Transcript Frequency-oriented diachronic approach to the study of

Frequency-oriented diachronic
approach to the study of variation in
prefixes in the aspectual system of
the Russian language
Valery Solovyev, Vladimir Bochkarev
Kazan University
The main components of the Russian
aspectual system
• Verbs: Perfective vs. Imperfective (делать –
сделать ‘do’)
• Perfective: Natural vs. Specialized (сделать
‘do’, переделать ‘redo’)
- Natural Perfective has the same lexical meaning as the
corresponding simplex verb
- The lexical meaning of the Specialized Perfectives is different
from the corresponding simplex verb
• Aspectual pairs: Perfective and Imperfective
with the same lexical meaning
Aspectual pairs: problems
• What does it mean to have the same meaning?
• Semantics is not formalized
• How to distinguish between Natural and
Specialized Perfectives?
• Currently there are no operational criterion
allowing to establish whether the pair of verbs
are aspectual or not
• Even professional aspectologists often can not
come to an agreement about the aspectual pairs
• The existence of the “diffuse zone" between
Natural and Specialized Perfectives (L. Janda)
Questionnaire 1997 (Горбова, ВЯ, 2011 )
+ aspectual
pair
- aspectual
pair
+/- with
remarks
no answer
прыгать – прыгнуть
(jump)
14
7
6
5
видеть – увидеть
(see)
12
10
5
5
идти – пойти
(go)
10
14
3
5
есть – поесть
(eat)
8
13
4
7
гулять – погулять
(walk)
7
14
5
6
петь – пропеть
(sing)
6
17
2
7
кричать – закричать
(shout)
5
16
6
5
Corpus
• Google Books Ngram Viewer
(https://books.google.com/ngrams)
• More than 67 billions words in Russian, 200
times greater than in the Russian National
Corpus
• Diachronic corpus: from 1800 up to 2008
• A graphical representation of the frequency of
use of words and phrases
New frequency-based approach to
semantics
• Presupposition-1:
If words have the same lexical semantics, but
belong to different grammatical categories, the
changes in frequency should be in sync
Example:
читать (to read, inf) – читал (read, 3sg.past)
r = 0.888, p < 10-29
Frequency changes are synchronized
Example (continuation):
читать (to read, inf) – читал (read,
3sg.past) – читая (reading, conv)
r = 0.860. The frequency of the word is very
small, it is difficult to compare
Example (continuation):
читать – читал – читая (scaled graph)
Multiply all frequencies of the word читая by 4.
Then see that frequencies change synchronously
Natural Perfective
Example 1: глядеть-поглядеть (look)
r = 0.821. Frequencies change synchronously 
lexical semantics coincides
New frequency-based approach to
semantics
• Presupposition-2:
If words have different lexical semantics, then
there is a period of time when one of the words
will be in more demand, and their frequencies
will not change sync
Specialized Perfective.
Example: строить - перестроить
(build-rebuild)
r = 0.025, p = 0.722. The shape of the curves differs
 lexical semantics is not the same
Criterion for what constitutes an
aspectual pairs
 Have in Google Books Ngram Viewer graphs of
the same shape
• Advantage of the criterion: can be applied to an
unlimited amount of examples and do not rely on
author’s intuition
• Limitation of the criterion: requires many
datapoints per pair
Бить (beat) and its perfectives
• Dictionaries indicate more than 10 meanings
of бить
• Different meanings ​have different natural
perfectives, for some there are no perfectives
• Individual meanings of words, but not
the whole words form aspectual pairs
Meanings of бить. Refinements
Janda’s database:
• Бить 1 пробить (ring, chime)
• Бить 2 побить (beat)
• Бить 3 разбить 1 (break into smaller pieces)
also can be added
• Бить 3a разбить 2 (defeat)
Бить врага – разбить врага
(defeat the enemy)
r = 0.975, p = 0. бить and разбить in this
meaning are aspectual pair
Meanings of бить. Additions
Shvedova’s Semantic Dictionary (1998)
• Бить 4 забить 1 (drive [as piles into ground])
Is забить Natural Perfective or Specialized
Perfectives?
And also there exist:
• Бить 5 набить 1 (beat [slap face])
Бить – забить
The semantics are obviously different, but …
Бить сваи – забить сваи
‘drive piles/beams’
The semantics is same 
in this sense it is the aspectual pair
Бить – набить
r = 0.299, p < 0.0001. The semantics are
different, but …
Бить морду – набить морду
‘beat face’
r = 0.394, p < 0.0001. In this sense it is the
aspectual pair
‘набить морду’ or ‘побить морду’?
‘побить морду’ clearly inappropriate
Колоть – заколоть (stab)
• Is it the aspectual pair?
Зализняк, Микаэлян, ВЯ, 2012 – No
Кузнецова, Янда, ВЯ, 2013 – Yes
“… не очень молодые люди … маршировали там с утра до
вечера и кололи чучела штыками”
‘… not so very young people … marched there from morning to
night and stabbed the scarecrow with bayonets’
“А сержант … настиг … фон Враницкого и молча заколол его
штыком”
‘A sergeant … caught … von Vranitzky and silently stabbed him
with a bayonet’
• Is there identity of lexical meanings?
• Can one example to prove the identity of semantics?
Колоть штыком – заколоть штыком
‘to stab with bayonet’
The curves have not the same shape.
Coincide only peaks during World War 2
Колоть свинью – заколоть свиней
‘stab pig’
In this sense колоть–заколоть is the aspectual pair
Possibilities of the method - 1
The method does not replace the traditional
linguistic methods, but complements them.
1) It gives arguments in favor of aspectual pairs
(if the frequency curves are synchronized)
2) It gives arguments against aspectual pairs
(if the frequency curves are different)
Triplets: primary imperfectives
vs. secondary imperfectives
• An example of a triplet: драить-надраитьнадраивать (polish)
• Two pairs: (1) perfective-primary imperfective;
(2) perfective-secondary imperfective.
Which are aspectual pairs?
• Do all members of a triplet evolve in the same
way?
дохнуть-подохнуть-подыхать
‘croak’
Primary imperfective approaches with other
жечь-сжечь-сжигать
‘burn’
Secondary imperfective approaches with other.
After 1920: r = 0.902, r = 0.798, p < 10-20
Possibilities of the method – 2
• It is a convenient way of analyzing data on the
evolution of aspectual pairs and triplets
• Frequency triplets in most cases (75%) are
evolving towards the semantics convergence
Evolving of usage frequency of
perfectives and imperfectives
The frequency ratio of aspects varies
linearly
Secondary imperfectives are washed
out of the language
Conclusion
• A new diachronic frequency-based approach
to lexical semantics has been presented
• It can be used to distinguish Natural Perfective
and Specialized Perfectives
• It allows for studying the scheme of the
evolution of triplets
• The frequency of imperfectives falls during the
last 200 years
Thank you for attention!
Maslov’s test. (Горбова, ВЯ, 2011 )
+ aspectual
pair
- aspectual
pair
+/- with
remarks
no answer
прыгать – прыгнуть
(jump)
9
0
0
0
видеть – увидеть
(see)
8
0
0
1
идти – пойти
(go)
7
1
0
1
есть – поесть
(eat)
4
2
2
1
гулять – погулять
(walk)
3
3
2
0
петь – пропеть
(sing)
3
3
2
0
кричать – закричать
(shout)
5
1
2
1