Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.philol.msu.ru/~otipl/new/main/courses/rm/dgbq-RM2004-WXP.doc Дата изменения: Sun Nov 14 23:07:00 2004 Дата индексирования: Sat Dec 22 21:28:32 2007 Кодировка: koi8-r

BASIC QUESTIONS IN ESTABLISHING THE VERB CLASSES OF CONTEMPORARY RUSSIAN

W. U. Dressler (Wien) N. V. Gagarina (Санкт-Петербург)


Introduction

In perfect agreement with V.V. Ivanov, who has often insisted on the
importance of semiotics for linguistics, the functionalist model of Natural
Morphology which we espouse (cf. Dressler et al. 1987) emphasizes the
semiotic basis of morphology. One relevant semiotic principle is the
reliability of signs, as highlighted by Morris (1938). Reliability implies,
on the paradigmatic axis, a) easy identificability of a sign, and on the
syntagmatic axis, b) reliable chaining of signs and of their parts. On a
lower functional level, we have the two main functions of inflectional
morphology: c) morphosemantic and morphotactic motivation of inflectional
paradigms, d) providing syntax with the (primarily synthetic) morphological
expression of morphosyntactic categories. Now we claim that, within
inflectional morphology, these four functions are best fulfilled by
productive patterns, i. e. productive categories, rules and classes of
paradigms (cf. Dressler 1997a, Dressler et al. 1996).
The central role of productivity is acknowledged in syntax, where nobody
would propose a model just in order to account for unproductive
constructions, in many models of phonology - particularly in Natural
Phonology, where truly phonological processes must be fully productive (cf.
Dressler 1985) - and, at least implicitly, in those models of word
formation which follow Aronoff's (1976) postulate that the main task of
word formation theory is to account for what is a potential, not an actual,
word: a potential complex word (i. e. a word derived via compounding or a
process of derivational morphology), in contrast to an actual word, is
derived via a rule which is not unproductive.
In models of inflectional morphology, however, emphasis is laid on the
distribution of patterns, without consideration for productivity. For
example, Jakobson, in his famous morphological writings (e. g. on "Russian
Conjugation", 1948/1971), devotes much more space to unproductive classes
and irregular, isolated paradigms than to productive ones.
The question of productivity of verbal classes has long history of
discussing. Karcevskij (1927) has established five opened classes (and
about 400 unproductive verbs of closed classes). This classification is
still used in the grammars of Russian for students (e. g. Miloslavskij
1981, Lekant 1992, Belo?apkova 1997). Unproductive groups of verbs are very
heterogenous and of variable membership [seven (Karcevskij 1933), seventeen
(Vinogradov 1972), twenty (?vedova 1980)].
In many recent publications, the concept of default plays a great role
(e. g. Fraser & Corbett 1994): the respective default is determined
according to descriptive simplicity (e. g. elsewhere condition), again
without consideration for productivity.

Sketch of our model

In contrast to all these models, we postulate, in strict parallel to all
other rule components of grammar, productivity as a primitive property of
inflectional morphology on the level of the potential system of a language
(grammatical competence), whereas type and token frequency are derived
properties on the levels of accepted norms (i. e. of language as social
institution) and of performance (parole), respectively. Furthermore we
distinguish morphological richness and complexity in the following way (cf.
Dressler 1997b): morphological richness of a language is constituted by its
productive morphological patterns, whereas morphological complexity
consists of richness plus all the unproductive morphological patterns.
For inflectional classes, we assume (cf. Dressler et al. 1996, Dressler
1997a) a class hierarchy which consists, in descending order, of: the
inflectional system (e. g. of verbs), macroclasses, traditional classes,
subclasses, [((sub)sub)...classes, if needed], microclasses. An
inflectional microclass is the smallest subset of an inflectional class
above the paradigm, definable as the set of paradigms which share exactly
the same morphological generalizations, but may differ via the application
of phonological processes (e. g. of final devoicing). An isolated paradigm
is a paradigm which differs morphologically or morphonologically from all
other paradigms, i. e. it does not form a microclass of its own but is
considered a satellite to the most similar microclass(es). Thus all
suppletive paradigms are isolated paradigms. Productive microclasses form
the core of hierarchically higher classes. In ascending hierarchically from
the terminal node of a microclass up to the dominating node of the
respective macroclass, we group those microclasses together which share a
certain amount of inflectional properties, and so on. Finally, all the
(sub/micro)classes of a macroclass share at least one exclusively identical
unmarked-category realization. In an inflecting-fusional language, such as
Russian, there is a preference for binarity, i. e. for having a higher node
dominate just two immediately lower nodes. In descending from the highest
node down to the terminal nodes of microclasses, we follow the principles
of inheritance and of default inheritance (cf. Fraser & Corbett 1994), i.
e. the obligatory properties of a higher node are inherited by all lower
nodes dominated by it, whereas a default property of a higher node a)
should become obligatory in the majority of lower nodes dominated by it, b)
but should be violated by a restricted number of lower nodes, c) but may be
inherited as such by a node of intermediary rank, provided that some other
property is added on this level.
As to the bases of morphological rules, we differentiate: word-based
inflection, where the base of an inflectional paradigm is the unmarked
autonomous inflectional word form, e. g. Nom. Sg. jazyk 'language', which
has no correspondence in Russian verb inflection; b) stem-based inflection,
where the root plus a thematic marker, i. e. less than an autonomous
inflectional word form, is the base of a paradigm, e. g. the thematic
vowels -a- and -e- of the first and second verbal microclasses of our
classification; c) root-based inflection, where the base of the (athematic)
paradigm is a root (the non-autonomous lexical base, without a thematic
marker), sc. nothing intervenes between root and affix, as in the paradigm
of nes-ti 'to carry' and where the root is excluded as an autonomous word
form (e. g. n'os). In Russian verbs, (athematic) root-based inflection is
always unproductive, whereas all productive microclasses are stem-based, i.
e. thematic.

Degrees of productivity

Such as many other concepts of naturalness theory, also productivity of
microclasses is gradual. This gradualness corresponds to the following
hierarchy of criteria:
a) Wurzel's (1984) secondary productivity in the integration of loan words
with unfitting properties, i. e. the highest degree of inflectional
productivity must obtain in order to integrate loan words (sc. it is more
difficult to integrate words coming from a foreign language than indigenous
words), 2) even loan words with unfitting properties, and to 3) accomodate
these according to the properties of an existing inflectional (micro)class.
Clearly a microclass must have optimal productivity in order to overcome
the difficulties of 1) and 2) and to enforce accommodation 3). Examples
are: E. to tape, print, click > tapat', printit', kliknut' (in
computerese).
In contemporary Russian, verbs constitute only nine percent of the whole
corpus of loan words (Kotelova 1983). Generally, many more new nouns are
created than verbs (cf. Zemskaja 1992: 126). Usually loan verbs are derived
from loan nouns by means of using already fitting morphological patterns
(in our case, the thematic vowel that determines the assignment of verbs to
productive classes) and they are not transferred (with or without
morphonological changes) from a foreign language to Russian, like it is
usually with nouns, e. g. kurort, zeitnot, etc. Quite generally, it is more
difficult for foreign verbs then nouns to integrate directly into a
language (cf. also Dressler 1997a: 7).
b) Wurzel's (1984) primary productivity in the integration of loan words
with fitting properties. Here only the first difficulty of a) has to be
overcome, because the class-defining properties of the loan word already
fit into the receiving microclass. For example, English verbs such as to
fax, quizz are athematic. In case that they are integrated into the
athematic microclass of nes-ti, then we have an instance of criterion b).
If, however, they become thematic and thus are accomodated and lose an
unfitting property, then criterion a) obtains as in case with faksit',
faksanut'.
c) assignment of indigenous neologisms formed by conversion, i. e. when no
suffix of derivational morphology predetermines the choice of inflectional
class. cl) Since proper names represent a marked subset of the indigenous
lexion, verbs formed via conversion from proper names indicate a higher
degree of productivity than verbs formed from common nouns. Example cited
after Zemskaja (1997) is: Bobbit > bobbitnut' and example cited after
Leva?ov (1996) is: Gajdar > gajdarit'.
d) inflection class change from a less to a more productive or stable
microclass. This is a criterion of rather little importance, because the
shift may occur even to an unproductive microclass from another, less
stable, i. e. from a recessive microclass. An example is class shift from
the unproductive microclass kapl'u - kapat' 'to drop' to the productive
microclass kapaju - kapat'. Karcevskij (1927) noted 37 verbs which moved to
the productive class.
e) word formation productivity of indigenous affixations, is the last and
hierarchically lowest criterion, because it provides direct productivity
evidence only for word formation, whereas for inflection it proves only
minimal stability. Highest type frequency occurs within the microclass
-ovat', with suffix -izirova- which derives verbs from loaned nouns, e. g.
dollar-izirova-t', kompjuter-izirova-t'. Such verbs have almost the same
level of frequency as nouns in -izacija (Zemskaja 1996, Bojarkina 1993),
and constitutes the majority in the corpus of new "verb dictionary".

Hierarchical verb classification

As the basis for our classification we have taken the traditional
correlation between the (open) stem (OS) of the infinitive and of the close
stem (CS) of the present/future (1. Sg.). Using theoretical assumptions of
the theory of Natural morphology (Dressler 1987, Kilani-Schoch 1988,
Dressler & Dziubalska-Ko?aczyk 1997, in press) we propose the following
hierarchical scheme of verb classes in Russian in a bottom-up manner,
starting with microclasses. Similar microclasses form subclasses
(sub(sub(sub))classes, if necessary), and these constitute macroclasses. We
define an inflectional macroclass as the highest, the most general type of
classes, which consists of several hierarchically lower (sub)classes and
microclasses. In the following classification, all microclasses which are
not explicitly identified as productive, are unproductive.
Macroclass 1 - OS (open stem) ends in a vowel (V) vs. CS (close stem) has
consonant (C) addition to the OS
Subclass I - V in the OS vs. addition of j in the CS
Subsubclass I.1 - single final V in the OS
Subsubsubclass I.1.1 - V in the OS does not change (when adding j) in the
CS. The microclasses have the following stem alternations:
MC 1: a/aj (gu'a-t' - gul'aj-u 'to walk') - fully productive (a-
productivity, i. e. productive even according to criterion a)
MC 2: e/ej (bole-t' - bolej-u 'to be sick') - slightly productive (e-
productivity)
MC 3: i/ij (gni-t' - gnij-u 'to rot')
MC 4: u/uj (du-t' - duj-u 'to blow')
Subsubsubclass I.1.2 - final V in the OS is not equal to the final V in the
CS. The microclasses have the following stem alternations:
MC 5: i/Ьj (p'i-t' - p'j-u 'to drink')
MC 6: y/oj (my-t' - moj-u 'to wash')
Subsubclass I.2 - final V is a part of a pluriphonemic thematic suffix
(V)va. The microclasses have the following stem alternations:
MC 7: ova/uj (komandova-t' - komanduj-u 'to comand') - fully productive (a-
productivity)
MC 8: eva/uj (pleva-t' - pl'uj-u 'to spit')
MC 9: va/j (dava-t' - daj-u 'to give')
Subclass II - V in the OS vs. addition of С other than j in the CS
MC 10: y/yv (yv') (ply-t' - plyv-u 'to swim')
MC 11: a/an (an') (sta-t' - stan-u 'to become')
Macroclass 2 - OS ends in V vs. CS without this final V
Subclass I - С + single V vs. С without V
Subsubclass I.1 - С before V in the OS vs. the same (palatalised (С')) С
without V in the CS

Subsubsubclass I.1.1 - С + a vs. С without a. The microclass has the
following stem alternations:
MC 12: ja/j (seja-t' - sej-u 'to seed')
Subsubsubsubclass I.1.1.1 - other than j С in the OS + a. The microclasses
have the following stem alternations:
MC 13: Ca/C(C') with no mobile root V (vra-t' - vr-u 'to lie')
MC 14: Ca/C(C') with mobile root V (bra-t' - ber-u 'to take')
Subsubsubclass I.1.2 - С with e vs. С without e
MC 15: Ce/C(C') with no mobile root V (smotre-t' - smotr'-u 'to look')
MC 16: Ce/C(C') with mobile root V (umere-t' - umr-u 'to die')
MC 17: Ci/C' (kuri-t' - kur'-u 'to smoke') - fully productive (a-
productivity)
MC 18: Co/C' (boro-t'-s'a - bor'-u-s' 'to struggle')
Subsubclass I.2 - С before V in the OS alternates with C1 in the CS. Due to
lack of space we simplify greatly in not differentiating further
microclasses according to distinct synchronic morphonological rules. Thus,
the microclasses have the following stem alternations:
MC 19: Ca/ C1 (pisa-t' - pi?-u 'to write', plaka-t' - pla?-u 'to cry', sypa-
t' - sypl'-u 'to poor')
MC 20: Ce/ C1 (v'id'e-t' - vi?-u 'to see', lete-t' - le?-u 'to fly', kipe-
t' - kipl'-u 'to boil')
MC 21: Ci/ C1 (hodi-t' - ho?-u 'to go', kruti-t' - kru?-u 'to turn', krasi-
t' - kra?-u 'to paint')
Subclass II - С + V is part of a pluriphonemic thematic suffix nu
MC 22: OS of the past maintains suffix -nu (prygnu-t' - prygn-u 'to jump
once') - fully productive (cl-productivity)
MC 23: OS of the past has neither the suffix -nu nor the -l (past macs.)
(otvergnu-t' - otverg 'to reject')
Macroclass 3 - stem of the infinitive ends in С
Subclass I - infinitive with dental suffix (default -ti)
Subsubclass I.1 - final С of stem does not alternate in the CS with other С
(С')
MC 24: non-altemating root V (polz-ti - polz-u 'to crawl')
MC 25: alternating root V in the past (nes-ti - nes-u 'to carry', n'os:
past. masc.)
Subsubclass I.2 - final С of the stem alternates with the other C1
MC 26: st'(sti)/d (d') (klas-t' - klad-u 'to put')
MC 27: sti/b (b') (skres-ti - skreb-u 'to scratch')
Subclass II - infinitive with final ?
MC 28: ?/k (1.Sg., 3.Pl) (pe? - pek-u 'to bake')
MC 29: ?/g (1.Sg., 3.Pl) (mo? - mog-u 'to be able')
Some examples of isolated paradigms: le? - l'ag-u 'to lie down', odet' -
oden-u 'to put on clothes, shoes', est' - em 'to eat'.

Conclusion

In contrast to previous classifications of Russian verbs, our
classification is based on homogeneous paradigm microclasses, which
marginalizes isolated paradigms. This follows from our view on
morphological productivity representing the core of morphology as an active
mechanism, which is quite different from pure storage of more or less
idiosyncratic morphological forms. Furthermore generalizations on
unproductive, but still homogeneous microclasses can be easier put to
active morphological creativity (e. g. in poetical occasionalisms) and
should therefore be distinguished from idiosyncratic, isolated paradigms.
Of course, minimicroparadigms of just two or three exactly parallel
paradigms (e. g. MC 3, 4) represent a transition between isolated
paradigms, that must be acquired via rote learning, and abstract patterns
of microclasses with higher type frequency. According to the principles of
(default) inheritance we have established a greater "vertical" depth and
smaller "horizontal" breadth of subclassification than in previous work
(where ten or more equipollent groups and/or classes are established), even
when they apply the same principle of focussing on the relations between
open and closed stems. Many more details have to be investigated, including
independent, substantial evidence from diachronic change, first language
acquisition and poetic license. But we hope that even now our basic
position has become clear.