CC BY
38
УДК 81'34
Р. К. Потапова, В. В. Потапов
Потапова Р. К.1, доктор филологических наук, профессор, академик Международной академии информатизации, заведующая кафедрой прикладной и экспериментальной лингвистики, директор Института прикладной и математической лингвистики ФГПН МГЛУ; e-mail: [email protected]
Потапов В. В., доктор филологических наук, главный научный сотрудник НИР кафедры прикладной и экспериментальной лингвистики ФГПН МГЛУ, ст. научный сотрудник филологического факультета МГУ; e-mail: [email protected]
ОСНОВНЫЕ ПРОБЛЕМЫ РОССИЙСКОЙ СУДЕБНОЙ ФОНЕТИКИ
Статья включает следующую информацию: русский язык и языки ближнего зарубежья, стратегия судебных фонетических исследований в России, анализ современной ситуации, затрагивающей проблему идентификации говорящего по голосу и речи. Статистический и содержательный анализ вопросников, связанных с качественными характеристиками корпуса экспертов в сфере судебной фонетики, показал, что все группы экспертов нуждаются в лингвистах со специальными речевыми и языковыми знаниями в области классической фонологии, экспериментальной фонетики, речевой акустики, восприятия речи, вариативности и фонетической интерференции, прежде всего русского языка и языков ближнего зарубежья, а также русского языка и языков дальнего зарубежья.
Ключевые слова: судебная фонетика; экспериментальная фонетика; речевая акустика; восприятие речи; интерференция.
Potapova R. K.
D. Sc., Professor, Member of the International Academy of Informatization, Head of the Department of Applied and Experimental Linguistics, Director of Institute of Applied and Mathematical Linguistics, Faculty of the Humanities and Applied Sciences, MSLU; e-mail: [email protected]
Potapov V. V.
D. Sc., Principal Research Fellow, Department of Applied and Experimental Linguistics, Institute of Applied and Mathematical Linguistics, Faculty of the Humanities and Applied Sciences, MSLU; e-mail: [email protected]
MAIN PROBLEMS OF RUSSIAN FORENSIC PHONETICS
Abstract - The paper includes information on the following topics: Russian and the languages of the former USSR; the strategy of forensic phonetic investigations
1 R. K. Potapova for many years is a full member of International Association for Forensic Phonetics and Acoustics (IAFPA).
P. K. nomanoBa, B. B. nomanoe
in Russia; the analysis of the current situation around the issue of speaker identification by voice and speech. Statistical and content analysis of questionnaires concerning qualitative characteristics of forensic phonetic experts corpus showed that: all experts teams need more linguists with special speech and language knowledge in the field of classical phonology, experimental phonetics, speech acoustics, speech perception; variety and phonetic interference regarding first of all Russian and languages of the former USSR, Russian and foreign languages of other countries of the world.
Key words: forensic phonetics; experimental phonetics; speech acoustics; speech perception; interference.
I. Introduction
Some considerable difficulties and obstacles on the domain of forensic phonetics investigations in Russia is the multilingual communication in society and the process of the spoken language interference. Russian is the official language of the Russian Federation. It belongs to the eastern group of Slavic languages, which are a part of the Indo-European family of languages. Russian is one of the official languages of the UN. It is being used as the language for international dialogue in the former Soviet republics [1; 3; 6; 7; 8; 9; 14].
Languages of the peoples of Russia today include:
1. Indo-European group of languages:
a) Slavic languages: the Byelorussians, the Bulgarians, the Poles, the Russians, the Ukrainians;
b) Baltic languages: the Latvians, the Lithuanians;
c) Iranian languages: the mountain Jews, the Ossetins, the Tajiks, Tats;
d) New Indian languages: the Gypsies;
e) Romanic languages: the Moldavians, the Romanians;
f) Germanic languages: the Jews, the Germans;
g) Armenian: the Armenians;
h) Modern Greek: the Greeks.
2. Caucasian group of languages:
a) Cartvell languages: the Georgians;
b) Northern Caucasian languages:
- Abkhazian-Adygei languages: the Abazians, the Abkhazians, the Adygei, the Kabardinians, the Circassians;
- Nakho-Dagestan languages: the Ingush, the Lezghins, the Chechens and others.
3. Afro-Asiatic group of languages:
a) Semitic languages: the Assyrians; the Jews.
4. Ural group of languages:
a) Finno-Ugric languages: the Votics, the Karelians, the Komi, the Mari, the Mordvinians, the Lapp, the Udmurts, the Finns, the Khanty, the Estonians.
5. Altaic group of languages:
a) Turkic languages: the Azerbaijanians, the Bashkirs, the Kazakhs, the Karachais, the Kirghiz, the Tatarts, the Crimean Tatars, the Siberian Tatars, the Turkmen, the Uzbeks, the Khakas, the Chuvash, the Yakuts and others;
b) Mongolian languages: the Buryats, the Kalmyks;
c) Tungus-Manchurian languages: the Evenks, the Evens and others.
6. Paleoasian group of languages (a relatively defined language community):
a) Chukchi-Kamchatka languages: the Koryaks, the Chukchi;
b) Eskimo-Aleutian languages: the Aleutians, the Eskimos;
c) Yenisei languages: the Khets;
d) Uycagir-chuvanian languages: the Yucagirians;
e) Nivkhanian language: the Nivkhs.
7. Chinese-Tibetan group of languages: the Chinese.
8. Korean: the Koreans.
The Indo-European group of languages is represented in the Russian Federation by Slavic languages (e. g., the Russian language), and Iranian languages (e. g., the Ossetian language, the Tatian language). The Altaic family unites Turkic (Tatar, Chuvash, Bashkir, Yakut, etc.), Mongolian (Buryat, Kalmyk) and Tungus-Manchurian languages (Evenk, Even, etc.). Northern Caucasian languages include the Abkhazian-Adygei group of languages and the Nakho-Dagestan language group. The Ural languages are divided into Finno-Ugric and Samodian languages. Paleoasian languages include the Chukchi-Kamchatka languages. Many of the above languages are also found outside the Russian Federation (e. g., the Eskimo is found in the USA and Canada, the Lapp in Finland, etc.).
The analysis of a sample based on 130 languages of the former USSR allows to state the following [9; 13]:
- No less than 70 % - 57.7 % of subjects consider Russian to be the native language of their nationality.
Р. ^ Потапова, В. В. Потапоe
- No less than 70 % - 35 % of subjects have a good command of Russian.
- Russian is the most widespread language.
- 163.5 million named Russian their native language.
- 69 million named Russian their second language, which they master freely.
The change of the native language went faster in cities as well as in national groups living outside of the republics, in the surroundings of other peoples.
II. Method. Discussion
The main tasks of the forensic phonetics departments: methodical management of the forensic phonetic laboratories; development of a methodical base for conducting forensic phonetic examinations and research.
The methodical aspects of speaker identification by spoken language parameters and on searching the characteristic fragments of the speech signal having similar articulation and voice are based on automatic and semiautomatic models of speaker identification techniques.
Statistical and content analysis of questionnaires concerning qualitative characteristics of forensic phonetic expert corpus showed that:
- all expert teams need more linguists with special speech and language knowledge on the domain of classical phonology, experimental phonetics, speech acoustics, speech perception;
- all linguists in this teams need knowledge regarding the linguistic variety and interference of spoken languages of the former USSR (e. g., Russian - Moldavian, Russian - Georgian, Russian - Tatar, etc.);
- linguistic peculiarities of forensic phonetic investigations include substitution of several phonemes, intonation constructions especially for question-contour, rhythmic delimitation of spoken utterances, prosodic details of syllable structures, etc.
In recent years there has existed a steady tendency for growth in the number of forensic phonetic examinations conducted in the forensic phonetic laboratories of Russia.
The increase in the number of laboratories lead to a sharp rise in the number of experts. Each expert team includes linguists and engineers on acoustics.
• In the expert-criminalistics laboratories the following tasks are carried out: Technical examination of the phonogram: qualification of the available phonogram as the original or a copy; authentication of the phonogram: establishing signs of technical and electronic montage; identification of the recorder on which the phonogram in question was taped. Noise filtering of low-quality phonograms and retrieving the conversation content "word-for-word". Speaker identification via speech phonograms: differentiating utterances as belonging to different speakers and estimating the number of speakers involved in the conversation; determining the suitability of the speech signals for identification; speaker identification on the basis of speech phonograms.
• The hardware and software devices used: Intensive work is being carried out now aimed at creating a unified system for performing speaker identification on the basis of the "Technique of speaker identification through phonograms of Russian speech using the automatic system, e. g. versions "Fonex", "Ethnos".
• The systems provide:
• input of speech signals of unlimited duration to the IBM PC;
• segmentation of the speech signal, preparation of speech segments for analysis; preliminary estimation of the phonogram suitability for identification analysis;
• the acoustic analysis of the phonogram: calculating identification features, i.e. speech parameters received through the three-dimensional speech representation (intensity, frequency, duration) and reflecting the amplitude, spectral and time characteristic, and comparing them with the statistical decision thresholds received during the training of the system. Acoustic parameters include parameters that describe statistical characteristics of the amplitude-frequency spectrum of the speech signal in general (integral features), features of the spectral structure of speech fragments, comparable in a context (phrase and word features) and parametrical characteristics measuring the pitch, formants, sound duration (microanalysis features of sounds); the linguistic (aural) analysis ofthe phonogram: revealing, analysing and comparing linguistic features of speech with Russian pronunciation standards, using the atlas of Russian dialects, accents and other languages interference examples; making the identification decision [5; 10; 11; 12; 16; 17; 18].
• The system is aimed at solving the following tasks:
• registering voice and speech phonograms of the known person;
P. ^ Пomanoea, B. B. Пomanoe
• examining phonograms of the unknown person using the phonograms of known people;
• examining phonograms of the unknown person using the phonograms of anonymous speakers;
• storing phonograms of the unknown person (anonymous speaker).
III. Conclusion
The complete solution of the speaker identification task by voice and speech involves implementation of the following algorithm:
• Recording the initial spoken language text on the analogue / digital carrier or in the memory of the system.
• Auditory analysis of phonograms.
• Adaptive suppression of stationary and non-stationary noises in the recorded signal.
• Duration and frequency analysis of the speech signal.
• Semi-automatic segmentation of the speech signal.
• Calculating of speech parameters both in the temporal and frequency domains, relevant for speaker identification.
• Implementing the set of measures and classification rules within the framework of the speech database on making expert decisions on speaker identification [17; 18].
The other original method of speaker identification based on searching the characteristic fragments of the speech signal having similar articulation and comparing their high-frequency formant structure is described. The method is "transparent" from the theoretical standpoint and rather formal. In practice, the method is only one of the possible procedures used in phonogram analysis. It must and can be easily combined with other methods of speaker identification, based on the comparison of the formant structure of speech in particular, as well as with auditory analysis, articulation-linguistic analysis and psychological and social analysis [2; 4; 5; 10; 11; 12; 16; 17; 18].
The typical sequence of the actions taken by the expert: 1. Listening and preliminary viewing of the dynamic spectrograms of the phonograms being compared, preliminary search of the speech spectrum segments with a distinct and clearly structured formant pattern with the fourth and higher formants. Usually the minimal unit considered is the syllable.
2. Optimizing the graphic and numerical representation of dynamic spectrograms of the selected segments of the compared phonograms by the criterion of the correspondence of their formant structures to the theory (phonetics and acoustics of speech production) and the expertise of the expert. The following optimum characteristics are selected: the type of spectral analysis, spectral resolution and the order of spectrum model, the type of spectrum normalization and smoothing, the type and ratio of frequency, amplitude and time scales, the frequency range, the type of graphic representation, etc.
3. Choosing phonogram fragments where the dynamics of formant trajectories of the first four formants can be clearly defined and the disappearances of formants and occurrence of "extra" spectral maxima can be interpreted by the expert. It is important to state that the measurements taken are measurements of formants and not of the vocal tract characteristics or characteristics of the recording situation. The formant structure determined on the phonogram fragments chosen for comparison should be typical for the speaker.
4. Searching for segments of the second phonogram, corresponding to the selected segments of the first phonogram that have spectral cuts with conterminous formant frequencies for the first three formants with the presence on the spectrogram of the fourth and fifth formants. The phonemic quality of the pronounced sounds does not play a significant role. The method requires searching not equally pronounced phonemes, but identical instant articulations within the framework of an arbitrary phonetic context.
5. Defining the real accuracy of the current formant measurements. Probability calculation of arbitrary concurrency or non-concurrency of the spectral sections found and "equalized" on the first three formants. Estimation of the necessary number of concurrent or non-concurrent formant structures of the equalized fragments within the framework of the current measurements for making an accurate decision.
6. Comparing the frequencies of the fourth, fifth and higher formants for the found "equalized" spectral fragments. Making the decision on their concurrence or non-concurrence in each case. Searching for the necessary number ofarticulation independent pairs of the "equalized" spectral sections of the compared phonograms. Normally, 5 or 6 articulation independent, concurrent or non-concurrent pairs of "equalized" sections are enough.
Р. К. Потапова, В. В. Потапов
7. Making the identification decision. If the frequencies of the "higher"
formants of the spectral sections comparable in the three formants are
identical (non-identical), it can serve as a basis for making the identification
decision.
The following conclusions can be made:
• When studying the languages of the former USSR, alongside with the languages of the Russian Federation it is necessary to include the languages of the CIS and a number of former Soviet republics in order to reveal various forms of language migration, processes of convergation/ divergation regarding Russian, characteristics of the spoken language interference.
• It is necessary to pay greater attention to the process of the "dissolution" of languages, and phonetic interference regarding Russian and the languages of the former USSR.
• It is worthwhile to take a special look at the quantitative and qualitative parameters of the languages that form autonomous republics and autonomous regions of the Russian Federation and the CIS states.
• Apparently, it should be considered perspective improving the existing questionnaires taking into account the peculiarities of the Russian Federation and the CIS countries.
REFERENCES
1. Агеева Р. А. Какого мы роду-племени? Народы России: имена и судьбы. Словарь-справочник. - М. : Academia, 2000. - 424 с.
2. Голощапова Т. И. Идентификация иноязычных дикторов автоматическими методами: материалы Междунар. конф. «Информатизация и информационная безопасность правоохранительных органов». - М. : Академия управления МВД, 2010. - С. 372-377.
3. Государственные языки в Российской Федерации: Энциклопедический словарь-справочник / гл. ред. В. П. Нерознак. - М. : Academia, 1995. - 400 с.
4. Женило В. Р. Компьютерная фоноскопия. - М. : Академия МВД России, 1995. - 308 с.
5. Интерференция звуковых систем / отв. ред. Л. В. Бондарко, Л. А. Вербицкая. - Л. : Изд-во Ленинград. ун-та, 1987. - 278 c.
6. Красная книга народов России. - М. : Academia, 1994. - 117 с.
7. Лингвистический энциклопедический словарь / гл. ред. В. Н. Ярцева. -М. : Советская энциклопедия, 1990. - 685 с.
8. Население СССР: По данным Всесоюзной переписи населения 1989 г. / Госкомстат СССР. - М. : Финансы и статистика, 1990. - 45 с.
9. Потапов В. В. Краткий лингвистический справочник: языки и письменность. - 2-е изд., перер. и доп. - М. : Фонд «Развития фундаментальных лингвистических исследований», 2014. - 272 с.
10. Потапова Р. К., Потапов В. В. Язык, речь, личность. - М. : Языки славянской культуры, 2006. - 496 с.
11. Потапова Р. К., Потапов В. В. Речевая коммуникация: От звука к высказыванию. - М. : Языки славянских культур, 2012. - 464 с.
12. Потапова Р. К., Потапов В. В., Лебедева Н. Н., Агибалова Т. В. Меж-дисциплинарность в исследовании речевой полиинформативности / под ред. Р. К. Потаповой. - М. : Языки славянской культуры, 2015. - 352 с.
13. Энциклопедия: Русский язык / гл. ред. Ю. Н. Караулов. - М. : Большая Российская энциклопедия, 1997. - 721 с.
14. Языки народов СССР / под ред. В. Виноградова: в 5 т. - Т. 1.: Индоевропейские языки. - М.-Л. : Наука, 1966. - 658 с.
15. Языковая ситуация в Российской Федерации: 1992. - М. : РАН, 1992. -223 с.
16. Popov N. F., Linkov A. N., Fesenko A. V., Kurachenkova N. B., Baicharov N. V., Karlin I. P., Timofeev I. N., Potapova R. K. The interactive expert system for forensic speaker identification used in Russia // Intern. Workshop "Speech and Computer" (SPECOM'1999). - M., 1999. - P. 37-53.
17. Potapova R., Potapov V. Auditory and visual recognition of emotional behaviour of foreign language subjects (by native and non-native speakers) // SPECOM 2013. - LNCS, vol. 8113. - Heidelberg: Springer International Publishing Switzerland, 2013. - P. 62-69.
18. Potapova R., Potapov V. Associative mechanism of foreign spoken language perception (forensic phonetic aspect) // SPECOM 2014. - LNCS. -Vol. 8773. - Heidelberg: Springer International Publishing Switzerland, 2014. -P. 113-122.
