THE SLAVIC DUAL: MORPHOSYNTACTIC FEATURE ECONOMY AS A by Tatyana Slobodchikoff

THE SLAVIC DUAL: MORPHOSYNTACTIC FEATURE ECONOMY AS A by Tatyana Slobodchikoff
THE SLAVIC DUAL: MORPHOSYNTACTIC FEATURE ECONOMY AS A
FACTOR IN LANGUAGE CHANGE
by
Tatyana Slobodchikoff
______________________
© Tatyana Slobodchikoff 2013
A Dissertation Submitted to the Faculty of the
Department of Linguistics
In Partial Fulfillment of the Requirements
For the Degree of
DOCTOR OF PHILOSOPHY
In the Graduate College
THE UNIVERSITY OF ARIZONA
2013
2
THE UNIVERSITY OF ARIZONA
GRADUATE COLLEGE
As members of the Dissertation Committee, we certify that we have read the dissertation
prepared by Tatyana Slobodchikoff
entitled The Slavic Dual: Morphosyntactic Feature Economy as a Factor in Language
Change
and recommend that it be accepted as fulfilling the dissertation requirement for the
Degree of Doctor of Philosophy
_______________________________________________________________________
Dr. Heidi Harley
Date: 5/02/2012
_______________________________________________________________________
Dr. Michael Hammond
Date: 5/02/2012
_______________________________________________________________________
Dr. Simin Karimi
Date: 5/02/2012
_______________________________________________________________________
Dr. Elly van Gelderen
Date: 5/02/2012
Final approval and acceptance of this dissertation is contingent upon the candidate’s
submission of the final copies of the dissertation to the Graduate College.
I hereby certify that I have read this dissertation prepared under my direction and
recommend that it be accepted as fulfilling the dissertation requirement.
________________________________________________ Date: 5/02/2012
Dissertation Director: Dr. Heidi Harley
3
STATEMENT BY AUTHOR
This dissertation has been submitted in partial fulfillment of requirements for an
advanced degree at the University of Arizona and is deposited in the University Library
to be made available to borrowers under rules of the Library.
Brief quotations from this dissertation are allowable without special permission, provided
that accurate acknowledgment of source is made. Requests for permission for extended
quotation from or reproduction of this manuscript in whole or in part may be granted by
the copyright holder.
SIGNED: Tatyana Slobodchikoff
4
ACKNOWLEDGEMENTS
I owe the inspiration for writing of this dissertation to my parents, Ljubov'
Aleksandrovna and Gennadij Fedorovič Stepanenko, who instilled in me the love for
knowledge from my early childhood. My late father was the greatest inspiration to me.
He always had the deepest appreciation for Russian literature, Russian culture, and the
grammatical structure of Old Church Slavic. A Russian philologist by training, he was
always excited when he would find an interesting Old Church Slavic suffix or a
grammatical rule. I think it was he who instilled in me natural curiosity and love for
languages.
I would like to express my gratitude for constant support from my dissertation
advisor, Heidi Harley, who helped me during the writing process of this dissertation.
Heidi has a brilliant theoretical mind. I would always walk away with new insights
whenever I met with her. Heidi would come up with a theoretical insight that would lead
me to discover new and interesting aspects of the Slavic dual. She was a great support in
my duel with the Slavic dual.
Mike Hammond, my dissertation committee member on the "Russian side," was
always happy to meet and help me with whatever questions I had. From the very
beginning of the writing process, Mike gave me an invaluable piece of advice, "Open that
dissertation file every day!" It was simple, but it really worked. I made sure I opened my
dissertation file almost every day. Mike's global thinking always inspired me, and I
would leave his office reenergized and ready to write more meaningful prose.
I owe a world of gratitude to my long term mentor and dissertation committee
member, Elly van Gelderen, who inspired me to become a linguist in the first place.
Elly's enthusiasm about linguistics was mind-blowing to me. She would always respond
immediately, and when I say it, I really mean it. She would respond within minutes to my
numerous questions about economy principles, features, corpora, etc. Elly's advice was
always very encouraging and very helpful. I think that because of her constant
encouragement I was able to successfully finish this dissertation.
I would also like to thank Simin Karimi who gave me very detailed comments on
all of my dissertation drafts. Simin read my chapters very carefully and would often
compliment me on my writing style. Simin was a great support when it came to some
details which I left unclear in my dissertation. She would always stress the importance of
clarity in my writing. This was very helpful to me.
I would like to express my enormous gratitude to Dr.Viktor Baranov, the director
of Project Manuscript and Professor at the Iževsk State Technical University for his help
with providing access to the diachronic digital corpora of Old Russian and Old Church
Slavic texts. Dr.Baranov was always very prompt responding to my numerous questions
about the the intrecasies of the Project Manuscript corpus. I am also very grateful to Dr.
Roumyana Pancheva for her help and assistance with the University of Southern
California parsed corpus of Old South Slavic manuscripts.
Last but not least, I would like to express my enormous gratitude to my family,
my husband, Misha, and my son, Kolja, who kept me on track by showing that there is a
life outside of my dissertation. Misha was always ready to listen how morphosyntactic
5
features split and fuse, and what defines [±singular] as opposed to [±augmented]. I think
that he still remembers the definitions. He was very patient and very encouraging. Kolya
is my greatest joy in life. He would always put a smile on my face whenever I had a bad
day.
6
DEDICATION
To my parents, Gennadij Fedorovič and Ljubov' Aleksandrovna Stepanenko, with love
and gratitude.
7
TABLE OF CONTENTS
LIST OF TABLES ............................................................................................................ 10
LIST OF FIGURES .......................................................................................................... 13
ABSTRACT ...................................................................................................................... 14
CHAPTER 1 INTRODUCTION ...................................................................................... 15
1.0. The Goals of the Dissertation ................................................................................. 15
1.1. The Problem of the Slavic Dual ............................................................................. 19
1.2. The Hypotheses ...................................................................................................... 25
1.3. The Proposal: A Principle of Morphosyntactic Feature Economy ........................ 27
1.4. Theoretical Assumptions ........................................................................................ 29
1.5. Methodology and Data ........................................................................................... 31
1.6. The Structure of the Dissertation ........................................................................... 37
CHAPTER 2 DUAL NUMBER IN THE WORLD’S LANGUAGES FROM A
TYPOLOGICAL PERSPECTIVE.................................................................................... 40
2.0. Introduction ............................................................................................................ 40
2.1. Humboldt's Typology of Dual Number .................................................................. 41
2.2. Jespersen’s Typology of the Dual .......................................................................... 53
2.3. Plank's Typology of the Dual: Humboldt Revisited .............................................. 57
2.4. Corbett's Constraint on Ranges of the Dual and Plural .......................................... 73
2.5. Cysouw: Dual as a Restricted Group ..................................................................... 80
2.6. The Typology of the Slavic Dual ........................................................................... 85
2.7. Summary ................................................................................................................ 91
CHAPTER 3 DERIVATION OF THE SLAVIC DUAL IN DISTRIBUTED
MORPHOLOGY .............................................................................................................. 93
3.0. Introduction ............................................................................................................ 93
3.1. The Architecture of Grammar in Distributed Morphology .................................... 95
3.2. Derivations of the Slavic Dual in DM .................................................................. 102
3.3. Impoverishment, Fission, and Morphosyntactic Simplification of the Slavic Dual ..
................................................................................................................... 124
8
TABLE OF CONTENTS - Continued
3.4. Summary .............................................................................................................. 130
CHAPTER 4 THE TWO NUMBER FEATURES OF THE SLAVIC DUAL AND THE
PATTERNS OF CROSS-LINGUISTIC VARIATION IN SOUTH, WEST, AND EAST
SLAVIC .......................................................................................................................... 131
4.0. Introduction .......................................................................................................... 131
4.1. The Two Number Features of the Slavic Dual ..................................................... 133
4.2. The Two Patterns of Variation: The Bimorphemic and Monomorphemic Duals ....
................................................................................................................... 144
4.2.1. The Bimorphemic Dual in Contemporary Standard Slovenian....................... 145
4.2.2. The Bimorphemic Dual in Contemporary Upper Sorbian ................................ 148
4.2.3. The Bimorphemic Dual in Contemporary Lower Sorbian ................................ 151
4.2.4. The Distribution of the [-singular] and [-augmented] Features in the
Bimorphemic Dual in Slovenian, Upper, and Lower Sorbian ................................... 153
4.2.5. The Monomorphemic Dual and Its Loss in Kashubian ..................................... 158
4.2.6. The Monomorphemic Dual and Its Loss in Old Russian .................................. 164
4.3. Summary .............................................................................................................. 180
CHAPTER 5 THE PRINCIPLE OF MORPHOSYNTACTIC FEATURE ECONOMY
AND THE DIACHRONIC CHANGES IN THE SLAVIC DUAL................................ 182
5.0. Introduction .......................................................................................................... 182
5.1. Morphosyntactic Markedness of Feature Values of the Dual .............................. 184
5.2. The Proposal: The Principle of Morphosyntactic Feature Economy ................... 188
5.3. The Principle of Morphosyntactic Feature Economy and Diachronic Change .........
................................................................................................................... 192
5.4. The Reanalysis of the Old Slovenian Dual as a More Economical Dual in
Contemporary Standard Slovenian.............................................................................. 197
9
TABLE OF CONTENTS - Continued
5.5. The Reanalysis of the Old Sorbian Dual as a More Economical Dual in
Contemporary Upper and Lower Sorbian ................................................................... 203
5.6. The Reanalysis of the 19th-Century Kashuabin Dual as Plural in Contemporary
Kashubian ................................................................................................................... 211
5.7. The Reanalysis of the Old Russian Dual as Plural in Contemporary Standard
Russian ................................................................................................................... 215
5.7.1. The Reanalysis of the 2nd Person Dual as Plural in Old Russian................... 215
5.7.2. The Reanalysis of the 1st Person Dual as Plural in Old Russian .................... 220
5.8. Summary .............................................................................................................. 225
CHAPTER 6: CONCLUSION ....................................................................................... 226
6.1. The Evolution of Dual Number and the Principle of Morphosyntactic Feature
Economy ................................................................................................................... 226
6.2. Theoretical Implications of the Morphosyntactic Feature Economy ................... 228
APPENDIX A ABBREVIATIONS ................................................................................ 231
APPENDIX B ORTHOGRAPHICAL SYSTEMS AND TRANSLITERATION
SYMBOLS ...................................................................................................................... 233
REFERENCES ............................................................................................................... 237
10
LIST OF TABLES
Table 2.1 Humboldt's Notional Typology of Dual Number .......................................................... 45
Table 2.2 Humboldt's Extensional Typology of Dual.................................................................... 46
Table 2.3 Humboldt's Notional, Extensional, and Conceptual Typologies of the Dual ................ 50
Table 2.4 Humboldt's Conceptual Typology of the Dual .............................................................. 51
Table 2.5 Jespersen's Typology of the Dual (1965:205) ................................................................ 53
Table 2.6 Plank's Typology of Languages with Dual Number (1989:299) ................................... 58
Table 2.7 Language Examples for Plank's Typology (1989:299) .................................................. 58
Table 2.8 Language Change in Humboldt's Typology (1827) ....................................................... 59
Table 2.9 Language Change in Plank's Typology (1989) .............................................................. 60
Table 2.10 Diachronic Typology of the Dual in Slavic Languages ............................................... 61
Table 2.11 Plank's Typology of Pronominal Dual Based on Person Restriction (1989:301) ........ 63
Table 2.12 Languages in Plank's Pronominal Typology (1989:301) ............................................. 64
Table 2.13 Typology of the Pronominal Dual in Slavic ................................................................ 65
Table 2.14 Plank's Verbal Agreement Typology (1989:306) ........................................................ 69
Table 2.15 Languages in Plank's Verbal Agreement Typology (1989:306) .................................. 69
Table 2.16 Personal Pronouns and Verbal Agreement in Slavic ................................................... 71
Table 2.17 Dual Pronouns in Slovenian (Jakop 2008:59).............................................................. 77
Table 2.18 Typology of the Slavic Dual (Belič 1899) ................................................................... 85
Table 4.1 Dual/Plural Syncretism in Contemporary Standard Slovenian .................................... 134
Table 4.2 Navajo Personal Pronouns (Young & Morgan 1980:22) ............................................. 134
Table 4.3 Old Russian Personal Pronouns in the Ostromir Gospel (1056-1057) (Corpus
Manuscript) .................................................................................................................................. 141
11
LIST OF TABLES - Continued
Table 4.4 Present Tense Verbal Suffixes in Old Russian (Možejko and Ignatenko 1988:141) ... 142
Table 4.5 The Distribution of Bimorphemic and Monomorphemic Dual Pronouns ................... 144
Table 4.6 Bimorphemic Dual Pronouns in Contemporary Standard Slovenian (Derganc
2003:166) ..................................................................................................................................... 146
Table 4.7 Present Tense Agreement Suffixes in Contemporary Standard Slovenian (Derganc
2003:167) ..................................................................................................................................... 147
Table 4.8 Bimorphemic Dual Pronouns in Upper Sorbian (Schaarschmidt 2002:2426) ............. 149
Table 4.9 Present Tense Agreement Suffixes in Upper Sorbian (Stone 1993:645) ..................... 150
Table 4.10 Bimorphemic Dual Pronouns in Lower Sorbian (Stone 1993:622) ........................... 152
Table 4.11 Present Tense Verbal Suffixes in Lower Sorbian (Stone 1993:647) ......................... 153
Table 4.12 Dual/Plural Pronominal Stem Syncretism in the 16th Century Slovenian ................ 155
Table 4.13 Dual/Plural Pronominal Stem Syncretism in the 16th Century Sorbian .................... 156
Table 4.14 The Distribution of the [-singular] and [-augmented] features in the Bi morphemic
Dual Pronouns in Slovenian, Upper & Lower Sorbian ................................................................ 158
Table 4.15 The Dual Pronouns in Kashubian (1860s) (Duličenko & Lehfeldt 1998:58-59) ....... 159
Table 4.16 Dual and Plural Nominative Pronouns in Kashubian (1860s) (Duličenko & Lehfeldt
1998:58-59) .................................................................................................................................. 160
Table 4.17 Kashubian Personal Pronouns (Stone 1993:773) ....................................................... 160
Table 4.18 Kashubian Personal Pronouns (Hopkins 2001:36) .................................................... 161
Table 4.19 The Evolution of the Kashubian Dual Pronouns ....................................................... 162
Table 4.20 Present Tense Verbal Suffixes in Kashubian (Hopkins 2001:36) .............................. 162
Table 4.21 Proto-Slavic Personal Pronouns (Sussex and Cubberley 2006) ................................. 165
12
LIST OF TABLES - Continued
Table 4.22 Old Russian Personal Pronouns in the Ostromir Gospel (1056-1057) (Corpus
Manuscript www.manuscripts.ru) ................................................................................................ 166
Table 4.23 The Morphological Structure of the Old Russian Dual Pronouns ............................. 167
Table 4.24 The Evolution of the 2nd Person Dual Pronoun vy in Old Russian .......................... 174
Table 4.25 The Evolution of the 1st Person Dual Pronoun vě in Old Russian ............................ 179
Table 5.1 Dual Pronouns in Contemporary Standard Slovenian ................................................. 202
Table 5.2 Bimorphemic Nominative Dual Pronouns in Contemporary Upper and Lower Sorbian
..................................................................................................................................................... 209
Table 5.3 The Evolution of the Kashubian Dual Pronouns ......................................................... 211
13
LIST OF FIGURES
Figure 1.1 Diachronic Changes in the Slavic Languages .............................................................. 21
Figure 2.1 Universal Hierarchy of Person Noyer (1997:114) ........................................................ 67
Figure 2.2 The Animacy Hierarchy (Corbett 2000:56).................................................................. 74
Figure 2.3 Greenberg's Number Hierarchy .................................................................................... 74
Figure 2.4 The 1st Pattern of Ranges of Dual and Plural in Language X ...................................... 75
Figure 2.5 The 2nd Pattern of Ranges of Dual and Plural in Language X .................................... 75
Figure 2.6 An Impossible Pattern of Dual and Plural in Language X ........................................... 75
Figure 2.7 Typology of Pronominal Number Marking (Cysouw 2009:205) ................................. 81
Figure 2.8 Cysouw (2009) vs. Traditional Number Markedness in the Pronominal ..................... 81
Figure 2.9 Lower Sorbian Dual-Unified-We Paradigm Cysouw 2009:206) ................................. 84
Figure 2.10 Markedness of the Dual (Žolobov 2001) .................................................................... 89
Figure 3.1 The Architecture of Grammar in DM (Harley and Noyer 1999) .................................. 96
Figure 4.1 Lattice-Theoretic Approach to Nominal Predicates (Harbour 2008:67) .................... 136
Figure 4.2 Lattice-Theoretic Semantics of the [±singular] Feature (Harbour 2008:69) .............. 138
Figure 4.3 Lattice-Theoretic Semantics of the [±augmented] Feature (Harbour 2008:69)......... 139
14
ABSTRACT
Dual number marked on personal pronouns and verbal agreement suffixes disappeared in
the majority of Slavic languages except for three - Slovenian, Upper Sorbian, and Lower
Sorbian. Previous studies do not provide a principled account why (i) the
monomorphemic dual was reanalyzed as bimorphemic in Slovenian, Upper, and Lower
Sorbian, and why (ii) it was replaced by the plural in the majority of Slavic languages
including Russian and Kashubian. In this dissertation, I investigate diachronic changes in
the morphosyntactic category of the Slavic dual which occurred in the 11th-15th
centuries. In this dissertation, I present new diachronic data obtained through digital
corpora of Old Slavic manuscripts. Within the framework of Distributed Morphology, I
propose a new principle of Morphosyntactic Feature Economy which explains these two
different patterns of diachronic change in the Slavic dual pronouns and verbal agreement.
I argue that the principle of Morphosyntactic Feature Economy is one of the factors
which drives diachronic change in the category of number in Slavic languages.
Morphosyntactic Feature Economy is a principle of efficient computation. It plays an
essential role in restructuring of the morphosyntactic category of the Slavic dual.
Morphosyntactic restructuring of the Slavic dual, triggered by its morphosyntactic and
semantic markedness, results in a morphosyntactically 'simpler' category of number. As a
consequence of the application of the principle of Morphosyntactic Feature Economy, the
language learner acquires a featurally restructuctured category of number, which is
simpler and more computationally efficient.
CHAPTER 1 INTRODUCTION
1.0.
The Goals of the Dissertation
In this dissertation, I investigate the reasons for diachronic changes which
occurred in the Slavic dual pronouns and verbal agreement in the time period from the
11th century until present. There are three goals which I pursue throughout this
dissertation. The main goal is to determine which factors internal to linguistic
computation are responsible for diachronic change in the inflectional category of the
Slavic dual. The second goal is to propose a principled explanation of two different
patterns of diachronic change in the grammatical category of the dual across the South,
West, and East branches of Slavic languages. The third goal, which follows from my
proposal, is to predict possible diachronic changes in the category of the dual crosslinguistically.
The first question which needs to be answered concerns language-internal factors
which lead to diachronic changes in the inflectional category of the dual. This question,
though not new in linguistic theory, remains largely unexplored within the Distributed
Morphology approach to the architecture of grammar, which I follow in this dissertation
(Noyer 1997, Calabrese 2011). Diachronic change is a change in the internal grammar of
one generation of speakers which becomes an input to the grammar of the consequent
generation. How can we explain this type of change, expressed morphologically and
16
phonologically, in terms of a linguistic computation which a language learner needs to
make?
I propose a new principle of Morphosyntactic Feature Economy, and argue that it
is one of the factors which drives diachronic change in the category of number.
Morphosyntactic Feature Economy, which I propose in this dissertation, is a principle of
efficient computation. It plays an essential role in restructuring of the morphosyntactic
category of the dual. Morphosyntactic restructuring of the dual, triggered by its
markedness, results in a morphosyntactically 'simpler' category of number. As a
consequence of the application of the principle of Morphosyntactic Feature Economy, the
language learner acquires a featurally restructuctured category of number.
There are three options available for a language learner to pursue to make the
Slavic dual less marked and easier to acquire in terms of its computational efficiency.
The 1st option allows a language learner to reanalyze a morphosyntactically marked dual
as a less marked, bi-morphemic dual morphologically realized by a combination of two
morphemes: a plural morpheme and the numeral two (1a). The 2nd option entails a
reanalysis of the dual as a less marked plural (1b). The 3rd option allows a language
learner to reanalyze a morphosyntactically marked dual as a so-called 'constructed' dual
morphologically represented by a combination of a plural and a singular morpheme (1c).1
(1)
Diachronic Reanalysis of the Slavic Dual
a). dual → bi-morphemic dual = plural + two
b). dual → plural
1
The first observation of the 'constructed' dual construction goes back to Hale's work on Hopi (1997). The
'constructed' dual was later systematically analyzed by Noyer (1997) and Nevins (2006, 2008, 2011).
17
c). dual → 'constructed dual' = plural + singular
The second question which I address in this dissertation concerns two different
patterns of diachronic changes which occurred in the Slavic dual pronouns across three
different language subgroups - South, West, and East Slavic (2).
(2)
Two Patterns of Diachronic Change in the Slavic Dual
a). dual → 'renewed' dual
Slovenian and Sorbian
b). dual → plural
Russian and Kashubian
The first pattern of diachronic change is exemplified in three contemporary languages Slovenian (South Slavic), Upper, and Lower Sorbian (West Slavic). In Slovenian and
Sorbian, the dual continued to be used by the speakers in an 'innovated', bi-morphemic
structure. The second pattern of diachronic change is more pervasive, and occurred in all
of the other Slavic languages presented here by Russian (East Slavic) and Kashubian
(West Slavic). In Russian and Kashubian, the dual was lost and reanalyzed by the
speakers as the plural.
In this dissertation, I propose that the reason why two different patterns of
diachronic change emerged was due to the two different morphological repairs (Fission
and Impoverishment) which the Slavic dual was subject to at Morphological Structure . I
argue that the morphosyntactic repairs of Fission and Impoverishment were triggered by
the Principle of Morphosyntactic Feature Economy which applied at Morphological
Structure, a separate level of representation of morphosyntactic features within the
Distributed Morphology model of grammar (Noyer 1997, Harley and Noyer 1999).
18
The third goal which I attempt to achieve in this dissertation is to predict possible
patterns of diachronic change in the category of dual cross-linguistically. If my proposal
of the principle of Morphosyntactic Feature Economy is on the right track, dual pronouns
should never be expressed by a single morpheme. The dual may be either bi-morphemic,
may be lost and replaced by the plural, or expressed by a 'constructed' dual in which a
dual reference is expressed via a combination of a plural pronoun and a singular verb. In
this dissertation, I show that the predictions of the principle of Morphosyntactic Feature
Economy are borne out in a number of languages typologically unrelated to Slavic. In
Manam (Austronesian), Hebrew (Afro-Asiatic), and Hopi (Uto-Aztecan), the dual has a
consistent bi-morphemic structure. A much wider variety of languages with the dual
remain to be analyzed to see if the predictions of the proposed principle of
Morphosyntactic Feature Economy hold.
19
1.1.
The Problem of the Slavic Dual
“Судьба двойственнаго числа
достаточно для того, чтобы получить хотя бы приблизительное представленіе о
томъ, какимъ образомъ двойственнoe числo, такъ
употреблявшееся въ старо2
А.И. Беличъ (1899)
The diachronic changes in the Slavic dual, although well documented, have been
a puzzling problem for linguistic theory since the 19th century. There are two diachronic
patterns of change which are attested in the Slavic dual (3 repeated from 2). The
reanalyzed dual pronouns of Slovenian and Sorbian received no principled explanation,
and were described as being "innovated" by the speakers (Derganc 1988, 1998, 2003;
Jakop 2008a). Similarly, dual pronouns which were replaced by the plural forms in
Russian were treated as 'lost' due to markedness of the dual as a more restricted type of
number (Žholobov 2001). However, no principled account was offered to explain why
the dual was marked, and what factors contributed to the loss of the Russian dual.
(3)
2
Two Patterns of Diachronic Change in the Slavic Dual
a). dual → 'renewed' dual
Slovenian and Sorbian
b). dual → plural
Russian and Kashubian
"The fate of the dual number in Slavic languages has not been analyzed enough to even get a remote idea
of how the dual, which was so consistently used in Old Church Slavic and still used in some of the
contemporary Slavic languages, completely disappeared in the other Slavic languages."
20
Bi-morphemic dual pronouns, which consist of a plural stem and the numeral two
or the dual suffix -j, continue to be used in Contemporary Standard Slovenian and
Contemporary Standard Upper and Lower Sorbian (4-5). Russian, as an example of a
Slavic language which lost the dual and replaced it with the plural, makes only a singular
~ plural distinction in the pronouns (6).
(4)
Slovenian
Mi-dva
bo-va
šl-a
po levi poti, vi-dva pa po desni.
1.PL.two be.1.DU.FUT
go.PRT-DU.MASC on left road, 2.PL-two and on right
‘The two of us will take the road on the left while the two of you the one on the
right.’
(Derganc 2003:169)
(5)
Upper Sorbian
Wó-j stej
najbóle skók
2.PL-DU be.2.DU most
jump
‘You two are the fastest.’
(Scholze 2007:124-125)
(6)
Contemporary Standard Russian
Vy gljadi-te
v oba.
1.PL look-1.PL.IMP in both
'You, beware.'
Diachronic changes in the Slavic dual are schematically represented in (Figure
1.1).3 As it is shown in (Figure 1.1), Proto- Indo European and Proto-Slavic had three
numbers - singular, dual, and plural (Sussex and Cubberly 2006:222). Approximately in
the 5th century AD, Proto-Slavic split into three branches - West, South, and East Slavic.
3
The presence of the dual is indicated by a star (*) in Figure 1.1.
21
Over time, each of the three branches gave rise to a number of different, but still
typologically and structurally related Slavic languages.
Figure 1.1 Diachronic Changes in the Slavic Languages
*Proto-Indo-European (3700-3400 BC)
*Proto-Slavic (2000-1500 BC - 5th AD)
*West Slavic
*Old Polish
*Old Czech
*South Slavic
*Old Slovak *Old Sorbian
Polish *Old Kashubian Czech
Slovak *Upper *Lower
Sorbian
*East Slavic
*Old Church Slavic
*Old Russian
Russian Belorusian Ukrainian
Kashubian
*Slovenian Bosnian Croatian Serbian Bulgarian Macedonian
* denotes that the language has dual
22
The dual, which was a part of the number system of Proto-Slavic, underwent a
series of diachronic changes. While the dual was a very robust grammatical category in
Old Church Slavic (11th century), it did not survive in all of its South Slavic descendants
besides Slovenian. In the East Slavic branch, Old Russian (11th century) had a very
robustly expressed dual number marked on nouns, pronouns, and verbal suffixes.
However, the Old Russian dual did not survive in any of the contemporary East Slavic
languages including Russian, analyzed in this dissertation.
Initially, diachronic change in the dual in the West Slavic branch was similar to
that of the East Slavic branch. Old Sorbian (9th-13th centuries) like Old Russian used to
have dual number in its grammar.4 However, over time, the dual survived only in two
West Slavic languages - Upper and Lower Sorbian. Upper Sorbian is a West Slavic
language spoken in eastern Germany in the region of Upper Lusatia with a cultural center
in Bautzen (spelt as Budyšin in Upper Sorbian) . It is a minority language due to the
political and economic dominance of German as the official language of Germany. It is
estimated that there are no more than 15,000 of native speakers of Upper Sorbian and all
of them are bilinguial (Elle 2000).
Compared to Elle (2000) estimates, the statistics from www.ethnologue.com show
a smaller number of native speakers of Upper Sorbian. According to Ethnologue,
currently, Upper Sorbian is spoken by 13,300 people. The language is classified as
"developing" which means that it "is in vigorous use, with literature in a standardized
4
I follow Schaarschmidt's chronological timeline in the historical development of Old Sorbian (1998:20).
23
form being used by some, though this is not yet widespread or sustainable"
(www.ethnologue.com).
Upper Sorbian still retains dual number marked on nouns, pronouns, and verbs
(Schaarschmidt 2002). The dual is undergoing decline, especially in the main dialects
spoken near Bautzen, and it is reported to be completely lost in some southern dialects of
Upper Sorbian, such as Rodewitz/Rodecy spoken near Bautzen (Schaarschmidt 2002:23).
Lower Sorbian, the other West Slavic language which retained dual number in its
grammar, is spoken in eastern Germany near the town of Cottbus ( spelt Chośebuz in
Lower Sorbian). Similar to Upper Sorbian, Lower Sorbian is a minority West Slavic
language. In comparison to Upper Sorbian, Lower Sorbian has a lot fewer native
speakers. According Ethnologue, Lower Sorbian is an endangered language spoken by
approximately 6,670 people. According to Ethnologue, Lower Sorbian is a "shifting"
language in which "intergenerational transmission is in the process of being broken, but
the child-bearing generation can still use the language, so it is possible that revitalization
efforts could restore transmission of the language in the home" (www.ethnologue.com)
Another minority West Slavic language, which used to have dual number in its
earlier historical development, is Kashubian. Kashubian belongs to the Lechitic subgroup
within West Slavic languages. It is spoken in the north-west part of Poland near the city
of Gdansk. According to Ethnologue, Kashubian is on the brink of extinction; it is
classified as a moribund (dying) language used only by community elders
(www.ethnologue.com). Currently, no children are acquiring Kashubian as their native
24
language. It is estimated that Kashubian is spoken by approximately 3000 people (Sussex
and Cubberly 2006:97).
Currently, great efforts are made by the Polish government to revitalize
Kashubian as a regional language. According to the Euromosaic Study conducted by the
European Commission, in 2002 Kashubian was introduced in the public school system in
Poland.5 It was used as a primary language of instruction in kindergarten and elementary
schools. Such language revitalization initiatives carried out by the European Commission
(within the European Union) are extremely important not only in regards to Kashubian,
but also in regards to revitalization of any endangered minority language spoken in
Europe.
5
More information about Kashubian can be found on the European Commission website
http://ec.europa.eu/languages/euromosaic/pol3_en.htm.
25
1.2. The Hypotheses
In this dissertation, I argue that the Slavic dual is represented by a marked feature
combination - [-singular -augmented] whose markedness should be eliminated. I further
propose that there are three hypotheses which can be postulated to account for the
diachronic changes in the Slavic dual (7).
(7)
Hypothesis 1: [-singular -augmented] → [-singular] [-augmented]
Hypothesis 2: [-singular -augmented] → [-singular]
Hypothesis 3: [-singular -augmented] → *[-augmented]
According to Hypothesis 1, the marked feature combination [-singular
-augmented] of the dual is split into two syntactic terminal nodes - [-singular] and
[-augmented]. The [-singular] and [-augmented] features can be realized by two separate
Vocabulary Items or morphemes. Hypothesis 1 is borne out. It is empirically confirmed
by the pattern of diachronic change in the Slovenian and Sorbian duals. Both Slovenian
and Sorbian dual pronouns have a bi-morphemic structure. The 1st morpheme is the
plural stem which is encoded by the [-singular] feature. The 2nd morpheme is the
numeral dva ('two') or the dual suffix -j which is encoded by the [-augmented] feature.
According to Hypothesis 2, the [-augmented] feature is deleted from the marked
[-singular -augmented] feature bundle, making the dual less marked. Hypothesis 2 is
borne out in all of the other Slavic languages including Russian and Kashubian which
underwent a diachronic change from the dual to plural. In Russian and Kashubian, the
plural morpheme, which replaced the dual, is encoded by the feature [-singular].
26
According to Hypothesis 3, the marked [-singular] feature of the dual is deleted
from the marked [-singular -augmented] feature combination. As a result, the reanalyzed
structure has the feature [-augmented]. This would entail that the dual is reanalyzed as the
singular since the semantics of the feature [-augmented] corresponds to referential
cardinality 1.
In a formal semantic framework, the [+augmented] feature means that given some
predicate P, it would pick out a reference set x if x has a proper subset y (8) (Link 1983,
Harbour 2008, Nevins 2011). The negative value ('-') of the [augmented] feature means
that it picks out a reference set without a proper subset. The only cardinality which does
not have a proper subset is number 1. When the [-augmented] feature appears in isolation,
without any other number feature, it refers to the referential cardinality of number 1.
(8)
Definition of the [+augmented] Number Feature
[+augmented] = λ P ∃y [P(x) ∧ P (y) ∧ y ⊂ x]
Diachronic change from cardinality 2, the dual, to cardinality 1, the singular, is
impossible for formal semantic reasons. Therefore, Hypothesis 3 is ruled out.
Empirically, no attested language that I am familiar with has undergone a historical
change from the dual to the singular.
In this dissertation, I test Hypotheses 1 and 2 and argue that these two hypotheses
are borne out and supported by the empirical data. To account for Hypotheses 1 and 2, I
put forward a new proposal which accounts for diachronic change in terms of economy of
morphosyntactic features.
27
1.3.
The Proposal: A Principle of Morphosyntactic Feature Economy
I propose a new principle of Morphosyntactic Feature Economy which provides a
principled explanation for two different patterns (Hypotheses 1-2) of diachronic change
in the Slavic dual (9). I claim that the principle of Morphosyntactic Feature Economy
operates on a morphosyntactic representation of the Slavic dual which is marked and
computationally inefficient. Assuming that the Faculty of Language (part of our cognitive
system) is designed for optimal computation , I stipulate that the proposed principle
belongs to the "third factor" principles of efficient computation (Chomsky 2005, 2008).
(9)
The Principle of Morphological Feature Economy
A marked [-singular -augmented] feature combination of the dual cannot be
realized at Phonological Form without eliminating markedness of its features at
Morphological Structure.
As shown in the diagram below, the principle of Morphosyntactic Feature
Economy targets a marked [-singular -augmented] feature bundle of the Slavic dual at
Morphological Structure before morphosyntactic features are filled with phonological
content through Vocabulary Insertion at Phonological Structure (10). There are two ways
to make the morphosyntactic representation of the dual less marked and therefore more
computationally efficient. One option is to split the marked feature bundle of the dual
into two separate terminal nodes via Fission, which happened in Old Slovenian and
Sorbian. The other option is to delete the [-augmented] feature via a morphological
repair operation of Impoverishment, which occurred in Old Russian and Kashubian.
28
(10)
The Principle of Morphological Feature Economy
Morphosyntactic Representation of the Slavic Dual at MS
MS
Dual [-singular -augmented]
[F marked F marked]
Marked
Morphosyntactic Repairs at MS
Fission
Impoverishment
(Old Slovenian and Sorbian)
(Old Russian and Kashubian)
[-singular] [-augmented]
[-singular]
Less marked
Less marked
Phonological Form
29
1.4.
Theoretical Assumptions
I assume the architecture of grammar within the theory of Distributed
Morphology (DM) developed by Harley and Noyer (1999) (11). In Distributed
Morphology, the derivation of a 'word' is distributed among three different components of
grammar - Narrow Syntax, Morphology, and Logical Form. Narrow Syntax computes
hierarchically organized syntactic structure by means of operations Merge and Copy.
This syntactic structure is spelled out by the morphological component of grammar, and
its meaning is determined by the component of Logical Form. At Phonological Form,
morphosyntactic features or feature bundles are filled with phonological material
corresponding to roots and affixes.
(11)
The Architecture of Grammar in DM (Harley and Noyer 1999)
Narrow Syntax (NS)
Morphological Structure (MS)
Logical Form
Impoverishment, Fission, and
Fusion
(LF)
Vocabulary Insertion
Phonological Form (PF)
30
In the model presented above, Morphology contains a level of representation
called Morphological Structure. Morphological Structure ,which mediates Narrow Syntax
and Phonological Form, is crucial for my proposal. I show that at Morphological
Structure a marked feature bundle of the dual is subject to morphological repairs which
eliminate its markedness before the dual reaches Phonological Form. Diachronic changes
in the Slavic dual pronouns provide empirical evidence for two types of repair operations
- Impoverishment and Fission which 'simplify' a marked morphosyntactic structure of the
dual.
31
1.5.
Methodology and Data
In this dissertation, I obtained original diachronic data from digitized corpora of
Old Church Slavic, Old Russian, Old Slovenian, and Old Sorbian manuscripts and texts.
The reason for collecting original diachronic data was to construct pronominal and verbal
agreement paradigms to more accurately trace diachronic changes in the Slavic dual.
Previous researchers (Iordanskij 1960, Xaburgaev 1989) who analyzed the Slavic dual
made incorrect generalizations about the diachronic data due to their reliance on incorrect
secondary sources. In my own analysis of the Slavic dual, my goal was to use only
reliable digitized diachronic sources to avoid mistakes in the data which could lead to
incorrect analysis and predictions.
Before I started writing this dissertation in 2010, digital corpora of Old Church
Slavic and Old Russian manuscripts were very scarce and highly unreliable. Historical
linguists had to rely on either facsimiles of Old Church Slavic and Old Russian texts
published long before the advance of modern technology, or on secondary sources of
information. As a result of such approaches to diachronic data, thorough and systematic
analysis of the Slavic dual pronouns and verbal agreement paradigms was flawed with
errors.
The diachronic data which I analyzed in this dissertation spans the 11th through
the 15th centuries. I examined the following old Slavic manuscripts and texts listed in
(12). Throughout the dissertation, I cite exact examples from these texts. The diachronic
data from Old Church Slavic and Old Russian texts presented in this dissertation have not
been published before, and are original. However, there were some instances when I had
32
to resort to secondary sources of data in the analysis of Old Slovenian and Old Sorbian
pronouns and verbal agreement. Whenever I used an example taken from a secondary
source, I cited a direct reference within the text.
(12)
Old Slavic Manuscripts and Texts
Old Church Slavic
The Codex Marianus (Mariinskoe Evangelie) (11th century)
Sava's Book, 1st and 2nd parts (Savvina Kniga) (11th century)
Old Russian
The Ostromir Gospel (Ostromirovo Evangelie) (1056-1057)
Savva's Book, the 2nd and 3rd parts (Savvina Kniga) (11th century)
The Izbornik (1076)
Archangel Gospel (1092)
The Lives of Boris and Gleb (Skazanie o Borise i Glebe) (12th century)
Tolkovyj Apostol (1220)
Simonov Gospel (1270)
Čudov Apostl (1355)
The Primary Chronicle from the Laurentian Codex (1377)
(Povest' Vremennyx Let po Lavrent'evskomu Spisku)
The Primary Chronicle from the Ipatjev Codex (ca. middle of the 15th century)
(Povest' Vremennyx Let po Ipat'evskomu Spisku)
The Primary Chronicle from the Radziwiłł Codex (ca. end of the 15th century)
(Povest' Vremennyx Let po Radzivillovskomu Spisku)
33
Old Slovenian: Dalmatian Bible (1584) in translation of Yurij Dalmatin
Old Sorbian: Jakubica's New Testament (1548)
Lower Sorbian: Bramborski Serbski Casnik (1848-1880)
I collected the diachronic data used in this dissertation from the following digital corpora
(13).
(13)
Digital Corpora of Old Slavic Manuscripts
Old Church Slavic
The parsed corpus of Old South Slavic manuscripts, University of Southern
California (http://www-bcf.usc.edu/~pancheva/ParsedCorpus.html).
Old Russian
Project Manuscript (www.manuscripts.ru), the Iževsk State Technical University,
the Russian Ministry of Education and Science.
The Digital Library of Old Russian Literature, the Institute of Russian Literature,
the Russian Academy of Sciences,
(http://pushkinskijdom.ru/Default.aspx?tabid=2070).
Lower Sorbian
Niedersorbisches Textkorpus
http://www.dolnoserbski.de/korpus/?rec=de
Slovenian: Scholarly Digital Editions of Slovenian Literature, a corpus of early
Slovenian manuscripts http://nl.ijs.si/e-zrc/bs/index-en.html
In addition to the digital corpora, I consulted the following printed Old Slavic and
Russian manuscripts and dictionaries (14-15).
34
(14)
Printed Old Slavic Manuscripts
Old Russian
The Ostromir Gospel (Ostromirovo Evangelie) (1056-1057) originally published
by A.Vostokov in 1843 and then reprinted.
Savva's Book, the 2nd and 3rd parts (Savvina Kniga) (11th century) jointly
published by the Institute of Slavic Studies of the Russian Academy of Sciences,
Russian Federal Archival Agency, and the Institute of the Bulgarian Language of
the Bulgarian Academy of Sciences.
Russian
Bibliya Vetxago i Novago Zavĕta. 1922. Londonǔ: Britanskago i Inostrannago
Biblejskago Obščestva.
(15)
Dictionaries
Old Church Slavic Dictionary (10-11th centuries)
Old Russian Dictionary (11-14th centuries)
Upper Sorbian-Russian Dictionary
In my analysis of the Slavic dual, I used the method of corpus analysis to identify
the contexts in which dual pronouns and dual verbal suffixes occurred as opposed to the
plural. The reason for this approach was to find contexts in which dual pronouns,
formally syncretic (identical in morphological form) with the plural, had a dual reference.
Based on the occurrences of dual and plural pronouns and verbal agreement suffixes in
35
Old Church Slavic and Old Russian manuscripts, I constructed pronominal and verbal
suffix paradigms.
During the period of diachronic data collection for this dissertation, I was
fortunate to find a very well researched and annotated digital corpus of Old South Slavic
texts developed by Dr. Pancheva and her research team at the University of Southern
California. 6Among the available Old Church Slavic digital manuscripts, I was mostly
interested in the Codex Marianus, one of the earliest Glagolitic four-fold gospels. It dates
back to the 11th century, and remains one of the most important Old Church Slavic
manuscripts to date. I used the data from the Codex Marianus to establish a pattern of
development of the Slavic dual pronouns and verbal agreement.
Due the efforts of Russian historical and computational linguists and the financial
support from the Russian Foundation for the Humanities, Project Manuscript, a new and
expansive collection of digitized Old Church Slavic and Old Russian manuscripts was
launched in 2005 to provide researchers with access to reliable and searchable diachronic
data.7 Access to the digital diachronic corpora of Project Manuscript allowed me for the
first time to analyze in full detail such old Russian manuscripts, as The Ostromir Gospel
(1056-1057), The Izbornik (1076), and the three chronologically different versions of The
Primary Chronicle (Povest' Vremennmyx Let) - The Laurentian Codex (1377), The
Ipatjev Codex, and The Radziwiłł Codex (ca. 1490). These three versions of the Old
6
I am very grateful to Dr. Roumyana Pancheva for her help and assistance with the USC parsed corpus of
Old South Slavic manuscripts.
7
I would like to express my enormous gratitude to Dr.Viktor Baranov, the director of Project Manuscript
and Professor at the Iževsk State Technical University for his help with providing access to the diachronic
digital corpora of Old Russian and Old Church Slavic texts.
36
Russian The Primary Chronicle have never been previously analyzed to the extent which
I achieved through the use of a morphological analyzer and other corpus tools.
Obtaining original Old Slovenian and Old Sorbian diachronic data was very
challenging due to the lack of available and reliable digital corpora. In my analysis of the
diachronic changes dual in Old Slovenian, I used secondary sources which were cited
accordingly. The Jakubica's New Testament (1548), one of the most important Old
Sorbian manuscripts, was not available in a digital corpus. In the analysis of the Sorbian
dual, I had to rely on secondary sources as well. Some of the Lower Sorbian texts dating
to the 19th century were accessible to me through a Niedersorbisches Textkorpus. It is
my hope that in the near future, more digital corpora will be developed to study and
preserve such highly endangered West Slavic languages, as Upper and Lower Sorbian.
37
1.6.
The Structure of the Dissertation
In Chapter 2, I present the most influential typological approaches to the
classification of the world’s languages with dual number and identify the place of the
Slavic dual within these typologies. I begin with the first typological discussion of
languages with the dual proposed by Wilhelm von Humboldt in his famous paper “On the
Dual Form” (1827, 1997). I follow up with a typology of the dual suggested by Otto
Jespersen (1965). Then, I outline the critique of Humboldt’s typology proposed by Plank
(1989), and present further implicational universals concerning the dual suggested by
Greenberg (1963), Corbett (2000), and Cysouw (2009). Finally, I present the typology of
the Slavic dual proposed by Belič (1899), Iordanskij (1960), and Žolobov (1998).
In Chapter 3, I present the framework of Distributed Morphology which I adopt in
this dissertation for the formal analysis of dual number in the Slavic languages (Harley
and Noyer 1999). I show a step-by-step DM derivation of dual number marked on Slavic
pronouns and verbs (by agreement), and explain why morphological operations of
Impoverishment and Fission, which operate on morphosyntactic representations of the
dual at Morphosyntactic Structure, are key for my proposal of diachronic changes in the
Slavic dual.
In Chapter 4, I explore one of the most interesting aspects of dual number
inflection in Slavic; namely a mismatch between its formal representation in Narrow
Syntax and Morphological Structure. There are two questions about the morphosyntactic
representation of the Slavic dual which I address in this chapter. First, what are the
number features which represent the Slavic dual in Narrow Syntax? Second, what are the
38
patterns of cross-linguistic variation in the realization of the Slavic dual at Phonological
Form?
I assume that dual number marked on Slavic personal pronouns and verbal
agreement suffixes is featurally complex, and is represented as a combination of the
[-singular -augmented] number features in Narrow Syntax (Noyer 1997, Harbour 2011,
Nevins 2011). I define the [±singular] and [±augmented] number features in a formal
semantic framework and explain how the [-singular -augmented] features combine to
deliver the dual semantics. Assuming Ritter's (1995) proposal that the number features
are hosted by a Number (Num) node, I argue that the Slavic dual is represented by a fully
specified morphosyntactic feature bundle [-singular -augmented] in Narrow Syntax.
I identify two patterns of cross-linguistic variation in the PF representation of the
Slavic dual: bi-morphemic and monomorphemic. I show that the bi-morphemic dual
pronouns occur in the three contemporary Slavic languages- Slovenian (South Slavic),
Upper, and Lower Sorbian (West Slavic). The monomorphemic dual pronouns were
found in Old Russian (East Slavic) and Kashubian (West Slavic), but over time were
replaced with the plural pronouns. Based on corpora analyses, I argue that in Slovenian
and Sorbian, bimorphemic dual pronouns diachronically developed from their
monomorphemic 'ancestors' while in Old Russian and Kashubian, the monomorphemic
dual pronouns were replaced by the plural pronouns in contemporary Russian and
Kashubian.
In Chapter 5, In this final chapter of the dissertation, I present my proposal of the
principle of Morphosyntactic Feature Economy and argue that it provides a principled
39
explanation for the two different patterns of diachronic change which occurred in the
Slavic dual pronouns. In the first pattern of diachronic change, the dual in Old Slovenian
and Old Sorbian was reanalyzed as a bi-morphemic structure and continued to be used by
contemporary speakers of both languages. In the second pattern, the dual in Old Russian
and Kashubian was 'lost,' i.e. it was reanalyzed as the plural.
In Chapter 6, I summarize the results of the dissertation, discuss theoretical
implications of the principle of Morphosyntactic Feature, and outline directions for future
research. I discuss the implications of the principle of Morphosyntactic Feature economy
within a larger linguistic context. I explore what the principle of Morphosyntactic Feature
Economy predicts in relation to morphosyntactic markedness, grammaticalization,
linguistic typology, and the architecture of grammar in DM.
40
CHAPTER 2 DUAL NUMBER IN THE WORLD’S LANGUAGES FROM A
TYPOLOGICAL PERSPECTIVE
2.0. Introduction
In this chapter, my aim is to discuss the relevant typological approaches to the
classification of the world’s languages with dual number and to identify the place of the
Slavic dual within these typologies. I begin with the first typological discussion of
languages with the dual proposed by Wilhelm von Humboldt in his famous paper “Űber
den Dualis" ("On the Dual Form”) (1827). I follow up with a typology of the dual
suggested by Otto Jespersen (1965). Then, I outline the critique of Humboldt’s typology
proposed by Plank (1989), and present further implicational universals concerning the
dual suggested by Greenberg (1963), Corbett (2000), and Cysouw (2009). Finally, I
present the typology of the Slavic dual proposed by Belič (1899), Iordanskij (1960), and
Žolobov (1998, 2001).
41
2.1. Humboldt's Typology of Dual Number
“Nach dieser Vorstellung ist der Dualis gleichsam ein Collectivsingularis der Zahl zwei,
da der Pluralis nur gelegentlich, nicht aber seinem ursprünglichen Begriff nach, die
Vielheit wieder zur Einheit zurückfürt. Der Dualis theilt daher, asl Mehrheitsform, und
als Bezeichnung eines gescholossenen Ganzen zugleich die Plural und Singular-Natur.”
“According to this view, the dual is, as it were, a collective singular of the number two,
since the plural only occasionally, but not by virtue of its original concept, turns plurality
back into the concept of oneness. As a form of the plural and at the same time as an
indication of a unified whole, the dual form has the nature both of a plural and a
singular.”
Wilhelm von Humboldt (1827:9)
Wilhelm von Humboldt’s pioneering paper “On the Dual Form” (1827) was one
of the first attempts to subject the category of the dual to cross-linguistic empirical
scrutiny and to present descriptive generalizations about the dual based on his innovative
methodology. In contrast to the ‘holistic’ comparative method with a focus on an entire
grammar of a language that was used by his 19th century contemporaries, Humboldt’s
novel methodology was to concentrate solely on one grammatical category, such as the
dual, to arrive at a more systematic structural typology of the world’s languages.
In his lecture “On the Dual Form” (1927) addressed to the Berlin Academy of
Sciences, Humboldt delivered a ground-breaking analysis of the category of dual number
examined cross-linguistically. The novelty of Humboldt’s approach consisted in the
investigation of the dual in its different morphological instantiations in the world’s
languages and in the examination of the concepts underlying dual forms. Humboldt was
convinced that a combination of a thorough empirical analysis of morphological dual
forms as well as the concepts that give rise to those forms would give a new explanation
of cross-linguistic differences in languages with the dual and would be key to
42
understanding of the concept of dual number. By using this new method in the realm of
comparative language study, Humboldt avoided what he called “a one-sided historical or
philosophical” approach to typology since he approached the problem of the dual both
conceptually and empirically (1997:112).
Humboldt’s reasoning for studying the dual as opposed to any other grammatical
category is worth mentioning. He presents several reasons for his choice. First, Humboldt
considers that dual number by itself can be more easily isolated from the grammatical
structure of a language since this category is “less deeply rooted” compared to pronouns
and verbs which are “so deeply rooted” in the grammatical structure of a language that
become more difficult to extract (1997:116). Second, the dual does not occur in a great
number of languages, which narrows down the scope of linguistic investigation.
The third reason for choosing the dual is its diverse cultural geography. As
Humboldt states, the dual can be found on the one hand in “uncivilized nations among the
Greenlanders and the New Zealanders,” and on the other hand in the Attic Greek, a
dialect spoken by the most highly educated.8 The fourth reason for analyzing the dual is
its accessibility for cross-linguistic study since one compares a single grammatical
category in a variety of languages which allows to investigate “a smaller field” in “the
smallest detail” (Humboldt 1997:117).
Last but not least, Humboldt mentions that the dual as a grammatical category is
closely related to the plural. Humboldt argues that due to this relation, the dual should be
It seems that the term ‘uncivilized nations’ might have been quite appropriate for the 19th century
linguistics. It is inappropriate in current linguistic theory and discourse. Its usage though, does not detract
from the main point that Humboldt was making.
8
43
studied not separately, but together with the plural. Humboldt does not carry out a
specific investigation of the plural, but draws on empirical evidence from languages with
restricted and unrestricted plurals to discover the true nature of dual number. Having
presented his reasons for studying the dual, Humboldt concludes his rationale with a final
remark that the category of dual rather than the category of number is a much more
focused object of linguistic analysis which allows for the application of pure reasoning
and meticulous empirical investigation.
Almost all of Humboldt’s above-mentioned observations about the dual hold true.
However, a few comments are in order. Humboldt makes the insightful observation that
dual forms can be easily isolated in morphological systems cross-linguistically. However,
for a more comprehensive study of the dual within one language as well as crosslinguistically, dual forms of nouns cannot be studied in isolation from pronouns and
verbal agreement. Such an ‘isolationist’ approach to dual forms would not work for many
Indo-European languages, including Slavic, where number and person are fused together
in a single morpheme in nouns, pronouns, and verbal agreement.
Humboldt’s empirical observation about the rarity of occurrence of dual number
in the world’s languages and its diverse geography was a remarkable insight. He remarks
that the dual occurs in a small number but in a large variety of typologically unrelated
language families. These language families include Afro-Asiatic (Arabic, Maltese);
Austronesian (Malay, Tagalog, Pampangan, Tahitian); Eskimo-Aleut (Inuktitut);
Iroquoian (Cherokee); Saami (Uralic); and Indo-European: Indo-Iranian (Sanskrit,
44
Avestan), Hellenic (ancient Greek), Baltic (Lithuanian), Celtic (Welsh), Germanic
(Gothic), and Slavic.9
Humboldt’s insight that dual is thus very well suited for cross-linguistic
comparison is especially important. Putting dual forms to empirical scrutiny crosslinguistically allows one to make crucial generalizations about how the category of
number is realized in a variety of languages, and how the concept of number in general is
represented in the mind of a speaker.10 Lastly, Humboldt stresses the necessity to analyze
the dual in conjunction with the plural since the latter can be further classified into a
restricted (paucal for two or several) and an extended plural (many) in some languages
which affects conceptual representation of the dual in the speaker’s mind.
Humboldt’s linguistic typology of languages with the dual is based on three
factors: (1) notion (Begriff), (2) extension (Umfang), and (3) concept (Vorstellung). In
terms of notions (Begriffen), he divides all languages into three classes: (1) languages
whose notion of the dual comes from the opposition between the speaker (‘I’) and the
addressee (‘You’); (2) languages that derive the dual based on objects found in natural
pairs; and (3) languages in which the dual derives not from any kind of phenomenon, but
from an abstract concept of duality (Table 2.1).
9
Humboldt does not discuss any details of the dual in Slavic.
This latter point was not originally intended by Humboldt, as he stated himself when presenting his
reasons for the study of the category of dual number rather than the category of number. However, it
naturally follows from empirical cross-linguistic generalizations about the dual and its comparison to the
singular and plural.
10
45
Table 2.1 Humboldt's Notional Typology of Dual Number
‘Notion’ (Begriff)
(1) I vs. You
(2) natural pairs of objects
(3) abstract duality
In languages of the 1st class, dual forms are based entirely on the idea of person.
Dual forms reflect the opposition between the speaker and the addressee; namely between
the 1st and the 2nd person. In such languages, dual forms remain tied to pronouns, and
depend on whether the 1st or the 2nd person is attributed more significance. There are
some languages where dual forms are restricted only to the 1st person plural; that is to the
1st person inclusive (‘we’ which includes the speaker and the addressee) whereas other
languages make a distinction between the 1st and the 2nd person in the dual.
Languages of the 2nd class derive dual forms from objects that come in natural
pairs. The bilateral symmetry of the human body supplies a number of examples for this
class: eyes, ears, brows, legs, hands and all other bodily features that come in twos. In
these languages, the dual does not extend beyond nouns that denote natural pairs.
The 3rd class is represented by languages in which dual forms occur in all parts of
speech without any restriction. Humboldt assumes that dual forms in these languages
derive from an abstract concept of duality. He does not explain how languages of this
class have reached this level of abstraction to ascribe dual notions to all parts of speech.
One can assume that Humboldt’s 3rd class most probably represents the concept of
46
agreement in languages with dual nouns and pronouns. However, Humboldt does not
make any explicit reference to agreement per se.
Concluding his ‘notional’ typology, Humboldt remarks that it is quite obvious that
languages may belong to more than one of the above classes, or to all three classes at
once. How is this possible? Humboldt gives an answer to this question with a very
important observation about language change. Over time, languages originally belonging
to the 3rd class where the dual permeates the entire grammar of the language may retain
the dual either in pronouns or nouns, or both, which puts such languages in either the 1st
or 2nd classes, or both.
Humboldt further classifies languages with the dual according to the category of
extension (Umfang). In this classification, he divides all languages with the dual into
three classes: (1) languages whose dual is limited only to pronouns; (2) languages whose
dual is restricted only to nouns; and (3) languages in which the dual is present in all parts
of speech (Table 2.2).
Table 2.2 Humboldt's Extensional Typology of Dual
‘Extension’ (Umfang)
CorrespondingLanguages
(1) Dual Pronouns
Malay, Tagalog, Pampangan (Austronesian);
Chaima (Carib); Tamang (Sino-Tibetan)
(2) Dual Nouns
Totonac (Totonacan); Quechua (Quechuan)
(3) Dual Parts of Speech
Sanskrit (Indo-Iranian); Arabic (Semitic);
Inuktitut (Eskimo-Aleut); Araucanian; Saami
(Uralic).
47
Humboldt stresses that classification according to ‘extension’ is based exclusively
on his collection of empirical data from languages with the dual. He insists, however, that
an accurate typology of languages with the dual must follow from general principles
independent of any empirical data. The last Humboldtian observation is particularly
important for this dissertation since a theory of the dual construed from independent
principles would allow to explain how the dual is represented in the speaker’s mind, and
account for directions of diachronic change in languages with dual number. A theoretical
framework which is not built on language-specific principles would allow us to account
for the emergence and disappearance of dual number cross-linguistically. I hope that this
dissertation provides such a theoretical framework which explains the emergence,
gradual loss, or retention of dual number in Slavic languages.11
After presenting his notional and extensional classifications of the dual – the
former based on the emergence of dual notion (Begriff) and its corresponding linguistic
forms and the latter based on the distribution of dual forms in the world’s languages,
Humboldt explores how the concept (Vorstellung) of dual number is represented in the
speaker’s mind. This is where Humboldt’s remarkable insight about conceptual structure
of the dual comes to light.
Humboldt argues that the dual should be conceived as a collective singular
realized by the numeral two rather than a concept arbitrarily corresponding to the numeral
two. Humboldt arrives at this concept of the dual by comparing it to the plural. On the
one hand, as a collective singular of two, a dual can be conceived as a plural in reference
11
The details of my proposal of the principle of Morphosyntactic Feature Economy in the Slavic dual are
presented in Chapter 5.
48
to the cardinality of one. On the other hand, a dual can be conceived as a singular, which
is a unified whole for the cardinality set of two. According to such mental representation,
the dual has the properties of both the singular and the plural.
This conception of the dual as a collective singular of two was an unprecedented
theoretical breakthrough that Humboldt presented in his 1827 lecture. Humboldt’s
conceptual theory of the dual has important implications for current theoretical
frameworks in which the dual is treated as a combination of the negative values of the
two number features - [±singular] and [±plural] (Hale 1997; Noyer 1997; Harbour 2008,
2011; Nevins 2006, 2008, 2011).12 Theories of markedness (Nevins 2011, Harbour 2011)
of dual number and its syncretism with the singular and plural naturally follow from such
treatment of the dual as having the properties of both the singular and plural.
Humboldt concludes his lecture “On the Dual Form” by making another
observation about a conceptual link between the categories of dual number and person.
He states that dual as a category of number is absolutely essential to the grammar of
language on the one hand. On the other hand, the dual is intrinsically connected to the
category of person since the 1st person plural inclusive is inherently dual. Humboldt
provides evidence of pronominal duals in Austronesian (Malay, Tagalog, Pampangan);
Sino-Tibetan (Tamang); and Carib (Chaima) language families to show that the
cardinality of two is tied to the inclusion or exclusion of the addressee in the plural as
well as specification of cardinality of two in the dual.
12
In Noyer's (1997), Nevins's (2006, 2008, 2011), and Harbour's (2008, 2011) work, the [±plural] feature
corresponds to the [±augmented]. Formal semantic definitions of the [±singular] and [±augmented] number
features are given in Chapter 4.
49
The conceptual connection between dual number and person pointed out by
Humboldt is of particular importance to this dissertation which investigates in detail
diachronic change in the pronominal dual in Slavic languages. Even though none of the
Slavic languages ever had an inclusive/exclusive distinction in the plural, dual/plural
syncretism that occurred over time in Slavic languages was essentially tied to the 2nd
person.13
Both the conceptual and extensional typological classifications of the dual
proposed by Humboldt were unprecedented in their intellectual and empirical scope.
However, they had some theoretical and empirical limitations. The conceptual
classification (Table 2.1) identifies different ‘sources’ of dual forms in various languages,
but does not provide an adequate explanation for abstract duality and its origin in
language. Humboldt’s extensional classification of the dual is not quite as empirically
encompassing as would have been desirable (Table 2.2).14
More importantly, Humboldt does not observe that the two typologies of the dual
show clear correspondences. The conceptual classification of the dual according to the
‘notion’ (Begriff) and the extensional classification according to the degree of extension
(Umfang) of dual forms in a language clearly correspond to each other (Table 2.3).
Languages of the 1st notional class which exhibit dual forms due to a distinction between
the speaker and the addressee are likely to have pronouns referring to a group of
cardinality of 2. Languages of the 2nd notional class that have dual forms based on the
13
Gamkrelidze and Ivanov (1984) present a different view. They propose that PIE personal pronouns and
consequently Proto Slavic distinguished two persons – 1st and 2nd and two numbers – singular and plural.
There was an inclusive ~ exclusive distinction in the 1st person plural which developed due to a binary
structure of PIE grammatical system.
14
This was likely due to the limitation of empirical resources that were available for Humboldt at that time.
50
idea of natural pairs are likely to have nouns referring to a group of cardinality of 2.
Languages of the 3rd notional class with dual pronouns and nouns will probably have dual
marked on all parts of speech. However, Humboldt does not point out these
correspondences.
Table 2.3 Humboldt's Notional, Extensional, and Conceptual Typologies of the Dual
‘Notion’ (Begriff)
‘Extension’ (Umfang)
‘Concept’
(Vorstellung)
(1) I vs. You
(1) dual pronouns
(2) natural pairs of
objects
(3) abstract duality
(2) dual nouns
Collective singular of
two
(3) dual parts of speech
One of the most important implications of Humboldt’s conceptual (Vorstellung)
theory of the dual is its mental representation as a collective singular of two. No other
comparative scholar of the 19th century was able to put forward such a novel
understanding of the concept of dual number. In this dissertation, I follow Humboldt’s
original insight on the conceptual representation of dual number, and its relation to the
singular and plural as shown in (Table 2.4). However, in my theoretical proposal of the
principle of Morphosyntactic Feature Economy in the Slavic dual I represent dual number
in terms of two morphosyntactic number features - [±singular] and [±augmented].
51
Table 2.4 Humboldt's Conceptual Typology of the Dual
‘Concept’ (Vorstellung)
Collective singular of two
Dual's Relation to
Singular
Dual relates to singular
as a unified whole (i.e.
cardinality 1)
Dual's Relation to
Plural
Dual relates to plural as
cardinality of 2 > 1
In his 1827 paper “On the Dual Form,” Humboldt was the first to draw linguists’
attention to the complexity of morphological, typological, and conceptual aspects of the
category of dual number. Humboldt’s innovative method combined a thorough empirical
investigation of morphological realization of the dual cross-linguistically with an
investigation of its conceptual representation. This novel method of linguistic analysis
allowed Humboldt to realize empirical and conceptual challenges that dual number poses
for both comparative and theoretical linguistics and propose three different typologies of
the dual: notional (Begriff), extensional (Umfang), and conceptual (Vorstellung).
Although neither the notional nor extensional classifications of the dual were
devoid of some drawbacks that I pointed out above, Humboldt’s conceptual classification
is one of great significance and far-reaching implications. According to his conceptual
typology, the dual is represented as a collective singular of the number two. This
conceptual representation emphasizes that a dual can be understood as a singular since it
is a unified whole restricted to two entities. At the same time, a dual can be thought of a
plural since it denotes two entities compared to one.
Humboldt’s conceptual representation of the dual as both singular and plural
allows one to understand the ‘dual’ nature of the dual and test its representation crosslinguistically to discuss linguistic universals concerning number systems. Thus
52
conceived, the dual is no longer considered “an exotic luxury” of some languages, but
rather a quite natural phenomenon that languages had in the past and continue to have in
the present (Humboldt 1997:136).
53
2.2. Jespersen’s Typology of the Dual
“Number might appear to be one of the simplest natural categories,
as simple as ‘two and two are four.’ Yet on closer inspection
it presents a great many difficulties, both logical and linguistic.”
Otto Jespersen (1965:188)
At the beginning of the 20th century, the problem of typological classification of
languages with a dual was approached by a Danish linguist, Otto Jespersen. As opposed
to Humboldt, Jespersen tried to incorporate the idea of linguistic change into his typology
of languages with a dual. Jespersen observed that nominal duals cross-linguistically were
not immune to diachronic change. Some languages with nouns denoting natural pairs
employed plural marking instead of the expected dual, whereas other languages preserved
dual marking for natural pair objects. The concept of diachronic change in nominal duals
was captivating for Jespersen who tried to capture it in his typology.
Jespersen (1965) divides all languages with a nominal dual into two classes: 1).
those where a dual is used to refer to duality of two referents, and a plural is used to mark
natural pairs whose duality is semantically obvious; 2). those where the dual only is used
to mark objects naturally found in pairs (Table 2.5).
Table 2.5 Jespersen's Typology of the Dual (1965:205)
Dual for Two Referents
Plural for Natural Pairs
sg. nuna ‘land’ Inuk.
du. nuna-k
pl. nuna-t
pl. issai ‘his eyes’ Inuk.
pl. siutai ‘his ears’
pl. talê ‘his arms’
Dual for Natural Pairs
osse ‘the eyes’ anc.Gr.
aksī ‘the eyes’ Skr.
aki ‘the eyes’ O.Lith.
54
Jespersen observes that many living languages with a nominal dual, such as
Inuktitut, mark the duality of two objects by using a distinct dual suffix, but indicate
duality of natural pairs such as eyes, arms, ears, etc with a plural marker. He does not
explicitly state whether plural instead of the dual marking in Inuktitut nouns denoting
natural pairs is a result of diachronic change.15 In the majority of Slavic languages, plural
marking ousted dual in nouns that refer to natural pairs. For example, dual suffixes
marked on nouns denoting natural pairs were replaced with the plural in Russian,
Belorusian, Ukrainian, Polish, Czech, Slovak, Bulgarian, Macedonian, Slovenian, and
some dialects of Upper Sorbian (Stone 1993:614, Sussex and Cubberley 2006:225).
Jespersen notes that many extinct languages, such as ancient Greek and Sanskrit,
as well as languages in earlier stages of their development, such as Old Lithuanian,
preserved a dual marker for naturally paired objects, but lost it in subsequent stages of
language change. Regarding living Indo-European languages, Jespersen observes that
only Sorbian and Slovenian continued to preserve the dual in their grammar. There is a
degree of ambiguity in this observation. If it refers to duals of ‘regular’ (not natural pairs)
Slovene and Sorbian nouns, it is quite true that they have morphologically distinct
suffixes in the dual. If this comment is directed towards duals of paired nouns, it does not
hold true as stated above.
In contrast to Meillet (1924) who considered the disappearance of the dual as a
movement from a 'primitive' mentality to more abstract thinking, Jespersen views
linguistic change as simplification in the grammar. He regards reduction of any
15
One might make an assumption that diachronically the dual got replaced by the plural in Inuktitut nouns
denoting natural pairs.
55
unnecessary distinctions, such as the dual, as a progressive phenomenon. Jespersen states
that is hard to show in detail a causal connection between the loss of the dual and
advancement of civilization where it was used. For example, he notes that the Greek dual
was kept the longest in continental Greece while it was lost in the colonies where the
civilization was more advanced.
Jespersen remarks that the use of dual forms can be quite restricted. He notes that
in Greek dual forms were used for poetic reasons. For example, in Homer’s works dual
forms are quite frequent although they are “an artificial archaism” (Jespersen 1965:206).
Some Old Germanic languages, such as Gothic also exhibit restricted distribution of the
dual. In Gothic, pronominal dual is found only in the 1st and 2nd person.
Despite the lack of a principled explanation for the loss of dual in many IndoEuropean languages, Jespersen makes an insightful observation about different ‘traces’ of
dual number that are left behind. In some languages, an ‘old’ dual form tends to be
syncretic with the plural such as in Old Norse, whereas in others, it is syncretic with the
singular, such as in Russian (cf. 13-14). As Jespersen (1965:207) notes, the Old Norse
pronoun Þau was a dual form which was also used as a plural (16). In Old Russian, the
suffix -a used to be a dual inflection marked on nouns with *a, *ja stems. In
Contemporary Standard Russian, a former dual suffix -a is a genitive singular of
masculine nouns modified by quantified DPs with paucal numerals dva, tri, četyre (two,
three, four) (17).
(16)
Old Norse
Þau
‘they two’
56
(Jespersen 1965:207)
(17)
Contemporary Standard Russian
dv-a
mužik-a
two-DU.MASC.NOM peasant- SG.MASC.GEN
‘two peasants’
Jespersen’s contribution to the problem of classifying languages with dual number
is significant in that he tried to capture the phenomenon of diachronic change in his
linguistic typology. The disappearance of the dual in many Indo-European languages was
well noted by Jespersen, but was left unaccounted for. Since Jespersen’s typology
focused only on nominal duals, but did not consider pronominal duals, it had no
predictive power to account for linguistic change in a wide variety of Indo-European and
non-Indo-European languages. In addition to his typology of nominal duals, Jespersen
made a remarkable observation about dual/plural and dual/singular syncretism, but his
typological framework could not account for it.
57
2.3. Plank's Typology of the Dual: Humboldt Revisited
Humboldt’s (1827) typological classifications of the dual remained influential in
the realm of typology and comparative linguistics. However, his classifications were
devoid conceptual and empirical drawbacks. Some of the problems in Humboldt’s
classifications of the dual were taken up more than a century and a half later by another
German typologist, Frans Plank. Plank’s (1989) main aim in critiquing Humboldt’s
typologies was to test descriptive generalizations about the dual and to propose novel
solutions where these generalizations fail.
In what follows, I focus on: (i) Plank’s revised extensional typology of the dual,
(ii) his novel typology of languages with pronominal dual, and (iii) his typology of verbal
agreement with dual nouns and pronouns. These three aspects are most relevant to the
topic of this dissertation since it investigates diachronic change in the pronominal
paradigm and verbal agreement system in the Slavic languages.
Humboldt’s original extensional classification of the dual included three classes
of languages: (1) with dual pronouns, (2) with dual nouns, and (3) with dual parts of
speech (Table 2.2). The major drawback of this typology is the lack of implicational
relations that can be established between dual nouns, dual pronouns, and dual agreement
forms. Plank (1989) proposes to solve this problem by suggesting a new typology of
languages with dual number. The novelty of Plank’s approach is in assigning a binary [±]
value to each of the three Humboldtian extensional classes – (1) dual nouns, (2) dual
pronouns, and (3) dual agreement forms. Bivalence of a feature for each class gives rise
to eight theoretically plausible language types with dual number (Table 2.6).
58
Table 2.6 Plank's Typology of Languages with Dual Number (1989:299)
a
b
ab
c
d
e
f
g
Personal
Pronouns
+
-
+
+
-
-
+
-
Nouns
-
+
+
+
-
+
-
-
Non-Pronominal
Agreement
Forms
-
-
-
+
-
+
+
+
Language types (a), (b), and (c) follow from Humboldt’s extensional typology
whereas types (ab), (d), (e), (f), and (g) are suggested by Plank (1989). Type (ab) has
both nominal and pronominal dual, but lacks dual agreement. Plank notes that languages
of this type are quite rare, and provides no examples (Table 2.7). Type (g) is a complete
opposite of type (ab) in that it has dual agreement forms, but lacks dual nouns and
pronouns. Plank (1989) suggests that type (g) might not have any members entirely, or
might be represented by Chamorro (Austronesian) and Hupa (Na-Dene) (Table 2.7). A
theoretically possible type (d) does not have any dual forms in all of the three categories:
nouns, pronouns, and agreement forms.
Table 2.7 Language Examples for Plank's Typology (1989:299)
a
Languages
b
ProtoSemitic
Hopi Irish
Polish
earlier
stages of
Slavonic
ab
c
IndoEuropean
Semitic
d
-
e
f
g
North
Gothic Chamorro
Semitic
Siroi
Hupa
Akkadian Kewa
Eastern
Dizi
Libyan
Arabic
59
Types (e) and (f) are interesting in that they both have dual agreement, but are
asymmetric in the presence or absence of dual nouns or pronouns (Table 2.6). Type (e)
has nominal dual whereas type (f) has pronominal dual. Type (e) is attested only in two
languages: North Semitic Akkadian (extinct) and Eastern Lybian Arabic (Table 2.7).
Pronominal dual languages of type (f) are more attested than type (e) and include Gothic
(Germanic), Siroi and Kewa (Trans-New Guinea), and Dizi (Afro-Asiatic) (Table 2.7).
In contrast to Humboldt’s extensional typology that has 3 extensional classes,
Plank’s (1989) typology with 8 types provides a more precise account of diachronic
change in languages with dual number. Plank (1989) argues that the existence of
language types (e) and (f) accounts for diachronic change in the category of number in a
way that was impossible in Humboldt’s extensional classification.
Recall that in Humboldt’s extensional classification, languages of type (c) where
dual is marked on nouns, pronouns, and agreement forms could only change to languages
of either type (a) or (b) with pronominal or nominal dual but crucially without dual
agreement (Table 2.8). It is not clear in Humboldt's typology why languages of type of
(a) and (b) have no agreement with nominal or pronominal arguments referring to dual
number.
Table 2.8 Language Change in Humboldt's Typology (1827)
Language Type
type (a)
Dual Pronouns
yes✓
Dual Nouns
no✓
Dual Agreement
no
type (b)
no
yes
no
type (c)
yes
yes✓
yes
60
In contrast to Humboldt (1827), Plank (1989) suggests that a language of type (c),
where dual is marked on all parts of speech, can shift to either type (e) or (f). That is, a
language of type (c) while retaining dual agreement could either lose a pronominal dual
and become a language of type (e), or it could lose a nominal dual and transfer to type (f)
(Table 2.9). Crucially, languages of both type (e) and (f) keep dual agreement.
Table 2.9 Language Change in Plank's Typology (1989)
Language Type
type (c)
Dual Pronouns
yes✓
Dual Nouns
yes✓
Dual Agreement
yes✓
type (e)
no
yes
yes✓
type (f)
yes✓
no
yes✓
Humboldt’s typology could not predict a pattern of diachronic change where
languages of type (c) could shift to an (e) or (f) type while keeping dual agreement. The
reason for that is simple. Humboldt did not recognize that each category – nouns,
pronouns, and as well as agreement can have a [±] value. For Humboldt, agreement is
monovalent; that is, it has a negative value [-agr]. Plank’s typology represents an advance
over Humboldt’s. It provides a better account of diachronic change since Plank’s
agreement feature is bivalent [±agr] and can predict that type (c) languages can lose
either pronominal or nominal dual while having a positive [+agr] value for dual
agreement.
Diachronic change in the category of dual number lends itself for an interesting
typological investigation especially in the Slavic languages where dual number has
undergone different changes in different subgroups of Slavic. Following Plank's (1989)
61
typological classification, I propose my own diachronic typology of Slavic languages
with dual number (Table 2.10). Proto-Slavic (2000-1500 BC - 5th AD) was a language of
type (c). It had dual pronouns, nouns, and a rich system of verbal, adjectival, quantifier,
and numeral agreement. Old Church Slavic was also a type (c) language with dual nouns,
pronouns, and agreement forms. All of its South descendants – modern Croatian, Serbian,
Bosnian, Bulgarian, and Macedonian, besides Slovenian, eventually lost dual pronouns,
nouns, and agreement, and gradually became languages of type (d).
Table 2.10 Diachronic Typology of the Dual in Slavic Languages
Language Type
c (+dual N, +dual Pro,
+dual Agr)
d (-dual N, -dual Pro, -dual Agr)
Proto-Slavic
Old Church Slavic
Old Russian
Slovenian
Upper Sorbian
Lower Sorbian
Croatian
Serbian
Bosnian
Bulgarian
Macedonian
Russian
Belorusian
Ukrainian
Polish
Czech
Slovak
Slovenian is the only representative in the South Slavic subgroup that has
remained a type (c) language (Table 2.10). It has generally preserved the dual in nouns,
pronouns, and agreement forms. The only exception are nouns denoting natural pairs,
such as noge (‘feet’), roke (‘hands’), oči (‘eyes’) which occur only in the plural. The
remaining of Slovenian nouns have a special dual suffix that signifies two entities.
Slovenian pronouns have a most extensive dual paradigm where dual forms are
62
morphologically distinct from the singular and plural. Slovenian agreement forms in
verbal inflection also differentiate the dual in the 1st, 2nd, and 3rd persons.
The majority of West Slavic languages – Polish, Czech, and Slovak – had been
originally type (c) languages, but they lost dual number in the course of history thus
becoming languages of type (d) (Table 2.10). Among the languages of the West Slavic
subgroup, Upper and Lower Sorbian are the two type (c) languages that have preserved
the dual in pronouns, nouns, and agreement forms (Table 2.10). In the East Slavic branch,
Old Russian (11th century) also started out as a type (c) language. Over time, its East
Slavic descendants – modern Russian, Belorusian, and Ukrainian have all gradually
become type (d) languages (Table 2.10).
Besides a new extensional typology of languages with the dual, Plank (1989) also
proposes a novel typology of languages with pronominal dual based on the person
restriction. He suggests that pronominal dual will vary depending on the inclusion or
exclusion of the 1st, 2nd, or 3rd person giving rise to eight theoretical possibilities (Table
2.11). This typology of pronominal dual is based on type (a) which is purely pronominal
without dual nouns or dual agreement. Types (ab), (c), and (f) are excluded from this
typology since they have either dual nouns or agreement in addition to dual pronouns.
Plank (1989) supports pronominal dual typology with empirically attested examples
for each language type (Table 2.12). While I will not contend empirical validity of all of
these examples, I will challenge the Slavic examples as they do not fit in Plank's
pronominal typology. Based on my diachronic corpus studies, I provide a different
typology of the pronominal dual in Slavic.
63
Table 2.11 Plank's Typology of Pronominal Dual Based on Person Restriction
(1989:301)
1st Person
a1
+
a2
+
a3
+
a4
+
a5
-
a6
-
a7
-
a8
-
2nd Person
+
-
+
-
+
+
-
-
3rd Person
+
-
-
+
+
-
+
-
Plank (1989) asserts that there are a lot of languages of type (a1) where the dual is
not restricted to any person. These languages belong to a diverse genetic pool including
Polynesian, Australian, Papuan, Austro-Asiatic, Samoyedic, Chukotko-Kamchatkan,
Khoisan, and Indo-European (Table 2.12).
In Plank’s pronominal typology, languages of type (a2) that restrict pronominal
dual to the 1st person are less numerous. They are represented by Penutian, Siouan, UtoAztecan, Pama-Nyungan, Austronesian, Chadic, and Niger-Congo language families
(Table 2.12). Restriction of the dual to the 1st and 2nd person as in type (a3) is attested in a
now extinct Gothic (Old Germanic), in Slovenian at some point in its history, and
Huavean, Athapascan, and Australian languages. Restriction of the pronominal dual of
type (a5) to the 2nd and 3rd person is comparatively rare (Table 2.12). There are only three
languages that are attested for (a5) type: Classical Arabic, Aleut, and possibly ancient
Greek at some point in its history (Table 2.12).
There is only one example of type (a6), an Afro-Asiatic language Dizi, which
restricts the dual to the 2nd person (Table 2.12). Restriction of the pronominal dual to the
3rd person as in type (a7) is extremely rare. There only two languages that are attested for
64
this type: some varieties of South Arabic and a now extinct Gulf language, Tunica (Table
2.12).
Among the Indo-European languages in Plank’s pronominal dual typology, there
are three Slavic languages: Old Church Slavic, Slovenian, and Kashubian whose
placement in Plank's typology appears to be incorrect. Based on diachronic data from the
corpus analysis of Old Church Slavic manuscripts, I propose a different typology of the
pronominal dual in Slavic (Table 2.13).
Table 2.12 Languages in Plank's Pronominal Typology (1989:301)
a1
a2
a3
Zuni
Gothic
Languages/ Polynesian
Australian
Maidu
Slovenian
Language
Papuan
Dakota
Huave
Families
Munda
Southern Chiricahua
Ob-Ugric
Pauite
Apache
Lappic
Yidiny
Hopi
Samoyedic
Tagalog
Western
Koryak
Pangasinan
Desert
Sanskrit
Gude
ancient
Margi
Greek
Duru
Old
Church
Slavic
Kashubian
Lithuanian
Nama
Hottentot
Khoisan
Old
Egyptian
a
4
a5
a6
a7
classical
Arabic
Aleut
ancient
Greek
Dizi
South
Arabic
Tunica
a
8
65
Table 2.13 Typology of the Pronominal Dual in Slavic
Pronominal Type
Slavic Languages
a1
Slovenian
Upper- and Lower
Sorbian
1st Dual ✓
2nd Dual ✓
3rd Dual ✓
a2
Kashubian
a3
Old Church Slavic
1st Dual ✓
2nd Dual No
3rd Dual No
1st Dual ✓
2nd Dual ✓
3rd Dual No
Plank claims that Old Church Slavic belongs to type (a1). However, this claim is
disproven by my diachronic corpus studies. As my analysis of the Old Church Slavic
manuscripts showed, pronominal dual in Old Church Slavic is restricted only to the 1st
and 2nd persons. That Old Church Slavic did not have a dedicated 3rd person pronoun is
also confirmed by many Slavic scholars (Gasparov 2001, Lunt 2001, Krivčik and
Možejko 1985, Xaburgaev 1986, Ivanova 1977).16 Therefore, an absence of the 3rd person
pronoun places Old Church Slavic in type (a3) (Table 2.13).
Slovenian is classified by Plank (1989) as a language with pronominal dual of
type (a3), where dual number is differentiated only in the 1st and 2nd persons of its
pronominal paradigm (Table 2.12, p.24). This classification is incorrect since dual
number in Slovenian is attested in all three persons in the pronouns (Derganc 1988, 2003;
Jakop 2008). Therefore, Slovenian pronominal dual belongs to type (a1) (Table 2.13).
Kashubian differentiates the dual only in the 1st person in its pronominal paradigm
(Stone1993:773). Therefore, it belongs to (a2) type, but not (a1) as suggested by Plank
(Table 2.13).
Since there is no 3rd person pronoun in Old Church Slavic, demonstrative pronouns were used in contexts
referring to third party entities.
16
66
Based on empirical evidence from various languages, Plank (1989) draws the
following implicational universal about the person restriction in the pronominal dual:
(19)
If only one person does not differentiate a dual, it will not be the 2nd (305).
This implicational universal explains why languages of type (a4) are unattested. In type
(a4), dual is differentiated in the 1st and 3rd persons, but not in the 2nd (Table 2.11).
Exclusion of the 2nd person contradicts Plank’s universal (19). If only one person does
not differentiate a dual, it cannot be the 2nd. In other words, it can either be the 1st or the
3rd person.
As Plank (1989) argues, restrictions of the dual to the 2nd person-only and 3rd
person-only patterns are attested quite rarely in languages. As shown in Table 2.12, Dizi
(Afro-Asiatic) is the only attested language of type (a6) which restricts its pronominal
dual to the 2nd person. South Arabic and a now extinct Tunica are only two attested
languages of type (a7) which have pronominal dual only in the 3rd person. These
typological observations allow Plank to conclude that restrictions to the 1st person are
much better attested in a greater number and variety of languages. More precisely, the 1st
person is much likelier to join the 2nd person whenever the pronominal dual is restricted
to two persons. Plank (1989) summarizes these observations in the following
implicational generalizations:
(20)
If only one person differentiates a dual, it will very likely be the 1st rather than the
2nd and the 3rd (305).
67
(21)
If only two persons differentiate a dual, the 1st is much likelier to be one of them
than the 3rd (305).
Implicational generalization (16) states that when one person differentiates a dual
in a pronominal paradigm, the 1st person wins over the 2nd and 3rd as suggested by
Plank’s data. Implicational universal (21) suggests that when two persons differentiate a
dual, the 1st person is preferred over the 3rd. Rarity of languages of type (a5) lends
support to Plank’s universal (21). There are only three languages of type (a5), classical
Arabic, Aleut, and ancient Greek where the dual is restricted to the 2nd and 3rd persons.
Plank’s typology of pronominal duals reveals a close correlation between dual
number and the 2nd person. As universal in (19) states if a language does not differentiate
a dual in one person, it cannot be the 2nd person. This universal holds true as evidenced
by the absence of any attested languages of type (a4 Table 2.12). Plank’s (1989) typology
of pronominal duals confirms that the 2nd person has a special status. The person of the
addressee is more ‘privileged’ than the 3rd and less ‘privileged’ than the 1st. This
correlation is reflected in the hierarchy of persons (Figure 2.1) which has also been
recognized by other scholars (Benveniste 1971, Jespersen 1965, Forschheimer 1953,
Noyer 1997, Corbett 2000).
Figure 2.1 Universal Hierarchy of Person Noyer (1997:114)
1st >2nd >3rd
68
Plank’s universal (19) has special significance for the typology of Slavic
languages since the Slavic dual is inherently connected to the 2nd person. The
phenomenon of dual/plural pronominal syncretism in Old Church Slavic and Old Russian
occurs specifically in the 2nd, but not in the 1st or 3rd person.
Above, I have presented two typologies proposed by Plank’s (1989): (i) an
extensional typology of languages with dual number (Table 2.6) and (ii) a pronominal
typology of duals constrained by the person restriction (Table 2.11). Now I will turn to
(iii) Plank’s typology of verbal agreement which is especially relevant for Slavic
pronominal agreement.
Plank (1989) notes that another area left unexplored by Humboldt’s (1927) work
on the dual is in the domain of verbal agreement. Plank argues that languages of type (c)
and (f) in the extensional typology of the dual (Table 2.6) can be further sub-classified
into two types according to person restrictions in agreement and pronominal paradigms:
1). languages where all persons in non-pronominal agreement paradigms differentiate a
dual from a singular and plural; 2). languages without a dual for any person in their nonpronominal agreement forms, although pronouns themselves differentiate a dual in some
or all persons (Table 2.14).
As Table 2.14 shows, languages of type (c) and (f) differ according to the person
restriction in the pronominal and agreement forms. In languages of type (c) exemplified
by Vedic Sanskrit, Old Church Slavic, Ugaritic, Aleut, Tunica, Siroi, and Kewa all
persons differentiate a dual from a singular and plural in agreement forms whereas not all
persons are marked for dual in the pronouns (Table 2.15). Languages of type (f), such as
69
some Old Germanic languages, are not marked for dual in any person in their verbal
agreement (Table 2.15).
Table 2.14 Plank's Verbal Agreement Typology (1989:306)
Personal
Pronouns
Agreement Forms
Presence of
Agreement
type (c) +dual Pro,
+dual N, +dual Agr
1st Dual
2nd Dual
3rd Dual
1st Dual ✓
2nd Dual ✓
3rd Dual ✓
Yes
Yes
yes
type (f) +dual Pro,
-dual N, +dual Arg
1st Dual
2nd Dual
3rd Dual
1st Dual
2nd Dual
3rd Dual
No
No
No
Table 2.15 Languages in Plank's Verbal Agreement Typology (1989:306)
Languages
type (c)
Verbal Agreement for 1st, 2nd, 3rd Person Dual
✓
Old Church Slavic
Vedic Sanskrit
Ugaritic (extinct)
Aleut
Tunica (extinct)
Siroi
Kewa
type (f)
yes
yes
yes
yes
yes
yes
yes
Old Germanic (except
Gothic)
No
Plank (1989:307) claims that there is an implicational relation between agreement
forms and dual number which is formulated as follows:
70
(22)
If a dual, or any other number, is differentiated for a particular person of nonpronominal agreement forms, there will also be a dual, or other number, for this
person in pronouns.
This implicational generalization between dual number in agreement forms and pronouns
does not hold true as attested by my own diachronic corpus studies of Old Church Slavic
(Table 2.16). In Old Church Slavic, all persons are marked for dual in agreement forms,
but only the 1st and 2nd persons differentiate a dual in a pronominal paradigm (Gasparov
2001, Lunt 2001, Krivčik & Možejko 1985, Xaburgaev 1986, Ivanova 1977). Plank
admits a few other exceptions to this universal that occur in Tunica and Aleut, but does
not recognize that Old Church Slavic and Kashubian analyzed in this dissertation do not
follow his typology.
Slovenian, Upper and Lower Sorbian follow Plank’s universal (18) since the dual
is differentiated in all persons of agreement forms and therefore, is present in all persons
in the pronouns. Kashubian is an interesting example of a West Slavic language with a
pronominal dual only in the 1st person and an absence of verbal agreement with the dual
in the 2nd and 3rd persons. Verbal agreement forms make only a singular/plural
distinction (Stone 1993:773,776).
Plank’s (1989) revisiting Humboldt’s (1827) lecture “On the Dual Form” comes
as no surprise. As a nineteenth-century linguist, Humboldt proposed a very
comprehensive typology of the dual approached from a an extensional, notional, and
conceptual perspectives. However, Humboldt did not notice the existence of binary [±]
values for each extensional class - dual pronouns, dual nouns, and dual agreement forms.
71
Plank’s (1989) recognized the bivalence of each category and proposed eight extensional
classes instead of original three.
Table 2.16 Personal Pronouns and Verbal Agreement in Slavic
Language
Personal Pronouns
Agreement Forms
Presence of
Agreement
type (c) Old
Church Slavic
1st Dual
2nd Dual
3rd Dual
1st Dual ✓
2nd Dual ✓
3rd Dual ✓
Yes
Yes
No
type (c)
Slovenian
1st Dual ✓
2nd Dual ✓
3rd Dual ✓
1st Dual ✓
2nd Dual ✓
3rd Dual ✓
Yes
Yes
Yes
type (c) Upperand Lower
Sorbian
1st Dual ✓
2nd Dual ✓
3rd Dual ✓
1st Dual ✓
2nd Dual ✓
3rd Dual ✓
Yes
Yes
Yes
type (c)
Kashubian
1st Dual
2nd Dual
3rd Dual
1st Dual
2nd Dual
3rd Dual
No
No
Another novelty present in Plank’s (1989) work is a typology of languages with
pronominal dual according to the person restriction. By assigning each of the three
persons in a pronominal paradigm a [±] value, Plank further arrived at eight pronominal
types which classified a variety of attested languages with pronominal dual marked in
different persons. The same bivalent approach allowed Plank to establish a typology of
dual pronouns and verbal agreement. Both pronominal and verbal agreement typologies
72
go beyond Humboldt’s ideas about the dual due to Plank’s breakthrough approach to
recognizing bivalence of grammatical categories.
A major drawback of Plank’s (1989) typology is an absence of typological
correlation between the dual, singular, and plural in a singular/dual/plural number system.
In order to establish such a correlation, it is necessary to consider the place of the dual
within a singular/dual/plural number system drawing on empirical evidence from a
variety of typologically unrelated languages.
Considering the dual in relation to the singular and plural in a singular/dual/plural
number system opens up an avenue to discover its conceptual composition. Therefore, the
next step in a typological and theoretical investigation of the dual is to recognize
bivalence of features that comprise the grammatical category of number, which was
successfully achieved by a number of linguists for a wide variety of languages (Hale
1997; Noyer 1997; Harley and Ritter 2002; Cowper 2005; Nevins 2006, 2011; Harbour
2008). In this dissertation, I take a step further and analyze morphosyntactic
representation of the dual in the South, West, and East Slavic languages.
73
2.4. Corbett's Constraint on Ranges of the Dual and Plural
Previous scholars (Humboldt 1827, Jespersen 1965, Plank 1989) who proposed
different typologies of dual number did take into consideration one very important factor.
They did not discuss patterns of constrains on dual number in various typologically
unrelated languages. This task was taken up by a British typologist, Greville Corbett, in
his seminal monograph Number (2000).
In his book Number, Corbett (2000) makes three proposals relevant for this
dissertation as they relate to dual number in Slavic. The first proposal concerns languages
with a singular/dual/plural number system where the dual is constrained according to the
Animacy Hierarchy. The second proposal focuses on diachronic change in number
systems, particularly on the rise and fall of the dual in various typologically unrelated
languages including the Slavic. The third proposal concerns the directional dependencies
between number and other morpho-syntactic features, such as person, case, and gender.
In what follows, I present these three proposals and explain how they relate to the Slavic
dual.
I will briefly outline Corbett’s 1st proposal about the Animacy Hierarchy as a
constraint for number systems which include the dual. According to the Animacy
Hierarchy, the more ‘animate’ a nominal is, the more likely it is to mark dual (Figure
2.2). According to Corbett (2000), the top members of the hierarchy – the 1st and 2nd
person pronouns are more likely to be marked for dual number than the 3rd person, kin,
human, animate, and inanimate nouns respectively.
74
Figure 2.2 The Animacy Hierarchy (Corbett 2000:56)
speaker > addressee > 3rd person > kin > human > animate > inanimate
(1st person (2nd person
pronouns) (pronouns)
The reason for Corbett's proposal of Animacy Hierarchy comes from inadequacy
of the Number Hierarchy (Greenberg 1963) as a general constraint for possible number
systems (Figure 2.3). Corbett (2000:39) contends that the Number Hierarchy derived
from Greenberg’s Universal 34 (No language has a trial number unless it has a dual. No
language has a dual unless it has a plural (1963:94)) has two problems. The first problem
is that the Number Hierarchy does not account for number systems which have a
paucal.17 The second problem is that the Number Hierarchy does not account for
languages with facultative (optional) numbers.
Figure 2.3 Greenberg's Number Hierarchy
singular > plural > dual > trial
While I will not comment on the 1st reason for rejecting the Number Hierarchy
due to the absence of paucal number in Slavic, I will address Corbett’s 2nd objection the
Number Hierarchy based on optionality of dual number in Slovenian. Corbett (2000:43)
claims that the Slovenian dual is optional and cites a plural form of the noun noge (‘legs’)
where the dual would be expected. This claim of the optionality of the Slovenian dual is
challenged by Jakop (2008:3) and Derganc (2003:172) who explain that the noun ‘legs’
belongs to a special class of nouns denoting natural pairs which occur only in the plural.
17
I adopt Corbett’s definition of paucal which refers to “a small number of distinct real world entities with
“no specific upper bound” and whose “lower bound will vary according to the system in which it is
embedded” (2000:22).
75
In the rest of the Slovenian nouns, the dual is not optional, but obligatory. Therefore, the
data from Slovenian does not provide support for the inadequacy of the Number
Hierarchy.
Regarding the distribution of number values in languages with a
singular/dual/plural system, Corbett (2000) proposes a Constraint on Ranges for Different
Number Values which governs the ranges for dual and plural numbers. This constraint
follows from the Animacy Hierarchy, and predicts two possible patterns of distribution of
the dual and plural while ruling out a third pattern of distribution which is not attested. In
the 1st pattern, the ranges of the dual and plural are identical (Figure 2.4). In the 2nd
pattern, the dual has a smaller range than the plural (Figure 2.5). A language, in which the
distribution of the dual is greater than that of the plural is impossible (Figure 2.6).18
Figure 2.4 The 1st Pattern of Ranges of Dual and Plural in Language X
1st > 2nd > 3rd > kin > human > animate > inanimate
Range of pl ***********************************************
Range of du ***********************************************
Figure 2.5 The 2nd Pattern of Ranges of Dual and Plural in Language X
1st > 2nd > 3rd > kin > human > animate > inanimate
Range of pl ***********************************************
Range of du ************
Figure 2.6 An Impossible Pattern of Dual and Plural in Language X
1st > 2nd > 3rd > kin > human > animate > inanimate
Range of pl ************
Range of du ******************************
18
In the patterns of ranges of the dual and plural shown in (Figures 2. 4-2.6), the singular is not shown.
Corbett assumes that “the singular is taken as given, being implied by the opposition with the plural”
(2000:90).
76
Although, Corbett’s two patterns of ranges of dual and plural descriptively might
account for the distribution of Slavic dual vs. plural, they do not explain why the ranges
the way they are. It is evident that ranges of the Slavic dual are going to be smaller than
those of the plural because i). the dual is a more marked number value compared to the
plural which follows from Greenberg’s Universal 34 and the Number Hierarchy; ii). the
dual is used less frequently than the plural.
More importantly, the two patterns proposed by Corbett (2000) do not provide a
principled explanation for the direction of diachronic change in the Slavic dual over time.
The Constraint on Ranges for Different Number Values does not give a deep theoretical
explanation why i).the Slavic dual was preserved in Slovenian, Upper and Lower
Sorbian, and Kashubian, and ii). why a singular/dual/plural number system in Old Church
Slavic and Old Russian changed to a singular/plural number system in the majority of
their descendant modern Slavic languages.
Corbett’s proposal on the rise and fall of the dual is especially relevant to this
dissertation. Corbett attributes the rise of dual number in a singular/dual/plural system to
the numeral ‘two’ based on the evidence from Lihir (Austronesian) whose dual, trial, and
paucal can be traced back to the numerals ‘two,’ ‘three,’ and ‘four.’ Corbett (200:267)
further strengthens his argument about the source of the dual by citing evidence from
Slovenian and Breton which ‘renewed’ their duals by adding the numeral ‘two.’19
19
Corbett (2000:268) claims that Breton, has a ‘new’ dual based on the numeral daou (‘two’). This claim is
challenged by Press (1986) and Heinecke (1998) who argue that Breton does not have a morphological
dual, but simply a derivational form with the meaning “double” or “a pair of.” A derivational form with a
dual meaning is formed by adding the numeral ‘two’ only to nouns denoting body parts (1). The reason
why Breton does not have a genuine dual comes from the fact that a derived noun daou-N can take a plural
77
Slovenian is an extremely interesting example of the ‘renewal’ of the dual in the
pronouns. By the 16th century, the Slovenian dual was weakened, and the 1st and 2nd
person plural pronouns mi and vi had begun to be used in dual contexts with dual
agreement markers on the verb (Jakop 2008:57). To resolve the problem of the
syncretism (identity of morphological form) between the plural and dual, the dual was
‘innovated’ in the 1st, 2nd, and 3rd person by the addition of the numeral dva (‘two’)
resulting in ‘new’ dual forms (Table 2.17).
Table 2.17 Dual Pronouns in Slovenian (Jakop 2008:59)
masc
mídva
vídva
onadva
1 du
2 du
3 du
fem/neut
Médve
Védve
onidve ~ onédve/onidve
The loss of the Slavic dual and its morphosyntactic traces in Russian and SerboCroatian are well noted by Corbett (2000). Agreement is one area where morphosyntactic
remnants of the Slavic dual are especially evident. In Contemporary Standard Russian,
noun phrases quantified by the paucal numerals dva (‘two’), tri ( 'three'), četyre ('four')
require a noun to appear in the genitive singular (23). Corbett claims that the genitive
suffix (2). Therefore, Breton clearly does not have a dual, but rather a bi-morphemic derived noun confined
to body parts that come in natural pairs.
(1) daou-lagad
two-eye
‘eyes’
Breton
(2) daou-lagad-où
two-eye-PL
‘pairs of eyes’
Breton
(Press 1986:71)
78
singular suffix -a dates back to the original dual suffix -a used with some Old Russian
masculine nouns. Corbett's claim is confirmed by my own Old Russian data (24). In (24),
the Old Russian masculine noun brother, quantified by the numeral two is marked by the
dual suffix -a.
(23)
Contemporary Standard Russian
dva brat-a
two table.GEN.SG.
'two brothers'
(24)
Old Russian
vid- ĕ
dǔv-a
brat-a
see-3rd.past two-masc.du.acc brother-masc.du.acc
'He saw two brothers.'
(The Ostromir Gospel 60.1)
Corbett (2000) further argues that as in Russian, in Serbo-Croatian, masculine
nouns modified by the numerals two, three, and four also appear in a special, so-called
'count form' which is a 'remnant' of the dual (25). Similar to Russian, the noun man is
marked by the genitive singular suffix -a. Corbett further notes that attributive modifiers,
such as dobr-a, are obligatorily marked by the 'count form' -a via agreement with the
noun. The agreement suffix -a on the adjective is not the same as the genitive form which
might be expected. Corbett concludes that the 'count' suffix -a on the attributive modifier
is due to the presence of the numeral two.
(25)
Serbo-Croatian
dva dobr-a
two good-COUNT
‘two good men'
čovek-a
man-GEN.SG (= COUNT)
(Corbett 2000:270)
79
In sum, Corbett's (2000) observations regarding the morphosyntactic remnants
of diachronic changes in the Slavic dual are descriptively correct. However, his proposal,
the Constraint on Ranges of Dual and Plural Number, cannot account for the direction of
diachronic change in the Slavic dual. Corbett's typological observations cannot account
why the majority of the Slavic languages including Russian and Kashubian developed a
singular/plural number system with the exception of Slovenian, Upper, and Lower
Sorbian which maintained a singular/dual/plural system.
80
2.5. Cysouw: Dual as a Restricted Group
Cysouw (2009) surveys a great variety of pronominal paradigms in many
typologically unrelated languages. His work on number and person marking in
pronominal paradigms makes three important contributions relevant for the analysis of
the Slavic pronominal dual and verbal agreement in this dissertation. First, Cysouw
redefines dual number in the pronominal domain and presents a novel typology of
number marking based on the concept of RESTRICTION. Second, he introduces the concept
of markedness reversals in number systems of various languages which challenges
traditional understanding of this concept in terms of Greenberg’s Universal 34: "No
language has a dual [number] unless it has a plural" (1963:94). Third, Slavic languages
with duals are shown to fit into a "dual-unified-we" typology (Cysouw 2009:206).
Cysouw (2009) proposes a novel typology of number marking in the pronominal
domain. His new typology of number consists of three categories: singular, group, and
restricted group. This typology is derived based on the idea of restriction. Restriction
applies to groups which can be restricted either to a minimum number of participants
(dual) or to a small number (paucal).
Cysouw’s typology of number marking is shown in (Figure 2.7). The category
group corresponds to the traditional plural. However, as opposed to the traditional
understanding of a plural defined by the number of participants involved, for Cysouw
(2009), group is defined qualitatively by the kind of participants not by their number.
Because a group is not restricted by the number of participants, it is an unrestricted group.
Therefore, group is an unmarked category in relation to the restricted group.
81
Figure 2.7 Typology of Pronominal Number Marking (Cysouw 2009:205)
Restricted
Group
Group
1+2
1+2+3
1+3
2+3
3+3
1
2
3
The category restricted group corresponds to the traditional dual. In this category,
the number of participants is restricted to what is minimally needed; namely, two
persons. Thus, the 1st and the 2nd persons are minimally needed to form an inclusive
(including the addressee) dual. The 1st and the 3rd persons are needed to from the an
exclusive (excluding the addressee) dual. The category restricted group is more marked
than group.
In Cysouw’s (2009) typology of number markedness is handled differently from
the traditional approach. For Cysouw, there are two unmarked categories – singular and
group since both are not restricted by number. Restricted group is a marked category
since it is restricted by number (Figure 2.8). Traditionally, singular is an unmarked
category in opposition to the plural and dual which are marked.
Figure 2.8 Cysouw (2009) vs. Traditional Number Markedness in the Pronominal
Traditional Division: Unmarked for
Number
SINGULAR
Marked for
Number
-
GROUP
(plural)
Cysouw's approach:
Unmarked for Number
-
RESTRICTED GROUP
(dual, trial, etc)
Marked for Number
82
Cysouw’s (2009) understanding of number markedness in the pronominal domain
is based on the concept of restriction in that a restricted group is more marked than simply
group. However, this interpretation of markedness of is not absolute. Cysouw points out
three types of markedness reversals: 1). referential; 2). morphological; 3). structural.
Cysouw argues that these three types of markedness reversals provide evidence against a
well-established marked status of the dual in relation to non-singular (plural) (Greenberg
1963, Jakobson 1984).
Referential markedness reversal occurs when an unmarked meaning of the
restricted GROUP is dual; that is, two persons are meant. A group (plural) referring to more
than two persons is marked by adding an extra morpheme. Cysouw cites Navajo (Na
Dene) as an example of referential markedness reversal. In Navajo, the form nxíh can be
used to mean either dual or plural. To express reference to more than two persons the
prefix da- is added to express plural 26).
(26)
Navajo
dual/plural
nxíh
‘we’
plural
da-nxíh
‘we, more than two’
(Cysouw 2009: 194)
Morphological markedness reversal occurs when the plural is realized by a more
morphologically complex form than the dual. Cysouw provides examples from Nganasan
(Uralic), Damana (Chibchan), and Kwamera (Austronesian) to illustrate his point.
Morphological markedness of phonological exponents of plural pronouns in relation to
dual is not unusual and received a principled explanation within a feature-theoretic
83
framework (Nevins 2006). Nevins (2006) argues that it is crucial to distinguish between
markedness of phonological exponents (morphophonological realization) and
markednedness of abstract features. According to Nevins's (2006) approach to
markedness, morphological complexity of the plural in contrast to less complex dual does
not mean that plural forms are more marked. To determine markedness, it is necessary to
analyze a morphosyntactic, featural representation of a dual or a plural form at the level
of Morphological Structure.
Structural markedness reversal is most puzzling. It occurs when more structural
distinctions are made within the dual than the plural. Cysouw cites pronominal paradigm
of Samo (Trans-Guinea) where more person distinctions are made in the dual than in the
plural. Cysouw notes that languages with structural markedness reversals are not
numerous. According to traditional observations (Greenberg 1966), I found that in Old
Church Slavic, Old Russian, Slovenian, Upper and Lower Sorbian, and Kashubian, the
dual shows fewer distinctions in gender and case.
Cysouw (2009) proposes a ‘dual-unified-we’ typology of pronominal systems
with the dual. The dual-unified-we typology is "the major paradigmatic structure with
dual marking but without an inclusive/exclusive opposition" (206). It is claimed to be one
of the four most frequent pronominal paradigms in the world’s languages (Ingram 1978).
Cysouw argues that Proto-Indo-European as well as modern Indo-European languages
including Lithuanian and Lower Sorbian are good examples of this typology. Lower
Sorbian personal pronouns are shown to have a complete ‘dual-unified-we’ paradigm
(Figure 2.9).
84
Figure 2.9 Lower Sorbian Dual-Unified-We Paradigm Cysouw 2009:206)
Group
1
ja
2
ty
3 wono/wona
Restricted
Group
my
mej
1+2+(3)
1+3
wy
woni
wej
wonej
2+3
3+3
Cysouw (2009) shows that ‘dual-unified-we’ pronominal paradigm is found not
only in Indo-European, but is attested in a wide variety of the world’s languages. For
example, it is found in Uralic languages Khanti and Mansi spoken in Siberia. In SouthEast Asia, Tibeto-Burman languages, Meithei, Kham, Mon-Kher, and Hmong Njua have
the dual of this type. Eskimo-Aleut, Trans New Guinea, and Australian languages display
the ‘dual-unified-we’ paradigm as well.
85
2.6.
The Typology of the Slavic Dual
A Serbian linguist, Alexander Belič, was the first among Slavic linguists to
address the typological, morphosyntactic, and diachronic complexities of the Slavic dual
(1899). Belič (1899) proposed the following typology of the Slavic dual (Table 2.18).
According to Belič, the Slavic dual was classified into 3 different types. The 1st type,
'free' dual, corresponds to natural pairs of objects, such as body parts (Table 2.18, (2)).
The 2nd type, bound dual, occurs with nouns modified by the numeral two or the
quantifier both (Table 2.18, (2)). The 3rd type, conjunctive dual, is found with two nouns
which are connected by the conjunction i ('and') and the verb is marked by a 3rd person
dual suffix -te (Table 2.18, (3)).
Table 2.18 Typology of the Slavic Dual (Belič 1899)
Types of the Slavic Dual
(1) Free Dual
a). distributive dual
(2) Bound Dual
a). anaphoric dual
Examples
uš-i (‘ears’), oč-i (‘eyes’), glaz-a ('eyes')
nodz-ĕ učenikomǔ ('legs of the disciples')
dǔv-a sǔn-a ('two sons')
ob-ĕ sestr-ĕ ('both sisters')
ta ... idos-te ('those two ... went')
(3) Conjunctive Dual
iĕkovǔ i ioanǔ rĕs-te ('Jakob and John said')
a). pronominal dual
radui-ta va sę (‘You two rejoice.’)
86
The 1st type of dual, the free dual is characterized by the following examples. The
Old Church Slavic nouns uš-i (‘ears’), oč-i (‘eyes’), bok-a ('sides'), glaz-a ('eyes'), ruts-ĕ
('hands'), nodz-ĕ ('legs') are all marked by the dual suffixes -i, -a, or -ĕ depending on the
noun declension type. Within the 1st type of the free dual, Belič also identifies a subtype,
a distributive dual, in which the paired noun marked by a dual suffix occurs with a plural
noun. The entire nominal phrase has a distributive meaning since the dual meaning of the
parired noun is distributed over a number of plural entities. For example, in the nominal
phrase, nodz-ĕ učeniko-mǔ ('pairs of legs belonging to each of a plural number of
disciples'), the paired noun legs is marked by a dual suffix -ĕ, and the meaning of
pairedness is distributed over the plurality of disciples (27).
(27)
Old Church Slavic
nodz-ĕ učenik-o-mǔ
leg-DU disciple-TH-PL.DAT
'pair of legs of each of the disciples'
(Žolobov 2001:17)
The bound dual type also has a subtype, identified by Belič as an anaphoric dual.
The anaphoric dual entails an anaphoric relation of the demonstrative pronoun marked by
a dual suffix to its antecedent, a syntactic dual subject used in the previous sentence. For
example, in the sentence ta ... idos-te, the demonstrative pronoun ta appears in the dual
number (28).
(28)
Old Church Slavic
ta
ido-s-te
this.DU.MASC go-PAST-3.DU
'Those two went.'
(Žolobov 2001:17)
87
The conjunctive dual, the 3rd type, is further classified by Belič as having a
pronominal dual subtype. The pronominal dual has the 1st and 2nd person pronouns
appearing as the syntactic subject and the verb is marked by a dual suffix via agreement
(29). The 3rd person dual is not part of the pronominal paradigm of Old Church Slavic.
(29)
Old Church Slavic
radui-ta
va
sę
rejoice-2.DU.IMP 2.DU.NOM REFL
‘You two rejoice.’
(Savva's Book 123:9, Math 28.9)
The typological classification of the Slavic dual developed by Belič (1899) has
important implications for the diachronic analysis of the Slavic dual presented in this
dissertation. Based on Old Church Slavic data, Belič was the first among Slavic linguists
to notice that the dual across all types identified in his typology (Table 2.18) was
undergoing gradual replacement by the plural. Belić (1899) came to the conclusion that
dual number, as a grammatical category, became unstable also in the pronominal
paradigm of personal pronouns. Belič (1899) suggested that the use of the plural
pronouns instead of the dual in Old Church Slavic could be attributed to the "identity of
some dual and plural forms" (1190). Despite extensive research of Old Church Slavic
manuscripts, Belić (1899) could not give a principled reason for the gradual replacement
of dual pronouns by their plural counterparts.
The typology of the Slavic dual proposed by Belič (1899) was further tested and
refined by a Russian Slavic linguist, A.M. Iordanskij (1960) in his book Istorija
Dvojstvennogo Čisla v Russkom Jazyke (The History of Dual Number in the Russian
language). Using Old Russian manuscripts as his sources of data, Iordanskij (1960)
88
analyzed the loss of dual number marked on Old Russian pronouns, nouns, numerals,
verbs, and adjectives. Iordanskij's contribution to the study of the historical development
of the Slavic dual was in his detailed analysis of the dual in Old Russian in contrast to
Belič's (1899) focus on the dual in Old Church Slavic. One of the major drawbacks of
Iordanskij's (1960) study of the Old Russian dual was his reliance on some secondary
sources of data, which lead him to make incorrect generalization about the diachronic
development and loss of dual number in Russian.
Regarding the diachronic development of the 1st and 2nd person pronouns,
Iordanskij argued that the loss of the dual began well before the 11th century, and
continued throughout the 12th century (1960:25). He claimed that the 2nd person dual
pronoun va was the first one to be replaced by the plural form vy (30). This claim is
incorrect for two reasons. First, it was not confirmed by my own corpora studies and
analysis of the Old Russian dual which I present in Chapter 4. In all of the Old Russian
manuscripts which I analyzed in this dissertation including the earliest, The Ostromir
Gospel (1056-1057), the only 2nd person pronoun which occurred was vy, not va.
Second, the dual verbal agreement suffix -ta in (30) shows that the subject still denotes 2
entities, not many. What we observe in (30) is not a total replacement of the dual by the
plural as claimed by Iordanskij (1960), but simply dual/plural syncretism of the form vy.
(30)
Old Russian
vy ubo nbsĭnaja človĕk-a es-ta
2.PL EMPH heavenly man-2.DU be-2.DU.PRES
'You two are heavenly men.'
(Skazanie o Borise i Glebe,
Iordanskij 1960:17)
89
Another Russian linguist, O.F. Žolobov (1998, 2001), conducted a very extensive
study of diachronic changes in the Old Russian dual. The main goals of this new study
were to correct the shortcoming of Iordanskij's (1960) initial study and to provide a
theoretical explanation for the loss of the dual in Old Russian. In his studies, Žolobov
(1998, 2001) relied only on primary data from a variety of Old Russian manuscripts and
birch barks dating from the 11th up to the 15th centuries thus eliminating Iordanskij's
(1960) data errors.
Žolobov (2001) argued that the reason for the loss of the dual could be accounted
for in terms of grammatical markedness.20 He claimed that the Old Russian dual was a
marked member of the dual/plural grammatical opposition (Figure 2.10). The dual, as a
marked member of the morphological opposition, underwent the process of neutralization
and was subsumed by the plural. When the dual was no longer a marked member of the
opposition, it became an unmarked type of number within the plural.
Figure 2.10 Markedness of the Dual (Žolobov 2001)
'one'
'two and more'
singular
'two'
dual
marked member
20
'two and more'
plural
unmarked member
Žolobov (2001) did not assume any specific notion of markedness. Perhaps, he followed Jakobsonian
tradition regarding morphological markedness.
90
Žolobov (2001) accounted for the loss of the dual in Old Russian in terms of its
grammatical markedness. However, Žolobov (2001) did not explain why the dual was a
marked category in a dual/plural grammatical opposition. It remains unclear in his
analysis what semantic or morphological factors contributed to the grammatical
markedness of the dual and its eventual disappearance in Old Russian.
91
2.7
Summary
In this chapter, I presented typological approaches to the study of dual number
cross-linguistically. The first typological classification of languages with dual number
began with the pioneering typological work of Wilhelm von Humboldt (1827). In his
conceptual typology, Humboldt (1827) proposed a crucial semantic relation between the
dual and the singular as well as the dual and the plural. Humboldt’s original conceptual
representation of the dual as both singular and plural was essential in my proposal of the
principle of Morphosyntactic Feature Economy presented in Chapter 5.
Humboldt's original typology of the dual was further developed by Jespersen
(1965), Plank (1989), Corbett (2000), and Cysouw (2009). Each of these linguists made
significant contributions to the typological study of the dual. I explored implications of
the proposed typologies in relation to the Slavic dual throughout this chapter.
Finally, I discussed typological approaches to the Slavic dual suggested by Belič
(1899), Iordanskij (1960), and Žolobov (1998, 2001). Belič's contribution to the study of
the Slavic dual was very significant. Working with the Old Church Slavic data, he was
one of the first Slavic linguists to recognize that the dual had already been undergoing
diachronic change in the 11th century. As a result of this diachronic change, the dual was
replaced by the plural in some instances in the Old Church Slavic manuscripts. The
dual/plural diachronic change in Old Church Slavic, a member of the South Slavic
branch, was also found in Old Russian, a representative of the East Slavic branch. In
Chapters 4 and 5 of this dissertation, I investigate in detail why the dual survived in
92
Slovenian, Upper, and Lower Sorbian (South Slavic) whereas it completely disappeared
in Russian (East Slavic) and Kashubian (West Slavic).
93
CHAPTER 3 DERIVATION OF THE SLAVIC DUAL IN DISTRIBUTED
MORPHOLOGY
3.0.
Introduction
In this chapter, I present the framework of Distributed Morphology (DM) which I
adopt in this dissertation for the formal analysis of dual number in the Slavic languages. I
show a step-by-step DM derivation of dual number marked on Slavic nouns, pronouns,
and verbs (by agreement), and explain why morphological operations of Impoverishment
and Fission, which operate on morphosyntactic representations of the dual at
Morphosyntactic Structure, are key for my analysis of diachronic changes in the Slavic
dual.
In Section 3.1., I lay out the architecture of grammar which I assume within the
theoretical framework of DM, originally proposed by Halle and Marantz (1993), and
further refined by Noyer (1997), Harley and Noyer (1999), and Embick and Noyer
(2007). I adopt the architecture of grammar proposed by Harley and Noyer (1999) since it
differentiates a separate level of Morphosyntactic Structure (MS) crucial for my analysis
of diachronic changes in dual number in Slavic.
In Section 3.2., I show how dual nouns, pronouns, and verbs bearing dual
agreement are derived in DM. I make two important distinctions regarding agreement and
case. The first distinction is drawn between syntactic and morphological agreement, and
the second one is between syntactic Case and morphological case.
94
In the framework of Minimalist syntax (Chomsky 2001, 2008 a, b), I assume that
syntactic agreement occurs between uninterpretable and interpretable features in a
syntactic derivation, and is carried out via an operation of feature valuation called Agree.
In contrast, morphological agreement is a purely post-syntactic operation implemented
via addition of Agreement nodes (AGR) at the level of Morphological Structure (MS)
(McFadden 2004, Bobaljik 2008, Embick and Noyer 2008).
Similar to syntactic vs. morphological agreement, I assume that the unvalued
structural Case features of a DP are valued by the functional heads T and v in a syntactic
derivation. In contrast to syntactic structural Case valuation which occurs strictly in the
narrow syntax, I assume that Kase nodes and morphological case features are absent in
the syntactic structure of a ‘word,’ and are added after syntax at MS prior to Vocabulary
Insertion (Bobaljik 2008, Embick and Noyer 2008).
In Section 3.3., I focus on the morphological operations of Impoverishment and
Fission which apply at MS. I assume that Impoverishment and Fission are motivated by
markedness of the [-singular -augmented] feature bundle of the dual (Calabrese 2011,
Harbour 2011). I further claim that application of both Impoverishment and Fission
results in simplification of the morphosyntactic representation of the Slavic dual, and
hence featural economy. In Section 3.4., I summarize the key theoretical assumptions and
concepts presented in this chapter.
95
3.1.
The Architecture of Grammar in Distributed Morphology
The theory of Distributed Morphology (DM) is a syntactic approach to
morphology originally proposed by Halle and Marantz (1993), and further refined by
Noyer (1997), Harley and Noyer (1999), and Embick and Noyer (2008). In this
dissertation, I adopt the version of DM developed by Harley and Noyer (1999). The
central architectural premise of DM is that words, like phrases and sentences, have an
internal structure whose derivation is syntactic. Syntax is the only component of grammar
which generates a syntactic hierarchical structure containing morphosyntactic features of
roots and affixes.
DM as an approach to morphology is called ‘distributed’ because the lexical
properties of a morpheme – its phonological exponence, morphosyntactic features, and its
semantic meaning are distributed among different components of the grammar, and are
not collected together in a single lexical item. In this respect, DM sharply contrasts with a
Lexicalist approach to morphology (Lieber 1992, DiSciullo and Williams 1987) which
assumes that all of the lexical contents of a morpheme are bound together in one lexical
item represented in the lexicon. DM rejects the idea of a lexical item derived in the
lexicon and belonging exclusively to morphology.
In DM, the notion of a morpheme is abstract. According to the model of grammar
in Figure 1, a morpheme is derived at the three levels of representation: syntactic,
morphological, and phonological. In the syntax, a syntactic numeration is composed by
selecting morphosyntactic features from a universal set of features (e.g.[+past], [Det],
[+sg], [Root], etc) made available by Universal Grammar (UG). Morphosyntactic
96
features combine to form binary syntactic structures via the syntactic operations of
Merge, Move, and Copy. By the end of a syntactic derivation, morphosyntactic features
are positioned at syntactic terminal nodes in a generated syntactic structure (Figure 3.1).
Figure 3.1 The Architecture of Grammar in DM (Harley and Noyer 1999)
Morphosyntactic Features
Syntactic Operations
(Merge, Move, Copy)
Morphological Structure (MS)
(Merger, Fusion, Fission,
Impoverishment)
Phonological Form (PF)
(Vocabulary Items, Readjustment)
Logical Form (LF)
Conceptual Interface (CI)
(“Meaning”)
Encyclopedia
(non-linguistic knowledge)
As shown in Figure 3.1, at the level of Morphological Structure (MS),
morphosyntactic features can be subject to morphological operations of Merger, Fusion,
97
Fission, and Impoverishment which can alter their morphosyntactic structure. After a
morphosyntactic structure undergoes any change at MS, it is sent off to the level of
Phonological Form (PF) where a phonological exponent supplies its phonological
features through insertion of a Vocabulary Item (VI).
Simultaneously, a morphosyntactic structure is also sent off to Logical Form (LF)
where its semantic meaning is computed by interpreting a syntactic derivation into which
morphosyntactic features enter. A mental Encyclopedia which contains non-linguistic
knowledge about objects and entities in the real world is accessed to obtain non-linguistic
information. Finally, the full semantic meaning of a morphosyntactic structure is
established at the Conceptual Interface (CI).
I adopt the three major tenets of DM which set it apart from other theories of
morphology: 1). Late Vocabulary Insertion, 2). Underspecification of Vocabulary Items,
3). Syntactic Hierarchical Structure All the Way Down. The first tenet of DM, Late
Vocabulary Insertion, means that morphosyntactic features at syntactic terminal nodes in
a syntactic derivation do not have any phonological content, and are ‘paired’ with their
Vocabulary Items (VIs) ‘late’ after a syntactic derivation has been composed.
Vocabulary Insertion takes place after syntax at Phonological Form (PF), when
morphosyntactic features at syntactic terminal nodes are discharged by VIs which fill
them with phonological content. A Vocabulary Item (VI) is a relation between a
phonological exponent (phonological feature matrix) and its context for insertion, which
specifies a certain morphosyntactic environment (31). Some examples of VIs are given in
(32).
98
(31)
Vocabulary Item
/phonological exponent/ ↔ context for insertion
(32)
Examples of Vocabulary Items
a). /dva/ ↔ Num [-sg -aug]
numeral two in Russian
b). /vy/ → elsewhere
2nd person plural pronoun in Russian
c). /brat/ ↔ [DP [D___]]
root of brat (‘brother’) in Russian
The second tenet of DM, Underspecification of Vocabulary Items, refers to the
idea that a phonological exponent of a Vocabulary Item can be underspecified (i.e. does
not contain all of the morphosyntactic features that a terminal syntactic node does)
relative to a syntactic terminal node into which it is inserted. In contrast to underspecified
VIs, morphosyntactic feature bundles (sets) at syntactic terminal nodes are fully
specified.
In a realizational model of grammar such as DM morphosyntactic features and
phonological exponents are independent of each other. Therefore, there is no need for
phonological exponents to carry a full set of morphosyntactic features. Thus,
Underspecification of Vocabulary Items follows naturally from the DM architecture of
grammar where syntactic terminal nodes, which carry fully specified morphosyntactic
features, and Vocabulary Items, which realize these features, reside at different
grammatical levels – syntax and PF.
A paradigm of the numeral dva (‘two’) in Contemporary Standard Russian is an
example of application of the principle of Underspecification of VIs (33). The Russian
numeral dv-a (‘two’) in the nominative case exhibits masculine and neuter gender marked
99
by the same suffix –a whereas feminine gender is marked by the suffix –e. Since the
same VI /a/ realizes both masculine and neuter gender of the numeral ‘two,’ we conclude
that the suffix –a is underspecified for gender.
(33)
The Numeral dva (‘two’) in the Nominative Case in Contemporary Standard
Russian
Gender
Suffixes
Masculine
-a
Neuter
-a
Feminine
-e
The phenomenon where one underspecified VI can be inserted into two or more
distinct syntactic nodes represented by different sets of morphosyntactic feature bundles
is called syncretism. Underspecification is one way in which syncretism can be derived in
DM. As we will see in Section 3.3. Impoverishment is another source of syncretism.
According to this definition of syncretism given above, the VI /a/ is an instance of
masculine/neuter gender syncretism in the numeral inflection of Contemporary Standard
Russian.
Underspecification of the VI /a/ leads to a syncretism when more than one VI can
be inserted into the same syntactic terminal node. In the case of the Contemporary
Standard Russian numeral dva, two VIs are eligible for insertion at the syntactic terminal
node fully specified for the feminine gender (34). When two or more VIs are competing
100
for Vocabulary Insertion into the same syntactic terminal node, their competition is
resolved via the Subset Principle (35).
(34)
Vocabulary Items
/e/ ↔ [-sg, -aug, +fem]
/a/ ↔ elsewhere
(35)
The Subset Principle (Halle & Marantz 1997:428)
Where several Vocabulary items meet the conditions for insertion, the item
matching the greatest number of features specified in the terminal morpheme
must be chosen.
The VI /e/ has three features in its context for insertion whereas the VI /a/ is
completely underspecified; it is a default, or an elsewhere item whose context for
insertion is devoid of any features. According to the Subset Principle, the VI /e/ wins the
competition, since its subset has more features fitting the set of features at a syntactic
terminal node specified for the feminine gender than that of its counterpart /a/. Thus, the
VI /e/ is inserted at a syntactic terminal node which is specified for the feminine gender
of the numeral dve (‘two’). The elsewhere VI /a/ is inserted in the masculine and neuter
contexts.
The third tenet of DM, Syntactic Hierarchical Structure All the Way Down,
highlights its major premise as a syntactic theory of morphological structure. This
hypothesis suggests that hierarchical structure into which root and affixes enter is
determined by the syntactic hierarchical structure generated by Merge and Move
operating on bundles of morphosyntactic features. It follows that in the default case,
syntactic and morphological structures are isomorphic; that is, there is one-to-one
101
mapping relationship between a syntactic terminal node and an affix corresponding to
this node.21
21
In many cases, no such isomorphy obtains due to operations of Impoverishment and Fission which apply
at the level of Morphological Structure. These operations are discussed in Section 3.3. of this chapter.
102
3.2.
Derivations of the Slavic Dual in DM
In DM, the derivation of a complex morphological object (‘word’), which consists
of a root and affixes occurs strictly in the syntax. Since the syntax is the only module of
grammar which generates hierarchical structure, the morphological structure of a ‘word’
is simply its syntactic structure.22 Following Embick and Noyer (2007), I assume that the
internal structure of a ‘word’ is represented as a complex syntactic head derived via head
movement by adjoining one syntactic head to another one in a successive manner (36).
(36)
Complex Head (Embick & Noyer 2007:303)
Z
Y
Z
[F]
X
Y
[F]
√Root
X
[F]
A complex head in (6) has a root (√Root) and several affixes (X, Y, and Z)
represented as syntactic heads which are positioned at syntactic terminal nodes in a
binary syntactic structure. Each syntactic head has a bundle of fully-specified
morphosyntactic features, labeled as [F].
The derivation of a complex morphological head proceeds as follows. A
syntactically complex head is derived via syntactic head movement (Travis 1984) when
one syntactic head adjoins another one through successive adjunction. The syntactic
2222
I use the term word descriptively, and not a as unit of lexicon as in Lexicalist theories espoused by
Lieber (1992) and DiSciullo and Williams (1987).
103
structure in (37) illustrates a complex morphological object (‘word’) before head
adjunction, and the structure in (38) shows this ‘word’ after head adjunction.
(37)
Syntactic Structure Before Head Adjunction (Embick & Noyer 2007:303)
ZP
Z
YP
Y
XP
√P
X
√Root
(38)
Complex Head After Head Adjunction (Embick and Noyer 2007:303)
ZP
Z
YP
Y
Z t
XP
X
Y
√Root
X [F]
[F]
t
√P
t
It is crucial for my analysis of dual number features in Slavic, which enter into
agreement relations with verbs and concord relations with nouns, to make explicit
assumptions about agreement and case features. Regarding agreement, I draw a clear
distinction between syntactic agreement and morphological agreement. Assuming
104
Minimalist syntax (Chomsky 2001, 2008 a, b), syntactic agreement occurs between
uninterpretable and interpretable features in a syntactic derivation, and is carried out via
an operation of feature valuation called Agree.
In the framework of DM assumed in this dissertation, morphological agreement is
a post-syntactic operation implemented via addition of Agreement nodes (AGR) at the
level of Morphological Structure (MS) (McFadden 2004, Bobaljik 2008, Embick and
Noyer 2007).
Similar to agreement features, I distinguish between abstract syntactic Case
features and morphological case features.23 In the spirit of Chomsky’s (2001, 2008 a, b)
Minimalist syntax, I assume that unvalued structural Case features present on a DP in a
syntactic derivation are valued by the functional heads T and v. In contrast to abstract
syntactic Case features, morphological case features are added to syntactic terminal nodes
at MS, and are consequently spelled-out by VIs at PF (Bobaljik 2008, Embick and Noyer
2007).
Let me first address the distinction between abstract syntactic agreement and
morphological agreement. In Minimalism (Chomsky 2001, 2008 a, b), syntactic
agreement is carried out via an operation Agree which assigns values to uninterpretable
features of the functional heads T and v. Uninterpretable phi-features (person and
number) on T and v enter a syntactic derivation unvalued. It means that these features
23
Inherent and lexical are non-structural cases which play a role in case marking of Slavic DPs. In Slavic
and in other languages, lexical case is unpredictable, and is valued by a lexical head, such a verb, or a
preposition (Woolford 2006, Richards 2007). Inherent case (such as dative in Slavic) is valued by a lexical
head which also assigns it a theta role. Therefore, inherent case is predictable because it is linked to
assigned theta-roles.
105
have no semantic value and need to be valued (‘agree’) by interpretable features phifeatures (person, number, and gender) of a DP.
An operation Agree assigns values to the unvalued phi-features on the heads T
and v. Once valued, uninterpretable features still have no semantic interpretation, and can
cause a syntactic derivation to “crash” at both LF and PF levels. To avoid a potential nonconvergence of a syntactic derivation, valued uninterpretable features are deleted at LF
before they reach the level of Vocabulary Insertion at PF.
In contrast to abstract syntactic agreement which occurs in narrow syntax,
morphological agreement is an operation which applies after the syntax at the level of MS
before Vocabulary Insertion at PF. Morphological agreement applies when AGR nodes
are added to a complex syntactic head by language-specific rules in accordance with
morphological requirements of a particular language.
First, I show how abstract syntactic agreement occurs in a clause according to the
Minimalist approach (Chomsky 2001, 2008 a, b). Second, I illustrate how morphological
agreement applies after syntax at the level of MS. Finally, I focus on the morphological
structure of an Old Church Slavic verb bearing dual agreement and show how an AGR
node is added to a complex verbal head.
To see how abstract syntactic agreement works, let us take an Old Church Slavic
sentence from the Codex Marianus (39). Its syntactic structure is shown in (40).
(39)
Old Church Slavic
чашѫ
моѭ испиета
čaš-ę
mojǫ ispi-je-ta
cup-3.SG.FEM.ACC my drink-TH-2.DU.PRES
106
‘You two will drink all of my cup.’24
(Codex Marianus, Math. 20.23)
(40)
TP
DP
proi
T
T
[+pres]
[u phi]
Agree
vP
DP
proi
[2, du]
v
[u Case] [i Asp]
[u phi]
v'
√P
√ISPI
Agree
DP
[3,sg]
[u Case]
Abstract syntactic agreement, or valuation of uninterpretable features on T and v
in the syntactic structure (40) proceeds in the following way. The functional head T has
uninterpretable phi-features - person and number, and an interpretable present tense
feature.25 The uninterpretable person and number features on T have no values, and as
such cannot be interpreted at LF or realized at PF.
In order to get valued, the head T with the uninterpretable person and number
features probes for the closest c-commanded DP which has interpretable person and
number features. The probe T finds its goal, a pro subject DP, which has interpretable 2nd
24
Old Church Slavic verbal stems of perfective aspect (which is lexical, i.e. Aktionsart ) in the present
tense can also express a future event (Xaburgaev 1986:190, Lunt 2001:81).
25
I assume (Chomsky 2008 a, b) that CP is a phase whereas TP is not. T can bear both uninterpretable phifeatures (person and number) and interpretable Tense feature only if it is selected by C. Since T is selected
by C, T “inherits” its uniterpretable phi-features and interpretable tense features from C. For more details
on phasehood of CP and not TP, see Chomsky (2008 a, b).
107
person and dual number features, and an unvalued structural Case feature. The
uninterpretable person and number features of T are matched with the interpretable 2nd
person and dual number features of a pro DP through an operation of feature valuation
called Agree. Once valued, the uninterpretable features on T are deleted before they reach
LF since they cannot be interpreted according to the Principle of Full Interpretation
(Chomsky 2001).
The functional head v bears uninterpretable person and number features and an
interpretable Aspect feature (Asp), which I adopt from Pesetsky and Torrego (2001). Like
T, the v head probes for the closest c-commanded DP with interpretable person and
number features. The probe v finds its goal, the object DP with the interpretable 3rd
person and singular number features, and an unvalued structural Case feature.26 The
operation Agree applies when the uninterpretable phi-features of v are matched with the
interpretable phi-features of the object DP.27 For the vP phase to be fully interpreted at
LF, the valued uninterpretable phi-feature features on v are deleted before LF.
In what follows, I show how morphological agreement applies strictly after syntax
at the level of MS. Let us focus on morphological agreement of a ‘verb.’ In the syntax,
the structure of a ‘verb,’ i.e. a complex verbal head contains a √ROOT, a verbalizing v
node, a Tense (T) node, but crucially no AGR node (41).
26
For the purpose of clarity of exposition, I do not provide a full structure of the DP in (40). I return to the
full structure of the DP later in this section.
27
According to the Minimalist Approach, the head v is “the functional category which heads verb phrases
with full argument structure unlike unaccusatives and passives” (Chomsky 2008 a:17).
108
(41)
Syntactic Structure of a Complex Verbal Head in NS (Embick and Noyer
2007:306)
T
v
√ROOT
T
[+tense]
v
I assume that an AGR node is added after syntax at the level of MS prior to
Vocabulary Insertion by a language-specific rule in accordance with morphological
requirements of a particular language (42).28 My assumption is generally in line with
Embick and Noyer’s (2007) views on AGR node adjunction with one crucial difference.
For Embick and Noyer (2007), AGR nodes are added at PF before Vocabulary Insertion,
whereas I assume that this process occurs earlier at the level of MS before PF in
accordance with the model of DM adopted from Harley and Noyer (1999).
(42)
Structure of a Complex Verbal Head with an AGR node at MS
T
v
√ROOT
28
T
v T
AGR
[+tense] [P, #]
I assume that structurally an AGR node is added by the process of adjunction (Embick and Noyer
2007:306).
109
To illustrate how an AGR node is adjoined at MS, let us take an Old Church
Slavic verb ispi-je-ta with a dual agreement suffix -ta from sentence (39). The root of this
verb ispi- is followed by a thematic vowel –je.29 The suffix –ta bears the features of
present tense, the 2nd person, and dual number of a pro-drop subject. This suffix is an
example of a portmanteau morpheme (very common in Slavic) in which tense, person,
and number features are fused together in one VI. In the syntax, the Old Church Slavic
verb isp-je-ta has the structure shown in (43). Crucially, the syntactic structure in (43)
does not have an AGR node.
(43)
Syntactic Structure of the Old Church Slavic verb ispi-je-ta in NS
T
v
√ISPI
T
[+pres]
v
An AGR node is added to the Tense node at MS by a language-specific rule,
which requires that in Old Church Slavic an AGR node must appear on a finite Tense
node (44). The rule in (44) introduces an AGR node which at this point in the derivation
is featureless as shown in (45). I assume that the features of a pro-drop subject are coped
onto this AGR node by a feature-copying process at MS (46). In the structure (47), an
added AGR node is no longer featureless, but already ‘acquired’ its phi-features, i.e. the
2nd person and dual number through feature-copying.
The verb ispiti belongs to the 3rd conjugation class according to its stem in the 2nd person singular form
ispi-je-ši (Krivčik & Možejko 1985:133).
29
110
(44)
OLD CHURCH SLAVIC Finite T Rule at MS
T finite → [T AGR]
(45)
The Structure of the OLD CHURCH SLAVIC verb ispi-je-ta with a
Featureless AGR at MS
T
v
√ISPI
(46)
T
vT
[+pres]
AGR
[…]
Feature Copying at MS
A phi-feature which is present on a syntactic terminal node in the narrow syntax is
copied onto an AGR node at MS.
(47)
The Structure of ispi-je-ta with AGR Features
T
v
√ISPI
T
v T
[+pres]
AGR
[2, du]
In (47), an AGR node adjoins to the T head, and makes it possible for an
operation of Fusion to apply at the level of MS and fuse two nodes together into one
before Vocabulary Insertion at PF (Halle and Marantz 1993, Noyer 1997, Embick 2010).
The T and AGR nodes are fused into one morphosyntactic object under the sisterhood
relation in the syntactic configuration (47) by a Fusion rule (48). A Fusion rule (48)
111
applies at MS when the T node with [+pres] tense feature and the AGR node with the
phi-features [2, du] are fused into one node.
(48)
Fusion Rule for T and AGR Nodes in Old Church Slavic
[T +pres]
[AGR 2 du] → [T/AGR +pres, 2, du]
At the level of MS, the final structure of the verb ispi-je-ta looks as shown in (49).
A theme node (TH) is added to the syntactic structure to host a theme vowel –je. At the
level of PF, Vocabulary Insertion applies and inserts the following VIs: the root /ispi/, a
phonologically null VI /∅/ for the verbalizing v node, a theme vowel /je/, and a single
exponent /ta/ for a fused T/AGR node (50).
(49)
The Final Structure of the Old Church Slavic verb ispi-je-ta at MS
T
v
T
√ISPI
v
v
(50)
T /AGR
[+pres, 2, du]
TH
VIs for the Old Church Slavic verb ispi-je-ta
/ispi/ ↔ /√ ISPI
/∅/ ↔ /v
/ta/ ↔ /[+pres, 2, du]
Now, I turn to the distinction between abstract syntactic Case and morphological
case. I will briefly explain the notion of syntactic case features. The subject and object
112
DPs enter a syntactic derivation with unvalued structural (nominative and accusative)
Case features as shown in the syntactic structure (40). Valuation of unvalued Case
features proceeds in the following way. When a subject or an object DP with an unvalued
structural Case feature is merged in a syntactic derivation, it makes T or v active for an
operation Agree to apply and match their uninterpretable phi features with interpretable
phi-features of a DP. Unvalued structural Case features are not features of the syntactic
heads T and v, but are assigned values by T and v under operation Agree. In the syntactic
structure (40), the functional head T values abstract nominative case of its pro-subject,
and the functional head v values abstract accusative case of its DP object.
In contrast to syntactic structural Case valuation which occurs strictly in the
narrow syntax, I assume that Kase nodes and morphological case features are absent in
the syntactic structure of a ‘word,’ and are added after syntax at MS prior to Vocabulary
Insertion. Following (Marantz 1991), I assume that Kase nodes are added at the level of
MS. As for addition of case features, my assumption generally follows Embick and
Noyer’s (2008) views on post-syntactic addition of morphological case features with one
crucial difference. According to Embick and Noyer (2007), morphological case features
are added to case/number nodes at PF whereas I assume that addition of morphological
case features occurs at a separate level of MS prior to PF.
In what follows, I show how morphological case nodes (K) and case features are
added to the syntactic structure of Slavic dual nouns and pronouns. Let us examine the
case paradigm of an Old Russian ‘noun’ žen-a (‘woman’). The Old Russian noun žen-a is
feminine in gender and appears in three numbers: singular, dual, and plural. It exhibits six
113
distinct case suffixes in the singular, three case suffixes in the dual, and five case suffixes
in the plural (51).
(51)
Case Suffixes of an *a-stem Old Russian Noun žen-a (‘woman’)
Number/Case
Nom
Acc
Gen
SG
-a
-u
-y
DU
PL
Dat
-ĕ
-ĕ
-y
Loc
-u
-ǐ
Instr
Voc
-oju
-o
-ma
-xǔ
-mǔ
-mi
For clarity of exposition, let us focus on the noun žen-a-ma in the dual dative
form (52).30 This noun is composed of the root žen-, a thematic vowel –a, and the suffix
–ma which marks dual number and dative case. Old Church Slavic is a language without
determiners. Following Perelsvaig’s (2007) argument on the universality of the DP
projection in determinerless languages, I assume that Old Church Slavic has a DP
projection in its noun phrase. In the model of DM (Harley and Noyer 1999, Embick and
Noyer 2007) adopted in this dissertation, I assume that syntactic structure of any
morphological form of a noun is composed of a root, nominalizing head n, a number head
(Num) – all dominated by the Determiner (D) head as shown in (53).
(52)
Old Russian
reč-e
žen-a-ma
say-3.AOR woman-TH-3.DU.DAT
‘He told the two women.’
(Ostromir Gospel, Math. 28.5, 203:5)
30
The thematic vowel -a shows up in the dative case of *a-stem nouns in the dual.
114
(53)
Narrow Syntactic Structure of an Old Russian Dual Noun žen-a-ma
D
D
Num
n
√
Num
[-sg -aug]
n
As shown in (53), in the syntax, the Num head contains the number features
[-sg -aug] which specify a dual number in a singular/dual/plural number system of Old
Russian.31 Crucially, the syntactic structure in (53) does not contain any Kase nodes or
morphological case features which indicate the dative case of the noun. This assertion
follows from the assumption established earlier that morphological case is a postsyntactic phenomenon; that is, no Kase nodes or case features are present in the narrow
syntactic computation of a morphological object (Marantz 1991, McFadden 2004,
Bobaljik 2008, Embick and Noyer 2007).
After the syntactic derivation in (53) is complete, it is subject to operations that
add morphological Kase nodes and case features at the level of MS. Following Marantz
(1991), I assume that a Kase node headed by K is added to host morphological case
features at the level of MS. Since Old Russian is language where number and case are
fused together in one portmanteau suffix, it is possible for a K head to adjoin to the Num
31
In Section 4.1. of Chapter 4, I provide a detailed account of dual number features.
115
head as shown in (54). A terminal node labeled TH is added to host a thematic vowel –a,
which is inserted at PF.
(54)
Syntactic Structure of an Old Russian Noun žen-a-ma with an Adjoined K
node at MS
D
D
Num
n
√
Num
n
n
Num
K
[-sg -aug] […]
TH
After a Kase node has been adjoined to the syntactic structure at MS, a level,
which mediates narrow syntax and PF, morphological case features can be added to the K
node. In order to add case features to the K node, Old Russian morphological cases, such
nominative, accusative, genitive, dative, locative, instrumental, and vocative need to be
decomposed into case features.
At this point in the derivation of the dual dative form žen-a-ma, three questions
arise: 1). which case features should be added to the K node at MS? 2). are case features
syntactically, morphologically, or semantically motivated? and 3). are case features
language-specific or universal?
The first question raised deals with theoretical approaches to case feature
decomposition. Case feature decomposition is not a trivial matter, and has been subject to
116
debate among linguists. There are at least three reasons for this debate. First, motivation
of case features is not always clear in the literature and is questionable (Halle 1997,
Müller 2004, Franks 2002, Baylin 2004). Second, there is a question of whether case
features are language-specific, or universal (McFadden 2004). Third, case categories
might not be represented as flat bundles of case features, but might have hierarchical
structure (Caha 2009).32
I argue that case features should not be assigned arbitrary labels simply to capture
generalizations about syntactic distribution of case-marked DPs and to describe attested
patterns of syncretism in inflectional categories, but rather case features should constrain
the grammar of a language based on syntactic and semantic grounds. I further claim that
decomposition of case features of nominal arguments (pronouns and nouns) is
constrained by two factors: i) by syntactic configurations in which DPs appear and ii) by
their argument structure; that is, by the semantic roles that DPs bear with respect to their
predicates. I sum up my claim about how case features are determined in (55).
(55)
Determination of Case Features Criterion
Case Features are determined by two factors:
1) The syntactic configuration in which a case-marked DP appears.
2) The argument structure of a DP within a verbal predicate.
32
It is not crucial for my proposal of the Principle of Morphosyntactic Feature Economy and for my
analysis of the Slavic dual what particular case decomposition theory is assumed. The proposed principle of
Morphosyntactic Feature Economy applies regardless of the how case features are decomposed.
117
I follow Jakobson’s (1984, 1985) fundamental insight that morphological cases
are not atomistic, i.e. that cases are not defined in terms of indivisible features, such
[+dative], [+instrumental], [+genitive], etc., but are decomposed into features which form
natural classes. Abundant instances of case syncretism, especially in Slavic languages
support the idea of natural classes of case features.
Determining case features in Slavic is a complex issue, which has been pursued
by Franks (2002) and Müller (2004) in Minimalism and DM.33 I claim that case features
should be syntactically and semantically motivated according to the case determination
criterion established in (55). Thus, I assume four binary case features - [±structural],
[±internal], [±inherent], and [±quantificational]. The feature [+structural] refers to a
syntactic structural Case valued by any of the following functional heads: C, T, v, Neg,
Pred, and Q. According to a well-established Minimalist assumption (Chomsky 2001,
2008a), T values structural nominative case whereas v values structural accusative case.
Following Bailyn (2004), I assume that besides the nominative and accusative, Slavic has
other structural cases which are valued by the following functional heads: C values
dative, Q values genitive, and Pred values instrumental.
To distinguish an internal argument from its external counterpart (Kratzer 1996), I
assume that the feature [+internal] specifies a case on a complement DP valued by the v
head in the vP projection. The feature [+inherent] is valued by a lexical head which also
33
Franks (2002) proposes four binary case features [±oblique], [±marginal], [±indefinite], and
[±quantified]. Although Franks’s proposal is a viable option to capture Russian case syncretisms and as
well as genitive and dative of quantification, it does not allow for the instrumental to be structural. I suggest
that instrumental case can be represented by the [±quant] feature; that is it can be both structural assigned
by the Pred head and lexical assigned by the lexical V or P head. I do not adopt Müller’s (2004) [±subject],
[±governed], and [±oblique] case features since they do not follow from the Minimalist view of case and
argument structure assumed in this dissertation.
118
assigns it a theta role. Therefore, cases which bear [+inherent] feature are predictable due
theta-roles of their DPs. The feature [+quantificational] is assigned to a structural case
which refers to a quantificational or scope relation.
Based on the five binary case features which I assume, the six primary cases in
Old Russian are decomposed as shown in (56).34
(56)
Decomposition of Case Features in OR
Nom Acc
Gen
Dat
Loc
Instr
Structural
+
+
±
±
-
±
Internal
-
+
-
-
-
-
Inherent
-
-
-
±
-
-
Quantificational
-
-
±
±
-
-
In the discussion above, I presented and motivated case features which can be
chosen for addition at a Kase node at the level of MS. The case features of a dative DP
žen-a-ma are determined by the argument structure in which this DP appears. In the
syntax, the dative DP žen-a-ma does not have structural case, rather it has inherent case
valued by the lexical verbal head reče (‘said’), which also assigns this DP the theta-role
of a goal.
Based on the syntactic configuration and argument structure of the verb reče
(‘said’), the case features [-struc, -int, +inh, -quant] are added at the Kase node of the
dual dative DP žen-a-ma. At MS, the structure of the DP žen-a-ma with an added [-struc,
34
I exclude the vocative case in the singular and focus on the six primary cases.
119
-int, +inh, -quant] case feature bundle at the K head looks as shown in (57). In the
structure in (34), the K node and the Num node are in a sisterhood relationship, and a
Fusion rule can apply to fuse the two nodes together (58).
(57)
Addition of case Features at MS
D
D
Num
n
√
n
TH
(58)
Num
Num
K
[-sg -aug] [-struc, -int, +inh, -quant]
n
Fusion Rule for OR Nominal Num/K Node
[Num -sg -aug]
[K -struc, -int, +inh, -quant] → [Num/K -sg, -aug, -struc, -int,
+inh, -quant]
The final structure of the dual noun žen-a-ma before Vocabulary Insertion is shown in
(59).
120
(59)
The Structure of the Dual Noun žen-a-ma at MS
D
D
Num
n
√
Num
n
n
Num/K
[-sg, -aug] [-struc, -int, +inh, -quant]
TH
The final step in the derivation of the dual dative form žen-a-ma is Vocabulary
Insertion at PF. The VIs for the inflectional suffixes of an Old Russian *a-stem noun žena-ma are shown in (60). The syncretic nominative and accusative cases are realized by
the VI /ĕ/ which is completely underspecified. The genitive and locative cases are
realized by the VI /u/ with the feature [-quant]. The dative and instrumental cases are
realized by the most specific VI /ma/ due to the number of features listed in the context
for its insertion.
(60)
VIs for Old Russian Dual Nominal Suffixes
Nom/Acc
Gen/Loc
Dat/Instr
/ĕ/ ↔ elsewhere
/u/ ↔ [-quant]
/ma/ ↔ [±struc, -quant]
The morphosyntactic derivation of dual pronouns proceeds similarly to the
derivation of dual nouns. For example, let us take the dual pronoun subject vy, which
appeared as early as the 11th century in Old Russian in the Ostromir Gospel (1056-57)
(61). I assume that in narrow syntax, the subject DP vy has a structure with a Number
121
(Num) node and a Person (P) node headed by the D head (62). A DP-internal Num head
is assumed following Ritter (1991, 1995). I further postulate a Person (P) node to be
adjoined to the Num node (Bianchi 2006).35
(61)
Old Russian
ne boi-ta
vy
sę
NEG fear-2.DU.IMP 2.DU.NOM
‘Don’t be afraid you two.’
REFL
(Ostromir Gospel, Math. 28.5, 203:5)
(62)
Structure of the Old Russian 2nd person Dual Pronoun vy in NS
D
D
D
u spr, u part
u sg, u aug
Num
√
Num
P
[-sg -aug] [-spkr, +part]
The interpretable phi-features of the 2nd person dual pronoun vy are located on
the Num and P heads. The Num head hosts the number features [-sg -aug] representing
dual number, and the P head hosts the person features – [-spkr, +part] representing the 2nd
person (Harley 2008). The D head of the DP has the uninterpretable phi-features: 2nd
person and dual number features, which need to be valued. Valuation of the
uninterpretable person and number features on D occurs when the interpretable 2nd person
and dual number features are copied from the P and Num heads onto the D head (63).
35
Bianchi (2006) argues for a separate syntactic projection of Person in the structure of Italian DPs.
122
(63)
Syntactic Structure of the Old Russian 2nd Dual Pronoun vy after Valuation
of D
D
D
Num
√ P
Num
[-spr, +part] [-sg, -aug]
D
-spr, + part
-sg -aug
As I have previously established, morphological case features do not appear in the
structure of a DP in the narrow syntax. Kase nodes and Kase features are added to the
structure of a DP at MS. The 2nd person dual pronoun vy which occurs in the nominative
case has its nominative case features added at the K node as shown in (64).
(64)
Structure of the Old Russian 2nd Dual Pronoun vy at MS
D
D
D
-spr, +part
-sg -aug
Num
√ P
Num
[-spr, +part]
Num
[-sg -aug]
K
[+struc, -int, -inh, -quant]
The last step in the morphosyntactic derivation of the dual pronoun vy is to fuse
person, number, and case features before Vocabulary Insertion at PF. A language-specific
123
fusion rule applies to Old Russian at MS to fuse person, number, and case features in one
portmanteau morpheme (65). The final structure of the Old Russian dual nominative
pronoun before Vocabulary Insertion is shown in (66).
(65)
Fusion Rule for OR P/Num/K Nodes
[P -spr, +part]
[Num -sg, -aug]
[K+struc, -int, -inh, -quant] →
[P/Num/K -spr, +part, -sg, -aug, +struc, -int, -inh, -quant]
(66)
Structure of the Old Russian Dual Pronoun vy with a Fused P/Num/K Node
at MS
D
D
D
-spr, +part
-sg, -aug
Num
√vy
P/Num/K
[-spr, +part, -sg, -aug, +struc, -int, -inh, -quant]
124
3.3. Impoverishment, Fission, and Morphosyntactic Simplification of the
Slavic Dual
It is well known that there is no isomorphism in the mapping between syntactic
terminal nodes and phonological form. If this were the case, then for each syntactic node
in the narrow syntax there would be one corresponding VI at PF. However, such one-toone correspondence obtains only in the unmarked case. In many instances,
morphosyntactic features residing at syntactic nodes are altered by the following
operations – Impoverishment, Fission, and fusion. All of these morphological operations
apply at MS before Vocabulary Insertion at PF (67).
(67)
Impoverishment, Fission, and Fusion at MS
Narrow Syntax (NS)
Morphological Structure (MS)
Impoverishment, Fission, and
Fusion
Phonological Form (PF)
Vocabulary Insertion
Impoverishment is an operation which deletes certain morphosyntactic features at
the level of MS prior to Vocabulary Insertion. In the literature (Bonet 1991, Halle 1997,
125
Noyer 1997, Bobaljik 2002, Harley 2008) Impoverishment rules have been used to
explain syncretism and metasyncretism of inflectional categories. When an
Impoverishment rule applies, a certain morphosyntactic feature or features is/are deleted
in a specific context. As a result of deletion, the same VI can be inserted into several
syntactic nodes represented by different morphosyntactic feature bundles. Due to
Impoverishment, fewer VIs can realize a greater number of morphosyntactic feature
bundles in syntactic nodes. This shows that there is no isomorphism between syntactic
nodes in the narrow syntax and their phonological exponents at PF.
I argue that due to the markedness of its feature values, the Slavic dual is subject
to Impoverishment of other orthogonal features, such as person, gender, and case. An
example of Impoverishment of case features can be seen in the dual suffixes of Old
Russian nouns (68).
(68)
Old Russian Dual Nominal Suffixes (Žolobov 2001:51)
Declension Class
*a-stem
*ja-stem
*o-stem
*jo-stem
*u-stem
*i-stem
consonant-stem
Nom/Acc
Gen/Loc
Dat/Instr
-ĕ
-i
masc. –a
neut. –ĕ
masc. –ja
neut. –i
-y
-i
-i
-u
-ju
-u
-ma
-ma
-ma
-ju
-ma
-u
-u
-u
-ma
-ma
-ma
It is striking that in the dual there are three suffixes which realize six cases of Old
Russian nouns in each of the seven nominal declension classes. There is a clear pattern of
126
Nom/Acc, Gen/Loc, and Dat/Instr instances of case syncretism which cuts across seven
declension classes of Old Russian nouns. The same pattern of syncretism occurs in the
Old Russian pronouns. When the same pattern of syncretism occurs across different
grammatical categories, it is called metasyncretism. Thus, the nominal suffixes in (68)
demonstrate metasyncretism of case features in the dual.
Metasyncretism of case features in the nominal dual can be accounted for in a
principled way via rules of Impoverishment. Impoverishment is a very powerful
mechanism which, in principle, can delete any features from a morphosyntactic
representation in a syntactic node. However, not any feature can be deleted. I assume that
the grammar of a language contains markedness statements containing information about
default values of different morphosyntactic features (Calabrese 2011, Harbour 2011).
These markedness statements constrain feature deletion. According to grammatical
markedness statements, deletion of a feature occurs only if a certain value of a feature is
marked, or if it appears in an environment of another marked feature value.
I claim that the marked feature bundle of the dual; that is the feature combination
[-sg -aug] triggers deletion of certain case features to capture Nom/Acc, Gen/Loc, and
Dat/Instr metasyncretisms. I propose Impoverishment rules for the accusative, genitive,
and dative cases in (69a-c). At the terminal node of the accusative Kase, the feature [+int]
is deleted to yield Nom/Acc case syncretism. Likewise, at the genitive Kase node the
[+struc,+quant] features are deleted to yield Gen/Loc case syncretism. Lastly, at the
Dative Kase node, the [+inh, +quant] features are deleted to yield Dat/Instr case
syncretism.
127
(69)
Impoverishment Rules for Old Russian Dual Nominal Suffixes
a. Acc Kase [+int] → ∅/ Num˚ [-sg -aug]
b. Gen Kase [+struc, +quant] → ∅/ Num˚ [-sg -aug]
c. Dat Kase [+inh, +quant] → ∅/ Num˚ [-sg -aug]
In contrast to Underspecification, which cannot explain why case features have to
be always underspecified in the dual across different VIs representing seven declension
classes of Old Russian nouns, Impoverishment provides a principled explanation of the
metasyncretic patterns because case features are deleted from a morphosyntactic
representation before Vocabulary Insertion can apply. An Impoverishment analysis
entails that the absence of case features in the dual of Old Russian nominal suffixes is not
simply an accident of vocabulary item specification, but is a systematic phenomenon in
Old Russian morphosyntax.
Fission is another operation of MS which is key to my analysis of diachronic
changes in the Slavic dual. Fission (Noyer 1997, 1998; Halle 1997; Embick and Noyer
2007) accounts for cases where a single morphosyntactic feature bundle is spit into
several positions of exponence, i.e. several places in the morphosyntactic structure where
additional phonological exponents can be inserted.
An example of Fission can be seen in the dual pronouns of Contemporary
Standard Slovenian (70). In Contemporary Standard Slovenian, the dual nominative in
the 1st, 2nd, and 3rd persons consists of two morphemes - the plural pronoun mi/ vi/ oni
and the numeral dva/dve (‘two’). Historical data show that by the 16th century, Slovenian
dual and plural pronoun forms in the 1st and 2nd persons had become syncretic, and were
128
realized by the VIs /mi/ and /vi/. To distinguish dual forms from the plural, new dual
forms midva and vidva appeared in 16th century Slovenian in the 1st and 2nd person
nominative (Derdanc 1998, 2003, Jakop 2008b).
(70)
Dual Personal Pronouns in Contemporary Standard Slovenian (Derganc
2003:166)
Num/Person
DU 1.
2.
3.
Nom
Acc
Gen
Loc
Dat
Instr
mi-dva,
mi-dve/medve
na-ju
na-ma
vi-dva,
vi-dve/ve-dve
ona-dva,
oni-dve/onedve
va-ju
va-ma
nji-ju
nji-ma
To account of the emergence of ‘new’ dual pronouns in the 16th century
Slovenian, I suggest that Fission applied at MS to split a marked feature bundle [-sg -aug]
of the dual into two features [-sg] and [-aug] (71). As a result of Fission, the Slovenian
dual forms midva and vidva are no longer syncretic with the plural forms mi and vi. More
importantly, a marked feature combination [-sg -aug] has become morphosyntactically
simplified due to creation of two simpler morphemes – i and -dva.
(71)
16th Century Slovenian Dual Pronoun Fission Rule
[-sg -aug] Num → [-sg] Num [-aug] Num
129
At PF, two VIs are inserted into two positions of exponence created by Fission
(72). The VI /dva/ is inserted in the context of the [-aug] feature while the VI /i/ is the
elsewhere form which realizes both the dual and plural.
(72)
/dva/ ↔ [-aug]
/i/ ↔ elsewhere
Fission is a strategy that languages employ to reduce markedness of feature
values. Based on diachronic changes in the Slavic dual, I claim that Fission is used to
reduce value markedness of dual number features in the morphosyntactic representation
by creating two positions of exponence.
130
3.4.
Summary
In this chapter, I have demonstrated how Slavic dual nouns, pronouns, and verbs
(by agreement with a pro/nominal) are derived in Distributed Morphology (Harley and
Noyer 1999). I have further shown that the Slavic dual is represented by a portmanteau
morpheme which is featurally complex. It contains a bundle of phi-features – person,
number, and gender as well as case features. These features are distributed among
different components of grammar.
Dual number is represented by the [-sg -aug] features which are present in Narrow
Syntax. Morphological Structure, a component, which mediates Narrow Syntax and
Phonological Form, plays a crucial role in restructuring of morphosyntactic
respresentation of the Slavic dual. At MS, Kase nodes and Kase features are added.
More importantly, I have demonstrated that economy operations of
Impoverishment and Fission, which apply at MS, offer a key explanation of diachronic
changes in the Slavic dual. Both Impoverishment and Fission apply to reduce markedness
of feature values of the dual and result in simplification of morphosyntactic
representation of the dual.
131
CHAPTER 4 THE TWO NUMBER FEATURES OF THE SLAVIC
DUAL AND THE PATTERNS OF CROSS-LINGUISTIC VARIATION
IN SOUTH, WEST, AND EAST SLAVIC
4.0.
Introduction
One of the most interesting aspects of dual number inflection in Slavic is a
mismatch between its formal representation in Narrow Syntax and Morphological
Structure. There are two questions about the morphosyntactic representation of the Slavic
dual which I address in this chapter. First, what are the number features which represent
the Slavic dual in Narrow Syntax? Second, what are the patterns of cross-linguistic
variation in the realization of the Slavic dual at Phonological Form?
In Section 4.1., I assume that dual number marked on Slavic personal pronouns
and verbal agreement suffixes is featurally complex, and is represented as a combination
of the [-singular, -augmented] number features in Narrow Syntax (Noyer 1997, Harbour
2011, Nevins 2011). I define the [±singular] and [±augmented] number features in a
formal semantic framework and explain how the [-singular -augmented] features combine
to deliver the dual semantics. Assuming with (Ritter 1995) that the number features are
hosted by a Num node, I argue that the Slavic dual is represented by a fully specified
morphosyntactic feature bundle [-singular -augmented] in Narrow Syntax.
In Section 4.2., I identify two patterns of cross-linguistic variation in the PF
representation of the Slavic dual: bimorphemic and monomorphemic. I show that the
bimorphemic dual pronouns occur in the three contemporary Slavic languages- Slovenian
(South Slavic), Upper, and Lower Sorbian (West Slavic). The monomorphemic dual
132
pronouns were found in Old Russian (East Slavic) and Kashubian (West Slavic), but over
time were replaced with the plural pronouns. Based on corpora analyses, I argue that in
Slovenian and Sorbian, bimorphemic dual pronouns diachronically developed from their
monomorphemic 'ancestors' while in Old Russian and Kashubian, the monomorphemic
dual pronouns were replaced by the plural pronouns in contemporary Russian and
Kashubian. In Section 4.3., I summarize my findings regarding the two different patterns
of cross-linguistic variation in the Slavic dual pronouns.
133
4.1.
The Two Number Features of the Slavic Dual
I begin with the analysis of the morphosyntactic structure of the Slavic dual in
Narrow Syntax. I argue that in Narrow Syntax the Slavic dual is represented by a
combination of the two features - [-singular -augmented]. I support my argument for this
featural representation by providing evidence of the distribution of the [-singular] and
[-augmented] features in the bimorphemic dual forms in Contemporary Standard
Slovenian , Upper Sorbian, and Lower Sorbian. I show that dual pronouns in Slovenian
and Sorbian are composite and consist of two elements: the plural form and the numeral
two or the dual suffix. The two features which make up the dual are distributed in the
following way: the [-singular] feature is encoded by the plural form whereas the
[-augmented] feature is expressed by the numeral two or a dual suffix.
The Slavic dual has a complex morphosyntactic composition and cannot be
viewed as a simplex feature, such as [dual]. It was noticed by Hale (1973) and Silverstein
(1976) that number features, such as [singular], [dual], [trial], [quadral], … and [plural]
are not atomic. If the feature [dual] were atomic, then it would be impossible to explain
why the dual syncretizes with the plural and the singular cross-linguistically.
Instances of dual/plural and dual/singular syncretism are well attested in the
world’s languages. In Contemporary Standard Slovenian , the pronominal stems mi, vi,
oni are syncretic in the dual and plural (Table 4.1). In Hopi (Uto-Aztecan), the dual and
singular syncretize in the verbal inflection marked by the null suffix on the verb paki
(73a-b) while the dual and plural sincretize in the pronominal inflection encoded by the
134
suffix -ma (73b-c). In Navajo (Na Dene), the dual is syncretic with the plural and realized
by the stem nihí (Table 4.2).
Table 4.1 Dual/Plural Syncretism in Contemporary Standard Slovenian
Num/Perso
n
DL
1.
2.
3.
PL
(73)
1.
2.
3.
Nom
Acc
mi-dva, midve/me-dve
vi-dva, vidve/ve-dve
ona-dva, onidve/one-dve
mi/me
vi/ve
oni/one
Gen
Loc
Dat
na-ju
na-ma
va-ju
va-ma
nji-ju
nji-ma
na-s
va-s
nji-h
na-m
va-m
nji-m
Hopi
a.
miɁ maana paki
that girl.SG enter.SG
‘That girl entered.’
b. mi-ma
maana-t paki
that- PL girl-PL enter.SG
‘Those (two) girls entered.’
c. mi-ma ma-man-t yɨŋya
that- PL RED-girl-PL enter.PL
‘Those girls (many) entered.’
(Jeanne 1978:73)
Table 4.2 Navajo Personal Pronouns (Young & Morgan 1980:22)
Singular
1. shí
2. ni
3.bí/hó
Dual
Nihí
Nihí
*
Plural
da-nihí
da-nihí
daa-bí/daa-hó
Instr
na-mi
va-mi
nji-mi
135
Based on the cross-linguistic evidence of instances of dual/plural and
dual/singular syncretism, it is reasonable to assume that atomic number features
[singular], [dual], and [plural] are not viable and should be decomposed into
combinations of the two bivalent features - [±singular] and [±augmented].36 A bivalent
approach to number features is more economical since there are only two features which
are posited with binary values instead of several features in a monovalent approach.
Following Noyer’s (1997) fundamental insight, I assume that the two bivalent
features [±singular] and [±augmented] are sufficient to encode a singular ~ dual ~ plural
number distinction (74). In Noyer’s system, the dual is represented as a combination of
the [-singular -augmented] features. The key observation about the featural composition
of the dual is that it shares the [-augmented] feature with the singular and the [-singular]
feature with the plural.
(74)
Feature-Based Representations of Number Categories (Noyer 1997)
Singular
[+singular -augmented]
Dual
[-singular -augmented]
Plural
[-singular +augmented]
Following Link (1983) and Harbour (2008), I assume a lattice-theoretic approach
to the semantics of nominal predicates. According to this approach, the denotation of a
nominal predicate represents a set of atoms and the set of subsets of the atoms. The
lattice-theoretic semantic approach, which I assume, is shown in Figure 4.1.
36
The category of number in pronouns can be also represented by means of a feature geometry (Harley &
Ritter 2002). In Harley and Ritter’s feature geometric approach, number features are privative (i.e. not
bivalent).
136
Figure 4.1 Lattice-Theoretic Approach to Nominal Predicates (Harbour 2008:67)
.
. . .
.
.
.
.
.
.
.
.
.
s
t
u
v
w
.
. . .
x
The bottom row of the lattice in Figure 4.1 represents atoms. For example, s, t, u,
v, w, x are atoms. The points of intersection of the lines above the atoms represent sets.
For example, the point of intersection above the atoms s and t represents a set {s, t}. The
next point of intersection above t corresponds to a set {s, t, u}. The lattice in Figure 4.1
illustrates that the lowest row of elements represents individual atoms or singular
number. The level above single atoms represents sets consisting of two atomic members.
Such sets of two atoms correspond to dual number. The level above dyads represents sets
which consist of three members, the triads, etc.
I define the features [±singular] and [±augmented] in a formal semantic
framework where features are represented as truth-conditional predicates (Noyer 1997;
Harbour 2008, 2011b; Nevins 2011). The feature [+singular] applies to a predicate (P)
and picks out atomic elements which satisfy P, where x is an atom relative to P if and
only if there is no y which is a subset of x (75 a). The [-singular] feature picks out
137
referents which are non-atomic, i.e. not atoms (75 b). I assume feature negation as shown
in (76). Minus values of numbers features are defined as negation of the plus values.
(75)
The Definition of the [±singular]
a. [+singular] = (λ x) [atom (x)]
iff ¬ ∃y [P(y) ∧ y ⊂ x]
b. [-singular] = ⌐ (λ x) [atom (x)]
(76)
Feature Negation
[+F] = ⌐ [+F]
The feature [±singular] breaks up the lattice assumed above into atomic and non-atomic
regions (Figure 4.2). The [-singular] feature defines atomic elements while the [-singular]
feature defines non-atomic elements.
A formal semantic definition of the [±augmented] feature is given in (77). The
feature [±augmented] is interpreted in terms of a set and its proper subset relation. The
feature [+augmented] applies to a predicate P and picks out properties which hold of x
and its proper subset y (77 a). The [-augmented] feature is interpreted as a negation of the
[+augmented] feature, which means that that for a predicate (P) y has no proper subset x
(77 b). If interpreted in isolation, the [±augmented] feature breaks up the lattice into two
partitions (Figure 4.3).
A formal semantic definition of the [±augmented] feature is given in (77). The
feature [±augmented] is interpreted in terms of a set and its proper subset relation. The
feature [+augmented] applies to a predicate P and picks out properties which hold of x
and its proper subset y (77 a). The [-augmented] feature is interpreted as a negation of the
138
[+augmented] feature, which means that that for a predicate (P) y has no proper subset x
(77 b). If interpreted in isolation, the [±augmented] feature breaks up the lattice into two
partitions (Figure 4.3).
Figure 4.2 Lattice-Theoretic Semantics of the [±singular] Feature (Harbour
2008:69)
[ + singular]
. . .
(77)
.
.
.
.
.
.
The Definition of the [±augmented]
a. [+augmented] = λ P ∃y [P(x) ∧ P (y) ∧ y ⊂ x]
b. [-augmented⟧ = ⌐ λ P ∃y [P(x) ∧ P (y) ∧ y ⊂ x]
. . .
[ - singular]
139
Figure 4.3 Lattice-Theoretic Semantics of the [±augmented] Feature (Harbour
2008:69)
[ + augmented]
. . .
.
.
.
.
.
.
. . .
[ - augmented]
The [±singular] and the [±augmented] features combine to produce singular,
dual, and plural number. When these two features co-occur, they are interpreted by
function application (78). According to a rule of semantic composition in (78), the
[±singular] feature is a function of the [±augmented] feature. This is a key point to
semantic composition of the dual and plural.
(78)
Semantic Composition
[±singular] and [±augmented] = [±augmented] ([±singular])
Let us consider the semantic composition of the singular. The [+singular] feature
selects elements which are atoms without proper subsets (75 a). If a predicate (P) is
[+singular], its reference set contains only one member, such as {a} whose cardinality
equals 1. Since the reference set {a} is atomic, it contains no proper subsets. Therefore,
140
for the value [+singular], the reference set is [-augmented]. Thus, the singular or the
reference set of cardinality 1 is defined as [+singular -augmented].
The featural composition of the plural proceeds in the following way. The plural
has a cardinality set which is greater than 2, and contains at least 3 members in a set, such
as {a, b, c}. The reference set {a, b, c} is non-atomic; therefore, the plural is [-singular]
(75b). The [-singular] feature is interpreted as a function of the [+augmented] feature.
According to the definition of the [+augmented] feature, a non-atomic property of the
[-singular] should also hold of its proper subset The reference set of the plural {a, b, c}
contains at least one proper subset, such as {a, b} which is non-atomic. Since the
reference set of the plural contains a non-atomic proper subset, the feature [+augmented]
applies. Thus, the plural or a reference set of the cardinalty 3 is a combination of the
[-singular +augmented] features.
The dual is a combination of the [-singular -augmented] features. A formal
semantic definition of the dual is given in (79). According to this definition, the dual
consists of a non-atomic x without a proper subset y which is also non-atomic.
(79)
The Dual
[-singular -augmented]
= [-augmented ([-singular])
= (λP) (λx: P(x)) ¬∃y [P(y) ∧ y ⊂ x] ((λx) [¬ atom (x)])
= (λ x : ¬atom (x)) ¬∃y[¬atom (y) ∧ y ⊂ x]
The semantic composition of the dual proceeds in the following way. The
referential cardinality of the dual equals 2, and its reference set consists of two elements,
141
such as {a, b}. This reference set is non-atomic, and hence is [-singular]. The reference
set {a, b}contains no proper subsets which are non-atomic since its proper sets {a} and
{b}are atomic. Due to the absence of a proper non-atomic subset in the set {a, b}, the
feature [-augmented] applies. Thus, a dual or a dyad consists of a minimum of two
elements in its subset which are atomic.
I have established that semantically the dual is composed of the two features [-singular -augmented]. Now I will illustrate how the number features of the Slavic dual
are represented in Narrow Syntax. Let us consider the 2nd person nominative dual
pronoun vy in Old Russian from the pronominal paradigm obtained from the Ostromir
Gospel (1056-1057), one of earliest Old Russian manuscripts (Table 4.3).
Table 4.3 Old Russian Personal Pronouns in the Ostromir Gospel (1056-1057)
(Corpus Manuscript)
Num/Person
SG
DU
PL
Nom
Acc
Gen
Loc
Dat
Instr
1.
azŭ
men-e
mŭn-ě
mŭno-jǫ
2.
ty
teb-e
teb-ě
tob-jǫ
1.
vě
ny
na-ju
na-ma
2.
vy
va
va-ju
va-ma
1.
my
na-sŭ
na-mŭ
na-mi
2.
vy
va-sŭ
va-mŭ
va-mi
I assume that in NS number features are hosted by a special designated Number
node (Num) within the DP (Ritter 1995). According to this assumption, the NS structure
of the Old Russian 2nd person dual pronoun vy is as shown in (81). In the syntactic
142
structure (81), the terminal Num node has a fully specified morphosyntactic feature
bundle [-singular -augmented] which represents the dual. The Person node (P)
represented by the [-speaker, +participant] features is a complement to the Num head.
(80)
The Structure of the OR 2nd Person Dual Pronoun vy in Narrow Syntax
D
D
Num
P
[-spr, +part]
Num
[-sg -aug]
The NS representation of an OR dual pronoun as the [-singular -augmented] feature
combination is confirmed by evidence from verbal agreement. In Old Russian, present
tense verbs are marked for dual number by the suffixes –vě and –ta which are different
from the plural suffixes (Table 4.4).
Table 4.4 Present Tense Verbal Suffixes in Old Russian (Možejko and Ignatenko
1988:141)
Person/Number
1.
2.
3.
Singular
-u
-šĭ
-tĭ
Dual
-vě
-ta
-ta
Plural
-mŭ
-te
-tŭ
The Ostromir Gospel provides evidence that the 2nd person pronoun vy had
become syncretic with the plural as early as in the 11th century. When the 2nd person dual
pronoun vy, which is syncretic with the plural, occurs in a sentence, the verbal suffix –ta
indicates agreement with the dual subject (81). The 2nd person plural pronoun vy shows
plural agreement marked by the suffix –te when the subject of the sentence is plural (82).
143
(81)
Old Russian
je-go
že vy
glje-ta
jako slĕpŭ
rodi-sę
that-ACC EMPH 2.PL. NOM say-2.DU.PRES as
blind
born-REFL
‘You two say that he was born blind.’
(Ostromir Gospel, John 9.19)
(82)
Old Russian
i
vy že
sŭvĕdĕtelĭstvu-je-te
and 2.PL also testify-TH-2. PL.PRES
‘And you (more than two) will also testify.’
(Ostromir Gospel, John 15.27)
144
4.2.
The Two Patterns of Variation: The Bimorphemic and
Monomorphemic Duals
There are two patterns of variation in the PF realization of the dual in Slavic
languages: bimorphemic and monomorphemic. Bimorphemic dual pronouns are found in
Contemporary Standard Slovenian, Upper Sorbian, and Lower Sorbian whereas the
monomorphemic dual pronouns used to occur in Old Russian (11th-14th centuries) and
Kashubian in the 19th and the first part of the 20th centuries (Table 4.5).
Table 4.5 The Distribution of Bimorphemic and Monomorphemic Dual Pronouns
Bimorphemic Dual
Monomorphemic Dual
Contemporary Standard Slovenian
Kashubian
Contemporary Upper Sorbian
Old Russian
Contemporary Lower Sorbian
I have claimed that in NS the Slavic dual is represented by a combination of the two
number features - [-singular -augmented]. My claim for this combinatorial representation
of the dual is supported by the distribution of the [-singular] and [-augmented] features in
the bimorphemic dual forms of personal pronouns in Contemporary Standard Slovenian,
Upper Sorbian, and Lower Sorbian. I analyze the bimorphemic structure of dual pronouns
in the three contemporary Slavic languages and show how the features [-singular] and
[-augmented] are distributed in the dual forms.
145
Dual pronouns in Contemporary Standard Slovenian, Upper Sorbian, and Lower
Sorbian are bimorphemic in their morphological structure and are composed of the plural
stem and the numeral two, or the dual suffix -j as shown in (83).
(83)
The Structure of Bimorphemic Dual Pronouns in Contemporary Standard
Slovenian, Upper Sorbian, and Lower Sorbian
Dual = plural + dva/-j (‘two’)
In the structure of the Contemporary Standard Slovenian dual pronouns mi-dva/vidva/oni-dva, the numeral dva (‘two’) is very transparent. In the structure of the Upper
Sorbian dual pronouns mó-j/wó-j/wona-j, the dual suffix –j had originated from the
nominative suffix –j of the numeral dwa-j (‘two’) (Derganc 1998).
The origin of the dual suffix –j in the structure of the Lower Sorbian dual pronouns
me-j/we-j/wone-j is different. Since the nominative of the numeral dwa in Lower Sorbian
does not have a dual suffix –j, it is suggested that this suffix had stemmed from the
suffix –j in the genitive forms of the personal pronouns na-ju, wa-ju, je-ju and the
numeral dwe-ju (Schuster-Šewc 2000).
4.2.1. The Bimorphemic Dual in Contemporary Standard Slovenian
The Contemporary Standard Slovenian pronominal dual is one of the clearest
examples of the combination of the plural morpheme and number two. In Contemporary
Standard Slovenian, the morphemic structure of the pronominal dual is very transparent.
146
The dual form consists of the plural stem mi/vi/oni and the numeral dva (‘two’) (Table
4.6).
Table 4.6 Bimorphemic Dual Pronouns in Contemporary Standard Slovenian
(Derganc 2003:166)
Num/Person
SG
1.
2.
3.
Nom
jaz
ti
on, ona, ono
DL
mi-dva, midve/me-dve
vi-dva, vidve/ve-dve
ona-dva, onidve/one-dve
mi/me
vi/ve
oni/one
1.
2.
3.
PL
1.
2.
3.
Acc
Gen
men-e
teb-e
njega, nje
Loc
Dat
na-ju
men-i
teb-i
njem, njej
njemu,
njej
na-ma
va-ju
va-ma
nji-ju
nji-ma
na-s
va-s
nji-h
Instr
men-oj
teb-oj
njim, njo
na-m
va-m
nji-m
na-mi
va-mi
nji-mi
I argue that in Narrow Syntax the numeral dva (‘two’) in Contemporary Standard
Slovenian bimorphemic dual pronouns is represented by the Number node. Let us
analyze the syntactic structure of the 1st person dual pronoun midva. In NS, the [-singular
-augmented] feature bundle of the dual is hosted by the Num node, a complement of the
D head (84).37
The 2nd and 3rd dual pronouns vidva and onidva have the same syntactic structure except for the
difference in the person features of the P node.
37
147
(84)
The Structure of the Contemporary Standard Slovenian Dual Pronoun midva in Narrow Syntax
D
D
Num
P
[+spr, -part]
Num
[-sg -aug]
The evidence for the presence of the feature combination [-singular-augmented] of
the Num node in NS comes from verbal agreement. In Contemporary Standard
Slovenian, verbs are marked for dual number by the suffix –va in the 1st person and the
suffix –ta in the 2nd and 3rd persons (Table 4.7). As we can see, the dual verbal suffixes
are morphologically different from their plural counterparts.
Table 4.7 Present Tense Agreement Suffixes in Contemporary Standard Slovenian
(Derganc 2003:167)
Number
SG
DU
PL
1
-m
-va
-mo
Person
2
-š
-ta
-te
3
-a
-ta
-jo
Let us examine an example of verbal agreement with dual subjects in Contemporary
Standard Slovenian. The suffix -va shows agreement with the 1st person dual midva, and
the suffix –a attached to a participle shows agreement with the dual number and
masculine gender of the same pronoun (85).
148
(85)
Mi-dva
bo-va
šl-a
po levi poti, vi-dva
pa po desni.
1.PL.two be.1.DU.FUT
go.PRT-DU.MASC on left road, 2.PL-two and on right
‘The two of us will take the road on the left while the two of you the one on the
right.’
(Derganc 2003:169)
4.2.2. The Bimorphemic Dual in Contemporary Upper Sorbian
The morphemic composition of the pronominal dual in Upper Sorbian is less transparent,
although it is very similar to its Contemporary Standard Slovenian counterpart. In Upper
Sorbian, the 1st, 2nd, and 3rd person dual is composed of the plural stem mó/wó/won and
the suffix -j derived from the masculine numeral dwaj (‘two’) (Table 4.8).38 The 1st and
2nd person dual stems mó and wó which are part of the dual are phonologically different
from the stems my and wy of the plural.
There is a historical phonology explanation for this difference between the two stems.
The stems mó and wó were the result of labialization and diphthongization which had
occurred in Old Sorbian in the 16th century (Schaarschmidt 1998, Scholze 2007).
Schaarschmidt argues that in most Upper and Lower Sorbian dialects, the allophone y
became labialized to either u or ó (1998:127).
The labialization of y had originated in the Upper Sorbian dialects. The Old Sorbian high
central vowel [ɨ] (allophone of /i/) spelt as y in the plural stems my and wy was
labialized as /u/ after the labials /m/ and /w/. Then, the vowel /u/ was diphthongized as
/uo/ spelt as ó in the Upper Sorbian dual forms mó-j and wó-j.39 We can summarize
38
The 3rd person dual pronouns won-a-j and won-e-j have the structure identical to their 1st and 2nd person
counterpart in that they are composed of the pronominal stem won- and the dual suffix -j. The vowels a and
e are thematic vowels showing masculine vs. non-masculine gender.
39
For more details on labialization and diphthongization of the allophone [ɨ] in Upper and Lower Sorbian,
see Schaarschmidt (1998).
149
labialization and diphthongization of y as the following phonological change: Old
Sorbian y > Upper Sorbian ó.
Table 4.8 Bimorphemic Dual Pronouns in Upper Sorbian (Schaarschmidt
2002:2426)
Num/Person
Nom
Acc
Gen
Loc
Dat
Instr
SG 1.
ja
mn-je
mn-i
mn-u
2.
ty
teb-je
teb-i
tob-u
3.
wón (masc)
won-o/wone (neut)
won-a
(fem)
je-ho,
nje-hojón,
njón
(masc)
je/jo,
nje/njo
(neut)
ju, nju
(fem)
ni-m
je-ho, nje(masc/neut)
ho
(masc/neut) nj-ej (fem)
je-je, nje-je
(fem)
je-mu, njemu
(masc/neut)
je-j, nj-ej
(fem)
DL 1.
mó-j
na-ju
na-maj
2.
wó-j
wa-ju
wa-maj
3.
won-a-j
je-ju, ne-ju
jim-aj, njim-aj
Non-Masc
won-e-j
ni-m
(masc/neut
)
nj-ej (fem)
je-j,
nje-j
PL 1.
my
na-s
na-m
nam-i
2.
wy
wa-s
wa-m
wam-i
3.
won-i
ji-m, ni-m
ni-mi
Non-Masc
won-e
ji-ch, ni-ch
nich
je, nje
The NS representation of the Upper Sorbian dual pronouns is identical to its
Slovenian counterpart. The 2nd person dual pronoun wó-j has the syntactic structure as
150
given in (86). As I have argued above, the NS feature combination [-singular augmented] representing the dual finds its reflex in the verbal agreement. In Upper
Sorbian, verbs are marked by the suffix –moj in the 1st person and the suffixes –taj/tej in
the 2nd and 3rd persons to reflect agreement with dual subjects (Table 4.9).
(86)
The Structure of the Upper Sorbian 2nd Person Dual Pronoun wó-j in
Narrow Syntax
D
D
Num
P
[-spr, +part]
Num
[-sg -aug]
Table 4.9 Present Tense Agreement Suffixes in Upper Sorbian (Stone 1993:645)
Number
SG
DU
PL
1
-u
-moj
-my
Person
2
-š
-taj/-tej
-če
3
-∅
-taj/-tej
-u
The 1st and 2nd person dual pronouns show agreement with the verb (87-88). The 1st dual
pronoun mój triggers agreement with the verb be in the form smój while the 2nd person
dual pronoun wój induces agreement with the verb be in the form stej.
151
(87)
Upper Sorbian
Ha mój
smój
wostał-oj tam
and 1.PL-DU be.1.DU stayed- DU there
‘And we two stayed there.’
(Scholze 2007:124-125)
(88)
Upper Sorbian
Wój
stej
najbóle skók
2.PL-DU be.2.DU most
jump
‘You two are the fastest.’
(Scholze 2007:124-125)
4.2.3. The Bimorphemic Dual in Contemporary Lower Sorbian
In Lower Sorbian, the 1st and 2nd person dual pronouns me-j and we-j are identical
in their morphemic composition to their Upper Sorbian counterparts (Table 4.10). The
dual pronouns in Lower Sorbian consist of the plural stem me or we and the suffix –j,
which had originated from the genitive forms of the personal pronouns na-ju, wa-ju, je-ju
and the numeral dwe-ju (‘two’) (Schuster-Šewc 2000). The only difference between the
Lower Sorbian and Upper Sorbian pronominal duals is the vowel /e/ which follows the
labials in the forms me-j and we-j.
As in the case with Upper Sorbian, a phonological rule of labialization applied in
Old Sorbian to produce a Lower Sorbian vowel /u/ (Schaarschmidt (1998). Then some
other phonological rule applied which lead to a phonological change from /u/ to /e/ in the
Contemporary Lower Sorbian dual.40 Schaarschmidt (1998:128) suggests that in Lower
Sorbian, y generally changed into u after w with the exception of the 2nd person plural
pronoun wy. He further suggests that the change from y to u also occurred after other
40
To my knowledge, it is not exactly clear which phonological rule applied to account for the phonological
change from /u/ to /e/ in Contemporary Lower Sorbian. More research is needed to identify this rule.
152
labials, such as b, m, p. We can summarize labialization of y in Lower Sorbian as a
phonological change from Old Sorbian y > Lower Sorbian u > Contemporary Lower
Sorbian e.
Table 4.10 Bimorphemic Dual Pronouns in Lower Sorbian (Stone 1993:622)
Num/Person
SG
Nom
Acc
Gen
Loc
Dat
Instr
1.
ja
m-ě
mj-e
mn-jo
m-ě
mn-u
2.
ty
3.
won, won-o,
won-a
me-j
jo-go, jo, jo-go, jeju
je
na-ju
2.
we-j
wa-ju
3.
1.
won-e-j
my
2.
wy
3.
won-i
DU 1.
PL
teb-je
je-ju, je-j
tob-u
njo-m,
njej
ni-m, nje-ju
wa-ma
je-j
na-s
ni-ma
wa-s
ji-ch, je
jo-mu,
jej
na-ma
ji-ch
ni-ch
ji-ma
na-m
ni-ma
nam-i
wa-m
wam-i
ji-m
nim-i
In Lower Sorbian, syntactic agreement with dual subjects is reflected in the dual
suffixes which are morphologically distinct from the plural ones (Table 4.11). In the 1st
person dual, the verb is marked by the suffix –mej whereas in the 2nd and 3rd person it is
marked by the suffix –tej.
In the Lower Sorbian newspaper of the 19th century, the Bramborski Serbski
Casnik (1848-1880), we find examples of verbs marked for dual agreement. The verb
glědaš (‘look’) has the suffix –mej to show agreement with the 1st person dual pronoun
153
mej (89). Both the auxiliary verb byš (‘be’) and the verb groniš (‘say’) are inflected for
the 2nd person dual (90).
Table 4.11 Present Tense Verbal Suffixes in Lower Sorbian (Stone 1993:647)
Number
SG
DU
PL
(89)
(90)
1
-m
-mej
-my
Person
2
-š
3
-∅
-tej
-šo
-u
Mej
gledach-mej
wójadnom jaden na drugego.
1.DU.NOM look-1.DU.AOR. together one on other
‘We two looked at each other.’
(Bramborski Serbski Casnik
1852:25, Niedersorbisches
Textkorpus)
Wej
stej
grońi-l-ej
2.DU.NOM be.2.DU say-PERF-2.DU
‘We two have said.’
(Bramborski Serbski Casnik
1850:04, Niedersorbisches
Textkorpus)
4.2.4. The Distribution of the [-singular] and [-augmented] Features in the
Bimorphemic Dual in Slovenian, Upper, and Lower Sorbian
As I have argued, Contemporary Standard Slovenian, Upper, and Lower Sorbian
have pronominal duals which are composed of the plural stem and the numeral two or the
suffix –j derived from this numeral. I claim that the number features [-singular] and
154
[-augmented], which make up the dual, are distributed in a predictable way. The feature
[-singular] is encoded by the plural pronominal stem while the feature [-augmented]
corresponds to the numeral two or the dual suffix –j.
Let us consider the distribution of the [-singular] feature in the Contemporary
Standard Slovenian, Upper, and Lower Sorbian bimorphemic duals. The evidence that
this feature is encoded by the plural pronominal stem comes from cases of dual/plural
syncretism. In Contemporary Standard Slovenian personal pronouns, the dual and the
plural are syncretic and realized by the VI /mi/ and /vi/ (91). The dual/plural pronominal
stem syncretism follows from the fact that the dual and the plural share the feature
[-singular].
(91)
The Dual/Plural Pronominal Stem Syncretism in Contemporary Standard
Slovenian
Dual
Plural
1. mi-dva
mi
2. vi-dva
vi
3. oni-dva
oni
Diachronic data from the 16th century Old Slovenian found in Dalmatin’s (1584)
translation of the Bible provides more evidence for the dual/plural syncretism in the 1st
and 2nd person pronouns my and vy (Table 4.12). The 1st person plural pronoun my
appears with the verb marked by the dual suffix –va to indicate dual reference of the
subject (92). The 2nd person plural vy occurs with the verb marked by the suffix –ta to
155
show agreement with a dual subject (93). In both examples (92-93), the subject is used in
the plural form, but the dual agreement suffixes point to the dual reference.
Table 4.12 Dual/Plural Pronominal Stem Syncretism in the 16th Century Slovenian
Person
1.
2.
(92)
Dual
my
wy
Plural
My
Wy
Old Slovenian
Mojʃter, my hozhe-va, de
nama
ʃturiʃh,
kar
te
bo-va
teacher 1.PL desire-1.DU COMP 1.DU.DAT do-2.SG.PRES which 2.SG be-1.DU.FUT
proʃsila.
ask
‘Teacher, we two wish that you do for us two what we two will ask you.’
(Dalmatian Bible, Mr 10.35)
(93)
Old Slovenian
Vy ne-veʃ-ta,
kaj prosʃi-ta.
2.PL NEG-know-1.DU.PRES what ask-1.DU.PRES
‘You (two) don’t know what you (two) are asking for.’
(Dalmatian Bible, Mr 10.38)
As I have discussed earlier, in Upper and Lower Sorbian, the pronominal stems
mó/wó and me/we in the dual forms are phonologically different from the pronominal
stems my/wy of the plural forms due to phonological changes which had occurred in Old
Sorbian. Therefore, it is not apparent that pronominal stems are syncretic in the dual and
plural in Lower and Upper Sorbian.
I argue that diachronically the pronominal stems of the dual and plural were
syncretic in earlier stages of Sorbian since they shared the [-singular] feature. I support
156
this claim by presenting historical evidence of the dual/plural syncretism which had
occurred in Sorbian in the 16th century (Table 4.13).
Table 4.13 Dual/Plural Pronominal Stem Syncretism in the 16th Century Sorbian
Person
Dual
my
wy
1.
2.
Plural
My
Wy
In the 16th century, the Sorbian 2nd person pronoun wy began to syncretize in the dual
and plural. The verbal suffix -tai indicates that the pronoun wy has a dual reference (94).
At the same time, the dual/plural syncretism had also spread to the 1st person. The verbal
suffix –moj shows that the reference of the pronoun my is dual rather than plural (95).
(94)
Old Sorbian
Wy widz-i-tai
a schlysch-i-tai
1.PL see-TH-2.DU.PRES and hear- TH-2.DU.PRES
‘You two see and listen.”
(Derganc 1998:50)
(95)
Old Sorbian
My chze-moj
1.PL want- 1.DU.PRES
‘We two want to …’
so,
schtoz …
REFL COMP
(Derganc 1998:52)
Based on the diachronic evidence of the dual/plural pronominal stem syncretism
in 16th century Sorbian, I conclude that in Upper and Lower S the pronominal stems
my/wy encode the [-singular] feature. Similarly, in Contemporary Standard Slovenian, the
dual/plural syncretism of the pronominal stems mi/vi/oni is the consequence of sharing of
157
the [-singular] feature. Thus, I maintain that the [-singular] feature is encoded by the
dual/plural pronominal stem.
I argue that the distribution of the feature [-augmented] follows from the semantic
composition of the dual. Recall that bimorphemic dual pronouns in Contemporary
Standard Slovenian, Upper, and Lower Sorbian are composed of the plural stem and
either the numeral dva (‘two’) or the dual suffix –j. I argue that the numeral dva and a
dual suffix –j encode the [-augmented] feature.
In Narrow Syntax, the dual is represented by the [-singular -augmented] feature
bundle. According to a formal semantic definition of the of the dual, the [-augmented]
feature combines with the [-singular] feature via function application. The [-augmented]
feature takes the [-singular] feature as its function. Therefore, the [-augmented] feature is
interpreted as non-atomic reference set without a non-atomic proper subset. The only
non-atomic subset which contains no non-atomic members in its proper subset is number
2. Thus, the most natural linguistic expression of a dyad is the numeral two or a dual
affix. That is why the numeral dva and the dual suffix –j encode the [-augmented] feature
in Contemporary Slovenian, Upper, and Lower Sorbian.
In sum, I have argued that in NS the Slavic dual is represented by a combination
of the two features – [-singular -augmented]. These two features are distributed in
bimorphemic dual pronouns in Contemporary Slovenian, Upper, and Lower Sorbian in a
way which follows from their semantics. The [-singular] feature is encoded by the plural
pronominal stem while the [-augmented] feature is encoded by the numeral dva (‘two’) or
the dual suffix –j (Table 4.14).
158
Table 4.14 The Distribution of the [-singular] and [-augmented] features in the Bi
morphemic Dual Pronouns in Slovenian, Upper & Lower Sorbian
[-singular]
[-augmented]
mi/vi/oni
Dva
Upper Sorbian
mó/wó/won
-j
Lower Sorbian
me/we/won
-j
Contemporary Slovenian
4.2.5. The Monomorphemic Dual and Its Loss in Kashubian
While the dual is still a ‘living’ category of the grammar in Contemporary
Slovenian, Upper, and Lower Sorbian this is no longer the case in Kashubian. In the
course of the historical development of Kashubian in the 20th century, the
monomorphemic dual pronouns syncretized with the plural, and as a result the dual
number distinction was lost.
Dual pronouns in Kashubian have undergone significant diachronic changes. As
recorded by a 19th century Kashubian linguist Florian Ceyonowa in his Kurze
Betrachtungen über die kaßubische Sprache, als Entwurf zur Gramatik (Brief Notes on
Kashubian Language, an Outline of Grammar) (cir.1860), Kashubian distinguished the
1st, 2nd, and 3rd person dual personal pronouns (Table 4.15). In the nominative case the 1st,
2nd, and 3rd dual forms were ma, va, wón-ji, wón-e.
159
Table 4.15 The Dual Pronouns in Kashubian (1860s) (Duličenko & Lehfeldt
1998:58-59)
Num/Person
SG
1.
jó
2.
te
3.
wón,
wóna,
wóno
ma
DU 1.
PL
Nom
Acc
Gen
Loc
Dat
Instr
Mje
mną
ce
jeho, ją,
wóno
tobje
jeho, jeje
njim, nji
tobą
jemu, ji
njim, nją
na-ju
na-ma
va-ma
2.
va
va-ju
3.
ji-ch, nji-ch
ji-ma
njim-ji
1.
wón-ji,
wón-e
me
na-s
ną-m
na-mji
2.
ve
va-s
vą-m
va-mji
3.
wón-ji,
wóne
ji-ma
nji-mji
ji-ch
njich
As noted by Breza and Treder (1981:125) in Gramatyka Kaszubska, the 1st and 2nd
dual pronouns ma, wa, nama, wama were the original dual forms which dated back to the
12th century. As recorded in the Atlas Językowy Kaszubszczyzny i Dialektów Sąsednich
(AJK, 12th century), the 1st dual form ma meant ‘we two’ and the 2nd wa ‘we two.’ Over
time, the dual meaning of these forms was lost.
In the 19th century Kashubian, we still observe no dual/plural syncretism in the 1st
and 2nd person nominative since the dual pronouns are phonologically distinct from the
plural ones. The 1st person dual ma is distinct from the 1st person plural me, and the 2nd
person dual va is distinct from the 2nd person ve (Table 4.16).
160
Table 4.16 Dual and Plural Nominative Pronouns in Kashubian (1860s) (Duličenko
& Lehfeldt 1998:58-59)
Dual
Plural
1. ma
1. me
2. va
2. ve
3. wón-ji, wón-e
3. wón-ji, wón-e
As noted by Stone (1993:768), at the beginning of the 20th century the dual was still
used in Slovincian and some other eastern dialects of Kashubian. During the historical
development of Kashubian in the 20th century, the 2nd person dual va has disappeared,
and only the 1st person dual forms ma, naju, nama were still used (Table 4.17). However,
even the 1st person dual forms could not persist for a long time. As Stone (1993) further
reports, in the northern Kashubian dialects, the 1st person dual forms ma, naju, nama
were no longer used by the 1950s.
Table 4.17 Kashubian Personal Pronouns (Stone 1993:773)
Num/Person
SG
Nom
Acc
Gen
1.
jó
2.
të
DL
1.
ma
na-ju
PL
1.
më
na-s
2.
3.
wa
oni, onë
wa-ju
Loc
Dat
Instr
Mnie
cebi-e
mną
Tobie
tobą
na-ma
nó-m
na-mi
wa-ma
161
It appears that by the end of the 20th century, the dual was no longer a distinct
number category in the Kashubian personal pronouns. According to a relatively recent
dissertation by Hopkins (2001), the Kashubian personal pronouns distinguish only two
numbers: singular and plural (Table 4.18). 41 The 1st person plural pronoun is represented
as më in the Kashubian orthography, and is pronounced as /mə/. The 2nd person plural
pronoun written as wa is pronounced as /va/.
Table 4.18 Kashubian Personal Pronouns (Hopkins 2001:36)
Singular
Plural
1.
ja
më [mə]
2.
të
wa [va]
3.
won/wona/wono
woni/wone
The analysis of the diachronic changes in the Kashubian personal pronouns suggests
that the dual has become syncretic with the plural, and as a consequence of this syncretism
the dual/plural distinction was lost (Table 4.19). The 2nd person dual va /va/ became
syncretic with the 2nd person plural wa /va/ at some point in the development of
Kashubian in the 20th century. Likewise, the 1st person dual ma /mæ/ became syncretic
with the 1st person plural më /mə/. I suggest that the introduction of the phoneme /ə/ (the
so-called Kashubian schwa) resulted in the phonological change ma /ma/ > më /mə/ in
41
In Table 4.18, I provide the IPA transcription of the dual and plural to avoid confusion between
orthography and pronunciation.
162
the 1st person (Stone 1993:765).42 This phonological change allowed the dual to syncretize
with the plural in the 1st person.
Table 4.19 The Evolution of the Kashubian Dual Pronouns
Number/Person
19th Century
Kashubian
(Ceynowa 1860)
Kashubian in the
1950s
(Stone 1993)
21st Century
Kashubian
(Hopkins 2001)
DU 1.
ma
ma
-
2.
va
-
-
wón-ji, wón-e
me
më
më
2.
ve
wa
wa
3.
wón-ji, wón-e
oni, onë
woni, wone
3.
PL 1.
In contemporary Kashubian, the system of verbal agreement also shows the absence
of dual suffixes with the distinction made only in the singular and plural (Table 4.20). For
example, agreement with the 2nd person plural pronoun wa is marked by the suffix –ta
(96).
Table 4.20 Present Tense Verbal Suffixes in Kashubian (Hopkins 2001:36)
Number
SG
PL
42
1
-mə
Person
2
-š
-ta
3
-e
-
Topolińska (1974) and Stone (1993) note that the emergence of the new phoneme /ə/ in Kashubian was
due to shift from /i/ to /ə/ in the word initial position. According to Stone (1993), the new phoneme /ə/ was
first attested at the end of the 17th century.
163
(96)
Dzez wa jidze-ta?
where 2.PL go-2.PL.PRES
‘Where are you going?’
(Stone 1993:780)
In sum, diachronically, Kashubian has lost a dual number distinction in its personal
pronouns due to the dual/plural syncretism in the 1st and 2nd person. Based on historical
data summarized in (90), it is evident that the 2nd person dual pronoun va /va/ is the first
one to syncretize with the plural. This diachronic change is quite expected since in Old
Church Slavic the 2nd person dual vy was syncretic with the plural. Next, historical
change affected the 1st person dual ma which syncretized with the plural më due to
phonological changes. The 1st and 2nd person dual forms ma and va which were attested
by Ceynowa in the 1860s were ‘replaced’ by the plural forms më and wa by the middle of
the 20th century. As a result of this diachronic change, the dual is no longer a distinct
number category in contemporary Kashubian.
164
4.2.6. The Monomorphemic Dual and Its Loss in Old Russian
In Old Russian, dual number was a distinct grammatical category in the
pronominal and verbal agreement systems beginning from the 11th century when the
early Russian literary tradition started to get established and the first translations of the
Bible were made. However, already in the 11th century dual pronouns began to show the
first signs of decline as they were being replaced by the plural forms due to syncretism.
During the period from the 12th through the 15th centuries, the dual underwent historical
reanalysis as the plural. As a result of this reanalysis, by the end of the 15th century, the
dual was no longer a distinct grammatical category, and the number system of Russian
pronouns was reduced to two members of the opposition - singular and plural.
The grammatical category of dual number in Old Russian (11th-15th centuries),
follows its Proto-Slavic roots. In Proto-Slavic, personal pronouns distinguished three
numbers - singular, dual, and plural and three persons - 1st, 2nd, and 3rd (Table 4.21).
Already in PS, the 2nd person dual form vy was syncretic with the plural.43 The
dual/plural syncretism of the 2nd person points to the instability of the dual as a
grammatical category. The fact that the dual "was already losing ground in Proto-Slavic"
is confirmed by "having its range of cases reduced to three by the syncretization of
Nominative + Accusative, Genitive + Locative, and Dative +Instrumental" (Sussex and
Cubberley 2006:41).44
43
The 2nd person dual form va postulated in Proto-Slavic by Sussex and Cubberley (2006) is not syncretic
with the 2nd person plural form vy.
44
I assume that Sussex & Cubberley (2006) suggest nominative/accusative case syncretism only for the
2nd person since the pronominal stems vě and na are phonologically different in the 1st person.
165
Table 4.21 Proto-Slavic Personal Pronouns (Sussex and Cubberley 2006)
Num/Person
SG
DL
PL
Nom
Acc
Gen
Loc
Dat
Instr
1.
azŭ
men-e
mĭn-ě
mǔn-ojǫ
2.
ty
teb-e
teb-ě
tob-ojǫ
3.
jĭ, ja, je
-jĭ, ja, jǫ
1.
vě
na
jego,
jemĭ, jeji
jejě/jejǫ
na-ju
2.
va/vy
va
va-ju
va-ma
3.
ja, ji
ja, ji
ji-ju
ji-ma
1.
my
na-sŭ
na-mŭ
na-mi
2.
vy
va-sŭ
va-mŭ
va-mi
3.
ji, jě/ję, ja
ji-mŭ
ji-mi
jě/ję
ji-xǔ
jemu, jeji,
jmŭ, jejǫ
na-ma
I start my analysis with the Old Russian dual pronouns which occurred in the
Ostromir Gospel (1056-1057), the earliest Old Russian literary document. The Old
Russian personal pronouns found in the Ostromir Gospel were very similar to their ProtoSlavic counterparts in terms of person, number, and case distinctions. The Old Russian
pronouns showed a three-way number system - singular, dual, and plural and a two-way
person system - 1st and 2nd (Table 4.22). As opposed to Proto-Slavic, the Old Russian
3rd person pronouns were not part of the system of personal pronouns. They developed
later from the demonstrative pronouns onŭ, ona, ono and the so-called 'non-personal'
pronouns jĭ, ja, je (Možejko and Ignatenko 1988:129).
166
Table 4.22 Old Russian Personal Pronouns in the Ostromir Gospel (1056-1057)
(Corpus Manuscript www.manuscripts.ru)
Num/Person
Nom
SG
1.
azŭ
men-e
mŭn-ě
mŭno-jǫ
2.
ty
teb-e
teb-ě
tob-jǫ
1.
vě
ny
na-ju
na-ma
2.
vy
va
va-ju
va-ma
1.
my
na-sŭ
na-mŭ
na-mi
2.
vy
va-sŭ
va-mŭ
va-mi
DL
PL
Acc
Gen
Loc
Dat
Instr
The Old Russian dual pronouns vary in their morphological structure. In the dual,
the nominative and accusative forms are monomorphemic whereas the genitive/locative
and the dative/intrumental forms are bimorphemic (Table 4.23). Following Šaxmatov
(1957:152), I assume that in the 1st and 2nd person dual, the entire forms vě, vy, ny, va
are stems. I do not assume that v- and n- should be treated as stems, and -ě, -y, -a as
suffixes. In the 1st person, the monomorphemic nominative and accusative dual forms
consist of two phonologically different pronominal stems: the stem vě in the nominative
and the stem ny in the accusative. In the 2nd person, the pronominal stems vy and va
show some similarity, but are still phonologically different. The bimorphemic
genitive/locative and dative/instrumental forms consist of pronominal stems na- or vaand case suffixes -ju or -ma.
167
Table 4.23 The Morphological Structure of the Old Russian Dual Pronouns
Monomorphemic Dual
Bimorphemic Dual
Nom, Acc 1.vě, ny
Gen/Loc -ju
2. vy, va
Dat/Instr -ma
I found that the 2nd person dual form vy was the only form occurring in the
Ostromir Gospel (1056-1057). According to my findings, there were no instances of the
2nd dual form va proposed by other scholars (Samsonov 1973, Možejko and Ignatenko
1988). Due to the phonological identity of the dual and plural expressed by the form vy
dual/plural syncretism was possible in the 2nd person nominative. Žolobov (2001)
independently arrived at the same finding that the form vy was the only possible 2nd
person nominative dual form which occurred in Old Russian and could be traced to its
Proto-Slavic form. He states that the 2nd person nominative dual form vy was the only
form which was used in Old Russian ("в древнерусском употреблении в ИП дв.ч.
известна только форма вы" (Žolobov 2001:76)).
I argue that the Old Russian dual was morphosyntactically unstable due to its
syncretism with the plural in the 2nd person. Since the times of the Proto-Slavic (20001500 BC), number syncretism had occurred between the dual and plural in the 2nd person
nominative (Table 4.21). The same pattern of dual/plural syncretism could be seen in Old
Russian in the 11th century. The 2nd person form vy was syncretic in the dual and plural
as evidenced by the data from the Ostromir Gospel (1056-1057) (97-98). The 2nd
person pronoun vy refers to two individuals since the verb is marked by the dual suffix -ta
168
(97). The same pronoun vy has a plural reference since the verb is marked by the plural
suffix -te (98).
(97)
Old Russian (11th century)
ne boi-ta
vy
sja
NEG fear-2.DU.IMP 2.PL.NOM REFL
‘Don’t be afraid you two.’
(Ostromir Gospel, Math.28.5, 203:5)
(98)
Old Russian (11th century)
vy
tvori-te
dĕl-o
ots-a
vaš-ego
2.PL.NOM do-2.PL.PRES thing-ACC father-GEN your-GEN
'You (more than two) do things that your father does.'
(Ostromir Gospel, John 8.41)
Besides the Ostromir Gospel (1056-1057), the data from the other well-known
Old Russian manuscripts, such as the Izbornik (1076) and the Archangel Gospel (1092)
show that the dual was syncretic with the plural in the 2nd person. Dual reference of the
2nd person nominative dual vy is confirmed by dual agreement suffixes marked on verbs.
The 2nd person nominative dual vy occurs with the verbs marked by the dual suffix -ta,
such as tvori-ta, spse-ta, živĕ-ta , and gle-ta (99-101). The above data show that despite
the dual/plural syncretism, the dual still remained a distinct grammatical category in the
pronominal system in the 11th century.
(99)
Old Russian (11th century)
i
vy
tvori-ta
da
spse-ta
sę
and 2.PL.NOM do-2.DU.PRES COMP save-2.DU.PRES REFL
'And you two do so that you will save yourselves.'
169
(Izbornik, 112.2)
(100)
Old Russian (11th century)
tako i
vy
živĕ-ta
čad-ĕ moi
so
and 2.PL.NOM live-2.DU.PRES child-3.DU my
'And so you two of my children will live.'
(Izbornik, 112.2)
(101)
Old Russian (11th century)
jego
že
vy
gle-ta
jako slĕpǔ rodi-sę
3.SG.ACC EMPH 2.PL.NOM say-2.DU.PRES as
blind born-REFL
‘You (two) say that he was born blind.’
(Archangel Gospel, 9.1)
The 2nd person dual vy which co-occurred with the dual suffix -ta continued to
appear in the Old Russian manuscripts in the 12th and in the early 13th centuries. I argue
that the first evidence of its changed status as a plural pronoun is found in the Simonov
Gospel (1270) where the pronoun vy occurs with the plural suffix -te (102). The plural
suffix -te points to the fact that the pronoun vy has a plural reference set rather than the
dual as it was seen in the 11th century manuscripts.
(102)
Old Russian (13th century)
i že
vy
gl-te
jako slĕpŭ
rodi-sę
that-ACC EMPH 2.PL. NOM say-2.PL.PRES as
blind
born-REFL
‘You (two) say that he was born blind.’
(Simonov Gospel, 17.1)
In the 14th century, there was still some variation in the agreement suffixes which
encoded the 2nd person dual. For example, in the Laurential Codex (1377) of the Povest'
170
Vremennyx Let, the syncretic 2nd person dual/plural vy appears with the dual suffix -ta
(103-104).45
(103)
Old Russian (14th century)
i
reč-e
wlegǔ
askoldu
and say-1.SG.PRES Oleg
Askold
'And Oleg says to Askold and Dir."
i dirovi
and Dir
-
vy
nĕs-ta
knęz-ę
1.PL.NOM neg.be-2.DU.PRES prince-du
'You two are not princes.'
(Laurential Codex, 8.1)
(104)
Old Russian (14th century)
i
reč-e
stopolkǔ
posĕdi-ta
vy
sdĕ
and
say-1.SG.PRES Stopolk
sit-1.SG.PRES 1.PL.NOM here
'And Stopolk says, "You two sit here."
(Laurential Codex, 87.2)
In the same text of the Laurentian Codex (1377) are some examples where vy
occurs with the plural suffix -te when the reference is made to two individuals. The
pronoun vy appears with the verb ima-te marked with the plural suffix -te (106). It is
clear from the context of (105) that the plural pronoun vy refers to two people.
45
Povest' Vremennyx Let or the Russian Primary Chronicle is one of the major linguistic and historical
documents about the emergence of early Russia. This document has five versions written at different times:
1. Laurentian Codex (1377)
2. Hypatian Codex (ca. 1425)
3. Radziwiłł Codex(1490s)
4. Academy Codex (end of 15th century)
5. Khlebnikov Codex (16th century)
171
(105)
Old Russian (14th century)
i
rĕša
ima
muži
smyslenii
and told
3.DU.DAT men
smart
'And the smart men told the two of them:'
(106)
Old Russian (14th century)
počto vy
rasprę ima-te
meži
soboju
why
1.PL.NOM dispute have-1.PL.PRES
between selves
'Why do you have an argument between the two of you?'
(Laurential Codex, 73.1)
In the first half of the 15th century, there is a tendency towards the use of the 2nd
person plural vy with the plural suffix -te. However, there some cases where the pronoun
vy still occurs with the dual suffix -ta. A very interesting example comes from the
Ipatjevskja Letopis (1425) where in the same passage we come across the plural pronoun
vy occurring with both the dual suffix -ta and the plural suffix -te (107-108).
(107)
Old Russian (15th century)
-
ego
že
vy gle-te
antixrǔsta
3.SG EMPH 2.PL say-2.PL.PRES antichrist
'You two say that he is an antichrist.'
(Ipatjevskja Letopis, 65.2)
(108)
Old Russian (15th century)
jako že
vy gle-ta
so EMPH 2.PL say-2.DU.PRES
'So you two say ...'
172
(Ipatjevskja Letopis, 65.2)
The same passage from the Primary Russian Chronicle which appears in the
Ipatjevskja Letopis of 1425 and the Laurentian Codex of 1377 shows variation in the
agreement suffixes occurring with the plural pronoun vy. The pronoun vy is used with the
dual suffix -ta in the Ipatjevskja Letopis (1425) whereas in the Laurentian Codex (1377)
it is used with the plural suffix -te (109-110).
(109)
Old Russian (15th century)
počto vy
rasprju ima-ta
why
1.PL.NOM dispute have-1.DU.PRES
'Why are you two arguing?'
(Ipatjevskja Letopis, 80.2)
(110)
Old Russian (14th century)
počto vy
rasprę ima-te
meži
soboju
why
1.PL.NOM dispute have-1.PL.PRES
between selves
'Why do you have an argument between the two of you?'
(Laurentian Codex, 73.1)
By the end of the 15th century, we still find quite a number of instances where the
2nd person plural vy occurred with the dual suffix -ta. For example, in the Radziwiłł
Codex (ca. 1490) of the Primary Chronicle, in the context where two individuals were
referred two, the plural pronoun vy appeared with the dual agreement suffix -ta (111113). There was one instance where the plural pronoun vy co-occurred with the plural
suffix -te (114).
173
(111)
Old Russian (15th century)
i reč
wlegǔ kǔ askolodovi
and say.1.SG.PRES Oleg to Askold
'And Oleg says to Askold and Dir."
i dirovi
and Dir
vy
nĕs-ta
knz-ę
1.PL.NOM neg.be-2.DU.PRES prince-du
'You two are not princes.'
(Radziwiłł Codex 11.1)
(112)
Old Russian (15th century)
počto vy
ima-ta
rasprju meži
soboju
why
1.PL.NOM have-1.DU.PRES dispute between selves
'Why do you have an argument between the two of you?'
(Radziwiłł Codex 126.1)
(113)
Old Russian (15th century)
i
reč
stopolkǔ
sĕdi-ta
vy
zdĕ
and
say-1.SG.PRES Stopolk
sit-1.SG.PRES 1.PL.NOM here
'And Stopolk says, "You two sit here."
(Radziwiłł Codex 139.1)
(114)
Old Russian (15th century)
ego
že
vy gl-te
antixrǔsta
3.SG EMPH 2.PL say-2.PL.PRES antichrist
'You two say that he is antichrist.'
(Radziwiłł Codex 103.2)
174
The data from the Ipatjevskja Letopis (1425) and the Radziwiłł Codex (ca. 1490)
presented above show that the dual number distinction in the pronominal system of Old
Russian had begun to be gradually replaced with the plural number in the 15th century.
The instances of the plural vy co-occurring with the plural agreement suffix -te
demonstrate that the pronoun vy has a plural reference set.
In sum, during the 11-15th centuries, the 2nd person dual vy has undergone
considerable changes (Table 4.24). In the 11th-12 th centuries, the dual vy began to
syncretize with the plural in the 2nd person while retaining the dual agreement suffix -ta.
In the 13th century, the first instance of the plural form vy co-occurring with the plural
suffix -te appeared in the Simonov Gospel (1270). In the 14-15th centuries, the plural
agreement suffix -te appears with the plural form vy indicating plural reference. I assume
that the process of replacement of the dual suffix -ta with the plural suffix -te continued
into the 16th century (Šaxmatov 1957).
Table 4.24 The Evolution of the 2nd Person Dual Pronoun vy in Old Russian
Time Period
Agreement Suffix
Reference
11-12th centuries
2nd Person
Pronoun
plural vy
dual -ta
2
13th century
plural vy
14-15th centuries
plural vy
dual -ta
plural -te
dual -ta
plural -te
2
>2
2
>2
In contrast to the 2nd person nominative vy, the 1st person nominative vě was not
subject to syncretism as early as in the 11th century. The dual pronoun vě did not
syncretize with the plural my in the 1st person due to the phonological difference in the
175
pronominal stems. The 1st person dual vě occurred in a number of the 11the century Old
Russian manuscripts including the Ostromir Gospel (1056-1057) and the Archangel
Gospel (1092). The dual reference of the 1st person dual vě was indicated by the
agreement suffix -vě marked on the verb (115-116). Thus, in the 11th century the 1st
person dual was systematically used with the dual verbal agreement suffixes.
(115)
Old Russian (11th century)
vě-vě
vě
jako sĭ
jestĭ
synǔ najǫ
know-1.DU.PRES 1.DU.NOM
as
this is
son
our
'We two know that he is our son.'
(Ostromir Gospel, John 9.20)
(116)
Old Russian (11th century)
da
COMP
vě
ubo
1.DU.NOM as
vǔ pravĭdu
in truth
dostoina
bo po dělomǔ
naju
prijemlje-vě
worth
for on deeds
1.DU.GEN receive-1.DU.PRES
'We are punished justly, for we are getting what our deeds deserve.'
(Archangel Gospel, 110.2, Luke 23.41)
According to Žolobov (2001), in the 13th century, we still find the 1st person dual
pronoun vě occurring in some of the Old Russian manuscripts. For example, in the
Tolkovyj Apostol (1220), the 1st person dual vě occurs with the expected dual agreement
suffix -vě (117-118).46
46
I assume that the dual suffix -vě is an instance of long-distance agreement.
176
(117)
Old Russian (13th century)
da
vě vǔ jazyky
COMP 1.DU in
pegans
'For us two to go to the pegans.'
(Žolobov 2001:149, Tolkovyj Apostol, Gal.9.9)
(118)
Old Russian (13th century)
t ǔkmo
ništaja
da
pomni-vě
only
poor
COMP remember-1.DU.PRES
'So that we two don't forget the poor.'
(Žolobov 2001:149, Tolkovyj Apostol, Gal. 9.10)
The first instance of the 1st person plural pronoun my instead of the 1st person
dual vě occurred in the middle of the 14th century as evidenced by the data from the
Čudov Apostl (1355), an Old Russian translation of the New Testament (Žolobov
2001:149). In the Čudov Apostl (1355), the 1st person dual vě was replaced by the 1st
person plural my when the reference was made to two individuals (119).47 The plural
suffix -mǔ which occurs in the next sentence in the manuscript indicates reference to the
plural (120).
(119)
Old Russian (14th century)
da
47
my vǔ jazyky
The reference to two individuals in (119-120) can be understood from the context in which the pronoun
my appears.
177
1.PL in pegans
'For us two to go to the pegans.'
COMP
(Žolobov 2001:149, Čudov Apostl, Gal. 9.9)
(120)
Old Russian (13th century)
( )
tokmo
ništix
da
pominaje-mǔ
only
poor
COMP remember-1.PL.PRES
'So that we don't forget the poor.'
(Žolobov 2001:149, Čudov Apostl, Gal. 9.9)
In the late 14th century, we still find evidence of the 1st person dual vě
appearing with the dual agreement suffix -vě in the Laurentian Codex (1377) of the
Primary Russian Chronicle. The 1st person dual vě occurs with the dual suffix -vě (121122).
(121)
Old Russian (14th century)
vě
posle-vě
k bratu
svojemu
1.DU.NOM send-1.DU.PRES to brother own
'We two will send to our brother.'
(Laurentian Codex, 58.2)
(122)
Old Russian (14th century)
vě
es-vě
angl-a
1.DU.NOM be-1.DU.PRES angel-3.DU
'We two are angels.'
(Laurentian Codex, 65.1)
178
Let us contrast the above sentences (121-122) from the Laurentian Codex (1377)
with the same sentences in the Radziwiłł Codex (ca. 1490). The replacement of the 1st
person dual pronoun vě with the 1st person pronoun my was gradual. First, we find that in
the Radziwiłł Codex (ca. 1490) the 1st plural my occurred instead of the 1st person dual
vě even when the dual suffix -vě remained unchanged (123). Next, the 1st person plural
my appeared with the plural agreement suffix -i on the DP (124). Finally, we find
examples where the 1st person plural my occurred with the plural suffix -mǔ (125).
(123)
Old Russian (15th century)
my
posle-vě
ko bratu svojemu
1.PL.NOM send-1.DU.PRES to brother own
'We two will send to our brother.'
(Radziwiłł Codex, 101.1)
(124)
Old Russian (15th century)
my
anggl-i
1.PL.NOM angel-3.PL
'We are angels.'
(Radziwiłł Codex, 111.2)
(125)
Old Russian (15th century)
my vǔdae-mǔ
1.PL know-1.PL.PRES
'We two know.'
(Radziwiłł Codex 103.2)
179
In sum, the 1st person dual pronoun vě has undergone considerable diachronic
changes in the period from the 11th through the 15th centuries (Table 4.25). As opposed
to the 2nd person dual vy which syncretized with the plural already in the 11th century,
the 1st person dual vě was not syncretic with its plural counterpart my due to the
phonological difference in the stems. The 1st person dual vě was consistently used with
the dual agreement suffix -vě in the 11-13th centuries. In the 14-15th centuries, the 1st
person plural my gradually replaced the 1st person dual vě. As a result of these diachronic
changes, by the end of the 15th century, the 1st person plural my and the plural suffix -mǔ
were used even when the reference was made to two individuals.
Table 4.25 The Evolution of the 1st Person Dual Pronoun vě in Old Russian
Time Period
11-13th centuries
1st Person Pronoun
dual vě
Agreement Suffix
dual -vě
Reference
2
14-15th century
dual vě
plural my
plural my
dual -vě
dual -vě
plural -mǔ
2
2
>2
180
4.3. Summary
In this section, I have explored two main questions:(i) what are the number
features which represent the Slavic dual in Narrow Syntax and (ii) what are the patterns
of cross-linguistic variation in the realization of the Slavic dual at Phonological
Structure?
I have argued that in Narrow Syntax the pronominal Slavic dual is represented by
the two features - [-singular -augmented]. I have shown that Contemporary Standard
Slovenian, Upper, and Lower Sorbian have pronominal duals which are composed of the
plural stem and the numeral two or the suffix –j derived from this numeral. I have further
argued that the number features [-singular] and [-augmented], which make up the dual,
are distributed in a predictable way. The feature [-singular] is encoded by the plural
pronominal stem while the feature [-augmented] corresponds to the numeral two or the
dual suffix –j.
I have identified two patterns of cross-linguistic variation in the PF representation
of the Slavic dual- bimorphemic and monomorphemic. The bimorphemic pronouns occur
in the three contemporary Slavic languages - Slovenian, Upper, and Lower Sorbian. The
bimorphemic structure of the dual pronouns in these languages have emerged
diachronically as a result of the dual/plural syncretism in the pronominal number system.
The monomorphemic pronouns used to be a part of the singular/dual/plural pronominal
number system in Old Russian and Kashubian. However, over time these
monomorphemic duals were replaced by the plural pronouns due to syncretism,
181
especially in the 2nd person. As a result of this diachronic change, contemporary Russian
and Kashubian pronouns distinguish only between the singular and plural.
The next most important question which needs to be answered is why the two
different patterns in the PF realization of the Slavic dual emerged, and what principle or
mechanism is responsible for their emergence. I provide my arguments regarding this
question in the next section of the dissertation.
182
CHAPTER 5 THE PRINCIPLE OF MORPHOSYNTACTIC FEATURE
ECONOMY AND THE DIACHRONIC CHANGES IN THE SLAVIC DUAL
5.0. Introduction
In this final chapter of the dissertation, I present my proposal of the principle of
Morphosyntactic Feature Economy and argue that it provides a principled explanation for
the two different patterns of diachronic change which occurred in the Slavic dual
pronouns. In the first pattern of diachronic change, the dual in Old Slovenian and Old
Sorbian was reanalyzed as a bimorphemic structure and continued to be used by
contemporary speakers of both languages. In the second pattern, the dual in Old Russian
and Kashubian was 'lost,' i.e. it was reanalyzed as the plural.
In Section 5.1., I present my assumptions about markedness of the
[-singular -augmented] feature combination of the dual. Following Nevins (2008, 2011), I
adopt a feature value markedness approach according to which the
[-singular -augmented] feature combination of the dual is subject to both context-free and
context-sensitive markedness statements.
In Section 5.2., I propose and motivate the principle of Morphosyctactic Feature
Economy. Assuming that the Faculty of Language is designed for an optimal
computation (Chomsky 2008a,b), I propose that the principle of Morphosyntactic Feature
Economy is an interface condition through which the Faculty of Language can maximize
its computational efficiency. In Section 5.3., I outline my assumptions about diachronic
change within the framework of Minimalism and Distributed Morphology. In Sections
5.4.-5.7., I present the analyses of how the principle of Morphosyntactic Feature
183
Economy accounts for the diachronic changes of the dual in Slovenian, Sorbian,
Kashubian, and Russian. In Section 5.8., I summarize the chapter.
184
5.1.
Morphosyntactic Markedness of Feature Values of the Dual
The notion of morphological markedness is often difficult to define and can be
approached differently in morphological theory.48 Broadly defined, morphological
markedness is "the asymmetric treatment of two categories within an opposition where
equal patterning might otherwise be expected" (Nevins 2011:417).49 In current
morphological theory, we can indentify three types of morphological markedness:
i). category internal markedness (Greenberg 1966), feature markedness (Harley & Ritter
2002), and feature value markedness (Nevins 2008, 2011).
Category-internal markedness proposed by Greenberg (1963) has to do with cooccurrence restrictions within a certain morphological category. According to this
approach to markedness, a marked category implies the presence of an unmarked one. In
his famous Universal 34, Greenberg suggests that "no language will have a dual number
unless it also has a plural number" (1963).
Feature markedness (Harley and Ritter 2002) is expressed via a node-counting
metric which measures terminal nodes or dependency relations in a feature-geometric
representation of a grammatical category. According to this approach, a feature
combination of a category is marked if it has more nodes/dependency relations in its
feature-geometric representation. For example, the dual is more marked than the singular
48
Haspelmath (2006:26) identifies as many as 12 senses of markedness. None out of these 12 senses are
assumed here.
49
The concept of markedness of grammatical categories goes back to Jakobson who pointed out that "the
asymmetry of correlative grammatical form can be characterized as the antinomy of signalization of A and
non-signalization of A" (1984:12). Jakobsonian notion of markedness has to do with specification vs. nonspecification of a semantic distinction in a grammatical category.
185
and plural since the feature-geometric representation of the dual contains a greater
number of terminal nodes.
Feature value markedness proposed by Nevins (2008, 2011) is defined as an
asymmetrical treatment of one value of a binary morphosyntactic feature. One value of a
binary feature can be marked in regards to the morphosyntactic context in which it
occurs. Morphosyntactic markedness of feature values entails that a feature with a
marked value can be either the target or trigger for neutralization. In the case of the
dual/plural syncretism in Slavic, the marked [-augmented] feature (whose feature value is
marked) is the target for deletion.50
Following Nevins (2008, 2011), I adopt feature value markedness to determine
markedness status of the morphosyntactic representation of the dual. According to a
feature value markedness approach, the values of the [-singular,-augmented] feature
combination of the dual are subject to context-free and context-sensitive markedness
statements. The result of the application of these markedness statements is that in the dual
the co-occurrence of a marked [-singular] feature with a marked [-augmented] feature
makes this combination morphosyntactically marked (126b).
(126) Number Feature Combinations of the Singular, Dual, and Plural (Noyer
1997)
50
a. Singular
[+singular -augmented]
b. Dual
[-singular -augmented]
The pervasive case, person, and gender syncretisms show that in Slavic languages, the dual is also the
trigger for deletion of case, person, and gender features.
186
c. Plural
[-singular +augmented]
Let us consider markedness of values of the [±singular] and [±augmented]
features in more detail. I assume (Nevins 2008, 2011) that the [±singular] feature is
subject to a context-free markedness statement (127). Since the [+singular] feature occurs
only in one context, in the singular, it is unmarked. The [-singular] feature is marked
regardless of the context in which it occurs - in the dual (128b) or in the plural (128c).
(127) Context-Free Markedness Statement (Nevins 2011)
The marked value of the [±singular] feature is -.
(128) Morphosyntactic Markedness of Values of the Number Features
Number
+/-Singular
+/-Augmented
a. Singular
+sg = unmarked
-aug = marked
b. Dual
- sg = marked
-aug = marked
c. Plural
-sg = marked
+aug = unmarked
Neither value of the [±augmented] feature is context-free marked. Markedness of
the [±augmented] feature needs to be relativized according to the context in which it
occurs. The [±augmented] feature occurs in the context the dual and the plural. Both the
dual and the plural possess a marked [-singular] feature. Therefore, the [±augmented]
feature is subject to a context-sensitive markedness statement (128). According to the
context-sensitive markedness statement (129), when the [±augmented] feature occurs in
187
the context of the marked [-singular] in the dual its feature value is marked. Thus, in the
dual the values of the [-singular] and [-augmented] features are marked (130).
(129) Context-Sensitive Markedness Statement (Nevins 2011)
In the context of [-singular] feature, the marked value of the
[±augmented] feature is - .
(130) Markedness of the Dual [-singular -augmented] Feature Bundle
Dual [-singular MARKED -augmented MARKED]
In this section, I have presented my theoretical assumptions about feature value
markedness and the status of the morphosyntactic representation of the dual in this theory
of morphological markedness. Assuming that the [-singular,-augmented] feature
combination of the dual is marked in Narrow Syntax, I claim that its markedness needs to
resolved post-syntactically at Morphological Structure. In the next section, I argue that
two different patterns of diachronic change in the Slavic dual which result from its
morphosyntactic markedness are resolved via the application of the principle of
Morphosyntactic Feature Economy at Morphological Structure.
188
5.2.
The Proposal: The Principle of Morphosyntactic Feature Economy
The main claim of this dissertation is that the principle of Morphosyntactic
Feature Economy is a driving force behind two different patterns of diachronic change
which occurred in the Slavic dual (131). In the first pattern of diachronic change, the dual
in Old Slovenian and Old Sorbian was not 'lost,' but reanalyzed and continued to be used
by the speakers due to the historical change in (129a). In the second pattern, the dual in
Old Russian and Old Kashubian was 'lost,' i.e. it was reanalyzed as the plural (129b).
(131) The Two Patterns of Diachronic Changes in the Slavic Dual
a). [-singular -augmented] → [-singular] [-augmented] = 'reanalyzed dual'
b). [-singular -augmented] →[-singular] = 'plural'
I claim that the principle of Morphosyntactic Feature Economy operates on a
morphosyntactic representation of the Slavic dual which is marked and computationally
inefficient. Assuming that the Faculty of Language (part of our cognitive system) is
designed for an optimal computation (Chomsky 2008a,b), I propose that the principle of
Morphosyntactic Feature Economy is an interface condition through which the Faculty of
Language can maximize its computational efficiency. The principle of Morphological
Feature Economy is formulated as follows: a marked [-singular -augmented] feature
combination of the dual cannot be realized at Phonological Form without eliminating
markedness of its features at Morphological Structure (132).51
51
In the diagram (132), F stands for a morphosyntactic feature.
189
(132) The Principle of Morphological Feature Economy
Morphosyntactic Representation of the Dual at MS
MS
Dual [-singular -augmented]
[F marked, F marked]
Non-economical
Morphosyntactic Repairs at MS
Fission
Impoverishment
(Old Slovenian & Sorbian)
(Old Russian & Kashubian)
[-singular] [-augmented]
[-singular]
More economical
More economical
Phonological Form
As schematized in the diagram (132), the principle of Morphosyntactic Feature
Economy targets a marked [-singular -augmented] feature bundle of the Slavic dual at
Morphological Structure before morphosyntactic features are filled with phonological
content through Vocabulary Insertion at Phonological Structure. There are two options to
make the morphosyntactic representation of the dual less marked and therefore more
computationally efficient. One option is to split a marked feature bundle of the dual into
two separate terminal nodes via Fission, which happened in Old Slovenian and Sorbian.
190
The other option is to delete the [-augmented] feature via a morphological repair
operation of Impoverishment, which occurred in Old Russian and Kashubian.
According to the first option, a marked morphosyntactic representation of the dual
in Old Slovenian and Old Sorbian was split by Fission into two separate terminal nodes [-singular] and [-augmented], which became less marked and more computationally
efficient. As a result of this morphosyntactic repair, a reanalyzed dual continues to be part
of the pronominal and verbal agreement systems in Contemporary Standard Slovenian,
Contemporary Upper Sorbian, and Contemporary Lower Sorbian. The important point
here is that the morphosyntactic structure of the dual in Old Slovenian and Old Sorbian
was restructured into two separate terminal nodes by the application of the principle of
Morphosyntactic Feature Economy.
According to the second option, a marked morphosyntactic representation of the
dual in Old Russian and Old Kashubian was repaired via Impoverishment. The operation
of Impoverishment deleted the [-augmented] feature from the [-singular-augmented]
feature bundle of the dual which resulted in the syncretism of the dual with the plural by
means of sharing the [-singular] feature. The dual/plural syncretism resulted in the
replacement of the dual with the plural, which explains the absence of the dual as a
grammatical category of number in Contemporary Standard Russian and Contemporary
Kashubian.
I argue that the principle of Morphosyntactic Feature Economy which applies to
the morphosyntactic representation of the Slavic dual at Morphological Structure is
necessary to resolve a morphosyntactically marked and therefore, computationally
191
inefficient [-singular -augmented] feature combination of the Slavic dual. The [-singular augmented] feature combination puts an additional computational burden on the
Language Faculty which is biologically designed to be an optimal solution to the sensorymotor and conceptual-intensional interface conditions (Chomsky 2008a:135).52
Chomsky claims that "language is something like a snowflake, assuming its
particular form by virtue of laws of nature - in this case principles of computational
efficiency" (2008b:8). In other words, the Language Faculty is regarded as
computationally perfect, and a marked feature combination of the dual is an
'imperfection' which stands in the way of computational efficiency. To solve this
problem, the principle of Morphosyntactic Feature Economy applies via the
morphological repairs at Morphological Structure to make the morphosyntactic
representation of the dual less marked and hence more computationally efficient.
The principle of Morphosyntactic Feature Economy, which I proposed to account
for the two different patterns of diachronic change in the Slavic dual, makes two
important predictions about diachronic change of the dual cross-linguistically. First, it
predicts that if a language has a singular ~ dual ~ plural number system, the dual will
have a bimorphemic morphological structure. Second, if a language 'lost' a dual number,
the dual number will be replaced by the plural. The important consequence of the second
prediction is that monomorphemic duals should not occur cross-linguistically. Extensive
cross-linguistic research is needed to test if the two predictions of the principle of
Morphosyntactic Feature Economy are born out.
52
This statement is known as the Strong Minimalist Thesis (SMT) in Chomsky's Minimalist Program
(Chomsky 2008a).
192
5.3.
The Principle of Morphosyntactic Feature Economy and
Diachronic Change
I briefly outline my theoretical assumptions about the principle of
Morphosyntactic Feature Economy, Economy Principles which govern language, and
diachronic change. Following Chomsky (2005, 2008 a, b, c), I speculate that the principle
of Morphosyntactic Feature Economy falls within general Economy Principles not
specific to the Language Faculty (FL); namely, within "the third factor" principles of
efficient computation.
Chomsky (2005, 2008a) suggests that there are "three factors in language design:
(1) genetic endowment, which sets limits on the attainable languages, thereby making
language acquisition possible; (2) external data, converted to the experience that selects
one or another language within a narrow range; (3) principles not specific to FL" (2008
a:3). It is plausible to believe that the principle of Morphosyntactic Feature Economy
belongs to the last category of cognitive principles which reduce computational burden
in a syntactic derivation.
According to Chomsky (2005), the 'third factor' principles of language design
refer to language-independent principles of structural architecture and computational
efficiency. Chomsky (2005, 2008a) argues that principles of computational efficiency
apply to all living organisms including language which is a biological system. Principles
of computational efficiency imply that all biological systems are structured and organized
in the most efficient way.
193
Language, as a biological system, must optimally satisfy conditions which are
imposed by two interfaces - sensorimotor (SM) and semantic/conceptual-intentional
(C-I). Chomsky (2008a,b,c) further argues that the two interfaces are asymmetric in their
nature and the roles they play in computational efficiency. The conceptual-intentional
interface, i.e. the language of thought, is more important for efficient computation than
the sensorimotor interface. The sensorimotor interface is not crucial for efficient
computation since most of phonological systems of various languages "violate principles
of computational efficiency, while doing the best they can to satisfy the problem they
face: to map to the SM interface syntactic objects generated by computations that are
"well designed" to satisfy C-I conditions" (Chomsky 2008c:136). The key observation
about the C-I interface is that it is designed for opmitization of computational efficiency
of mapping of syntactic objects to C-I.
The Principle of Morphosyntactic Feature Economy which I propose belongs to
the "third factor" principles of efficient computation in that it simplifies the
morphosyntactic representation of the dual at Morphological Structure making this
representation more computationally efficient. At any given stage of language evolution,
including diachronic change in the Slavic dual, principles of efficient computation apply
to optimize linguistic computation. I argue that the Principle of Morphosyntactic Feature
Economy, as a principle of efficient computation, accounts for two different patterns of
diachronic change in the Slavic dual: 1). its reanalysis in Old Slovenian and Sorbian, and
2). its loss in Old Russian and Kashubian.
194
Before proceeding to explain how the Principle of Morphosyntactic Feature
accounts for diachronic changes in the Slavic dual, it is important to discuss how
diachronic change is treated in the most recent version Chomsky's (2008b) Biolinguistic
Program. Chomsky (2008b) differentiates two types of change: evolutionary (genomic)
change and historical change. Evolutionary change involved some sort of genetic
mutation which yielded the operation Merge. Historical change manifests itself only in
the "modes of externalization" - morphology and phonology of a language. However, it is
not suggested in Chomsky's (2008b) Biolinguistic Program how diachronic change,
which is only expressable in morphology and phonology, comes about and what
principles or mechanisms are responsible for it.
Within the Minimalist approach to syntactic (not morphological) diachronic
change, it has been extensively argued by van Gelderen (2008, 2009, 2011) that "the third
factor" principles, such as Feature Economy is responsible for various stages in language
change. Analyzing linguistic cycles, van Gelderen argues that Feature Economy
minimizes content of morphosyntactic features in order to maximize computational
efficiency of the syntactic derivation. In van Gelderen's (2011) formulation of Feature
Economy, semantic features (adjuncts) are reanalyzed by the language learner as
interpretable features (specifiers) which can value uninterpretable features (heads).
Uninterpretable features (affixes) act as probes in a syntactic derivation; they are most
economical and computationally efficient (133).53
53
In her most recent work The Linguistic Cycle: Language Change and the Language Faculty (2011), van
Gelderen analyzes the reasons for linguistic cycles. She argues that in a linguistic cycle shown in (133)
uninterpretable features are again 'renewed' by semantic ones.
195
(133) Feature Economy (van Gelderen 2011:17)
Minimize the semantic and interpretable features in the derivation:
Adjunct
semantic
Specifier
> [iF]
>
Head
[uF]
Affix
>
[uF]
Within the framework of Distributed Morphology, the principles of Economy as a
driving force of diachronic change have not received not very much attention in the
literature. Noyer (1997) suggests that diachronic change in language can be accounted for
by gain or loss of filters, which are formulated as co-occurrence restrictions on
morphosyntactic features. For example, Classical Arabic has a filter *[1dual] which is
assumed by a child acquiring this language. The filter *[1 dual] will impoverish the
morphosyntactic representation ‘1st dual’ agreement node by neutralizing it with a ‘1st
plural’ agreement node. The problem with this approach is that such filters seem to be
stipulated and do not stem from any underlying principle governing language.
In his most recent work, Calabrese (2011) suggests that diachronic change in
morphology can be brought about by marking statements which are activated or
deactivated in a specific language. For example, an active marking statement *[+plural,
+dual] characterizes a marked morphosyntactic feature configuration of the dual as
'costly' in terms of general principles of Language Economy. This marking statement
triggers a morphological repair operation which replaces the dual with the plural.
Calabrese further suggests that the reasons for morphological marking statements might
196
be in "the way in which functional or cognitive considerations are expressed in
grammatical terms (through morphological features)" (2011:291).54
In sum, the principle of Morphosyntactic Feature Economy which I have
proposed in this dissertation gives an insight into how the Faculty of Language 'deals'
with marked morphosyntactic 'imperfections' in order to maximize its computational
efficiency. It provides a principled explanation for the two patterns of diachronic change
in the Slavic dual: continued use of the dual in Slovenian and Sorbian and 'loss' of the
dual in Russian and Kashubian.
54
For further details on functional or cognitive motivations behind morphological marking statements, see
Calabrese (2011).
197
5.4.
The Reanalysis of the Old Slovenian Dual as a More Economical
Dual in Contemporary Standard Slovenian
In this section, I show how the principle of Morphosyntactic Feature Economy
applied to the Old Slovenian dual to produce a new and reanalyzed form of the dual in
Contemporary Standard Slovenian. Approximately in the 16th century, the Old Slovenian
dual became very unstable as a grammatical category (Derganc 1988, 2003; Jakop 2008).
I argue this instability of the Old Slovenian dual was due to the morphosyntactic
markedness of its morphosyntactic representation.
At the end of the 16th century, Slovenian literary tradition began to take roots
with the translation of the Bible (1584) by Jurij Dalmatin. This event marked the formal
establishment of Standard Slovenian language as there was considerable variation in
many of the Slovenian dialects (Greenberg 2008). As the Dalmatian's (1584) translation
of the Bible shows, the 1st, 2nd, and 3rd person dual pronouns appeared to be syncretic
with their plural counterparts.
Whenever instances of the dual/plural my, vy, ona syncretism occur in the
Dalmatian Bible (1584), verbal suffixes indicate dual reference. For example, the 1st
person dual pronoun my, which appears in the plural form, triggers dual agreement
marked by the suffixes -va and -a (134). The 2nd person pronoun dual vy (syncretic with
the plural) occurs with the dual form of the verb 'be' and the dual suffix -a which indicate
dual reference (135). Likewise, the 3rd person pronoun ona induces dual agreement
marked by the dual form of the verb 'be' and the dual suffix -a (136).
198
(134) Old Slovenian (16th century)
My bo-va
letu mejʃtu konzha-l-a ...
1.PL be-1.DU.FUT this place destroy-PART-DU.MASC
'We two will destroy this place..."
(Jakop 2008:363; Dalmatian Bible 1584: 1 Ms 19.13)
(135) Old Slovenian (16th century)
Vy, ʃta
me v’ neʃrezho pèrpravi-l-a ...
2.PL be.2.DU.PAST me in accident get-PART- DU.MASC
'You two got me into trouble ..."
(Jakop 2008:363; Dalmatian Bible 1584: 1 Ms 34.30)
(136) Old Slovenian (16th century)
Inu ona
ʃta
bi-l-a
oba-dva nagá
and 3.PL.MASC be.3.DU.PAST be- PART-DU. MASC both-two naked
'And they two were both naked.'
(Jakop 2008:367; Dalmatian Bible 1584: 1 Ms 2.25)
As I have argued in Section 5.1., the morphosyntactic representation of the dual is
more marked than that of the plural. Recall that in the morphosyntactic representation of
the dual [-singular -augmented], the value of the [-augmented] feature is marked in the
context of a marked [-singular] feature. In contrast, in the feature representation of the
plural [-singular +augmented], the value of the [+augmented] feature is unmarked. Thus,
the [-singular -augmented] feature combination of the dual is morphosyntactically
marked and its markedness needs to be eliminated.
I propose that the solution to the problem of reducing markedness of the Old
Slovenian dual lies in the application of the principle of Morphosyntactic Feature
Economy. At Morphological Structure, Old Slovenian dual pronouns my, vy, oni were
199
represented by the combination of the [-singular -augmented] features whose feature
values were marked (137).
(137) The Marked Structure of the Old Slovenian Dual Before Fission
D
D
Num
[-sg -aug]
The principle of Morphosyntactic Feature Economy applied to the marked feature
combination of the Old Slovenian dual to eliminate a marked feature bundle of the dual. I
propose that a Fission rule split the marked feature combination of the Old Slovenian
dual into two terminal nodes – [-singular] and [-augmented] (138).
(138) Fission Rule for the Old Slovenian Dual (16th century):
[-singular -augmented] → [-singular] [-augmented]
As a result of Fission, two separate terminal nodes were created in the
morphosyntactic structure shown in (139). The Number node which hosted the marked
[-singular, -augmented] feature combination was split into Num1 with [-singular] feature
and Num2 with the [-augmented] feature. The resultant morphosyntactic structure is less
marked and therefore is more computationally efficient (139).
200
(139) A Reanalyzed Old Slovenian Dual After Fission
D
D
Num
Num1
[-sg]
Num2
[-aug]
After the rule of Fission (136) had split the morphosyntactic combination of the
Old Slovenian dual into two terminal nodes, Vocabulary Insertion applied to realize the
two positions of exponence with phonological content. Two Vocabulary Items were
needed to fill two positions of exponence created by Fission (140). The [-singular] feature
was filled by Vocabulary Items /mi/, /vi/, or /ona/ (the plural pronominal stems). The
[-augmented] feature was realized by the numeral dva (‘two’) as the only minimal nonsingular VI which contains no non-atomic proper subsets.
(140) VIs for the Old Slovenian Dual Pronouns
/mi/, /vi/, or /ona/ ↔ Num [-singular]
/dva/ ↔ Num [-augmented]
The result of application of the principle of Morphosyntactic Feature Economy in
Old Slovenian is a reanalyzed bimorphemic structure of the pronominal dual.
Interestingly, evidence of a reanalyzed morphosyntactic structure of the Old Slovenian
dual is found in the Dalmatian Bible (1584), the same manuscript where the 'old' dual
forms also occur. As the data (141-143) show, the reanalyzed dual pronouns my-dva, vydva, ona-dva consist of a plural pronominal stem my, vy, ona and the numeral dva ('two').
201
Importantly, the verbs in (141-143) are marked by the dual suffixes -va and -ta to show
agreement with the 1st, 2nd, and 3rd person reanalyzed dual pronouns.
(141) Old Slovenian (16th century)
Oʃtani-te tukaj, dokler my-dva
supet k’vam
stay-IMP here until 1.PL-two
to
2.PL.DAT
'Stay here until we two come back to you.'
pride-va
return-1.DU.PRES
(Jakop 2008:363; Dalmatian Bible 1584: 2 Mz 2.14)
(142) Old Slovenian (16th century)
Vy-dva
bo-ta
rejs ta
Kelih py-l-a
2. PL- two be-2.DU truly this chalice drink-PAST-MASC.DU
'You two will truly drink from this chalice."
(Jakop 2008:363; Dalmatian Bible, Mr 10.39)
(143) Old Slovenian (16th century)
kadar ʃta
ona-dva
bi-l-a
na Puli
when be.2.DU.PAST 3.PL.MASC-two be-PAST-DU.MASC on field
'When they two were in the field ...'
(Jakop 2008:364; Dalmatian Bible, 1 Mr 4.8)
Since the 16th century, the 1st, 2nd, and 3rd person reanalyzed Slovenian dual
pronouns midva, vidva, onadva have become part of the grammatical system in
Contemporary Standard Slovenian (Table 5.1). The major diachronic change which
occurred in the number system of Slovenian pronouns was the morphosyntactic
reanalysis of the Slovenian dual as a less marked and more computationally efficient dual
(144).
202
Table 5.1 Dual Pronouns in Contemporary Standard Slovenian
Num/Person
DL
1.
Nom
mi-dva, midve/me-dve
vi-dva, vidve/ve-dve
ona-dva, onidve/one-dve
2.
3.
Acc
Gen
Loc
Dat
na-ju
na-ma
va-ju
va-ma
nji-ju
nji-ma
Instr
(144) Morphosyntactic Reanalysis of the Slovenian Dual
'marked' [-singular,-augmented] → [-singular] [-augmented] 'less marked'
As I have argued in this section, the principle of Morphosyntactic Feature
Economy was the driving force behind the morphosyntactic reanalysis of the Slovenian
dual. A singular/dual/plural number system of Old Slovenian pronouns was reanalyzed at
Morphological Structure via Fission which created a less marked and more economical
dual in Contemporary Standard Slovenian (144-145).
(145) Old Slovenian Pronoun Number System
Singular
[+singular -augmented]
Dual
[-singular -augmented]
Plural
[-singular +augmented]
(146) Contemporary Standard Slovenian Pronoun Number System
Singular
[+singular -augmented]
Dual
[-singular] [-augmented]
Plural
[-singular +augmented]
203
5.5.
The Reanalysis of the Old Sorbian Dual as a More Economical Dual
in Contemporary Upper and Lower Sorbian
In this section, I present my theoretical account which explains the reanalysis of
the Old Sorbian dual. I argue that the principle of Morphosyntactic Feature Economy
applied to a marked [-singular -augmented] morphosyntactic representation of the Old
Sorbian dual to reduce its markedness. The case of the Old Sorbian dual is very similar to
the Old Slovenian dual in that both languages employed the same morphosyntactic
strategy of Fission which resulted in less marked and more computationally efficient
representation of the dual.
In the 16th century, Old Sorbian dual pronouns became syncretic with their plural
counterparts as evidenced by the following data (146-147). The verbs in (146-147) are
marked by the dual suffixes -moj and -taj to indicate agreement with the subject in the 1st
and 2nd person and dual number. The important point here is that the verbal agreement
suffixes -moj and -tai reflect dual reference of the subject pronouns.
(147) Old Sorbian (16th century)
My chze-moj
1.PL want- 1.DU.PRES
‘We two want to …’
so,
schtoz …
REFL COMP
(Derganc 1998:52)
(148) Old Sorbian (16th century)
Wy widz-i-tai
a schlysch-i-tai
1.PL see-TH-2.DU.PRES and hear- TH-2.DU.PRES
‘You two see and listen.”
(Derganc 1998:50)
204
I claim that the dual/plural pronominal syncretism shown in (147-148) is a strong
indication of markedness of the [-singular-augmented] feature bundle which represents
the Old Sorbian dual at Morphological Structure. At Morphological Structure, the
terminal Number Node hosts a marked [-singular -augmented] feature combination of a
dual pronoun (149). For agreement to occur, the dual number features [-singular
-augmented] are copied to an Agreement node (AGR) within the verbal structure (150).
The morphological evidence for the presence of a marked [-singular -augmented] feature
bundle of the dual at Morphological Structure is the dual agreement suffixes -moj and -taj
(146-147).
(149) The Structure of an Old Sorbian Dual Pronoun at MS
D
D
Num
[-sg -aug]
(150) The Structure of an Old Sorbian Dual Verb at MS
T
v
√ ROOT
T
v T
[+pres]
AGR
[-sg -aug]
The dual/plural syncretism of the 1st and 2nd pronouns my and vy shown in (147148) is the result of Vocabulary Insertion via underspecification at Phonological
Structure. Since both the dual [-singular -augmented] and plural [-singular +augmented]
205
share the [-singular] feature, they can be realized by the same Vocabulary Items via
underspecification. The Vocabulary Item /ty/ (2nd person singular) gets inserted into the
[+singular] context as a more specific Vocabulary Item while the Vocabulary Items /my/
(1st person), or /wy/ (2nd person) are completely underspecified (151).
(151) Vocabulary Items for Old Sorbian Pronouns
/ty/ → Num [+singular]
/my/ or /wy/ → elsewhere
In terms of acquisition of number features, I suggest that when the dual/plural
pronominal syncretism emerged in Old Sorbian, the speaker was getting evidence of
markedness of the [-singular -augmented] feature bundle of the dual. At this point, the
same syncretic Vocabulary Item specified as [-singular] for both the dual and plural was
an input into the acquisition process. The [-singular] feature could trigger both dual and
plural agreement which was problematic for a language learner. Therefore, this conflict in
agreement had to be resolved.
I argue that the Principle of Morphosyntactic Feature Economy applied to solve
the dual/plural syncretism which posed an acquisition problem. Similar to the Old
Slovenian dual, a [-singular,-augmented] representation of the Old Sorbian was marked
(152). A Fission rule applied to the marked feature bundle of the dual to split it into
separate terminal nodes (153).
206
(152) The Marked Structure of the Old Sorbian Dual Pronoun before Fission
D
D
Num
[-sg -aug]
(153) Fission Rule for the Old Sorbian Dual
[-singular -augmented] → [-singular] [-augmented]
As a result of Fission, a marked [-singular -augmented] feature bundle hosted by
the Number node was split into two separate terminal number nodes - Num1 with the
[-singular] feature and Num2 with the [-augmented] feature (154). The reanalyzed
structure of the Old Sorbian dual became less marked and therefore more
computationally efficient.
(154) A Reanalyzed Old Sorbian Dual After Fission
D
D
Num
Num1
[-sg]
Num2
[-aug]
After the morphological repair of Fission took place at Morphological Structure,
Vocabulary Insertion occurred to fill the two terminal nodes Num1 and Num2 with
207
phonological material. As in the case with Old Slovenian, two Vocabulary Items were
needed to fill two positions of exponence created by Fission (155). The Vocabulary Items
/me/ (1st person), /we/ (2nd person), or /won/ (3rd person), the plural pronominal stems,
were inserted in the [-singular] context. The Vocabulary Item /j/, the dual suffix, was
inserted in the [-augmented] context since it is the only possible non-singular exponent.55
(155) VIs for the Old Sorbian Dual Pronouns
/me/, /we/, or /won/ ↔ Num [-singular]
/j/ ↔ Num [-augmented]
As I have shown, the monomorphemic Old Sorbian dual which was syncretic with
the plural underwent a historical reanalysis (156). As a result of this morphosyntactic
reanalysis, a reanalyzed bimorphemic dual emerged which was less marked and more
economical for syntactic computation.
(156) Morphosyntactic Reanalysis of the Sorbian Dual
'marked' [-singular -augmented] → [-singular] [-augmented] 'less marked'
We find examples of reanalyzed dual pronouns in one of the most important Old
Sorbian texts of the 16th century, Jakubica's New Testament (1548). The 1st, 2nd, and
3rd person dual pronouns me-y, we-y, wone-y have a bimorphemic morphological
55
As I have discussed in Chapter 4, in Sorbian, the dual suffix -j historically stems from the numeral dwaj
('two').
208
structure which consists of a plural pronominal stem me, we, wone and the dual suffix -y
(157-159).
(157) Old Sorbian (16th century)
Me-y ... moʃche-mey
1.PL-DU can-1.DU.PRES
'We two can.'
(Unger 1998:57)
(158) Old Sorbian (16th century)
Moʃche-tey we-y
pytz ...
can-2.DU.PRES 2.PL-DU drink
'You two can drink ...'
(Unger 1998:57)
(159) Old Sorbian (16th century)
A wone-y knomu reknuʃch-tey ...
and 3.PL-DU to him tell-3.DU.PRES
'And they two answer him ...'
(Unger 1998:58)
In the 19th century Lower Sorbian, we find evidence of reanalyzed bimorphemic
dual pronouns which consist of the plural pronominal stems me- and we- and the dual
suffix -j (160-1). In the examples given below (156-157), the dual reference of the 1st
person and 2nd person dual pronouns mej and wej is confirmed by the dual agreement
suffixes -mej and -tej.
(160) Lower Sorbian (19th century)
Me-j
gledach-mej
wójadnom jaden na drugego.
1.PL-DU
look-1.DU.AOR. together one on other
‘We two looked at each other.’
(Bramborski Serbski Casnik 1852:25, Niedersorbisches Textkorpus)
209
(161) Lower Sorbian (19th century)
We-j
stej
grońi-l-ej
2.PL- DU be.2.DU say-PERF-2.DU
‘We two have said.’
(Bramborski Serbski Casnik 1850:04, Niedersorbisches Textkorpus)
In Contemporary Upper and Lower Sorbian, dual pronouns continue to be part of
the grammar of these two languages (Table 5.2). As I have argued in this section, the
reason for the renewal of dual pronouns in the grammar of Sorbian is a more economical
morphosyntactic structure of the dual at Morphological Structure.
Table 5.2 Bimorphemic Nominative Dual Pronouns in Contemporary Upper and
Lower Sorbian
Num/Person
Upper Sorbian
Lower Sorbian
DU 1.
mó-j
me-j
2.
wó-j
we-j
3.
won-a-j
won-e-j
In sum, the principle of Morphosyntactic Feature Economy is the underlying
reason for diachronic change in the Sorbian dual pronouns. The pronominal system of
Old Sorbian used to have a marked [-singular,-augmented] feature bundle of the dual
(162b). A reanalyzed pronominal number system in Contemporary Upper & Lower
Sorbian retained the dual since it was reanalyzed into a more economical
morphosyntactic structure with two terminal nodes: [-singular] and
[-augmented] (163b).
210
(162) Old Sorbian Pronoun Number System
a. Singular
[+singular -augmented]
b. Dual
[-singular -augmented]
c. Plural
[-singular +augmented]
(163) Contemporary Upper & Lower Sorbian Pronoun Number System
a. Singular
[+singular, -augmented]
b. Dual
[-singular] [-augmented]
c. Plural
[-singular, +augmented]
211
5.6.
The Reanalysis of the 19th-Century Kashuabin Dual as Plural in
Contemporary Kashubian
The Kashubian dual underwent a different diachronic change compared to its
Slovenian and Sorbian counterparts. Unlike Slovenian and Sorbian, in which the dual
continued to 'survive' since it was reanalyzed as bimorphemic, the Kashubian dual was
'lost' since it was replaced by the plural (Table 5.3). I argue that the principle of
Morphosyntactic Feature Economy is responsible for the morphosyntactic reanalysis of
the Kashubian dual as plural.
Table 5.3 The Evolution of the Kashubian Dual Pronouns
Number/Person
19th Century
Kashubian
(Ceynowa 1860)
Kashubian in the
1950s
(Stone 1993)
21st Century
Kashubian
(Hopkins 2001)
DU 1.
ma
ma
-
2.
va
-
-
wón-ji, wón-e
me
më
më
2.
ve
wa
wa
3.
wón-ji, wón-e
oni, onë
woni, wone
3.
PL 1.
Similar to the Slovenian and Sorbian dual, the Kashubian dual was
represented by a marked [-singular -augmented] feature combination at Morphological
Structure (164). This markedness lead to the dual/plural syncretism which occurred
approximately in the middle of the 20th century. The dual/plural syncretism had to be
212
resolved via the principle of Morphosyntactic Feature Economy which applied at
Morphological Structure via the operation of Impoverishment.
(164) The Marked Structure of the Kashubian Dual at MS
D
D
Num
[-sg -aug]
The operation of Impoverishment deleted a marked [-augmented] feature in the
context of a marked [-singular] feature assuming context-sensitive markedness of the
[±augmented] feature (Nevins 2008, 2011) (165). As a result of the application of the
Impoverishment rule, the Kashubian dual retained only the [-singular] feature in its
morphosyntactic representation (166).
(165) Impoverishment Rule for Kashubian:
[-augmented] → ∅/ Num [-singular]
(166) The Kashubian Dual after Impoverishment at MS
D
D
Num
[-sg]
I suggest that after the rule of Impoverishment (166) deleted the [-augmented]
feature in the morphosyntactic structure of the dual, the Kashubian speaker of the next
generation did not have enough morphological evidence to posit the other '+' value of the
213
[augmented] feature. Due to the lack of distinction in the acquisition input, I suggest that
the entire [±augmented] feature was deleted via Impoverishment (167).
(167) Impoverishment of the [±augmented] in Kashubian
[±augmented] → ∅
As a result of Impoverishment of the [±augmented] feature, the Kashubian
pronouns were distinguished only by the [±singular] feature at the Morphological
Structure. Postsyntactic Vocabulary Insertion applied to fill the terminal nodes of the
[±singular] feature. Assuming with (Embick 2010) that Vocabulary competition is local,
i.e. it occurs at a single terminal node, Vocabulary Items competed to realize [+singular]
terminal node.
There were two VIs which were in competition for Vocabulary Insertion at the
[+singular] terminal node /të / and /wa/ (168). The VI /të/ won the competition and got
inserted in the [+singular] context as it is more specific. The VI /wa/ (2nd person) is the
elsewhere items whose context in non-specified.56
(168) VIs for Kashubian Pronouns
/të / → Num [+singular]
/wa/ → elsewhere
The application of the principle of Morphosyntactic Feature Economy provides a
principled explanation why the singular ~ dual ~ plural number system, which Kashubian
56
Vocabulary Insertion of the 1st person VI / më / would proceed in the same way. The context for
insertion would indicate the 1st person features instead of the 2nd person features.
214
pronouns used to have, was replaced by the singular ~ plural number system which
emerged in contemporary Kashubian (169-170). Due to markedness of the [-singular
-augmented] feature combination of the dual, the [±augmented] feature was deleted. As a
result, only one bivalent [±singular] feature was left to distinguish the singular from the
plural in Kashubian.
(169) Kashubian Pronoun Number System (19th- 20th century)
Singular
[+singular, -augmented]
Dual
[-singular, -augmented]
Plural
[-singular, +augmented]
(170) Contemporary Kashubian Pronoun System
Singular
[+singular]
Plural
[-singular]
215
5.7.
The Reanalysis of the Old Russian Dual as Plural in Contemporary
Standard Russian
The 'fate' of the dual in Old Russian is very similar to that of Kashubian. In the
11th century, the Old Russian dual showed the first signs of morphosyntactic instability
due to the dual/plural syncretism in its pronominal system. By the 15-16th centuries, the
Old Russian dual was lost; i.e. it was reanalyzed as plural (171b). I argue that the
application of the principle of Morphosyntactic Feature Economy is the reason for the
reanalysis of the Old Russian dual as plural.
(171) The Two Patterns of Diachronic Changes in the Slavic Dual
a). [-singular,-augmented] → [-singular] [-augmented] = 'reanalyzed dual'
b). [-singular,-augmented] →[-singular] = 'plural'
5.7.1. The Reanalysis of the 2nd Person Dual as Plural in Old Russian
First, I present the analysis of the diachronic changes which occurred in the 2nd
person dual since the 2nd person dual showed the dual/plural syncretism much earlier
than the 1st person dual. As early as in the 11th century, the Old Russian dual vy began to
syncretize with the plural in the 2nd person (172). Despite the dual/plural pronominal
syncretism, the 2nd person dual agreement suffix -ta was very robust, i.e. the verb was
always marked by the dual suffix -ta in the 2nd person.
216
(172) The Old Russian Dual in the 11th Century
jego
že vy
gle-ta
jako slĕpǔ rodi-sę
3.SG.ACC EMPH 2.PL.NOM say-2.DU.PRES as
blind born-REFL
‘You two say that he was born blind.’
(Archangel Gospel, 9.1)
Assuming (Nevins 2008, 2011) that the dual is represented by a marked
[-singular -augmented] feature combination in narrow syntax, I suggest that the Old
Russian dual had a marked morphosyntactic structure in narrow syntax (173). In the
[-singular -augmented] representation of the Old Russian dual both number features are
marked.
(173) The Marked Structure of the Old Russian (11th century) Dual in Narrow
Syntax
D
D
Num
[-sg -aug]
The evidence for the [-singular -augmented] representation of the Old Russian
dual in narrow syntax is in the fact that the dual number features [-singular, -augmented]
are copied from the Num node (173) to an AGR node at Morphological Structure (174).
In the MS structure of the verb gle-ta marked by the dual agreement suffix -ta, the AGR
node has the copied dual number feature [-singular -augmented]. At Phonological Form
the dual agreement features are phonologically realized by the verbal suffix -ta, which we
observed in the above example (172)
217
(174) The Structure of an Old Russian Verb gle-ta with AGR at MS
T
v
√ ROOT
T
v T
[+pres]
AGR
[-sg -aug]
I claim that dual/plural vy syncretism was the consequence of markedness of the
narrow syntactic representation of the Old Russian dual. At Phonological Form, the
dual/plural vy syncretism was the result of underspecifation of Vocabulary Items. There
were two Vocabulary Items competing to realize the [+singular] terminal node (175). The
Vocabulary Item /ty/ was inserted in the [+singular] context since it was more specific.
The Vocabulary Item /vy/ was the elsewhere item.
(175) VIs for the Old Russian Dual Pronouns
/ty/ → Num [+singular]
/vy/ → elsewhere
In the 11-14th centuries, the Old Russian dual vy continued to be syncretic with
the plural in the 2nd person. I argue that during this time period, the principle of
Morphosyntactic Feature Economy had not yet applied to the marked [-singularaugmented] representation of the Old Russian dual at Morphological Structure. The
operation of Impoverishment had not yet deleted the offending [-augmented] feature from
the [-singular -augmented] feature bundle.
218
The argument for the non-application of the principle of Morphosyntactic Feature
Economy to the Old Russian dual in the 11-14th centuries is supported by the presence of
the dual agreement suffix -ta marked on Old Russian verbs, such as gle-ta. As I have
shown above (173-174), the dual number feature combination [-singular -augmented]
was copied to an AGR node at Morphological Structure, and was phonologically realized
as the dual agreement suffix -ta. The dual/plural syncretism was simply resolved via
underspecification.
By the 15th century, the 2nd person dual vy was already reanalyzed as plural since
the Old Russian verb was marked by the 2nd person plural agreement suffix -te (176). I
argue that by the 15th century the principle of Morphosyntactic Feature Economy had
already applied via the operation of Impoverishment. The rule of Impoverishment deleted
the [-augmented] feature in the context of a marked [-singular] feature at Morphological
Structure (177).
(176) The Old Russian Dual Reanalyzed as Plural in the 15th Century
ego
že
vy gl-te
antixrǔsta
3.SG EMPH 2.PL say-2.PL.PRES antichrist
'You two say that he is an antichrist.'
(Radzivilow Codex 103.2)
(177) Impoverishment Rule for Old Russian (end of the 15th century):
[-augmented] → ∅/ Num [-singular]
219
The result of the application of the principle of Morphosyntactic Feature
Economy in the 15th-century Old Russian is a diachronic change from the dual to plural.
The morphosyntactic structure of a reanalyzed Old Russian dual as plural is shown in
(178). In the morphosyntactic structure (178), the terminal Number node retained only
the [-singular] feature similarly to Kashubian. The Old Russian pronoun vy was no longer
represented by the [-singular -augmented] feature combination, but already had an
impoverished [-singular] morphosyntactic representation.
(178) The Change from Old Russian Dual to Plural after Impoverishment at MS
D
D
Num
[-sg]
The reanalysis of the Old Russian dual as plural and its impoverished [-singular]
representation is supported by the plural agreement suffix -te marked on the verb, such as
gle-te. At Morphological Structure, the [-singular] feature of the Number node is copied
to an AGR node within the verbal structure (179). The AGR node with the copied
[-singular] feature is then realized by the VI /te/ at Phonological Structure.
(179) The Structure of an Old Russian Verb gle-te with AGR at MS
T
v
√ ROOT
T
v T
[+pres]
AGR
[-sg]
220
5.7.2. The Reanalysis of the 1st Person Dual as Plural in Old Russian
I argue that the principle of Morphosyntactic Feature Economy is the reason for
the reanalysis of the Old Russian 1st person dual my as plural. The principle of
Morphosyntactic Feature Economy applied via the operation of Impoverishment at
Morphological Structure. The marked [-singular-augmented] representation of the 1st
person dual my was impoverished to retain only the [-singular] feature, which became
less marked.
Unlike the 2nd person dual vy, which became syncretic with the plural in the 11th
century, the first person dual vě did not syncretize with the plural until the end of the 15th
century (cf. 180-181). Up until the end of the 15th century, the 1st person dual vě had
dual reference which is confirmed by the dual agreement suffix -vě marked on the verb
(180).
(180) The 1st Person Dual vě (end of the 14th century)
vě
posle-vě
k bratu
svojemu
1.DU.NOM send-1.DU.PRES to brother own
'We two will send to our brother.'
(Laurentian Codex 58.2)
By the end of the 15th century, the dual/plural my syncretism emerged, but there
were instances where the verb was still marked by the dual suffix -vě (181). I suggest that
221
the reason for the lack of dual/plural syncretism until the 15th century is the phonological
difference in the dual pronominal stem vě and plural stem my.
(181) The 1st Person Dual my (end of the 15th century)
my
posle-vě
ko bratu svojemu
1.PL.NOM send-1.DU.PRES to brother own
'We two will send to our brother.'
(Radziwiłł Codex 101.1)
I argue that in the instances where the 1st person dual was syncretic with the
plural, but the verb still retained the 1st person dual agreement suffix -vě , the principle of
Morphosyntactic Feature Economy did not apply to the marked [-singular -augmented]
feature bundle of the dual. The evidence for this is the presence of the dual suffix -vě
marked on the verb.
I assume that similar to the 2nd person dual, the dual/plural my syncretism is the
result of markedness of the [-singular, -augmented] feature bundle of the 1st person dual
in narrow syntax (182). The evidence for the [-singular -augmented] structure of the 1st
person dual in narrow syntax is the addition of the AGR node with the copied [-singular augmented] number features phonologically realized as the suffix -vě (183).
222
(182) The Marked Structure of the 1st Person Dual in Narrow Syntax
D
D
Num
[-sg -aug]
(183) The Structure of an Old Russian Verb posle-vě with AGR at MS
T
v
√ ROOT
T
v T
[+pres]
AGR
[-sg -aug]
After the AGR node was adjoined to the morphosyntactic structure, Vocabulary Insertion
applied and this node was phonologically realized by the suffix -vě. The dual/plural my
syncretism was resolved via underspecification of Vocabulary Items (184).
(184) VIs for the Old Russian Dual Pronouns
/ja/ → Num [+singular]
/my/ → elsewhere
By the end of the 15th and perhaps by the beginning of the 16 centuries, the 1st
person the Old Russian dual my underwent reanalysis as plural (185). I argue that by this
time in the development of Old Russian, the principle of Morphosyntactic Feature
Economy had applied since the verb bears the plural agreement suffix -mǔ.
223
(185) The 1st Person Dual my (end of the 15th century)
my vǔdae-mǔ
1.PL know-1.PL.PRES
'We (two) know.'
(Radziwiłł Codex 103.2)
The principle of Morphosyntactic Feature Economy applied via the operation of
Impoverishment, which deleted the marked [-augmented] number feature in the context
of the marked [-singular] feature (186). The result of Impoverishment was reanalysis of
the dual as plural since only the [-singular] number feature was hosted by the Number
node (187). Thus, the 1st person dual was reanalyzed as plural.
(186) Impoverishment Rule for Old Russian (end of the 15th century):
[-augmented] → ∅/ Num [-singular]
(187) The Change from Old Russian Dual to Plural after Impoverishment at MS
D
D
Num
[-sg]
The reanalysis of the 1st person dual as plural is evidenced by the plural
agreement suffix on the verb. At Morphological Structure, the [-singular] feature was
copied to an AGR node which was phonologically realized by the suffix -mǔ (188).
224
(188) The Structure of an Old Russian Verb vǔdae-mǔ with AGR at MS
T
v
√ ROOT
T
v T
[+pres]
AGR
[-sg]
The principle of Morphosyntactic Feature Economy was a driving force for the
morphosyntactic reanalysis of the Old Russian dual as plural, at the end of the 15th
century. As a result, the singular ~ dual ~ plural pronominal number system of Old
Russian was reanalyzed as a singular ~ plural system in Contemporary Standard Russian
(189-190). The only feature which distinguishes singular and plural pronouns in
Contemporary Standard Russian is [±singular].
(189)
(190)
Old Russian Pronoun Number System
Singular
[+singular -augmented]
Dual
[-singular -augmented]
Plural
[-singular +augmented]
Contemporary Standard Pronoun Number System
Singular
[+singular]
Plural
[-singular]
225
5.8.
Summary
In this chapter, I have argued that the principle of Morphosyntactic Feature
Economy provides a principled explanation for the two pattern of diachronic change of
the Slavic dual within the framework of Distributed Morphology. I have shown that the
principle of Morphosyntactic Feature Economy applied to a marked [-singular augmented] representation of the Slavic dual via the morphological repairs of Fission and
Impoverishment. The reason for these repairs was elimination of the marked [-singular augmented] feature combination of the Slavic dual.
In Old Slovenian and Old Sorbian, the principle of Morphosyntactic Feature
Economy applied via Fission to split the morphosyntactic feature bundle of the dual into
two separate terminal nodes - [-singular] and [-augmented]. As a result, markedness of
the feature bundle was eliminated, and the dual was restructured as bimorphemic in
contemporary Slovenian and Sorbian. In Old Russian and Kashubian, the principle of
Morphosyntactic Feature Economy applied via Impoverishment whereby the
[±augmented] feature was deleted. Due to the deleted [±augmented] feature, markedness
of the [-singular, -augmented] feature combination was eliminated, and contemporary
Russian and Kashubian employ a singular ~ plural pronominal number system.
226
CHAPTER 6: CONCLUSION
6.1.
The Evolution of Dual Number and the Principle of
Morphosyntactic Feature Economy
The origin of dual number in natural language is not clear cut. Some researchers
argue that conceptualization of dual number might have stemmed from natural symmetry
of the human body. "According to Humboldt, duality indeed is a fundamental category of
human experience, cognition, and feeling" (Plank 1989:319). Žolobov (1998, 2001) also
suggests cognitive underpinnings of the dual. He argues that natural pairs of human body
parts gave rise to grammaticalization of natural pairs and consequently to
grammaticalization of dual number as a grammatical category.
If the hypothesis that dual number originated from natural pairs of human body
parts and was conceptualized in human thought, is on the right track, we expect the
following stages of grammaticalization. First, natural symmetry of the human body was
conceptualized through natural pairs, and natural pairs, such as legs, eyes, ears, hands
were grammaticalized as paired nouns. Next, natural pairs gave rise to dual pronouns as
more abstract concepts.
Analyzing linguistic cycles, van Gelderen argues that Feature Economy
minimizes content of morphosyntactic features in order to maximize computational
efficiency of the syntactic derivation (191repeated from 133). If we follow van
Gelderen's (2008, 2009, 2011) principle of Feature Economy, grammaticalization of dual
number becomes clearer. Interpretable number features of nouns (DPs) in specifier
227
positions change to uninterpretable features of syntactic heads, and consequently become
uninterpretable features corresponding to agreement affixes.
(191) Feature Economy (van Gelderen 2011:17)
Minimize the semantic and interpretable features in the derivation:
Adjunct
semantic
Specifier
> [iF]
>
Head
[uF]
Affix
>
[uF]
These stages of grammaticalization of dual number are attested in Slavic languages. For
example, in Old Church Slavic and Old Russian, the majority ( approximately 95%) of
sentences contain pro-drop subjects which points to the uninterpretable features on T and
v syntactic heads.
The Principle of Morphosyntactic Feature Economy is motivated not only by
morphosynatactic markedness of the [-singular -augmented] features bundle, but also by
semantic markedness of dual number. According to Sauerland (2008), the dual is more
semantically marked than the plural. If we take into account cognitive origin and
conceptual evolution of the dual, this argument is well supported. Dualities are more
difficult to conceptualize than pluralities. Therefore, I agree with Sauerland (2008) that
dual number is more semantically marked. Diachronic change from the dual to the plural
in Old Russian and Kashibian confirms semantic markedness of dual number.
228
6.2.
Theoretical Implications of the Morphosyntactic Feature Economy
In this dissertation, I proposed that the principle of Morphosyntactic Feature
Economy is a driving force behind the two different patterns of diachronic change in the
Slavic dual. The next step is to determine what predictions this principle makes for dual
pronouns attested cross-linguistically. According to Hypothesis 1 postulated for a
possible diachronic change in Slavic dual, a marked [-singular -augmented]
representation is split into two syntactic terminal nodes - [-singular] [-augmented]. The
two syntactic terminal nodes should be realized by two separate Vocabulary Items which
results in a bimorphemic dual pronoun. I have shown that Hypothesis 1 is borne out in
the Slovenian and Sorbian dual pronouns. Will it be borne out in dual pronouns crosslinguistically?
Let us consider some the dual in languages typologically unrelated to Slavic. For
example, a dual pronoun from Manam (Austronesian) has a complex structure which is
bimorphemic. It consists of a plural suffix -di and a dual suffix -ru (192). Another
example of a bimorphemic dual is attested in the nominal dual in Hebrew (Afro-Asiatic)
(193). In Manam and Hebrew, the pro/nominal duals are bimorphemic. The plural suffix
is encoded by the [-singular] feature, and the dual suffix or numeral two is encoded by
the [-augmented] feature. In combination, both features yield dual number, or referential
cardinality of 2.
229
(192) Manam (Austronesian)
a.
áine
ŋara
woman that
'that woman'
b.
áine
ŋara-di-a-ru
woman that-3PL-EP-DU
'those two women'
c.
áine
ŋara-di
woman that-3PL
'those women'
(Nevins 2011:423)
(193) Hebrew (Afro-Asiatic)
Svu’–ay-im
week-two-PL
‘two weeks’
(Ritter 1995:409)
Another example of the distribution of the [-singular] and [-augmented] features can be
found in Hopi (194). In Hopi, both the pronominal and nominal duals are 'constructed' in
that they are encoded by a combination of the [-singular] and [-augmented] features.
(194) Hopi (Uto-Aztecan)
a. Pam taaqa wari.
that man-SG run.SG.PERF
‘That man ran.’
b.
Puma taaqa-t wari.
those man-PL run. SG.PERF
‘Those two men ran.’
c.
Puma taa-taq-t
yuɁtu.
those RED.man-PL run.PL.PERF
‘Those men ran.'
230
More extensive cross-linguistic analysis of pronominal duals is needed to
establish that Hypothesis1 holds cross-linguistically. I have already shown that
Hypothesis 2 is borne out in Russian (East) and Kashubian (West), as examples of Slavic
languages from two different subgroups. In the future, more research emphasis on nonIndo-European languages with dual number is required to establish that the principle of
Morphosyntactic Feature Economy which I proposed in this dissertation accounts for
diachronic change in a principled way.
231
APPENDIX A ABBREVIATIONS
ACC
accusative
AGR
agreement
AOR
aorist
AUG
augmented
C
complementizer
CP
complementizer phrase
DAT
dative
DP
determiner phrase
DU
dual
FEM
feminine
FUT
future tense
GEN
genitive
IMPF
imperfective
INF
infinitive
INH
inherent
INSTR
instrumental
INT
internal
K
case
LOC
locative
MASC
masculine
NEG
negative
NOM
nominative
232
NUM
number
P
person
PART
participant
PERF
perfective
PL
plural
PRES
present
PAST
past
QUANT
quantificational
REFL
reflexive
SG
singular
SPR
speaker
STRUC
structural
T
tense
TH
theme vowel
TP
tense phrase
v
"small" v
vP
"small" verb phrase
V
verb
VI
vocabulary item
VOC
vocative
VP
verb phrase
233
APPENDIX B ORTHOGRAPHICAL SYSTEMS AND TRANSLITERATION
SYMBOLS
Old Church Slavic and Old Russian
Cyrillic
,
( )
,
Transliteration Symbol
A
B
V
G
D
E
Ž
Dz
Z
I
I
G
K
L
M
N
O
P
R
S
T
U
F
X
W
Ts
Č
Š
Št
Ǔ
Y
Ĭ
Ĕ
Ju
Ja
Je
Ę
Ję
Ǫ
Jǫ
Ps
Th
i, v
234
Slovenian
A, a
B, b
C, c
Č, č
D, d
Dž, dž
E, e
F, f
G, g
H, h
I, i
J, j
K, k
L, l
Lj, lj
M, m
N, n
Nj, nj
O, o
P, p
R, r
S, s
Š, š
T, t
U, u
V, v
Z, z
Ž, ž
235
Russian
Cyrillic
A, a
Б, б
В, в
Г, г
Д, д
Е, е
Ё, ё
Ж, ж
З, з
И, и
Й, й
К, к
Л, л
М, м
Н, н
O, o
П, п
Р, р
С, с
Т, т
У, у
Ф, ф
Х, х
Ц, ц
Ч, ч
Ш, ш
Щ, щ
Ъ, ъ
Ы, ы
Ь, ь
Э, э
Ю, ю
Я, я
Transliteration Symbol
A
B
V
G
D
E
Ë
Ž
Z
I
J
K
L
M
N
O
P
R
S
T
U
F
X
Ts
Č
Š
šč
''
Y
'
È
Ju
Ja
236
Upper and Lower Sorbian
A, a
B, b
C, c
Č, č
D, d
Dź, dź
E, e
ě
F, f
G, g
H, h
Ch, ch
I, i
J, j
K, k
Ł, ł
L, l
M, m
N, n
ń
O, o
ó
P, p
R, r
ř, ŕ
S, s
Š, š
Ś, ś
T, t
Ć, ć
U, u
W, w
y
Z, z
Ž, ž
Ź, ź
237
REFERENCES
Bailyn, John Frederick. 2004. The case of Q. In Proceedings of FASL 12, ed. by Olga
Arnaudova, Wayles Browne, Maria Luisa Rivero, and Danijela Stojanovic , 1-36,
Ann Arbor: Michigan Slavic Publications.
Belič, Alexander. 1899. K dvojstevnnomu čislu v staroslavjanskom. Izvestija Otdelenija
Russkogo Jazyka i Slovesnosti. 4. 1159-91.
Benveniste, Emile. 1971. The nature of pronouns. Problems in general linguistics, 217–
22. Coral Gables, FL: University of Miami Press.
Bianchi, Valentina. 2006. On the syntax of personal arguments. Lingua.116. 2023-2067.
Bobaljik, Jonathan. 2002. Syncretism without Paradigms: Remarks on Williams 1981,
1994. In Yearbook of Morphology 2001, ed. by G. Booij and J. van Marle, 53-86,
Dordrecht: Kluwer.
Bobaljik, Jonathan. 2008. Where's Phi? Agreement as a postsyntactic operation. In In Phi
Theory: Phi-Features across Modules and Interfaces, ed. by Daniel Harbour,
David Adger, and Susana Bejar, 295-328, Oxford: Oxford University Press.
Bonet, Eulàlia. 1991. Morphology after Syntax: Pronominal Clitics in Romance. Doctoral
dissertation, MIT.
Breza, Edward, and Jerzy Treder. 1981. Gramatyka Kaszubska. Gdańsk : Zrzeszenie
Kaszubsko-Pomorskie.
Caha, Pavel. 2009. The Nanosyntax of Case. Doctoral dissertation, University of Tromsø.
Calabrese, Andrea. 2011. Investigations on markedness, syncretism, and zero exponence
in morphology. Morphology. 21:283–325.
Chomsky, Noam. 2001. Derivation by phase. In Ken Hale: A Life in Language, ed. by
Michael Kenstowicz, 1-52, Cambridge: MIT Press.
Chomsky, Noam. 2005. Three factors in language design. Linguistic Inquiry. 36.1-22.
Chomsky, Noam. 2008 a. Approaching UG from below. In Interfaces +Recursion:
Chomsky's Minimalism and the View from Syntax-Semantics, ed. by Uli Sauerland
and Hans-Martin Gärtner, 1-30, Berlin: Mouton de Gruyter.
238
Chomsky, Noam. 2008 b. The Biolinguistic Program: Where Does It Stand Today? Ms.
Chomsky, Noam. 2008 c. On phases. In Foundational Issues in Linguistics Theory:
Essays in Honor of Jean-Roger Vergnaud, ed. by Robert Freidin, Carlos P. Otero,
and Maria Luisa Zubizarreta, 133-166, Cambridge: MIT Press.
Corbett, Greville G. 2000. Number. Cambridge: Cambridge University Press.
Cowper, Elizabeth. 2005. A note on number. Linguistic Inquiry. 36:441–455.
Cysouw, Michael. 2009. The paradigmatic structure of person marking. Oxford: Oxford
University Press.
Derganc, Alexandra. 1988. On the history of the dual in Slovene and Russian. Wiener
Slawistischer Almanach. 22.237-47.
Derganc, Alexandra. 1998. Parallels in the development of the dual in Slovene and
Sorbian. Slavisticna Revija. 216.45-54.
Derganc, Alexandra. 2003. The dual in Slovenian. Spachtypologie und
Universalienforschung. 56. 165-81.
Di Sciullo, Anna Marie, and Edwin Williams. 1987. On the Definition of Word.
Cambridge: MIT Press.
Duličenko, Aleksandr and Werner Lehfeldt. 1998. Kurze Betrachtungen über die
kaßubische Sprache, als Entwurf zur Gramatik. G ttingen: Vandenhoeck and
Ruprecht.
Elle, Ludwig. 2000. Die heutige Situation der Sorbischen Sprache und Konzepte zu ihrer
Revitalisierung. In Maintenance, revitalization, and development of minority
languages, ed. by Šatava, Leoš, and Susanne Hose, 17-21, Bautzen/Budyšin:
Sorbisches Institut/Serbski Institut.
Embick, David. 2011. Localism versus globalism in morphology and phonology.
Cambridge: MIT Press.
Embick, David, and Rolf Noyer. 2007. Distributed morphology and the
syntax/morphology interface. In Oxford Handbook of Linguistic Interfaces, ed.
by Giliian Ramchand and Charles Reiss, 289-324, Oxford: Oxford University
Press.
Forchheimer, Paul. 1953. The category of person in language. Berlin: Walter de Gruyter.
239
Franks, Steven. 2002. A Jakobsonian Feature Based Analysis of the Slavic Numeric
Quantifier Genitive. Journal of Slavic Linguistics. 10.141-181.
Gelderen, Elly van. 2008. Where did late merge go? Grammaticalization as feature
economy. Studia Linguistica. 62. 287-300.
Gelderen, Elly van. 2009. Introduction. In Cyclical Change, ed. by Elly van Gelderen, 112, Amsterdam: John Benjamins.
Gelderen, Elly van. 2011. The Linguistic Cycle: Language Change and the Language
Faculty. Oxford: Oxford University Press.
Gamkrelidze, T.V., and V.V. Ivanov. 1984. Indoevropejskij yazyk i
indoevropejcy:Rekonstrukcija i istoriko-tipologičeskij analiz prajazyka i
protokultury. Tbilisi: Publishing House of the Tbilisi State University.
Gasparov, Boris. 2001. Old Church Slavonic. München: Lincom Europa.
Greenberg, Joseph H. 1963. Some universals of grammar with particular reference to the
order of meaningful elements. In Universals of language, ed. by Joseph H.
Greenberg, 73–113. Cambridge, MA: MIT Press.
Greenberg, Mark. 2008. A Short Reference Grammar of Standard Slovene. München:
Lincom Europa.
Hale, Kenneth. 1973. Person marking in Walbiri. In A Festschrift for Morris Halle, ed. by
Stephen R. Anderson and Paul kiparsky, 308-344. New York: Holt, Rinehart, and
Winston Inc.
Hale, Kenneth. 1997. Some observations on the contribution of local languages to
linguistic science. Lingua. 100.71-89.
Halle, Morris. 1997. Distributed Morphology: Impoverishment and Fission. MIT Working
Papers in Linguistics. 30.425-449.
Halle, Morris, and Alec Marantz. 1993. Distributed Morphology and the Pieces of
Inflection. In The View from Building 20: Essays in Linguistics in Honor of
Sylvain Bromberger, ed. by Ken Hale and Samuel Jay Keyser, 111-76,
Cambridge: MIT Press.
Hopkins, Paul. 2001. The Phonological Structure of the Kashubian Word. Doctoral
disseratation, State University of New York.
240
Harbour, Daniel. 2008. Morphosemantic Number: From Kiowa Noun Classes to UG
Number Features. Dordrecht: Springer.
Harbour, Daniel. 2011a. Descriptive and explanatory markedness. Morphology. 21.223–
45.
Harbour, Daniel. 2011b. Valence and atomic number. Linguistic Inquiry. 42. 561-594.
Harley, Heidi, and Rolf Noyer. 1999. Distributed morphology. Glot International. 4.3-9.
Harley, Heidi. 2008. When is a syncretism more than a syncretism? Impoverishment,
Metasyncretism. and Underspecification. In Phi Theory: Phi-Features across
Modules and Interfaces, ed. by Daniel Harbour, David Adger, and Susana Bejar,
251-294, Oxford: Oxford University Press.
Harley, Heidi, and Elizabeth Ritter. 2002. Person and number in pronouns: A featuregeometric analysis. Language.78. 482–526.
Haspelmath, Martin. 2006. Against markedness (and what to replace it with). Journal of
Linguistics 42:25–70.
Humboldt, Wilhelm von. 1827. Űber den Dualis. Lecture. Gelesen in der Akademie der
Wissenschaften am 26. April 1827.
Humboldt, Wilhelm von. 1997. On the dual form. In Essays on language, ed. by T.
Harden and D. Farrelly, 111-136, Frankfurt am Main: Peter Lang.
Ingram, David. 1978. Typology and universals of personal pronouns. In Universals of
Human Language, ed. by Joseph H. Greenberg, 213-248, Stanford: Stanford
University Press.
Iordanskij, A.M. 1960. Istorija dvojstvennogo čisla v russkom jazyke. Vladimir.
Ivanova, T.A. 1977. Staroslavjanskij jazyk. Moskva: Vysšaja Škola.
Jakobson, Roman. 1984. Russian and Slavic grammar. Berlin: Mouton.
Jakobson, Roman. 1985. Izbrannye Raboty. Moskva: Progress.
Jakop, Tjaša. 2008 a. The Dual in Slovene Dialects. Bochum: Universitätsverlag Dr. N.
Brockmeyer.
Jakop, Tjaša. 2008 b. Dvojinske oblike v delih protestantskih piscev. Online:
http://www.centerslo.net/files/file/simpozij/simp27/28_Jakop.pdf
241
Jeanne, Laverne Masayesva. 1978. Aspects of Hopi Grammar. Doctoral dissertation,
MIT.
Jespersen, Otto. 1965. The philosophy of grammar. New York: W.W. Norton &
Company.
Kratzer, Angelika. 1996. Severing the External Argument from its Verb. In Phrase
Structure and the Lexicon ed. by J. Rooryck and L. Zaring, 109-137, Dordrecht:
Kluwer.
Krivčik, V. F., and N. S. Možejko. 1985. Staroslavjanskij jazyk. Minsk: Vyšejšaja Škola.
Lieber, Rochelle. 1992. Deconstructing morphology. Chicago: University of Chicago
Press.
Link, Godehard. 1983. The logical analysis of plurals and mass terms: A latticetheoretical approach. In Meaning, use, and interpretation of language, ed. by R.
Bäuerle, C. Schwartze, and A. von Stechow, 302-323, Berlin: Walter de Gruyter.
Lunt, Horace L. 2001. Old Church Slavonic grammar. Berlin: Mouton de Gruyter.
Marantz, Alec. 1991. Case and Licensing. Proceedings on the Eastern States Conference
on Linguistics. 8. 234-253.
McFadden, Thomas. 2004. The Position of Morphological Case in the Derivation: A
Study of the Syntax-Morphology Interface, Doctoral Dissertation, University of
Pennsylvania.
Meillet, Antoine. 1924. Le slave commun. Paris: E. Champion.
Možejko, N. S., and A. P. Ignatenko. Drevnerusskij jazyk. Minsk: Vyšejšaja Škola.
Müller, Gereon. 2004. A Distributed Morphology Approach to Syncretism
in Russian Noun Inflection. In Proceedings of FASL 12, ed. by Olga Arnaudova,
Wayles Browne, Maria Luisa Rivero, and Danijela Stojanovic , 353-374, Ann
Arbor: Michigan Slavic Publications.
Nevins, Andrew. 2006. Dual is still more marked than plural. Online:
http://ling.auf.net/lingbuzz/000235
Nevins, Andrew. 2008. Cross-modual parallels in the study of phon and phi. In Phi
Theory: Phi-Features across Modules and Interfaces, ed. by Daniel Harbour,
David Adger, and Susana Bejar, 329-367, Oxford: Oxford University Press.
242
Nevins, Andrew. 2011. Marked triggers vs. marked targets and impoverishment of the
dual. Linguistic Inquiry. 42.413-44.
Noyer, Rolf. 1997. Features, Positions, and Affixes in Autonomous Morphological
Structure. Garland Publishing: New York and London.
Noyer, Rolf. 1998. Impoverishment theory and morphosyntactic markedness. In
Morphology and Its Relation to Syntax and Phonology, ed. by Lapointe S.G.,
D.K. Brentari, and P.M. Farrell, 264-285, Stanford: CSLI.
Perelstvaig, Asya. 2007. On the Universality of DP: A View from Russian. Studia
Linguistica. 61. 59-94.
Pesetsky, David, and Esther Torrego. 2001. T-to-C Movement: Causes and
Consequences. In Ken Hale: A Life in Language, ed. by Michael Kenstowicz,
355-426, Cambridge: MIT Press.
Plank, Frans. 1989. On Humboldt on the dual. In Linguistic Categorization, ed. by
Roberta Corrigan and Michael Noonan, 293-333, Amsterdam: John Benjamins
Publishing Company.
Richards, Norvin. 2007. Lardil “case stacking” and the structural/inherent case
distinction.
Online: http://ling.auf.net/lingbuzz/000405
Ritter, Elizabeth. 1991. Two functional categories in Noun Phrases: Evidence from
Modern Hebrew. In Syntax and Semantics 25: Perspectives on Phrase Structure,
ed. by Susan Rothstein, 37-62, New York: Academic Press.
Ritter, Elizabeth. 1995. On the syntactic category of pronouns and agreement. Natural
Language and Linguistic Theory. 13. 405-43.
Samsonov, N.G. 1973. Drevnerusskij jazyk. Minsk: Vyšejšaja Škola.
Sauerland, Uli. 2008. On the semantic markedness of phi-features. In Phi Theory: PhiFeatures across Modules and Interfaces, ed. by Daniel Harbour, David Adger,
and Susana Bejar, 57-82, Oxford: Oxford University Press.
Schaarschmidt, Gunter. 1998. A historical phonology of the Upper and Lower Sorbian
Languages. Heidelberg: Universitätsverlag C. Winter.
Schaarschmidt, Gunter. 2002. Upper Sorbian. München: Lincom Europa.
243
Scholze, Lenka. 2007. Das grammatishe System der obersorbischen Umgangssprache.
Doctoral Dissertation, Universität Konstanz.
Silverstein, Michael. 1976. Hierarchy of features and ergativity. In Grammatical
Categories in Australian Languages, ed. by R.M.W. Dixon, 112-171, Canberra:
Australian Institutes of Aboriginal Studies.
Schuster-Šewc, Heinz. 2000. Zur Genese der Dualflexion im Sorbischen. Das Sorbische
im slawischen Kontext:Ausgewälte Studien. 24.90-93.
Stone, Gerald. 1993. Sorbian. In The Slavonic Languages, ed. by Bernard Comrie and
Greville G. Corbett, 593-685, London: Routledge.
Stone, Gerald. 1993. Cassubian. In The Slavonic Languages, ed. by Bernard Comrie and
Greville G. Corbett, 759-794, London: Routledge.
Sussex, Ronald, and Paul Cubberley. 2006. The Slavic Languages. Cambridge:
Cambridge University Press.
Šaxmatov, Aleksej. 1957. Istorićeskaja Morfologija Russkogo Jazyka. Moskva:
Gosudarstvennoe Učebno-Pedagogičeskoe Izdatel'stvo.
Topolińska, Zuzanna. 1974. A Historical Phonology of the Kashubian Dialects of Polish.
The Hague: Mouton.
Travis, Lisa. 1984. Parameters and Effects of Word- Order Variation. Doctoral
Dissertation, MIT.
Unger, Mike. 1998. Studien zum Dual: Eine Darstellung am Niedersorbischen Neuen
Testament Miklawuš Jakubica (1548). Specimina Philologiae Slavicae. 118.
München: Verlag Otto Sagner.
Woolford, Ellen. B. Lexical case, inherent case, and argument structure. Linguistic
Inquiry 37.111-130.
Xaburgaev, G.A. 1986. Staroslajanskij Jazyk. Moskva: Prosveščenie.
Young, Robert, and William Morgan. 1980. The Navajo Language. Albuquerque:
University of New Mexico Press.
Žolobov, Oleg Feofanovich. 1998. Simvolika i istoricheskaja dinamika slavjanskogo
dvojstvennogo čisla. Frankfurt am Mein: Peter Lang.
244
Žolobov, O.F., and V.B. Krys'ko. Istoričeskaja grammatika drevnerusskogo jazyka:
Dvojstvennoe čislo. Moskva: Azbukovnik.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement