The Many Faces of Perfective Aspect in Russian

1
The Many Faces of Perfective Aspect in Russian

• Laura A. Janda
• University of Tromsø
• laura.janda_at_hum.uit.no
• http//hum.uit.no/lajanda/

2
Overview

• Cluster Model Three Metaphors
• Solid vs. Substance gt Perfective vs.
  Imperfective
• Travel vs. Motion gt Construal of Completability
• Granular vs. Fluid gt Construal of
  Singularizability
• Cluster Model Predictions
• Complex Act Perfectives and Biaspectuals
• Single Act Perfectives and Allomorphy
• Natural Perfectives and Empty Prefixes
• Semantic Profiles
• Constructional Profiles
• Grammatical Profiles

3
Slavic Aspect

• Contrasts perfective vs. imperfective (no
  progressive and no neutral aspect)
• Is independent of tense and other verbal
  categories
• Implements imperfective (as unmarked) where other
  languages would have perfective
• Has a complex and seemingly incoherent array of
  uses
• Has a very complex system for aspectual
  derivation

4
Problem

• Model of aspectual pairs has a long tradition
• Vinogradov 1938, axmatov 1941, Bondarko 1983,
  Certkova 1996, Zaliznjak melev 2000,
  Timberlake 2004

• Suspicions that aspectual relationships involve
  more complex clusters have arisen
• Isacenko 1960, Bertinetto Delfitto 2000,
  Tatevosov 2002, Janda forthcoming

5
What is an aspectual cluster?

• An aspectual cluster is a group of verbs joined
  via transitive relationships on the basis of
  aspectual derivational morphology
• All verbs in a cluster are aspectually related to
  a single lexical item
• In addition to Imperfective Activity verbs, an
  aspectual cluster can include four types of
  Perfective verbs
• Natural Perfective, Specialized Perfective,
  Complex Act, Single Act

6
Four types of Perfectives

• Natural Perfective
• napisatp write, svjazatp tie, o(b)cipatp
  pinch/pluck, ?krepnutp get stronger
• Specialized Perfective
• perepisatp rewrite, razvjazatp untie,
  pererabotatp revise, vdutp blow in,
  vycipatp pluck out
• Complex Act Perfective
• popisatp write a while, porabotqtp work a
  while, podutp blow a while, pocipatp
  pinch/pluck a while, poskripetp squeak a
  while
• Single Act Perfective
• dunutp blow once, cipnutp pinch/pluck
  once, skripnutp squeak once

7
The three metaphors

• Solid vs. Substance
• Construal of event aspect type
• Perfective vs. Imperfective
• Travel vs. Motion
• Construal of Completability
• Natural Specialized Perfective vs. Complex Act
  Perfective
• Granular vs. Fluid
• Construal of Singularizability
• Single Act Perfective

8
Metaphor 1

• Solid vs. Substance gt Perfective vs. Imperfective

9
Traditional Feature Analyses

• Boundedness, Totality, Definiteness, Change vs.
  Stability, Sequencing vs. Simultaneity, Exterior
  vs. Interior, Figure vs. Ground, Punctuality vs.
  Durativity, Resultative
• Lack intricacy needed to account for uses
• Are ultimately new synonyms for perfective vs.
  imperfective

10
The Two Types of Matter

• Discrete Solid Object

• Fluid substance

11
Properties of Matter and Parameters of Aspect

• Inherent Properties -- correspond to inherent
  structure of situations and act as default values
• Interactional Properties -- correspond to
  discourse structure, and can override Inherent
  Properties
• Human Interactional Properties -- correspond to
  pragmatic structure, and can override Inherent
  Properties

12
A. - G. Properties inherent to types of matter

• A. Edges
• B. Shape
• C. Integrity
• D. Countability
• E. Streamability
• F. Penetrability
• G. Conversions

13
A. Edges

• Perfective
• Has edges 1)

• Imperfective
• Has no edges 2)

14
B. Shape

• Perfective
• Can have various shapes 3), 4), 5)

• Imperfective
• Has no shape but can spread 6), 7), 8), 9)

15
C. Integrity

• Perfective
• A unique occurrence 10)

• Imperfective
• Continuous processes and repetitions 11), 12)

16
H. K. Interactions of types of matter and
discourse structure

• H. Compatibility
• I. Dynamicity
• J. Salience
• K. Contiguity

17
H. Compatibility

• Perfective Sequencing and future 24), 25), 26)

• Imperfective
• Simultaneity and present 27), 28), 29), 30)

18
H. Compatibility, contd.

• Perfective embedded in imperfective
• Interruption of ongoing action 31)

19
I. Dynamicity

• Perfective moves story along 32)

• Imperfective slows story down 32)

20
J. Salience

• Perfective obvious, foregrounded events 32)

• Imperfective backgrounded events 32), 33)

21
Metaphor 2

• Travel vs. Motion gt Construal of Completability
• Natural Specialized Perfective vs. Complex Act
  Perfective

22
Travel vs. Motion

• One can travel to a destination
• or
• One can move without a destination
• This distinction is grammaticalized in Russian
  motion verbs idtii walk (somewhere) vs.
  xoditi walk (around, back and forth)
• This can be likened to the Completability of an
  action

23
Completability

• Pisatel pieti knigu.
• The writer is writing a book.

• Professor rabotaeti v universitete.
• The professor is working at the university.

Note that Completability is a scale involving
various kinds of construal.
24
Completability

• Many verbs are Ambiguous
• Completable
• Pisatel pieti knigu A writer is writing a
  book
• Non-Completable
• Pisatel pieti knigi A writer writes books
• Some verbs are Non-Completable stonati moan
• But some can be Completable if specialized
• rabotati work gt pererabotatp revise
• Few verbs are unambiguously Completable
• krepnuti gt okrepnutp get stronger

25
What Completability means for aspectual
derivation

• Only verbs that can be construed as Completable
  have Natural Perfectives
• pisati write gt napisatp write, krepnuti
  get stronger gt okrepnutp get stronger
• Only verbs that can be construed as
  Non-Completable have Complex Act Perfectives
• pisati writegt popisatp write a while,
  stonati moangt postonatp moan a while,
  rabotati workgt porabotatp work a while
• Verbs that can be Completable if specialized have
  Specialized Perfectives
• pisati writegt perepisatp rewrite, rabotati
  work gt pererabotatp revise

26
Metaphor 3

• Granular vs. Fluid gt Construal of
  Singularizability
• Single Act Perfective

27
Granular vs. Fluid

• Substances can be
• Particulate, like sand
• Continuous, like water

This can be likened to Singularizability of an
action
28
Singularizability

• Malcik duli na oduvancik.
• The boy was blowing on the dandelion.
• Malcik dunulp na oduvancik.
• The boy blew once on the dandelion.

• Professor rabotali v universitete.
• The professor was working at the university.

29
What Singularizability means for aspectual
derivation

• Verbs that can be construed as Non-Completable
  and have a Complex Act Perfective can also have a
  Single Act Perfective
• cipati pinch/pluck pocipatp pinch/pluck
  a while gt cipnutp pinch/pluck once
• duti blow podutp blow a while gt dunutp
  blow once
• skripeti squeak poskripetp squeak a while
  gt skripnutp squeak once
• rabotati work porabotatp work a while gt
  rabotnutp work once NB Some are formed
  ad-hoc

30
Singularizability and motion verbs

• The Non-Completable motion verbs can also be
  construed as Singularizable
• xoditi walk can refer to multiple round-trips,
  in which case there is a Single Act Perfective
  sxoditp make a single round trip

On sxodilp v magazin He went to the store (and
came back once)
31
Advantages of the cluster model

• The cluster model is more accurate than the
  pair model
• Cluster structures are highly constrained and
  transparently motivated by meanings of verbs
• Verbs with Completable construals form Natural
  Perfectives
• Verbs with Non-Completable construals form
  Complex Act Perfectives
• Verbs with Granular construals form Single Act
  Perfectives
• Motion verbs play a prototypical role in the
  system

32
Cluster Model Predictions

• Complex Act Perfectives and Biaspectuals
• Single Act Perfectives and Allomorphy

33
Complex Act Perfectives and Biaspectuals
34
Biaspectual verbs

• Can express both Imperfective and Natural
  Perfective with the same morphological form gt
  indicates strong tendency for Completability,
  which should hinder formation of Complex Act
  Perfectives
• Over 90 are foreign borrowings
• All foreign verbs have ova- suffix, which gives
  verbal inflection but does not designate aspect
• Empirical study tests prediction of Cluster Model

35
Empirical study

• Hypothesis
• Bi-aspectual borrowed verbs are strongly
  Completable (telic), so they will be unlikely to
  form Complex Act Perfectives with po-
• Imperfective borrowed verbs will be more likely
  to form Complex Act Perfectives with po-

36
Empirical study of Biaspectuals

• Methodology
• Cull all foreign verbs from a single source
• Sort Biaspectual vs. Imperfective
• Collect data on frequency of unprefixed and po-
  prefixed (Complex Act Perfective) forms

37
Results of empirical study

• 555 foreign verbs in Wheeler 1972/1992
• 349 (63) Bi-aspectual
• 206 (37) Imperfective

Is this significant?
Yes! Logistic regression model using Pearsons
statistic yields 107.37 and the associated
p-value is lt.0001
38
Single Act Perfectives and Allomorphy
39
A little problem...

• The Cluster Model claims that Single Act
  Perfectives are formed both with the suffix -nu
  (as in cixnut sneeze once) and with the prefix
  s- (primarily for motion verbs like sxodit go
  someplace and come back once).

??????
40
But...

• this is a strange combination of -nu and s- and
  there is very little in the scholarly literature
  to support grouping these two morphemes together

41
Allomorphy hypothesis

• Suffix -nu and prefix s- are allomorphs if
• They are in complementary distribution
• They have the same function
• Databases of Single Act Perfectives with -nu and
  s-
• Statistical Analysis of distribution

42
-nu database

• 295 Imperfective verbs form Single Act
  Perfectives with -nu
• collected by Anastasia Makarova
• data from vedova et al. 1980, Zaliznjak 1980 and
  Exploring Emptiness database at UiT

???????? ??? ???????????

• includes both -nu and -?nu verbs like pleskat
  splash which forms plesnut and pleskanut
  splash once
• includes both reflexive and non-reflexive verbs
  like kacat/kacnut, kacatsja/kacnutsja
  swing/swing once

43
s- database
????????

• 105 Imperfective verbs form Single Act
  Perfectives with s-
• collected by Laura Janda with help from Anastasia
  Makarova
• data from 17 V Dictionary, Zaliznjak 1980 and
  Isacenko 1960
• includes 11 motion verbs like xodit/sxodit
  go/go someplace and come back once
• includes both reflexive and non-reflexive verbs
  like lovcit/slovcit, lovcitsja/slovcitsja be
  sneaky/do one sneaky thing

44
Are -nu and s- in complementary distribution?
45
chi-squared 259.3 (plt0.0001, df5) Cramers V
0.8 (enormous effect)



46
Do -nu and s- have the same function?

• Yes, they can both mean do something once
• There is one verb that uses both -nu and s- to
  form synonyms xvastat xvastnut/xvastanut,
  sxvastat brag once
• There are several verbs that use -nu and s-
  simultaneously sgrustnut, sgrustnutsja, feel
  sad once smetnut, smetnutsja leap sideways
  (once of animals), struxnut do one cowardly
  thing

???????? ??? ?????????
47
The function of -nu and s- is not entirely
identical...

• With -nu we usually extract a single cycle from a
  series of repeated events cixat/cixnut
  sneeze/sneeze once, lizat/liznut lick/lick
  once
• With s- we often have something that actually
  only happened once maloduestvovat/smaloduestvo
  vat act cowardly/do once cowardly thing

Potential series of events
Real series of events
-nu
s-
48
Evaluation of the Allomorphy hypothesis

• Are -nu and s- in complementary distribution?
• Do -nu and s- have the same function?
• Is the Allomorphy hypothesis confirmed?
• Is the Cluster Model confirmed?

• Statistically speaking, almost.
• As far as we can tell, almost.
• Mostly.
• Yes.

49
Natural Perfectives and Empty Prefixes

• Semantic Profiles
• Constructional Profiles
• Grammatical Profiles

50
Thought experiment
Thats weird!

• Imagine a language with aspect
• Two values Imperfective and Perfective
• Perfective prefix Imperfective
• Prefixation contributes only Perfective
• How many prefixes does this language need?

Reasonable answer ONE prefix but Russian has
19! (Krongauz 199864 ,99)
51
Exploring Emptiness research group at the
University of Tromsø
Investigating verbal morphology traditionally
claimed to be semantically empty
Olga Lyashevskaya
Tore Nesset
Svetlana Sokolova
yours truly
Julia Kuznetsova Anastasia Makarova (not
pictured)
52
Natural Perfectives and Linguistic Profiles

• 2000 imperfective verbs in Russian form a
  Natural Perfective with an empty prefix
• 19 prefixes supposedly have the same empty
  meaning
• How can we prove that the empty prefixes are
  not empty?
• Linguistic Profiles
• Semantic Profiles
• Constructional Profiles
• Grammatical Profiles
• Databases and statistical analyses

53
Natural Perfectives and Empty Prefixes

• Semantic Profiles
• with Olga Lyashevskaya
• Semantic Classes assigned by the Russian National
  Corpus

54
(No Transcript)
55
chi-squared 106.24 df 10 p-value lt
2.2e-16 Cramers V 0.3 (moderate effect)
56
Natural Perfectives and Empty Prefixes

• Constructional Profiles
• Forthcoming PhD dissertation by Svetlana Sokolova

57
Constructional Profiles of Verbs

• Relative frequency distributions of constructions
• Constructional profiles show that near synonyms
  can behave very differently
• Constructional profiles can show that the empty
  prefixes are not so empty

58
Gruzit load one verb with three empty
prefixes?

• It is traditionally claimed that in aspectual
  pairs such as pisat/napisat write,
  morozit/zamorozit freeze, obedat/poobedat
  eat lunch the prefixes na-, z?-, p?- are
  empty (have zero meaning)
• A few verbs have more than one empty prefix
  gruzit load has three Natural Perfectives
  nagruzit, zagruzit, pogruzit
• The constructional profiles of these three verbs
  are very different, indicating that the prefixes
  cannot be empty (How could there be three
  different zeroes?)

59
Relevant Constructions

• Accusative marks load (locatum)
• Acc na/v Acc (nagruzit jaciki na teleku
  load boxes onto the wagon)
• Acc (zagruzit ugol budet problematicno loading
  the coal will be problematic)
• Accusative marks container (location)
• Acc Inst (on nagruzil sanki proviziej he
  loaded the sleighs with provisions)
• Acc (nagruzili telegi i uexali v gorod they
  loaded the wagons and rode to town)
• Data 935 sentences from the Russian National
  Corpus

60
chi-squared 452.827 (plt0.0001, df6) Cramers
V 0.507 (large effect)

61
Natural Perfectives and Empty Prefixes

• Grammatical Profiles

62
Distribution of Paradigm Forms for Natural
Perfectives formed with prefixes za-, na-, pro-,
s-
198,132 datapoints representing 467 verbs in RNC
63
chi-squared 7823.7 (plt0.0001, df9) Cramers V
0.1 (small effect)
64
Conclusions

• Cluster Model provides a better analysis of
• Perfective vs. Imperfective than feature analysis
• Aspectual relations than traditional pair model
• Cluster Model makes predictions borne out by the
  data
• Biaspectual verbs are strongly Completable and
  avoid formation of Complex Act Perfectives
• Suffix -nu and prefix s- collaborate in a state
  of near-allomorphy to form Single Act Perfectives
• Cluster Model helps debunk empty prefix myth by
  showing relationship between prefixes and
• Semantic profiles
• Constructional profiles
• Grammatical profiles