ACTA LINGUISTICA LITHUANICA

LI (2004), 1–14

Research possibilities for the morphology of Knygos nobažnystės (1653) using a relational database

DALIA JAKULYTĖ

Klaipėda University

The article presents the relational database created for the investigation of the morphology of old Lithuanian writings. A logical data model was developed and implemented within the Microsoft Access database management system (DBMS). All the words of two parts of the Knyga Nobažnystės were entered into the database. An engine was created for morphological (inflectional) data selection, so that the words can now be selected by any assigned property or set of properties. Such a database serves as a tool for data sorting, selection and correction, as well as for text and index media storage. It allows the researcher to deal with various linguistic problems, such as noun stem alternation, authorship of texts etc. in an efficient, quick and accurate manner.

0. INTRODUCTION

One of the greatest problems faced by linguists studying old texts is the selection of material. The main sources of material are the texts themselves (usually photocopies of books), dictionaries, and word registers. Data are transcribed from these sources (by hand or computer), paper card indexes are compiled, and they are sorted and arranged in a particular order. Selecting material from such a card index is a long and fairly complicated task: the entire index must be examined, the required cards selected, sorted, and so on; once the work is complete, the selected cards must be returned to their places in the general index, otherwise it will no longer be suitable for use.

Computer technologies help solve such problems. Computer corpora and databases of old texts are created for lexicographic research, etc.

One of the essential features of computerized corpora is the possibility of performing automatic text analysis, e.g., automatically generating text concordances, that is, presenting a searched-for word in a minimal context. For automatic language analysis, texts must be annotated so that a concordance program searches for linguistic units not only by formal but also by grammatical features. (Marcinkevičienė 1997). However, automatic programs for recognizing grammatical forms of language and the like can be created and applied only to the standard language, while the grammatical features of old writings must be annotated manually. For example, the Lithuanian-language corpus comprising 100 million words (donelaitis.vdu.lt/tekstynas), prepared at the Centre of Computational Linguistics of the Faculty of Humanities at Vytautas Magnus University, consists of publications from the period of independence, selected so that they as far as possible...

2 |

D A L I A J A K U L Y T Ė

The corpus reflects contemporary written Lithuanian. Its search system allows users to search for a word, obtain statistical information about a selected word, search for a specific grammatical form of a selected word (e.g., nepriklausomybės), or search for all forms of a word using the symbol “*” (nepriklausomyb*). The Institute of Mathematics and Computer Science of the University of Latvia has also made available online texts of old written monuments from the 16th–18th centuries with similar search options (users can search for a specific word or part of one using the symbol “%”), as well as archived files of a dictionary of word-form frequencies and an index (www.ailab.lv/senie). Similar work is also being carried out at the Institute of the Lithuanian Language: texts of old writings are entered or scanned using computers, and indexes of word forms are compiled. These are produced semi-manually: a computer program merely arranges the word forms alphabetically, indicating their location in the text and providing context. Textual information, as well as the grammatical and lexical meaning of a specific word form, is entered by hand (Ambrazas 1998).

The morphology of old texts has traditionally been studied using the same kinds of card indexes as those used for dictionaries, but with a different method of use: for example, entries are sorted not only by lexemes but also by other features (parts of speech, morphemes, derivation, form, etc.). Accordingly, the structure of the computer database and its material-selection capabilities must also be adapted.

The idea of creating a database for the study of the morphology of old texts arose in the course of continuing the research on Knygos Nobažnystės¹ begun by Antanas Jakulis. Until 2004, when a facsimile edition of Knygos Nobažnystės was published (Pociūtė, ed., 2004), Lithuanian linguists had access only to a microfilm made in 1968 (held at the Central Library of the Academy of Sciences). In that same year, the Centre for Baltic Studies at Klaipėda University acquired photocopies of Knygos Nobažnystės (Sumos evangelijų ir Maldų Krikščioniškų) made by A. Jakulis, while the library of the Faculty of Arts had xerox copies of the hymnal. Using these, A. Jakulis compiled a lexical register of Knygos Nobažnystės, of which only one part has been published—the vocabulary of Sumos evangelijų (Jakulis 1995). A paper card index of the entire vocabulary was compiled for this lexical register: words were written on individual cards, with their base forms, meanings, and grammatical forms noted (Fig. 1); the cards were grouped by base form and arranged alphabetically. However, while continuing his research into the language of Knygos Nobažnystės, the compiler himself used the index and resorted the cards according to a system known only to him. Essentially, it needed to be compiled anew, but the aim was to make it suitable for repeated use. It was therefore decided to create an electronic card index—a database for collecting, sorting, or filtering various morphological, lexical, semantic, and other data, and serving as a tool for researchers studying the morphology of old texts.

The creation of such a database involves two main stages: design and implementation. During the design stage, decisions are made about what kinds of data researchers studying the morphology of old texts need; the specific nature, features, and interrelationships of those data are defined; and the questions the database being developed should answer are determined. The result of this stage is a logical data structure. It is described in Section 2.1 of this article. The second-

¹ Plačiau apie Knygą Nobažnystės ir jos istoriją žr. Dainoros Pociūtės (2004), Inges Lukšaitės (2001), Zigmo Zinkevičiaus (1988: 213–223), Juozo Tumelio (1967, 1968) ir kt. darbuose.

KNYGOS NOBAŽNYSTĖS DUOMENŲ BAZĖ | 3

This stage—the implementation of the designed data structure in one of the database management systems—is described in Section 2.2. The created and implemented database of the morphology of old writings can now be used for research: that is, after loading the data and tagging the required features, the data can be selected, edited, and sorted as desired. The second section of the article describes the application of the created database of the morphology of old writings to the morphological study of a specific source—Knygos nob ažnystės.

1. CREATING A MORPHOLOGY DATABASE OF OLD WRITINGS

1.1. Logical data structure

The main elements of a lexical-index card file are the forms of words used in a text (here and throughout, each word form—that is, each occurrence of any lexeme—will be called a “word”). One card represents one word and its features (attributes) (see Fig. 1).

FIG. 1. KNYGOS NOBAŽNYSTĖS LEXICAL INDEX CARD

Each word is unique. Even if the same word, in the same form and with the same meaning, is used several times, each occurrence will be in a different place and a different context; if the cards were numbered, identical word forms would have different numbers.

Therefore, the primary element of the database must also be a word with a unique number. Its attributes are its form, source, base forms, meaning, etc. Some attributes (e.g., part of speech, paradigm) are shared by the entire lexeme. To define lexeme attributes, an element called “Lexeme” is created and linked to the “Word” element in a “one-to-many” relationship; that is, one “Lexeme” record may correspond to several “Word” records. The generalized logical data structure for the morphology of old writings is shown in Fig. 2.

4 |

DALIA JAKULYTĖ

FIG. 2. LOGICAL DATA STRUCTURE²

More details about the specific nature of the morphology data, the planned elements, their attributes, and the relationships between them are given in the article on the creation of the morphology database for Knygos nobanžystės (Jakulytė 2001).

1.2. Implementation of the Morphology Database

The electronic equivalent of a card index is a database. Specialized software—database management systems (DBMS)—is used to process, view, select, and print accumulated data, as well as to enter new data. Currently, relational database management systems are the most widely used (e.g., Visual

² An information element stored in a database—an entity—is represented by a rectangle in a logical data structure. The entity’s name is indicated at the top, and its attributes are listed in the lower section. A “one-to-many” relationship is represented by a branching line.

KNYGOS NOBAŽNYSTĖS DUOMENŲ BAZĖ

| 5

FoxPro, dBase, Paradox, etc.). The DBMS MS Microsoft Office suite includes Access, which has all the characteristic features of a relational DBMS and is well integrated with the other applications in the suite. Data are stored in tables. Each row in the main table is a separate record (a primary element and its attributes). Lists of recurring elements (attributes) are stored in auxiliary tables. Elements are entered into the main table, while attributes are selected from auxiliary tables. By linking tables, various queries can be made to select information from one or more tables (i.e., to retrieve the required cards from the card index and arrange them as desired).

Based on the logical data structure developed, the morphology database for old writings was implemented using the DBMS Microsoft Access 2003. Two main tables were created: “Words” (for word forms) and “Basic Forms” (for lexemes)³. The auxiliary tables contain lists of attributes. All auxiliary tables are linked to the main tables through a “one-to-many” relationship. The tables of the morphology database for old writings and the relationships between them are shown in Fig. 3.

FIG. 3. SCHEMA OF DATABASE TABLES AND RELATIONSHIPS BETWEEN THEM IN THE DBMS MS ACCESS

³ The original table names have been retained for now. They will probably not be changed—the names are rather difficult to change, and users generally do not see them.

6

Dalia Jakulytė

Some auxiliary tables have already been filled in, while others are adapted to each written monument. The completed tables are those for grammatical categories (shown on the left in Fig. 3). Each consists of several fields: an identifier, attribute name(s), and notes. Only the identifier is entered into the main table from an auxiliary table, but data from any field can be used when selecting data; therefore, for users’ convenience, some attributes are defined using both Lithuanian and international terms (cf. the list of case-category attributes in Table 1). The “Context,” “Chapter,” “Text,” and “Authors” tables are adapted to each written monument. The “Basic Forms” (lexemes) and “Meanings” tables are initially empty, but data can be imported from a database created for another written monument.

TABLE 1. LIST OF CASE-CATEGORY ATTRIBUTES

Case category

ID Case Latin Notes

0 -

1 nom. N. Nominative

2 gen. G. Genitive

3 dat. D. Dative

4 acc. A. Accusative

5 instr. I. Instrumental

6 ines. In. Inessive

7 illat. Il. Illative

8 ades. Ad. Adessive

9 allat. Al. Allative

10 voc. V. Vocative

21 poss. gen. G. posses. Possessive genitive

KNYGOS NOBAŽNYSTĖS DUOMENŲ BAZĖ | 7

2. USING THE DATABASE FOR MORPHOLOGICAL RESEARCH ON KNYGOS NOBAŽNYSTĖS

2.1. Data entry

Before entering morphological data for a particular written monument, the database is adapted. For example, when studying the language of KN, the text’s structure and authorship must be taken into account (see Jakulis 1982, 1984 for more detail). Different parts of the book were prepared or translated by different authors, so the “Text” table lists the text types (gospel, sermon, prayer, hymn...), while the “Authors” table lists known KN authors. The “Chapter” table is also filled in, listing all KN chapters (see Fig. 6).

Data entry begins with compiling the text. The compiled text was loaded into a table in line-by-line form, and identification numbers were assigned to the lines. The text was then split into words, which were assigned identification numbers; line IDs, text IDs, chapter IDs, and the original page and line numbers were recorded. All these operations were performed automatically using text-editor macros created by the author in Microsoft Word. Because word boundaries in this text often do not coincide with the spaces between words, the resulting table is reviewed and corrected. The reviewed tables are loaded into the corresponding tables of the morphology database, “Context” (renamed “KNtext”) and “Words.”

IDs for other features (gender, number, case, tense, etc.) are entered manually (not automatically) in the database’s “Words” table. For ease of entry, IDs are numbers corresponding to the ordinal position of each category; thus, for example, to specify the form of the word Kryžtaus (masculine singular genitive), one need only enter “1,” “1,” “2” in the respective fields (cf. Fig. 4).

FIG. 4. FRAGMENT OF THE “WORDS” TABLE. SPECIFYING A FORM

Before specifying the principal forms, the corresponding record is created in the “Principal forms” table, and is then entered in the “Principal forms” field of the “Words” table or selected from the drop-down list (Fig. 5).

8 |

Dalia Jakulytė

FIG. 5. FRAGMENT OF THE “WORDS” TABLE. SPECIFICATION OF LEMMAS

At present, the words from SE, Pas, MKr and K have been entered into the database, and those of their features that require no further research or have already been investigated have been marked. Thus, for example, following Antanas Jakulis’s hypothesis (Jakulis 1982), authorship of the SE sections has been marked, while only the identifier, slot, presumed lemmas, and part-of-speech ID have been entered in the “Lemmas” table. Lemmas can be revised and other features specified later, after reviewing all the word forms (see section 3.2).

Not all features are specified for several reasons:

– the features specified for selecting material for morphological research (location, form, and lemmas) are sufficient; many features are intended for other kinds of research, such as lexicological or semantic studies;

– some features are relevant only to a researcher here or there (e.g., word formation, base word, etc., will be of interest only to specialists in word formation), while some should be specified by the relevant specialist (e.g., meanings by a lexicologist), etc.;

– many features are debatable or uncertain; some depend on the point of departure for the research: for example, should the word kalnas ‘mountain’ be considered a root word or a suffixed derivative, that is, is it associated in the speaker’s mind with kilti, kelti ‘rise, raise’ or not? And what about delnas ‘palm’? How the root should be specified—as it is (kal-), or with the basic ablaut grade (kel-)—also depends on the researcher’s needs, for example, whether they want to retrieve words with a in the root or all derivatives of a particular root⁴.

– it is often difficult to determine the inflectional type as well (specified in the “Lemmas” table), since the Knygoje Nobžny5tės contains numerous variant forms, for example: bjauras / bjaurus, garba / garbė, and so on. In each such case, it is first necessary to conduct careful research and determine what accounts for the variation (perhaps it is determined by semantics, perhaps different authors use different forms, etc.);

– it seems that specifying the stem of a particular form (in the “Words” table) should cause somewhat fewer problems, but even here doubts arise: if ausis ‘ear’ has an i-stem, ausiai an io-stem, and ausų a consonant stem, then what kind of stem does ausį have?

Therefore, the researcher is left to enter such features. Different modes for viewing the material in the database management system are convenient for this purpose.

⁴Indeed, two fields are provided for this purpose: “Root” and “Base root”.

KNYGOS NOBAŽNYSTĖS DUOMENŲ BAZĖ

9

2.2. Reviewing, sorting, and editing the material

The material can be reviewed directly in auxiliary tables linked to the main table. In the “Chapters” table, we can view the entire text of each chapter (Fig. 6).

FIG. 6. EXCERPT FROM THE “CHAPTER” TABLE. REVIEW OF THE CONTENTS OF CHAPTER 13

10 |

Dalia Jakulytė

FIG. 8. DATA REVIEW IN THE “BASIC FORMS” TABLE

In addition, all data can be sorted by any attribute in ascending or descending order, and can be searched and filtered.

2.3. Data selection

For selecting the material, the DBMS Microsoft Access provides special queries (Query). Here the material (e.g., words) is selected according to any feature or group of features. For the user's convenience, forms are created in which the desired selection criteria are entered. The author has created such a form for selecting morphological data (inflection) (Fig. 9); forms are also being created for selecting vocabulary and word formation, etc.

FIG. 9. QUERY FORM FOR SELECTING INFLECTIONAL DATA

KNYGOS NOBAŽNYSTĖS DUOMENŲ BAZĖ

11

The required parameters are entered in the created form (their lists are provided in the lower part of the form). For example, to retrieve the present-tense forms of verbs of the mokyti, moko, mokė type, enter “511” in the “Inflection type” field and “1” in the “Tense” field. Clicking the “Select” button produces the following result:

TABLE 2. SELECTION RESULTS FOR PRESENT-TENSE FORMS OF VERBS OF THE MOKYTI, MOKO, MOKĖ TYPE

Lizdas Pagrindinės formos Žodis Forma Kontekstas Autorius Skyrius pildyti išpildyti, -pildo/-pildžia, -pildė įpildžia ties.es.3 204623 1 S14 laikyti užlaikyti, -laiko, -laikė vžiláyka sangr.ties.es.3 204224 1 S13 valgyti valgyti, valgo, valgė valgį ties.es.3 205703 1 S17 sakyti sakyti, sako, sakė fakote ties.es.dgs.2 206213 1 S18 klausyti klausyti, klauso, klausė klaufa ties.es.3 204727 1 S14 mokyti mokyti, moko, mokė mokia ties.es.3 200821 1 S02.....

Having noticed that present-tense forms occur with both o- and ia-stems, we refine the selection by specifying that only ia-stem forms be retrieved (stem: 2). The following result is produced:

TABLE 3. SELECTION RESULTS FOR IA-STEM FORMS OF VERBS OF THE MOKYTI, MOKO/MOKIA, MOKĖ TYPE

Slot Main forms Word Form Context Author Section rodyti rodyti, rodo / rodžia, rodė rodžiame pres.1pl 223129 1 S57 pildyti išpildyti, -pildo/-pildžia, -pildė įpildžia pres.3 204623 1 S14 rodyti rodyti, rodo / rodžia, rodė rodžia pres.3 201219 1 S03 mokyti mokyti, moko, mokė mokia pres.3 200821 1 S02 mokyti mokyti, moko, mokė mokia pres.3 200219 1 S01 mokyti mokyti, moko, mokė mokia pres.3 221313 2 S53

The selection can be further refined, for example, by retrieving all present-tense forms of each of these verbs:

12 |

DALIA JAKULYTĖ

TABLE 4. PRESENT-TENSE FORMS OF THE VERB IŠPILDYTI

SELECTION RESULTS

Slot

Main forms

Word

Form

Context

Author

Section

pildyti

to fulfill, -fulfills/-fulfills, -fulfilled

fulfills

pres. 3rd sg.

204623

1

S14

to fill

to fulfill, -fulfills/-fulfills, -fulfilled

fulfills

pres. 3rd sg.

304612

2

M36

By carrying out the selection systematically, narrowing the data range and refining the criteria, material can be gathered to address any specific question. For example, the data already presented in Table 3 make it possible to conclude that verbs of this type with the stem iš are more likely to be used by the first author, that is, the author of the first and last sermons.

CONCLUSIONS

The appropriate use of information technology tools makes it possible to considerably reduce the time required by researchers of old writings. The morphological database of old writings created by the author makes it possible to address various linguistic questions quickly and accurately (e.g., the mixing of noun stems, text authorship, and so on). The only time-consuming task is data entry, while data selection and sorting take just a few seconds; the data themselves are preserved and remain unchanged. The DBMS tools for searching, selecting, sorting, editing, and generating reports form a new part of the methodological toolkit available to the morphology researcher, while the logical model developed makes it possible to study language comprehensively and systematically. No less important is the fact that data can be imported into the morphological database from other sources, and can also be readily converted into other formats (e.g., the Historical Dictionary of the Lithuanian Language being compiled at the Institute of the Lithuanian Language).

The morphological database can be used not only to study morphology, but also to store data and present it in various forms to users with different needs. For example, a form could be created for theologians or literary scholars that would display the text for reading, explain unfamiliar words, and provide space for writing comments.

The Knygos nobanžystės morphology database created for a specific source is merely a tool, an intermediate step in which the results of current and future research on this book will be collected. Such a database can be published only after all the data (or some essential portion of it) have been entered, but this is not a one-person job. As a tool for studying the morphology of Knygos nobanžystės, it can also be used at present (it is available online at www.ik.ku.lt/~jakulyte/apie.htm).

ABBREVIATIONS

DBMS – Database Management System.

ID – Identifier, the unique name of a database item.

Knygos nobažnystės duomenų bazė

| 13

K – catechism, the final section of MKr (“KATECHIZMAS ABA TRVMPAS PAMOKSLAS Wieros Krikščionißkos del waykialu mažu.”).

KN – Knyga nobažnystės („KNIGA Nobazniftes Krikščionißzkos/ Ant gá r bos Diewuy TRAYCEY Szwentoy Wienatijam: Ant wartoi má Bažniczoms dides Kunigiftes Lietuvos ißduotá. KIEDAYNISE, DRVKAWOIA, IOCHIMAS IVRGIS RHETAS, Meatu Poná, 1653.“).

MKr – Christian Prayers (“MALDOS KRIKSČIONIŠKOS, Wifokiam meatu ir reykalams bendriems prigulinčios: Ant garbos Diewa Traycey Szwentoy wiena Tewá, Sunaus, ir Dwasios S. O del naudos Bažniczios io, ant ßwieta ißzlayftos <…> KIEDAYNISE, SPAUDE, JOCHIMAS JURGIS RHETAS, Meatu Poná, 1653.”). The third part of KN, containing 58 prayers and a catechism (see K).

Pas – Passion, the final section of SE (“PASSYA ABA HISTORYA Apie kančia ßmerti, ir pákałimą Poná mufu IEZVSA Chryftufá, iš wiku kiáturiu Ewangeliftu pátogay ßurinkta”).

SE – Sum of the Gospels (“SUMMA Abá Trumpas ißz̧guldimas EWANIELIV SZWENTU Per Wifus meatus/ dienomis Nedelos/ Bažnicžiofe Krikščionißkoſe ßkaytomu/ O ánt gá r bos Wienam DIEVUY, ſu wala Wiriaßniuju del pážitka Wierniems, didey Kunigiftey Lietuvos, ir del naminiku pabažnu ánt užiwoima namifzkia ißduotá. KIEDAYNISE, DRVKAWOIA, IOCHIMAS IVRGIS RHETAS, Meatu Poná, 1653.”). The second part of KN, containing 60 Gospel readings and sermons, as well as the Passion (see Pas).

LITERATŪRA

AMBRAZAS, S., ALEKNAVIČIENĖ, O., ZINKEVIČIUS V. 1998: Istorinis lietuvių kalbos žodynas ir senųjų raštų kompiuterizavimas. Lietuvių kalbotyros klausimai 39, 192–210.

BIRŽIŠKA, V. 1990: Aleksandrynas I. Antrasis fotografuotinis leidimas, Vilnius: Lietuvos kultūros fondas.

JAKULIS, A. 1982: „Sumos evangelijų“ ir „Maldų krikščioniškų“ autorystės klausimu. Baltistica 18 (2), 173–178.

JAKULIS, A. 1984: „Sumos evangelijų“ ir „Maldų krikščioniškų“ ryšiai su ankstesniais lietuviškais raštais. Kalbotyra 35 (1), 31–41.

JAKULIS, A. 1995: „Knygos nobažnystės“ leksika, Klaipėda: Sauliaus Jokužio leidykla-spaustuvė.

JAKULYTĖ, D. 2001: Informacinių technologijų panaudojimo galimybės senųjų raštų kalbos tyrimams („Knygos nobažnystės“ morfologijos duomenų bazė). Vārds un tā pētīšanas aspekti. Rakstu krājums 5, 152–168.

LUKŠAITĖ, I. 2001: Knygos nobažnystės (1653) parengimo kultūrinė aplinka. Knyga nobažnystės krikščioniškos (1653) – XVII a. Lietuvos kultūros paminklas, Kėdainiai: Kėdainių krašto muziejus, 4–29.

MARCINKEVIČIENĖ, R. 1997: Tekstynų lingvistika ir lietuvių kalbos tekstynas. Lituanistica 1 (29), 58–78.

POCIŪTĖ, D. 2001: Knygos nobažnystės giesmių šaltiniai ir kalvinistinio pamaldumo slinktys: nuo tikėjimo tikro prie tobulo. Knyga nobažnystės krikščioniškos (1653) – XVII a. Lietuvos kultūros paminklas, Kėdainiai: Kėdainių krašto muziejus, 41–54.

POCIŪTĖ, D., reng., 2004: Knyga nobažnystės krikščioniškos 1653, Vilnius: Lietuvių literatūros ir tautosakos institutas.

POCIŪTĖ, D. 2004: Lietuviškosios bažnyčios knyga. Knyga nobažnystės krikščioniškos 1653, Vilnius: Lietuvių literatūros ir tautosakos institutas, xiii–lxv.

TUMELIS J. 1967: Knygos nobažnystės krikščioniškos (1653) pirmojo leidimo klausimu. Bibliotekininkystės ir bibliografijos klausimai 6, 132.

TUMELIS J. 1968: Dar kartą apie „Knyga nobažnystės krikščioniškos“ (1653). Surastas pilnas pirmojo leidimo egzempliorius, Literatūra ir menas 44 (1146), 12.

ZINKEVIČIUS, Z. 1988: Lietuvių kalbos istorija III. Senųjų raštų kalba, Vilnius: Mokslas.

Dalia Jakulytė Klaipėdos universitetas Baltų kalbotyros ir etnologijos katedra H. Manto g. 84, 5808 Klaipėda

Gauta 2004 11 12