Earticle

The parallel corpus is today accepted as a useful tool in the translational studies, because the researchers can observe many translational examples with the parallel corpus. This paper aims to introduce a statistical method in France, called textometry, which enable to extract translational equivalences from parallel corpora or translation corpora. The textometry execute various statistical methods targeted at textual units (words, phrases, sentences, etc.) in a parallel corpus divided into the tokens. Using quantitative evidences by the textometry, researchers can verify the translational relation of linguistic units in translational texts.

Corpus-based studies have become a general method of investigation in modern translation studies. One of the reasons to use a parallel corpus is to permit the direct examination of existing translation resources produced by translators. To extract precise data from a corpus and for the efficient study of these data, researchers have proposed techniques based on quantitative algorithms. Poly-cooccurrece is one of these quantitative methods as it allows the exploration of the distribution of concurrent occurrences of a particular word in a corpus. The aims of this study are as follows. First, to extract the concurrent occurrences of words that represent mutual translations in a French-Korean parallel corpus using poly-cooccurrence. Second, to compare lexical cooccurrences of these words. And third, to validate the advantage of this method in a corpus-based translation studies. Traditional methods such as KWIC (Keyword in Context) and simple-cooccurrence merely produced a list of concurrent words and failed to explain the correlation between the words. However, poly-cooccurrence creates a tree diagram which describes the distributions of the words and therefore we can explicitly comprehend the lexical context of translation units and its translation context by exploring hierarchically concurrent words in bi-textual resources.

This paper discusses the application of a comparable corpus composed of translation texts for the analysis of the translation characteristics of Korean-French and French-Korean translations. In most translation studies, researchers focus on parallel corpora, rather than comparable corpora. However, parallel corpora are not widely used in this field due to their dependence on computational preprocessing such as the alignment of translation units. When the purpose of a study is not the direct comparative study of translation data, a comparable corpus is preferred in order to make general observations on the translation characteristics of translation texts. In this paper, by using Korean-French and French-Korean translation texts as a comparable corpus, we attempt to apply quantitative analysis to assess the validity of this kind of corpus in translation studies. In doing so, it is our hope that researchers can make full use of translation texts assisted by computer-based tools without the challenges inherent in the construction of a corpus units.

Since the late 20th century, corpus-based studies have become a general method of investigation in modern linguistics. A corpus (monolingual or multilingual corpus) permits the linguists to explore the real linguistic data used by a speech community. However most of the studies focus on the high-frequency words in a corpus, whereas the low-frequency words are excluded from the subject of the analysis. The low-frequency words also consist of a corpus; therefore they are important linguistic elements as much as the high-frequency words. The purpose of this paper is to look into translation equivalences between the low-frequency words in a parallel corpus, a French Novel Moderato Cantabile, and two Korean translations, the Moderato Corpus. For the purpose of researching the translation phenomenon of low-frequency words in the Moderato Corpus, this paper uses the quantitative methods of lexical frequency, co-occurrence and contextual exploration. The result of these analyses demonstrates that the low-frequency nouns have different translational situations as much as the high-frequency ones. Surprisingly, they have also a definitive equivalence between the French word and the Korean word. Therefore, this study contends that low-frequency words can contribute to translation studies and, especially, enhance the automatic translation machine as a translation reference.