Derivation versus inflection in three

inflecting languages

Stela Manova
Universität Wien

1. Introductionr

This paper deals with derivation and inflection from a typological perspec-
tive. Derivation and inflection are seen as constituting a continuum between
the poles of prototypical derivation and prototypical inflection (cf. Dressler
1989). As is well known from cognitive psychology, categories organized on
prototypes are easy to classify when prototypical instances are concerned, but
can be problematic in cases of non-prototypical ones. Therefore, in order to
contribute to the discussion on the demarcation of derivation and inflection,
I will focus on the morphological behavior of non-prototlpical derivation
and inflection. The categories I analyze, denominal diminutives, formation of
females from males and imperfectivization, are interesting examples of non-
prototypicality, since they allow for two types of expression, derivational and
inflectional. Data from three Slavic languages, Bulgarian, Russian and Serbo-
Croatian, all representing the inflecting(-fusional) type, serve as evidence for
the discussion.
A salient feature of the inflecting type noted by Skaliöka (1979) is the clear
distinction between derivational and inflectional suffixes. On the basis of this
typological characteristic and the principle of constructional iconicity pos-
tulating correspondence between addition of meaning and addition of form
(Dressler 2000:290),I assume the following morphotactic structure for a pro-
totypical Slavic word: eREF - Roor - DSUFF - (rr'r) - rsurr. Thematic markers
are given in brackets, since they occur as stem-forming elements only in verbal
morphology.
234 Stela Manova

I will speak of derivational realization of a category if its formal exponent is

in the derivational slot of the word (see 1a, 2a,3a) and of inflectional realization
if a category is formally expressed either by material in the inflectional slot of
the word or by a ru (see lb, 2b, 3b). The following examples from Bulgarian
illustrate derivational and inflectional realizations of denominal diminutives,
formation of females from males and imperfectivization:
Diminutives
( 1) a. rresc stol 'chair' -> DrM NEUr stol-t-e vs.
b. v'rsc kotel'caldron' -> DIM Nnur kotl-o-e

Gender (formation of females from males)

(2) a. v'ASC, car 'king, tsar' -+ FEM car-ic-a'queen' vs.

b. turtsc zabravan 'chuckle-head' -+ rnu zabravan-o-a
Aspect ( imperfectivization )

(3) a. plv raz-grom-j-a / raz-grom-i-i'llyou defeat'2 -+ IMeFV raz-grom-

jav-a-m I raz-grom-jav-a-i vs.
b. vrv ob-misl-j-a I ob-misl-i-i'Ilyou think over'-+ IMIFV ob-misl-o-ja-
m I ob-misl-o-ja-i

As is typical for inflecting languages, Bulgarian, Russian and Serbo-Croatian

express inflectional properties in more than one way, which results in morpho-
logical organization based on inflectional classes. I define inflectional class with
Aronoff (1994:64) as "a set of lexemes whose members each select the same set
of inflectional realizations". Due to lack of space, and since the categories I an-
alyze enter productive/major classes, I will deal mainly with productive/major
classes (see TablesI,2,3,4, 5, & 6).
When evaluated by the traditional criteria for distinguishing between
derivation and inflection (e.g. word-class change, obligatoriness, morpheme
order, etc.), the categories I am interested in exhibit nearly the same behavior.
However, when situated within the inflectional systems of Bulgarian, Russian
and Serbo-Croatian, formation of females from males and imperfectivizaton
differ from denominal diminutivization in regard to inflection class assign-
ment. In what follows I will try to show that when the traditional demarcation
criteria fail to determine the status of a category in an inflecting language, the
typologically adequate mechanism of inflection class assignment does not. I
claim that non-prototypical inflection is marked by membership in a partic-
ular inflectional class, whereas derivation cannot be identified inflectionally.
(Note that the connection of a particular inflectional class with a single deriva-
 Derivation versus inflection in three inrlecting languages 23i

tional suffix does not mean that derivation can be identified inflectionallv i ct-.
Manova 2003b). For example, in Russian, abstract nouns in -o,st' le.g. krn-.lrvl
'beautiful' --> krasivost"beauty') constitute the main part of class 4 ict-. Table
2). However, in this language there are also other suffixes for abstract nouns
which do not take the inflection of class 4, the inflection class of nouns in
-ost'(e.g. krasivyj 'beautiful' --> krasota 'beauty', class 2, Table 2). ln contrast,
the numerous Russian suffixes deriving females from males all select the same
inflection, namely class 2, Table 2.)
First, I will briefly present diminutivization, formation of females from
males and imperfectivization in each of the three languages.

2. Diminutivization

In Bulgarian,3 Russian and Serbo-Croatian, a vety productive rule N --+ N +

DrM suFF + rsuFF derives diminutives from nominal bases. The suffixes below
express diminutive meaning and can occupy the derivational slot of the noun
according to the rule mentioned above. Since suffixes in the derivational slot
determine the paradigmatic properties of the word, everl'where the inflection
is also written. If no inflectional suffix is given, it means that the respective
derivational suffix selects zero inflection, i.e. class i in the three languages, see
Tables 1, 2, & 3.

Bg.i -ec; -k-a; -ic-a; -tic-a; -itk-a; -c-e; -enc-e; -ic-e; -l-e; -t-e
x.: -ec, -ik, -ok/-ök/-ek, -tik, -ic-a, -k-a, -otk-a, -ik-o, -k-o, -c-o/-c-e, -ec-o
sc.: -it, -tit, -ak, -etak, -itak, -ic-a, -tic-a, -c-e, -anc-e, -aic-e, -enc-e, -eiö-e,
-eljak, -uljak

Diminutive suffixes take different inflection and as can be seen in Tables l, 2,

& 3, they can be found in nearly all declensional classes in Bulgarian, Russian
and Serbo-Croatian.
Due to the fact that Bulgarian diminutives (or at least some of them) have
their diachronic origin in inflection, in modern Bulgarian an unproductive rule
N -+ + rsuFF involving only inflectional change also diminutivizes nominal
N
bases. I will call diminutives formed according to this rule inflectional, in order
to distinguish them from diminutives derived by special diminutive suffixes.
Inflectional diminutives always take the inflection -e, which can be attached to
animate nouns as well as to nouns denoting objects, regardless of their gender:
236 Stela Manova

(4) wl'sckotel 'caldron'-+ DrM xlvr kotl-e

wr,o,scbik 'bull'-+ DrM NEUr biö-e
nnl.;t devojk-a 'maiden' -+ DrM Nnvr devojö-e
In Bulgarian linguistics, one can find two different explanations of the dia-
chronic origin of such diminutives: the first one sees them as continuants of
old vocative forms (cf. Mladenov 1929:224;Miröev 1963: 150), i.e. the modern
Bulgarian DrM otöe (<- otec 'father'), ouvr staröe (<- starec 'old man') should be
derived from the oBg. voc otröe,voc starnte. The second explanation connects
inflectional diminutives with the old church Slavic af- stems (cf. Georgiev
1985: 164) and seems to be more probable. It is motivated by the plural forms of
inflectional diminutives which, as is usual for nouns originating from the ong.

Table 1. Bulgarian nominal inflection (productive classes), cf. Manova & Dressler
(200 I )

la.(mono- 1b.(po1y- 2. 3. 4.
syllables) syllables)

SG -0 -n -o -e, Lws: -t,

-(j)u
SG DEF -ta fo -to

PL -ove -t -x -a -ta
PL DEF -te -te -te -te -te

Table 2. Russian nominal inflection (major classes), cf. Corbett (1991:36)

1. 2.

NOM -0 -a -o (-e)
ACC = NOM OT GEN -u -o
GEN -a 'v -a
DAT -u -e -u
INSTR -om -oj -om
LOC -e -e -e

PL

NOM -v -v -a -1
Äcc - NOM OT GEN
GEN -ov -ej
DAT -am -am -am -jom
INST -amt -ami -ami -jami
LOC -ax -CLX -ax -jax
 Derivation versus inflection in three inflecting languages 237

Täble 3. Serbo-Croatian nominal inflection (major classes)

1a.(mono- ib.(po1y- 2. l. 4.
syllables) syllables)

NOM -0 -0, -a -o/-ea -o

-i
rws:-o, -u, -e,
voc -e -e/-u or - Nott -o - NoM -i
ACC = NOM OT GEN = NOM OI GEN -u = NOM = NOM
GEN -a -a -a -i
DAT-LOC -u -u -i -u -i
INST -om -om -0m -om/-em -i

NOM-VOC -ovi -T -e -a -t
ACC -0ve -e -e -a -i
GEN -ova -a -a -a -i
DAT-LOC-INST -otima -ima -ama -ima -ima

pf-stems, has an -ef- amplification (er otieta, starierri). A second argument in

favour of the g/-stem origin of the diminutives in question is the fact that the
oeg. gr-declension was semantically related to the meaning of smallness, since
gt-stems originally consisted only of nouns denotingyoung of animals. Accord-
ing to Georgiev (1985), diminutives in -l-e (oru petl-e <- petel'cock, rooster')
and -t-e (owr vojnii-e <- vojnik 'soldier') where -l- and -ö- are parts of the re-
spective stems, served as bases for the development of the diminutive suffixes
-öe and -le, as in owr zäb-ie <- zäb'tooth' and DIM mäL-le <- mä2'man' .
Although both Russian and Serbo-Croatian are also descendants of ocs,
the diminutive suffix -e exists only in Serbo-Croatian where it is bound to the
semantic pattern 'young of animais', e.g. lisic-a'fox' -+ Iisiö-e, ptic-a'bird' -->
ptiö-e.s As for nouns such as vojniö-e'soldier voc' which are homophonous
with the respective Bulgarian diminutives (see above), in Serbo-Croatian, these
forms express vocative case, without any diminutive meaning at all. It seems
that the existence of the category of vocative which often inflects with the suf-
fix -e (see Table 3), has blocked the development of the diminutivization rule
with the same suffix. On the other hand, modern Russian has no vocative,
but even the semantics 'young of animals' is always expressed by suffixes in
the derivational slot, e.g. medveä-onok'bear-cub', tel-önok'calf', etc., the old
gt-stems being preserved in the plural (Nou pL medveä-ata, tel-jata).
To sum up, in Bulgarian, Russian and Serbo-Croatian, diminutives are usu-
ally derived according to a very productive rule requiring a diminutive suffix
in the derivational slot of the noun. The numerous diminutive suffixes take
48 Stela Manova

different inflections which allot diminutive formations to different inflectional

classes. In addition, Bulgarian and Serbo-croatian have inflectional diminu-
tives in -e (class 4 in Bulgarian, cf. Täble 1, and with a peculiar declension in
Serbo-Croatian (cf. Endnote 5).

3. Gender MALES -+ FEMALES

Although gender is not usually marked by a special suffix in the noun, the
category divides the lexicon into classes which trigger agreement. Therefore
slavic grammars consider gender a classificatory category for nouns. There are
only two instances when gender has a morphological exponent of its own in
the noun: 1) when females are derived from nouns denoting males, and 2) in
cases of adjective-to-noun syntactic conversion (cf. Spencer 2002). However,
syntactic conversion is irrelevant for us, since adjectives which undergo this
change preserve their adjectival inflection. Moreover gender in the adjective is
prototypical inflection (cf. Dressler 1989), i.e. always overtly marked, and thus
beyond the scope ofthis paper.
In Bulgarian, Russian and Serbo-Croatian, female nouns with nominal in-
flection are derived from males by the rule uesc N + csuFF + rsuFF -(j)a --,
pe^a N with the following gender suffixes:

sg.; -k-a, -in-ja; -kin-ja; -ic-a, -es-a, -is-a

u.: -k-a, -ix-a, -ic-a, -nic-a, -i-a, -n-a, -in-ja, -ux-e, -ess-a, -is-a
sc.'. -k-a, -ic-a, -inj-a, -kinj-a, -es-a

In addition to the forms derived with special gender suffixes, Bulgarian, Rus-
sian and serbo-croatian possess a set of nouns formed according to the un-
productive pattern MASC N + rsuFF -+ FEM N involving only affixation with the
inflection -a:
(5) vg. söprug 'husband' --> säprug-a'wife'
R. suprug'husband' --> suprug-a'wife'
sc. suprug'husband' --> suprug-a'wife'

In the oldest Slavic texts, in one and the same source, one finds used paral-
lel forms such as rab-a and rab-yni, both meaning 'slave-rEu, servant-rpu',
and derived from the masculine nounrabs 'slave, servant'(see SJS in the ref-
erences). Therefore, it is diftcult to establish which type of expression, that
derived with a special gender suffix or that formed by addition of the inflection
-a, is diachronically older.
 Derivation versus inflection in three inflecting languages 239

Nouns for females, whether formed with a special gender suf8x or derived
inflectionally, always take the inflection of the declension class of nouns termi-
nating in -a (class 2) in the three languages (see Täbles 1, 2, & 3), this rvithout
any exceptions. Note, however, that whereas nouns derived with a special gen-
der suffix are always feminine, nouns terminating in -a are not (e.g. Bg MASC
balta'falher' ), i.e. if gender is not expressed morphologicalll', it does not per-
fectly correspond to a particular inflectional class. For discussion on gender
and declension class assignment in Bulgarian see Manova & Dressler (2001),
for Russian Corbett (1991:34-43) and Fraser & Corbett (1995).

4. Aspect

It is well-known that most Slavic verbs can be organized in aspectual triples.

An aspectual triple consists of a basic verb, i.e. a primary imperfective
verb (rurrvl) -+ a perfective verb (prv) -+ a secondary imperfective verb
(rurrv2), where pvp and ruprv2 do not differ semantically. Each member of
the triple exhibits its own morphotactic structure (Manova 2002):

(6) rMPpvl = Roor + rM + rsuFF

Bg. stro- j- a 'I build'
PFV = PREF+ IMPFVI
ng. do- stroja 'I complete building'
rvrpr,v2 = pREF-Roor+ ASUFF+ TM+ ISUFF
eg. do- stro- jor- a- m 'I compiete building'
The assumption of triples implies that the primary verbal lexicon consists of
imperfective verbs only. However, there are also basic verbs (i.e. without pre-
fixes or aspectual suffixes) which are perfective (in Bulgarian, for example,
some 80 verbs,6 cf. Stojanov 1993:335). Thus like gender, the category of as-
pect is a classificatory one for basic verbs, since without being overtly marked
for aspect they are classified as either perfective or imperfective.
In regard to rnapEv2 verbs, there exists a fundamental distinction between
Bulgarian, on the one hand, and Russian and Serbo-Croatian, on the other:
whereas nearly all ruppvl verbs in Bulgarian have prv and Ivrppv2 forms,
in Russian and Serbo-Croatian, if the perfectivizing prefix has only aspectual
meaning and there is no significant semantic difference between rvrprvl and
rrv, imperfectivization is blocked and the basic trrprvl form is used instead of
rlrpEv2. Consider:
24o Stela Manova

(7) Bg. piio'Iwrite', piiei --> napiia 'I write down', napiöei --> napisvam, napis-
vai
R.pisat"to write' --> napisat''to write down' -> +napisyvat'
sc. pisctti 'to write' --> napisati 'to write down' -+ +napisivati

However, if perfectivization
involves a significant semantic change, all three
forms exist in the three languages. For example:
(8) sg. piia'lwrite', püei --> podpiia'I sign', podpiiei -+ podpisvam, podpisvai
R.pisat"to write' --> podpisat''to sign' --> podpisyvat'
sc. pisati 'to write' --> potpisati 'to sign' potpisivati
-
Bulgarian, Russian and Serbo-Croatian express imperfectivization either by a
suffix in the derivational slot or inflectionally by a ru only. However deriva-
tional and inflectional realizations of the category of aspect differ in terms
of productivity and, as can be seen from the next examples, productive rules
always require an aspectual suffix in the derivational slot of the verb.
Bulgarian (cf. Table 4, class 3):
(9) a. Productive suffixes: -u-a-, -(j)av-a- (both class 3)
ppv kaäa'I say', kahei -+ rMpFV kaz-v-a-m, kaz-y-a-i
ppv izora 'I plow', izoreö --> ruprv2 izor-av-a-m, izor-av-a-i
b. Unproductive suffixes: -(j)a-, -uv-a- (both class 3)
vw izgovorja 'I articulate', izgovorii -+ ruprv2 izgovar-ja-m, izgovar-
ja-i
vw kupja 'I buyi kupii --> rwerv kup-uv-a-m, kup-uv-a-i

Täble 4. Bulgarian verbal inflection (major classes), based on Andrejöin ( 1978)

l. z.

I sc pnEs -( j)a -(j)-o -(j)a-m

l-v-a-m,
-(j)at-a-m,
-uv-a-m)
2 sc pnrs -e-i -i-i -a-i
3 sc pnrs -e -i -u
l.l. 1.2. 2.r. 2.2.
1 scAorist -o-x -(j)a-x -i-x -(j)o-" -(j)a-x
I sc Imperfect -(j)a-x -e-x -e-x -(j)a-x -(j)a-x
 Derivation versus inflection in three inflecting languages 241

Russian (cf. Table 5, class 1):

( 10) a. Productive suffixes: -yv-a-; -iv-a (both class 1)

pEv perepisat' 'to copy' --> ]o/rrv2 perepis-W-a-t'
pFv oplakat"to mourn' -+ ruprv2 oplak-iv-a-t'
b. Unproductive suffixes: -v-a-; -(j)a- (both class 1)
vrv sogret"to warm' -+ rrrappv2 sogre-v-a-t'
vnv ob'jasnit', to explain -+ ruprv2 ob'jasn-ja-t'

Serbo-Croatian (cf. Täble 6, classes 1 & 2):

(11) a. Productive suffixes: -(j)av-a-/ I sc pREs -(j)av-a-m (class l) and
-(j)iv-a-/ I sc pnns -uj-e-m (class 2)
pev odöbriti 'to approve, to permit' -+ ruppv2 odobr-äv-a-ti
ppv zavöriti 'to weld, to solder' -> rnprv2 zavar-iv-a-ti
b. Unproductive suffixes: -v-a-, -(j)a- (both class l)
pev döbiti 'to receive, to obtain -+ ruprv2 dobi-v-a-ti
prv zäviti 'to wrap' -+ rvrprv2 zavi-ja-ti

Table 5. Russian verbal inflection (productive classes), based on Isaöenko (1982)

1. 2.

Infinitive -a-t' -ov-a-t', -e-t' -nu-t' -t-t'

[-yv-a-t', -ev-a-t'
-iv-a-t'l
I sc pnss -aj- u -uj- u -ej- u -n- u -(j)u
2 sc pnns -aje-i' -uj-e-i' -eje-i' -ne-i' -i-i'
3 sc pnr,s -ajel -uj-e-t -eje-t -ne-t -i-t

Table 6. Serbo-Croatian verbal inflection (productive classes), based on Dressler et al.

( 1ee6)

L 2. 3. 4.

Infinitive -a-tt -ov-a-fi, -i-ti -nu-tl

[-(j)av-a+i] -ev-a-ti,
-(j)iv-a-ti
I sc pnEs -a-m -uj-e-m -t-m -ne-m
2 sc pnns -a-i -uj-e-i -t-< -ne-s
3 sc pnns -cI -uj e -ne
Imperative -a-j -uj -1 -ni
242 StelaManova

Each of the three languages possesses a restricted number of imperfectivizing

ASUFFS, two productive and two unproductive, but only three of them use the
derivational slot of the verb. In addition, ASUFFS always combine with the rrra
-a- only, which makes them similar to conjugation class markers, i.e. esurps
can be seen as complex rrras of the type -(V)va-. The same holds for the suffixes
-ova-/-eva- in Russian and Serbo-Croatian (see class 2 in Tables 5 & 6).
All rnappv2 verbs in Bulgarian and Russian are marked by membership in
a particular conjugation class (class 3 in Bulgarian and class 1 in Russian, see
Tables 4 & 5), whereas Serbo-Croatian rnrppv2 verbs go into two conjugation
classes, class 1 and class 2 respectively (Table 6), and this when derived with
productive suffixes (11a). However, class I and class 2 are phonologically com-
plementary and the most salient feature of the Serbo-Croatian vowel system,
accentuation, governs the distribution of the two productive aspectual suffixes
-(j)av-a- (class 1) and -(j)iv-a- (class 2). According to Babiö (1991), the exact
rule is: if the ppv verb has a long stressed vowel, then secondary imperfec-
tivization uses the suffix -(j)iv-a-, e.B.vnv isklluö-i-tl 'to turn off' -+ ruprv2
iskljuö-iv-a-tl; if the ppv has a short stressedvowel, its respective ruprv2 verb
is derived with the suffix -(j)ava-, e.g. pFV ogräniö-i-ti'to limit' -+ rnappv2
ograniö-äv-a-ti.
Both suffixes -(j)av-a- and -(j)iv-a- differ in their degree of productivity
and verbs derived with the suffix -ov-a- (class 2) have rnrppv2 forms with -lv-a-
(class 2), though with the inflection suffixes of class l, i.e. -iv-a-ti / -iv-a-m
instead of -iv-a-ti /-uj-e-m,7 e.g.: pFV dar-öv-q-ti 'to present, donate' (<- dar
'present, gift'), enns I sc dar-uj-e-m --> rMpF.v & prv dar-iy-a-ti, pRns I sc
dar-iv-a-m. This shift from verb class 2 to verb class I speaks for greater pro-
ductivity of class I in comparison to class 2 (Dressler 1997) and assigns to class
I a status of the default class for rnprv2. It should also be noted that due to
regional and dialectal variants, the Serbo-Croatian verbal system sometimes
allows for doublets derived with both -(j)av-a- and -(j)iv-a- from one and the
same base.
In Russian up to the 19th century (cf. Isaöenko 1982:227-229), besides the
productive suffi,xes -yv-a-/-iv-a- (class 1 in Table 5), the suffixes -oy-a-/ -ey-
a- (i.e. class 2) were also imperfectivizing. Afterwards -ov-a- and -ey-a- werc
restricted to nominal and adjectival bases only, and thus specializedfor deriva-
tion (cf. also the Russian Academy Grammar 1980:337ff.). This specialization
of the suffixes collects all Russian ruppv2 verbs in class 1 (cf. Table 5), and
allots noun-to-verb and adjective-to-verb derivations to more than one verb
class: soyet'advice, council' --> sovet-ov-a-t''to advise'; ipric'syringe' --> ipric-
ev-a-t'; 'to syringe'; pjatn-o'spot, stain' --> pjatn-a-t' 'to spot, to stain'; kamen'
 Derivation versus inflection in three inflecting languages 243

'stone' --> kamen-e-f' 'to stone'; sekretar"secretary' --> sekretar-i-t' 'to work as
a secretary'; pust-oj '.-pty' --> pust-ov-a-t' 'to be empty (for space)'; slab-yj
'weak, feeble' -+ slqb-e-t''to lose weight; to weaken'; iist-yj 'clean' --> üst-i-t'
'to clean'.
The same holds for verbal derivations in Serbo-Croatian and Bulgarian:
sc. säv(j)et 'advice, council' --> sfu(j)et-ov-a-ti'to advise'; söstr-a'sister' -->
söstr-i-ti'to accept as a sister'; vösl-o 'oar, scull, paddle' --> vösl-a-ti'to row,
paddle, scull':' bijöl'white' --> bijöl-i-ti, bijöl-je-ti'to whiten, bleach ; Bg. sävet
'advice, council' --> sävet-y-a-m'I advise', sävet-v-a-i; petn-o 'spot, stain' -->
petn-j-a'I spot, slain', petn-i-i; bjal'white' -+ bele-ja (se)'I turnlbecome
white', bele-e-i (se).
To sum up: Bulgarian and Russian always connect tuprv2 verbs with a
particular inflectional class. Serbo-Croatian rvrppv2 verbs are distributed into
two phonologically complementary classes, of which the class of -(j)av-a- (class
1, Table 6) is the more productive one, expressing imperfectivization in general.
By contrast, the output ofderivations to verb goes into differentverb classes.
ruprv2 verbs in the three languages are always marked by the ru -a-, which
is the default marker for imperfectivity, i.e. there are only very few verbs with
the ru -a- which are not imperfective. Note, however, that in Russian and
Serbo-Croatian, if a verb has the rM -A-, this does not automatically assign
it to class 1, and such a verb could belong to class 2 or to an unproductive verb
class in both languages.

5, Derivation versus inflection

Now I will briefly evaluate the three categories, diminutives, formation of

females from males and imperfectivization, according to some of the tradi-
tional criteria for distinguishing between inflection and derivation (surveys in
Dressler 1989; Plank 1994; Booij 2000):

L Change of word class: Derivation, unlike inflection, may be word-class-

changing.
The three categories I analyze are word-class-preserving by definition, i.e.
if we diminurivize a noun, the result is a noun, nouns denoting females are
derived from nouns denoting males, and since aspect is a verbal category, im-
perfectivization always involves only verbs. However, this does not mean that
the suffixes expressing the three categories cannot be word-class-changing. The
most frequent case of word class change involves addition of aspectual suf-
244 Stela Manova

fixes to nominal and adjectival bases from which verbs are derived, e.g. the
above-cited sg. sävet 'advice, council' --> sävet-y-a-n'I advise', savet-v-a-i (cf.
Section 4). In Bulgarian, the unproductive imperfectivizing suffix -uv-a isvery
productive for derivation of verbs from nominal bases, e.g. sän'dream' -->
sän-uy-a-ffi 'I dream', prorok 'prophet' --> prorok-uv-a-m'Iprophetl etc. How-
ever, the specialization of the oid aspectual suffixes n. & sc. -ov-a-l -ev-a only
for derivation of verbs from nominal and adjectival bases shows that in Rus-
sian and Serbo-Croatian, aspectual suf{ixes tend to be word-class-preserving.
Thus paradoxically, aspectual suffixes are word-class-changing only in Bulgar-
ian where the category of imperfectivization has a full set of forms (i.e. seems
to be inflectional, cf. Section 4, ex. 6). It should also be mentioned that the ad-
dition of the rM -a- which is an unproductive realization of imperfectivization
can also be word-class-changing (cf. the verbalizations at the end ofSection 4).
Gender suffixes, when added to adjectives or verbs, derive common gen-
der nouns. Consider: Bg. MASC & rEl. pijan-ic-a (n. nrasc & rnrr p'janica)
'drunkard' formed from the adjective pijan (n. p'janyj)'drunk' by addition of a
gender suffix and without a masculine counterpart8 as well as the derivation ng.
bäbrja'I chatter, babble', bäbrii --> nasc & pnrtbäbr-ica 'babbler' (there is no
masculine noun*bäbrik, cf. rnu üstnica <- MASC öistnik 'fastidious person').
Such common gender nouns can refer to males and females but have feminine
morphology (class 2 in the three languages).
As for diminutives, unlike languages such as German where diminutive
suffixes can change the word class of the base (e.g. lieb 'dear' -+ Liebchen
'the dear-orr,r', cf. Dressler & Merlini Barbaresi 1994:I03f .), in the three Slavic
languages I discuss diminutivization is usually word-class-preserving. Word-
class-changing diminutivization is extremely rare and not mentioned in the
literature. Yet, an example could be sc. ADI zelen'green' --+ N zelen-if 'little
green tree' (cf. Babiö 1991: 190).
Thus, according to the criterion ofword class change, the three categories
tend to be derivation, diminutivization being the least derivational one.

2. obligatoriness 6 syntactic relevance: Inflection is obligatory, whereas deriva-

tion is optionai. However, since s1'ntactically required agreement is obliga-
tory, obligatoriness is connected with the criterion of syntactic relevance (cf.
Dressler 1989:6) according to which, inflection is relevant to syntax (Anderson
1982:587). Thus, due to the syntactic agreement between subject and predica-
tive positions, feminine gender may be obligatory in sentences with a female
subject in Bulgarian, Russian and serbo-croatian. Gender agreement is oblig-
atory with inhabitatives, e.g. Bg. Tja e irlandka 'she is an Irishwoman' and not
 Derivation versus inflection in three inflecting languages 245

*Tja e irlandec'She is an Irishman'. This even in the plural rvhere fe 'ther/ is

gender neutral, i.e. if we refer to a group of female persons, we must sat' Ie -srr
irlondki 'They are Irishwomen' (and not *Te sa irlandcl 'They are lrishmen'),
although v'l^sc irlandci'Irishmen' which expresses the meaning'inhabitants of
Ireland' in general is the form expected with a gender neutral subject. Ciearlr-,
the gender neutral subject te'they' cannot require gender. Therefore, it is not
sure that the obligatory feminine gender with inhabitatives is a case of svn-
tactic agreement (cf. also Booij 2002:82). Gender agreement is optional rvith
agent nouns, i.e. both Tja e uütelka'She is a teacher-peu' and Tja e utitel'She
is a teacher-uesc' are possible, but the latter is more respectful (on gender
agreement, see also Dressler & Doleschal 1991).
Aspect can also be required by certain verbs, e.g. regardless oftheir aspect
phase verbs Ilke to begin, to continue and to end combine only with imperfec-
tives, as can be seen from the following examples from Bulgarian:

(12) da zapotna-erv da podpisvam-rMpFv' I begin to sign'

zapoövam-trrvw da podpisyam-rMpFv' I begin to sign'

but not
*
da zapotna-vrv da podpiia-prv ' I begin to sign
*zapoövam-nnrvrv da podpila-pEv ' I begin to sign'

Obligatoriness is here due to the nature of the perfective aspect which views an
activity as a whole and is thus incompatible with the focus on the start/end or
the development, whereas the imperfective is. Since in the above examples, the
imperfective verb is not required by the aspect of the introductory verb, but by
its semantics, (12) does not represent syntactic agreement.
As for diminutives, the use of a diminutive noun requires further diminu-
tivization. Therefore from the Bulgarian examples below, (13) and (1a)
are well-formed sentences, (15) is strange but acceptable, whereas (16) is
impossible.

(13) Viä mu räö-itka-ta, pröst-te-ta-ta j to tolkova

Look his hand-prrr,r-ner, finger-oru-pI,-DEF its are-3pr. so
bel-i
white-pr
'Look at his hand-oilr, its fingers-orrrr are so white'
246 Stela Manova

(14) ViL mu röö-itka-ta, präst-öe-ta-ta j to tolkoua

Look his hand-orna-ppr, finger-orrra-pl-DEF its are-3pr so
bel-itk-i
white-oru-pr
'Look at his hand-orrr.t, its fingers-olM are so white-oru'

0, ?ViZ mu rät-iöka-ta, präst-i-te j ta tolkova

Look his hand-orlr-onn, finger-rr-onr its are-3pr so
bel-itk-i
white-orrr,r-pr
'Look at his hand-pru, its fingers are so white-oru'
(16) *Vi| mu rät-itka-ta, präst-i-te j to tolkova bel-i
Look his hand-orna-par, finger-er-pnp its are-3pr so white-pr
'Look at his hand-oru, its fingers are so white'

These examples resemble to some extent the use of honorifics, i.e. if one speaks
in a given way, (s)he should keep it up. Thus, the obligatory diminutivization
in the above sentences does not look like agreement required by syntax.
To sum up: in regard to obligatoriness, the three categories behave like in-
flection, but since they do not participate in syntactic agreement, according
to the criterion of s;,ntactic relevance, they seem to be derivation (or at least
inherent inflection, cf. Booij 2000).

3. Productivity'.Inflectron is typically more productive than derivation.

Of the three categories discussed, diminutives show the greatest degree
of productivity. It is possible to derive a diminutive (usually even more than
one) from nearly every noun, whereas imperfectivization can, as already dis-
cussed (cf. Section 4, ex.7), be lexically blocked in Russian and Serbo-Croatian.
Moreover, verbs of foreign origin are often biaspectual, e.g. sg. organiziram'I
organize'l n. organizovat'/ sc. organizirati.The same should be said for the pro-
ductivity of the rule deriving females from males - loanwords denoting males
(e.g. professor, director) are usually double gender nouns. However, if loanwords
are often paired for gender in colloquial sty'e (rg. direktor-k-a / x. direktor-i-a
/ sc. direktor-ic-a),instances such as Bg. MASC strelec'marksman'-+ rerra o and
MASC letec'pilot' -+ FEM oe undoubtedly speak for the restricted productivity
of the females-from-males formation, in Bulgarian at least.
According to this criterion, diminutivization seems to be inflection,
whereas formation of females from males and imperfectivization are less in-
flectional. Yet, diminutivization should be a peculiar type of inflection, since it
usually allows for more than one derivation from the same base (e.g. ug. kniga
'book' -+ DrM kniL-ka & otw kni1-le) and may apply recursively (e.g. eg. kniä-le
 Derivation versus inflection in three inflecting languages 247

'book-orrt' --+ kniä-le-nce'book-oru-Drivr' --> kni2-le-nce-nce'book-oru-oru-

orvr).
4. Order of morphemes: Inflection is more peripheral than derir.ation.
If denominal diminutives, formation of females from males and imperfec-
tivization are expressed in the derivationai slot, their suffixes precede the inflec-
tion. When the three categories are reaiized inflectionalll', gender and diminu-
tive suffixes are word-final, whereas rlrs u'hich are the inflectional equivaient
of imperfectivization are between derivational and inflectional suffixes (cf. the
prototypical form ofthe Slavic rvord in Section t and ex. (3a, b)).
According to this criterion, the three categories (except Russian diminu-
tives which should be derivation, cf. Section 2) are neither inflection nor
derivation.
5. Overt analogue criterion: All three categories, except Russian diminutives,
can be expressed either in the derivational or in the inflectional slot of the word,
i.e. when realized inflectionally the three categories have overt analogues (see
examples (1), (2), & (3)) which, according to Booij (2000), distinguish deriva-
tion from inflection. In fact, an overt analogue of a morphological change is a
traditional prooffor conversion I zero derivation (cf. Sanders 1988) and since
it has always been assumed that conversion operates only in derivation, it is ex-
pected that the overt analogue criterion should distinguish between derivation
and inflection. Thus, if a derivational rule has an overt analogue, it is a case of
conversion, however if an inflectional rule has an overt analogue, the morpho-
logical theory even lacks a label for it. The problem was discussed in detail in
Manova (2003a).
6. Change of inJlectional class. In Scalise (1984:110), one reads that deriva-
tional rules can change the declensional class of nouns and the conjugational
class of verbs, whereas inflectional rules cannot. According to this criterion, all
three categories represent derivation, since the rule deriving females is always
inflection-class-changing, and diminutivization and imperfectivization can be
inflection-class-changing as well as inflection-class-preserving.
According to these demarcation criteria, it seems that there is no strik-
ing difference between denominal diminutives, derivation of females from
males and imperfectivization: the three categories are word-class-preserving
by rule, although their suffixes can be word-class-changing; without being
syntactically relevant, the three categories can be obligatory in certain con-
texts; they obey the expected morpheme order; have overt analogues; and can
be inflection-class-changing. (A11 these, of course, with some nuances in the
three languages). Only in respect to productivity, diminutivization appears
248 Stela Manova

to be more productive (inflectional) than females-from-males formation and

imperfectivization.
However, when situated within the inflection systems of Bulgarian, Russian
and Serbo-Croatian, the three categories in question clearly differ in regard to
inflection class assignment. Therefore, I suggest inflection class assignment as
a criterion for demarcation between derivation and inflection. I formulate the
criterion in the following way:

7 . Inflection class assignment; If a category can be identified either with a par-

ticular inflectional class or with complementary classes, it is expressed inflec-
tionally and therefore represents (non-prototypical) inflection. If a category
cannot be identified inflectionally, i.e. its output belongs to different (unre-
lated) inflectional classes, it is derivation.
This criterion assigns to nominal diminutives the status of derivation and
to formation of females from males and imperfectivization the status of in-
flection. (Note that this holds only for Bulgarian, Russian and Serbo-Croatian
and does not exclude that in other languages, the same categories can behave
differently according to inflection class assignment).
Of course, now the question about the reliability of the criterion of inflec-
tion class assignment arises. And really, why should a single criterion be enough
evidence for the derivational or inflectional status of a category? There are at
least two reasons:
First, it is not very probable that the most salient feature of an inflecting
language, its inflection class organization, should be irrelevant for the differen-
tiation between derivation and inflection.
Second, only on the basis of their inflection class organization can Bulgar-
ian, Russian and Serbo-Croatian successfully distinguish between the following
four types of inflectional categories: i) inflection such as case or definiteness
identified as a cell of the paradigm of an inflectional class; 2) categories such
as nominal number which constitute a part (i.e. involve more than one cell) of
the paradigm of an inflectional class (e.g. singular and plural subparadigms);
3) categories such as gender and aspect which can, when expressed by an overt
suffix, be identified with a particular inflectional class or with complementary
classes, e.g. gender (cf. Stump 1998) as formation of females from males and
aspect in terms of imperfectivization.l0 (Note, however, that if this type of in-
flection is not overtly expressed, it does not perfectly corfespond to a particular
inflectional class. For example, although masculine nouns usually belong to in-
flectional class I in the three languages, they can be found in other inflectional
classes as well. The same holds for feminine and neuter nouns.); 4) categories
 Derivation versus inflection in three inflectrng languages 249

such as the above-discussed diminutives which cannot be identified inf'Iection-

ally, since their output enters more than one inflectional c1ass.
Clearly, the inflection class system of an inflecting language cannot distin-
guish between different types of derivation.

6. Conclusion

In Bulgarian, Russian and Serbo-Croatian, non-prototypical categories such as

denominal diminutivization, formation of females from males and imperfec-
tivization allow for two types of forms, derivational and inflectional. Of these,
in harmony with the semiotic principle of constructional iconicity (correspon-
dence between addition of meaning and addition of form), only derivational
realizations are productive. This preference for derivational forms over inflec-
tional ones is category-specifically realized: diminutivization has developed a
rich set of derivational suffixes and its output has thus dispersed within the
inflectional systems of the three languages; formation of females from males
exhibits numerous gender suffixes, but has been left connected with a single
noun class only; imperfectivization uses a few ASUFFS and is also connected
with a particular inflectional class, the two classes of Serbo-Croatian rupEv2
verbs being phonologically complementary. This makes denominal diminu-
tives unidentifiable inflectionally and according to the criterion of inflection
class assignment, defines them as derivation, whereas formation of females
from males and imperfectivization, since identifiable with a particular inflec-
tional class (or complementary classes), represent inflection.
In addition to the typologically-adequate connection of non-prototypical
inflection with particular inflectional classes, for distinguishing between deriva-
tion and inflection, Bulgarian, Russian and Serbo-Croatian use language-
specific strategies: in order to underline the inflectional status of imperfec-
tivization, Bulgarian has developed a full set of aspectual forms, Russian and
Serbo-Croatian have reduced the number of their aspectual suffixes special-
izing -ova-l-eva- only for derivation. Although Serbo-Croatian Iuppv2 verbs
have a less clear inflectional status than Bulgarian and Russian ones, the dis-
tribution of rrrppv2 verbs into conjugation classes in Serbo-Croatian is not
chaotic, but governed by a language-specific feature - the accentuation of
the vowels.
Thus, on the one hand, non-prototypical derivation and inflection in Bul-
garian, Russian and Serbo-Croatian with their two types of forms, derivational
and inflectional, confirm the assumption of a derivation-inflection contin-
25o Stela Manova

uum. on the other hand, however, the three non-prototypical categories I

analyzed undoubtedly show that inflecting languages tend to make distinction
between derivation and inflection, and this even in cases where the traditional
demarcation criteria do not.

Acknowledgements

This paper is a revised version of a talk given at the llth International Mor-
phology Meeting, Vienna, February 2004. I would like to thank those present
in vienna, and particularly ursula Doleschal and Andrew spencer for helpful
comments. I am also very grateful to: Johannes Reinhart for useful suggestions
on an earlier draft; Matthew Baerman and Greville corbett for comments on
obligatoriness and agreement; Heiner Eichner and Heinz Miklas for discussion
on the diachronic development of diminutives, gender and aspect in slavic;
and to wolfgang Dressler for his detailed comments on the final version. Any
errors are mine. This research was supported by the Anniversary Fund (Ju-
biläumsfonds) of the Austrian National Bank (grant p-r0366); the support is
gratefully acknowledged.

Notes

r' Abbreviations: ACc - accusative, en1 - adjective, esurr - aspectual suf6x, eg. Buigarian,
-
oer - dative, ulr - diminutive, oru supr - diminutive suflrx, DSUFF - derivationai suffix,
psÄa * feminine, cEN
- genitive, csuFF - gender suflix, rlrprv 1 - (primary) imperfective,
rvppv2 - secondary imperfective, rNSTR - instrumental, rsuFF * inflectional suffix, roc
locative, rw - loanword, MASC - masculine, N - noun, NEUT neuter, NoM nominative,
-
- -
oeg. - Old Bulgarian, ocs - Old Church Slavic, ppv perfective, rr_ plural, rnnr preflx,
- - -
PREs - present, n. - Russian, sc. - serbo-croatian, sc * singular, ru - thematic marker, voc
- vocative,
z. since Bulgarian has no infinitive, all Bulgarian verbs are given in their basic form, 1 sc
rnrs, and in 2 sc pnns which exhibits all possible inflectional suffixes.

3. In contrast to ali other Siavic languages, Bulgarian nominal morphology has lost the cat-
egory ofcase and developed the category ofdefiniteness. Thus in a Bulgarian noln, number
and definiteness are always overtly marked, definiteness being expressed by suffrxes, whereas
in Russian and Serbo-Croatian nouns, the categories of case and number are always overtly
signalled (see Tables 1,2, & 3).
4. Diminutives derived with the suffixes -ce, -ance, -ence, -a!ce, and -eice either have the
inflection of class 3 or take the amplification -f, as is usual for a minor inflection class, cf.
Bariö et aL. (1995:144).
 Derivation versus inflection in three inflecting languages 2Sr

5. Such nouns, as descendants of ocs et-stems, have -er- amplification rrith the endings of
class 3 in the oblique singular cases, their No\r pL is iormed either rrith the collectir.e suffjx
-ad (class 4 inflection in the oblique cases) or n.ith -r'll-ci (inilection ofclass 1 r.
6. Such verbs have only the last two forms of the aspectual triple, e.g.: Bg. pF\- xt,ärl-.ia'I
throw', ntärl-i-i -+ ruprv xvärl-ja-m, rvärl- ja-i, R. prv bro-.-l-t' 'to throrr''._ rrrprr. lrro,.-n-
t'; sc. ppv bäc-i-ti'to throw'--+ wrppv bdc-a-ti.
7. Some regional variants allow both -iy-a-ti / -iy-a-nt and -ir-a-ti /-ui-e-nt.
8. In Serbo-Croatian, there also exists a common gender noun ulsc & FE\t pi.iailica 'drunk-
ard'. However, in this language in contrast to Bulgarian and Russian, tiom the adjectir-e piinn
'drunk', one can derive the masculine noun pijnnac'drunkard'.

9, Cf. u,r.sc pluvec'swimmer' +

rcu pluvkinja; and MASC kosmonayl 'cosmonaut' + FE\I
kosmonavtka (colloquial) which is semantically related to letec'pt\ot'.
ro. Note that what holds for imperfectivization is also correct for perfectivization with the
thematic marker -n- / -nrr- (the onlv perfectir.izing suffix in the three languages). As can be
seeninTäbles5and6,perfective -n-/-nu-verbsevenconstituteverbclassesoftheirownin
Russian and Serbo-Croatian.

References

Anderson, Stephen R. ( 1982). "Where's morphology?". Linguistic Inquiry, lj,571472.

Andrejöin, Ljubomir (1978). Osnovna bdlgarska gramatika IA basic grammar of Bulgarian].
Sofija: Nauka i izkustvo.
Aronoff, Mark (1994). Morphology by l*elf: Stems and Inllectional Classes. (= Linguistic
Inquiry Monographs, 22.) Cambridge, Mass.: MIT Press.
Babiö, Stjepan (1991). Tttorba rijeöi u hrvatskom knji2et'nom jeziku: nacrt za gramatiku lAn
outline of Croatian grammar: word formation).2 izd. Zagreb: Djela Hrvatske akademije
znanosti i umjetnosti.
Bariö, Eugenija, Mijo Lonöari6, Dragica Maliö, Siavko Paveöiö, Mirko Peti, Vesna Zeöeviö, &
Marija Znika (1995). Hrvatska gramatika lCroatian grammarl. Zagreb: Skolska knjiga.
Booij, Geert (2000). "lnflection and derivation'l Booij et al. (Eds.) 2000. 360-369.
Booij, Geert (2002). The Morphology of Dutch. Oxford University Press.
Booij, Geert, Christian Lehmann & ioachim Mugdan (Eds.). (2000) . Morphologie /
Morphology. An International Handbook on Inflection and Word-Formation. Yol. l.
Berlin: Walter de Gruyter.
Corbett, Greville G. (1991). Gender. Cambridge: Cambridge University Press.
Dressler, Wolfgang U. (1989). "Prototypical differences between inflection and derivation".
Zeißchrift für Phonetik, Sprachwissenschafi und Kommunikationsforschung, 42,3-10.
(1997). "On Productivity and Potentiality in Inflectional Morphology'l CIASNET
Working Pap er s, 7, 2-22.
(2000). "Naturalness'l Booij et al. (Eds.) 2000.288-296.
& Ursula Doleschal ( 1991 ). "Gender agreement via derivational morphology". Aca
Linguistica Hungarica, Vol. a0 ( 1-2), I I 5-l 37 . ( I 990-1 99 I ).
z5z Stela Manova

Katarzlna
Dziubalska-Kolaczyk, & Antigona Kariöiö ( 1996). 'A contrastive analysis
- - ofverbal
-, inflection classes in Polish and Serbo-Croatian". Suyremena lingvistika, 411 42,
127-t38.
& Lavinia Merlini (1994). Morphopragmatics: Diminutives and Intensifers
- - in- ltalian, German, andBarbaresi
Other Languages. Berlin: Mouton.
Fraser, Norman M., & Greville G. Corbett (1995). "Gender, Animacy, and Declension Class
Assignment: A Unified Account for Russian". In Geert Booij & Jaap van Marle (Eds.),
Yearbook of Morphology ß9a @p.123-150). Dordrecht: Kluwer.
Georgiev, Vladimir (1985). "Preosmisieni padelni formi fReanalyzed case forms]'1 In
Vladimir Georgiev (Ed.), Problemi nabäIgarskija ezik (pp. 164-t68). Sofija: Izdateistvo
na BAN.
Isaöenko, Aleksandrv. (1982). Die russische sprache der Gegenwart. Formenlehre.4. Auflage.
München: Max Hueber.
Manova, Stela (2002). "Between inflection and derivation: On morphotactic expression
of aspect and gender in Bulgarian, Russian and Serbo-Croattan". Wiener Slayßtisches
Jahrbuch, 48, 203-217.
(2003a). Conversion and Subtraction in Bulgarian, Russian and Serbo-Croatian.
Ph.D. Dissertation, University of Vienna.
(2003b). "An input-oriented approach to inflection-class assignment illustrated
with Bulgarian nominal inflection'l Wiener Slavistisches Jahrbuch, 49, 103-118.
& Wolfgang U. Dressler (2001). "Gender and declensional class in Bulgarian'l
Wiener Linguistische Gazette, 67-69, 45-81.
Miröev, Kiril ( 1963). Istoriteska gramatika na bäIgarskija kni\oven ezik f{istoical Grammar
ofBulgarian]. Vtoro izdanie. Sofija: Nauka i izkustvo.
Mladenov, stefan (1929). Geschichte der bulgarischen sprache. Berlin und Leipzig: walter de
Gru)ter & Co.
Plank, Frans ( 1994). "Inflection and derivation". In Robert E. Asher (Ed.), The Encyclopedia
of Language and Linguistics (pp. 1671-167B). Oxford: Pergamon press.
Russian Academy Grammar = Russkaja grammatika. (1980). Tom I. Fonerika, Fonologija,
Udarenie, Intonacija, Slovoobrazovanie, Morfologija. Moskva: Nauka.
sanders, Gerald (1988). "zero Derivation and the overt Analogue criterion". Michael
Hammond & Michael Noonan (Eds.), Theoretical Morphology (pp. 155-175). San
Diego, Calif., etc.: Academic Press.
Scalise, Sergio (1984). Generatite Morphology. Dordrecht: Foris.
SJS = Slowikjazyka staroslovönsköho. Lexicon Linguae Palaeosloyenicae. ( 19S2). Vol. 3. Praha:
Academia.
Skaliöka, Vladimir ( 1979). Typologische Studien.Wiesbaden: Braunschweig.
Spencer, Andrew (2002). "Gender as an inflectional category". Journal ofLinguistics, 38,279-
3t2.
stojanov, stojan (1993). Gramatika na bälgarskija kniäoven ezik lA grammar of standard
Bulgarianl. V izdanie. Sofija: Universiretsko izdatelstvo "Sv. Kl. Oxridski'l
Stump, Gregory T. (1998). "Inflection". In Ardrew Spencer & Arnold Zwicky (Eds.), The
Handbook of Morphology (pp. t3-a3). Oxford: Blackweli.