Methods for determining equivalents of verbs expressing emotional states in the Czech-Polish translation on the material from the InterCorp parallel corpus
Abstract
Methods of determining equivalents of verbs expressing emotions in translation from Czech to Polish, based on texts from the InterCorp parallel corpus This study focuses on the problem of determining Polish equivalents of Czech verbs. The aim of the analysis is to explore which method can best handle this task and what factors determine the choice of the equivalent.
The object of the analysis are Czech lexical units referring to various mental or emotional states. Although Czech is a language closely related to Polish and false friends might be naively expected to pose the main challenge, finding an appropriate Polish equivalent for most lexemes of this type involves many issues common in translation from typologically more distant languages.
This concerns both the task of decoding the meaning of the lexeme in the source text according to the conceptual patterns of the source language as well as encoding the meaning in the target language according to its conceptual patterns. To study factors responsible for an optimal choice of an equivalent, the inquiry is carried out on texts from a parallel corpus.
Parallel texts offer the chance to examine many occurrences of the source and target lexemes in their contexts, investigated from the morphosyntactic, syntactic, semantic and pragmatic perspectives, including details about the situation in which the lexemes are used and details about the text itself. Given the challenging choice of psych verbs, the texts that best meet the research needs are fiction texts or texts of a similar type.
InterCorp is a parallel corpus that includes a substantial share of such texts. On the basis of available lexicons and the corpus data, the task to find the best fitting equivalent can then be approached from various perspectives, resulting in a sequence of steps, each invoking a method of searching for an equivalent.
The perspectives combine the approaches of translatology, theoretical linguistics and corpus linguistics. Through the prism of translation, the author shows the possibilities of applying various linguistic theories (valency theory, Case Grammar, Pattern Grammar) and corpus tools (Treq, Sketch Engine), demonstrating the merits of using the parallel corpus, but also identifying its limitations.
Building on these resources, an algorithm to determine Polish equivalents for the Czech verbs is proposed. The agenda is reflected in the outline of the book, which consists of thirteen chapters, best seen as forming four parts - theoretical and data-related (Chapters 1-4), analytical (Chapters 5-10), proposal-related (Chapters 11-12), and a summary (Chapter 13).
Bibliography and indexes conclude the volume. Whenever appropriate, the text is amply illustrated with examples and accompanied by tables, summarizing statistical data.
After a short introduction, Chapter 2 discusses theoretical issues concerning equivalence and translation, as well as presenting various definitions of equivalence and the relationship between traditional translation and automatic alignment. The next chapter includes comments on the correlation of translation and contrastive studies (Chapter 3).
The InterCorp parallel corpus is presented in Chapter 4, together with specifics of its Czech-Polish section and details about Treq, the database of lexical equivalents extracted from the parallel corpus. Chapter 5 introduces the methodology used in the six following case studies, each concerned with a Czech verbal lexeme.
The theoretical part proceeds with chapters presenting these case studies: zdát se 'seem' (Chapter 6), trápit (se) 'bother, worry, torment' (Chapter 7), soucítit 'sympathize' (Chapter 8), závidět 'envy' and žárlit 'be jealous' (Chapter 9), mít rád 'like' and milovat 'love' (Chapter 10). The following chapter presents an algorithm that facilitates the determination of equivalents.
The procedure consists of the following steps: dictionary lookup, automatic extraction of equivalent pairs from the parallel corpus, valency-based analysis, analysis in terms of Case Grammar (deep cases), analysis in terms of Pattern Grammar, and analysis employing other methods or resources: Word Sketch, WordNet, WoSeDon, Cognitive Grammar (Chapter 11). The algorithm is exemplified in the following chapter, a case study of the Czech verb toužit 'desire' (chapter 12).
Each of the steps is described in detail and exemplified to show the extent of usability of the overall methodology and of each of the individual approaches, including cases where the step does not lead to a desirable outcome. The concluding summary indicates further research perspectives (Chapter 13).
The monograph presents an interdisciplinary analysis, combining the approaches of linguistics and translatology, together with elements of statistics and computational linguistics, based on data from a parallel corpus. The analysis results in the design of a procedure aiming at the choice of the most fitting lexical equivalent.
Recent years have seen a number of research results published in the field combining corpus linguistics and translation. However, most of them concern English.
Yet this study is unique not only due to the languages studied, but also due to the methods employed, especially methods from the borderline of translation studies and comparative analysis.