Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
THE SLAVONIC
AND EAST EUROPEAN
REVIEW
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
334 D- J- L- JOHNSON
Endung -a zu erklaren, miissen deshalb
Scheitern verurteilt angesehen werden (p.
This is a counsel of despair. If one insists th
of -a must be provable by the norms of th
is doomed to failure. The defective textual material makes it difficult
to resurrect the structure of synchronic segments of the language as a
whole and well-nigh impossible to reconstruct with clarity the system
of particular dialects. The texts are based on dialects which borrowed
the termination when it was already well established elsewhere. The
norms of the innovating dialect were clearly not those of the language
of the texts.
This does not mean, however, that it is neither possible nor
profitable to formulate a theory to explain the development of -a
even though there is no hope of proving beyond doubt its validity on
the material of the texts. To be acceptable, though, such a theory
must fulfil certain conditions. First, it must be based on a postulated
set of changes that are intrinsically likely. If -a is to be accounted
for morphologically, those changes must clearly be within the bounds
of the principles that delimit the possible types of analogical change.
Second, those changes must be seen as part of the developing system,
and any assumption about the structure of the innovating dialect
must not be in conflict with what is known about the general lines of
development of Serbo-Croat in the period in question. The postulated
structures should fit into the known pattern of variations. Third,
though the postulated structures may not be provable by the formal
norms of the texts it is clearly desirable that there should be some
clues to the existence of such structures in the irregular, variant forms
which must certainly in many cases reflect norms of other dialects. If
such criteria are applied to the theories of Belie and Karlgren, it be?
comes clear that their unacceptability is not simply a question of not
being provable in terms of the norms of the language of the texts.
Karlgren motivates the development of -a in the following way:
with the disappearance of final jers [i>, b], which originally marked
the genitive case in the plural (0-, u-, ^-declensions) the symmetry of
the paradigm was disturbed and the need was felt for the addition of
a new ending. This motivation clearly infringes the second criterion
for an acceptable theory. The loss of final jers in Serbo-Croat must
have taken place by the second half of the 1 ith century at the latest if
consequent changes associated with the loss were to have had time to
establish themselves in the language by the 12th century, when the
first texts appear. The first sporadic notations of -a occur in the
14th century. Making allowance for the inevitable time-lag before the
reflection of the new ending in the texts, it would thus seem that at
the very least estimation two hundred years elapsed between the loss
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 335
of jers and the beginning of the development of -a in
plural. Or, to put it another way, the language happily to
ending in this case form for at least two hundred years w
veloping a new ending, and many dialects did so for c
longer. The loss of jers could, therefore, have had nothing
the motivation of the new ending. As regards the process
-a arose, Karlgren postulates that it developed in the f
declension paradigm under the influence of the masculine
sion. He assumes the existence of the following pattern of
the genitive and locative plurals:
Masc. jo-Declension Fem. a-Declension
Gen. pl. muz -1 zen -0
Loc. pl. muz -ih zen -ah
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
336 D. J. L. JOHNSON
However, even if wc accept the existence of -i, -
paradigm, serious doubts arise as to the practical r
posed analogy. If j0-stem masculines were used
plural with the ending -1 they were almost certai
the ending -0 (zero). If they were capable of having
plural, it was also possible for them to take -e
phonological developments in the history of Serbo
ceased to be any phonetic necessity for the distinc
soft stem endings. This, together with the mixing
declension masculines, produced a situation where
characterised by a considerable latitude of free var
to case endings. So, for instance?
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 337
Plural paradigm
No special relationship b
speakers were to see arbit
transfer them to another
analogy is not arbitrary,
elimination of a grammar
system (often rules apply
classes). It does not alter
structure.
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
338 D. J. L. JOHNSON
Yet even if we accept that -I, -ih was not co
stems it does not help us to see how this resulted
context of speech, in the development of -a. For
would need to postulate a situation in which a
grammar structure was occupied by a rule gov
ship of the locative to the genitive plural. The ru
to a locative plural termination in Vowel + h,
spond a genitive of that same vowel, long, min
If at the same time there was a subsidiary 'exc
effect that in ^-declension feminines, the relatio
to genitive pl. was Vah/0 it is easy to see that th
peripheral, and be eliminated in favour of th
difficult, though, to see how a general rule su
could arise unless there were a specially tight stru
the genitive and locative cases. For one case form
in terms of another is unusual. The closeness of th
two, and the similarity of the terminations, wou
guage with an inherently unstable situation. It s
that such a rule could be maintained without the
and entering a state of free variation.
What was the situation in Serbo-Croat at the p
There certainly were points of contact between
dual there was no formal distinction between the
of the dual as a separate category, but the con
forms with plural meaning, this lack of distinct
influenced the relationship of the two cases in t
no formal distinction between the two cases in th
adjectives and pronouns. There was some seman
the two cases, as, for instance, in adverbial time
other hand, the two cases were clearly distinguis
declension of all nouns except the {-stems, and th
part semantically distinct. Karlgren assumes the f
the two cases in the plural. He puts this forth as
porting argument for his theory. He does not, ho
sider it crucial for his case. There are rare examp
of the locative ending being used for the gen
KpaHHaHex).5 Examples of the use of locative for
become fairly frequent in Dalmatian texts fro
Karlgren quotes an early example of the use
genitive from Danicic.6 He fails, however,
Danicic's view, the rare examples of this are
scribal error, engendered by the identical adjectiv
5 F. Miklosich, Monumenta Serbica, Vienna, 1858, no. 38, p. 3
6 D. Danicic, Istorija oblika srpskoga iii hrvatskoga jezika, Belgr
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 339
cases. Neither does he mention Daniele's opinion that the la
fusion of the two cases with ^-sterns was the consequence of
tion of the genitive plural in -a. He does not mention th
found not a single example of confusion in 14th-century te
that the confusion invariably occurs only with an adjective
That the confusion always takes the form of the use of the l
the genitive, and not vice versa, emphasises the role of the
here. Danicic found only two examples in the 16th century o
of genitive for locative.7 Thus, although there may have be
association of the genitive and locative cases in the plural, t
tionship fell far short of merger or free variation. Moreove
dialects the links between locative and dative proved d
That being so, there must be serious doubt whether there c
have existed in the language such a rule of grammar as
engendered the proposed analogy.
Nevertheless, if we allow for that possibility, we then com
face with the final, irreconcilable contradiction of Karlgren
If we postulate the existence of that structure which al
generate the new ending in -a, namely that the relationshi
the locative and genitive was governed by the general rule
but that the feminine ^-declension represented an exception
must conclude that the situation in the other declension
follows:
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
340 D. J. L. JOHNSON
was eliminated. But if these other clas
this case the basis for the spreading of
the other hand, one admits the pre
masculines and neuters which would make the extension of -a to
them possible, then the general force of the grammar rule on whi
the postulated analogy is based is destroyed. If such endings we
present, and therefore such relationships between locative and gen
tive as eh/0 or ih/0, then there could be no such rule and therefore
analogy.
It is possible to envisage that there may have been small groups of
words or individual words that provided exceptions to the general
rule, as did the nouns in -a. But, if the general rule was to be a rule of
sufficient generality to engender the analogy under discussion, these
groups or individuals must have occupied a position of very marginal
importance in the system. It is difficult to see how they could have
been a vehicle for the subsequent extension of -a.10
Thus Karlgren, in the elaboration of his theory, makes not one
assumption unsupported by textual evidence, but many. In the end,
there also remains the inescapable contradiction. If the grammar rule
that was alone capable of generating -a was to have the universality
essential for the process, it presupposes a structure that makes im?
possible the extension of-a. If we presuppose a structure under which
the extension of -a could take place, then there could have been no
rule of sufficiently general force to generate -a.
Belic's theory is preferable to Karlgren's in at least one respect. He
makes only one assumption that is not supported by textual evidence,
about the structure of the dialect that innovated -a. Unlike Karlgren,
who assumes the presence of structures that cannot be shown to have
existed elsewhere, either before or after, Belie postulates only that a
particular feature developed earlier in the dialect in question than it
is shown by texts to have done elsewhere.
Belic's argument is as follows. With the loss of the dual as a
separate category, the old terminations marking the dative and in?
strumental dual did not disappear but continued in use with plural
meaning side by side with the old plural endings,
9 If, as is likely, -a first developed in the feminine a-declension, there would have been a
transitional stage of free variation of the old and new ending, e.g., zen -0/zen -a. Since both
masculine and neuter o-declension nouns also had 0 ending in the genitive plural, the same
variation began to develop with them too, e.g. grad -0/grad -a.
10 Masculine nouns with genitive plural in -ov could have been such a group. It seems
that -ov was generalised as a suffix in free variation with non-suffixed forms at an early
date so that there was obtained a paradigm such as gradovi gradove gradov gradovom,
etc. In such cases, the relationship of loc. to gen. pl. would have been the same as for the
mass of o-declension nouns?eh/o. There may have been a small class of nouns, though,
that did not generalise -ov as a suffix, but retained it as a gen. pl. termination, e.g.,
dinari, dinarov, dinareh.
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 34I
e.g. in the feminine a-declension:
Dat. pl. -am, -ama
Inst. pl. -ami, -ama
This use engendered in both cases the appearance of a free
-am/-ama where the form -ama was analysed by the lingui
sciousness of the speakers into -am-j-a where -f-a repre
optional addition to the normal ending -am. Since the genit
was the only case of the plural to be monosyllabic (Belie
primarily a creation of disyllabic oxytonic <z-stems), this o
was transferred to it,
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
342 D. J. L. JOHNSON
plural. The process of analogy cannot r
of unrelated phones or morphs: it r
structural norm from one part of the
tion lies in the psychological need for
each individual example must be sou
occupied in the language structure by
forms, and in the relationship of the in
The nature of the process itself presup
should have a structural identity with
The structure of the -am/-ama altern
starting point would have been repr
sciousness of the speakers as stem + ca
mark would structurally subdivide i
appendage that could be added at the
stem + (ending ? a)
stem + (_C ? a)
If one excludes the nominative and acc
was no possibility of the extension of
were marked by a vowel ending, then
the structure of the dative and instrumental cases could become
dominant in the plural paradigm, and therefore be extended by
analogy to the other cases, that is, the genitive and locative. Since the
structural element whose analogical transference is here in question
is the morph + a, it is clear that the structure of any recipient case
would have to exemplify the formula:
stem+(_C)
The genitive plural manifestly did not consist of such a structure an
therefore could not possibly have been the object of such an analog
The structure of the genitive plural was:
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 343
A + a added to the old structure of the genitive plural w
evitably and simultaneously destroy that structure; a +a
zero eliminates the zero. What we would obtain is not a variant of
the old case mark, but a new ending of totally different quality. If
new form zena were created, and it were to exist side by side with th
old form, zen, the relationship between, the two would not be of th
same type as that between zenam/zenama, but of the same type
existed, for example, in Church Slavonic in the genitive singular of
the masculine declension:
rhiNoy/chiNd
stem + (_C + a)
then the case most open to the play of analogical forces was clearly
the locative case which had the structure:
stem + (_C)
This is precisely the structure which one might have expected to de?
velop the morph ? a, that is to say by analogy with zenam/zenama,
zenah/zenaha. There is absolutely no evidence of such a form as
zenana having existed, and Belie certainly does not postulate one,
even though the locative plural in -ah was, according to all indica?
tions, very much a live one, well beyond the period of the develop?
ment of -a in the genitive plural. The absence of ? a in the locative
case implies that a more accurate description of the structure of the
dative-instrumental plural would be given by the formula:
stem + (_m ? a)
that is, ? a was felt to be applicable only to basic endings terminating
in the consonant m. This is another reason for thinking the extension
of ? a to the genitive case impossible.
Apart from this basic implausibility, Belic's theory has many other
weaknesses. It is difficult by its means to explain the length of the
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
344 D- J- L- JOHNSON
new genitive plural in -a. If one accepts t
theory, then there existed such forms in
where 4- a was an optional appendage associa
speakers with the final ?a in the alterna
virtue of its origin ? a was short, atonic,
marker. How did it become long, capable of
dependent case marker ? Belie ascribes its le
the /-declension ending -I. But it is diffic
could be associated. -I was not only long; it
marker, never an appendage; it regularly
syllables and some disyllables (e.g. mod
bolesti).13 Clearly as long as ?a was an op
ated with the ? a of -am/-ama it could not
How did it lose this association and beco
marker ? Belie, it seems, was aware of the d
not state it explicitly. He is vague about t
the suggested developments. He does, how
portance to a hypothetical shift of stress in o
This was an analogy proceeding from the
paradigm.
Since zene
zenEm So also zena > zena
zenama
zenah
13 Another point of difference between the two structures was that the stem vowel pre?
ceding -I showed phonemic opposition between long and short vowel, e.g., stvari,
\kti, the gen. pl. of ^-declensions did not oppose length to shortness in the stem.
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 345
of many classes of noun, including some of the f-declen
norn. pl. radosti, gen. pl. radosti; dubr&ve, dubrav; sinovci,
jeleni, jSlen; glumci, glumac; novel, novae. On the other han
basically oxytonic feminine nouns retained stem stress in t
ative and accusative plural, e.g. strane, gore. Moreover, the d
of intensity between the stressed and pretonic syllable is at
in question likely to have been slight and the difference in t
of the stem vowel in such words as zena, as between the ge
other plural cases, was almost certainly a more marked
feature than the difference in stress position. It would be di
accept the idea of a purely analogical stress movement from
on to a case-marking termination, let alone on to an option
age at approximately the same period, when there was gest
beginning to develop, the general shift of stress by one sy
wards the beginning of the word, which became a characteri
of precisely those same dialects that produced the great inn
in the plural declensions including that of -a. It should be o
here that the advancement of the stress probably took plac
all from open final vowel on to a preceding long syllable, th
cisely in such forms as would have been created by Belic's s
shift (zena > zenii). However, even if we accept that by loss o
alternative the forms with +a had become the sole forms in the
genitive plural, and that therefore + a had become a case marking
ending (even perhaps bearing stress) it is still not easy to accept the
idea of the transference of length from -i, as length of termination was
not otherwise a regular mark of the genitive plural.
Accepting for the moment Belic's theory, the gen. pl. at this
period would have been represented by:
Masc. Neut. Fem. i Fem. a
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
346 D. J. L. JOHNSON
/-declension, but it was far from dominant in
genitive plural.
As we have seen, the only way for ? a to have los
as optional appendage and to have become an
marker was through the suppression of the altern
had partaken, that is, the loss of genitive forms
This must have preceded any eventual lengthening
therefore must have affected baritones and oxy
zen; lipa/lip > zena; lipa). If this had happened, a
of Karlgren's theory, any foundation for the subs
to other genders was destroyed. This, perhaps, is n
blow to Belic's theory, since, although he is of t
developed first with feminines, and then spread t
does not exclude the possibility of a simultaneou
ment among masculines:
Dat. pl. zubom/zuboma
therefore by analogy
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 347
The alternation -om/-em etc. was not originally a free variation
the choice of ending was phonologically determined by the ha
or softness of the preceding consonant. This situation was earl
scured in Serbo-Croat (a) by the transference of masculine i-st
the 0-declension (e.g. gost, gostem); (b) by the hardening of so
palatal consonants. Although in most dialects the factor of
logical determination was never entirely removed, there was n
theless created a substantial area of free variation. The frontier
between the area of determination and free variation was fixed
complex detailed rules that were subject to frequent shifting (
KHe3eM, khc3om; npnjaTejLeM, npnjaTejbOM; 3aKOHOM, 3aKOHeM.)15
Given this situation it is easy to understand that there would b
considerable resistance to the imposition of a further set of free va
tions upon those already existing. There is a good motivation for
dropping of-oma with the loss of the dual category. The situation
the instrumental was even more complex. The alternation -i/-mi w
not phonologically determined, since -mi derived from the i-
^-declensions. Insofar as there were any rules determining cho
they could only have been governed by lexical factors. Such rules
difficult to maintain so there was naturally much mixing (e
KpajibMH, BHHorpaAMH, Tpi>3MH; KpajiH, rpa^H, Tpb3H). The exten
to the instrumental of the ending -om, and the establishment wit
the continuation of old dual endings of a possible further three se
of free variation would place an intolerable burden on grammatica
structure:
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
348 D. J. L. JOHNSON
reduced any analogical force the alternation m
had.
The texts also show very few examples of the m
instrumental cases on the basis of the old plural
mixing of the two cases begins only on the basis
-im/-ima. In contrast, there are very numerous e
ing of the two cases in the feminine on the basi
-am/-ami. Though the interpretation of the
difficult, this seems to imply a resistance on the
such mixing. Since the motive force behind the
that the two cases were not distinguished in the
of masculines to it is another indication that the a
was never a dominant feature of the masculi
therefore, extremely unlikely that there cou
structure in the masculine paradigm that could h
accordance with Belic's theory. The chronolo
theory also gives cause for serious doubt. It invo
of a fairly lengthy sequence of changes.
One might well ask how many generations passed in the course of
this ten-point development ? Even if some of the changes may have
followed quite quickly upon the preceding change, it means that the
first stage must surely have been taking place very early in the de?
velopment of Serbo-Croat. It is inconceivable that the whole process
could be squeezed into the span of a single generation as is implied by
Belie (although the texts do not show 1. until the 15th century, he
assumes it to have taken place in the 13th-14th centuries, i.e. shortly
before or at the same time as -a is already appearing in texts). Neither
could one postulate, in order to help matters, that the development
of length was a later one. Early spellings of the gen. pl. as -bb show
17 Danicic has only one example before 16th century (op. cit. p. 96).
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 349
that at the time of its earliest appearance it was already lon
the first stage in the development was a very early one, it is
surprising that it should not find its way into the texts until
after the first appearance of -a.
The theory for the explanation of-a which I should like t
involves making an assumption about the innovating dial
not supported by unambiguous textual evidence. It is b
chronological assumption consisting of postulating for the d
question the earlier existence of a structure that later becam
elsewhere. It is a morphological explanation accounting
generation of -a by analogy. It does not, however, I trust, i
basic theoretical principles about how analogy works. Only
cal processes of the commonest type are involved. Neither is
assumption entirely without positive textual indications.
The assumption concerns the feminine z-declension. Th
paradigm inherited by Serbo-Croat at the beginning of i
was as follows:
N kosti
A kosti
G kosti
D kostem
I kostmi
L kosteh
18 Belie* derives cakavic kostan, kostah from kostim, kostih. He also postulates such
forms in his explanation of Montenegrin dialect gen. and loc. pl. -Ih: (Istorija, p. 80).
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
350 D. J. L. JOHNSON
Dat. -om/-em
Inst. -i/-mi21
Loc. -eh/-eh (-ih)
It will be seen that in the dative case the i-stem termination -em re?
inforced and was absorbed into the already existing alternation
-om/-em that had originated in the different terminations for the
hard and soft 0-stems. In this alternation the ending -em became
tightly associated with a characteristically masculine and neuter end?
ing. The dominance of this association is illustrated by the develop?
ment by analogy of a locative form in -oh in accordance with the
proportion:
Dat. -om/-em
Loc. -oh/-eh22
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 351
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
352 D. J. L. JOHNSON
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 353
The likelihood that oci exerted some influence on the dual z-declen?
sion forms is strong, but the probabilities are that any such influence
would have made itself felt before the loss of the dual as a separate
category, not afterwards. When the dual forms of the z-declension
were still alive the closely similar forms of oci could well have in?
fluenced them in the direction kostju>kostiju, kostma > kostima,
thus reinforcing the other factors leading to a generalisation of i vowel
in both the plural and dual of the dative and instrumental, and in the
locative plural. The resultant identity of form, ocima, kostima, could
then have allowed the words oci and usi which were usual in dual
forms, to play a role in the preservation of the dual form as
alternate plural, kostim/kostima. If there had been no such develop
ment, the link between the z-declension and oci which existed only i
the dual would have been broken with the loss of that category. Th
z-declension nouns would in all probability have lost their old du
forms, as indeed happened in other dialects25 and the forms ocima,
etc. would have become isolated as irregular plurals. As such, th
could hardly have subsequently caused the development of -im/-im
in the mass of i declension nouns. If the now plural paradigm of oc
had exerted such an unlikely influence, one wonders why it should
have been limited to the dative and instrumental, and not also have
been felt in the genitive ? The question also arises?why should such
a development have taken place in some areas but not in other
Belic's other suggestion that the z-declension's -im/-ima arose unde
the influence of the masculine paradigm is also extremely improbabl
The processes by which the old set of possible masculine termination
in dative and instrumental could have generated independentl
-im/-ima would have been complex. They would have required t
synchronic existence in a single case form of at least seven differe
endings in free or partially free variation (that is not counting the
endings -mi, -ma which could possibly be assumed to be lost an
which were not essential to the process). These endings would ha
embodied three totally different types of alternation:
1. -om/-oma
-em/-ema ? a
-im/-ima
2. -om/-em/-im vowel alternations
3. -i/-im ? m
There is, in my opinion, no likelihood of
tions really existing in speech. In any case
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
354 D- J- L- JOHNSON
himself elsewhere expressed doubt abo
process.26
Apart from the above general considerations pointing to the
existence of z-declension forms -im, -imi, -ih prior to the appearance
of the new endings -im, -ima, there are some textual indications.
26 Dvojina, p. 128.
27 R. Aitzetmueller and L. Sadnik, Handworterbuch zu den altkirchenslavischen Texten, The
Hague, 1955.
28 Rjecnik hrvatskoga iii srpskoga jezika JA2JJ, VI, p. 912.
29 Quoted from Danicic, Rjehiik iz knjizevnih starina srpskih, Belgrade, 1863, p. 447. He
gives an incorrect reference for this sentence.
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 355
The form of the words in agreement with kletvi makes it h
likely that it represents a Serbianisation of an Old Church
0-stem termination as in ?npbBbie kjictbh aflaMOBH (M
The form molitvih is evidence both of the existence in som
an z-declension ending -ih, and also of the extension to the
the vacillation between z- and 0-stems. The existence of molit
by side with molitvah is a strong indication of the existenc
forms as crkvih side by side with crkvah.
4. 6ojiapHMH. 3 times.
= (?m)i
30 Istorija . . ., p. ii8.
31 Leskien, Grammatik der serbo-kroatischen Sprache, Heidelberg, 1914, p. 438.
32 The Slavonian dialect forms such as klupija do not represent such an amalgam. They
are due to the superimposing of advancing -a on to an earlier state when the gen. pl. was
marked both by -0 and -i.
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
356 D. J. L. JOHNSON
masculine ending it could for a limited period in som
to mark the instrumental plural of the class of n
only plausible explanation of this fact is that -imi w
of the feminine z-declension. The evidence suggests
of mixing of the old masculine z-stems and 0-stems
proceeded faster in the singular than in the plur
spread to the plural, nouns of the type bolare would
better position to resist change than others, since th
stems in singular and plural, and had always been
adherence to two different paradigms. Retaining
firmly their specifically z-declension type in the pl
could well have been sporadically affected by ne
within the feminine z-declension.
The number of these textual indications may seem small, but if
one bears in mind that there are in all the texts up to the end of the
15th century no more than a handful of examples of the dative, in?
strumental and locative plurals of z-declension feminines (and some
of those are in texts of Church Slavonic style) then these few ex?
amples assume relatively greater significance.
If by the middle of the 13th century some dialect had developed
a plural paradigm where all the oblique cases were characterised by
the same initial vowel in the termination, and where the genitive was
marked by that vowel alone and long, then an extension of this
pattern to the other feminine declension would generate -a:
kosti zen + a
kostim zenam
kostimi zenami
kostih zenah
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
SERBO-CROATIAN GENITIVE PLURAL IN -a 357
Norn. sing. upbKbBb npbKBa
pp. 99, 191 pp. 197, 416
Acc. sing. itpbKbBb i^pbKBy
pp. io, 17 pp. 87, 417
Gen. sing. npbKBH upbKBe
pp. 13, 116 pp. io, 14
Inst. Sing. UpbKBlK) UpbKBOMb
p. 195 pp. 134, in
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms
358 D. J. L. JOHNSON
D. kletvim kletvam
I. kletvimi kletvami
L. kletvih kletvah
This content downloaded from 139.18.20.5 on Tue, 20 Sep 2016 10:05:08 UTC
All use subject to http://about.jstor.org/terms