Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje

Application of Descriptive Terminology Research Findings in Terminology Practice JOLANTA KOVALEVSKAITĖ Vytautas Magnus University AGNĖ BIELINSKIENĖ Vytautas Magnus University KEYWORDS: descriptive terminology, terminology teaching/learning, terminology practice, informative context, specialized corpus INTRODUCTION The tools and methods used in descriptive terminology (also called terminography in Lithuanian linguistics) can be useful in practical terminology work. In this context, we understand terminology practice as an activity that encompasses not only the work of practitioners (terminologists, translators, editors, etc.), but also student training. Drawing on descriptive terminology research, terminology teaching can be organized so that practical problems are solved inductively, with terminology theory integrated into the teaching/learning process (cf. the English problem-solving principle, see Korkas et al. 2010; learning through discovery, see Alcina 2011). Terminology as a subject is included in linguistics, translation studies, and other programs; introductory terminology courses or courses tailored to specific subject fields are taught at various levels of study, and seminars for practitioners and summer schools are organized (Picht 2011: 21), yet little has been written about terminology teaching/learning to date (Alcina 2011: 8). Because of its broad applications in the information society, the importance of terminology teaching and the need for terminologists are growing (Bucher 2007). Recently, terminology skills have been strongly emphasized in discussions of the competencies required of translators (Korkas et al. 2010; Maskaliūnienė et al. 2012; Sikora 2013). The aim of this article is to present how the findings of descriptive terminology research (sources, tools, and methods) can be applied in termi- Terminologija | 2015 | 22 149 in terminology practice, since much attention is paid here to analyzing the use of terms in texts (or corpora). Analysis of term usage is useful in addressing issues relevant to terminology management: investigating the general and distinguishing features of a concept, determining the concept’s place in a concept system, evaluating the quality of a definition and/or compiling a corpus-based definition of a term, and so on. 1. KEY PRINCIPLES OF DESCRIPTIVE TERMINOLOGY One of the branches of descriptive terminology work—the corpus-based terminology work (Eng. corpus-driven terminology management, Budin 2012)—focuses primarily on term usage and the analysis of usage in a corpus. Context should be considered the most important factor in the formation of terminological meaning (cf. Zeller 2005); that is, it is precisely from context that one can determine whether a particular word or combination of words is a term or a non-term. This question often arises for practitioners working in the field of terminology. Descriptive terminology is characterized by different principles, sources, and tools for terminology management than prescriptive terminology (Marcinkevičienė 2000; Budin 2012). Prescriptive terminology relies mainly on documents, standards, approved term collections, or databases, whereas the primary sources for descriptive terminology are corpora, especially specialized corpora comprising texts from the subject field being analyzed (for more on compiling specialized corpora in Lithuanian, see Marcinkevičienė 2000; Bielinskienė et al. 2015). In descriptive terminology, specialized corpora are primarily needed to obtain new terms in the subject field being analyzed, but they are equally important for collecting other information about a term, such as definitions, contexts of use, collocational behavior, and frequency. Terms in such corpora are studied using automatic term recognition and analysis tools. The methods, tools, and sources of descriptive terminology can be used when carrying out, or learning to carry out, practical terminology work. 2. RESULTS OF DESCRIPTIVE TERMINOLOGY RESEARCH AND SCIENTIFIC PROJECTS In Lithuania, in addition to the ongoing terminology work and regularly updated resources of the Terminology Centre of the Institute of the Lithuanian Language and the Terminology Subcommittee of the State Commission of the Lithuanian Language (Republic of Lithuania- 150 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje ... (the Lithuanian Terminology Bank, Terminynas, the Terminology Database of the Standardization Department, EuroTermBank, etc.), research projects related to the compilation of Lithuanian terminology resources are also being carried out; for example, the SNOMED CT medical classification and terminology system1 and the LiDA bilingual (English–Lithuanian) social sciences thesaurus adapted for the Lithuanian language2 have been developed, among others. Several tools and resources adapted for Lithuanian, as well as a corpus-based methodology for defining terms, were developed during the ŠIMTAI 1 and ŠIMTAI 2 projects (VDU)3. Elements of this methodology (e.g., the notion of informative contexts) can be applied in practice when analyzing term usage and defining a term. 2.1. Resources and tools The study of the ŠIMTAI 1 and ŠIMTAI 2 projects was conducted in accordance with the principles of descriptive terminology; therefore, the first step was to prepare the necessary resource—a 4-million-word specialized corpus of education and science texts (automatically morphologically annotated). This corpus consists of documents from the Ministry of Education and Science and other institutions, covering education, higher education, and related areas (for more, see Bielinskienė et al. 2015: 24). The corpus represents the specific area of education and science selected for the study, so it was expected that documents of this kind would contain mostly terms from this field. On the basis of this corpus, the online Dictionary of Education and Science Terms (hereinafter ŠMTŽ) and the freely accessible online Ontology of Education and Science Terms were created (for more, see Markiewicz et al. 2013). Several tools were also developed: 1) a tool for automatic term identification in a corpus; 2) the tool JungLe, designed to assign a single headword form to several inflected forms of the same term (Boizou et al. 2012); and 3) the term-context analysis tool Qafe. Since data selected with the Qafe tool have been and can be used in terminology teaching and learning, we will present its operating principles in greater detail. The purpose of the tool Qafe is to automatically analyze the context of individual terms and identify the subject-field information conveyed in it, the so-called 1 http://www.lmb.lt/index.php?-538029741. 2 http://www.lidata.eu/idex_tez.php. 3 Automatic identification of education and science terms is a two-part project funded by the Research Council of Lithuania and carried out at the Centre of Computational Linguistics of Vytautas Magnus University in 2009–2010 (ŠIMTAI 1) and 2010–2012 (ŠIMTAI 2). Terminologija | 2015 | 22 1 5 1 informative contexts (see Section 2.2), in which the features of the concept designated by the term and its semantic relations to other concepts are apparent. Having received the contexts selected by the Qafe tool, a terminologist can more quickly systematize and analyze information about the concept designated by the term. The Qafe tool searches for paralinguistic patterns that reflect the structure characteristic of informative contexts (e.g., term definitions) and other elements of that structure—lexical, grammatical, and so on. It was established that definitions, which generally reflect a hyperonymic/hyponymic relation, can be found in a text by searching for patterns such as [Tn is N], [Tn – N], [Tn is known as], and so on4 (the patterns are discussed in more detail in Bielinskienė et al. 2015: 76–78). Using the set of the aforementioned paralinguistic patterns, the Qafe tool automatically selects potential informative contexts5. For example, when analyzing the term compulsory subjects, seven informative contexts were automatically identified, and nearly all of them were collected using the pattern [T– Nn], for example, compulsory subjects [T] – [dash] these are subjects [Nn], providing the foundations of knowledge and competencies in the chosen field of study. This informative context conforms to the usual structure of a definition: definitions generally present a hyperonym and distinguishing features when defining a concept. Thus, searching for informative contexts can also automatically identify ready-made definitions that may occur in corpus texts. The linguistic information used by the Qafe tool can be applied directly or indirectly. By using the data extracted by the tool directly, a researcher can automatically collect informative contexts and analyze them. If the tool is not used, its operating principles can still be applied; for example, knowing the structure of the patterns used by the tool makes it possible to search manually in a term concordance for passages that indicate subject-field information, and so on (for more on this, see Section 3). All or part of the information obtained from informative contexts can be used to define terms. 4 Tn – the nominative form of a term; N – a noun; is, is known as, a dash – lexical and graphic signs required in the context being searched. 5 Šiame straipsnyje ir kitur (pvz., Bielinskienė et al. 2015) vartojami terminai galimi terminai, galimi informatyvieji kontekstai. Tokia skirtis reikalinga, nes tarp automatiškai gautų duomenų gali pasitaikyti ir netikslių. 152 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje 2.2. Methodology for Defining Terms on the Basis of Corpora Since the ŠIMTAI 2 project aimed to provide corpus-based definitions for the terms described in ŠMTŽ, one of the project stages was devoted to developing and applying a methodology for defining terms on the basis of corpora. Subject-field information for defining terms was obtained automatically using the Qafe tool, which extracted contexts of a term that were considered potential knowledge-rich contexts. In the project mentioned, knowledge-rich contexts formed part of the linguistic resources prepared and used to define terms6. By analysing texts from a specific subject field or specialised corpora, it is possible not only to identify the usage characteristics of a particular term (frequency of use, most common contextual partners, and so on), but also to gather information about the concept denoted by the term. The main subject-field information about a term is conveyed through semantic relations, which are realised in knowledge-rich contexts (English knowledge-rich contexts; see Meyer 2001). Ingrid Meyer (2001: 281) regards a knowledge-rich context as one that contains at least one characteristic of the concept designated by a term. In other words, such a context must indicate a feature of the concept or a relation between that concept and other concepts. I. Meyer (2001) skiria du informatyviųjų kontekstų tipus: apibrėžiamuosius (angl. defining knowledge-rich contexts) ir aiškinamuosius (angl. explanatory knowledge-rich contexts). Apibrėžiamųjų kontekstų pagrindas – hiperonimijos ir hiponimijos, rečiau meronimijos7 semantiniai ryšiai. Informatyvieji kontekstai su hiperonimais ir hiponimais yra artimi klasikinei apibrėžties formai: T (terminas) = X (hiperonimas) + skiriamieji požymiai, pvz.: akreditavimas – tai vertinimas, naudojamas aukštojo mokslo sistemoje (vertinimas yra akreditavimo hiperonimas). Apibrėžiamaisiais kontekstais, kuriuose dažniausiai atsispindi esminė informacija apie sąvoką, paprastai pasinaudojama sudarant apibrėžtis. Explanatory contexts differ from definitional contexts in that they do not contain a hyperonym, but simply indicate one or more characteristics of a concept, which are not necessarily essential or distinguishing (Meyer 2001: 288). Explanatory contexts are important because they provide other subject-field knowledge about a term and its use; for example, they show what a term can, must have, usually is, is known as, and so on. Such contexts can be primary information 6 Other data obtained from the corpus are presented in the Dictionary of Education and Science Terms: inflected forms of the term, collocates, examples, concordance; see ŠMTŽ – http://daukantas.vdu.lt/moksliniai-terminai/. 7 Other works may use the terms partonymy and partonyms. Terminologija | 2015 | 22 153 when formulating a definition or be incorporated into the definition as a particular component (conveying a characteristic), e.g.: Graduates who have successfully completed the programme graduates will have acquired the set of skills required for a particular professional activity. Knowledge-rich contexts are also valuable because they are examples of authentic usage drawn from the concordance of a particular term. It is important to note that examples of authentic usage are not as ideal as some definitions presented in terminology databases or dictionaries of terms. Therefore, I. Meyer (2001: 284) calls contexts that have characteristics typical of dictionary definitions, such as a colon or dash after the term, dictionary defining knowledge-rich contexts (English dictionary defining knowledge-rich contexts). I. Meyer calls knowledge-rich contexts drawn from actual usage, whose sentence structures are freer and more varied, naturally occurring defining knowledge-rich contexts (English naturally occurring defining knowledge-rich contexts). These latter contexts make up the largest part of a term’s context in a corpus. Characteristics of a concept mentioned in such naturally occurring contexts are not necessarily distinguishing, just as the nearest concept mentioned is not always to be considered the closest hyperonym (English closest genus), as required when writing a definition. Thus, when analysing a term’s usage in a corpus, the knowledge-rich contexts often also contain information that is not necessarily essential. 3. ANALYSIS OF TERM USAGE IN A SPECIALIZED CORPUS The following aspects are very important for terminology and terminography: term, concept, concept characteristics, concept relations, concept systems, and principles for writing definitions. Therefore, when analyzing term usage, it is important to identify the general and distinctive characteristics of a concept, its place in the concept system, assess the quality of the definition and/or prepare a corpus-based definition of the term. In prescriptive terminology, terms are defined primarily by considering the place of the concepts they designate in a hierarchical concept system and their relations to other concepts (Marcinkevičienė 2000: 6). Thus, the ability to identify the general and distinctive characteristics of a concept is essentially needed in order to learn about the concept itself and/or prepare a definition of the term. Using a specialized corpus from a particular field, one can examine concordances of specific terms and thereby determine what concept a term designates, what its characteristics are, and which other concepts it is related to. In this way, the concept system is constructed “bottom-up,” that is, on the basis of information about term usage. 154 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje 3.1. Analysis of term context from the perspective of domain information For information about the concept designated by particular terms to be found in a specialized corpus, the corpus must represent the subject area, and the term must occur fairly frequently in it; the researcher will then obtain a larger concordance and, consequently, more information about the term. Examples from the concordance of the term scientific work in the corpus: (1) The third reason is the bureaucratization of scientific work. (2) Scientific work was most often carried out in Germany, France, Poland, Latvia, and Sweden (Table 4 of Appendix 3). (3) Research (otherwise known as scientific work) comprises research classified as basic and applied research. (4) In addition to scientific work, researchers must also engage in teaching. It can be seen that the same term is used in different ways in these sentences: in sentences 1 and 2 the term is mentioned, but although the discussion concerns matters related to scientific work, no domain information about the concept designated by this term is provided. Examples 3 and 4 also mention other terms related to the concept scientific work: research, basic research, applied research, researchers, teaching. Such fragments of a term’s context are particularly important for terminography, because they can help explain how the entity designated by the term is understood, what its purpose or nature is—this helps define the concept designated by the term. Thus, a terminographer needs to select from the entire term context only those passages that mention other terms related to the term being analyzed or provide ready-made definitions. If a large number of term-concordance lines have to be processed manually and the concordance is extensive, analyzing the term in the corpus becomes inefficient and time-consuming. After all, identifying the most informative contexts requires reading the entire concordance and examining the content of each sentence’s context, which is not easy for a linguist, translator, or student who lacks specialized knowledge of the terminology of the field under study. In this case, it is useful to employ tools that at least partially automate the analysis of term contexts. A tool such as Qafe, described in Section 2.1, functions as a kind of filter, separating the most informative contexts from other contexts of the term that are less important because they provide no information about the concept. Of the terms described in the 224 ŠMTŽ entries, about 60 percent of the definitions were written on the basis of information from informative contexts. Thus, Terminologija | 2015 | 22 1 5 5 When compiling definitions for these terms, the terminologist either adopted a ready-made definition found in the corpus or formulated one on the basis of informative contexts and/or all the term’s concordance lines. Some informative contexts were not used because the automatically selected set also includes unsuitable contexts—that is, contexts that, although they formally match one of the paralinguistic patterns, nevertheless reflect neither a semantic relation nor any feature of the concept being analyzed (for a detailed discussion of how the patterns used by the tool Qafe might be improved and the accuracy of automatic analysis increased, see Bielinskienė et al. 2015, section 5.2.5). 3.2. Questions relevant to terminology management and their resolution based on analysis of term usage In addressing issues that arise in terminological work, informative contexts of both suitable and unsuitable quality can be used. Four principal theoretical and practical aspects of terminology management are discussed below (for these and other aspects of terminology theory relevant to terminology management, for example the problem of distinguishing a term from a non-term, see also Korkas et al. 2010: 129–130). The analysis is based on concordance lines for terms in the specialized corpus and/or on fragments of the concordance lines of a particular term drawn from informative contexts identified by the tool Qafe8. 3.2.1. IDENTIFYING FRAGMENTS OF TERM CONTEXT THAT CONVEY SUBJECT-SPECIFIC INFORMATION To develop the skills needed to identify in a text the features of the concept denoted by a term and the characteristics of the term’s usage, one should analyze a concordance for a term of medium frequency. Drawing on the notion of informative context (described in section 2.2), this makes it possible to learn to locate parts of a term’s context that are significant from a terminological perspective. As noted, an informative context should indicate: (a) at least one feature of the concept denoted by the term, which need not be essential but should characterize the concept in some respect—that is, indicate how the concept is understood, the purpose or features of the relevant subject, and so on—or (b) a relation to other concepts. The contexts below show that the concept denoted by the term academic degree is characterized by the fact that it is awarded by a university (example 1), and also 8The concordances for all terms included in ŠMTŽ can be consulted. The concordances do not separately mark possible informative contexts, and until the tool Qafe is available to all users, informative contexts will have to be identified manually. 156 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje an academic degree is required when applying for certain positions (example 2): (1) The University has the right to award academic degrees when the Ministry of Education and Science grants it the right to doctoral studies, either independently or jointly with other Lithuanian or foreign research and higher education institutions. (2) Therefore, only one qualification requirement should be set for this position—to hold an academic degree, the completion of which demonstrated the abilities and qualifications required of a research staff member. A further review of the context of this term can reveal what kinds of academic degrees exist (doctoral and habilitated doctorate degrees), how they are obtained, and so on. Essential information about the concept (its principal distinguishing feature) is conveyed by the definition found in the context (for definitions, see 3.2.4): Academic degree – a degree attesting to a person’s academic qualification, awarded upon successful defense of a dissertation. Incidentally, this definition is also included in ŠMTŽ as the one that best captures the essence of the term. Not all automatically identified informative contexts are suitable—that is, not all reflect features of the concept or its relation to other concepts. For example: In grade 8, economics is a compulsory subject, with one lesson per week. This context for the term compulsory subject is classified as unsuitable because it conveys no subject-specific information about the term. Although we learn that economics can be a compulsory subject in grade 8, this information does not reveal the features of the concept (that is, what a compulsory subject is). In the next example, too, we find no information significant from a terminological perspective about the concept denoted by the term national qualifications system, its nature, and so on: The National Qualifications System is a priority policy area. If the term concordance is large and the Qafe context-filtering tool is unavailable, the same task can be performed using elements of the paralinguistic models applied in automatic analysis (see 2.1), which can be used to search for informative definitional or explanatory contexts. For example, structural elements such as be, consider, constitute, comprise, as well as a dash, colon, etc., can be used to identify definitional contexts (cf. Castagnoli 2006: 167). Terminologija | 2015 | 22 157 When analyzing the use of Lithuanian terms, one can observe that such words and graphic signs are common in definitions, e.g.: The University Council is a collegial governing body. A department is the main academic unit of a faculty, carrying out study programmes or parts thereof, as well as research. The author of a scientific work is considered to be a natural person. A research institute consists of centres, departments and/or research laboratories. Scientific activity also includes scientific teaching, research services, the translation and editing of scientific literature, etc. Furthermore, it should be borne in mind that definitional contexts generally convey semantic relations between a term and another concept (hyperonymy/hyponymy or meronymy) (for semantic relations, see 3.2.2). Explanatory contexts can be identified by lexical elements recurring near the term, such as can, must, must be, can be, may have, is + a participle, and other frequently recurring verbs, especially in the present tense, e.g.: The right to award academic degrees is held by the University. Higher education must be equally accessible to everyone. The study system is based on the number of credits. Thus, searching texts for the aforementioned elements (which are often also collocates of a term in a concordance) can help locate informative contexts for terms more easily and use them in further terminological work: to determine the systematicity of terms (see 3.2.3) and formulate definitions (see 3.2.4). As mentioned earlier, some automatically selected informative contexts are unsuitable. In addition, for various reasons, some informative contexts are not automatically recognized (this problem is discussed in greater detail in section 5.2.3 of Bielinskienė et al. 2015). Importantly, unsuitable or unselected contexts can also be used to learn to identify informative contexts and other relevant information about the concept denoted by a term. By comparing the informative contexts for a particular term selected automatically by the Qafe tool with the full concordance for that same term, it is possible to identify those the tool has missed. For example, two informative contexts were not automatically identified in the 54-line concordance for the term compulsory subject: (1) Compulsory subjects constitute the foundations of general university education and education in the chosen field of study. (2) The general university education subject block is divided into two subgroups: compulsory subjects and alternative elective subjects. 158 Jolanta Kovalevskaitė | Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje Agnė Bielinskienė By analyzing such cases, it is possible to determine the reasons why these contexts were not automatically selected by the tool or why they did not fit any of the models used. Sentence structure is often very complex and does not reflect the models used by the tool, or the concepts expressing semantic relations are distributed across a much broader context than that covered by the model. 3.2.2. IDENTIFICATION OF SEMANTIC RELATIONS BETWEEN CONCEPTS IN THE TERM CONTEXT An example of a term's usage is considered a defining context if it reflects the semantic relations conveyed between concepts. By evaluating informative contexts selected automatically or manually, one can develop the ability to identify other concepts in the domain under analysis and to analyze the semantic relations between these concepts. For informative contexts, and especially for defining contexts, which differ from explanatory contexts precisely because they convey semantic relations, the most relevant relations are hyperonymy/hyponymy and meronymy. Using the methodology described in (2.2) as a theoretical basis, one can learn to identify certain elements in a context that may be perceived as indicators of a semantic relation. First, one can examine lexical units that are usually found in constructions conveying hyperonymy/hyponymy and meronymy. Attention should be paid to the previously mentioned lexical elements be, consider, which indicate that hyperonyms and hyponyms may be expected in the sentence. In the example Science valleys are primarily research centers in which the potential of Lithuanian scientists will be concentrated; the best scientists, it is expected, and foreign scientists will also work there it can be seen that the structure of the sentence fragment consists of three parts: the term (science valleys), which in this case is a hyponym; the relation indicator (are); and the phrase (research centers), which is a broader concept than that of the term under analysis—a hyperonym. In another example The author of a scholarly work is considered a natural person the hyperonym is natural person, the hyponym is the term author of a scholarly work, and the link between them is the word considered. Thus, defining contexts must contain a hyperonym, which, in addition to indicating a concept's characteristic, also specifies the term's place in the conceptual system. Incidentally, it should be borne in mind that sometimes the term under analysis itself may become a hyponym of another, more general concept. Terminologija | 2015 | 22 159 Another semantic relation—meronymy—is identifiable from the words include, comprise, whose lexical meanings themselves already indicate that the constituent parts of an entity will be listed: Research institutes comprise centres, departments and/or research laboratories. Research activity also includes research training, research services, and the translation and editing of scholarly literature, etc. These semantic relations can also be identified by the fact that hyperonymy / hyponymy and meronymy are conveyed visually through elements that set text apart, such as certain punctuation marks: dashes, parentheses, and colons. Semantically related concepts occur to the left and right of these marks or between them (e.g., between parentheses): Research centre – a structural unit of a university or faculty (the hyperonym of research centre is unit); <...> per kokį laikotarpį šie mokslo darbai (mokslo straipsniai) yra paskelbti ir kokiuose mokslo leidiniuose (mokslo darbų hiponimas mokslo straipsnis); Su jais tiesiogiai siejamos ir kai kurios pagrindinės šių mokslų kryptys: istorija, filologija, menotyra, etnologija, komunikacija ir informacija (mokslų krypties hipo-nimai yra istorija, filologija, menotyra, etnologija ir t. t.). Since these punctuation marks denote both types of semantic relation, it is difficult to differentiate between them automatically. However, one can observe that in the case of meronymy several concepts are listed consecutively, and conjunctions are sometimes inserted as well: Research activity comprises the creation, accumulation and dissemination of scientific knowledge, research and experimental development, as well as innovation activity. Terms that are synonymous can also be identified in a text by similar indicators. In Lithuanian-language texts, relations of identity between term meanings can be identified by definitional verbs call, say, the conjunction or, and so on. A synonym is often given in parentheses, sometimes with a clarifying abbreviation (i.e.) (Stankevičienė 2013: 135–138). In analysing terms in the fields of education and science, informative contexts containing synonyms were not sought, but it is evident that synonyms between terms can also be identified in context on the basis of formal features. An informative context may also contain no semantic relations, even though its usage pattern is suitable for identifying semantic relations, e.g.: The most highly evaluated part of the programs is research work. National Qualifications System is a priority policy area. 160 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje The examples show that scientific work is not a hyponym of the overly abstract phrase part of a programme, but national qualifications system – policy direction is. Thus, it should be remembered that not all automatically identified contexts will convey systematic semantic relations. In many cases, this may also be determined by the nature of the texts in the corpus: in texts such as press releases and the like, concepts are not explained but merely mentioned, so the term contexts obtained from such texts are unlikely to contain systematic concepts. 3.2.3. SYSTEMATICITY OF TERMS IN TEXTS. IDENTIFYING THE GENERAL AND SPECIFIC FEATURES OF THE CONCEPT UNDER ANALYSIS Typically, if terms are obtained from a corpus, they have one major advantage: they are systematic and interrelated (Marcinkevičienė 2000: 7). This is because texts of certain types (e.g., textbooks, administrative documents) present entire sets of mutually compatible terms (Rimkutė 2010: 72). Thus, a term’s informative contexts show which other concepts the concept under analysis is related to. In the example Compulsory subjects form the foundations of general university education and education in the chosen field of study we can identify several concepts: the concept under analysis, compulsory subject, and other concepts related to it: general university education, field of study, foundations of education in a field of study. It follows that we can learn that the concept field of study includes the concepts general university education and education in a field of study, and that compulsory subjects belong to them as well (other contexts also mention elective subjects). By analysing related concepts, it is possible to understand their relationships and construct a hierarchical system of concepts, as well as distinguish the general and distinguishing features of a term. This is very important when compiling terminological dictionaries and/or creating term bases. Ontologies, which record relationships between concepts, are a useful resource for analysing conceptual systems. In the field of education and science, the most recent publicly available ontology in Lithuanian is the Ontology of Education and Science Terms (Markiewicz et al. 2013). This term ontology comprises 465 general concept classes, whose hierarchy is based on definitions of education and science terms prepared using a specialised corpus. According to the definition of the term rector (A single-person university governing body that directly manages Terminologija | 2015 | 22 1 6 1 the activities of the university, acts in its name and represents it), this term is assigned to the “administrator” subclass within the “participant” class, as are the terms vice-rector, dean, vice-dean, which share the same feature (see figure). Fig. Fragment of the Ontology of Education and Science Terms: the “administrator” subclass within the “participant” class Thus, by checking a particular term in the field of education and science against the ontology, we can see which other concepts in that field it is related to. When analysing the systematicity of terms, including when writing definitions, it is important to identify the general and distinguishing features of a concept. For example, to identify the general and distinguishing features of the concept compulsory subject, it is necessary to find informative contexts that mention the nearest concept at the same level. Compulsory subject – a subject in a study programme that is compulsory to study and provides the foundations of general university education and education in the chosen field of study. By examining the informative contexts of the two related terms compulsory subjects and elective subjects, it would already be possible to specify the general feature shared by the concepts mentioned above (“a subject in a study programme”) and the distinguishing features of the concept compulsory subject (“study is compulsory”, “provides the foundations of general university education and education in the chosen field of study”), as expressed in the definition given in ŠMTŽ (A subject in a study programme that is compulsory to study and provides the foundations of general university education and education in the chosen field of study). 162 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje 3.2.4. FORMULATING AND/OR EVALUATING A TERM DEFINITION Another important stage of terminological work is formulating a term definition or evaluating an existing definition. As is well known, in descriptive terminology all stages of terminological work are linked to a corpus and its use (Marcinkevičienė 2000). The same applies to formulating a definition. Having examined the concept’s characteristics and the term’s use in the corpus, one can formulate a term definition based on corpus data and the central idea of descriptive terminology that context is the factor shaping terminological meaning. Studying context brings the concept into sharper focus and avoids a possible mismatch between the theoretical content of a term and its content in actual use (Zeller 2005: 314); the resulting definition is therefore more precise. Definitions are formulated in several ways on the basis of corpus data. One approach, also used in compiling ŠMTŽ, is to draw on definitions found in the term’s context. As mentioned earlier, informative contexts may contain ready-made definitions, which can simply be adopted. For example, the definition chosen for the term academic degree was one found in the corpus, since the term’s use in the corpus corresponded to the distinguishing features recorded in that definition: Academic degree – a degree attesting to a person’s academic qualification, awarded upon successful defense of a dissertation. If informative contexts contain no ready-made definitions, an overall picture of the concept and its definition can be assembled by combining information from several informative contexts or concordance lines. For example, ŠMTŽ gives the following definition: Peer review – a method of evaluating and ensuring the quality of research, in which a scholarly work is reviewed by other researchers in the same field or branch of science. This definition is assembled from the following concordance lines (the information used in the definition has been spaced out in the examples): Peer review is one of the m e t h o d s o f e v a l u a t i n g r e s e a r c h. It was thought that quantitative indicators could not reliably reveal the quality of research; accordingly, peer review was considered the only valid m e t h o d o f e v a l u a t i o n. Peer review means that o t h e r r e s e a r c h e r s i n t h e s a m e f i e l d assess research articles for validity, significance, and originality <...>. Terminologija | 2015 | 22 163 A suitable definition found among definitional contexts may be reformulated to reflect the topic prevailing in informative contexts (e.g., where the legal aspect predominates), an essential feature, or frequently recurring formulations (Bielinskienė et al. 2015: 100): Academy of Sciences – an institution or organization bringing together the country's most distinguished scholars or (and) scholars in a particular field (or fields) of science, and fulfilling representative, expert, coordinating, and science-popularization roles on behalf of science. In the example presented, the first distinguishing feature (bringing together <...> scholars) is taken from the defining context (cf. Lithuanian Academy of Sciences (hereafter – Academy of Sciences) is a budgetary institution bringing together the most distinguished Lithuanian scholars and foreign scholars whose work is connected with Lithuania), while the function performed (fulfilling representative roles on behalf of science <...> ) is formulated on the basis of the predominant information in the concordance. The methods mentioned for writing a definition on the basis of a corpus may be useful to those writing definitions (most relevantly, terminologists) and to those editing definitions found in other sources (translators compiling terminology databases and, of course, editors). Indeed, examining usage may reveal that an existing definition is incomplete because the concept denoted by the term has changed, or for a similar reason. One can check how well a corpus-based definition was written, or evaluate it, by comparing one's own definition with a terminologist's definition written according to the same principle, for instance, one provided in the ŠMTŽ. For example, drawing on informative contexts and an ontology, students formulated the following definition for the term scientific article: Definition 1: A peer-reviewed scholarly publication with a formal structure, published in periodical, continuing, or one-off scholarly publications. They compared their own definition with the definition of the term scientific article given in the ŠMTŽ: Definition 2: An article published in the scholarly press and containing the scholarly apparatus customary in a particular field of science, as well as meeting other criteria for scholarly quality in that field. Definition 2 is based on an adjusted and shortened defining context from the corpus: Defining context: Scientific article – an article published in the scholarly press or (and) containing the scholarly apparatus customary in a particular field of science (footnotes or (and) a bibliography, or (and) formulas, or (and) diagrams, or (and) a description of the methodology, or (and) statistical tables, and so on), and meeting the criteria for scholarly quality recognized in that field. 164 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje As can be seen, in definition 1 the hypernym is chosen more precisely (scientific publi-cation rather than article), mentioning the distinguishing feature “peer-reviewed” and thus differentiating it from non-peer-reviewed publications. Both definitions indicate where a scientific article may be published, but definition 1 specifies this in greater detail—published in periodical, continuing, or one-off scientific publications—whereas definition 2 is more concise—published in the scientific press. There would be no need to list these if the conceptual system included the concept scientific press. In definition 2, the concept scholarly apparatus is explained in detail in the defining context; therefore, instead of the feature named in definition 1, having a formal structure, the more precise feature would be the one taken from the defining context: having the scholarly apparatus customary in a specific field of science. In practical work, one often has to choose the most suitable definition from among several already given in other sources. The principles set out in terminology theory (adequacy, systematicity, conciseness, etc.) can, of course, help with this. Nevertheless, in addition to these principles, it is useful to compare the information about a term found in lexicographic sources with the information that emerges from the term’s usage in a corpus. Terms are often used in ways that do not entirely match their definitions (Kaupaitytė 2010), or usage reveals that the concept designated by a term does not entirely correspond to the concept given for that same term in a lexicographic source. Comparing a definition written on the basis of a corpus with definitions given in dictionaries or terminology databases makes it possible not only to verify the term’s definition, but also to compare definitions written according to different principles. For example, in the Dictionary of Study Terms (Smetona, Pukelis 2014), some terms are the same as those given in ŠMTŽ. Let us compare the entries for the term course unit in the two dictionaries: in the Dictionary of Study Terms (Smetona, Pukelis 2014): A part of a study programme consisting of several study outcomes that are related and integrated in terms of study content, with a defined number of credits and an objective, and based on study outcomes whose achievement is assessed in a single final assessment (an examination or another form; a cumulative grading system may be applied). ŠMTŽ: 1. The principal constituent unit of a study programme, taught continuously over one or several semesters; 2. A field or branch of study (in international statistics). Both definitions mention the essential feature that a course unit is part of a study programme; thus, the study programme is the core concept with which the term under analysis is associated in both cases. The term’s concordance Terminologija | 2015 | 22 165 Since the corpus contains no information about such concept characteristics as a specified number of credits, an objective based on study results, the achievement of which is assessed in a single final assessment, they are not included in the ŠMTŽ definition. The duration of instruction is a characteristic that emerged in the terms’ informative contexts; therefore, the ŠMTŽ definition notes that a study subject is taught continuously for one or several semesters. The informative contexts revealed that this term denotes two concepts, so ŠMTŽ provides definitions for both concepts (1, 2). It may be added that definitional contexts can be useful as examples of better or poorer definitions, by analyzing which terminographers acquire the skill to distinguish an adequate definition from an incomplete or flawed one (Meyer 2001: 284). The article presents examples from the fields of education and science because the corpus used as a source was specifically from these fields. However, practitioners dealing with terminology in other fields may also need corpora from those fields. Thus, when applying descriptive terminology methods, in addition to existing corpora, it is possible to use corpora created with special tools⁹ for the intended task (Do-It-Yourself (DIY) specialized corpora, Zanettin 2002, cit. from Castagnoli 2006). To compile such a corpus quickly, texts in the relevant field published online are most often collected, as they reflect current concepts and the latest terms—something that cannot always be expected of a static corpus, i.e., one compiled during a certain period and no longer updated (Castagnoli 2006: 160). CONCLUDING REMARKS So far, only a few resources and tools have been developed in Lithuania using the principles of descriptive terminology. The first attempts, undertaken in the ŠMTTAI 2 project, to obtain data for a terminology dictionary and ontology in the fields of education and science may be applied to address current terminology issues encountered by practicing specialists who work directly with texts and obtain terminology from them or create terminology for a particular field themselves, as well as by students studying terminology, translation, and editing. The possibility of analyzing terms directly in their contexts of use can help identify and address issues relevant to terminology. The methodology for defining terms on the basis of corpora discussed in this article provides 9 For example, the WebBootCat tool offered by SketchEngine: http://www.sketchengine.co.uk/. 166 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje the possibility of identifying a term’s informative contexts, determining hierarchical relations among concepts, and formulating term definitions. The aspects of terminology work described here are not equally relevant to the target groups of researchers mentioned: for instance, terminologists are most often responsible for writing a definition, whereas translators or editors filling a terminology database may need to choose the best one from several options or edit a draft definition. For terminologists, translators, and editors alike, if the terms being analyzed are absent from terminology databases and dictionaries, it may be important to identify the connections or types of relations among the concepts denoted by the terms, as well as to define those concepts. ŠALTINIAI Švietimo ir mokslo tekstynas. VDU Kompiuterinės lingvistikos centras, 2010. Švietimo ir mokslo terminų ontologija. Prieiga per internetą: http://meta.vdu.lt/webprotege/ (žiūrėta 2015-05-15). ŠMTŽ – Švietimo ir mokslo terminų žodynas. Prieiga per internetą: http://daukantas.vdu.lt/moksliniai-terminai (žiūrėta 2015-06-15). LITERATŪRA Alcina A. 2011: Teaching and learning terminology: new strategies and methods. – Teaching and learning terminology: new strategies and methods (ed. A. Alcina). Benjamins Current Topics 35, Amsterdam: John Benjamins, 1–11. Bielinskienė A., Boizou L., Grigonytė G., Kovalevskaitė J., Rimkutė E., Utka A. 2015: Lietuvių kalbos terminų automatinis atpažinimas ir apibrėžimas. Monografija, Kaunas: Vytauto Didžiojo universitetas. Boizou L., Grigonyte G., Rimkutė E., Utka A. 2012: Automatic Inference of Base Forms for Multiword Terms in Lithuanian. – Proceedings of the Fifth International Conference Human Language Technologies – The Baltic Perspective 2012, 27–35. Bucher A.-L. 2007: Terminology Work the Swedish way. – Terminologija 14, 37–48. Budin G. 2012: Trends Analysis: eLexicography, Terminology, and Global Content Management. Lexicograffiti. European Parliament, Luxembourg 23rd of January, 2012. Prieiga per internetą: https://term-coord.files.wordpress.com/2012/01/presentation.budin.pdf (žiūrėta 2015-06-15). Castagnoli S. 2006: Using the Web as a Source of LSP Corpora in the Terminology Classroom. – Wacky! Working papers on the Web as Corpus (eds. M. Baroni, S. Bernardini), Bologna: Gedit, 159–172. Kaupaitytė D. 2010: Švietimo terminų apibrėžtys ir vartosena: kompetencija ir kvalifikacija. Magistro baigiamasis darbas, Kaunas: Vytauto Didžiojo universitetas. Korkas V., Rogers M. 2010: How much terminological theory do we need for practice? An old pedagogical dilemma in a new field. – Terminology in Everyday Life. Terminology and Lexicography Research and Practice. Eds. F. Steurs, M. Thelen, Amsterdam: John Benjamins, 123–136. Marcinkevicienė R. 2000: Terminografija ir tekstynas. – Terminologija 6, 5–22. Markiewicz L., Rimkutė E. 2013: Švietimo ir mokslo terminų žodynas ir ontologija. – Terminologija 20, 56–74. Maskaliūnienė N., Kaminskienė L. 2012: Bendrosios kompetencijos ir vertėjų rengimas. – Vertimo studijos 5, 113–130. Meyer I. 2001: Extracting Knowledge-rich Contexts for Terminography: A Conceptual and Methodological Framework. – Recent Advances in Computational Terminology. Eds. D. Bourigault, Ch. Jaquemin, M. L’Homme, Amsterdam: John Benjamins, 279–302. Picht H. 2011: The Science of Terminology: History and Evolution. – Terminologija 18, 6–26. Rimkutė E. 2010: Lietuvių kalbos terminų automatinio nustatymo galimybės. – Kalbų studijos 16, 71–78. Sikora I. 2013: Contemporary approach to terminological competence, management and terminology teaching on the basis of courses for translators offered by Polish higher education institutions. – Proceedings of the 19th European Symposium on Languages for Special Purposes, July 8–10, Vienna, 500–509. Terminologija | 2015 | 22 167 Smetona A., Pukelis K. (sud.) 2014: Studijų terminų žodynas. Prieiga per internetą: http://www.skvc.lt/uploads/ documents/files/Kita informacija/Leidiniai/Zodynas.pdf (žiūrėta 2015-06-15). Stankevičienė S. 2013: Gamtos mokslų kalba: gretinamasis vokiečių ir lietuvių kalbų leksinių-semantinių aspektų tyrimas. Daktaro disertacija, Vilnius: Vilniaus universitetas. Zeller I. 2005: Automatinis terminų atpažinimas ir apdorojimas. Daktaro disertacija, Kaunas: Vytauto Didžiojo universitetas. RESULTS OF THE DESCRIPTIVE TERMINOLOGY RESEARCH FOR TERMINOLOGY PRACTICE The paper discusses the descriptive approach, which only recently has been more widely used in Lithuanian terminology practice. Terminologists, translators, proofreaders and other practitioners dealing with terminology can benefit from methods and tools that are applied in descriptive terminology management. Furthermore, the approach of descriptive terminology can be adopted for terminology teaching and learning purposes. The paper aims to describe possible applications how the descriptive terminology research (resources, tools and methods) can be used in the terminological practice. Firstly, we introduce resources and tools developed during the project ŠIMTAI 2 (Automatic Identification of Education and Science Terms). The resources and tools have been developed for the semi-automatic analysis of Lithuanian terms: i.e. specialized annotated corpus, tools for automatic extraction of terminology and a tool for semi-automatic detection of knowledge-rich contexts for terms in the specialized corpus, which then can be used for preparing definitions of terms. Knowledge-rich contexts are discussed in more detail. Knowledge-rich contexts contain conceptual characteristics about the analysed term, which are reflected by semantic relations and attributes. Applying the approach of the corpus-based descriptive terminology, we show a) how knowledge-rich contexts can be detected; b) how particular semantic relations can be identified in knowledge-rich contexts; c) how the conceptual system can be constructed by analysing the usage of terms; d) how knowledge-rich contexts of a term can be used for its definition writing, and evaluation. We exemplify the possibilities of methods and tools of descriptive terminology for the practitioners who are working in the field of Lithuanian terminology. Gauta 2015-06-15 Jolanta Kovalevskaitė Vytauto Didžiojo universitetas Kompiuterinės lingvistikos centras K. Donelaičio g. 52–206, LT-44248 Kaunas E. paštas [email protected] Agnė Bielinskienė Vytauto Didžiojo universitetas K. Donelaičio g. 52, LT-44248 Kaunas E. paštas [email protected] 168 Jolanta Kovalevskaitė Agnė Bielinskienė Deskriptyviosios terminologijos tyrimų rezultatų panaudojimas terminologijos praktikoje