The new book: Polyakov V'N' Solovyev V'D' COMPUTER METHODS AND MODELS IN TYPOLOGY AND HISTORICAL LIN

1
The new bookPolyakov V.N. Solovyev V.D.
COMPUTER METHODS AND MODELS IN TYPOLOGY AND
HISTORICAL LINGUISTICSKazan KSU, 2006.
Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
2
Abstract

• In the monography the issues received as a result
  of application of computer and mathematical
  methods to the analysis of the Database
  Languages of the World are presented in the
  regular form . The Database contains the
  description of 315 languages, basically, Europe,
  Northern and the Central Asia on more, than to
  3800 features. Computer methods which allow to
  estimate degree of similarity of structure of
  languages are described. These methods can be
  applied to specification of genetic
  classification. The mathematical model of
  distribution of language features is constructed.
  The monography can be of interest as for
  linguists of various specialities (typology,
  historical linguistics), and for mathematicians
  and experts in computer technologies, who
  interested by application of mathematical methods
  in the humanities.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
3
Contents (part 1)

• Chapter 3. THE DATABASE SOFTWARE
• 3.1. The primary goals solved by means of the
  kernel of the DB
• The 3.2.Windows-version of the kernel of the DB
• 3.3. The DB installation
• 3.4. Work with the Database Languages of the
  World
• 3.5. Pair comparison of languages at the level of
  the classes and at the level of the whole essay
• 3.6. Search of lines in essay
• 3.7. How to work with model and the language
  essay
• 3.7.1. Commands of the main menu
• 3.7.2. Export of language essays in a DB from a
  file
• 3.7.3. Import of language essays in a file
• 3.8. Navigation within the model and within the
  essays
• 3.9. Development of the Web-version of the DB
• 3.9.1. The site viewing
• 3.9.2. Viewing of the essays of language
• 3.9.3. Comparison of essays
• 3.10. DB examination
• 3.11. DB destination

Introduction Chapter 1 TYPOLOGY AND
CLASSIFICATION OF LANGUAGES 1.1. Ways of
classification of languages 1.2. Correlations
between various classifications 1.3. Mathematical
methods 1.4. Databases 1.5. The resume under
chapter 1 Chapter 2 THE DATABASE LANGUAGES OF
THE WORLD 2.1 History of the DB creations 2.2.
Characteristics of a content of the Database
Languages of the World 2.3. The principles of
the Database Languages of the World 2.3.1. A
binary principle 2.3.2. A hierarchical
principle 2.3.3. A paradigm principle 2.3.4. A
data presentation format pros and cons 2.4.
Characteristics of the model of the language
essay 2.5. The resume under chapter 2
4
Contents (part 2)

• Chapter 4. NEW POSSIBILITIES OF QUANTITATIVE
  RESEARCHES
• 4.1. The calculation of the measures of pair
  similarity of languages
• 4.1.1. Approaches to the calculation of the
  similarity measures
• 4.1.2. Taking in account of the structure and
  volume of feature space
• 4.1.3. A technique of evaluation of calculations
• 4.1.4. Some results of calculations of similarity
  measures
• 4.1.5. Preliminary discussions of results
• 4.2. Clasterization
• 4.2.1. The general information on the problem of
  clasterization
• 4.2.2. Clasterization of languages
• 4.2.3. Clasterization of features
• 4.2.4. Evaluation of results by the method Data
  Splitting
• 4.2.5. Architecture of the software system
• 4.2.6. Revealing of possible errors in data
• 4.2.7. Calculation of frequencies of occurrence
  of features in the chosen group of languages
• 4.3. The intellectual analysis of data
• 4.4. Discussion of preliminary results in the
  field of clasterization and intellectual analysis
  of data
• 4.5. The resume under chapter 4

Chapter 5. MODEL OF STRUCTURAL EVOLUTION OF
LANGUAGES 5.1. The general reasons 5.2.
Borrowings vs. Parallel evolution 5.3. Diagrams
of feature distribution 5.4. The analysis of
mathematical model of dynamics of features 5.5.
The resume under chapter 5
5
Contents (part 3)
Bibliography Appendix A. Table ?.1. A genetic
accessory of languages of a DB Appendix ?. Table
?.1. The list of the classifying features
presented in the DB Appendix B. Example of essay
(Swedish language) Appendix ?. Results of
comparison of essays of Danish and Swedish
languages Appendix ?. The special teaching
course Databases For Researches In Language
Typology And Historical Linguistics Appendix
E. Table ?.1. Quantitative card of features for
section 2.5.4 Compound and Complex Sentence for
IE-languages Appendix ?. Dynamics of statistical
universal features Appendix ?. Distribution of 11
dead languages used in one of calculations on
time axis Appendix ?. The symbol index Appendix
?. The short glossary Appendix ?. The
personality index

• Chapter 6. DEVELOPING OF NEW METHODS OF
  VERIFICATION OF GENETIC HYPOTHESES
• 6.1. Calculation quantitative portraits of
  language families and branches
• 6.2. Classification of features
• 6.3. An example of quantitative portrait of
  IE-languages (section "Compound and Complex
  Sentence)
• 6.4. Automatic detection of genetically
  significant features
• 6.5. A technique of verification of genetic
  hypotheses
• 6.6. The resume under chapter 6
• Chapter 7. THE PHENOMENON OF TYPOLOGICAL SHIFT
• 7.1. The nature of the language universal
  features
• 7.2. The nature of the language rare features
• 7.3. A phenomenon of typological shift
• 7.4. The resume under chapter 7
• THE CONCLUSION

6
Chapter 1 TYPOLOGY AND CLASSIFICATION OF
LANGUAGES

• In chapter 1 of the monography the description of
  the basic approaches of different classification
  of languages are presented, the main results
  received in recent works with application of
  mathematical methods are mentioned, and the basic
  problems standing in this feield are listed.
• The resume under chapter 1
• Languages can be classified on the various bases
  on an origin generality (genetic classification),
  on a territorial proximity of zones of
  distribution (areal), on a similarity of
  structural properties (typological).
• Between these classifications there are
  significant correlations, however there are no
  rigid dependences.
• In genetic classification of languages
  considerable successes are reached by application
  of a comparative-historical method, for the basic
  language families protolanguages are
  reconstructed.
• Possibly, comparative-historical method in the
  classical form has practically settled its
  possibilities reconstruction on depth is
  considered the big 10 thousand years impossible,
  many delicate questions remain without the
  answer.
• With introduction in a scientific turn all new
  linguistic data and creation of extensive
  databases essentially new possibilities of
  researches with application of mathematical and
  computer methods have appeared.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
7
Chapters 2 THE DATABASE LANGUAGES OF THE WORLD
and Chapter 3 THE DATABASE SOFTWARE

• In the second and third chapters the detailed
  description of structure of the Database
  Languages of the World and toolkit accompanying
  it are given. Methods of work with the DB are
  described.
• The resume under chapter 2
• The Database Languages of the World contains
  detailed descriptions almost all languages of
  Europe, Northern and Central Asia and is one of
  two largest typological databases in the world.
• The architecture of the DB is focused on support
  of mathematical methods of researches.
• In the field of language typology and historical
  linguistics there is a number of unresolved
  problems and questions. The Database and
  connected mathematical models and methods could
  be potentially applied to resolve them.
• The resume under chapter 3
• The complex of software is developed, allowing to
  input and to edit the description of languages in
  the used format, search and comparison of
  languages.
• The DB is accessible in various variants DOS,
  Windows, Web and Excel-versions exist.
• The DB can be used for research, educational and
  reference purposes.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
8
Chapter 4 NEW POSSIBILITIES OF QUANTITATIVE
RESEARCHES IN THE FIELD OF LANGUAGE TYPOLOGY AND
HISTORICAL LINGUISTICS

• In the fourth chapter the key mathematical
  apparatus for the developed approach for
  calculation of measures of language similarity is
  described. In traditional typological researches
  languages are classified on the basis of a small
  number of features. The offered measures of
  similarity use all available data set and the
  strict mathematical apparatus. It gives new
  prospect in classification of languages. However
  the choice of an adequate measure of similarity
  in itself has appeared as a challenge. In the
  chapter the description of the general
  theoretical apparatus and a technique of a choice
  of the suitable measure of similarity is given.
  For calculation of measures of similarity and
  classification of languages the wide spectrum of
  methods is used. Also ordination of languages by
  means of statistical package R is applied.
• The resume under chapter 4
• The measure of similarity of languages is a base
  parameter which can be useful to the decision of
  many classification problems. However a choice
  from continual set of different measures of a
  concrete measure of similarity which would be
  adequate to studied object is a difficult
  independent problem.
• The general principles of construction of
  demanded measures of similarity are developed and
  by means of numerical experiments the group of
  typical measures is analysed.
• The software is created, allowing to build
  classifications of languages by means of strict
  methods of the cluster analysis.
• Possibilities of application of other methods of
  the mathematical analysis are shown. Ordination
  by means of the package of statistical researches
  R is to be seemed as the most perspective from
  this group of methods.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
9
Chapter 5 MODEL OF STRUCTURAL EVOLUTION OF
LANGUAGES

• In the fifth chapter the mathematical model of
  distribution of language features is described.
  There is a number of the postulates in its
  grounds that was formulated after the analysis of
  the DB. They correlate well with postulates of
  evolution of lexicon (formulated by Morris
  Swadesh). The new concept of LF-diagrams
  (reflecting dependence of number of features on a
  degree of their prevalence) is formulated. It was
  appeared to be a powerful tool of the analysis of
  mechanisms of borrowing.
• The resume under chapter 5
• The values of language features collected in a DB
  can be analyzed by many various ways.
• The analysis of distribution of the features
  existing exactly in two languages, has allowed to
  reveal defining character of borrowing for
  evolution of languages.
• The offered method allows to study contacts
  between the language groups, taking place during
  evolution, by means of strict mathematical
  methods.
• The LF-diagrams describing dependence of number
  of features from degree of their prevalence are
  entered. It is absolutely new interesting
  linguistic object of the researches which have
  appeared only after creation of the DB.
• The mathematical model of distribution of the
  features, explaining appearance of LF-diagrams is
  constructed. The offered model confirms recently
  stated hypothesis (Dahl Ö. 2004) about a
  competition of young features.
• The model allows to analyze dynamics of features
  in the whole DB and within separate language
  groups. The received results well correlate with
  earlier received classical methods.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
10
Chapter 6 DEVELOPING OF NEW METHODS OF
VERIFICATION OF GENETIC HYPOTHESES

• In the sixth chapter a new methodology of the
  analysis of genetic reletionship of the languages
  are entered . It is based on frequencies of
  features. Earlier in works on typology such terms
  as unical feature, statistical universal feature
  were used. In thesis for a doctor's degree of E.
  Yaroslavceva the concept of rare feature is
  entered. These terms are specified with reference
  to use in the DB and for objectives of this
  research.
• In the same chapter application of the offered
  methods to check of hypotheses about genetic
  relationship of languages is shown.
• The resume under chapter 6
• Classification of features on the frequency of
  the occurrence, specifying earlier offered by
  E.Yaroslavceva is proposed.
• The technique of application of the entered
  apparatus for automatic revealing of genetic
  markers (the features having strong correlation
  with relationship of languages) is described.
• The new method of an establishment of a
  generality of an origin of the languages, based
  on comparison of rare features is offered.
• The method is verified by its evaluation on
  groups of languages with precisely established
  relationship (Romance and German languages, the
  Altay macro-family).
• The method is applied to several languages
  isolates (Sumerian, Nivkh). The received results
  confirm some of earlier stated hypotheses.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
11
Chapter 7 THE PHENOMENON OF TYPOLOGICAL SHIFT

• In the seventh chapter numerical data on modern
  and extinct languages are compared. Developed
  above the techniques allows to reveal global
  tendencies of language development. In
  particular, the phenomenon of typological shift
  consisting in increase of frequency of
  high-frequency features and washing away of
  low-frequency is revealed.
• The resume under chapter 7
• Quantitative methods can be most effectively
  applied to the description of macroevolution of
  languages and language features.
• Presence of enough full descriptions of 52
  extinct languages allows to fullfill regular
  comparison of distribution of features in the
  languages existing on the average 2 thousand
  years ago, and modern. It gives the new tool of
  the analysis of language evolution.
• The phenomenon named typological shift is
  described. It means that for the specified
  interval of time frequency features
  (statistically universal) became even more
  frequency, and rare features have undergone to
  washing away.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
12
CONCLUSION

• In the book new techniques of quantitative
  researches on the material of DB Languages of
  the world are considered. The book describes the
  methods of calculation of pair measures of
  similarity, methods of the cluster analysis and
  the intellectual analysis of data.
• Quantitative portraits have been calculated under
  the genetic index of the DB. It was possible to
  reveal new statistical regularities of evolution
  of languages by means of these calculations . The
  phenomenon of typological shift on the Euroasian
  space is found out. It also explains presence of
  the strong typological background interfering
  revealing of genetic similarity of languages by
  direct computer methods.
• The facts and laws described have allowed to look
  in a new fashion at the phenomenon of the
  language variety and to construct new model of
  structural evolution of languages.
• The new technique of verification of the genetic
  hypotheses, based on using of rare features has
  been offered. The technique gives good conformity
  with traditional representations for languages of
  the Romance-German group, the Altay macro-family.
  Data in support of genetic relationship of
  Sumerian and Semitic languages, Nivkh and
  Chukotko-Kamchatkan have been got. The
  relationship of Old Japanese to the Altay
  macro-family has not proved to be true.
• As a whole it is possible to tell, that the
  Database Languages of the World is a valuable
  computer linguistic resource, and its value for
  scientific community will increase in due course
  only.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
13
POST SCRIPTUMThe review of the new results
received after the book exit

• Regarding a content
• Descriptions of 3 Baltic languages and language
  of Basques are formulated, results of examination
  of 34 languages are obtained.
• In the field of software products
• The new version of a site is developed and placed
  on the Internet
• The new interface concept of the DB is developed,
  the prototype model of the reference and
  educational version of the DB is executed.
• In the part of quantitative researches
• The new measures of similarity having an
  indicator of quality on 10 above, described in
  the book, are created.
• The technique of use of methods of philogeny on
  the DB is mastered, genetic trees are received.
• Parameters of typological shift are specified.
• Genetic markers on the families of Eurasia are
  revealed.

Visit to Max Planck Institute of Evolutionary
Anthropology, Leipzig, May 2008.
14
PPS. FUTURE DIRECTIONS

• Cooperation with WALS techniques
• Issue of version of DB for English-speaking
  scientist
• Improving all products
• Issue of reference book on content of the DB
• Verifying of main genetic hypothesis and
  information about ancient areal contacts between
  languages of Eurasia on the material of the DB
• New similarity measures discovering (on division
  of areal and genetic markers)
• and so on .

15

• Thank you for attention !