251–252, 286: PIE *‑bʰos, *‑mos; forms with
*bʰ and *m are not related | Weiss 2009/2011: 199, 207–208: *‑bʰos
(outside IIr.) reflects *‑bʰi̯os, either phonologically or analogically; BS and
Gmc. forms continue *‑mos | Krahe 1942/1967: 40: Gmc. forms reflect
instr. pl. PIE *‑mis | Boutkan 1995: 259, 261–262, 264, 269–271, 275, 278:
PIE, PGmc. *‑mus; [see also o-stems, § 3.16.5] | Kloekhorst 2008a: 105,
214: Hi. dat.–loc. ‑aš is from PIE *‑os(?)
PBS Olander 2009: 188–189: reflex of expected unaccented C-stem form is
perhaps preserved in ORu. zvěŕ em, Ru. détjam | Endzelīns 1971: 165:
Baltic and Slavic endings are modelled on i-stems | Stang 1966: 223:
Li. ‑i‑ is from i-stems; [see also o-stems, § 3.16.5] | Otrębski 1956: 53,
268 Chapter 3
57; [see also o-stems, § 3.16.5] | Endzelīns 1923: 323: Lv. form is from
i̯o-stems
PS Vondrák 1908/1928: 7 | Hujer 1910: 16–18, 147–150, 154–155: Li. ‑ms
reflects *‑mas < PIE *‑mos; Slavic ‑mъ is from *‑mon < PIE *‑mom, a vari-
ant of *‑bʰ(i̯)om found in Ved. dat. sg. túbhyam, pl. yuṣmábhyam; Li. ‑mus,
OPr. ‑mans reflect *‑mons, a contamination of *‑mon and *‑mos; vari-
ants with *bʰ and *m existed side by side in PIE | Meillet 1924/1934:
394–395: Slavic ‑mъ goes back to *‑mus or *‑mos | Vaillant 1958: 187–
189: variation in Slavic points to recent substitution of inherited C-stem
ending by o- and i-stem endings; [see also o-stems, § 3.16.5] | Bräuer
1969b: 10, 55, 72, 83 | Arumaa 1985: 115–117: origin of *m is uncertain;
original form was probably *‑bʰos | Igartua 2005a: 348–349: PIE
*‑bʰ/mos; [see also o-stems, § 3.16.5] | Aitzetmüller 1978/1991: 95, 98,
101; [see also u-stems, § 3.16.3]
PIE In Proto-Indo-European the dative and ablative plural were not distin-
guished formally. This situation is preserved in Indo-Iranian. In consonant
stems the ending was the dative-ablative plural marker attached directly to the
stem-final consonant.
The Hittite dative-locative marker ‑aš has been traced back to a Proto-
Indo-European o-stem locative plural ending *‑osu with apocope of *‑u (e.g.
Kammenhuber 1969: 305; Sihler 1995: 253; but cf. Melchert 1994: 182 with refer-
ences). It is more likely, however, that the Hittite marker reflects an element
*‑os, perhaps the original Proto-Indo-European dative-ablative plural ending
(e.g. Jasanoff 2009a: 140–141). The Greek form continues the locative plural
form. In Latin the i-stem ending has been introduced. The Germanic forms
may contain the dative-ablative plural ending *‑mos < PIE *‑bʰi̯os or the instru-
mental plural ending *‑bʰis. Evidence for one of the endings is limited to OEng.
dem. pron. þǣm, num. tvǣm, where the umlauted vowel points to PGmc. *‑miz
from PIE *‑bʰis (see e.g. Kroonen 2013: 529). In other instances it is likely that
the Germanic forms simply represent the merger of the dative-ablative and
instrumental plural endings (thus also e.g. Brugmann loc. cit.).
There are two main problems concerning the reconstruction of the case–
number marker of the dative-ablative plural in Proto-Indo-European: the ini-
tial consonantism and the vocalism of the marker. While the latter problem is
relevant only for the dative-ablative plural marker, the former also applies to
other markers.
As for the initial consonant, it is an old and well-known problem that the
various Indo-European branches seem to disagree about the initial consonan-
tism of the case–number markers of the dative-ablative-instrumental dual,
Nominal Inflection 269
the dative-ablative plural and the instrumental plural. Some branches point to
PIE *bʰi̯ (Indo-Iranian), others to *bʰ (Italic, Celtic, Armenian, Greek) and still
others to *m (Balto-Slavic, Germanic). The Anatolian and Tocharian evidence
(see Katz 1998: 248–250) is too meagre to contribute decisively in solving the
question.
Various hypotheses have been advanced to solve the apparent discrep-
ancy among the branches (see e.g. Brugmann loc. cit.; Matzinger 2001: 188).
A currently widespread view holds that in Proto-Indo-European the dative-
ablative plural marker contained *m, while the instrumental plural marker
contained *bʰ. After the dissolution of the proto-language *m was generalised
in some branches, *bʰ in others (Hirt 1895; Hoffmann apud Eichner 1974b: 29
n. 9a; Rix loc. cit.; Beekes 1985: 143–144; loc. cit.; Katz 1998: 248–249, following
Beekes; Meier-Brügger 2003: 197; cf. the criticism in Matzinger 2001: 189; Hill
2012: 179–181). Other authors trace the *m of Balto-Slavic and Germanic back
to pronominal forms like PIE dat. sg. *te/osmōi̯ or dat. pl. *smos (preserved in
Hi. ‑šmaš, TochB ‑me), whence it spread to other case forms in these branches
(Matzinger 2001: 193–194; Jasanoff 2009a: 140–141; Rasmussen loc. cit.; cf. the
criticism in Melchert & Oettinger 2009: 65). Hill (2012: 186–192), on the other
hand, assumes that the starting point of the alternation between *m and *bʰ
was a pre-Proto-Indo-European development of *‑n̥ m‑ to *‑n̥ bʰ‑ in the n-stems,
from which *bʰ subsequently spread (for a recent assessment of this idea see
Vijūnas 2013: 95–96).
While these hypotheses are theoretically imaginable, I do not find them
convincing. First of all, the choice of *bʰ or *m in the individual branches
seems to be arbitrary. Also, the presence of *i̯ in the dative-ablative plural and
dative-ablative-instrumental dual in Indo-Iranian requires additional hypoth-
eses, for instance that *i̯ was introduced from the instrumental plural (thus
e.g. Kümmel 1997: 118; Meier-Brügger 2003: 197). Alternatively, PIE *‑bʰi̯os,
retained in Indo-Iranian, was remade to *‑bʰos by analogy with instr. pl. *‑bʰis
(Rasmussen 1989b: 139 n. 22). In Jasanoff’s (2009a) view, the non-Anatolian
Indo-European dative plural *‑bʰ(i̯)os consisted of an adverbial form *bʰi fol-
lowed by the original dative plural marker *‑os, preserved in Hi. dat.–loc. pl.
‑aš (followed by Melchert & Oettinger 2009: 63–64; Weiss loc. cit.; see also Neu
1991: 14).
A straightforward solution to the problem is to assume that PIE *bʰi̯ was
retained in Indo-Iranian, developed into *bʰ in Italic, Celtic, Armenian and
Greek, and became *m in Balto-Slavic and Germanic. The idea of a regular
development of *bʰi̯ to *bʰ is sometimes found in the literature (e.g. Hackstein
2012b: 112; according to Weiss 2009/2011: 207 n. 10, “[i]t is not clear whether this
development was phonological or analogical”; see also Jasanoff 2009a: 141 n. 6).
270 Chapter 3
PBS In the attested Baltic and Slavic languages the consonant-stem ending
PIE *‑bʰi̯os > PBS *ˌ‑mas [4|7] has largely been replaced with the i-stem ending
PBS *‑imas < *‑ibʰi̯os; in Latvian the i̯o-stem ending is used. This process may
have already started in pre-Proto-Balto-Slavic, but there are indications in both
Baltic and Slavic that the original ending without *i survived until a relatively
late stage, at least in some cases. Archaic-looking accentuations such as ORu.
zvěŕ em, Ru. détjam may preserve a trace of the original consonant-stem end-
ing; in root nouns, where the ending was attached directly to the root, a form
ending in *‑bʰi̯ós would become unaccented by the mobility law [4] (Olander
2007a, loc. cit.). A more direct trace is constituted by the dative plural forms in
PS *‑āmə (CS *‑amъ) from pre-PS *‑ānmas; see below.
In Lithuanian the case–number marker PBS *‑mas was reduced to ‑ms,
an intermediate stage being attested in OLi. ‑mus (Ferrell 1965b: 98–99 n. 6;
Kazlauskas 1970, whose view is accepted by Stang 1975: 49; Olander 2005, also
for the following discussion). The usual Latvian dative-instrumental plural
marker ‑m is historically the dative-instrumental dual form, but in old texts
and (rarely) in dialects the original dative plural marker is preserved as ‑ms
(Mühlenbach 1903: 70–72; Endzelīns 1923: 296–298). Since PBS *u is preserved
in final syllables in Latvian, this form cannot reflect *‑mus (cf., however,
Endzelīns 1923: 296, who suggests that *u was lost in long adjective forms of
the type *labiemus-jiemus). Old and dialectal Latvian ‑ms therefore supports
the reconstruction PBS *‑mas.
Old Prussian has two different dative plural markers: ‑mans in nouns such as
waikammans, mergūmans and in pronouns such as 1pl. noūmans, 2pl. ioūmans;
and ‑mas, which is restricted to personal pronouns, e.g. 1pl. noūmas, 2pl.
ioūmas. The latter ending probably reflects PBS *‑mas, although, admittedly, a
reduced form *‑ms would perhaps be more in accordance with the general Old
Prussian phonological development (cf. Stang 1966: 119). The former ending
may have its ‑n‑ from the accusative plural ‑ans (thus e.g. Brugmann loc. cit.;
Berneker 1896: 196–197; Ferrell 1965b: 98; Kortlandt 1975/2011a: 48; Rasmussen
272 Chapter 3
loc. cit.; but cf. Endzelīns 1944: 87: “die alten Preussen haben doch ihre Sprache
nicht aus einer Grammatik erlernt, wo auf den Dativ der Akkusativ folgt”).
PS PBS *ˌ‑mas yielded PS *‑mə [17] (CS *‑mъ [29]); for the development of
PS *‑mə to ONovg. ‑mъ, not *‑me, see [29] and Olander 2012: 335. The origi-
nal ending was in most instances replaced with the i-stem ending PS *‑imə
(CS *‑ьmъ [29]), as in OCS ‑ьmъ, ORu. ‑ьmъ, or by the o-stem ending PS *‑amə
(CS *‑omъ [29]), as in ONovg. ‑omъ, OCz. ‑óm.
In the attested Slavic languages only the type formed with the suffix ‑jan‑,
referring to inhabitants of a place, preserves the original ending, e.g. ORu.
Vavilon’amъ, OSrb. dubrovьčamь, reflecting PS *‑ām‑ (CS *‑am‑) from *‑ānm‑
(Brugmann 1897: 387; Meillet 1924/1934: 423; Shevelov 1982: 370; Igartua loc. cit.;
for the Old Serbian ending see Loma 2013: 260; Vaillant loc. cit. 216–219).
PS *gastiˈmə (CS *gostь̀ mъ) PBS *‑ˈimas PIE dat.–abl. *‑ibʰi ̯os
alongside older ‑i‑ is due to influence from ā- and ē-stems; [see also o-stems,
§ 3.16.5] | Otrębski 1956: 42; [see also o-stems, § 3.16.5] | Endzelīns
1923: 319–320: Lv. ‑īm for ‑im is based on analogy with ā- and ē-stems; [see
also o-stems, § 3.16.5]
PS Vondrák 1908/1928: 7 | Hujer 1910:—[see C-stems, § 3.16.1] | Meillet
1924/1934: 419–420; [see also C-stems, § 3.16.1] | Vaillant 1958: 137: OCS
variant ‑emъ is taken from i̯o-stems; [see also o-stems, § 3.16.5] | Bräuer
1969a: 157: PIE *‑imos or *‑imus | Arumaa 1985: 129 | Igartua
2005a: 270: PIE *‑ibʰ/mos; CS *‑ьmъ reflects *‑imon; [see also o-stems,
§ 3.16.5] | Aitzetmüller 1978/1991: 75; [see also u-stems, § 3.16.3]
PBS PIE *‑ibʰi̯os regularly yielded PBS *‑ˈimas [7], preserved in Li. ‑ims and in
Old Latvian dat.–instr. forms like loudims. The remaining Latvian forms have
been remade according to existing patterns, the variants ‑īm and ‑im being
original dative-instrumental dual forms. Old Prussian has imported an n from
the accusative plural, as in the other paradigms (see § 3.16.5).
PS The old ending PBS *‑ˈimas yielded PS *‑ˈimə [17] (CS *‑ьmъ [29]), pre-
served in the old Slavic dialects. The Old Church Slavonic variant ‑emъ partly
originates in the i̯o-stems (cf. van Wijk 1931: 175).
PIE The expected ending *‑ubʰi̯os consisted of the zero grade of the u-stem
suffix followed by the dative plural marker *‑bʰi̯os (see § 3.16.1 for the marker).
The ending is preserved in Indo-Iranian and Latin. In Germanic the attested
forms may reflect the dative or instrumental plural. The Greek form is based
on the locative plural. In Hittite the dative-locative marker ‑aš is added to the
stem.
PBS PIE *‑ubʰi̯os yielded PBS *‑ˈumas [7], reflected in OLi. ‑umus and Li. ‑ums
(see § 3.16.5). The Latvian dative-instrumental plural is originally the dative-
instrumental dual form; I have not found any examples of the expected Old
and dialectal Latvian ending *‑ums.
in the document; ‑ъmъ in this text would represent actual ‑omъ, which may
reflect PS *‑umə or *‑amə (cf. Zaliznjak 1995/2004: 113, 548).
PBS PIE *‑ah₂bʰi̯os yielded PBS *‑ˈā̰mas [1|7]. The original ending is pre-
served in Lithuanian and Old Latvian (see § 3.16.1). As in the other stems, the
ending has been replaced with the dative-instrumental dual ending in modern
Latvian, and the n of the accusative plural has been introduced in Old Prussian.
PS The reflex of PBS *‑ˈā̰mas is PS *‑ˈāmə [13|17] (CS *‑amъ [29]; *‑jamъ
[20|27|29]). The ending is preserved in the old Slavic dialects.
PIE Brugmann 1909–1911: 259: PIIr. *‑ai‑ for PIE *‑o‑ is from pronouns; [see
also C-stems, § 3.16.1] | Beekes 1995/2011: 212: PIE *‑omus; [see also
C-stems, § 3.16.1] | Rasmussen forthc. a § 4: PIE *‑oi̯bʰi̯os; OCS ‑mъ
Nominal Inflection 277
locative marker ‑aš seems to go back to PIE *‑os, which some authors regard as
the original dative plural marker (see § 3.16.1).
PBS PIE *‑obʰi̯os regularly yielded PBS *‑ˈamas [7]. The ending was preserved
in Baltic, with the usual changes in the case–number marker (see § 3.16.1). In
Latvian the reflex of the thematic vowel was replaced with ‑ie‑ from the pro-
nominal declension, and in most dialects ‑ms was substituted with ‑m from the
dative-instrumental dual (see § 3.16.1).
PS PBS *‑ˈamas regularly yielded PS *‑aˈmə [13|17] (CS *‑omъ [29]; *‑jemъ
[20|29]), preserved in South and West Slavic. In East Slavic the historical rela-
tionship between frequent ‑omъ and rare ‑ъmъ is not clear (see § 3.16.3 on the
u-stems). Old Czech attestations of expected ‑ém etc. in the soft stems are
rare; the more common endings ‑óm etc. are imported from the hard stems
(Trávníček 1935: 41–42, 299).
PIE Brugmann 1909–1911: 262–267: Ved. ‑bhiḥ, Av. ‑bī�š,̆ Arm. ‑bk‘, ‑wk‘ and
probably OIr. ‑b point to PIE marker *‑bʰis; Gmc. *‑mz probably reflects
*‑mis, but may also contain *‑mos; Li. ‑mis, dial. ‑mi, Lv. ‑mis, ‑mi, OCS ‑mi
point to PBS *‑mīs, *‑mī | Beekes 1995/2011: 187, 189: PIE instr. pl. *‑bʰi
is preserved in Hom. ναῦφι | Rasmussen forthc. a § 2: PIE *‑bʰis; [see
Nominal Inflection 279
PBS PIE *‑bʰihs yielded PBS *‑mī�s̰ [1]; for the replacement of *bʰ with *m,
which may be analogical, see § 3.16.1. Seen in isolation, standard Li. ‑mis is
inconclusive with respect to the quantity of the vowel. However, Žemaitian
‑mis points to a long vowel since PBS *‑mis would have yielded **‑mẹs (Stang
loc. cit., 128; Olander 2005: 277 with further references). This means that stan-
dard Li. ‑mis is the result of a shortening by Leskien’s law, showing that we are
dealing with a form containing a laryngeal after the vowel, viz. PBS *‑mī�s̰ . This
reconstruction is supported by the circumstance that the marker attracted the
280 Chapter 3
accent by Saussure’s law in words with mobile accentuation, e.g. Li. mintimìs
from PBS *minˈtimī�s̰ (Olander 2004: 409–410).
Judging from Old Lithuanian forms like akmemis < from *‑enmī�s̰ , presenting
an ending without *‑i‑ before the marker, the original consonant-stem ending
was preserved in Proto-Balto-Slavic (Stang loc. cit.). At a later stage the conso-
nant-stem ending was substituted with the i-stem ending ‑imis in Lithuanian.
In Latvian the i̯o‑stem ending is used.
PS PBS *‑mī�s̰ yielded PS *‑mī [12|13|17] (CS *‑mi [29]). The old ending was
replaced almost everywhere with the i-stem ending PS *‑imī (CS *‑ьmь), e.g.
OCS masc. kamenьmi, fem. materьmi, or with the o-stem ending PS *‑ū (CS *‑y),
e.g. OCS neut. slovesy. As pointed out by Vaillant (loc. cit.), the variation found
in Slavic may indicate that the proto-language had still preserved the conso-
nant-stem ending. This ending is indeed still found in the ‑jane type, where e.g.
ORu. pol’ami from *‑āmī reflects *‑ān‑mī (cf. § 3.16.1).
PBS Olander 2009: 191: PBS *‑ˈimī�s̰ is from PIE *‑íbʰi(h)s with regular retention
of the accent | Endzelīns 1971: 154 [see dat. pl., § 3.16.1] | Stang
1966: 213; [see also ā-stems, § 3.17.4] | Otrębski 1956: 42 | Endzelīns
1923: 319–320 [see dat. pl., § 3.16.1]
PS Vondrák 1908/1928: 8 | Hujer 1910: 151–153 [see C-stems, § 3.17.1] |
Meillet 1924/ 1934: 419–420 | Vaillant 1958: 137 | Bräuer 1969a:
157: PIE *‑imīs | Arumaa 1985: 129: OCS ‑mi is from *‑mīs, also in
Baltic forms | Igartua 2005a: 271: PIE *‑ibʰ/mis; [see also ā-stems,
§ 3.17.4] | Aitzetmüller 1978/1991: 75; [see also u-stems, § 3.17.3]
PIE The ending PIE *‑ibʰi(h)s consisted of the zero grade of the stem-suffix
followed by the instrumental plural marker. The details of the marker are dis-
cussed in § 3.17.1. The ending is well preserved in the old Indo-European lan-
guages. In Latin, dat.–abl. ‑ibus reflects the dative plural ending.
PBS The Proto-Indo-European variant *‑ibʰihs yielded PBS *‑ˈimī�s̰ [1] (for the
substitution of *bʰ with *m see § 3.17.1). This ending is preserved in Lithuanian,
which shows regular shortening of the acute *ī�̰ by Leskien’s law and, in words
with mobile accentuation, advancement of the accent from the first to the sec-
ond syllable of the ending by Saussure’s law (see § 3.17.1). In standard Latvian
the original dative-instrumental dual is used as the plural; the ending ‑ims
found in Old Latvian reflects the dative plural ending PBS *‑imas. A form aci-
mis, with a reflex of the original instrumental plural ending *‑imī�s̰ , is report-
edly found in the dialect of Rucava near the border to Lithuania (Endzelīns
1923 loc. cit.).
PS PBS *‑ˈimī�s̰ is regularly reflected as PS *‑iˈmī [12|13|17] (CS *‑ьmi [29]), pre-
served in the old Slavic dialects.
PIE The ending PIE *‑ubʰi(h)s consisted of the zero grade of the stem-suffix
followed by the instrumental plural marker (see § 3.17.1). The ending is pre-
served in the old Indo-European languages. In Germanic the instrumental plu-
ral merged with the dative plural. La. ‑ubus reflects the dative plural ending.
PBS PIE *‑ubʰihs yielded PBS *‑ˈumī�s̰ [1], with the usual substitution of *bʰ
with *m (see § 3.17.1). The ending is preserved in Lithuanian, while in Latvian
the original dual form is used.
PS PBS *‑ˈumī�s̰ is reflected as PS *‑uˈmī [12|13|17] (CS *‑ъmi [29]). The end-
ing is preserved in Old Church Slavonic and Old Russian, whereas in the Old
Novgorod dialect only the o-stem ending is attested. Gebauer (1896: 325)
assumes that forms like *synmi have been replaced with o-stem forms of the
type syny in Old Czech. According to Trávníček (1935: 322), on the other hand,
forms in ‑mi existed in Old Czech but were not attested in the texts.
Nominal Inflection 283
PIE Brugmann 1909–1911: 264: PIE *‑ā‑bʰ‑, *‑ā‑m‑; [see also C-stems,
§ 3.17.1] | Beekes 1995/2011: 200: PIE *‑h₂bʰi | Rasmussen forthc.
a § 5: PBS *‑ā́mis; PIE probably *‑ah₂bʰis; length of Slavic ‑ami is prob-
ably from i̯o-stem ending *‑i̯õi̯s | Debrunner & Wackernagel 1930/
1975: 124; [see also C-stems, § 3.17.1] | Rix 1976/1992: 135: instr. Myc.
/‑āpʰi/ reflects PGk. *‑āpʰi < PIE *‑eh₂bʰi; OCS ‑ami has *m from dat.
pl. *‑mos; [see also C-stems, § 3.17.1] | Sihler 1995: 272: Gk. ‑αις, La.
‑īs are innovations based on o-stem instr. pl. | Weiss 2009/2011: 229:
PIE *‑eh₂bʰis | Krahe 1942/1967: 22: Gmc. forms reflect instr. PIE
*‑āmis | Boutkan 1995: 225, 229: Gmc. dat. pl. forms reflect *‑ōmus;
[see also o-stems, § 3.17.5] | Kloekhorst 2008a: –
PBS Olander 2009: 190–191: PBS *‑ˈā̰mī�s̰ is from PIE *‑áh₂bʰi(h)s with reg-
ular retention of the accent | Endzelīns 1971: 145 [see dat. pl.,
§ 3.16.4] | Stang 1966: 200–201: Li. ‑omis is from *‑āmī�s;́ original long
*ī is shown by NWŽem. dial. ‑mis (not *‑mẹs) | Otrębski 1956: 26–27:
Li. ‑omis perhaps contains *ī | Endzelīns 1923: 307–308 [see dat. pl.,
§ 3.16.4]
PS Vondrák 1908/ 1928: 8 | Hujer 1910: 151–153 [see C-stems,
§ 3.17.1] | Meillet 1924/1934: 399 | Vaillant 1958: 84 | Bräuer
1969a: 107, 127: CS *‑( j)ami reflects *‑(i̯)āmīs | Arumaa 1985: 154:
CS *‑mi is from *‑mis [read “*‑mīs”] | Igartua 2005a: 239–241:
PIE *‑ābʰ/mis; Baltic and Slavic marker perhaps reflects *‑mih₁s or
*‑min/ms | Aitzetmüller 1978/1991: 88, 91; [see also u-stems, § 3.17.3]
284 Chapter 3
PIE The ā-stem ending consisted of the suffix *‑ah₂‑ and the instrumental
plural marker *‑bʰi(h)s (see § 3.17.1). The ending is preserved in Indo-Iranian.
In Mycenaean Greek the original ending lives on in instrumental plural forms
like a‑ni‑ja‑pi /hāniāpʰi/, although the absence of a final *‑s shown by Hom.
adv. βίηφι ‘by might’ does not match the evidence of the other languages. The
Attic dative plural form φυγαῖς is based on analogy with the o-stem instrumen-
tal plural. The same goes for the Latin ablative in ‑īs, OLa. ‑eis. In Germanic the
instrumental plural of the ā-stems merged with the dative plural.
PS PBS *‑ˈā̰mī�s̰ became PS *‑ˈāmī [12|13|17] (CS *‑ami [29]; *‑jami [20|27|29]),
preserved in the old Slavic dialects.
PIE Brugmann 1909–1911: 267–269: OCS ‑y, ‑ji are not from PIE *‑ōĩs, but
from u-stem ending PIE *‑ūs | Beekes 1995/2011: 212: PIE *‑ōis <
*‑oh₁eis < *‑o‑eis(?) | Rasmussen forthc. a § 4: Li. ‑aĩs, Slavic ‑y reflect
PIE *‑ō̃is̯ | Debrunner & Wackernagel 1930/1975: 66, 105–108: Ved.
Nominal Inflection 285
‑áiḥ is from PIE *‑ōis (with uncertain vocalism), perhaps also in Slavic
‑y | Rix 1976/1992: 140–141: Gk. ‑οῖς reflects PIE *‑ōi̯s < pron. *‑ooi̯s;
Gk. dial. ‑οῖσι is loc. ending | Sihler 1995: 263–264: Gk. ‑οις, La. ‑īs are
from PIE *‑ōi̯s with “enigmatic” internal structure; Italic endings may also
reflect loc. *‑oi̯su | Weiss 2009/2011: 206–208, 224: La. ‑īs < OLa. ‑eis <
‑ois reflects instr. *‑ōis (and perhaps loc. *‑oisu), also in OCS ‑y | Krahe
1942/1967: 11: Gmc. forms reflect instr. PIE *‑omis | Boutkan 1995: 196–
198: PGmc. instr. pl. *‑mis (not dat. pl. *‑mus) is seen in OEng. pron. þǣm,
twǣm; in masc. o-stems we expect generalisation of dat. pl. *‑amus; [see
also dat. pl., § 3.16.5] | Kloekhorst 2008a: [see C-stem instr. sg., § 3.7.1]
PBS Olander 2009: 190: Li. ‑aĩs, PS *‑ˈū reflect PIE *‑ṓi̯s without hia-
tus | Endzelīns 1971: 137: Li. ‑ais, Lv. dial. ‑is correspond to Ved. ‑āiḥ;
standard Lv. uses the dat. form | Stang 1966: 65, 70, 186: Li. ‑aĩs, Lv. ‑is,
OCS ‑y go back to PIE *‑ōis; Li. ‑ais shows early shortening of *ōi before
*‑s; Lv ‑iem is of same origin as dat. ending | Otrębski 1956: 16: Li. ‑ais
reflects PIE *‑ōís [read “*‑ōis”?] with shortening | Endzelīns 1923:
295–299 [see dat. pl., § 3.16.5]
PS Vondrák 1908/1928: 7 | Hujer 1910: 160–164: Slavic ‑y is not from
PIE *‑ō̃is but has replaced expected *‑i with back formation from i̯o-
stem ‑ji < PIE *‑i̯ōis | Meillet 1924/1934: 410: Slavic ‑y < *‑ū reflects
*‑ōis | Vaillant 1958: 37–38: Slavic ‑y reflects PBS *‑uois < PIE *‑ōis <
*‑oīs, containing instr. sg. *‑ō followed by *‑ī�s;̆ East Baltic endings reflect
*‑ais with *a from other case endings | Bräuer 1969a: 27, 75: CS *‑y,
*‑ji perhaps reflect PIE *‑ōi̯s regularly; CS *‑y is not neut. u-stem instr. pl.
*‑ūs | Arumaa 1985: 142–144: PIE *‑ōis | Igartua 2005a: 176–184: PS
*‑ū perhaps reflects PIE *‑ōis | Aitzetmüller 1978/1991: 82, 84: OCS ‑y,
‑ji are unclear; ‑y is neither from PIE *‑ōis, nor based on analogy with i̯o-
stem ending
2012 (pp. 172–173: PIE *‑ōi̯s from pre-PIE *‑omis, following Cowgill 1985a: 108),
both with references to earlier literature; Hill’s proposal has been criticised by
Vijūnas 2013: 95–96.
PBS PIE *‑ōi̯s is retained as PBS *‑ˈōi̯s. Li. ‑aĩs and Lv. ‑is are probably the
result of a shortening of the long diphthong before a word-final consonant
(Stang loc. cit.; Kortlandt 1975/2011a: 48; 2012a: 256). The Latvian paradig-
matic dative-instrumental plural ending is originally a pronominal dative-
ablative-instrumental dual form; the old ending is preserved in adverbial
expressions like viênis pràtis ‘(of the) same opinion’. In Old Prussian the instru-
mental was perhaps no longer used in nouns, but the expression sen wissan
swaieis ‘with all one’s own people’ seems to preserve the original form (Stang
1966: 178).
PS PBS *‑ˈōi̯s yielded PS *‑ˈū [12|14|15|17] (CS *‑ỳ [29]; *‑jì [20|29]), with the
early development of pre-PS *‑i̯ə̄u̯ to PS *‑ū mentioned in [17]. The ending is
preserved in the old Slavic dialects.
*‑su | Rix 1976/1992: 113, 157–158: Gk. ‑σι is from PIE *‑su, with ‑i
from dat. sg. and instr. pl. ‑φι; *s was reintroduced by analogy with
s-stems and stems ending in a stop | Sihler 1995: 253: Gk. *‑si is from
*‑su, with ‑i from loc. sg. | Weiss 2009/2011: 199, 208 | Krahe 1942/
1967: – | Boutkan 1995: 270–271 | Kloekhorst 2008a: 214: Hi ‑aš is
not related to loc. pl. PIE *‑su; [see also dat. pl., § 3.16.1]
PBS Olander 2009: 188–189, 193: ORu. o zvěŕ ’ax, Ru. détjax may preserve
reflex of expected unaccented root-noun form | Endzelīns 1971:
165: Baltic and Slavic endings are modelled on i-stems | Stang
1966: 223: Li. ‑yse is modelled on i-stems; Lv. ‑’uos stems from (i)i̯o-
declension | Otrębski 1956: 54 | Endzelīns 1923: 323: Lv. ‑’uos is
based on i̯o-stems
PS Vondrák 1908/1928: 7: old C-stem ending is preserved in Slavic n-stems
such as ORu. Pol’axъ, OCz. Dol’as | Hujer 1910: 146–147: original Slavic
‑sъ is preserved in C-stems like OCz. Dol’as; Slavic ‑jasъ (for *‑jǫsъ?) from
*‑i̯ōnsu is analogical to endings beginning with a vowel | Meillet
1924/1934: 395–396, 324: Slavic *‑sъ is preserved in C-stems like OCz.
Pol’as; *‑jasъ may be the regular result of *‑jęsъ < *‑jans in OCz., ORu.,
Čak., but is analogical in OSrb. and Sln. | Vaillant 1958: 187–189: varia-
tion in Slavic points to recent substitution of inherited C-stem ending
with o- and i-stem endings; original *‑sъ is preserved in OCz. Dol’as,
v Polas | Bräuer 1969b: 10, 55, 73, 83 | Arumaa 1985: 118 | Igartua
2005a: 350–353 | Aitzetmüller 1978/1991: 95, 98, 101, 104–105; [see also
u-stems, § 3.18.3]
PBS As in the dative and instrumental plural, Lithuanian and the Slavic lan-
guages show the reflex of the i-stem ending PIE *‑isu > PBS *‑ˈisu in the con-
sonant stems. However, in a few instances the original i-less ending is found
in Slavic (see below), indicating that the introduction of the i-stem ending
was independent innovations in Baltic and Slavic. This assumption is compat-
ible with the idea that i-stem forms with an unaccented ending such as ORu.
o zvěŕ ’ax, just like the corresponding dative plural forms, reflect original root
288 Chapter 3
nouns ending in *‑sú, which regularly became unaccented by the mobility law
[4] (Olander 2007a, loc. cit.). The Latvian ending is that of the i̯o-stems.
PS In the attested Slavic languages PBS *‑su was replaced with the i-stem
ending reflecting PS *‑iˈxu (CS *‑ixъ [29]) or, as in Old Czech, the o-stem ending
PS *‑ai̯ˈxu (CS *‑ěxъ [22|29]).
In a marginal case, the type in *‑jane (see § 3.16.1 on the dative plural), there
are traces of *‑su and, with analogical *x from the o‑, i‑ and u-stem ending,
*‑xu (see e.g. Węglarz 1933 and the above-mentioned literature). For instance,
preserved *‑sъ is found in Old Czech place names like Dolas, Brěžas (see e.g.
Gebauer 1896: 19, 77–78; Čornejová 2007); the quantity of the a in Old Czech is
uncertain (Zubatý 1893: 498 n. 2; Węglarz 1933: 34 n. 1 with references). Reflexes
of *‑xъ are found in ORu. Pol’axъ, ONovg. (na) pogoščaxъ, OSrb., OCr. děčaxь. CS
*‑asъ cannot reflect *‑ansu, which would yield **‑ǫsъ, but *a was introduced by
analogy with the dative and instrumental plural endings *‑amъ, *‑ami, which
show regular reduction of *‑ānm‑ to *‑ām‑ (see Brugmann 1897: 441–442 and
§ 3.16.1; cf., on the one hand, Zubatý 1902: 227 and Hujer loc. cit., who regard
*‑amъ, *‑ami as analogical; and, on the other, Meillet loc. cit., according to
whom *‑asъ may be regular in some Slavic dialects; and different still Węglarz
1933: 40, who assumes replacement of *‑i̯ǫm‑, *‑i̯ǫs‑ by *‑’anm‑, *‑’ans‑, which
then yielded *‑’am‑, *‑’as‑).
PIE Brugmann 1909–1911: 253: PIE *‑isu, perhaps also *‑isi | Beekes 1995/
2011: 203 | Rasmussen forthc. a § 6: PBS *‑iṣu, PIE *‑isu | Debrunner
& Wackernagel 1930/1975: 72, 162 | Rix 1976/1992: 157–158: Gk. ‑εσι for
dial. ‑ισι has suffix-vowel of gen. sg.; [see also C-stems, § 3.18.1] | Sihler
1995: 311, 314–315: PIE *‑isu; Gk. ‑εσι for dial. ‑ισι has new stem-vowel *e;
Nominal Inflection 289
[see also C-stems, § 3.18.1] | Weiss 2009/2011: 242, 246: PIE *‑isu is not
preserved in La. | Krahe 1942/1967: – | Boutkan 1995: 240: PIE
*‑isu | Kloekhorst 2008a: 873: HLuv. adv. tara/i‑su-u and Milyan adv.
trisu may correspond directly to Ved. triṣú, Li. dial. adv. trisù
PBS Olander 2009: 191, 193: PBS *‑ˈisu reflects PIE *‑ísu | Endzelīns 1971:
154: Li. ‑yse probably consists of acc. pl. plus a postposition; Lv. forms in
‑îs correspond to Li. forms in ‑yse or ‑ysu | Stang 1966: 213: Li. akisù
preserves original *‑isu, also in Ved. ‑iṣu, OCS ‑ьxъ; Li. ‑yse has long vowel
from ill. ‑ysna and ā-stem ‑ose, ė-stem ‑ėse | Otrębski 1956: 42: Ved.
‑iṣu, OCS ‑ьxъ reflect *‑isu, preserved in Li. dial. forms like avisù; Li. ‑yse
has long vowel by analogy with other paradigms | Endzelīns 1923: 321:
Lv. ‑îs is either acc. pl. plus postposition or an old form in *‑u
PS Vondrák 1908/1928: 7 | Hujer 1910: 146–147: Slavic ‑ьxъ is from PIE
*‑isu | Meillet 1924/1934: 419–420: Slavic ‑ьxъ reflects PIE *‑isu |
Vaillant 1958: 137: Slavic ‑ьxъ reflects PIE *‑isu; OCS ‑exъ is from i̯o-stems;
Li. ‑yse, Lv. ‑îs have ī from loc. sg. | Bräuer 1969a: 157 | Arumaa
1985: 129 | Igartua 2005a: 271–272: Slavic ‑ьxъ reflects PIE *‑isu; Li. ‑yse
is from acc. pl. *‑īs plus postposition *en | Aitzetmüller 1978/1991: 75;
[see also u-stems, § 3.18.3]
PIE The expected ending *‑isu, consisting of the stem-suffix *‑i‑ and the loc-
ative plural marker *‑su, is attested in Vedic and Balto-Slavic. In Attic Greek
the ending has been recomposed by the new stem-vowel ‑ε‑ and the remade
locative plural marker ‑σι (see § 3.18.1). The Latin ending ‑ibus continues the
dative plural. Hittite does not preserve the ending. According to Kloekhorst
(loc. cit.), HLuv. adv. tara/i‑su-u ‘three times’ and Milyan adv. trisu ‘three times’
may be equated with Ved. triṣú and Li. dial. adv. trisù ‘by threes’. Eichner (1992:
61–62, 73–74), on the other hand, traces these and other Anatolian forms back
to *‑is‑wé, “an endingless locative [. . .] of a ‑wo-derivative”.
PS The regular result of PBS *‑ˈisu is PS *‑iˈxu [12|13] (CS *‑ьxъ [29]). The
ending is preserved in the old Slavic dialects. It is likely that in at least some
instances the e of the Old Church Slavonic variant ‑exъ has been introduced
through analogical levelling with the new instrumental singular and dative
290 Chapter 3
plural forms ‑emь, ‑emъ (cf. van Wijk 1931: 175, whose proportion “‑ьmь : ‑emь,
‑ьmъ : ‑emъ = ‑ьxъ : x” does not work as it is based on two distinct chronological
layers, namely the old i-stem forms ‑ьmь, ‑ьmъ and the new forms ‑emь, ‑emъ).
*‑usu | Vaillant 1958: 112: Slavic ‑ъxъ, only preserved in ORu., has been
replaced with ‑oxъ in other dialects with o from nom. ‑ove, gen. ‑ovъ; Lv.
‑ûs has been influenced by loc. sg. | Bräuer 1969a: 148 | Arumaa
1985: 129 | Igartua 2005a: 297–298: Slavic ‑ъxъ reflects PIE *‑usu |
Aitzetmüller 1978/1991: 72: Li. ‑se for OLi. ‑su is due to influence from loc.
sg.; Gk. ‑σι for *‑su has been remade on the model of dat.–loc. sg. in ‑ι; OCS
‑oxъ has o from nom. ‑ove, gen. ‑ovъ
PS PBS *‑ˈusu yielded PS *‑uˈxu [12|13] (CS *‑ъxъ [29]), preserved in Old
Russian and probably in Czech, though not in the earliest texts (Trávníček 1935:
322; but cf. Gebauer 1896: 325); the usual Old Czech ending ‑iech is that of the
o-stems. Old Church Slavonic ‑oxъ has o from nom. ‑ove, gen. ‑ovъ.
PIE The ending *‑ah₂su consists of the suffix *‑ah₂‑ and the locative plural
marker *‑su. The regular reflexes of the ending are preserved in Indo-Iranian
and Balto-Slavic. The Attic ending ‑αῖς is modelled on the instrumental plural
of the o-stems. The old locative plural ending was reshaped to *‑āsi in Greek
(§ 3.18.1), preserved in Old Attic with dative plural function, e.g. δικεσι /‑ɛ̄si/,
and in Attic in the adverb Ἀθήνησι ‘at Athens’. An earlier stage /‑āhi/ is likely
to be represented by Myc. ku‑na‑ke‑ta‑i (Hajnal 1995: 21). In Latin, where the
Nominal Inflection 293
original locative plural ending was lost, ‑īs < *‑ai̯s is modelled on the o-stem
instrumental plural.
PBS PIE *‑ah₂su yielded PBS *‑ˈā̰su [1]. The original ending is preserved in
Old Lithuanian and in dialects, e.g. East Li. šakā̊sù. In standard Lithuanian a
new case–number marker ‑se was created by replacing the inherited ‑u with
the locative marker ‑e found in the singular (see § 3.8.5). The Latvian locative
plural marker cannot go back to *‑su since PBS *‑u is preserved in Latvian. It is
unclear which short vowel has been lost in the Latvian form.
PS PBS *‑ˈā̰su is continued as PS *‑ˈāxu [13] (CS *‑axъ [29]; *‑jaxъ [20|27|29]),
with *x for regular *s by analogy with the o‑, i‑, u‑stems and certain consonant
stems. The ending is preserved in the old Slavic dialects.
PS *takai ̯ˈxu (CS *tok||cěx́ ъ; PBS *‑ˈai ̯su PIE *‑oi ̯su
*gojíxъ)
PIE Brugmann 1909–1911: 251–252: PIE *‑oisu, perhaps also *‑oisi; *‑oi‑ origi-
nally belongs to pronouns | Beekes 1995/2011: 213: PIE *‑oisu has
*oi from pronouns | Rasmussen forthc. a § 4: OCS ‑ěxъ reflects PBS
*‑ai̯ṣu < PIE *‑oi̯su; OLi. ‑uosu has vocalism of acc. pl.; Li. ‑uosè has ‑è from
sg. | Debrunner & Wackernagel 1930/1975: 109–110: Ved. ‑eṣu, YAv.
‑aēšu, OCS ‑ěxъ go back to PIE *‑oisu, originally a pron. ending | Rix
1976/1992: 140–141: Ved. ‑eṣu, YAv. ‑aēšu, OCS ‑ěxъ reflect PIE pron. ending
*‑oi̯su; Myc. ‑oi̯hi and Gk. (Hom., dial.) ‑οισι (with reintroduced *s) have ‑i
for *‑u; Gk. ‑οις is instr. pl. ending; [see also C-stems, § 3.18.1] | Sihler
294 Chapter 3
1995: 253, 263–264: Gk. dial. ‑οισι, Myc. ‑oihi are from PIE pron. ending
*‑oi̯su with ‑i for *‑u from sg.; Hi. ‑aš is from original noun ending *‑osu
with apocope; Gk. ‑οις is instr. pl. ending; Italic endings perhaps represent
merger of instr. *‑ōi̯s and loc. *‑oi̯su | Weiss 2009/2011: 206–208, 224;
[see instr. pl., § 3.17.5] | Krahe 1942/1967: – | Boutkan 1995: – |
Kloekhorst 2008a: 104; [see dat. pl., § 3.16.5]
PBS Olander 2009: 191–192: PBS *‑ˈai̯su is from PIE *‑ói̯su | Endzelīns 1971:
137: OCS ‑ěxъ reflects PIE *‑oisu; Li. ‑uose, Žem. ‑unsi are from acc. ending
plus a postposition | Stang 1966: 186–187: Li. ‑uose reflects OLi. ‑uosu,
with ‑e from sg.; Li. ‑uo‑, Lv. ‑uo‑ have been introduced from ill. pl. ‑uosna
by analogy with ā-stems; Žem. ‑ẹisu reflects *‑oisu | Otrębski 1956: 16:
Ved. ‑eṣu, OCS ‑ěxъ are from PIE *‑oisu; Li. ‑uose, Li. dial. ‑unse contain
acc. pl. ‑uons plus postposition *en | Endzelīns 1923: 300: Li., Lv. loc.
pl. forms reflect acc. pl. and postposition
PS Vondrák 1908/1928: 7: OCS ‑ěxъ, Ved. ‑eṣu are from PIE *‑oisu, with *‑oi‑
originating in pronouns | Hujer 1910: 146–147: Slavic ‑ěxъ reflects
PIE pron. ending *‑oisu, replacing *‑osu | Meillet 1924/1934: 410: OCS
‑ěxъ, ‑jixъ corresponds to Ved. ‑eṣu | Vaillant 1958: 36: Slavic ‑ěxъ, ‑jixъ
from PBS *‑aišu, PIE *‑oisu, with *oi from pron. inflection or from loc.
sg.; Li., Lv. forms are remade on the model of ill. pl. | Bräuer 1969a:
27–28: PIE *‑oi̯‑ for *‑o‑ is pron. stem; 74 | Arumaa 1985: 144–145: PIE
*‑oisu, originally pron. ending | Igartua 2005a: 184: CS *‑ěxъ from PIE
*‑oisu | Aitzetmüller 1978/1991: 82, 84: OCS ‑ěxъ, ‑jixъ from PIE *‑oisu
with *oi from pronouns or from loc. sg.; [see also u-stems, § 3.18.3]
PBS PIE *‑oi̯su is reflected as PBS *‑ˈai̯su [7]. In Baltic the ending is preserved
in a few relics such as Žem. adv. pẹnkẹisu ‘by fives’ from *‑íesu with secondary
acute tone. Elsewhere in East Baltic the ending has been reshaped under the
Nominal Inflection 295
influence of the illative plural in ‑uosna and with replacement of ‑u with the
‑e of the locative singular (see § 3.8.5 and Stang 1966: 186–187). In Old Prussian
the form was lost.
PS PBS *‑ˈai̯su regularly yielded PS *‑ai̯ˈxu [12|13] (CS *‑ěxъ [22|29]; *‑jixъ
[20|22|29]). The ending is preserved in the old Slavic dialects. In stems end-
̆
ing in PS *ī�(n)k/g/x, where the velar was palatalised by the preceding *ī� ̆ in the
second palatalisation [23], the soft ending CS *‑ixъ replaced the phonetically
regular reflex *‑ěxъ, e.g. OCS otьcixъ. The expected reflex is preserved in the
pronouns OCS gen.–loc. sicěxъ, vьsěxъ (Vermeer 2003a: 411–414 and 2003b: 383,
both with references and discussion).
Chapter 4
Verbal Inflection
This system was significantly simplified in Slavic (cf. Andersen 2013: 19–20;
Ackermann 2014: 253–256). The original imperative and subjunctive forms
disappeared completely, as did the middle voice. The Proto-Indo-European
optative forms became the Slavic imperative. The old resultative aspect (the
“perfect”) is represented by a single form, PS prs. 1sg. *ˈu̯ āi̯dai̯ (CS *vě̋ dě). While
the imperfective present was preserved, the perfective and imperfective pret-
erite forms (which were augmentless, thus formally injunctives) merged into
a preterite commonly referred to as the aorist; the non-present forms that
survive in Slavic are the Proto-Indo-European sigmatic perfective preterite
(“sigmatic aorist”) and the thematic imperfective preterite (“thematic imper-
fect”). At a later stage, when the inherited preterite formations had merged
functionally to a general preterite, a new imperfective preterite was created,
containing a string of suffixes followed by the endings of the thematic pret-
erite (Andersen 2013: 30–31). Several non-finite forms are formed from the
verb in Proto-Slavic, including active and passive participles, an infinitive and
a supine.
The categories of the Slavic finite verb, at a stage before the periphrastic
forms had become an integrated part of the verbal system, were tense (present
vs. preterite), mood (indicative vs. imperative), aspect (imperfective vs. perfec-
tive) and person–number. From a morphological point of view, the root shape
in some verbs indicated tense (e.g. *ber‑ designating present and *bira‑ desig-
nating non-present); the suffix indicated aspect; and the inflectional endings
indicated tense, mood and person–number.
Most of the Proto-Indo-European person–number markers had an initial
consonant. Therefore, less fusion has taken place between the marker and a
preceding suffix in the development of the Proto-Indo-European verbal sys-
tem into Proto-Slavic. From this point of view the verbal system has retained
a more conservative character in Proto-Slavic than the nominal system, where
the border between the desinential suffix and the case–number marker is pre-
served intact across the various stem-types in the dative–instrumental dual
and in the dative and locative plural only.
The relatively clear-cut border between suffix and marker in the verbal
system has some bearings for the structure of this chapter. It is often possible
to treat two or more endings together, for instance the athematic and thematic
third-person singular endings PS*‑ti (CS *‑tь) and *‑eti (CS *‑etь). In the dual
and first- and second-person plural all endings (athematic and thematic pres-
ent, preterite, imperative) may be treated together, e.g. PS athem. prs. 2pl. *‑te
(CS *‑te), them. prs. *‑ete (*‑ete), athem. prt. *‑te, them. prt. *‑ete, athem. ipv.
*‑īte, them. ipv. *‑āi̯te. This leads to a simpler and clearer overview with less
repetition of references and analyses.
298 chapter 4
̋
1 *esˈmi (*jèsmь); *ˈu̯ āi̯dai̯ (*vě dě) *esu̯ ā (*jesva, *esˈmə (*jèsmъ, →*jesmè,
→*jesvě) →*jesmò)
̆
2 *eˈsi (*jèsь), *eˈsī / *eˈsēi ̯ (*jesì) *estā (*jesta) *esˈte (*jestè)
3 *esˈti (*jèstь) *este (*jeste) *sanˈti (*sǫ́ tь), *dāˈde/inti
(*dádętь)
with the exception of the first-person singular, the thematic endings consist
of the thematic vowel followed by the same markers as in the athematic conju-
gation (e.g. Szemerényi 1996/1999: 236–238; Cowgill 1985a/2006; 2006; Kümmel
1997: 120–122). According to a group of alternative views, there was no or only
a very limited overlap between the primary athematic and thematic end-
ings in Proto-Indo-European. This view is represented by e.g. Watkins (1969),
Kortlandt (1979b/2009, 1979/2007, 1997/2007) and Beekes (1995/2011: 258–261:
“The primary endings of these two systems were totally different”; for further
references see Szemerényi 1996/1999: 237 n. 2).
According to the standard view, the Proto-Slavic thematic present directly
reflects the corresponding Proto-Indo-European category. The Proto-Indo-
European thematic endings consisted of the thematic vowel followed by the
person–number marker. The first-person singular had a peculiar form in *‑oh
for expected *‑omi.
The primary thematic endings were preserved in Slavic with minor modifi-
cations (cf. the radically different view of Watkins 1969: 218: “Außer der 2. Pl.
‑ete kann man keine der aksl. Formen mit den entsprechenden Paradigmen
Skt. vahāmi, lat. uehō, got. ‑wiga usw. gleichsetzen”). Note that the Proto-Slavic
first-person dual and plural endings are reconstructed with the original vowel
timbre *‑a‑ here, reflecting PIE *‑o‑ (1du. *u̯ edau̯ ā, 1pl. *u̯ edaˈmə), although all
attested Slavic languages show ‑e‑. The reason for this reconstruction is that
the analogical replacement of *‑a‑ with *‑e‑ most likely took place after the
fronting of non-front vowels [20] in the i̯e-present (see §§ 4.12 and 4.15), a pro-
cess belonging to the post-Proto-Slavic period.
1 *ˌu̯ edān (*vȅdǫ; pišǫ̀ ) *u̯ edau̯ ā (→*vedeva, *u̯ edaˈmə (→*vedèmъ,
→*vedevě; *píševa, →*vedemè, →*vedemò;
→*piševě) *píšemъ, →*píšeme,
→*píšemo)
2 *u̯ edeˈxi (*vedèšь; *píšešь), *u̯ edetā (*vedeta; *u̯ edeˈte (*vedetè; *píšete)
*u̯ edeˈxī / *u̯ edeˈxē̆i ̯ *pišeta)
(*vedešì; *píšeši)
3 *u̯ edeˈti (*vedètь; *píšetь) *u̯ edete (*vedete; *u̯ edanˈti (*vedǫ́tь; →*píšǫtь)
*pišete)
300 chapter 4
1 *u̯ ēˈsu (*vě ́sъ) *u̯ ēsau̯ ā (*věsova, *u̯ ēsame (→*věsomъ, *věsome,
→*věsově) →*věsomo)
2 *ˈbēi̯ / *ˈbī (*bi̋ ), *ˌmer *u̯ ēstā (*věsta) *u̯ ēste (*věste)
(*mȇr), *ˌbū (*by̑ )
3 *ˈbēi̯ / *ˈbī (*bi̋ ), *ˌmer *u̯ ēste (*věste) *u̯ ēsen (*věsę)
(*mȇr), *ˌbū (*by̑ )
While the original Proto-Slavic athematic preterite paradigm is still clearly vis-
ible in Old Church Slavonic texts, the paradigm has largely been remade in
most Slavic languages by the introduction of a suffix CS *‑ox/s/š‑ or *‑ex/s/š‑
before the ending, e.g. ORu. aor. 1sg. vedoxъ, OCz. vedech corresponding to OCS
věsъ (alongside vedoxъ). In noting which forms preserve the regular reflexes I
have not taken this suffix into account. This means that the above-mentioned
forms are considered regular since they do, in fact, preserve the original ending.
Verbal Inflection 301
1 *u̯ edāi̯mi (*veděmь; ?) *u̯ edāi̯u̯ā (*veděva, *u̯ eˈdāi̯me (→*vedě mъ,̋ *veděme,
̋
→*veděvě; *pišiva, ̋
→*vedě mo; →*piši̋mъ, *piši̋me,
→*pišivě) →*piši̋mo)
2 *u̯ eˈdəi̯ (*vedì; *pišì) *u̯ edāi̯tā (*veděta; ̋ *piši̋te)
*u̯ eˈdāi̯te (*vedě te;
*pišita)
3 *u̯ eˈdəi̯ (*vedì; *pišì) – *ˈbūndan (*bǫ̋ dǫ)
Among the non-finite forms the Proto-Slavic infinitive formally reflects a Proto-
Indo-European i-stem locative singular (§ 3.8.2), while the supine is identical
to a u-stem accusative singular (§ 3.8.3). The masculine and neuter forms of
the present participle are inflected as (o)nt-stems (§§ 3.3.6 and 3.3.7), while
Verbal Inflection 303
ocs jesmь, jesmъ Li. →esù; OLi. ésmi, essmí; Ved. ásmi
ORu. jesmь, jesmъ; refl. →důmies OAv. ahmī; YAv. ahmi
ONovg. jesmь Lv. →ȩsmu Gk. εἰµί; Lesbian ἔµµι
OCz. jsem OPr. →asmai, →asmu 2 × La. sum; OLa. ESOM;
South Picene esum
Go. im; ONor. em
Hi. ēšmi
PIE Brugmann 1913–1916: 595–603: OCS ‑mь etc. reflect PIE *‑mi | Beekes
1995/2011: 259: PIE *‑mi | Rix 1976/1992: 250: PIE *‑mi | Sihler 1995:
458–459: OCS ‑mь is from PIE *‑mi, probably also in La. sum | Weiss
2009/2011: 384–385: La. sum, OLa. ESOM reflect PIE *h₁ésmi; OLi. ‑mì,
refl. ‑mies corresponds to OPr. asmai, reflecting PBalt. *‑mai, possibly
the result of contamination with middle endings | Krahe 1942/1967:
135–136: PIE *ésmi is preserved in Go. im | Boutkan 1995: 373–374:
Go. im etc. reflect *‑mi; loss of *‑i in Runic em is due to lack of sentence
stress | Kloekhorst 2008a: 578: Hi. ‑mi partly reflects PIE primary end-
ing *‑mi, partly PIE secondary ending *‑m plus present marker *‑i
PBS Olander 2009: 136: PS*esˈmi goes back to PIE *h₁ésmi | Kortlandt
1979b/2009: 155: Slavic ‑mь reflects PIE *‑mi; Li. ‑mì has acute tone from
2sg. ending; OPr. asmai reflects *‑moʔi | Endzelīns 1971: 202: OCS ‑mь
is from PIE *‑mi; Li. ‑mi, ‑mies reflects *‑mie < *‑mei, based on analogy
with 2sg. *‑ei; OPr. ‑mai may be from contamination of endings of pf.
*vaidai and *vai(d)mi; OPr. asmu, Li. dial. esmù, Lv. dial. ȩsmu have been
influenced by them. forms | Stang 1942: 225, 1966: 406–407: Li. ‑mi,
‑miesi < *‑míe and OPr. asmai reflect PBalt. *‑mái, originally a middle
304 chapter 4
ending, perhaps remade from *‑ai; PBalt. *‑mái may also represent con-
tamination of pf. *‑ai and athem. prs. *‑mi | Otrębski 1956: 180–181:
OLi. ‑mi is not from PIE *‑mi, but from *‑mie, based on analogy with 2sg.
*‑sie | Endzelīns 1923: 545–546: Lv. ‑mu has ‑u from them. stems
PS Vondrák 1908/1928: 111: CS *‑mь is from PIE *‑mi | Meillet 1924/1934:
309–310: Slavic ‑mь is from PIE *‑mi | Stang 1942: 213: OCS ‑mь reflects
PIE *‑mi | Vaillant 1966: 8: Slavic ‑mь is from PIE *‑mi | Arumaa
1985: 268–271: Slavic ‑mь reflects PIE *‑mi | Aitzetmüller 1978/1991:
176: OCS ‑mь is from PIE *‑mi
PIE The first part of *‑mi is the first-person singular active marker *‑m‑, also
found in the secondary ending. The second part is the hic et nunc marker *‑i,
which also appears in other, but not all, present forms of the verb. The ending
*‑mi is preserved in the old Indo-European languages.
PBS The Slavic evidence shows that PIE *‑mi was preserved in Proto-Balto-
Slavic. In Baltic the ending was altered in various ways.
OLi. ‑mi is usually taken together with the reflexive form, e.g. OLi. důmies;
by internal reconstruction, this leads to pre-Li. *‑mḛ̄, shortened in final posi-
tion according to Leskien’s law, but preserved when followed by an enclitic. It
seems to me, however, that in this case the application of internal reconstruc-
tion leads in the wrong direction. It leads to a better solution if we assume that
OLi. ‑mi directly reflects PBS, PIE *‑mi and that the reflexive form arose by a
proportional analogy of the type them. non-refl. vedù : refl. vedúos(i) :: athem.
non-refl. dúomi : refl. X, where X = dúomies(i).
The assumption that Li. ‑mi directly reflects PBS *‑mi < PIE *‑mi has the
advantage of explaining why these forms usually have initial accentuation in
Daukša’s Postilė (see Specht 1922: 30–31; Hirt 1929: 304). In the copula there are
six instances of initial accentuation but only two instances of final accentua-
tion (for the material see Skardžius 1935: 192, who regards the forms with initial
accent as secondary). If ‑mi reflected acute *‑mḛ̄, we would expect the first-
person singular to have final accentuation by Saussure’s law. While the forms
with final accent may easily have arisen by analogy with the thematic verbs,
those with initial accent are more difficult to explain as secondary (see e.g.
Senn’s attempt at doing so in 1935: 95, referring to 1929/1974: 257; and Skardžius
1935: 193).
OPr. asmai is likely to be based on original *‑mi. It is not clear what ‑ai is,
but Endzelīns (1944: 156; 1971: 202) may be on the right track when he identi-
fies it with the ‑ě of OCS vědě, which reflects the Proto-Indo-European perfect
Verbal Inflection 305
ending *‑h₂a plus the particle *‑i (§ 4.3.2; see also Jasanoff 2003: 75 n. 20). It is
possible that the first-person singular is the locus of the creation of the Old
Prussian present marker ‑ai, found in all persons outside the third person (this
is implied in Endzelīns 1944: 157–160; see also van Wijk 1918: 61).
The rarely attested form OPr. asmu may represent the same innovation as Li.
dial. esmù, Lv. ȩsmu, viz. a contamination of the old athematic ending *‑mi and
the thematic ending *‑oh (Endzelīns 1944: 156–157; 1971: 202); after a labial, ‑u
from unaccented *‑ō seems to be regular in Old Prussian (Stang 1966: 49–50).
Alternatively, ‑mu may be a phonetically conditioned variant of ‑mai (Cowgill
1985a/2006: 107; 2006: 562).
PS PBS *‑mi was retained as PS*‑mi (CS *‑mь [29]), preserved in the old
Slavic dialects.
PS *ˈu̯āi ̯dai ̯ (CS *vě ̋dě) PBS →*‑ai ̯ PIE pf. *‑h₂a
PIE Brugmann 1913–1916: 491, 595–603: Ved. ‑a, Gk. ‑α, Gmc. ‑∅ reflect PIE
*‑a; La. uīdī, OCS vědě have pf. middle ending | Beekes 1995/2011:
265–266: Ved. ‑a, Gk. ‑α go back to PIE *‑h₂e; OCS vědě, La. ‑ī, probably
also Old Hi. ‑ḫe reflect *‑a plus *‑i | Rix 1976/1992: 256: Gk. ‑α, Ved. ‑a,
Luvian ‑ḫa reflect PIE *‑h₂e; La. uīdī, OCS vědě are from *‑ai | Sihler
1995: 570–571, 587: PIE *‑h₂e; La. ‑ī < *‑ai, Hi. ‑ḫi, OCS vědě contain inher-
ited *‑a plus primary tense marker *‑i | Weiss 2009/2011: 392: Gk. ‑α,
Ved. ‑a go back to PIE *‑hxe; La. uīdī, OCS vědě contain PIE *‑h₂e and hic
et nunc ptcl. *‑i, also in Hi. primary ḫi-conjugation | Krahe 1942/1967:
102: Gmc. ‑∅ reflects PIE *‑a | Boutkan 1995: 332–334: Gmc. dialects
have ‑∅ from PGmc. *‑a < PIE *‑h₂e | Kloekhorst 2008a: 341–342: Hi.
306 chapter 4
‑ḫḫi, older ‑ḫḫe reflects *‑h₂e‑i, containing PIE pf. 1sg. ending *‑h₂e, also
in Ved. ‑a, Gk. ‑α, Go. -∅
PBS Olander 2009: – | Kortlandt 1979b/2009: 154: Slavic vědě contains old
pf. form plus *‑i from athem. primary endings | Endzelīns 1971: 202:
OCS vědě is old pf. *vaidai | Stang 1942: –, 1966: 406–407 | Otrębski
1956: – | Endzelīns 1923: 697–698: OCS vědě is old middle form
PS Vondrák 1908/1928: 215: OCS vědě contains old pf. middle ending *‑ai, cor-
responding to Ved. tutude, La. tutudī | Meillet 1924/1934: 310: Slavic ‑ě,
Ved. ‑e are from PIE pf. middle ending *‑ai | Stang 1942: 214: vědě is
an old pf. form containing the middle ending *‑ai, also in Ved. pf. middle
vidé, La. uīdī | Vaillant 1966: 6, 76: Slavic vědě has ‑ě from pf. middle
*‑ai, also in Ved. ‑e, La. ‑ī | Arumaa 1985: 315–316: Slavic vědě reflects
old middle form *‑ai, comparable to La. uīdī | Aitzetmüller 1978/1991:
233: ending of OCS vědě may be related to ‑ě‑ of prt. stem
PBS By analogy with the first-, second- and third-person singular and third-
person plural of the present the element *‑i was added to PIE *‑h₂a, yielding
PBS *‑ai̯ [1]. The ending is not preserved directly in Baltic, but it is possible that
it appears in OPr. 1sg. asmai (see § 4.3.1).
̋
PS PBS *‑ai̯ was retained as PS*‑ai̯ (CS *‑ě [22|29]). PS*ˈu̯ āi̯dai̯ (CS *vědě)
is preserved as a paradigmatic form in Old Church Slavonic, Old Slovene, the
Old Novgorod dialect and Old Czech. In Old Russian the form functions as a
particle. The old form was replaced with the synchronically more regular form
CS *vě̋mь in the Slavic languages.
Verbal Inflection 307
PIE Brugmann 1913–1916: 540, 595–603: PIE *‑ō; OCS ‑ǫ goes back to *‑ōⁿ,
reflecting *‑ō plus nasal from prt. *‑om | Beekes 1995/ 2011: 260:
PIE *‑oh | Rix 1976/1992: 250: Gk. ‑ω is from PIE *‑ō or *‑oh₂; Ved.,
YAv. ‑āmi, OCS ‑ǫ, Hi. tiyami may reflect more original ending PIE
*‑omi | Sihler 1995: 458–459: PIE *‑oh₂; OCS ‑ǫ < *‑ōm has nasal from
PIE secondary ending *‑om | Weiss 2009/2011: 395: Li. acute ‑ù points
to PIE *‑oh₂, from *‑oh₂e by “Jasanoff’s law” | Krahe 1942/1967: 96: Go.
‑a etc. reflect PIE *‑ō | Boutkan 1995: 308–310: Go. ‑a etc. are from PIE
*‑oh | Kloekhorst 2008a: –
PBS Olander 2009: 195: Li. acute tone points to PIE *‑oh; the ending
became unaccented in Slavic when the secondary ending *‑m was
imported | Kortlandt 1979b/2009: 154–156: Li. ‑ù, Lv. ‑u reflect PIE
*‑oh; OPr. ‑a reflects PIE *‑oh₁; Slavic ‑ǫ is from *‑am, consisting of pf.
ending *‑a plus secondary ending *‑m | Endzelīns 1971: 201: OPr.
girdiu, Li. ‑ù, Lv. ‑u reflect PIE *‑ō | Stang 1942: 225, 1966: 406: Li. ‑u,
refl. ‑úos(i), Lv. ‑u, refl. ‑uôs, go back to *‑úo < PIE *‑ō; OPr. crixtia probably
reflects *‑ō | Otrębski 1956: 180: Li. ‑u is from *‑úo < *‑ō | Endzelīns
1923: 545: Lv. ‑u, ‑uôs is from *‑uo < PIE *‑ṓ
PS Vondrák 1908/1928: 111: Li. ‑u is from PIE *‑ō; CS *‑ǫ perhaps reflects subj.
*‑ām, also in La. agam | Meillet 1924/1934: 310–312: Slavic ‑ǫ reflects
PIE *‑ō plus secondary ending *‑m | Stang 1942: 213–214: OCS ‑ǫ < *‑ān
reflects *‑ā from PIE *‑ō, plus *‑n from *mogъn or *mogon | Vaillant
1966: 8: Slavic ‑ǫ is from *‑ōm, a reduced form of *‑ōmi, corresponding to
Ved., YAv. ‑āmi, Hi. ‑aḫmi | Arumaa 1985: 268–271: PIE *‑ō; Slavic ‑ǫ
has not been explained satisfactorily, but may be based on an interplay
between primary and secondary endings | Aitzetmüller 1978/1991:
308 chapter 4
176, 210: OCS ‑ǫ is from *‑ā (< PIE *‑ō) plus secondary ending *‑m; soft ‑( j)ǫ
for *‑( j)ę < *‑i̯ām is based on analogy with e-verbs
PBS pie *‑oh is reflected as PBS *‑ˈō̰ [1], which yielded Li. ‑ù, refl. ‑úos(i), Lv.
‑u, refl. ‑uôs, with regular acute tone reflecting the word-final laryngeal.
PS The expected outcome of PBS *‑ˈō̰ is PS*‑ˈā [13|15] (CS *‑a [29]; *‑ja
[20|27|29]), but this ending is not attested in the Slavic languages. According
to the standard explanation, the *‑m of the secondary ending was attached to
*‑ā at a pre-stage of Slavic (before [19], where *‑m was lost in the secondary
ending; see § 4.4), yielding PS*ˌ‑ān [19] (CS *‑ǫ [28]; *‑jǫ [20|27|28]).
Vaillant (loc. cit.) has suggested that Slavic ‑ǫ reflects * ‑ōm from *‑ōmi, in
parallel to ā-stem instr. sg. *‑ǫ from *‑ān < *‑āmi (see also H. H. Hock 2007;
Hill 2013: 173–175). As mentioned in § 3.7.4, I do not think it is possible to
view them. prs. 1sg. ‑ǫ and ā-stem instr. sg. ‑ǫ as the outcomes of the same
apocope of final *‑i, since the former form acquired its *‑i after the dissolu-
tion of Proto-Balto-Slavic, whereas the latter ending had lost its *‑i already in
Proto-Balto-Slavic. In order to make the chronology work, we would have to
assume that Proto-Balto-Slavic had both original *‑ō̰ and innovated *‑ō̰m (from
*‑ō̰mi) in thematic verbs, which is not attractive. Such a chronological problem
does not seem to be addressed by the proponents of the apocope hypothesis.
However, the two views on the prehistory of PS 1sg. *ˌ‑ān may perhaps be
combined. We may assume that the athematic ending *‑mi was added to the
Verbal Inflection 309
̋
ps aor. *u̯ēˈsu (CS *věsъ) pbs →*‑am pie prt., inj. *‑m̥
̋
ps *ˈsēdu (CS *sě dъ) pbs *‑am pie prt., inj. *‑om
PIE
Brugmann 1913–1916: 595–596 (1897: 391): Gk. athem. ‑α is from PIE *‑m̥ ;
Ved. them. ‑am, Gk. ‑ον reflect PIE *‑om; OCS ‑ъ is from PIE unaccented
*‑om | Beekes 1995/2011: 260, 264: Gk. athem. ‑α reflects *‑m̥ ; OCS ‑ъ
is from PIE them. *‑om | Rix 1976/1992: 242–243: Gk. ‑ον, Ved. ‑am,
OCS ‑ъ reflect PIE them. *‑om | Sihler 1995: 458 | Weiss 2009/2011:
386–388 | Krahe 1942/1967: – | Boutkan 1995: – | Kloekhorst
2008a: 609: Hi. ‑(n)un reflects PIE secondary ending (vocalic) *‑m
PBS
Olander 2009: 137–138 | Kortlandt 1979b/2009: 155: Slavic ‑ъ from
early BS *‑um < *‑om represents secondary them. ending, replacing
original athem. ending | Endzelīns 1971: – | Stang 1942: –, 1966:
– | Otrębski 1956: – | Endzelīns 1923: –
PS Vondrák 1908/1928: 119: OCS ‑ъ is from *‑om | Meillet 1924/1934: 323:
Slavic them. ‑ъ corresponds to Ved. ‑am, Gk. ‑ον; athem. PIE *‑m̥ was
replaced with them. ending in Slavic | Stang 1942: 213: OCS ‑ъ corre-
sponds to Gk. ‑ον, Ved. ‑am, ending in PIE *‑m/n | Vaillant 1966: 15–16:
Slavic ‑ъ corresponds to PIE them. *‑om and athem. *‑m, *‑m̥ | Arumaa
1985: 268: Slavic ‑ъ is from PIE them. *‑om; sigm. ending Slavic ‑sъ goes
back to *‑som, remade from *‑sm̥ | Aitzetmüller 1978/1991: 179–180:
OCS ‑ъ is from *‑om
PS The regular reflex of PBS *‑am is PS*‑u [19] (CS *‑ъ [29]), preserved in the
old Slavic dialects.
Verbal Inflection 311
ps *u̯edāi ̯mь (CS *veděmь; ?) pbs →*‑a̰i ̯mi pie opt. *‑oi ̯h₁m̥
PIE
Brugmann 1913–1916: 557–558: PIE *‑oi̯m̥; Ved. ‑eyam for *‑ayam is based
on analogy with 2sg. ‑eḥ etc.; Gk. dial. ‑οια replaces *‑οα; Go. ‑au reflects
subj. *‑ō plus ptcl. *‑u | Beekes 1995/2011: 275: Go. ‑au is from *‑ajun
< PIE *‑oih₁m | Rix 1976/1992: 262: PIE *‑oi̯h₁m̥ is preserved in Ved.
‑eyam, Gk. dial. εξελαυνοια < *‑oi̯ia̯ ; Gk. ‑οιµι has introduced primary end-
ing *‑mi | Sihler 1995: 596–598: PIE *‑oi̯m̥; Gk. dial. ‑οια shows ana-
logical retention of *i̯; Ved. ‑eyam is analogical for *‑āyam | Weiss
2009/2011: 417: Ved. ‑eyam, Gk. dial. εξελαυνοια, Go. ‑au reflect PIE *‑oih₁m̥ ,
possibly with syllabic suffix *‑ih₁‑ | Krahe 1942/1967: 108: Ved. ‑eyam
(for *‑ayam), Gk. dial. εξελαυνοια, Go. ‑au, ONor. ‑a reflect PIE *‑oi̯m̥,
alongside *‑oim > PGmc. *‑ai(n) seen in WGerm. ‑e | Boutkan 1995:
321–323: PIE *‑oih₁m is preserved in Go. ‑au, ONor. ‑a; WGmc. forms are
analogical | Kloekhorst 2008a: –
PBS
Olander 2009: – | Kortlandt 1979b/2009: – | Endzelīns 1971:
– | Stang 1942: –, 1966: – | Otrębski 1956: – | Endzelīns 1923: –
PS Vondrák 1908/1928: 120 | Meillet 1924/1934: 266, 310: use of primary
ending in Slavic and Greek is not original | Stang 1942: 239–240:
Slavic ‑ěmь with primary ending has replaced expected *‑ojь from
PIE *‑oi̯m̥ | Vaillant 1966: 16: Slavic ‑ěmь contains primary ending,
replacing *‑ǫ < *‑oin < *‑oi̯m̥ | Arumaa 1985: – | Aitzetmüller
1978/1991: 192: Slavic has replaced secondary ending with primary
ending
form εξελαυνοια; in the usual form ‑οιµι the primary ending has been
added to the stem, like in Slavic. Similarly, in Vedic expected *‑aya has
been remade to ‑eyam, which has ‑e‑ from 2sg. ‑eḥ, 3sg. ‑et etc. (e.g. Sihler
loc. cit.) and ‑m from first-person singular forms with consonantal *‑m.
Hoffmann’s (1976a: 615 n. 12) suggestion that the suffix *‑ih₁‑ was always
syllabic, contrary to the Proto-Indo-European phonotactic rules, is in my
view considerably less attractive than assuming analogical developments
in Vedic and Greek (cf. also the elaboration of Hoffmann’s suggestion in
Jasanoff 2009b).
PS PBS *‑a̰ im ̯ i yielded PS*‑āi̯mi [13] (CS *‑ěmь [22|29]), preserved in OCS
otъpaděmь and in Croatian Church Slavonic, e.g. mozěm (see Vaillant 1930:
241–243). There are no clear attestations of the corresponding athematic form
(for an attempt at identifying reflexes of CS *‑jamь < *‑i̯ēmi in Croatian texts
see Vaillant 1930: 250–251).
ps *eˈsi (CS *jèsь), →*eˈsī / →*eˈsēĭ ̯ (*jesì) pbs *‑si pie *‑si
PIE
Brugmann 1913–1916: 582, 603–610: CS *‑šь reflects PIE *‑si, with *‑š‑ from
forms in *‑išь; OCS ‑ši has ‑i from athematic variant ‑si; PIE had a special
2sg. form in *‑ei, preserved in Li. ‑ì, Gk. Dor. ipv. ἄγει, Gk. ind. ‑εις (with ana-
logical ‑ς) | Beekes 1995/2011: 259–260: PIE athem. *‑si, them. *‑eh₁i;
Slavic athem. ‑si has been contaminated with them. ending; Slavic ‑ši
contains *‑eh₁i; Gk. them. ‑εις has added ‑ς from athem. ending | Rix
1976/1992: 250–251: athem. PIE *‑si; them. Gk. ‑εις has probably arisen by
metathesis from PIE *‑esi, seen in Ved. ‑asi, YAv. ‑ahi, Go. ‑is, OIr. biri, La.
‑is, Hi. tiyaši | Sihler 1995: 459–460: PIE *‑si is preserved in Hi. ‑ši, IIr.
*‑si; OCS ‑ši, containing *‑ī or a diphthong, is enigmatic; Gk. ‑εις reflects
*‑ει from *‑esi, plus ‑ς from secondary endings | Weiss 2009/2011:
384–385, 395: PIE athem. *‑si; OLi. esì probably reflects PBalt. *‑sei, which
seems to agree with OCS jesi; La. ‑is is from PIE them. *‑esi | Krahe
1942/1967: 96, 135–136: WGmc. forms point to PGmc. *‑is(i), while ONor.
berr points to *‑iz(i), both from PIE them. *‑esi; Go. is reflects PIE athem.
*esi | Boutkan 1995: 308, 310–313: PIE them. *‑eh₁i and athem. *‑si
merged to them. *‑esi in Gmc.; ohg ‑s, OSax. ‑is, OEng. ‑(e)s reflect *‑ési;
ONor. nemr reflects *´‑ezi | Kloekhorst 2008a: 751–752: Hi. ‑ši is from
PIE primary athem. *‑si and from athem. secondary *‑s plus present
marker *‑i
PBS
Olander 2009: 196: final accent in Slavic, e.g. ORu. živeší, is the regular out-
come of PIE them. *‑ési | Kortlandt 1979b/2009: 156–157: Li. ‑ì reflects
PIE them. *‑eh₁i, also found in Gk. and Celtic; when copula *esi was
314 chapter 4
replaced with *eseʔi in BS, the ending spread to other verb classes; Slavic
originally had four endings, ‑si in the copula, ‑sь in other athem. verbs,
‑ši in i-verbs, ‑šь in them. verbs.; the Slavic dialects generalised different
endings | Endzelīns 1971: 202–203: Baltic and Slavic athem. *‑sei may
be due to influence of them. *‑ei | Stang 1942: 225–230, 1966: 407–409:
Li. ‑i, refl. ‑íes(i), Lv. ‑i, refl. ‑iês from *‑íe may reflect *‑éi or *‑ái; OPr. ‑sei
may be a contamination of athem. *‑si and them. *‑ē̆i | Otrębski 1956:
181–182: Li. ‑i is from acute *‑ie < *‑ei; Li. ‑si is from *‑sie, also in OPr. waisei,
OCS věsi | Endzelīns 1923: 546–549: Lv. esi, Li. esì reflect *esíe, also in
OPr. assei, OCS jesi; Li. ‑i, Lv. ‑i may go back to PIE *‑ē̆i
PS
Vondrák 1908/1928: 112–113: CS *‑si reflects middle ending *‑sai, also in
OPr. assai ‘you (sg.) are’; Slavic (outside OCS) ‑šь reflects *‑si, with ‑š‑ <
*‑x‑ from verbs like *xvalisь, *vidisь; OCS ‑ši reflects *‑šь influenced by
*‑si | Meillet 1924/1934: 316–318: OCS postvocalic variant ‑š‑ for *‑s‑
has been introduced from verbs of the type *prosišь; ending of bereši,
prosiši may have arisen by contamination of them. *beri (with *‑i from
*‑ē̆i) and athem. *prosišь; it is possible that reflexes of *‑šь found in Slavic
dialects outside OCS represent a shortening of *‑ši | Stang 1942: 214–
215: OCS bereši has probably arisen by contamination of them. *beri (from
PIE *‑ē̆i, also in East Baltic *‑íe) and *prosišь, *jesь (from *‑si); Slavic forms
in ‑š have been shortened from ‑ši | Vaillant 1966: 8–10: Slavic ‑si, ‑ši
reflect *‑sēi, consisting of PIE athem. *‑si plus them. *‑ēi, still preserved
in East Baltic; Slavic ‑š‑ has been generalised from i-presents; apparent
Slavic reflexes of *‑šь are the results of a reduction of *‑ši | Arumaa
1985: 271–273: Baltic and Slavic point to (acute) *‑sei | Aitzetmüller
1978/1991: 176: PIE *‑si; Baltic and Slavic forms reflect *‑sei with *‑ei of
unclear origin, probably not reflecting alleged PIE them. *‑ei; Slavic ‑š‑
has been generalised from i-presents
PBS PIE athem. *‑si, them. *‑esi yielded PBS *‑si, *‑ˈesi. The athematic end-
ing is perhaps preserved in OLi. éssi, with initial accentuation indicating a
non-acute short final syllable (see below). The thematic ending is most likely
reflected in PS*‑exi (CS *‑ešь), with secondary *x (see below). Both Baltic and
Slavic show several remade variants of the inherited endings. The most difficult
problems in Baltic and Slavic are, as I see it, the final diphthong found in the end-
ing in Old Prussian and perhaps in Slavic, and the ending without *s in East Baltic.
Old Prussian shows the thematic ending in the forms giwassi 2 ×, gīwasi 1 ×,
where the inherited thematic vowel *‑e‑ has been replaced with ‑a‑, as in the
third-person form (§ 4.9). The final vowel ‑i probably represents a diphthong
(Stang 1966: 408), also found in OPr. athem. assai, assei. The Old Prussian end-
ing is usually identified with OCS athem. ‑si, them. ‑eši (see below), but in my
view it is more likely that the diphthong arose in the prehistory of Old Prussian,
perhaps originating in the first-person singular (see § 4.3.1).
As for the lacking *‑s in the East Baltic thematic ending, I noted above
that, difficult as the Baltic ending may be, I do not think it justifies the
reconstruction of an additional ending in Proto-Indo-European. The idea
has been put forward that the starting point was the second-person singu-
lar of the verb ‘to be’, *ˈesi, which was reinterpreted as *ˈes‑i, leading to the
replacement of the thematic ending *‑e‑si with *‑e‑i (Rasmussen 1987a/1999:
118–122; Kümmel 1997: 121 n. 13; Petit 2010: 243); the suggestion also works if
one assumes that *‑a‑si, with analogical o-grade in Baltic as seen in OPr. ‑asi,
316 chapter 4
PS On the basis of the attested Slavic material we may reconstruct PSathem.
*‑si (CS *‑sь [29]), and *‑sī (*‑si [29]) or *‑sē̆i ̯ (*‑si [22|29]); them. *‑eˈxi (CS
*‑ešь [21|29]; *‑ješь [21|29]), and *‑eˈxī (*‑eši [21|29]; *‑ješi [21|29]) or *‑eˈxē̆i ̯
(*‑eši [21|22|29]; *‑ješi [21|22|29]). The ending presents several difficulties.
Regarding the quantity of the final vowel, it is unclear if the apparent
reflexes of CS *‑ь are old or if they represent late shortenings of CS *‑i. The
long vowel is attested in early South and East Slavic texts, including the Old
Novgorod dialect (Stang 1969b: 133–134; Zaliznjak 1995/2004: 136, 138); it is also
found in modern Slavic dialects, mainly in the copula. Elsewhere reflexes of CS
*‑ь are found. Some authors assume that all instances of ‑ь have been short-
ened from ‑i (thus e.g. Meillet loc. cit.; Vaillant loc. cit.; Stieber 1969–1973/1989:
203–204; Mareš 1978: 204–205); Meillet adduces the apparently parallel loss of
unaccented ‑i attested in Ru. inf. být’, ipv. bud’ vs. nestí, nesí. I do not consider
this scenario more attractive than assuming that at least some instances of
Verbal Inflection 317
final ‑ь reflect PS*‑i from PBS, PIE *‑i, a view that is also quite widespread in
the literature.
While the short vowel PS*‑i (CS *‑ь) historically presents no problems, the
long vowel or diphthong of the variant PS*‑ī or *‑ē̆i ̯ (CS *‑i) is enigmatic. CS
*‑i is usually connected with the final diphthong of OPr. ‑sai, ‑sei. However, as
seen above, Old Prussian also shows a diphthong in the first-person singular
and in the first- and second-person plural. The Old Prussian diphthong has,
in my opinion, most likely arisen in the first-person singular (§ 4.3.1), whence
it may have spread to the other persons and numbers. There is therefore no
particular reason to look for a historical connection between the diphthong in
the Old Prussian second-person singular ending and the possible diphthong of
the corresponding Slavic ending.
Before we dismiss the connection entirely, however, we ought to have a look
at the possibilities for an inherited diphthongal ending in the second-person
singular of the thematic present in Balto-Slavic. The first possibility is that
Proto-Indo-European had a primary thematic second-person singular ending
*‑ē̆i ̯ or *‑eh₁i. As noted above, evidence for a Proto-Indo-European primary the-
matic second-person singular ending *‑ē̆i ̯ or *‑eh₁i hardly exists outside Baltic.
Even if Proto-Balto-Slavic did have a thematic ending *‑ē̆i,̯ whether inherited
from Proto-Indo-European or created analogically in pre-Proto-Balto-Slavic,
the contamination of this ending with the inherited athematic ending *‑si to
*‑sē̆i ̯ does not seem particularly straightforward.
Even less attractive is the proposal that the Baltic and Slavic second-per-
son singular ending reflects a middle ending *‑sai̯ (e.g. Milewski 1932: 21) or
*‑soi̯ (e.g. Cowgill 1985a/2006: 107; 2006: 554–555). The idea is unacceptable for
at least two reasons (see also Meillet 1908: 412; Kuznecov 1961: 92): first, the
middle endings have been lost everywhere in Balto-Slavic, and—in contrast
to the retention of a perfect ending in the first-person singular, which makes
good sense from a functional point of view (§ 4.3.2)—there is no reason why
a middle form should be preserved in the present second-person singular and
nowhere else. And second, Proto-Indo-European final *‑a/oi̯ in all probability
yielded PS*‑ai̯ [7] (CS *‑ě [22|29]) > OCS etc. ‑ě, not **‑i.
As there is no comparative evidence from outside Slavic for the final vowel
of the person–number marker CS *‑si, we can do no better than to reconstruct
a Proto-Slavic athematic ending *‑sī or *‑sē̆i,̯ and thematic *‑exī or *‑exē̆i ̯ along-
side inherited *‑si, *‑exi. I have not been able to find a plausible source of *‑ī or
*‑ē̆i,̯ nor a motivation for the partial substitution in Slavic of inherited *‑i with
a long vowel or diphthong.
Another irregularity appears in the thematic ending, where the ruki out-
come of *x cannot be regular. If it is correct that the contraction of *‑ei̯e‑ to *‑ī‑
[24|25] in Slavic ei̯e-verbs was a significantly later process than the ruki change
318 chapter 4
[12], it follows that the *x of Slavic is regular only in originally athematic verbs
whose root ended in *ī� ̆ i̯ ū̆ u̯ r k (similarly Kortlandt 1979b/2009: 157). When
these verbs were thematicised, *x was retained and spread to the other the-
matic verbs; only the few remaining athematic verbs retained old *s.
ps *ˈbēi ̯ / *ˈbī (CS *bi ̋ ), *ˌmer (*mȇr), pbs *ˌ‑s pie prt., inj. *‑s
*ˌbū (*by̑)
OCS bi, umrě, →umrětъ, by, →bystъ Li. – Ved. aor. inj. gāḥ
ORu. bi, ja, →jatъ, by, →bystь; ONovg. by Lv. – OAv. prs. inj. mraoš; YAv. ?
OCz. kry, mřě, by OPr. – Gk. aor. ἔβης
La. –
Go. –
Hi. →ēšta
PIE Brugmann 1913–1916: 408, 425–426, 609–610: OCS da, ję go back to *da‑s,
*jęs‑s; OCS 2sg. forms in ‑(s)tъ are from 3sg. | Beekes 1995/2011: 264:
OCS děla contains *‑ās | Rix 1976/1992: 243: Gk. ‑ς, Ved. ‑ḥ, Hi. ‑š are
from PIE *‑s | Sihler 1995: 459: Gk. ‑ς, Ved. ‑ḥ reflect PIE *‑s | Weiss
2009/2011: 386–387: La. ‑s, Gk. ‑ς are from PIE *‑s | Krahe 1942/1967:
– | Boutkan 1995: 361–362: Gmc. forms reflecting *dēs correspond to
Ved. root aor. ádhāḥ, with *‑si from prs. | Kloekhorst 2008a: 687: Hi.
‑š, found after vowels, reflects PIE *‑s; Hi. ‑tta, found after consonants, is
from ḫi-conjugation
PBS Olander 2009: – | Kortlandt 1979b/2009: 159: Slavic ‑tъ is originally
an enclitic ptcl. | Endzelīns 1971: 202–203: Li. ‑aĩ, Lv. ‑i replace earlier
*‑ās | Stang 1942: –, 1966: – | Otrębski 1956: 184: Li. vilkaĩ, Lv. vìlki
have been remade from *‑ās | Endzelīns 1923: –
PS Vondrák 1908/1928: 126: OCS 2–3sg. ‑tъ is due to influence from prs.
3sg. | Meillet 1924/ 1934: 322–323 | Stang 1942: 64–73, 219–223:
OCS 2–3sg. pitъ, bystъ are old root. aor. or ipfv. forms, perhaps contain-
ing pf. 2sg. ending PIE *‑to or *‑tos, also in Ved. véttha, Gk. οἶσθα, Go.
waist | Vaillant 1966: 54–56: 2sg. forms are from 3sg. | Arumaa
1985: 271–273, 276: Slavic da, bě reflect root or sigmatic aorists; forms like
Verbal Inflection 319
PS PBS *‑s yielded PS*‑∅ [17]. The form with a zero ending is preserved in
the old Slavic dialects. The loss of word-final *‑s entailed the merger of the
preterite second-person singular forms with the third-person singular forms,
which had lost their *‑t in pre-Proto-Balto-Slavic [9]. From this point onwards
the second- and third-person singular forms of the preterite have a common
development, most clearly seen in the addition of a new ending ‑tъ, originat-
ing in the present third-person singular. The second- and third-person singular
forms are treated together in § 4.10.1.
ps *ˈsēde, *ˌu̯ede (CS *sě̋de, *vȅde) pbs *ˌ‑es pie prt., inj. *‑es
PIE
Brugmann 1913–1916: 610: OCS ‑e is from *‑es | Beekes 1995/2011:
260: PIE *‑es | Rix 1976/1992: 243: Gk. ‑ες, Ved. ‑aḥ, OCS ‑e, Hi. daškeš
reflects PIE *‑es | Sihler 1995: 456, 459: Gk. ‑ες, Ved. ‑aḥ reflect PIE
*‑es | Weiss 2009/2011: 387: Ved. ‑aḥ is from PIE *‑es | Krahe
1942/1967: 102–103: OEng. bǣre, OSax. bāri, ohg bāri may reflect PIE
*‑es | Boutkan 1995: 334–335: WGmc. *‑i is from PIE them. aor. end-
ing *‑es or represents Gmc. pluperfect ending | Kloekhorst 2008a:
687–688
320 chapter 4
PBS Olander 2009: 197: unaccented PBS *ˌ‑es reflects PIE *‑és | Kortlandt
1979b/2009: 156: Slavic ‑e is from PIE *‑es | Endzelīns 1971:
202–203 | Stang 1942: –, 1966: – | Otrębski 1956: – | Endzelīns
1923: –
PS Vondrák 1908/1928: 119: Slavic ‑e is from *‑es | Meillet 1924/1934:
322 | Stang 1942: 64: OCS ‑e reflects *‑es | Vaillant 1966: 16: Slavic
‑e is from *‑es | Arumaa 1985: – | Aitzetmüller 1978/1991: 179: OCS
‑e is from *‑es
PBS PIE *‑es is reflected as PBS *ˌ‑es [4]. The ending was lost in Baltic.
PS PBS *ˌ‑es yielded PS*ˌ‑e [17] (CS *‑e [29]), retained in the old Slavic dia-
lects. That the form was unaccented in Proto-Slavic cannot be seen in the par-
adigm of the thematic aorist, which only comprised immobile verbs (accent
paradigms a and b; see Dybo 1961: 37; Ackermann 2014: 23–24). In the para-
digm of the sigmatic aorist, however, the thematic second-person singular
form *ˌu̯ ede (CS *vȅde) makes this accentuation clear. This also applies to the
third-person singular form.
ps →*dādi ̯i (CS *dad’ь) pbs *‑i ̯ḛ̄s pie opt. *‑i ̯eh₁s
PIE
Brugmann 1913–1916: 545–552: OCS ‑ždь somehow reflects PIE ipv. *‑dʰi
and opt. marker *‑i̯ē‑, perhaps originally in the verb daždь | Beekes
1995/2011: 275: PIE *‑i̯éh₁s | Rix 1976/ 1992: 261 | Sihler 1995:
552–553, 596: OLa. siēs, Gk. εἴης reflect PIE *‑i̯éh₁s; in later La., sī‑ was
generalised | Weiss 2009/2011: 416–417: OLa. siēs, Gk. εἴης reflect PIE
*‑i̯éh₁s | Krahe 1942/1967: 107, 110, 137, 141: Germ. has generalised *‑ī‑
from pl. | Boutkan 1995: 337–338, 463: Go. ‑eis reflects *‑ih₁s with
generalised zero grade | Kloekhorst 2008a: –
PBS
Olander 2009: – | Kortlandt 1979b/2009: – | Endzelīns 1971:
– | Stang 1942: 247, 1966: 439: Li. dial. duõ is an old formation, perhaps
reflecting a PIE aor. ipv. | Otrębski 1956: – | Endzelīns 1923: 686: it
is very questionable if Lv. duod, Li. dial. dúodi reflect PIE *dōdʰi
PS Vondrák 1908/1928: 120: OCS daždь has replaced *dadi̯ā | Meillet 1924/
1934: 331: it is unclear how PIE *ēdi̯ēs has resulted in CS *ědjь | Stang
1942: 241–242: OCS daždь for expected *‑i from PIE opt. *‑i̯ēs, *‑i̯ēt is
perhaps due to contamination of *‑’i and ‑ьjь, reflecting alternative opt.
endings PIE *‑i̯ēs and *‑ei̯əs; or it may be a contamination of *‑ždi and
*‑zdь from opt. *‑di̯ēs and ipv. *‑d‑dʰi | Vaillant 1966: 35–36: OCS ‑ždь
goes back to ‑ždi, with *‑ji (for *‑jě, *‑ja) from pl. forms | Arumaa 1985:
309–311: Slavic forms are unclear | Aitzetmüller 1978/1991: 193
PBS pie *‑i̯eh₁s yielded PBS *‑i̯ḛ̄s [1]. The ending is not preserved in Baltic.
PS The reflex of postconsonantal *i̯ in the attested Slavic forms of the athe-
matic imperative second- and third-person singular forms indicates that the
inherited alternation between PIE *‑i̯eh₁‑ in the singular and *‑ih₁‑ in the dual
and plural was preserved until a relatively late stage. However, the expected
reflex of PBS *‑i̯ḛ̄s, PS*‑i̯ē [13|17] (CS *‑ja [27|29]) is not found. Brugmann’s idea
(loc. cit.) that OCS daždь has arisen by contamination of PIE opt. *‑i̯eh₁‑s and
ipv. *‑dʰi is perhaps possible, but it requires that we assume that the imperative
existed alongside the optative until a late point in the prehistory of Slavic; how-
ever, we do not find any other indications of retention of the original impera-
tive formation in Slavic.
322 chapter 4
I find Vaillant’s solution (loc. cit.) more attractive. The inherited second- and
third-person singular form *dādi̯ē (or, depending on the relative chronology,
2sg. *dādi̯ēs, 3sg. *dādi̯ē) was first remade to *dādi̯ī (or 2sg. *dādi̯īs, 3sg. *dādi̯ī)
by analogy with the dual and plural marker *‑ī‑. Subsequently, and certainly
after the loss of word-final *‑s, the final vowel underwent shortening to *‑i (cf.,
somewhat differently, Vaillant 1930: 254–256). The irregular shortening, strictly
speaking, of a final vowel in the imperative form in pre-Proto-Slavic has a par-
allel in dialectal Lithuanian imperative forms like ím, as compared with refl.
iḿiẽs pointing to original *‑iẽ. The shortening in pre-Proto-Slavic did not have
any consequences for the morphological system. At the stage when *‑i̯ī was
shortened to *‑i̯i in the athematic imperative form, the imperative second- and
third-person singular forms of thematic presents still ended in a diphthong
*‑əi̯ (or the older form *‑ai̯ in the third-person singular), which could not be
shortened.
In this scenario all Slavic imperative forms may be derived from old optative
forms. It is not because of analogy that the second- and third-person singular
forms of the athematic imperative are identical in Slavic; these forms merged
phonologically. The drawback, of course, is that this requires the assumption
of an irregular shortening of final *‑ī to PS *‑i.
ps *u̯eˈdəi̯ (CS *vedì; *pišì) pbs *‑ˈai ̯s pie opt. *‑oi ̯h₁s
PIE
Brugmann 1913–1916: 557–561: Gk. φέροις, Ved. bháreḥ, Slavic beri
reflect PIE *‑ois | Beekes 1995/2011: 275: PIE *‑oih₁s | Rix 1976/
1992: 261 | Sihler 1995: 596–597: PIE *‑oi̯s | Weiss 2009/2011:
– | Krahe 1942/1967: 108: Go. ‑ais reflects PGmc. *‑aiz (> ONor. ‑ir, OEng.
‑e) or *‑ais (> OSax. ‑es, ohg ‑ēs), both from PIE *‑ois | Boutkan 1995:
321, 324: Go. ‑ais, ONor. ‑ir reflect PGmc. *‑ais < PIE *‑oih₁s | Kloekhorst
2008a: –
Verbal Inflection 323
PBS Olander 2009: 198: PS*‑ˈai̯ reflects PBS *‑ˈais, the regular outcome of
PIE *‑ói̯h₁s | Kortlandt 1979b/2009: – | Endzelīns 1971: 203:
OPr. immais, weddeis preserve old opt. form | Stang 1942: 247, 1966:
423–425, 434–440: OPr. wedais, Gk. ‑οις, Go. ‑ais, Ved. ‑eḥ reflect PIE opt.
*‑ois, also partly preserved in Li. dial. ím, refl. iḿiẽs, Lv. lìec | Otrębski
1956: 227: OLi., Li. dial. ipv. ‑i, OLi. refl. kęlieś may represent prs. 2sg.
form | Endzelīns 1923: 686: Lv. ved may represent PBalt. ipv. *vede
PS Vondrák 1908/1928: 119–120: Slavic ‑i reflects *‑ois, with regular reflex of
*oi in final syllable closed by consonant | Meillet 1924/1934: 329–
330: Slavic ‑i is from *oi in final syllable | Stang 1942: 239: OCS ‑i is
from opt. *‑ois | Vaillant 1966: 35–36: Slavic 2–3sg. ‑i for *‑ě is from
athem. verbs | Arumaa 1985: 309–311: Slavic ‑i probably reflects PIE
*‑ois, with secondary acute tone; or it may be due to influence from i̯e-
verbs | Aitzetmüller 1978/1991: 192: OCS ‑i (for *‑ě) from *‑ois is due to
levelling with the types znaji and nosi
PIE pie *‑oi̯h₁s was made up by the thematic vowel *‑o‑, the optative suffix
*‑ih₁‑ and the secondary ending *‑s. While the laryngeal of the thematic opta-
tive suffix does not surface in any Indo-European language, its former presence
is made apparent by the acute tone of the suffix in the plural in Slavic, e.g. Čak.
(Novī) pecȉte (see Hollifield 1980: 27; Rasmussen 1989b: 223–225; cf. the differ-
ent approach of Jasanoff 2009b).
PBS pie *‑oi̯h₁s yielded PBS *‑ˈai̯s [1|7], preserved in OPr. wedais and in
Slavic. For the accentuation of the form see the third-person singular ending
(§ 4.11.2). It is questionable if the Latvian imperative and the Lithuanian k-less
imperative continue the old optative form (thus e.g. Stang loc. cit.); if they
do, the process by which final *‑s disappeared is unclear, as it is generally in
the East Baltic second-person singular forms. The Latvian imperative forms
are sometimes thought to reflect the original Proto-Indo-European thematic
imperative in *‑e (e.g. Forssman 2001: 212), but I agree with Stang (1966: 435–
436) that this is not likely.
PS pbs *‑ˈai̯s yielded PS*‑ˈəi̯ [12|17] (CS *‑i [22|29]; *‑ji [20|22|29]), preserved
in the old Slavic dialects. The ending triggers the second palatalisation [23]
of a stem-final velar outside the Old Novgorod dialect, e.g. PS*peˈkəi̯ (CS
*pek||cì [22|23|29]) > OCS pьci, showing that the ending did indeed contain a
diphthong.
324 chapter 4
ocs jestъ, jestь (rare), je Li. 3ps. ẽsti; OLi. refl. Ved. ásti
ORu. jestь, je; ONovg. jestь, je důstis OAv. astī; YAv. asti
OCz. jest, je; OPo. (rare) jeść Lv. 3ps. iêt Gk. ἐστί
OPr. 3ps. ast, 3sg. La. est
→astits Go. ist
Hi. →ēšzi, ēšza (OS) (rare)
ocs vedetъ, pridetь (rare), bǫde Li. 3ps. →vẽda; refl. Ved. bhárati
(rare); pišetъ, bьjetь (rare), igraje →vẽdas(i) OAv. baraⁱtī; YAv. baraⁱti
(rare) Lv. 3ps. →vȩ̀lk; refl. Gk. →φέρει; Myc. →e‑ke
ORu. vedetь; pišetь; ONovg. →ceļ̂as /ekʰei/
poidetь, poide; šletь, prisъle OPr. 3ps. →senrīnka La. legit
OCz. vede; píše Go. bairiþ
Hi. →daškizzi
PIE
Brugmann 1913–1916: 582, 610–616: PIE athem. *‑ti, them. *‑eti are pre-
served in ORu. ‑tь, ‑etь; in Baltic, ‑a reflects old inj. sg. form, with vocalism
from 1pl. Li. ‑ame; Gk. ἄγει is created on the analogy of 2sg. ἄγεις | Beekes
1995/2011: 259–260: PIE athem. *‑ti; PIE them. *‑e was preserved in ORu. ‑e
and enlarged by a ptcl. ‑tъ in ocs; Gk. them. ‑ει contains *‑e plus primary
*‑i | Rix 1976/1992: 251: Gk. athem. ‑σι, ‑τι are from PIE *‑ti; Gk. them.
‑ει probably reflects *‑ei̯t by metathesis from PIE *‑eti | Sihler 1995:
461–463: OCS ‑tъ is unexpected for ‑tь from PIE *‑ti; OCS ‑e, Baltic forms,
Gk. ‑ει point to PIE them. t-less ending, alongside *‑eti | Weiss 2009/
2011: 385: athem. ORu. jestь reflects PIE *‑ti; OCS ‑tъ and Slavic endingless
forms are unclear, but may be due to an early loss of ‑ь followed either
Verbal Inflection 325
PS PBS athem. *‑ti, them. *‑ˈeti were retained as PS*‑ti (CS *‑tь [29]),
*‑eˈti [13] (CS *‑etь [29]; *‑jetь [29]). The original forms are preserved in
East Slavic, including Old Russian, and in the rare Old Polish variant jeść,
alongside jest. They are sporadically found in the Old Church Slavonic texts,
especially in athematic verbs (see Diels 1932–1934/1963: 227; cf., however, Vaillant
1948/1964: 227).
However, we also find two unexpected sets of endings: ‑tъ, ‑etъ in Old
Church Slavonic and ‑∅, ‑e in all of West Slavic as well as in South and East
Slavic dialects. For the distribution of the endings see Miller 1988: 8–16. The
situation in the present third-person singular is paralleled by the one observ-
able in the third-person plural, where we find reflexes of PS*‑inti, *‑anti in East
Slavic, but ‑ętъ, ‑ǫtъ in Old Church Slavonic and *‑ę, *‑ǫ in various dialects.
The following discussion of the third-person singular forms is also relevant
for the third-person plural forms.
As for the Slavic thematic short ending ‑e, the idea that it is simply an archa-
ism preserving the original Proto-Indo-European ending, represented by e.g.
Watkins, Kortlandt and Beekes, has been criticised above.
An alternative view points to the interaction between primary and second-
ary endings observable in other verbal endings, tracing the short ending back
to the Proto-Indo-European secondary ending (e.g. Fortunatov 1908: 29–32;
Kul’bakin 1929: 307; 1961: 102–103; apparently also Belić 1932 [see Vaillant 1932a:
243] in a publication that was not accessible to me). A related view maintains
that in certain North-West Russian dialects there is a semantic difference in
the third-person singular and plural between a “current present” (“актуальное
настоящее”) in ‑t and a “not current present” (“неактуальное настоящее”) in
‑∅ (Ryko 2000). A similar phenomenon has been observed in Old Novgorod
texts, where short forms are found mainly in conditional and final clauses
(Zaliznjak 1995/2004: 137). Since the function of the “not current” forms resem-
ble that of the Indo-Iranian injunctive, it is suggested that there is a relation-
ship between the Slavic short forms and a Proto-Indo-European injunctive
form in *‑∅ or *‑t (Ryko 2000: 129–132; Zaliznjak 1995/2004: 137 refers to this
suggestion as an attractive idea [“привлекательная идея”]). To me, however, it
328 chapter 4
seems more likely that the distribution of forms in ‑t and ‑∅ in Russian dialects
is secondary.
The zero ending has also been explained as the direct reflex of PIE *‑ti,
through an apocope of *‑i and then, since final stops were not allowed at this
stage, loss of *‑t (e.g. Meillet loc. cit.; Stang 1942: 218; 1969b: 136–139; loc. cit.;
Cowgill 2006: 550–551). I find this hypothesis most attractive. In general, older
Slavic texts present forms with ‑t, whereas the short forms are found at younger
stages. While the language of the Old Church Slavonic texts usually has forms in
‑tъ, the dialects most closely related to Old Church Slavonic, namely Bulgarian
and Macedonian, usually show the short forms. Likewise, while forms in ‑tь
are dominant in Old Russian, the short forms are found in modern Russian
dialects. Even in dialects where no direct traces of PS*t are found in the third-
person forms, the former presence of an additional syllable is indicated by the
quantity of the final vowel in forms like Štk. prs. 3sg. nèsē, prs. 3pl. nèsū, Po.
wiodą, OCz. vedou (Meillet 1913a: 235; Kümmel 1997: 121 n. 14; Cowgill 2006: 551;
see also Kortlandt loc. cit., on Čak. [Vrgada] ‑ẽ, Slk. ‑ie pointing to a lost jer; but
cf. Vondrák 1908/1928: 113 and Endzelīns 1931/1979: 582–583, according to whom
the long vowel in the singular is analogical from the plural). As pointed out by
Stang (1969b: 138) the loss of *‑ti was facilitated by the fact that the forms were
undoubtedly very frequent and often relatively long, and that even the short-
ened third-person endings were easily recognisable by the speakers.
The hypothesis of a sporadic development of PS*‑ti to *‑t to ‑∅ in the third-
person singular and plural forms of the present in Slavic is not unproblem-
atic. The details of the development and the scenario in which it took place
are most conveniently treated together with the Old Church Slavonic reflex of
PS*‑ti, viz. ‑tъ, in these forms.
The endings found in Old Church Slavonic are athem. ‑tъ, them. ‑etъ, along-
side less common instances of ‑tь, ‑etь and ‑∅, ‑e. There is no agreement on
the origin of the element ‑tъ, usually thought to be identical with the ‑tъ found
in certain monosyllabic aorist second- and third-person singular forms in
Old Church Slavonic (see § 4.10.1). In the following I shall give an overview of
the most important approaches to the problem (see also Stang 1942: 216–217;
Kuznecov 1961: 94–96; Miller 1988: 16–19).
Fortunatov (1908) identifies OCS ‑tъ with OPr. ‑ts, e.g. in astits, tracing it
back to a Proto-Balto-Slavic demonstrative pronoun *tas which was added as
an enclitic to the secondary ending in Slavic (see also van Wijk 1916: 116, who
thinks that “Fortunatov in der guten Richtung gesucht hat”). However, it is dif-
ficult to understand why, in Fortunatov’s scenario, secondary endings should
be used in the present tense. Besides, the use of ‑tъ not only in the singular but
also in the plural in Slavic requires additional hypotheses which make the idea
less attractive; note that in Old Prussian, ‑ts is only used with singular subjects.
Verbal Inflection 329
There is also no good evidence for the enclitic use of pronouns at the relevant
pre-stage of Slavic. Further criticism of the hypothesis has been adduced e.g.
by Meillet (1913a: 234–235), van Wijk (1918: 112–114), Stang (1942: 216: “unhalt-
bar”, 219; 1969b: 135), Obnorskij (1953: 117–118) and Cowgill (1985a/2006: 106;
2006: 551–552).
According to a related view, OCS ‑tъ in the present third-person singular
and plural and the aorist second- and third-person singular is an enclitic par-
ticle *tu ‘then’, apparently also seen in OPr. tīt ‘thus’ and stwi ‘here’ (Kortlandt
1979b/2009: 159; 1998/2009: 285; in 1979b: 62 Kortlandt followed Fortunatov, but
see Kortlandt 1985: 113 n. 1 [removed in 1985/2009: 52]). In OPr. ‑ts the particle
had been reanalysed as a pronoun and received an additional *‑s. As long as
the existence of such a particle is not confirmed by more transparent evidence,
and its function and the motivation for its introduction in third-person forms
only have not been established, this hypothesis is difficult to accept.
Phonetically, OCS ‑tъ may reflect PIE ipv. 3sg. *‑tu, e.g. Ved. ástu (Forssman
1981; cf. Hackstein 2007: 147 n. 40, who suggests that the Old Church Slavonic
reflex ‑tъ from PIE *‑ti may be partly due to influence from the imperative
ending). However, it is hard to believe that an imperative form would be used
in the present indicative (see also Stang 1942: 216).
Finally, several scholars have advanced hypotheses wherein OCS ‑tъ directly
reflects PIE *‑ti. Some scholars are not concerned with the details of this
development, e.g. Berneker (1904: 370–371) and Pedersen (1905a: 322). Others
are more explicit, e.g. Aitzetmüller (1978/1991: 177), who maintains that *‑t’ь
was depalatalised to ‑tъ, a development also suggested by others, e.g. Vondrák
(1908: 135). This development is difficult to understand if ь was still a real
(front) vowel at the relevant time; and it certainly was, since the change must
have been carried out before the earliest attested Old Church Slavonic texts.
Lekov (1934: 85–86) has suggested that the substitution of the palatalised *t in
the primary ending *‑tь with the unpalatalised *t in the secondary ending *‑t
entailed a change of *‑ь to *‑ъ (Lekov’s publication is not accessible to me, but
see Stang 1942: 217, also with criticism of the view). I do not find this morpho-
logical development realistic.
Among the views that ‑tъ somehow reflects *‑ti, a more fruitful approach
is taken by Stang (1942: 217–219; 1969b: 134) and Vaillant (1950: 207–208; loc.
cit.), elaborating on Meillet’s views (Meillet 1913a, loc. cit.; see also Cowgill
2006: 550–551). According to these scholars, inherited *‑tь was shortened to
*‑t, which yielded either ‑∅ or, with a paragogic vowel, ‑tъ. As pointed out by
Stang (1942: 217–218), the further history of the form shows that it was subject
to irregular treatment in the individual Slavic languages. Although the hypoth-
esis of an irregular loss of *‑i is not without its weaknesses, it is, in my opinion,
the one that most convincingly accounts for the facts.
330 chapter 4
One problematic aspect of this view is that the assumed loss of the final
vowel created a closed syllable at a time where only open syllables were
allowed (Aitzetmüller loc. cit.). To overcome this difficulty we must imagine
a scenario where the reflexes of Proto-Slavic final *‑u, *‑i, *‑ə could be omit-
ted in informal speech but were preserved in neutral speech. Subsequently the
informal endings entered into the neutral register. Due to the restriction on
word-final consonants in this register, final *‑t either developed into *‑∅ or, as
in Old Church Slavonic, received a paragogic vowel ‑ъ. I find it unlikely that it
was original forms in PIE *‑t which received a paragogic vowel, as suggested by
Andersen (1998a: 445); the forms in *‑t had already disappeared in pre-Proto-
Balto-Slavic [9].
Apparent reflexes of earlier *‑tъ are also found in East Slavic, e.g. modern
Russian nesët. These forms are the relatively recent results of a phonetic hard-
ening of older ‑t’ and are thus not genealogically identical to the Old Church
Slavonic forms in ‑tъ (see e.g. Kuznecov 1956: 175–178; Stang 1969b: 135; cf.
V. Kiparsky 1967: 189–190 with an overview of hypotheses). The hardening of
ORu. ‑tь > ‑t’ to Ru. ‑t is sometimes mentioned as a parallel to the development
of CS *‑tь to OCS ‑tъ (e.g. Berneker 1904: 370–371). In my view, however, we are
dealing with two different types of processes: in the case of Old Church Slavonic,
an irregular loss of a final vowel followed by the addition of a paragogic vowel;
in the case of the Russian development, a depalatalisation of final ‑t’.
ps *ˈbēi̯ / *ˈbī (CS *bi ̋ ), *ˌmer (*mȇr), *ˌbū (*by̑) pbs *ˌ‑∅ pie prt., inj. *‑d
ocs bi; umrě, →umrětъ; by, →bystъ Li. – Ved. aor. inj. gā́t
ORu. bi; ja, →jatъ; by, →bystь; ONovg. by Lv. – OAv., YAv. prs. inj. mraot̰
OCz. kry; mřě; by OPr. – Gk. aor. ἔβη
La. –
Go. –
Hi. →ēšta (OS)
PIE
Brugmann 1913–1916: 408, 425–426, 615–616: OCS ję reflects *jęs‑t; OCS
umrětъ from older umrě has ‑tъ from prs. | Beekes 1995/2011: 264: OCS
děla reflects *‑āst | Rix 1976/1992: 243: Gk. ‑∅, Ved. ‑t, Hi. ‑t reflect
Verbal Inflection 331
PBS pie *‑d was lost in Proto-Balto-Slavic [5|9]. The athematic ending is not
preserved in Baltic.
PS PBS *‑∅ is continued as PS*‑∅. The zero ending is found in the type
PS*ˈbēi̯ / *ˈbī (CS *bi̋ ), which has an accented acute root-syllable, as shown by
Štk. bȉ (cf. van Wijk 1926; Stang 1957/1965: 134–135). This type may most likely
be traced directly back to a Proto-Indo-European sigmatic aorist *bʰḗi̯hst, and
does not have a variant form in ‑tъ in Old Church Slavonic.
By contrast, the aorist of verbs like OCS umrěti usually presents a second-
and third-person singular ending ‑tъ in conservative Old Church Slavonic man-
uscripts, viz. umrětъ; these forms were unaccented in Proto-Slavic, as shown by
Štk. mrȉje, ȕmrije. Less conservative Old Church Slavonic texts and the remain-
ing Slavic languages, including the Old Novgorod dialect, generally do not pre-
serve the ending ‑(s)tъ, e.g. OCS (Codex Suprasliensis) umrě and the Štokavian
form ȕmrije mentioned above. A variant of the umrětъ type is constituted by
the athematic verbs OCS byti, dati, jasti, which have aorist second- and third-
person singular forms in ‑stъ (bystъ, dastъ, jastъ) and in ‑∅ (by, da); these forms
are also unaccented, cf. e.g. Štk. bȋ, dȍbī. Тhe ‑s‑ appearing before ‑tъ in these
forms has probably been introduced by analogy with the present forms OCS
jestъ, dastъ, jastъ. Similarly, the Old Russian variants bystь, dastь etc. formed
from athematic verbs have ‑stь from the present forms.
Stang (1942: 67, 73) argues against regarding the variants without ‑tъ, i.e. the
type OCS umrě, by, as old. Instead he considers them to be more recent forms
that have arisen under the influence of the type OCS bi (see also e.g. Vaillant
loc. cit.; Rix et al. 1998/2001: 99 n. 10). It should not be forgotten, however, that
the forms without ‑(s)tъ are attested all over the Slavic world, whereas the ‑(s)
tъ forms find only limited support outside Old Church Slavonic. This fact lends
more weight to the view that regards the forms without ‑(s)tъ as the more
original ones. I do not agree with Vaillant (loc. cit.) and Aitzetmüller (loc. cit.)
that the accentuation of SCr. 2–3sg. pȋ (cf. 1sg. pȉh) presupposes the presence
of a final syllable *‑tъ. While it is true that Proto-Slavic final long vowels are
shortened in Serbo-Croatian, this does not hold true for monosyllables, cf. for
instance 2sg. pron. nom. SCr. tȋ from PS*ˌtū, where there is no reason whatso-
ever to assume a late loss of a final syllable.
As for the etymological substance of the element *‑tu/ə, several proposals
have been advanced. The idea that OCS ‑tъ reflects a perfect ending *‑to or
*‑tos (Meillet 1902: 139–142; Stang loc. cit.) is unlikely, first because the reflex
of a Proto-Indo-European perfect ending would be unique in a Slavic preterite
form; second, while *‑tos would indeed yield OCS ‑tъ [7|17|29] (whereas *‑to
Verbal Inflection 333
would yield *‑to [7|29]), the reconstruction of this form requires an additional
hypothesis, namely the analogical addition of the secondary ending *‑s to the
original form. Third, we would have to assume that the second-person singular
ending spread to the third person, which is less probable than a spread in the
opposite direction.
According to an alternative view (Vaillant loc. cit.; Aitzetmüller loc. cit.), OCS
‑tъ represents the preservation of the secondary ending *‑t, which would have
been lost by [9]. What makes this view difficult to accept is that it presupposes
a phonetic development, the retention of a PIE word-final stop in Slavic, that
runs counter to what we can observe elsewhere. It is true that sound change
is not always regular, but such scenarios are less attractive than ones operat-
ing with regular sound change or well-motivated analogical developments.
Vaillant’s implication that the retention of *‑t depended on sandhi conditions
is ad hoc. Furthermore, the choice of the vowel ‑ъ in order to preserve *‑t would
be thinkable in late Common Slavic dialects such as Old Church Slavonic, where
the realisation of ‑ъ was [ʊ] or [ə]. However, in my view the loss of word-final
*‑t already took place in pre-Proto-Balto-Slavic; I do not agree with Vaillant’s
view that “[l]’amuïssement des consonnes finales, dont ‑t, n’est pas ancien en
slave”. At that time the only word-final vowel to show up as ‑ъ in Old Church
Slavonic was *‑u; but I do not see why the speakers would chose precisely *‑u as
a paragogic vowel at this point, rather than *‑a, *‑e or *‑i. This is not accounted
for in Vaillant’s explanation.
Some authors have connected OCS ‑tъ in the second- and third-person singu-
lar forms of the aorist with the homophonous element found in the third-person
singular of the present, e.g. vedetъ (e.g. Fortunatov 1908: 27–28; Vondrák loc. cit.;
Kortlandt loc. cit.; 1998: 285; cf., however, the criticism of Vondrák’s view in Stang
loc. cit.; and see Ackermann 2014: 133–135). While I basically believe this con-
nection to be valid, I explain OCS ‑tъ differently from these authors (see § 4.9).
In order to understand how the present third-person singular ending ‑tъ
came to be used as an aorist second- and third-person singular ending at a
pre-stage of Old Church Slavonic, we may imagine a scenario where the pres-
ent third-person singular ending PS*‑(e)ti was shortened to *‑(e)t in informal
speech (see § 4.9). When informal *‑(e)t was imported into the neutral register,
the word-final stop was not tolerated and was either lost or received a para-
gogic vowel ‑ъ ([ʊ] or [ə]). Subsequently the stylistic alternation between ‑tъ
and ‑∅ in the third-person singular of the present was transferred to certain
monosyllabic aorist third-person singular forms and, since the second- and
third-person singular forms where identical everywhere outside the present,
also to the second-person singular.
334 chapter 4
It remains to be explained why ‑tъ was added to the second- and third-
person singular forms in verbs with originally a root aorist, but not in verbs
with a sigmatic or thematic aorist. This may be due to the fact that Proto-Slavic
root aorists contained a short vowel, e.g. PS2–3sg. *ˌmer (CS *mȇr), whereas
sigmatic aorists had a synchronic long vowel, e.g. *ˈbēi̯ / *ˈbī, *ˈkrū (CS *bȉ,
*kry̏ ). The forms of the sigmatic aorist, with their long vowel, had enough
phonological weight already. By contrast, the nucleus of the root-aorist forms
contained only a short vowel; by adding the new ending ‑tъ to the forms,
the speakers obtained a more harmonious system. This hypothesis may find
some support in the fact, noted already by Wiedemann (1886: 14, 21), that Old
Church Slavonic forms without ‑tъ are found predominantly in compound aor-
ist forms, e.g. priję (Zographensis). The Serbo-Croatian reflexes mrȉje and bȉ,
with the opposite quantitative distribution of that found in pre-Proto-Slavic,
are the results of later developments. The explanation presented here of the
Old Church Slavonic aorist second- and third-person singular forms in ‑tъ is
related to the one given by Kuryłowicz (1964: 157).
ps *ˈsēde, *ˌu̯ede (CS *sě̋de, *vȅde) pbs *ˌ‑e pie prt., inj. *‑ed
PIE
Brugmann 1913–1916: 611: Ved. ‑at, Gk. ‑ε, Osc. kúmbened, OIr. prs. ‑beir,
OCS ‑e go back to PIE *‑et | Beekes 1995/2011: 260: PIE *‑et | Rix
1976/1992: 243: Gk. ‑ε, Ved. ‑at, OCS ‑e, Hi. ‑et reflect PIE *‑et | Sihler
1995: 456, 460–462: Gk. ‑ε is from PIE *‑et or *‑e | Weiss 2009/2011:
387: OLa. FECED preserves secondary ending | Krahe 1942/1967:—
| Boutkan 1995: – | Kloekhorst 2008a: 800: Hi. postvocalic ‑t reflects
PIE *‑t
Verbal Inflection 335
PBS
Olander 2009: 197: PBS unaccented *ˌ‑e reflects PIE *‑ét | Kortlandt
1979b/2009: 159: Slavic ‑e is from PIE *‑et | Endzelīns 1971: [see prs.,
§ 4.9] | Stang 1942, 1966: [see prs., § 4.9] | Otrębski 1956: [see prs.,
§ 4.9] | Endzelīns 1923: [see prs., § 4.9]
PS Vondrák 1908/1928: 119: Slavic ‑e is from *‑et | Meillet 1924/1934:
322 | Stang 1942: 64: OCS ‑e is from *‑et | Vaillant 1966: 16: Slavic ‑e
is from *‑et | Arumaa 1985: 276 | Aitzetmüller 1978/1991: 179: OCS
‑e is from *‑et
PIE The ending *‑ed consisted of the e-grade of the thematic vowel followed
by the active third-person singular ending *‑d. The ending was preserved in the
ancient Indo-European languages.
PBS PIE *‑éd yielded PBS *ˌ‑e [4|5|9]. For the Baltic reflexes of this ending
see § 4.9.
PS PBS *ˌ‑e was retained as PS*ˌ‑e (CS *‑e [29]), preserved in the old Slavic
dialects. As shown by forms like Štk. aor. plȅte, zȁplete from PS*ˌplete (CS
*plȅte), the form was unaccented in Proto-Slavic, like the corresponding
second-person form.
ps →*dādi ̯i (CS *dad’ь) pbs *‑i ̯ḛ̄ pie opt. *‑i ̯eh₁d
PIE
Brugmann 1913–1916: 545–552: Slavic ‑ždь is from 2sg. | Beekes 1995/
2011: 275: PIE *‑i̯éh₁t | Rix 1976/1992: 231, 261: Gk. εἴη, Ved. syā́t, OLa.
SIED reflect PIE *‑i̯éh₁t | Sihler 1995: 552–553, 596: OLa. SIED , Gk. εἴη,
Ved. syā́t are from PIE *‑i̯éh₁t; in later La., sī‑ was generalised | Weiss
2009/2011: 416–417: OLa. SIED , Gk. εἴη reflect PIE *‑i̯éh₁t | Krahe 1942/
1967: 107, 110, 137, 141: Germ. has generalised *‑ī‑ from pl. | Boutkan 1995:
337–338, 464: Go. ‑i is from *‑īt < PIE *‑ih₁t, with generalised zero grade
and shortening of a high vowel after loss of *‑t | Kloekhorst 2008a: –
PBS
Olander 2009: – | Kortlandt 1979b/2009: – | Endzelīns 1971:
– | Stang 1942: 244–245, 1966: 425–426, 439: OLi. tedůd, tedůdi
may contain shortened reflex of PIE athem. opt. *‑ī‑; [see also 2sg.,
§ 4.8.1] | Otrębski 1956: – | Endzelīns 1923: –
PS Vondrák 1908/ 1928: 120; [see 2sg., § 4.8.1] | Meillet 1924/ 1934: –
| Stang 1942: 241–242: Slavic 3sg. is probably analogical from
2sg. | Vaillant 1966: 35–36; [see 2sg., § 4.8.1] | Arumaa 1985: 309–
311; [see 2sg., § 4.8.1] | Aitzetmüller 1978/1991: 193
PBS pie *‑i̯éh₁d yielded PBS *‑i̯ḛ̄ [1|5|9]. The ending was not retained in Baltic.
PS The regular reflex of PBS *‑i̯ḛ̄ would have been PS*‑i̯ē [13] (CS *‑ja [27|29]);
however, the endings actually found in Slavic should probably be traced back
to PS*‑i̯i. I assume that the third-person singular form underwent the same
development as the second-person singular form, with which it merged after
the loss of word-final *‑s in pre-Proto-Slavic [17]: pre-Proto-Slavic *‑i̯ē was
replaced with *‑i̯ī by analogy with the dual and plural forms, and *‑i̯ī was
shortened to PS *‑i̯i.
Verbal Inflection 337
ps →*u̯eˈdəi ̯ (CS *vedì; *pišì) pbs *‑ˈai̯ pie opt. *‑oi ̯h₁d
PIE Brugmann 1913–1916: 557–561: Slavic vedi, Li. tevediẽ reflect PIE
*‑oit | Beekes 1995/2011: 275: PIE *‑oih₁t | Rix 1976/1992: 231, 261: Gk.
‑οι, Ved. ‑et, Go. ‑ai are from PIE *‑oi̯h₁t | Sihler 1995: 596–597: Gk. ‑οι,
Ved. ‑et, Go. ‑ai reflect PIE *‑oi̯t | Weiss 2009/2011: 417: them. opt. was
eliminated in Italic and Celtic, possibly except OLa. OPETOIT | Krahe
1942/1967: 108: Go. ‑ai goes back to PIE *‑oit | Boutkan 1995: 321, 323:
Go. ‑ai reflects PGmc. *‑ai < PIE *‑oih₁t, with retention of the diphthong
before original *‑t | Kloekhorst 2008a: –
PBS Olander 2009: 198: PS*‑ˈai̯ reflects PBS *‑ˈai < PIE *‑ói̯h₁t | Kortlandt
1979b/2009: – | Endzelīns 1971: 243 | Stang 1942: 243–244, 1966:
422: Li. tenešiẽ, Gk. ‑οι, Go. bairai, Ved. ‑et reflect PIE *‑oit | Otrębski
1956: 227–228: Li. tedirbiẽ reflects PIE *‑oit | Endzelīns 1923: –
PS Vondrák 1908/1928: 119–120: Slavic ‑i is from *‑oit, with regular reflex
of *oi in a final syllable closed by a consonant | Meillet 1924/
1934: 329–330; [see 2sg., § 4.8.2] | Stang 1942: 239: OCS ‑i is from
*‑oit | Vaillant 1966: 35–36; [see 2sg., § 4.8.2] | Arumaa 1985: 309–
311 | Aitzetmüller 1978/1991: 192: OCS ‑i (for *‑ě) from *‑oit is due to
levelling with the types znaji and nosi
PS PBS *‑ˈai̯ would have yielded PS*‑ˈai̯ (CS *‑ě [22|29]; *‑ji [20|22|29])
according to the sound laws [22|29], but at some point in the prehistory of
Proto-Slavic the second-person singular ending PS*ˈ‑əi̯ (CS *‑i; *‑ji) was taken
over by the third-person singular, perhaps by analogy with the corresponding
athematic forms which had been identical since the loss of word-final frica-
tives [17]. The old Slavic dialects retain the ending.
PS →*u̯ēsau̯ā (CS *věsova, →*věsově) PBS *‑u̯ā̰ PIE prt., inj. *‑u̯ah₂
ps *sēdau̯ā (CS *sědova, →*sědově) pbs *‑au̯ā̰ pie prt., inj. *‑ou̯ah₂
PS *u̯edāi ̯u̯ā (CS *veděva, →*veděvě; pbs *‑ō̰i ̯u̯ā̰ pie opt. *‑oi ̯h₁u̯ah₂
*pišiva, →*pišivě)
PIE
Brugmann 1913–1916: 638–639: PIE primary *‑u̯ e/os(i), secondary
*‑u̯ ē̆/ō̆; Slavic ‑vě was probably influenced by 1du. pron. vě; Slavic ‑va may
be modelled on 2du. ‑ta or on nominal masc. nom.–acc. du. ‑a | Beekes
1995/2011: 271: PIE primary *‑u̯ es, secondary *‑u̯ e; Slavic ‑ě is analogi-
cal from pronouns; Hi. ‑wen(i) may originate in du. | Rix 1976/1992:
– | Sihler 1995: 454: PIE primary *‑u̯ os, secondary *‑u̯ ē̆ | Weiss
2009/2011: 384, 386: PIE primary *‑u̯ os; CS *‑va is from PIE secondary
ending *‑u̯ oh₁; vowel of OCS ‑vě is modelled on pron. vě | Krahe 1942/
1967: 97: Go. bairos reflects PIE *‑ou̯ es (“obwohl die lautliche Entwicklung
[. . .] nicht ganz klar ist”) | Boutkan 1995: 319–320, 324–325, 337: PIE
primary *‑ues, secondary *‑ue; Go. ‑os is not the direct reflex of *‑owes,
but perhaps of remade *‑ōwes; Go. secondary ending ‑wa may repre-
sent *‑ueh₁, replacing PIE *‑ue by analogy with 1du. pron.; Go. ‑u may
go directly back to PIE *‑ue | Kloekhorst 2008a: 1001; [see 1pl., § 4.15]
PBS
Olander 2009: – | Kortlandt 1979b/2009: – | Endzelīns 1971:
205 | Stang 1942: 236, 1966: 419–420: Li. ‑va, refl. ‑vos(i), reflects *‑vā́,
perhaps also in Slavic ‑va; BS *‑vā́ may have been influenced by 2du.
*‑tā́ | Otrębski 1956: 185–186: Li. ‑va, refl. ‑vos(i), is related to OCS
‑vě | Endzelīns 1923: –
PS Vondrák 1908/1928: 115, 120: Slavic ‑vě is probably modelled on 1du.
pron. | Meillet 1924/1934: 325: CS *‑va corresponds to Li. ‑va, refl.
‑vos(i) | Stang 1942: 223: Slavic ‑vě may correspond to Ved. secondary
‑va, with lengthening of the vowel, or it may have been influenced by
1du. pron. ‑vě; Slavic ‑va may have been influenced by 2du. ‑ta, or it could
correspond to Li. ‑va < *‑vā́ | Vaillant 1966: 14–15: CS *‑va seems sec-
Verbal Inflection 341
PIE Indo-Iranian points to a primary marker PIE *‑u̯ e/os, with the pres-
ent marker *‑i added in Avestan. In Hittite there is no dual category, but the
first-person plural endings prs. ‑weni, prt. ‑wen seem to contain the same *‑u̯ ‑ as
the dual endings in the remaining Indo-European languages (see e.g. Eichner
1975: 87; Kloekhorst loc. cit.; Melchert forthc. § 3.3.4.1). The n of the Hittite end-
ings is reminiscent of that of Gk. 1pl. ‑µεν, but the role it plays in the verbal
system is unclear.
The reflex of a primary thematic ending *‑ou̯ es is probably found in Gothic
‑os, although the match is possibly not perfect (see the discussion in Boutkan
1995: 319–320; cf. also Kroonen 2013: xxvi, who suggests a thematic first-person
dual ending PIE *‑oh₁u̯ es, yielding *‑ōwiz > *‑ōiz > Go. ‑os). It is likely that the
endings found in Balto-Slavic are the original secondary endings. The long
vowel of PBS *‑u̯ ā̰ is incompatible with the short vowel of Ved. ‑va, however;
Avestan is inconclusive in this respect. While the Gothic optative ending ‑aiwa
points to a final long vowel as in Balto-Slavic, Go. prt.-prs. magu and Early
Runic prt. waritu fit Ved. ‑va, if the forms are derived from PIE *‑u̯ e/o.
Since the Germanic preterite system generally reflects the original perfect
endings whereas the optative retains the Proto-Indo-European secondary end-
ings, we may perhaps speculate that Go. ‑u originally belonged to the perfect
system, and that ‑wa was the secondary ending. In that case we may reconstruct
a Proto-Indo-European secondary ending *‑u̯ ah₂ on the basis of Go. ‑wa and
PBS *‑u̯ ā̰, and a perfect ending *‑u̯ e/o preserved in Go. ‑u and Ved. ‑va (attested
both as a secondary ending and, in post-Saṃhitā texts, as a perfect ending).
It should be clear, however, that the paucity of attested forms in Indo-
Iranian, Germanic and Balto-Slavic, the unsettled status of the Anatolian evi-
dence and the loss of all first dual forms in the other Indo-European branches
make these reconstructions uncertain.
PS PBS *‑u̯ ā̰, *‑au̯ ā̰, *‑ī�ṵ ̯ ā̰, *‑a̰ iu̯ ̯ ā̰ are reflected as PS*‑u̯ ā [13] (CS *‑va [29]),
*‑au̯ ā [13] (*‑ova [29] → *‑eva; *‑jeva [20|29]), *‑īu̯ ā [13] (*‑iva [29]), *‑āi̯u̯ā [13]
(*‑ěva [22|29]; *‑jiva [20|22|29]). Alongside PS*‑u̯ ā, attested in West Slavic,
Old Serbian and, rarely, Old Russian (V. Kiparsky 1967: 191), the Slavic dialects
present more widespread reflexes of a person–number marker CS *‑vě (for the
distribution of the markers see Meillet loc. cit.). Solely on the basis of the dis-
tribution of the markers one would perhaps trace both *‑va and *‑vě back to
Proto-Slavic, or even only the latter (cf. Vaillant loc. cit.). However, the compar-
ison with Lithuanian ‑va, refl. ‑vos(i), makes it plausible that the more original
form was CS *‑va. The reflexes of CS *‑vě may have arisen due to influence from
the first-person dual pronoun PS*u̯ ē (§ 3.10.8).
While the preterite preserves the reflex of the thematic o-vowel, the present
ending underwent the same generalisation of the e-vowel as in the first-person
plural (§ 4.15).
PS *u̯edāi ̯tā (CS *veděta; *pišita) PBS →*‑a̰i ̯tā̰ PIE opt. *‑oi ̯h₁tom
PIE
Brugmann 1913–1916: 639–642: PIE primary *‑to/es or *‑tʰo/es is reflected
in Ved. ‑thaḥ, Go. ‑ts; PIE secondary *‑tom is reflected in Ved. ‑tam, Gk.
‑τον (where it was also used as the primary ending); relationship of BS
endings to other languages is unclear | Beekes 1995/2011: 271: PIE pri-
mary *‑the/os, perhaps with *h₁; PIE secondary *‑tom | Rix 1976/1992:
245, 252–253: Ved. ‑thaḥ, Go. ‑ts reflect PIE primary ending *‑tos; Gk. ‑τον
is from PIE secondary ending *‑tom | Sihler 1995: 454, 470, 605: PIE
*‑th₁es is not attested in Gk., where -τον corresponds to PIE secondary
ending *‑tom > Ved. ‑tam; Umbr. ipv. 2pl. etato < PItal. *‑tā is probably
original du. ending corresponding to OCS ‑ta, Li. ‑ta | Weiss 2009/
2011: 384, 386, 387: PIE primary *‑th₂es; South Picene videtas ‘you (pl.)
see’ may reflect primary 2du. *‑teh₂s, based on secondary *‑teh₂ reflected
in Umbr. ipv. 2pl. ‑to, Li. 2du. ‑ta, OCS ‑ta; but Ved. ‑tam, Gk. ‑τον point to
PIE *‑tom | Krahe 1942/1967: 97: Go. bairats is from PIE *‑et/tʰes, with
analogical ‑a‑ for *‑i‑ and regular development of *‑þs to ‑ts | Boutkan
1995: 319–320, 325, 337: PIE primary *‑the/os; Go. them. vowel ‑a‑ is ana-
logical; Go. ‑ts from PGmc. *‑þs may be analogical or regular; Ved. ‑tam,
Gk. ‑τον reflect PIE secondary *‑tom | Kloekhorst 2008a: –
PBS
Olander 2009: – | Kortlandt 1979b/2009: – | Endzelīns 1971:
205 | Stang 1942: 236, 1966: 420: Li. ‑ta, refl. ‑tos(i), from Proto-Li. *‑tā́
is identical to CS *‑ta | Otrębski 1956: 186: Li. ‑ta, refl. ‑tos(i), is from
*‑tā, also in OCS ‑ta | Endzelīns 1923: 554: Lv. 2pl. ‑t, ‑tās may repre-
sent original 2du. ending, corresponding to Li. ‑ta, Slavic ‑ta, or it may
have been remade on the model of 1pl. ‑mās
PS Vondrák 1908/1928: 115, 120: OCS ‑ta, Li. ‑ta go back to *‑tā | Meillet
1924/1934: 325–327: Li. ‑ta and CS *‑ta correspond to secondary Ved. ‑tām,
Verbal Inflection 345
Gk. ‑τᾱν, minus “la nasale finale inorganique”; oldest (pie) form of sec-
ondary ending was *‑tā | Stang 1942: 223: OCS ‑ta corresponds to Li.
‑ta | Vaillant 1966: 15: PBS *‑tā | Arumaa 1985: 282: BS forms may
be related to PIE ending *‑tām | Aitzetmüller 1978/1991: 179
PS PBS *‑tā̰, *‑etā̰, *‑ī�t̰ ā̰, *‑a̰ it̯ ā̰ yielded PS*‑tā [13] (CS *‑ta [29]), *‑etā [13]
(*‑eta [29]; *‑jeta [29]), *‑ītā [13] (*‑ita [29]), *‑āi̯tā [13] (*‑ěta [22|29]; *‑jita
[20|22|29]). To the extent that the forms are attested in the Slavic languages,
they generally preserve the original endings.
346 chapter 4
PIE
Brugmann 1913–1916: 639–642: PIE primary ending was probably *‑tes,
reflected in Ved. ‑taḥ, OCS ‑te; PIE secondary ending *‑tām is seen in
Ved. ‑tām, Gk. ‑την, Dor. ‑τᾱν; Gk. ‑τον is originally secondary 2du. end-
ing | Beekes 1995/2011: 271: Ved. ‑taḥ, OCS ‑te reflect PIE primary *‑tes;
Gk. ‑την, Dor. ‑τᾱν reflect secondary *‑teh₂m | Rix 1976/1992: 252–253:
‑τον is originally secondary 2du. ending | Sihler 1995: 454, 470: PIE
*‑tes was lost in Gk., which employs 2du. secondary ending | Weiss
2009/2011: 384, 387: PIE primary *‑tes; Ved. ‑tām, Gk. ‑την reflect PIE sec-
ondary *‑teh₂m | Krahe 1942/1967: – | Boutkan 1995: 319–320: PIE
*‑tes | Kloekhorst 2008a: –
PBS
Olander 2009: – | Kortlandt 1979b/2009: – | Endzelīns 1971:
205 | Stang 1942: –, 1966: – | Otrębski 1956: – | Endzelīns 1923: –
PS Vondrák 1908/1928: 115–116: OCS ‑te is perhaps from *‑tes, also in Ved.
‑taḥ; OCS variant ‑ta probably has ‑a from nominal masc. nom.–acc.
du. | Meillet 1924/1934: 325–327: CS *‑te and perhaps also *‑ta are
old | Stang 1942: 223–224: OCS *‑ta may be old, *‑te was perhaps orig-
inally the primary ending | Vaillant 1966: 15: OCS ‑te may go back to
PIE *‑tes (as in Ved. ‑taḥ) or to PIE *‑te; the Slavic variant ‑ta partly shows
the introduction of the general du. marker ‑a (as in 1du. ‑va, on the model
of 2du. ‑ta) and partly influence from nominal dual forms | Arumaa
1985: 283: OCS ‑te is replaced with ‑ta in other Slavic languages; both ‑te
and ‑ta are isolated in the IE languages | Aitzetmüller 1978/1991: 179:
PIE primary *‑tes or *‑tos, as in Ved. ‑taḥ; Gk. ‑τον may point to *‑te/o, with
secondary ‑ν; Ved. ‑tām, Gk. ‑την point to secondary *‑tām; OCS variant
‑ta has analogical ‑a
348 chapter 4
PBS In Balto-Slavic only the primary marker *‑tes survived, yielding PBS
*‑tes. The marker is not preserved in Baltic, where a reshaped version of the
third-person singular is used for in all third-person forms (§ 4.9).
PS PBS primary and secondary *‑tes, *‑etes are continued as PS*‑te [17] (CS
*‑te [29]), *‑ete [17] (*‑ete [29]; *‑jete [29]). The original ending is preserved
only in Old Church Slavonic texts, alongside a new ending ‑ta also found in the
remaining Slavic dialects (for the attestations in Old Church Slavonic see Diels
1932–1934/1963: 228). The final vowel of ‑ta was introduced by analogy with
PS1du. *‑u̯ ā and 2du. *‑tā, and partly also by analogy with the nominal mascu-
line nominative–accusative dual ending PS*‑ā (CS *‑a).
OCS jesmъ; Štk. →jèsmo Li. →ẽsam(e); OLi. Ved. →smási, smáḥ
ORu. jesmъ, →jesmy; →ésme, →esmé; refl. OAv. →mahī; YAv. →mahi
ONovg. →jesme; Ukr. (ne) →důdameś Gk. →ἐσµέν; Ion. →εἰµέν; Dor.
→damó Lv. →ȩsam →εἰµές
OCz. →jsme, →jsmy OPr. →asmai La. sumus
Go. →sijum; ONor. →erum
Hi. →ešuwani
Verbal Inflection 349
ocs →vedemъ; pišemъ; Štk. →nesémo, Li. →vẽdam(e); refl. Ved. →bhárāmasi,
→nèsēmo →vẽdamės bhárāmaḥ
ORu. →vedemъ; pišemъ; ONovg. Lv. →vȩ̀lkam; refl. OAv. →sə̄ṇghāmahī;
→možemъ; (ne) znajemo (i.e. ‑mъ); →ceļ̂amiês; dial. YAv. →barāmahi
Ukr. →vedemó; →píšemo →‑me; refl. →‑mēs Gk. →φέροµεν; Dor.
OCz. →vedem, →vedeme, →vedemy; OPr. →giwammai →φέροµες
píšem, →píšeme, →píšemy; Slk. dial. La. legimus
→budemo Go. bairam
Hi. →daškēweni (OS)
ps *u̯ēsame (CS →*věsomъ, *věsome, →*věsomo) pbs *‑me pie prt., inj. *‑me
PS *ˈsēdame (CS →*sě̋domъ, *sě̋dome, PBS *‑ame PIE prt., inj. *‑ome
→*sě̋domo)
PS *u̯eˈdāi ̯me (CS →*vedě̋mъ, *vedě̋me, PBS *‑ˈa̰i ̯me PIE opt. *‑oi ̯h₁me
→*vedě̋mo; →*piši ̋mъ, *piši ̋me, →*piši ̋mo)
PIE
Brugmann 1913–1916: 616–623: PIE primary *‑me/os, continued in Ved.
‑maḥ, Gk. Dor. ‑µες, La. ‑mus, OIr. ‑beram; PIE *‑me/osi, continued in
Ved. ‑masi, OIr. ammi, bermai; secondary *‑me/o, *‑mē/ō, continued
in Ved. ‑ma, ‑mā, Go. ‑um, ‑aima; Li. ‑me perhaps represents merger of
PIE *‑me and *‑mē; Slavic ‑mъ is unclear; ‑me may reflect *‑mes or *‑me;
‑mo may reflect *‑mos or *‑mo; ‑my has arisen under the influence of
1pl. pron.; in Slavic the them. vowel was replaced with *‑e‑ in them. prs.
vedemъ by analogy with other persons and i̯e-prs. znajemъ | Beekes
1995/2011: 259–261: PIE primary athem. *‑mes, secondary *‑me; primary
them. *‑omom, secondary *‑omo/e; OCS ‑mъ reflects PIE primary them.
ending *‑mom | Rix 1976/1992: 243–244, 251–252: Gk. ‑µεν, Hi. ‑wen
(with ‑w‑ from 1du. *‑u̯ e) are from PIE secondary ending *‑me plus *‑m
from 1sg.; Gk. Dor. ‑µες reflects PIE primary ending *‑mes | Sihler
1995: 463–464, 465: PIE primary ending *‑mos, secondary *‑me; OCS
‑mъ reflects PIE *‑mos; Li. ‑me represents secondary ending; Gk. ‑µεν is
secondary ending plus obscure nasal element; Gk. Dor. ‑µες represents
contamination of primary and secondary endings | Weiss 2009/2011:
385–386: PIE had accented *‑mes vs. unaccented *‑mos; Slavic ‑me is
either inherited or based on analogy with 2pl.; Li. ‑me, refl. ‑mės(i), reflects
*‑mē | Krahe 1942/1967: 97, 136: Go. ‑am reflects PGmc. *‑om(i)z
from PIE them. *‑omes; ONor. erum < PGmc. *ezum‑ reflects *ezm̥ ‑; Go.
sijum contains si‑ from 3pl. and ‑um from preterite | Boutkan 1995:
308, 313–317, 324: PIE primary them. *‑omo(m), athem. *‑mes; Go. ‑am,
ONor. ‑um contain contaminated *‑omes; Go. opt. ‑ma < *‑mē perhaps has
*‑ē from du. pron. *u̯ eh₁ | Kloekhorst 2008a: 1000–1001: Hi. prs. ‑weni,
prt. ‑wen may be related to 1du. ending of other IE languages
PBS
Olander 2009: 196: PBS *‑ˈamas reflects PIE *‑ómos | Kortlandt
1979b/2009: 160–161: Li. ‑me is from PIE *‑me, the acute of the reflex-
ive form being taken from sg.; OPr. ‑mai is from PIE *‑mo plus *‑i from
sg.; Slavic ‑my has arisen under the influence of 1pl. pron.; Slavic ‑me
reflects secondary athem. ending *‑me; Slavic ‑mo reflects PIE primary
athem. ending *‑mo, *‑mos; Slavic ‑mъ reflects PIE primary them. end-
ing *‑omom | Endzelīns 1971: 204–205: Li. ‑m(e), ‑mės(i), reflects *‑mē,
perhaps corresponding to Go. ‑ma, Ved. ‑mā; OPr. ‑mai probably has ‑ai
from 1sg. | Stang 1942: 234–235, 1966: 416–417: Li. ‑me, refl. ‑mės(i),
(from proto-Li. *‑mḗ) and Lv. dial. refl. ‑mēs reflect old secondary end-
ing, related to Ved. ‑ma (rarely ‑mā), Go. opt. bairaima; Lv. refl. ‑mies may
be analogical; OPr. ‑mai may be a new middle ending, or it may consist
of *‑ma or *‑mā plus ptcl. *‑i | Otrębski 1956: 184–185: Li. ‑me, refl.
‑mės(i), dial. ‑ma, refl. ‑mos(i), point to *‑mē, *‑mā, also reflected in Lv.
refl. ‑mēs, ‑mās; OLi. ‑mi, Lv. ‑m, refl. ‑miês, reflect *‑mie | Endzelīns
352 chapter 4
1923: 551–553, 689–690: oldest Lv. ending is ‑me from *‑mē, identical with
Li. ‑me; this may be the inherited secondary ending, perhaps identical
with Go. ‑ma and Ved. ‑mā
PS
Vondrák 1908/1928: 116–117, 120: Slavic ‑mъ is ambiguous; ‑me may corre-
spond to Li. ‑me and Gk. ‑µεν (or Dor. ‑µες); ‑my is based on analogy with
1pl. pron.; ‑mo may be old pf. ending *‑mo | Meillet 1924/1934: 313–316,
330–331: reconstruction of PIE ending is difficult; Li. ‑me, ‑mės reflects
long vowel; them. vowel *‑o‑ was replaced with *‑e‑ in them. present by
analogy with i̯e-present | Stang 1942: 222–223, 240–241: Slavic ‑mъ and
‑mo reflect PIE *‑mos or *‑mon or secondary ending *‑mo; Slavic ‑me is
from *‑mes or secondary ending *‑me, or it is based on analogy with 2pl.
*‑te; Slavic ‑my is perhaps modelled on 1pl. pron.; all Slavic endings may
reflect PIE *‑mos or *‑mon | Vaillant 1966: 11–12, 32–34: Slavic ‑my is
taken from 1pl. pron.; Slavic ‑me is modelled on 2pl. ‑te; Slavic ‑mъ reflects
PIE *‑mos, also in Ved. ‑maḥ, La. *‑mos; Slavic ‑mo reflects old pf. ending,
also in Ved. ‑má; Li. ‑me, refl. ‑mės(i), is modelled on 2pl. ‑te, refl. ‑tės(i);
OPr. ‑mai has ‑ai from 2sg. or from nom. pl. | Arumaa 1985: 278–279,
311: Slavic ‑my is influenced by 1pl. pron.; ‑me may be inherited as such or
go back to *‑mes as in Gk. Dor. ‑µες; ‑mo may reflect *‑mos, as in La. ‑mus,
or correspond to Baltic forms without *‑s | Aitzetmüller 1978/1991:
178, 192–193: OCS ‑mъ is either from PIE *‑mos (if development of PIE
*‑os > ‑ъ is accepted) or *‑mon; Slavic ‑mo either reflects (1) PIE secondary
ending (Ved. ábharāma) or pf. ending (vidmá), or (2) PIE *‑mos; Slavic
‑me was either influenced by 2pl. ‑te or inherited from PIE *‑mes > Gk.
Dor. ‑µες; Slavic ‑my has arisen under the influence of 1pl. pron.
PS The Proto-Balto-Slavic primary endings prs. *‑mas, *‑ˈamas and the sec-
ondary endings prt. *‑me, *‑ˈame, opt. *‑ī�m
̰ e, *‑ˈa̰ im
̯ e yielded PS prs. *‑mə [17]
(CS *‑mъ [29]), *‑aˈmə [13|17] (*‑omъ [29] → *‑emъ; *‑jemъ [20|29]), prt. *‑me
(*‑me [29]), *‑ame (*‑ome [29]), *‑īme [13] (*‑ime [29]), *‑ˈāi̯me [13] (*‑ě̋ me
[22|29]; *‑i̋me [20|22|29]). Although no attested Slavic dialect preserves the
354 chapter 4
which I do not agree (see Olander 2012: 322–325), it is not clear what the source
for the analogical *‑x would have been.
Thus despite a high number of hypotheses on the background of the
Common Slavic variant *‑mo, the problem has not found a convincing solution
yet. As mentioned above, however, it seems relatively clear that the ending
arose after the dissolution of Proto-Slavic as a reaction to the loss of the final
jer in the original primary ending PS*‑mə.
In the present first-person dual and first-person plural the reflex of the the-
matic vowel PIE *‑o‑ has been replaced with *‑e‑ in the attested Slavic dia-
lects, in contrast to the preterite where the original o-vocalism was preserved,
e.g. OCS prs. vedemъ, aor. věsomъ, sědomъ. The introduction of e-vocalism in
the present is usually explained as analogical based on the i̯e-present, where
PS*‑i̯a‑ became *‑i̯e‑ [20] (see Ul’janov 1888: 21; Brugmann loc. cit.; Meillet loc.
cit.; Stang 1942: 236 n. 2). Since I assume that the fronting of non-front vow-
els after palatal consonants [20] took place after the dissolution of the Slavic
proto-language (see the discussion in § 1.5.5), it follows that the replacement
of thematic *‑a‑ with *‑e‑ in the first-person dual and first-person plural of the
present tense had not yet been carried out in Proto-Slavic but was a Common
Slavic process.
ocs jeste Li. →ẽsat(e); OLi. éste, esté Ved. sthá, →sthána
ORu. jeste; ONovg. jeste Lv. →ȩsat OAv. stā, spašnuϑā; YAv. ?
OCz. jste OPr. →asti, →astai, →estei Gk. ἐστέ
La. →estis
Go. →sijuþ
Hi. →paitteni, →esteni
356 chapter 4
PS *dāˈdīte (CS *dadi ̋te) PBS *‑ˈī ̰te PIE opt. *‑ih₁te
ps *u̯eˈdāi ̯te (CS *vedě̋te; *piši ̋te) pbs *‑ˈa̰i ̯te pie opt. *‑oi ̯h₁te
PIE
Brugmann 1913–1916: 623–626: PIE *‑te is found in Gk. ‑τε, La. ‑tis (with
analogical ‑s), ipv. ‑te, OIr. ‑d, ‑th, Li. ‑te, OCS ‑te; primary ‑tha of IIr. may
be an innovation; Li. refl. ‑tės(i) is modelled on 1pl. ‑mės(i) | Beekes
1995/2011: 259–260: PIE had primary athem. *‑th₁e, them. *‑eth₁e, and
secondary athem. *‑te, them. *‑ete | Rix 1976/ 1992: 252: postu-
lated PIE primary ending *‑tes (> La. ‑tis) is doubtful | Sihler 1995:
464–465: PIE primary and secondary endings were *‑te; La. ‑tis is from
2du. *‑th₁es | Weiss 2009/2011: 386: La. ‑tis reflects PIE primary end-
ing *‑tes; aspiration in Vedic is taken from 2du.; PIE secondary ending
was *‑te; Li. ‑te, refl. ‑tės(i), reflects long vowel | Krahe 1942/1967: 97,
103, 109: Ved. ‑tha, Gk. ‑τε, PGmc. *‑þ(i), *‑ð(i) reflect PIE *‑t/tʰe; Go. prt.
‑uþ perhaps reflects PIE aor. ending, with u from 1 and 3pl. | Boutkan
1995: 308, 317–318, 321, 336, 338: Go. ‑iþ is from PIE primary *‑eth₁e; Gmc.
358 chapter 4
and Greek, the Greek second-person plural marker ‑τε would have to be traced
back to PIE *‑te, either because the Proto-Indo-European primary marker was
*‑te after all, or because the primary marker was replaced with the secondary
marker *‑te in Greek.
Here I have tentatively reconstructed *‑th₁e for the proto-language. In any
case, the question of the origin of the aspiration in Indo-Iranian does not have
any significance for the analysis of the Balto-Slavic material.
In the thematic present and preterite the marker was preceded by the
thematic vowel *‑e, in the athematic optative by the optative suffix *‑ih₁‑,
and in the thematic optative by the thematic vowel *‑o‑ and the optative
suffix *‑ih₁‑.
In the Baltic thematic present the inherited e-grade of the thematic vowel
was replaced with o-grade. The tone and vocalism of Li. dial. (Tverečius) ‑iẽ‑ in
refl. sėʒ́iẽťės are probably analogical, as in the first plural (see § 4.15).
PS PBS athem. prs. *‑te and them. prs. *‑ˈete yielded PS*‑te (CS *‑te [29]), *‑eˈte
[13] (*‑ete [29]; *‑jete [29]). Likewise, the preterite endings PBS athem. *‑te and
them. *‑ete yielded PS*‑te (CS *‑te [29]), *‑ete (*‑ete [29]). In the imperatives,
PBS *‑ī�t̰ e, *‑ˈa̰ it̯ e yielded PS*‑īte [13] (CS *‑ite [29]), *‑ˈāi̯te [13] (*‑ě̋ te [22|29];
*‑ji̋te [20|22|29]). The endings are preserved in the old Slavic dialects.
PS →*sanˈti (CS *sǫ́tь); *dāˈde/inti PBS *‑enti, *‑inti PIE *‑enti, *‑n̥ti
(CS *dádętь)
PIE
Brugmann 1913–1916: 626–637: OCS jadętъ, RuCS sǫtь go back to PIE
athem. *‑e/onti; RuCS dadętь has athem. *‑n̥ ti; RuCS ‑ǫtь reflects them.
*‑onti | Beekes 1995/2011: 234: PIE athem. ending was *‑nti; them.
ending was *‑o, preserved in Li., but elsewhere largely replaced with ana-
logical *‑onti | Rix 1976/1992: 252: Gk. εἰσί, Myc. e‑e‑si, Dor. ἐντί reflect
PIE athem. *h₁senti; Gk. ‑ουσι, Myc. e‑ko‑si, Dor. ‑οντι reflect PIE them.
*‑onti | Sihler 1995: 465–470, 549–550: La. them. ‑unt reflects PIE pri-
mary *‑onti; Sabellic distinguishes primary ‑nt from secondary ‑(n)s; Gk.
εἰσί for *εἱσί is analogical, Myc. e‑e‑si showing the expected form; La. sunt
reflects PIE *h₁sonti | Weiss 2009/2011: 386–387: La. has generalised
PIE primary *‑onti; La. sunt, OCS sǫtъ have arisen through contamination
of athem. PIE *‑enti and them. *‑onti | Krahe 1942/1967: 97–98, 136:
Go. sind goes back to PIE athem. *senti; PIE *‑onti is reflected in Gmc.
*‑anđ(i) (> Go. ‑and, ohg. ‑ant) and *‑anþ(i) (ONor. ‑a) | Boutkan
1995: 308: PIE them. *‑o(?), athem. *‑enti; Gmc. them. endings are based
on reshaped *‑onti | Kloekhorst 2008a: 189–190: Hi. ‑anzi replaces
more original ‑anza, with ‑i from other present endings; Hi. ‑anza repre-
sents athem. PIE *‑énti and them. *‑ónti
PBS
Olander 2009: 194, 197: PBS *‑ˈanti reflects PIE *‑ónti | Kortlandt
1979b/2009: 161–162: them. 3pl. PIE *‑o is preserved in Baltic, merging
with them. 3sg. PIE *‑e → *‑o; Slavic *‑ǫtь contains them. *‑o plus athem.
*‑nti | Endzelīns 1971: 203–204; [see 1923] | Stang 1942: 233–234,
1966: 411–412; [see 3sg., § 4.9] | Otrębski 1956: 183–184; [see 3sg.,
§ 4.9] | Endzelīns 1923: 549–551; [see 3sg., § 4.9]
PS Vondrák 1908/1928: 117–119; 212–219: CS *‑ǫtь reflects PIE *‑onti; OCS
dadętъ is from *‑n̥ ti, jadętъ is from *‑enti | Meillet 1924/1934: 320–322:
‑tъ, ‑∅ have arisen through reduction of ‑tь; athem. sǫtъ reflects PIE *‑onti,
also in La. sunt | Stang 1942: 215–219: Slavic ‑tъ, ‑∅ through reduction
of ‑tь | Vaillant 1966: 12–14: ‑tъ, ‑∅ result from reduction of ‑tь; appar-
ent athem. variant PIE *‑enti is not original | Arumaa 1985: 280–281:
PIE *sonti is preserved in La. sunt and ORu. sutь | Aitzetmüller 1978/
1991: 177–178, 210: OCS ‑tъ has replaced ‑tь through hardening; athem. sǫtъ
has taken over them. ending; soft ‑( j)ǫ‑ for *‑( j)ę‑ < *‑i̯an‑ is analogical
from e-verbs
PIE The athematic endings, accented *‑énti and unaccented *‑n̥ ti, consisted
of the third plural active marker *‑(e)nt‑ followed by the hic et nunc marker *‑i.
In thematic forms the zero grade of the ending was preceded by the o-grade of
the thematic vowel.
362 chapter 4
In Latin sunt the original athematic ending was replaced with the thematic
ending; Umbr. sent shows that the original ending was preserved in Proto-Italic.
PBS PIE athematic *‑n̥ ti, *‑enti yielded PBS *‑inti [2], *‑enti. Since the forms
have disappeared in Baltic and syllable-final *en and *in merge in Slavic by
[28], we cannot know whether both athematic endings or only one of them
survived into Proto-Balto-Slavic and Proto-Slavic; I have presented both end-
ings here. The reflex of PIE thematic *‑onti is PBS *‑ˈanti [7].
In the Baltic languages only one form for the third-person singular, dual and
plural is found in all verbal categories; the only exception is the clearly second-
ary Old Prussian facultative marking of the singular form through the element
‑ts of pronominal origin (see Stang 1966: 410–411 for a possible corresponding
plural form). According to the most widespread view, the Baltic third-person
forms historically reflect the third-person singular, which was generalised in
the dual and plural. Although the motivation for this constitutive development
of Baltic remains unclear, I still consider it the most plausible scenario.
An alternative view regards the Proto-Baltic third plural ending *‑a as the
direct reflex of a Proto-Indo-European primary thematic third plural ending
*‑o (Kortlandt loc. cit.; Beekes loc. cit.). In this scenario the analogical substitu-
tion of the thematic third singular ending *‑e with *‑o > *‑a in Baltic resulted in
a merger of the singular and plural forms, which was subsequently generalised
to the entire verbal system. The obvious drawback of this view is that it requires
the reconstruction of a Proto-Indo-European ending *‑o on the basis of only
one language branch, with very indirect support from other branches (see also
the criticism of Cowgill 1985a/2006: 106; 2006: 556; Villanueva Svensson 2010:
361–362; and cf. Kortlandt 2011b). I also find it implausible that the third sin-
gular ending *‑e would have been replaced analogically with *‑a when this led
to a merger of the third singular and plural forms; one would rather expect
the speakers of the language to try to avoid such merger. Kortlandt’s assump-
tion that the merger started as a phonetic process in i̯e/o-verbs does not sig-
nificantly improve the idea. In fact, this hypothesis amounts to assuming
that the thematic third singular was replaced with the third plural in Baltic, a
view which is considerably less plausible than the more widespread scenario,
according to which the replacement took place in the opposite direction.
PS The reflexes of the athematic endings PBS *‑inti, *‑enti are PS*‑inti, *‑enti
(both corresponding to CS *‑ętь [28|29]). As mentioned above, we cannot
know whether both endings were preserved in Proto-Balto-Slavic and Proto-
Slavic. It is not unlikely that one of the endings had been generalised in pre-
Proto-Slavic. In the third plural of the verb ‘to be’ the thematic ending was
Verbal Inflection 363
PS *u̯ēsin (CS *věsę) PBS *‑in PIE prt., inj. *‑end, *‑n̥d
PIE
Brugmann 1913–1916: 626–637: OCS věsę reflects PIE athem. *‑n̥ t, alongside
*‑e/ont; OCS nesǫ is from PIE them. *‑ont | Beekes 1995/2011: 259–261,
263–264: Slavic ‑ę is from PIE athem. *‑n̥ t, alongside *‑ent; Ved. ‑ur is from
pf.; PIE them. ending was *‑ont | Rix 1976/1992: 244–245: Gk. athem.
‑αν (for *‑α), OAv. ‑at̰ are from PIE *‑n̥ t; Gk. (Hom., Dor.) athem. ἔβαν is
from PIE *egʷh₂ent; Gk. (Hom.) 3sg. ἦεν is from PIE 3pl. *eh₁sent; Gk.
them. ‑ον, Ved. ‑an, OIr. ‑at, OCS ‑ǫ reflect PIE *‑ont | Sihler 1995: 465–
470: Gk. ‑αν is perhaps analogical for *‑α from PIE *‑n̥ t, with ‑ν from them.
verbs or from e.g. ἔβαν < PIE *egʷh₂ent; Dor. 3pl. ἦν, Att., Ion. 3sg. ἦεν, ἦν
reflect PIE 3pl. *eh₁sent, also in Ved. 3pl. ā́san; Gk. ‑ον, OCS ‑ǫ directly
reflect PIE *‑ont | Weiss 2009/2011: 386–387: La. ‑nt reflects primary
ending | Krahe 1942/1967: 103–104: Gmc. ‑un reflects PIE secondary
ending *‑n̥ t | Boutkan 1995: 336–337: PIE *‑n̥ t yielded PGmc. *‑unt (>
Go. ‑un) | Kloekhorst 2008a: 244–245: Hi. ending is ‑er, not ‑ir, and
reflects pf. ending PIE *‑ēr; PIE *‑ent is preserved in Luvian ‑anta
PBS
Olander 2009: – | Kortlandt 1979b/2009: 162: athem. Slavic ‑ę is
from PIE *‑ent; Slavic them. ‑ǫ is from PIE *‑ont | Endzelīns 1971:
– | Stang 1942: –, 1966: – | Otrębski 1956: – | Endzelīns 1923: –
PS Vondrák 1908/1928: 119: Slavic ‑ę, ‑ǫ reflect PIE *‑n̥ t, *‑ont | Meillet
1924/1934: 323–324: OCS ‑ę, Av. ‑at̰, Gk. ‑αν (with analogical ‑ν from ἔφερον)
reflect PIE *‑n̥ t; OCS ‑ǫ, Ved. ‑an, Gk. ‑ον go back to PIE *‑ont | Stang
1942: 213: OCS ‑ǫ corresponds to Gk. ‑ον, Ved. ‑an from PIE *‑nt | Vaillant
1966: 16–17: Slavic ‑ę, OAv. ‑at̰ are from PIE *‑n̥ t, also in Gk. ‑αν with ‑ν from
them. ‑ον; Slavic ‑ǫ, Ved. ‑an, Gk. ‑ον reflect PIE *‑ont | Arumaa 1985:
280–281: Slavic ‑ę, ‑ǫ reflect PIE *‑n̥ t, *‑ont | Aitzetmüller 1978/1991:
179–180
PBS PIE athem. *‑n̥ d and them. *‑ond yielded PBS *‑in [2|5|9] and *‑an
[5|7|9]. The endings were lost in Baltic.
Verbal Inflection 365
PS PBS *‑in, *‑an were retained as PS*‑in (CS *‑ę [28]), *‑an (CS *‑ǫ [28]). As
noted in § 4.2.6 the imperative third-person plural form OCS bǫdǫ historically
contains the thematic secondary ending.
I assume that pre-PS word-final *‑n from PIE *‑n(d) was preserved in Proto-
Slavic, in contrast to the reflex of PIE *‑m, which was lost, e.g. PIE, PBS i-stem
acc. sg. *‑im > PS*‑i [19] and PIE o-stem masc. acc. sg. *‑om > PBS *‑am [7] >
PS*‑u [19] (Olander 2010 with discussion of alternative solutions; see also the
proposal of Villanueva Svensson 2010: 362–363; and cf. Kortlandt 2011b).
The endings were generally preserved in the old Slavic dialects. In Old
Russian the variant in ‑utь is remade on the model of the present form.
Chapter 5
Concluding Remarks
In this study I have set out to reconstruct the inflectional endings of Proto-
Slavic and to discuss their historical background. In chapter 1, the introduc-
tion, I discuss some methodological and terminological issues of importance
for the rest of the study, including the concept of a proto-language, which I
define as the stage of a language just before the earliest innovation not shared
by all of its daughter languages, allowing for some variation. I touch upon the
relationship between Anatolian and non-Anatolian and conclude that in many
cases the reconstructions referred to as Proto-Indo-European in this study are
strictly speaking non-Anatolian Indo-European. Proto-Slavic is defined as the
stage of Slavic spoken immediately before the fronting of non-front vowels
after palatal consonants [20].
In chapter 2, devoted to a discussion of the phonological background of
the study, I present my views on the Proto-Indo-European, Proto-Balto-Slavic
and Proto-Slavic phonological systems. As a consequence of the definition
of Proto-Slavic given in the preceding chapter, the latter system still had oral
diphthongs, and there was no phonologically relevant opposition between
non-palatalised and palatalised consonants. This chapter also contains a rela-
tive chronology of the phonological developments leading from Proto-Indo-
European to Slavic.
The main part of the study comprises chapter 3 on the nominal system and
chapter 4 on the verbal system. For each Proto-Slavic ending I first reconstruct
the Proto-Indo-European form on the basis of the old Indo-European lan-
guages. I then establish the Proto-Balto-Slavic reflex of the ending on the basis
of the Balto-Slavic languages and of the Proto-Indo-European reconstruction.
Finally the Proto-Slavic ending is reconstructed on the basis of the attested
old Slavic languages and of the reconstructed Balto-Slavic form. In each of the
three steps—the reconstruction of the Proto-Indo-European, Proto-Balto-
Slavic and Proto-Slavic endings—I discuss problematic forms in the relevant
languages and point out possible solutions.
I must admit that a few Slavic inflectional forms are not satisfactorily
accounted for in this study. Interestingly, two of the diachronically most dif-
ficult inflectional forms belong to the verbal system. The first one is the present
second-person singular marker PS *‑sī or *‑sē̆i ̯ (CS *‑si), with a variant *‑xī or
*‑xē̆i ̯ (*‑ši), in which the long vowel or diphthong remains unexplained (§ 4.6).
The second one is the first-person plural marker -mo found in Serbo-Croatian,
Slovene, Slovak and Ukrainian, with a vowel that does not seem to be inher-
ited, yet is difficult to explain within Slavic (§ 4.15).
The study differs from other studies treating the same subject due to a com-
bination of several design choices. First, the study focuses on Proto-Slavic in
the strict sense, not on a more loosely defined Common Slavic (see § 1.5.5).
Second, all relevant phonological changes from Proto-Indo-European to Slavic
have been formulated explicitly and ordered chronologically (§ 2.3), and refer-
ence is made to them throughout in the treatment of the individual endings.
Third, the study treats all inflectional endings of Proto-Slavic, not only a subset
(chapters 3 and 4).
These design choices, I hope, have led to a more complete, coherent and
accessible presentation of the prehistory of the Proto-Slavic inflectional sys-
tem than what has previously been made available. On the other hand, they
have also exposed problems that would perhaps not have been revealed oth-
erwise, such as the complex of problems related to the loss of laryngeals [1],
the Common Indo-European vowel contractions [3], the rise of Balto-Slavic
accentual mobility [4] and the delabialisation of PIE *o [7].
Among the most important specific results of the study I shall emphasise
two. A long-standing chronological problem in the development of Slavic final
syllables may be solved if it is assumed that Slavic retained the distinction
between PIE *m and *n after short vowels in final syllalbes, e.g. PIE prt. 1sg.
*‑om > PS *‑u [7|19] (CS *‑ъ [29]) vs. 3pl. *‑ont > PS *‑an [7|9] (CS *‑ǫ [28]) (see
§§ 4.4 and 4.18 as well as Olander 2010).
Another problem, or rather complex of problems, concerns the outcome
of pre-PS *‑ā̆(R)s. Elaborating on and expanding existing ideas by scholars
such as Meillet and Zaliznjak, I have put forward the hypothesis that pre-PS *ā̆
was centralised to *ə̄ ̆ in final syllables closed by an obstruent. This hypothesis
explains a number of controversial forms, including OCS o-stem masc. nom. sg.
‑ъ from PS *‑ə (via [29]) < PBS *‑as (via [17]) < PIE *‑os (via [7]) and OCS ā-stem
nom. pl. ‑y < PS *‑ə̄ (via [29]) < PBS *‑ās (via [17]) < PIE *‑ah₂as (via [1|3]).
Moreover, it also accounts for the corresponding forms in the Old Novgorod
dialect, ‑e and ‑ě, in addition to several endings with a similar structure (see,
among others, §§ 3.3.14, 3.5.4, 3.13.6, 3.13.7, 3.14.4, 3.14.5 and 4.8.2; Olander 2012).
The solutions proposed here would probably not have arisen if the point of
departure had been the traditional Common Slavic system, if the phonologi-
cal developments had not been ordered in a rigorous chronological order, or if
only the apparently relevant endings had been included in the analyses.
Bibliography
In reprinted publications with indication of the original pagination the page numbers
referred to in the main text are the original ones.
Ackermann, Katsiaryna. 2014. Die Vorgeschichte des slavischen Aoristsystems: mit der
kommentierten Belegsammlung der Aoristformen und Formen des präteritalen passi-
ven Partizipiums im Altkirchenslavischen (Brill’s Studies in Indo-European Languages
& Linguistics 10). Leiden & Boston: Brill.
Adams, Douglas Q. 1988. Tocharian historical phonology and morphology (American
Oriental Series 71). New Haven, CT: American Oriental Society.
Aitzetmüller, Rudolf. 1978. Altbulgarische Grammatik als Einführung in die slavische
Sprachwissenschaft (Monumenta Linguae Slavicae dialecti veteris. Fontes et disser-
tationes 12). Freiburg i. Br.: Weiher. (2., verbesserte und erweiterte Auflage
(Monumenta Linguae Slavicae dialecti veteris. Fontes et dissertationes 30), 1991.)
Andersen, Henning. 1968. IE *s after i, u, r, k in Baltic and Slavic. Acta Linguistica
Hafniensia 11. 171–190.
———. 1969. Lenition in Common Slavic. Language 45. 553–574.
———. 1970. On some old Balto-Slavic isoglosses. In Velta Rūķe-Draviņa (ed.), Donum
Balticum. To Professor Christian S. Stang on the occasion of his seventieth birthday 15
March 1970, 14–21. Stockholm: Almqvist & Wiksell.
———. 1980. Morphological change: Towards a typology. In Jacek Fisiak & Werner
Winter (eds.), Historical morphology (Trends in Linguistics. Studies and Monographs
17), 1–50. The Hague, Paris & New York: Mouton.
———. 1985. Protoslavic and Common Slavic: Questions of periodization and termi-
nology. In Michael S. Flier & Dean S. Worth (eds.), Slavic linguistics, poetics, cultural
history: In honor of Henrik Birnbaum on his sixtieth birthday, 13 December 1985
(International Journal of Slavic Linguistics and Poetics 31–32), 67–82. Columbus,
OH: Slavica.
———. 1996. Reconstructing prehistorical dialects: Initial vowels in Slavic and Baltic
(Trends in Linguistics. Studies and Monographs 91). Berlin & New York: Mouton de
Gruyter.
———. 1998a. Slavic. In Paolo Ramat & Anna Giacalone Ramat (eds.), The Indo-
European languages, 415–453. London & New York: Routledge. English version of
“Le lingue slave”, in Le lingue indoeuropee, Bologna: il Mulino, 1993, 441–479.
———. 1998b. The Common Slavic vowel shifts. In American contributions to the
Twelfth International Congress of Slavists, Cracow, Aug.–Sept. 1998. Literature.
Linguistics. Poetics, 239–249. Bloomington, IN: Slavica.
———. 1998c. Диалектная дифференциация общеславянского языка: Парадокс
общих тенденций развития с раз личными локальными результатами. In
370 Bibliography
Robert A. Maguire & Alan Timberlake (eds.), American contributions to the Twelfth
International Congress of Slavists, Cracow, Aug.–Sept. 1998. Literature. Linguistics.
Poetics, 565–600. Bloomington, IN: Slavica.
———. 2006. Synchrony, diachrony, and evolution. In Ole Nedergaard Thomsen (ed.),
Competing models of linguistic change: Evolution and beyond. Vol. 279 (Amsterdam
Studies in the Theory and History of Linguistic Science, Series 4: Current Issues in
Linguistic Theory 279), 59–90. Amsterdam & Philadelphia: Benjamins.
———. 2009a. Noget om analogi. In Rita Therkelsen & Eva Skafte Jensen (eds.),
Dramatikken i grammatikken: Festskrift til Lars Heltoft, 1–10. Roskilde: Institut for
Kultur og Identitet, Roskilde Universitet.
———. 2009b. The satem languages of the Indo-European Northwest. First contacts?
In Angela Marcantonio (ed.), The Indo-European language family: Questions about
its status (Journal of Indo-European Studies, Monograph Series 55(2)), 1–31.
Washington, DC: Institute for the Study of Man.
———. 2012. The new Russian vocative: Synchrony, diachrony, typology. Scando-
Slavica 58. 122–167.
———. 2013. On the origin of the Slavic aspects: Aorist and imperfect. Journal of Slavic
Linguistics 21. 17–43.
———. 2014. Early vowel contraction in Slavic: 1. i-verbs. 2. The imperfect. 3. The vòlja/
súša nouns. Scando-Slavica 60. 54–107.
Anthony, David W. 2007. The horse, the wheel, and language: How Bronze-Age riders
from the Eurasian steppes shaped the modern world. Princeton, NJ, & Oxford:
Princeton University Press.
———. 2013. Two IE phylogenies, three PIE migrations, and four kinds of steppe pas-
toralism. Journal of Language Relationship / Вопросы языкового родства 9. 1–21.
Antonsen, Elmer H. 1994. The earliest attested Germanic language, revisited. North-
Western European Language Evolution 23. 41–68.
Anttila, Raimo. 1972. An introduction to historical and comparative linguistics.
New York: Macmillan; London: Collier-Macmillan.
Arumaa, Peeter. 1933. Untersuchungen zur Geschichte der litauischen Personalpronomina
(Eesti vabariigi Tartu Ülikooli toimetused / Acta et commentationes Universitatis
Tartuenis (Dorpatensis), B. Humaniora 32(2)). Tartu: Mattiesen.
———. 1964. Urslavische Grammatik. Einführung in das vergleichende Studium der sla-
vischen Sprachen. Vol. 1. Einleitung. Lautlehre (I. Teil: Vokalismus, II. Teil: Betonung).
Heidelberg: Winter.
———. 1985. Urslavische Grammatik. Einführung in das vergleichende Studium der sla-
vischen Sprachen. Vol. 3. Formenlehre. Heidelberg: Winter.
Bammesberger, Alfred. 1984. Studien zur Laryngaltheorie (Ergänzungshefte zur
Zeitschrift für vergleichende Sprachforschung 33). Göttingen: Vandenhoeck &
Ruprecht.
Bibliography 371
Bjørnflaten, Jan Ivar. 1990. The birch bark letters redeemed. Russian Linguistics 14.
315–338.
Boutkan, Dirk. 1995. The Germanic “auslautgesetze” (Leiden Studies in Indo-
European 4). Amsterdam & Atlanta: Rodopi.
Bräuer, Herbert. 1961. Slavische Sprachwissenschaft. Vol. 1. Einleitung, Lautlehre
(Sammlung Göschen 1191/1191a). Berlin: de Gruyter.
———. 1969a. Slavische Sprachwissenschaft. Vol. 2. Formenlehre (1. Teil) (Sammlung
Göschen 1192/1192a/1192b). Berlin: de Gruyter.
———. 1969b. Slavische Sprachwissenschaft. Vol. 3. Formenlehre (2. Teil) (Sammlung
Göschen 1236/1236a). Berlin: de Gruyter.
Brosman, Paul. 2002. Evidence in support of “Proto-Indo-Hittite”. Folia Linguistica
Historica 23(1–2). 1–21.
Brugmann, Karl. 1897. Grundriss der vergleichenden Grammatik der indogermanischen
Sprachen. 2. Bearbeitung. Vol. 1. Einleitung und Lautlehre. Strassburg: Trübner.
———. 1906. Grundriss der vergleichenden Grammatik der indogermanischen
Sprachen. 2. Bearbeitung. Vol. 2: Lehre von den Wortformen und ihrem Gebrauch, 1.
Teil. Allgemeines; Zusammensetzung (Komposita); Nominalstämme. Strassburg:
Trübner.
———. 1909–1911. Grundriss der vergleichenden Grammatik der indogermanischen
Sprachen. 2. Bearbeitung. Vol. 2: Lehre von den Wortformen und ihrem Gebrauch, 2.
Teil, 1. Lieferung: Zahlwörter. Die drei Nominalgenera. Kasus- und Numerusbildung
der Pronomina; 2. Lieferung: Bedeutung der Numeri beim Nomen und Pronomen.
Bedeutung der Kasus. Das Adjektivum. Die Adverbia nach Form und Gebrauch. Die
Präpositionen nach Form und Gebrauch. Strassburg: Trübner.
———. 1913–1916. Grundriss der vergleichenden Grammatik der indogermanischen
Sprachen. 2. Bearbeitung. Vol. 2. Lehre von den Wortformen und ihrem Gebrauch, 3.
Teil, 1. Lieferung: Vorbemerkungen. Verbale Komposita. Augment. Reduplizierte
Verbalbildungen. Die Tempusstämme im Allgemeinen. Präsens und starker Aorist.
Die s-Aoriste. Das Perfekt und sein Augmenttempus; 2. Lieferung: Zusammengesetzte
(periphrastische) Tempusbildungen. Die Modusbildungen. Die Personalendungen.
Der Gebrauch der Formen des Verbum finitum. Der Gebrauch der Formen des
Verbum infinitum. Partikeln im einfachen Satz. Strassburg: Trübner.
Buck, Carl Darling. 1955. The Greek dialects. Chicago & London: University of Chicago
Press. (Reprint, 1973.)
Byrd, Andrew. 2010. Reconstructing Indo-European syllabification. PhD dissertation,
University of California, Los Angeles.
Carrasquer Vidal, Miguel. 2011. Syllables, intonations and Auslautgesetze. In Tijmen
Pronk & Rick Derksen (eds.), Accent matters: Papers on Balto-Slavic accentology
(Studies in Slavic and General Linguistics 37), 19–57. Amsterdam & New York:
Rodopi.
Bibliography 373
Corominas, Joan & José E. Pascual. 1980. Diccionario crítico etimológico castellano e his-
pánico. Vol. 1. A–Ca (Biblioteca románica hispánica 5, diccionarios 7). Madrid:
Gredos. (1a reimpressión, 1984.)
Cowgill, Warren. 1965. Evidence in Greek. In Werner Winter (ed.), Evidence for laryn-
geals (Janua linguarum. Series maior 11), 142–180. The Hague: Mouton. (Klein 2006,
137–171.)
———. 1974. Indo-European languages. In The New Encyclopædia Britannica:
Macropædia (15th ed.). Vol. 9, 431–438. Chicago: Encyclopædia Britannica. (Klein
2006, 19–36.)
———. 1975. The origins of the Insular Celtic conjunct and absolute verbal endings.
In Helmut Rix (ed.), Flexion und Wortbildung: Akten der V. Fachtagung der
Indogermanischen Gesellschaft, Regensburg, 9.–14. September 1973, 40–70.
Wiesbaden: Reichert. (Klein 2006, 299–322.)
———. 1979. Anatolian hi-conjugation and Indo-European perfect: Instalment II.
In Erich Neu & Wolfgang Meid (eds.), Hethitisch und Indogermanisch: vergleichende
Studien zur historischen Grammatik und zur dialektgeographischen Stellung der
indogermanischen Sprachgruppe Altkleinasiens (Innsbrucker Beiträge zur
Sprachwissenschaft 25), 25–39. Innsbruck: Institut für Sprachwissenschaft der
Universität Innsbruck. (Klein 2006, 53–67.)
———. 1985a. The personal endings of thematic verbs in Indo-European. In Bernfried
Schlerath & Veronica Rittner (eds.), Grammatische Kategorien, Funktion und
Geschichte. Akten der VII. Fachtagung der Indogermanischen Gesellschaft, Berlin,
20.–25. Februar 1983, 99–108. Wiesbaden: Reichert. (Klein 2006, 69–76.)
———. 1985b. PIE *duu̯ o ‘2’ in Germanic and Celtic, and the nom.–acc. dual of non-
neuter o-stems. Münchener Studien zur Sprachwissenschaft 46. 13–28. (Klein 2006,
433–444.)
———. 2006. The personal endings of thematic verbs in Indo-European. In Klein
2006, 535–567. Longer version of Cowgill 1985a.
Debrunner, Albert & Jacob Wackernagel. 1930. Altindische Grammatik. Vol. 3.
Nominalflexion. Zahlwort. Pronomen. Göttingen: Vandenhoeck & Ruprecht.
(Unveränderter Nachdruck der 1. Auflage von 1929/1930, 1975.)
Derksen, Rick. 2008. Etymological dictionary of the Slavic inherited lexicon
(Leiden Indo-European Etymological Dictionary Series 4). Leiden & Boston: Brill.
———. 2011. The fate of the neuter o-stems in Balto-Slavic. In Tijmen Pronk & Rick
Derksen (eds.), Accent matters: Papers on Balto-Slavic accentology (Studies in Slavic
and General Linguistics 37), 59–66. Amsterdam & New York: Rodopi.
Diels, Paul. 1932–1934. Altkirchenslavische Grammatik. Vol. 1–2. Heidelberg: Winter.
(2. Ausg., 1963.)
Dolobko, Milij Gerasimovič. 1925. Die enklitischen Formen des Pronomens der 1. und
2. Person im Dativus dualis des Urslavischen. Zeitschrift für slavische Philologie 1.
336–442.
Bibliography 375
Dombrowski, Quinn. 2006. Palatalizations in the Old Novgorod dialect: Comparing the-
ory and data. BA thesis, Department of Slavic Languages and Literatures, University
of Chicago.
Dybo, Vladimir Antonovič. 1961. Ударение славянского глагола и формы старосла-
вянского аориста. Краткие сообщения Института славяноведения АН СССР
30. 33–38.
———. 1981. Славянская акцентология: опыт реконструкции системы акцент-
ных парадигм в праславянском. Москва: Наука.
———. 2000. Морфонологизованные парадигматические акцентные системы.
Типология и генезис. Vol. 1. Москва: Языки русской культуры.
Dybo, Vladimir Antonovič, Galina Igorevna Zamjatina & Sergej L’vovič Nikolaev. 1990.
Основы славянской акцентологии. Москва: Наука.
Ebeling, Carl L. 1963. Questions of relative chronology in Common Slavic and Russian
phonology. In Dutch contributions to the Fifth International Congress of Slavicists
(Slavistische drukken en herdrukken 45), 27–42. The Hague: Mouton.
Eichner, Heiner. 1974a. Untersuchungen zur hethitischen Deklination. Inaugural-
Dissertation der Philosophischen Fakultät der Friedrich-Alexander-Universität zu
Erlangen-Nürnberg (Teildruck).
———. 1974b. Zu Etymologie und Flexion von vedisch strī ́ und púmān. Die Sprache 20.
26–42.
———. 1975. Die Vorgeschichte des hethitischen Verbalsystems. In Helmut Rix (ed.),
Flexion und Wortbildung: Akten der V. Fachtagung der Indogermanischen
Gesellschaft, Regensburg, 9.–14. September 1973, 71–103. Wiesbaden: Reichert.
———. 1982. Studien zu den indogermanischen Numeralia: Rekonstruktion des urindo-
germanischen Formensystems und Dokumentation seiner einzelsprachlichen Vertre
tung bei den niederen Kardinalia ‘zwei’ bis ‘fünf’. Habilitationsschrift, Universität
Regensburg.
———. 1985. Das Problem des Ansatzes eines urindogermanischen Numerus
‘Kollektiv’ (‘Komprehensiv’). In Bernfried Schlerath & Veronica Rittner (eds.),
Grammatische Kategorien, Funktion und Geschichte. Akten der VII. Fachtagung der
Indogermanischen Gesellschaft, Berlin, 20.–25. Februar 1983, 134–169. Wiesbaden:
Reichert.
———. 1988. Sprachwandel und Rekonstruktion: Prinzipielles zur indogermanisti-
schen Rekonstruktion. In Christian Zinko (ed.), Akten der 13. Österreichischen
Linguistentagung, Graz 25.–27. Oktober 1985 (Arbeiten aus der Abteilung “Verglei
chende Sprachwissenschaft” Graz 1), 10–40. Graz: Leykam.
———. 1988–1990. Reklameiamben aus Roms Königszeit (Erster Teil). Die Sprache 34.
206–238.
———. 1992. 3. Anatolian. In Jadranka Gvozdanović (ed.), Indo-European numerals
(Trends in Linguistics. Studies and Monograph 57), 29–96. Berlin & New York:
Mouton de Gruyter.
376 Bibliography
———. 1981. Der Auslaut von altkirchenslavisch jestŭ, sǫtŭ. In H. Riggenbach (ed.),
Colloquium Slavicum Basiliense: Gedenkschrift für Hildegard Schroeder (Slavica
Helvetica 16), 145–151. Bern, Frankfurt am Main & Las Vegas, NV: Lang.
Forssman, Berthold. 2001. Lettische Grammatik (Münchener Studien zur Sprachwis
senschaft, Beih. 20). Dettelbach: Röll.
Fortson, Benjamin W., IV. 2004. Indo-European language and culture: An introduction
(Blackwell Textbooks in Linguistics 19). Malden, MA & Oxford: Blackwell.
Fortunatov, Filipp Fëdorovič. 1888. Phonetische Bemerkungen, veranlasst durch
Miklosich’s Etymologisches Wörterbuch der slavischen Sprachen. Archiv für sla
vische Philologie 11. 561–575.
———. 1895. Объ удареніи и долготѣ в балтійскихъ языкахъ. I. Удареніе въ прус-
скомъ языкѣ. Русскій филологическій вѣстникъ 33. 252–297. [non uidi]
———. 1897. Ueber accent und länge in den baltischen sprachen. Beiträge zur Kunde
der indogermanischen Sprachen 22. 153–188. German version of Fortunatov 1895.
———. 1908. Старославянское ‑тъ въ 3-мъ лицѣ глаголовъ. Извѣстія Отдѣленія
русскаго языка и словесности Императорской Академіи Наукъ 13(2). 1–44.
Fox, Anthony. 1995. Linguistic reconstruction: An introduction to theory and method.
Oxford: Oxford University Press.
Fraenkel, Ernst. 1928. Zur baltischen Morphologie. In Symbolae grammaticae in hono-
rem Ioannis Rozwadowski. Vol. 2, 19–25. Cracovia: Gebethner & Wolff.
Fritz, Matthias. 2011. Der Dual im Indogermanischen: Genealogischer und typologischer
Vergleich einer grammatischen Kategorie im Wandel. Heidelberg: Winter.
Gadolina, Margarita Anatol’evna. 1963. История форм личных и возвратного
местоимений в славянских языках. Москва: Издательство Академии наук
СССР.
Gălăbov, Ivan. 1973. Urslavische Auslautprobleme. Wiener slavistisches Jahrbuch 18.
5–17.
Galton, Herbert. 1956. Did sandhi exist in Old Slav? Indogermanische Forschungen 62.
167–176.
Gebauer, Jan. 1896. Historická mluvnice jazyka českého. Vol. 3. Tvarosloví, 1: Skloňování.
Praha & Vídeň: Tempský.
———. 1898. Historická mluvnice jazyka českého. Vol. 3. Tvarosloví, 2: Časování. Praha
& Vídeň: Tempský.
Gorrochategui, Joaquín. 1994. La declinación céltica de los temas en -ā y los datos his-
panos. In Roland Bielmeier & Reinhard Stempel (eds.), Indogermanica et Caucasica:
Festschrift für Karl Horst Schmidt zum 65. Geburtstag (Untersuchungen zur indoger-
manischen Sprach- und Kulturwissenschaft, NF 6), 316–330. Berlin & Boston: de
Gruyter.
Gorškova, Klavdija Vasil’evna & Georgij Aleksandrovič Xaburgaev. 1981. Историческая
грамматика русского языка. Москва: Высшая школа.
378 Bibliography
Griffith, Aaron. 2006. *‑n(C)s in Celtic. Die Sprache 45 [2005] (1–2). 44–67.
Hackstein, Olav. 1993. On the prehistory of dual inflection in the Tocharian verb. Die
Sprache 35. 47–70.
———. 2007. Ablative formations. In Alan J. Nussbaum (ed.), Verba Docenti: Studies
in historical and Indo-European linguistics presented to Jay H. Jasanoff by students,
colleagues, and friends, 131–153. Ann Arbor, MI: Beechstave Press.
———. 2012a. Collective and feminine in Tocharian. In Olav Hackstein & Ronald I.
Kim (eds.), Multilingualism and history of knowledge. Vol. 2. Linguistic developments
along the Silkroad: Archaism and innovation in Tocharian (Österreichische Akademie
der Wissenschaften. Philosophisch-historische Klasse. Sitzungsberichte 834; Irani
sche Onomastik 12), 143–177. Wien: Verlag der Österreichischen Akademie der
Wissenschaften.
———. 2012b. Review of Weiss 2009. Kratylos 57. 109–115.
Hafner, Stanislaus, František Václav Mareš & Manfred Trummer (eds.). 1988.
N.S. Trubetzkoy: Opera Slavica minora linguistica (Sitzungsberichte der Öster
reichischen Akademie der Wissenschaften. Philoso phisch-historische Klasse).
Wien: Verlag der Österreichischen Akademie der Wissenschaften.
Hajnal, Ivo. 1995. Studien zum mykenischen Kasussystem (Untersuchungen zur indo-
germanischen Sprach- und Kulturwissenschaft, Neue Folge 7). Berlin & New York:
de Gruyter.
Halla-aho, Jussi. 2006. Problems of Proto-Slavic historical nominal morphology (Slavica
Helsingiensia 26). Helsinki: Helsinki University Press.
Hamp, Eric P. 1976. On Slavic ev < *eu̯ . Zbornik Matice srpske za filologiju i lingvistiku
19(2). 13–14.
———. 1983. ja = Runic ek. International Journal of Slavic Linguistics and Poetics 27. 11–13.
———. 2011. Indo‑European ‘ego’, Slavic ja = Runic ek, and Celtic Ø. Slavia Centralis
4(1). 5–13.
Hansen, Bjarne Simmelkjær Sandgaard. 2014. The outcome of PIE *‑ē ̆i(̯ C)# and *‑ē ̆u̯(C)#
in Germanic. North-Western European Language Evolution 67. 149–172.
Havlová, Eva & Adolf Erhart (eds.). 1996. Etymologický slovník jazyka staroslověnského.
Vol. 6. klęti – kuditi. Praha: Academia.
Havránek, Bohuslav. 1928. Genera verbi v slovanských jazycích. Vol. 1 (Rozpravy Královské
české společnosti nauk. Třída filosoficko-historicko-jazykozpytná, nová řáda (8) 2).
Praha: Česká společnost nauk.
Heggarty, Paul. 2013. Europe and Western Asia: Indo-European linguistic prehistory.
In Immanuel Ness & Peter Bellwood (eds.), The encyclopedia of global human migra-
tion, 157–167. Oxford: Wiley-Blackwell. (Preprint version.)
Hill, Eugen. 2012. Hidden sound laws in the inflectional morphology of Proto-Indo-
European: A phonological account for the primary first singular of thematic verbs
and the instrumental of thematic nouns and adjectives. In Benedicte Nielsen,
Thomas Olander, Birgit Anette Olsen & Jens Elmegård Rasmussen (eds.), The Sound
Bibliography 379
———. 2014. A note on the Slavic genitive plural. In David Birnbaum, Michael S. Flier
& Cynthia M. Vakareliyska (eds.), Philology broad and deep: In memoriam Horace G.
Lunt, 143–150. Bloomington, IN: Slavica. (Quoted from preprint version.)
Johnson, D.J.L. 1972. The genesis of the Serbo-Croatian genitive plural in ‑ā. The Slavonic
and East European Review 50(120). 333–358.
Kallio, Petri. 2006. On the earliest Slavic loanwords in Finnic. In Juhani Nuorluoto (ed.),
The Slavicization of the Russian North: Mechanisms and chronology / Die Slavisierung
Nordrusslands: Mechanismen und Chronologie / Славянизация Русского Севера:
механизмы и хронология (Slavica Helsingiensia 27), 154–166. Helsinki: Dept. of
Slavonic and Baltic Languages and Literatures.
Kammenhuber, Annelies. 1969. Hethitisch, Palaisch, Luwisch und Hieroglyphen
luwisch. In Bertold Spuler (ed.), Handbuch der Orientalistik. Vol. 1. Abt.: Der nahe
und der mittlere Osten, 2. Bd.: Keilschriftforschung und alte Geschichte Vorderasiens,
1. und 2. Abschn.: Geschichte der Forschung, Sprache und Literatur, Lief. 2:
Altkleinasiatische Sprachen, 119–357. Leiden & Köln: Brill.
Kapović, Mate. 2006a. Reconstruction of Balto-Slavic personal pronouns: With emphasis
on accentuation. PhD dissertation, University of Zagreb.
———. 2006b. The development of Proto-Slavic quantity (from Proto-Slavic to mod-
ern Slavic languages). Wiener slavistisches Jahrbuch 51. 71–111.
———. 2007. The *vòl’ā-type accent in Slavic. In Mate Kapović & Ranko Matasović
(eds.), Tones and Theories: Proceedings of the International Workshop on Balto-Slavic
Accentology, Zagreb, 1–3 July 2005, 89–104. Zagreb: Institut za hrvatski jezik i
jezikoslovlje.
———. 2009. The accent of Slavic *ja(zъ) ‘I’. In Thomas Olander & Jenny Helena
Larsson (eds.), Stressing the past: Papers on Baltic and Slavic accentology (Studies in
Slavic and General Linguistics 35), 53–73. Amsterdam & New York: Rodopi.
———. Forthc. Povijest hrvatske akcentuacije. Fonetika. Zagreb: Matica hrvatska.
Karstien, Hans. 1936. Slavische Instrumentalformen auf -a. Zeitschrift für slavische
Philologie 13. 109–128.
Katz, Joshua T. 1998. Topics in Indo-European personal pronouns. PhD dissertation,
Harvard University.
Kazlauskas, Jonas. 1968. Lietuvių kalbos istorinė gramatika. Vilnius: Mintis.
———. 1970. On the Balto-Slavic dative plural and dual. In Thomas F. Magner &
William R. Schmalstieg (eds.), Baltic linguistics, 87–91. University Park, PA & London:
Pennsylvania State University Press.
Kellens, Jean. 1989. Avestique. In Rüdiger Schmitt (ed.), Compendium linguarum
Iranicarum, 32–55. Wiesbaden: Reichert.
Kessler, Brett & Annukka Lehtonen. 2006. Multilateral comparison and significance
testing of the Indo-Uralic question. In Peter Forster & Colin Renfrew (eds.),
384 Bibliography
———. 1990. The spread of the Indo-Europeans. Journal of Indo-European Studies 18.
131–140. (Kortlandt 2010a, 1–6.)
———. 1991. A note on the Tocharian dual. Tocharian and Indo-European Studies 5.
5–10. (Kortlandt 2010a, 155–157.)
———. 1993. Tokie šalti rytai. Baltistica 28. 45–48. (Kortlandt 2009, 147–149.)
———. 1997. Thematic and athematic verb forms in Old Irish. In Alexander M.
Lubotsky (ed.), Sound law and analogy: Papers in honor of Robert S.P. Beekes on
the occasion of his 60th birthday, 133–137. Amsterdam: Rodopi. (Kortlandt 2007,
107–111.)
———. 1997a. Baltic ē- and ī/jā-stems. Baltistica 32. 157–163. (Kortlandt 2009,
129–135.)
———. 1997b. PIE lengthened grade in Balto-Slavic. In Douglas Q. Adams (ed.),
Festschrift for Eric P. Hamp. Vol. 2 (Journal of Indo-European Studies, Monograph
Series 25), 26–31. Washington, DC: Institute for the Study of Man. (Kortlandt 2009,
61–64.)
———. 1998. The Old Prussian preterit. In A.A. Gippius, L.G. Nevskaja, T.M. Nikolaeva
& T.V. Civ’jan (eds.), Πολύτροπον: К 70-летию Владимира Николаевича Топорова,
144–147. Москва: Индрик. (Kortlandt 2009, 283–285.)
———. 2004. Balto-Slavic accentuation: Some news travels slowly. Baltistica 39. 13–17.
(Kortlandt 2009, 81–84.)
———. 2005. Holger Pedersen’s Études lituaniennes revisited. Baltistica 6 priedas. 151–
157. (Kortlandt 2009, 21–26.)
———. 2006. Balto-Slavic accentual mobility. Baltistica 41. 359–369. (Kortlandt 2009,
93–101.)
———. 2007. Italo-Celtic origins and prehistoric development of the Irish language
(Leiden Studies in Indo-European 11). Amsterdam & New York: Rodopi.
———. 2007. Gothic gen.pl. -e. Historische Sprachforschung 120. 237–240. (Kortlandt
2009, 125–127.)
———. 2008. Balto-Slavic phonological developments. Baltistica 43. 5–15. (Kortlandt
2009, 43–50.)
———. 2009. Baltica & Balto-Slavica (Leiden Studies in Indo-European 16). Amsterdam
& New York: Rodopi.
———. 2009. Accent retraction and tonogenesis. In Thomas Olander & Jenny Helena
Larsson (eds.), Stressing the past (Studies in Slavic and General Linguistics 35),
75–82. Amsterdam & New York: Rodopi. (Kortlandt 2009, 103–109.)
———. 2010a. Studies in Germanic, Indo-European and Indo-Uralic (Leiden Studies in
Indo-European 17). Amsterdam & New York: Rodopi.
———. 2010b. An outline of Proto-Indo-European. In Kortlandt 2010a, 37–45.
———. 2010c. Balto-Slavic accentuation revisited. Wiener slavistisches Jahrbuch 56.
61–81. (Kortlandt 2010a, 341–360.)
Bibliography 387
———. 1997. Selected writings on Indian linguistics and philology (Leiden Studies in
Indo-European 8; Kern Institute miscellanea 2). Amsterdam & Atlanta: Rodopi.
Kul’bakin, Stepan M. 1929. Le vieux slave (Collection de manuels publiée par l’Institut
d’études slaves 5). Paris: H. Champion.
Kümmel, Martin Joachim. 1997. Review of Beekes 1995. Philologia Fenno-Ugrica 2–3.
113–125.
———. 2000. Das Perfekt im Indoiranischen: eine Untersuchung der Form und Funktion
einer ererbten Kategorie des Verbums und ihrer Weiterentwicklung in den altindoira-
nischen Sprachen. Wiesbaden: Reichert.
———. 2007. Konsonantenwandel: Bausteine zu einer Typologie des Lautwandels und
ihre Konsequenzen für die vergleichende Rekonstruktion. Wiesbaden: Reichert.
———. 2013a. Zur Endung des Genitivs Plural im Indoiranischen. Indogermanische
Forschungen 118. 193–211.
———. 2013b. The Iranian reflexes of Proto-Iranian *ns. Orientalia Suecana 61.
138–145.
———. Forthc. Verwandte des Indogermanischen? Zur Frage des ‘Eurasiatischen’
und anderer Makrofamilien. https://www.academia.edu/342156/Verwandte_des_
Indogermanischen_Zur_Frage_des_Eurasiatischen_und_anderer_Makrofamilien.
Kuryłowicz, Jerzy. 1927. Les effets du ə en indoiranien. Prace filologiczne 11. 201–243.
———. 1952. L’accentuation des langues indo-européennes (Polska akademia umiejęt-
ności. Prace Komisji językowej 37). Kraków: Polska Akademia Umiejętności. (2e
édition, Polska akademia nauk. Komitet językoznawcze. Prace językoznawcze 17,
Wrocław & Kraków: Zakład Narodowy im. Ossolińskich, 1958.)
———. 1964. The inflectional categories of Indo-European. Heidelberg: Winter.
Kuznecov, Petr Savvič. 1956. К истории форм 3-го лица настоящего времени гла-
гола в русском языке. Slavia. Časopis pro slovanskou filologii 25(2). 175–183.
———. 1961. Очерки по морфологии праславянского языка. Москва: Академия
наук СССР.
Kwon, Kyongjoon. 2009. The early development of animacy in Novgorod: Evoking the
vocative anew. In Vit Bubenik (ed.), Grammatical change in Indo-European lan-
guages (Current Issues in Linguistic Theory 305), 43–53. Amsterdam & Philadelphia:
Benjamins.
Lamprecht, Arnošt. 1987. Praslovanština (Spisy University v Brně. Filosofická fakulta,
266). Brno: Univerzita J.E. Purkyně.
Lane, George Sherman. 1963. Bimoric and trimoric vowels and diphthongs: laws of
Germanic finals again. Journal of English and Germanic Philology 62. 155–170.
Laroche, Emmanuel. 1965. Études de linguistique anatolienne. Revue hittite et asia-
nique 23. 33–54.
Le Feuvre, Claire. 1998a. Études linguistiques sur les documents de Novgorod: leur
apport à la grammaire comparée des langues slaves. Revue des études slaves 70(1).
241–248.
Bibliography 389
———. 1998b. Études linguistiques sur les documents de Novgorod: leur apport à la
grammaire comparée des langues slaves. Thèse de doctorat, École pratique des
hautes études, IVe Section (Sciences historiques et philologiques).
———. 2007. 1. Vieux russe. LALIES: Actes des sessions de linguistique et de littérature
27. 5–112.
———. 2011. Mécanismes de réaffectation désinentielle et hiérarchie des oppositions
casuelles en slave. In Michèle Fruyt, Michel Mazoyer & Dennis Pardee (eds.),
Grammatical case in the languages of the Middle East and Europe: Acts of the
International colloquium Variations, concurrence et evolution des cas dans divers
domaines linguistiques, Paris, 2–4 april 2007 (Studies in Ancient Oriental Civilization
64), 345–358. Chicago: The Oriental Institute of the University of Chicago.
Lejeune, Michel. 1969. Essai de philologie mycénienne. XI. L’instrumental pluriel thé-
matique. Revue de philologie, de littérature et d’histoire anciennes 42. 219–229.
(Lejeune 1972, 253–266.)
———. 1972. Mémoires de philologie mycénienne. Vol. 3. 1964–1968 (Incunabula Graeca
43). Paris: Edizioni dell’Ateneo Roma.
Lekov, Ivan. 1934. Праславянскитѣ глаголни форми и отраженията имъ въ днѣш-
ните славянски езици (Списание на българската академия на наукитѣ 50.
Клонъ историко-филологиченъ и философско-общественъ 24). София:
Придворна печатница. [non uidi]
Leskien, August. 1876. Die Declination im Slavisch-Litauischen und Germanischen.
Leipzig: S. Hirzel.
———. 1909. Grammatik der altbulgarischen (altkirchenslavischen) Sprache.
Heidelberg: Winter. (2. und 3. Auflage, 1919.)
Leumann, Manu. 1926–1928. Lateinische Grammatik. Vol. 1. Lateinische Laut- und
Formenlehre (Handbuch der Altertumswissenschaft, Abt. 2, Teil 2, Band 1).
München: Beck. (Neuausgabe, 1977.)
Lewis, Henry & Holger Pedersen. 1937. A concise comparative Celtic grammar.
Göttingen: Vandenhoeck & Ruprecht. (3rd ed., 2nd impression with the supplement
of 1961 by Henry Lewis, 1989.)
Liddell, Henry George, Robert Scott, Henry Stuart Jones & Roderick McKenzie. 1843. A
Greek–English lexicon. Oxford: Clarendon. (9. ed., with a revised supplement,
1996.)
Lindstedt, Jouko. 1991. The notational fallacy in CS phonology. Scando-Slavica 37.
108–122.
Ljapunov, Boris Mixajlovič. 1905. Формы склоненiя въ старословянскомъ языкѣ.
Vol. 1. Склоненiе именъ. Одесса: “Экономическая” типографiя.
Loma, Aleksandar. 2013. Топономија бањске хрисовуље: Ка осмишљењу старо-
српског топономастичког речника и бољем познавању општесловенских
именославних образаца (Библиотека ономатолошки прилога 2). Београд:
Српска академија наука и уметности.
390 Bibliography
Issues in linguistics: Papers in honor of Henry and Renée Kahane, 644–657. Urbana,
IL, Chicago & London: University of Illinois Press.
Mareš, František Václav. 1962a. The Slavic verbal forms of the 3rd person plural bǫdǫ
and bǫ. International Journal of Slavic Linguistics and Poetics 5. 28–30.
———. 1962b. Ранний период морфологического развития славянского склоне-
ния (общая характеристика). Вопросы языкознания 1962(6). 13–21.
———. 1966. The Proto-Slavic and Early Slavic declension system. Travaux linguis-
tiques de Prague 1. 163–172.
———. 1969. Diachronische Phonologie des Ur- und Frühslavischen (Slavistische
Beiträge 40). München. (Mareš 1999, 17–100.)
———. 1978. Das slavische Konjugationssystem des Präsens in diachroner Sicht.
Wiener slavistisches Jahrbuch 24. 175–209.
———. 1986. Vom Urslavischen zum Kirchenslavischen. In Peter Rehder (ed.),
Einführung in die slavischen Sprachen, 1–19. Darmstadt.
———. 1993. Význam staroslověnských rukopisů nově objevených na hoře Sinaj. K hla-
holským rukopisům 3/N a 4/N. Slavia 62. 125–130.
———. 1999. Diachronische Phonologie des Ur- und Frühslavischen (Schriften über
Sprachen und Texte 4). Frankfurt am Main: Lang.
Martínez, Javier & Michiel de Vaan. 2001. Introducción al avéstico. Madrid: Ediciones
Clásicas.
———. 2014. Introduction to Avestan. English translation of Martínez & de Vaan 2001
by Ryan Sandell (Brill Introductions to Indo-European Languages 1). Leiden &
Boston: Brill.
Maslov, Jurij Sergeevič. 1981. Грамматика болгарского языка. Москва: Высшая
школа.
———. 1982. Граматика на българския език. Bulgarian translation of Maslov 1981 by
Blažo Blažev. София: Наука и изкуство.
Matasović, Ranko. 2004. The Proto-Indo-European syllabic resonants in Balto-Slavic.
Indogermanische Forschungen 109. 337–354.
———. 2005. Toward a relative chronology of the earliest Baltic and Slavic sound
changes. Baltistica 40. 147–157.
Mathiassen, Terje. 1989. Nochmals der Akk. Pl. der ā-Stämme im Ostbaltischen.
Baltistica 25. 123–125.
Matthews, W.K. 1960. Russian historical grammar. London: Athlone Press.
Matzinger, Joachim. 1997. Zu armenisch mek‘ ‘wir’. Historische Sprachforschung 110.
82–92.
———. 2001. Die ‘m-Kasus’ des Balto-Slawischen und Germanischen. In Heiner
Eichner, Peter-Arnold Mumm, Oswald Panagl & Eberhard Winkler (eds.), Fremd
und eigen: Untersuchungen zu Grammatik und Wortschatz des Uralischen und
Indogermanischen in memoriam Hartmut Katz, 183–208. Wien: Edition Praesens.
392 Bibliography
———. 1950. Urslavische Grammatik. Einführung in das vergleichende Studium der sla-
vischen Sprachen. Vol. 3. Formenlehre. Heidelberg: Winter.
Milewski, Tadeusz. 1932. Rozwój fonetyczny wygłosu prasłowiańskiego. Slavia 11. 1–32,
225–264.
Miller, Raymond H. 1988. The third person present tense and Common Slavic dialectol-
ogy. International Journal of Slavic Linguistics and Poetics 37. 7–33.
Morpurgo Davies, Anna. 1998. History of linguistics. Vol. 4: Nineteenth-century linguis-
tics. London & New York: Longman.
Moszyński, Leszek. 1972. О времени монофтонгизации праславянских дифтонгов.
Вопросы языкознания 1972(4). 53–67.
Mühlenbach, Karl. 1903. О слѣдахъ двойственнаго числа въ латышскомъ языкѣ.
Извѣстія Отдѣленія русскаго языка и словесности Императорской Академіи
Наукъ 8(1). 7–80.
Narten, Johanna (ed.). 1975–1976. Karl Hoffmann: Aufsätze zur Indoiranistik. Vol. 1–2.
Wiesbaden: Reichert.
Neu, Erich. 1979. Einige Überlegungen zu den hethitischen Kasusendungen. In Erich
Neu & Wolfgang Meid (eds.), Hethitisch und Indogermanisch: vergleichende Studien
zur historischen Grammatik und zur dialektgeographischen Stellung der indo
germanischen Sprachgruppe Altkleinasiens (Innsbrucker Beiträge zur Sprachwissen
schaft 25), 177–196. Innsbruck: Institut für Sprachwissenschaft der Universität
Innsbruck.
———. 1991. Etruskisch—eine indogermanische Sprache Altanatoliens? Historische
Sprachforschung 104. 9–28.
Nichols, Johanna. 1993. The linguistics geography of the Slavic expansion. In Robert A.
Maguire & Alan Timberlake (eds.), American contributions to the Eleventh
International Congress of Slavists, 377–391. Columbus, OH: Slavica.
Nielsen, Hans Frede. 2000. The early runic language of Scandinavia: Studies in Germanic
dialect geography. Heidelberg: Winter.
Nielsen Whitehead, Benedicte, Thomas Olander, Birgit Anette Olsen & Jens Elmegård
Rasmussen (eds.). 2012. The sound of Indo-European: Phonetics, phonemics, and mor-
phophonemics (Copenhagen Studies in Indo-European 4). Copenhagen: Museum
Tusculanum.
Nussbaum, Alan J. 1986. Head and horn in Indo-European (Untersuchungen zur indo-
germanischen Sprach- und Kulturwissenschaft 2). Berlin: de Gruyter.
Obnorskij, Sergej Petrovič. 1953. Очерки по морфологии русского глагола. Москва:
Академия наук СССР.
Oettinger, Norbert. 1979. Die Stammbildung des hethitischen Verbums (Erlanger Beiträge
zur Sprach- und Kunstwissenschaft 64). Nürnberg: Carl.
———. 1988. Der indogermanische Nominativ Dual aus laryngalistischer Sicht. In
Alfred Bammesberger (ed.), Die Laryngaltheorie und die Rekonstruktion des indoger-
manischen Laut- und Formensystems, 355–359. Heidelberg: Winter.
Bibliography 395
Olander, Thomas. 2005. The dative plural in Old Latvian and Proto-Indo-European.
Indogermanische Forschungen 110. 273–281.
———. 2006. Accentual mobility: The prehistory of the Balto-Slavic mobile accent para-
digms. PhD dissertation, University of Copenhagen.
———. 2007a. Once more on desinential accent in Balto-Slavic mobile paradigms.
Baltu filoloģija 16(1–2). 81–85.
———. 2007b. The Balto-Slavic mobile accent paradigms. In Mate Kapović & Ranko
Matasović (eds.), Tones and Theories: Proceedings of the International Workshop on
Balto-Slavic Accentology, Zagreb, 1–3 July 2005, 1–14. Zagreb: Institut za hrvatski jezik
i jezikoslovlje.
———. 2007c. The accentuation of Greek monosyllabic words. In Coulter George,
Matthew McCullagh, Benedicte Nielsen, Antonia Ruppel & Olga Tribulato (eds.),
Greek and Latin from an Indo-European perspective, 1–8. Cambridge: Cambridge
University Press.
———. 2009. Balto-Slavic accentual mobility (Trends in Linguistics. Studies and
Monographs 199). Berlin & New York: Mouton de Gruyter.
———. 2010. Proto-Indo-European final nasals in Slavic. Scando-Slavica 56. 84–98.
———. 2012. Proto-Indo-European *‑os in Slavic. Russian Linguistics 36. 319–341.
Oliver, Lisi (ed.). 1994. Calvert Watkins: Selected writings. Vol. 1. Language and linguis-
tics. (Innsbrucker Beiträge zur Sprachwissenschaft 80). Innsbruck: Institut für
Sprachwissenschaft der Universität Innsbruck.
Olsen, Birgit Anette. 1988. On the interpretation of four Vedic verbs: irajyáti and irad-
hanta—badhnā́ti and ubhnā́ti. Arbejdspapirer udsendt af Institut for Lingvistik,
Københavns Universitet 7. 85–100.
———. 1999. The noun in Biblical Armenian (Trends in Linguistics. Studies and
Monographs 119). Berlin & New York: Mouton de Gruyter.
———. 2001. Verb or noun? On the origin of the third person in IE. In Martin E. Huld,
Karlene Jones-Bley, Angela Della Volpe & Miriam Robbins Dexter (eds.), Proceedings
of the Twelfth Annual UCLA Indo-European Conference (Journal of Indo-European
Studies, Monograph Series 40), 65–79. Washington, DC: Institute for the Study of
Man.
———. 2004. The complex of nasal stems in Indo-European. In James Clackson &
Birgit Anette Olsen (eds.), Indo-European word formation (Copenhagen Studies in
Indo-European 2), 215–248. Copenhagen: Museum Tusculanum.
———. 2009. The conditioning of laryngeal breaking in Greek. In Rosemarie Lühr &
Sabine Ziegler (eds.), Protolanguage and prehistory: Akten der XII. Fachtagung der
Indogermanischen Gesellschaft, vom 11. bis 15. Oktober 2004 in Krakau, 348–365.
Wiesbaden: Reichert.
———. Forthc. An outline of Indo-European nominal derivation.
Orr, Robert. 1984. The locative singular of the consonant stems in Slavic: A new
approach. Canadian Slavonic Papers 26(2–3). 201–206.
396 Bibliography
———. 1988. A phantom sound-change: CS *-ŎM# > *-Ŭ#. Canadian Slavonic Papers
30. 41–61.
———. 2000. Common Slavic nominal morphology: A new synthesis. Bloomington, IN:
Slavica.
Oslon, Michail V. 2010. Review of Olander 2009. Вопросы языкознания 2010(2).
141–146.
Osthoff, Hermann. 1878. Kleine beiträge zur declinationslehre der indogermanischen
sprachen, I. In Hermann Osthoff & Karl Brugmann (eds.), Morphologische
Untersuchungen auf dem Gebiete der indogermanischen Sprachen. Vol. 1, 207–290.
Leipzig: Hirzel.
Otrębski, Jan. 1956. Gramatyka języka litewskiego. Vol. 3. Nauka o formach. Warszawa:
Państwowe Wydawnictwo Naukowe.
Pedersen, Holger. 1905a. Die nasalpräsentia und der slavische akzent. Zeitschrift für ver-
gleichende Sprachforschung 38 (Neue Folge 18). 297–421.
———. 1905b. Zur armenischen sprachgeschichte. Zeitschrift für vergleichende
Sprachforschung 38. 194–240.
———. 1924. Sprogvidenskaben i det 19. aarhundrede: metoder og resultater. København:
Gyldendal.
———. 1931. Linguistic science in the nineteenth century: Methods and results. English
translation of Pedersen 1924 by John Webster Spargo. Bloomington, IN: Indiana
University Press. (Midland Book Edition, by arrangement with Harvard University
Press, 1962.)
———. 1933. Zur Frage nach der Urverwandtschaft des Indoeuropäischen mit dem
Ugrofinnischen. In Liber Semisaecularis Societatis Fenno-Ugricae, 308–325. Helsinki:
Suomalais-Ugrilainen Seura.
———. 1935. Lit. iau. Studi Baltici 4. 150–154.
———. 1938. Hittitisch und die anderen indoeuropäischen Sprachen (Det Kgl. Danske
Videnskabernes Selskab, Historisk-filologiske Meddelelser 25(2)). København:
Levin & Munksgaard.
Penzl, Herbert. 1972. Methods of comparative Germanic linguistics. In Frans van
Coetsem & Herbert L. Kufner (eds.), Toward a grammar of Proto-Germanic, 1–42.
Tübingen: Niemeyer.
Petit, Daniel. 2010. Untersuchungen zu den baltischen Sprachen (Brill’s Studies in Indo-
European Languages & Linguistics 4). Leiden & Boston: Brill.
———. 2011. Review of Olander 2009. Historische Sprachforschung 123. 318–321.
———. 2013. Review of Olander 2009. Kratylos 58. 151–157.
Peyrot, Michaël. 2013. The Tocharian subjunctive: A study in syntax and verbal stem for-
mation (Brill’s Studies in Indo-European Languages and Linguistics 8). Leiden &
Boston: Brill.
Pinault, Georges-Jean. 1989. Introduction au tokharien. LALIES. Actes des sessions de
linguistique et de littérature 7. 5–224.
Bibliography 397
Rieken, Elisabeth. 1994. Der Wechsel -a- / -i- in der Stammbildung des hethitischen
Nomens. Historische Sprachforschung 107. 42–53.
———. 2008. The origin of the ‑l genitive and the history of the stems in ‑īl‑ and ‑ūl‑ in
Hittite. In Karlene Jones-Bley, Martin E. Huld, Angela Della Volpe & Miriam Robbins
Dexter (eds.), Proceedings of the 19th Annual UCLA Indo-European Conference, Los
Angeles, November 3–4, 2007 (selected papers) (Journal of Indo-European Studies
Monograph Series 54), 239–256. Washington, DC: Institute for the Study of Man.
———. 2009. Der Archaismus des Hethitischen: eine Bestandsaufnahme. Incontri
Linguistici 32. 37–52.
Ringe, Donald A. 1991. Evidence for the position of Tocharian in the Indo-European
family? Die Sprache 34. 59–123.
———. 2006a. A linguistic history of English. Vol. 1. From Proto-Indo-European to
Proto-Germanic. Oxford & New York: Oxford University Press.
———. 2006b. A sociolinguistically informed solution to an old historical problem:
The Gothic genitive plural. Transactions of the Philological Society 104(2). 167–206.
———. 2009. The Proto-Indo-European verb system. Handout. http://www
.ling.upenn.edu/~kroch/courses/lx310/ringe-handouts-09/pie-pgmc-vb.pdf.
Ringe, Donald A. & Joseph F. Eska. 2013. Historical linguistics: Toward a twenty-first
century reintegration. Cambridge: Cambridge University Press.
Ringe, Donald A., Tandy Warnow & Ann Taylor. 2002. Indo-European and computa-
tional cladistics. Transactions of the Philological Society 100. 59–129.
Rinkevičius, Vytautas. 2010. Review of Olander 2009. Baltistica 45. 369–379.
Rix, Helmut. 1976. Historische Grammatik des Griechischen. Laut- und Formenlehre.
Darmstadt: Wissenschaftliche Buchgesellschaft. (2., korrigierte Auflage, 1992.)
———. 1977. Das keltische Verbalsystem auf dem Hintergrund des indo-iranisch-
griechischen Rekonstruktionsmodells. In Karl Horst Schmidt (ed.), Indogermanisch
und Keltisch: Kolloquium der Indogermanischen Gesellschaft am 16. und 17. Februar
1976 in Bonn, 132–158. Wiesbaden: Reichert.
Rix, Helmut, Martin Kümmel, Thomas Zehnder, Reiner Lipp & Brigitte Schirmer (eds.).
1998. Lexikon der indogermanischen Verben. Wiesbaden: Reichert. (2., erw. und verb.
Aufl., 2001.)
Rohlfs, Gerhard. 1949. Historische Grammatik der italienischen Sprache und ihrer
Mundarten. Vol. 1. Lautlehre. Bern: Francke.
———. 1966. Grammatica storica della lingua italiana e dei suoi dialetti. Vol. 1.
Fonetica. Italian translation of Rohlfs 1949 by Salvatore Persichino. Torino: Giulio
Einaudi.
Ryko, Anastasija Igorevna. 2000. Семантическое распределение окончаний 3-го
лица презенса в северо-западных русских говорах. In Балто-славянские иссле-
дования 1998–1999, 114–133. Москва: Индрик.
400 Bibliography
Sabaliauskas, A. 1976. Dėl lie. formos nuodu ‘mudu’. Baltistica 12. 167.
Šaxmatov, Aleksej Aleksandrovič. 1915. Очеркъ древнѣйшаго перiода исторiи рус-
скаго языка (Энциклопедiя славянской филологiи 11(1)). Петроградъ: Отдѣленіе
русскаго языка и словесности Императорской Академіи Наукъ. (Репринтное
издание, Москва: Индрик, 2002.)
Schelesniker, Herbert. 1964. Beiträge zur historischen Kasusentwicklung des Slavischen
(Wiener Slavistisches Jahrbuch, Ergänzungsband 5). Köln & Graz: Hermann
Böhlaus Nachf.
Schenker, Alexander M. 1993. Proto-Slavonic. In Bernard Comrie & Greville G. Corbett
(eds.), The Slavonic languages, 60–121. London & New York: Routledge. (Paperback
edition, 2002.)
Scherer, Wilhelm. 1868. Zur Geschichte der deutschen Sprache. Berlin: Duncker.
Schindler, Jochem. 1964. Einige indogermanisch-uralische Wortgleichungen. Die
Sprache 10. 171–173.
———. 1974. Fragen zum paradigmatischen Ausgleich. Die Sprache 20. 1–9.
———. 1977. Notizen zum Sieversschen Gesetz. Die Sprache 23. 56–65.
———. 1989. Altindische Nominalmorphologie. Vorlesungen, WS 1988/1989. [Wien].
Schlerath, Bernfried. 1981. Ist ein Raum/Zeit-Modell für eine rekonstruierte Sprache
möglich? Zeitschrift für vergleichende Sprachforschung 95. 175–202.
Schmalstieg, William R. 1961. Primitive East Baltic *‑uo‑, *‑ie‑ and the 2nd sg. ending.
Lingua 10. 369–374.
———. 1965. Slavic o- and ā-stem accusatives. Word 21. 238–243.
Schmidt, Gernot. 1978. Stammbildung und Flexion der indogermanischen Personal
pronomina. Wiesbaden: Harrassowitz.
Schmitt-Brandt, Robert. 1998. Einführung in die Indogermanistik. München: Francke.
Schmitt, Rüdiger. 1981. Grammatik des Klassisch-Armenischen mit sprachvergleichen-
den Erläuterungen (Innsbrucker Beiträge zur Sprachwissenschaft 32). Innsbruck:
Institut für Sprachwissenschaft der Universität Innsbruck. (2., durchgesehene
Auflage, Innsbrucker Beiträge zur Sprachwissenschaft 123, 2007.)
Schwyzer, Eduard. 1939. Griechische Grammatik. Vol. 1. Allgemeiner Teil. Lautlehre.
Wortbildung. Flexion. München: Beck. (Vierte, unveränderte Auflage, 1968.)
Seebold, Elmar. 1984. Das System der Personalpronomina in den frühgermanischen
Sprachen: Sein Aufbau und seine Herkunft (Ergänzungshefte zur Zeitschrift für ver-
gleichende Sprachforschung 34). Göttingen: Vandenhoeck & Ruprecht.
Senn, Alfred. 1929. Kleine litauische Sprachlehre. Heidelberg: Groos. (Nachdruck,
Hildesheim: Georg Olms, 1974.)
———. 1935. Die athematischen Verba in Mikalojus Dauksza’s Schriften. Studi Baltici
4. 86–122.
Seržant, Ilja A. 2004. Zur Vorgeschichte des Inessivs im Urostbaltischen. Acta Linguistica
Lithuanica 51. 59–67.
Bibliography 401
Svane, Gunnar. 1958. Die Flexionen in štokavischen Texten aus dem Zeitraum 1350–1400.
Aarhus: Universitetsforlaget i Aarhus.
Szemerényi, Oswald. 1967. The perfect participle active in Mycenaean and Indo-
European. Studi micenei ed egeo-anatolici 2. 7–26. (Considine & Hooker 1987/1991,
1253–1272.)
———. 1970. Einführung in die vergleichende Sprachwissenschaft. Darmstadt:
Wissenschaftliche Buchgesellschaft. (4., durchgesehene Auflage, 1990.)
———. 1973. Marked–unmarked and a problem of Latin diachrony. Transactions of the
Philological Society 72. 55–74. (Considine & Hooker 1987/1991, 925–944.)
———. 1996. Introduction to Indo-European linguistics. Translation of Szemerényi
1970/1990, with additional notes and references. Oxford: Oxford University Press.
(Paperback ed., 1999.)
Tedesco, Paul. 1951. Review of Mikkola 1950. Language 27. 165–177.
Tichy, Eva. 2000. Indogermanistisches Grundwissen für Studierende sprachwissenschaft-
licher Disziplinen. Bremen: Hempen.
———. 2006. A survey of Proto-Indo-European. English translation of Tichy 2000.
Bremen: Hempen.
Toporov, Vladimir Nikolaevič. 1961. К вопросу об эволюции славянского и балтий-
ского глагола. Вопросы славянского языкознания 5. 35–70.
Torbiörnsson, Tore. 1921. En fornpolsk nybildning. In Uppsala Universitets Årsskrift 1921:
Filosofi, Språkvetenskap och Historiska Vetenskaper. Vol. 5. Språkvetenskapliga sälls-
kapets i Uppsala förhandlingar jan. 1919–dec. 1921, 45–62. Uppsala: Akademiska
bokhandeln.
———. 1923a. Die altbulgarische Umbildung der Partizipialformen. Slavia 1. 208–214.
———. 1923b. Eine altpolnische Neubildung. Archiv für slavische Philologie 38.
120–127.
Townsend, Charles E. 2007. Review of Igartua 2005. Journal of Slavic Linguistics 15.
171–173.
Trautmann, Reinhold. 1910. Die altpreussischen Sprachdenkmäler. Göttingen: Vanden
hoeck & Ruprecht.
Trávníček, František. 1935. Historická mluvnice československá: Úvod, hláskosloví a
tvarosloví (Řada spisů duchovědných 2). Praha: Melantrich.
Trubačёv, Oleg Nikolaevič (ed.). 1985. Этимологический словарь славянских языков:
праславянский лексический фонд. Vol. 12. *koulъkъ—*kroma/*kromъ. Москва:
Наука.
Trubeckoj, Nikolaj Sergeevič. 1922. Essai sur la chronologie de certains faits phoné-
tiques du slave commun. Revue des études slaves 2. 217–234. (Hafner, Mareš and
Trummer 1988, 37–54.)
Ul’janov, Grigorij Konstantinovič. 1888. Основы настоящаго времени въ старо-
славянскомъ и литовскомъ языкахъ. Варшава: Въ типографiи Марiи Земкевичъ.
Bibliography 403
———. 2000. On the status of the earliest Russian isogloss: Four untenable and three
questionable reasons for separating the progressive and the second regressive pala-
talization of Common Slavic. Russian Linguistics 24. 5–29.
———. 2003a. Comedy of errors or inexorable advance? Exploring the dysfunctional-
ity of the debate about the progressive palatalization of Slavic. In Jos Schaeken,
Hubrecht Peter Houtzagers & Janneke Kalsbeek (eds.), Dutch Contributions to the
Thirteenth International Congress of Slavists, Ljubljana, 2003, Linguistics (Studies in
Slavic and General Linguistics 30), 397–452. Amsterdam & New York: Rodopi.
———. 2003b. Leading ideas in the study of the progressive palatalization of Proto-
Slavic. International Journal of Slavic Linguistics and Poetics 44–45. 379–396.
———. 2006. Jagić’s chronology of the progressive palatalization. In Per Ambrosiani,
Irina Lysén, Elisabeth Löfstrand & Johan Muskala (eds.), jáko bl[a]gopěsnívajà ptíca:
Hyllningsskrift till Lars Steensland (Acta Universitatis Stockholmiensis 32), 229–240.
Stockholm: Stockholms universitet.
———. 2008a. Pedersen’s chronology of the progressive palatalization. In Peter
Houtzagers, Janneke Kalsbeek & Jos Schaeken (eds.), Dutch Contributions to the
Fourteenth International Congress of Slavists, Ohrid: Linguistics (Studies in Slavic
and General Linguistics 34), 503–571. Amsterdam & New York: Rodopi.
———. 2008b. Review of Halla-aho 2006. Scando-Slavica 54. 288–294.
———. 2014. Early Slavic dialect differences involving the consonant system. In Egbert
Fortuin, Peter Houtzagers, Janneke Kalsbeek & Simeon Dekker (eds.), Dutch contri-
butions to the Fifteenth International Congress of Slavists, Minsk: Linguistics (Studies
in Slavic and General Linguistics 40), 181–227. Amsterdam & New York: Rodopi.
Vijūnas, Aurelius. 2013. Review of Nielsen Whitehead et al. 2012. Kratylos 58. 85–99.
Villanueva Svensson, Miguel. 2010. Review of Kortlandt 2009. Baltistica 45. 359–379.
———. 2011. Indo-European long vowels in Balto-Slavic. Baltistica 46. 5–38.
———. 2013. Review of Nielsen Whitehead et al. 2012. Baltistica 48. 337–342.
Villar, Francisco. 1995. A new interpretation of Celtiberian grammar (Innsbrucker
Beiträge zur Sprachwissenschaft. Vorträge und Kleinere Schriften 62). Innsbruck:
Institut für Sprachwissenschaft der Universität Innsbruck.
———. 1906. Vergleichende slavische Grammatik. Vol. 1. Lautlehre und
Stammbildungslehre. Göttingen: Vandenhoeck & Ruprecht.
Vondrák, Václav. 1906. Vergleichende slavische Grammatik. Vol. 1. Lautlehre und
Stammbildungslehre. Göttingen: Vandenhoeck & Ruprecht. (Zweite stark ver-
mehrte und verbesserte Auflauge, 1924.)
———. 1908. Vergleichende slavische Grammatik. Vol. 2. Formenlehre und Syntax.
Göttingen: Vandenhoeck und Ruprecht.
———. 1908. Vergleichende slavische Grammatik. Vol. 2. Formenlehre und Syntax.
Göttingen: Vandenhoeck & Ruprecht. (Zweite Auflage, neubearbeitet von Dr.
O. Grünenthal, 1928.)
Bibliography 405
Page numbers for definitions are in boldface. Common Slavic shortening of long
diphthongs 118, 170. See also
accent 3–4, 9, 12–13, 18, 33–34, 40, 42, 43, 48, monophthongisation of oral
49, 49–50, 54, 68, 69, 106–107. See also diphthongs [22]
Dybo’s law and deglottalisation [13]; Common Slavic vowel contractions [25] 64,
mobility law [4]; prosody; Saussure’s 65, 92, 93, 226, 298, 317
law comparative method 1, 13–15, 19
accent-ablaut paradigms 68, 259, 260 Core Indo-European. See non-Anatolian
acrostatic 68 Indo-European
ad hoc hypotheses 57, 91, 138, 326, 333
amphikinetic 68 deaspiration of voiced aspirated stops.
Anatolian (vs. non-Anatolian Indo-European) See Winter’s law and deaspiration of
5, 8–9, 21–23, 74, 89, 112–113, 241, 296, voiced aspirated stops [6]
366 deglottalisation and Dybo’s law. See Dybo’s
apocope 106, 197, 214–215, 218, 221, 268, law and deglottalisation [13]
308, 328 delabialisation of *o to *a [7] 41–42, 48, 49,
assibilation of palatal stops [8] 41, 51, 51–52, 50, 51, 56, 58, 60, 61, 89, 90, 91, 92, 93,
52, 54, 110 97, 103, 104, 106, 110, 119, 135, 138, 140,
assimilation 57, 143, 239 157, 165, 178, 198, 206, 208, 209, 212,
Auslautgesetze (laws of final syllables) 221, 233, 257, 259, 261, 266, 271, 273,
10–11, 37 274, 276, 278, 294, 310, 317, 323, 332,
333, 337, 338, 341, 353, 359, 362, 364,
backing of *ē to *ā after palatalised 365, 367
consonants [27] 57, 60, 61, 62, 66, 100, delabialisation of *ō to *ā [15] 43, 49, 55, 57,
111, 118, 136, 159, 161, 165, 197, 219, 236, 67, 90, 151, 168, 197, 204, 251, 262, 286,
276, 284, 293, 308, 321, 336 308
backing of *e to *a before *u̯ [11] 53, 57, 63, depalatalisation 329, 330
65, 143, 148, 154, 155, 174, 183, 229, 261 devī ́ type 73, 101, 165, 184
Baltic linguistic unity 24, 25, 27, 59 devoicing of word-final obstruents [5] 40,
Balto-Slavic linguistic unity 24–25, 27 48, 50, 51, 58, 87, 91, 92, 93, 106, 108, 110,
Brugmann’s law 188, 220, 345 136, 332, 335, 336, 337, 364
diphthongisation of *ē̆ to *i̯ā̆ before
centralisation of *ā̆ to *ə̄ ̆ before word-final tautosyllabic *u̯ [10] 52–53, 53 57,
fricatives. See loss of word-final 128, 174, 183
fricatives, with centralisation of diphthongisation of syllabic sonorants [2]
preceding *ā̆ to *ə̄ ̆ [17] 48, 58, 86, 87, 90, 114, 117, 242, 248, 362,
collective formation 222, 233 364
Common Indo-European vowel contractions Dybo’s law and deglottalisation [13] 43, 54,
[3] 47, 48, 49, 50, 51, 55, 90, 117, 135, 55, 56, 81, 100, 102, 110, 113, 131, 148, 161,
136, 150, 151, 165, 167, 195, 206, 208, 211, 162, 165, 168, 170, 176, 191, 192, 193, 195,
212, 231, 248, 259, 261, 262, 264, 367 197, 200, 201, 204, 215, 217, 218, 219, 221,
Common Slavic 1, 4, 7, 29–31, 44–45, 59–60, 224, 228, 236, 240, 259, 261, 266, 276,
367. See also Proto-Slavic 278, 280, 281, 282, 284, 289, 291, 293,
Common Slavic shortening of final long 295, 308, 312, 321, 327, 336, 338, 342, 345,
vowels 43–44, 332 353, 360, 363
index 409
final long vowels, Common Slavic shortening. Kiparsky’s law (metathesis of word-final *i
See Common Slavic shortening of final and preceding dental; Gk.) 315
long vowels
final syllables, laws of. See Auslautgesetze labialisation of *i̯ to *u̯ after *ō [14] 49, 55,
first palatalisation of velars [21] 27, 28, 57, 151, 286
42, 43, 53, 54, 60–61, 62, 65, 66, 168, 186, laryngeal colouring 14, 38, 39–40, 184
316 laryngeals, loss. See loss of laryngeals [1]
forma difficilior 14, 15, 89 laws of final syllables. See Auslautgesetze
fronting of non-front vowels after palatal Leskien’s law 100, 113, 117, 131, 188, 215, 236,
consonants [20] 27–29, 31, 33, 42, 46, 245, 279, 281, 284, 304, 341, 353
56, 57, 59, 60–62, 62, 63, 64, 65, 66, 67, loanword relations (Slavic) 11, 15, 26, 28, 29,
73, 75, 76, 90, 91, 92, 93, 97, 100, 103, 106, 30, 31, 42, 44, 45, 46, 62, 81
118, 119, 131, 136, 150, 151, 159, 161, 165, 176, long diphthongs, Common Slavic shortening.
178, 180, 184, 186, 195, 197, 199, 211, 212, See Common Slavic shortening of long
219, 221, 234, 236, 248, 251, 259, 262, 265, diphthongs
266, 276, 278, 284, 286, 293, 295, 299, loss of laryngeals [1] 41, 42, 47–48, 48, 49,
308, 323, 338, 342, 345, 353, 355, 360, 50, 51, 56, 81, 90, 100, 101, 110, 113, 117, 131,
363, 366 136, 150, 165, 167, 176, 190, 192, 193, 195,
197, 198, 200, 201, 204, 206, 208, 209, 211,
glottalisation 41, 43, 48, 50, 54, 195, 242, 244, 212, 214, 217, 218, 219, 221, 228, 231, 236,
251. See also prosody 240, 248, 257, 259, 261, 262, 264, 276,
“Graeco-Aryan” model 296 279, 281, 282, 284, 293, 306, 308, 312, 321,
323, 336, 337, 338, 341, 353, 359, 367
haplology 146, 147 loss of *n between a high vowel and
Herkunfthypothese 22 word-final *s [16] 55, 56, 242, 244,
hic et nunc marker *-i (PIE) 304, 314, 326, 246
361 loss of word-final dentals after long vowels,
*hom (PIE particle) 109–110, 111, 112, 122, 123, with raising of the vowel [18] 46, 47,
154, 201, 237, 239 57, 82, 84
hysterokinetic 9, 68 loss of word-final fricatives, with
centralisation of preceding *ā̆ to *ə̄ ̆
*i (PIE hic et nunc marker). See hic et nunc [17] 13, 43, 47, 48, 51, 53, 54, 55, 56–57,
marker *-i (PIE) 57, 60, 61, 63, 64, 67, 68, 81, 90, 91, 92,
Indo-European homeland 23 95, 97, 98, 99, 103–104, 125, 126, 127, 128,
Indo-Uralic hypothesis 19 131–133, 206, 222, 226, 229, 231, 234, 242,
infinitive (PS) 172, 297, 302 244, 246, 248, 251, 254, 271, 272, 273, 274,
inflection (definition) 15–18 276, 278, 280, 281, 282, 284, 286, 319,
informal speech 330, 333 320, 321, 323, 331, 332, 336, 338, 348, 353,
injunctive (PIE) 296, 297, 302, 327 354–355, 367
intermediate proto-language 19, 20–21, 25 loss of word-final *m after short vowels, with
internal reconstruction 7, 8, 13–15, 26, 35, rounding and raising of preceding *a
130, 136, 167, 187, 196–197, 198, 203, 304, to *u [19] 47, 51, 52, 54, 57, 58–59, 61,
308, 315 86, 110, 114, 115, 116, 118, 119, 122, 123, 170,
iotation. See elimination of post-consonantal 180, 258, 259, 261, 262, 265, 266, 308,
*i̯ [26] 309, 310, 365, 367
410 index
loss of word-final stops [9] 47, 48, 50, 51, 52, palatalisation of velars. See first palatalisation
58, 87, 89, 91, 92, 93, 97, 106, 108, 136, of velars [21]; second palatalisation of
319, 330, 331, 332, 333, 335, 336, 337, 364, velars [23]
367 palatal stops, assibilation. See assibilation of
palatal stops [8]
macroparadigm 16, 17 paradigm (definition) 16–18
mesostatic 68 paragogic vowel 329–330, 333
metathesis of liquids 29–30, 44, 60, 62, 65, perfect (PIE) 4, 296, 297, 304–305, 306, 317,
66 332, 341, 364
metathesis of word-final *i and preceding permissive (Li.) 48, 337–338
dental (Greek). See Kiparsky’s law phonotactic rules 27, 28, 46, 61, 115, 178, 312
métatonie douce in monosyllabic words 240. post-consonantal *i̯, elimination. See
See also monosyllabic lengthening elimination of post-consonantal *i̯
middle voice (PIE) 296, 297, 306, 317 [26]
mobility law [4] 24, 42, 49, 49–50, 50, 51, 55, proportional analogy 97, 104, 131–132, 133,
56, 81, 98, 99, 103, 106, 108, 111, 115, 116, 170–171, 250, 290, 304
117, 119, 125, 127, 128, 130, 136, 147, 150, prosody 3–4, 12, 16, 18, 24, 34, 41, 47–48, 49,
151, 165, 167, 176, 178, 190, 195, 197, 198, 54, 111, 186. See also accent; glottalisa-
226, 229, 231, 233–234, 242, 244, 245, tion; quantity; tone
248, 250, 259, 261, 262, 264, 271, 285, proterokinetic 68
288, 320, 335, 338, 367 Proto-Baltic See Baltic linguistic unity
monophthongisation of nasal diphthongs Proto-Balto-Slavic See Balto-Slavic
[28] 54, 57, 58, 60, 61, 63, 65, 66, 67, linguistic unity
86, 87, 90, 91, 92, 93, 118, 123, 131, 159, Proto-Indo-European word-final voicing 40,
161, 165, 170, 248, 251, 308, 362, 363, 50
365, 367 proto-language 6, 18–21, 27, 59–60
monophthongisation of oral diphthongs Proto-Slavic 1, 7, 19, 25–31, 42–45, 68–80.
[22] 27, 28, 29, 43, 46, 48, 49, 52, 53, See also Common Slavic
55, 56, 57, 61, 62, 62–63, 63, 66, 68, 76,
126, 127, 128, 145, 147, 148, 150, 151, 157, quantity 3, 41, 43–44, 66–67, 79. See also
170, 172, 174, 176, 178, 181, 183, 186, 195, prosody
199, 206, 207, 208, 209, 211, 212, 234, 266,
288, 295, 306, 312, 316, 317, 323, 338, 342, raising and rounding of *a to *u before
345, 353, 360 word-final *m. See loss of word-final
monosyllabic lengthening 122–123. See also *m after short vowels, with rounding
métatonie douce in monosyllabic words and raising of preceding *a to *u [19]
raising of *e to *i before *i̯ [24] 64–65, 161,
nasal diphthongs, monophthongisation. 208, 226, 259, 317–318
See monophthongisation of nasal raising of long vowels before word-final
diphthongs [28] dentals. See loss of word-final dentals
neutral speech 330, 333 after long vowels, with raising of the
non-Anatolian Indo-European 8–9, 19, vowel [18]
21–24, 47, 74, 89, 112–113, 138, 241, 269, reinterpretation of vowel quantity as quality
296, 366 [29] 28, 31, 43, 48, 49, 50, 51, 52, 53, 54,
55, 56, 57, 58, 60, 61, 63, 64, 65, 66,
optative (PIE) 48, 56, 296, 297, 301, 302, 312, 66–67, 68, 76, 81, 82, 84, 90, 91, 92, 95,
321, 322, 323, 336, 337, 341, 353, 359 97, 98, 99, 100, 102, 103, 106, 108, 111, 113,
oral diphthongs, monophthongisation. See 114, 115, 116, 118, 119, 125, 126, 127, 128, 131,
monophthongisation of oral 136, 138, 140, 145, 147, 148, 150, 151, 157,
diphthongs 159, 161, 162, 165, 168, 170, 172, 174, 176,
index 411
178, 180, 181, 183, 184, 186, 189, 191, 192, Teeter’s law 10, 21
193, 195, 197, 199, 200, 201, 204, 206, 207, thematic imperfective preterite (“imperfect”)
208, 209, 211, 212, 215, 217, 218, 219, 221, (PIE) 297, 301
222, 224, 226, 228, 229, 231, 234, 236, third palatalisation of velars. See second
238, 240, 242, 244, 246, 248, 251, 254, palatalisation of velars [23]
258, 259, 261, 262, 265, 266, 271, 272, 273, tone 3, 12–13, 18, 40, 49, 79. See also prosody
274, 276, 278, 280, 281, 282, 284, 286,
288, 289, 291, 293, 295, 305, 306, 308, umlaut, progressive. See fronting of non-front
310, 312, 316, 317, 320, 321, 323, 327, 332, vowels after palatal consonants [20]
333, 335, 336, 338, 342, 345, 348, 353,
360, 362, 363, 367 velars, palatalisation. See first palatalisation
resultative aspect (PIE). See perfect (PIE) of velars [21]; second palatalisation of
root aorist (PIE) 300, 331, 334. See also root velars [23]
preterite vocative (status as paradigmatic form) 18,
root preterite (PIE) 319, 331. See also root 69
aorist voiced aspirated stops, deaspiration.
rounding and raising of *a to *u before See Winter’s law and deaspiration of
word-final *m. See loss of word-final voiced aspirated stops [6]
*m after short vowels, with rounding vowel contractions, Common Indo-European.
and raising of preceding *a to *u [19] See Common Indo-European vowel
ruki change [12] 41, 43, 48, 51, 53–54, 55, 56, contractions [3]
62, 81, 98, 99, 103, 126, 127, 128, 206, 234, vowel contractions, Common Slavic. See
240, 266, 280, 281, 282, 284, 286, 289, Common Slavic vowel contractions
291, 295, 318, 323 [25]
sandhi 11–12, 117, 172, 178, 214, 333 Winter’s law and deaspiration of voiced
Saussure’s law 117, 131, 190, 242, 245, aspirated stops [6] 41, 42, 50, 50–51,
279–280, 281, 304 51, 52, 53, 110, 155
Schwundhypothese 22 word formation 4, 15
second palatalisation of velars [23] 27, 28, word-final fricatives, loss. See loss of
28–29, 42, 46, 61, 62, 63, 63–64, 67, 133, word-final fricatives, with centralisa-
167, 178–179, 234, 295, 323 tion of preceding *ā̆ to *ə̄ ̆ [17]
shortening of final long vowels in Common word-final *m, loss after short vowels.
Slavic. See Common Slavic shortening See loss of word-final *m after short
of final long vowels vowels, with rounding and raising of
shortening of long diphthongs in Common preceding *a to *u [19]
Slavic. See Common Slavic shortening word-final obstruents, devoicing. See
of long diphthongs devoicing of word-final obstruents [5]
sigmatic aorist (PIE) 297, 310, 319, 331, 332 word-final stops, loss. See loss of word-final
sporadic (irregular) sound change and stops [9]
variation 21, 24, 25, 27, 42, 111, 122–123, word-final voicing in Proto-Indo-European.
138, 270, 309, 322, 328, 329, 330 See Proto-Indo-European word-final
stem 16, 16–18 voicing
subjunctive (PIE) 296, 297, 302
subparadigm 16, 32, 71, 73, 75 yodisation. See elimination of post-
supine (PS) 116, 297, 302 consonantal *i̯ [26]
syllabic sonorants, diphthongisation.
See diphthongisation of syllabic
sonorants [2]
syllabification rules (PIE) 40–41, 115