Text Entropy and Implementation of Translation Strategy | April 2015 | Translation Journal

April 2015 Issue

Read, Comment and Enjoy!

Join Translation Journal

Click on the Subscribe button below to receive regular updates.


Text Entropy and Implementation of Translation Strategy

Text Entropy and Implementation of Translation Strategy

Entropy as a measure of the informational ambiguity is an important informational characteristic of text. Translator’s activity, on the one hand, is objectively determined by the characteristics of the original text, the features of the language of translation, the laws and principles of translation; and on the other hand, it is a creative activity, conditioned by the translator’s subjective qualities and perspective. Thus, in specific contexts of communication different translators may solve the same problem in different ways.

The existence of the entropy of the source text and the target language creates the possibility of different interpretations and different rendering of the original in translation, as it opens the opportunity for the translator to actively participate in the creation of the communicatively relevant meaning of the target text. Therefore, we view entropy as an important objective factor influencing the choice of translation strategy. The aim of this study is to determine the influence of the source text and target language entropy on the choice of translation strategy. The choice of translation strategy as the research object entails the need to define translation entropy as the main concept of the research and the select the linguistic unit of analysis.

The source text entropy is defined as a measure of its informationalambiguity (informationalcapacity), which is comprised by the aggregate of information per unit and the number of possible variants of expression / interpretation of its sense [5].

One of the first specialized studies aimed at the practical definition of the entropy of a particular work of art is the work by Indian philologist Narashimha Ramayya"Linguistic Entropy in Othello of Shakespeare" [14] which is an attempt of introducing concepts, models and methods of cognition developed in physics and mathematics, classical and oriental philosophy into linguistic and literary analysis of the text. Viewing entropy as a measure of disorder in the system [14, p. 10], the author states that the speech entropy is generated by "the primary dialect" the regional dialect, colloquial language, and, in the last instance, by the literary language, while the linguistic structures inducing entropy are represented at the level of phonology, grammar and syntax [ibid, p. 3]). To linguistic phenomena associated with the existence of entropy, the author attributes slang (colloquialisms), idioms, unusual phrases, sentences with intermittent structure and even the omission of sentences. Considerable attention is given to the phonetic tools such as alliteration, assonance, rhyme, onomatopoeia; the researcher observes that "in the poetic work phonology becomes phraseology" [ibid, p. 23]. The author also introduces the concept of psychological and socio-linguistic entropy that impede decoding of the play by its readers [ibid, p. 27-60].

The main value of this work for the linguistic and translation research is, in our opinion, the definition of linguistic resources which are characterized by a high potential for the creation of the text entropy: such are mostly stylistic means of different levels characterized by polysemy (ambiguity) and unpredictability. Translation entropy in its turn is to be viewed as a combination of the source text entropy and the target language entropy.

Other studies of various art systems unveil different levels of information and entropy in the classical work of art and the avant-garde works, intrinsically focused on unpredictability of combination of elements: the texts of the avant-garde works are characterized by a high level of entropy, because sometimes even the combination of phonemes in them is not governed by any laws of the language; at the same time, at higher levels the violation of lexical or grammatical compatibility is observed [10]. However, the same results in high predictability because each successive element is predictably a-semantic, illogical, as well as the previous one, which results in message redundancy [ibid].

The previous studies show that the level of entropy depends on the type of text: newspaper articles are characterized as texts with low entropy (highly predictable) [7, p. 180; 9, p. 149], and texts with a high level of entropy are exemplified by avant-garde poems [10]. Quantitative (mathematical) studies of language entropy are based mainly on the number of characters (letters in the alphabet of the respective language) and the probability of their occurrence in the text [6; 8; 9], although the corpus-based researches sometimes operate the "word usage" as the unit of research [3].

In our opinion, the approach based on the number of characters completely eliminates the aspect of the content of messages and leads to the analysis of quantitative indicators in isolation from quality. We believe that the entropy of speech should be determined by the number of realized informative choices. The notion of choice as a move in solving a problem of translation is introduced by Jirí Levý [12]. Making a choice implies selecting one of the alternatives variants, differing semantically, stylistically, etymologically etc. The author of the original also selects alternatives offered by the source language means operating within the SL entropy. Based on the understanding of language/speech entropy developed by the school of N. Kolmogorov, we define the quantitative measure of entropy as the potential of delivering information through 1) the number of linguistic units, the use of which is not predictable, given the type and structure of the text, the previous context, stylistic norms etc (as indicated by A.S. Monin, a word that can be guessed from the previous context, does not add any information [7, p. 180]), and 2) the number of linguistic units that suggest other options for expressing the same meaning. According to Jirí Levý, “The translator, in his system of decisions, may take one step more or less than the author of the original did” [12, p. 151].

According to the hypothesis of the study, after interpreting the source text and forming its conceptual image (the text concept), the translator, correlating this conceptual image with the world view of the target language community, creates a "projection", a generic image of the future translation. Focusing on this ideal image guided by the relevant principles of translation and given extra-linguistic factors, the translator chooses a strategy of translation of the entire text – a global strategy.

In the course of translation, the translator is faced with the language reality of the text, which unfolds as a discrete continuum. This text presents the two types of speech fragments/ translation units: (1) predictable, suggesting unambiguous reading (at least in the given context and situation) and a standard way of translation and (2) unpredictable, i.e. having a high level of entropy. Such fragments allow for differing interpretations, the multiplicity of renderings, various embodiments in translation. It is the presence of such fragments involving multiple choices that opens a possibility for the translator to implement a global translation strategy through local strategies.

For example, in the following fragment only one word allows for the possibility of a strategic choice by the translator. Numerous direct (untranslated) borrowings reflect the specificity of Ernest Hemingway’s individual style, "transferring" the reader into the atmosphere of another country [2]. A global strategy for maintaining this effect of being in a strange environment (foreignizing strategy) is quite adequate to the author’s intention and conceptual sense of the original. However, in the course of translation, the possibility of transformation of the global strategy into the local strategy is possible only when translating the borrowing from Spanish (in the following example – the word «desencajonada»). This word is not international and for the English-speaking reader it comprises an unpredictable combination of graphic characters, the meaning of which is unknown. In other words, this word has a high level of entropy, while the other language units in the fragment are unambiguous and predictably combined in the text with no possibility of varying interpretations in translation:

They've never seen a desencajonada." (Hemingway. The Sun Also Rises)

Even one word opens the prospect of transformation of the global translation strategy into various local strategies. In translation into Ukrainian the foreign language element is reproduced without translation, which makes it possible to maintain the association with the foreign culture, as it was intended by the author. However, the foreignizing strategy in this case is mitigated by the introduction of the footnote with the translation from Spanish (this translation technique may be described as untranslated borrowing with a commentary):

Вони ніколи не бачили desencajonada1.

1Вивантаження биків (ісп.)(Хемінгуей. Фієста (І сонце сходить). Перекл. М. Пінчевський)

This local strategy can be described as mitigated foreignizing; its implementation results in maintaining the effect of being in a foreign country, but the readers’ perception of the referential situation is facilitated by the commentary. By expanding the volume of the text carrying the same meaning, its information entropy is reduced and less cognitive effort for the reader's perception of the text is required.

In the Russian translation the foreign borrowing is completely replaced by the explanation of its meaning in the target text:

Они никогда не видели выгрузки быков. (Хемингуэй. Фиеста (И восходит солнце). Перев. В. Литвинов)

Desencajonada, a Spanish culture-specific notion referring to the ‘introductory’ part of corrida in which the new bulls are first presented for the spectators to appraise them (the bulls are released from their transportation containers onto the arena; from ‘desencajonar’ – lit. “relase from the box”), is translated semantically as ‘unloading bulls’, without informing the reader of actual significance and culture-specific value of the denotatum.

In this case, the local strategy counters the global strategy – it can be described as a strategy of domestication, which approximates the referential situation to cognitive phenomena familiar to the reader; thus, the readers’ perception is automated (no additional cognitive effort is required to decode the given piece of the text), and the artistic effect intended by the author is neutralized. As a result, the target text has a significantly lower level of information entropy than the original; its interpretative potential is reduced. In the absence of compensatory strategies and techniques, such rendering can significantly distort the original style of Hemingway’s work, deprive it of its characteristic semantic capacity, its appeal to the reader’s creativity, featuring him as a participant of the event.

A similar impoverishment of the author's style is observed in translations of O. Henry’s stories, for example, in the case where the translators replace the direct speech with foreign accent represented using foreign words or graphons (intentional violation of the graphical shape of a word [1]) with standard literary speech similar in meaning:

Old Behrman, with his red eyes, plainly streaming, shouted his contempt and derision for such idiotic imaginings.

"Vass!" he cried. "Is dere people in de world mit der foolishness to die because leafs dey drop off from a confounded vine? I haf not heard of such a thing. (O. Henry. The Last Leaf)

In the Ukrainian translation the failure to recreate the artistic effect of speech characterization is only partially compensated for by introducing a descriptive commentary ‘with terrible German accent’:

Старий Берман з червоними очима, які помітно сльозилися, галасливо виявив свою зневагу, знущаючись із таких ідіотських вигадок.

– Що, – кричав він з жахливим німецьким акцентом, – хіба ще є такі дурні, щоб умирати через листя, яке осипається з клятого плюща? Вперше чую. (О. Генрі. Останній листок. Переклад М. Дмитренка)

In the Russian translation of this speech characteristic is completely lost, with no compensatory technique:

– Что! – кричал он. – Возможна ли такая гупость – умирать оттого, что листя падают с проклятого плюща!... (О. Генри. Последний лист. Перевод Н. Дарузес)

Translators both into Ukrainian and Russian in this case preferred the strategy of content to the strategy of form, although for literary artistic translation the strategy of reproducing the artistic form is of paramount importance since intention to focus on the form enables the translator to achieve the "poetic" equivalence [11 p. 150]. Using graphons (departing from the spelling rules for rendering incorrect pronunciation) or inserting some foreign words, the author increases the entropy of the source text: the direct speech becomes unpredictable and its automatic perception becomes impossible, requiring the reader to use additional cognitive efforts for understanding, appealing to his auditory perception, involving him in the process of co-creation of the sense. A significant reduction in the information entropy due to the loss of artistic device noticeably impoverishes the translation.

Unfortunately, in the analyzed translations of short stories by O. Henry similar local strategies of "semantic" translation prevail. If we apply the method of determining the global strategy through the proportion of local strategies (see: [16]), we must admit that the wrong choice of the global strategy has been initially made. As a consequence, the translation reproduces mostly the plot and compositional characteristics of the stories, the surface content (often – only the denotative meaning) of the characters’ speech, but the specific features of the artistic form, including puns, allusions, speech characteristics of the characters are often lost in translation.

The results of the analysis indicate that the information entropy of the original text offers the prospect of selecting a local translation strategy, and the entropy of the target language opens up the opportunities of choosing the language means to implement this strategy. In this case, the success of the translation solutions, ideally, is determined by the choice of the strategy enabling to recreate the variety of the readers’ renditions rooted in the author's text, i.e., to save an identical level of entropy in the translation. However, the examples demonstrate that actual translations often reduce the entropy of the source text by substituting multiple interpretations of the text by the translator’s only interpretation; consequently, the reader is less involved in the process of co-creation, the ability of the text “to generate the sense” [5] is reduced.

Similar observations were made as far back as mid-20th century by a Czech translation theoretician Jiri Levy: the fact that “the language of average and of bad translations very often has lost the freshness and the flavour a good author knows how to give to his wording” [13, p. 62] is explained by the translators’ preference of the most general and most predictable target language units: “When choosing from among several equivalents or quasi-equivalents for a foreign term, a translator inevitably tends to choose a general term, whose meaning is broader than that of the original one, and in consequence is devoid of some of its specific semantic traits… Within a group of semantically cognate expressions, the single words occur with different average frequencies in common usage, and therefore have different degrees of predictability; those that are used with the greatest average frequencies are the first to occur to the translator seeking the right expression”[ibid. p. 63 – 64]. Translator’s choice of a more general word sometimes results from the absence of a TL word with similar denotative, connotative and stylistic characteristics. In the following example the choice of a word etymologically related to French is informative:

He [Denis] was enamoured with the beauty of words.(A.Huxley.CromeYellow)

Denis, the main character, was looking for a word to describe the beauties of the English landscape, but only French words came to his mind, because those are most typically used by novelists. Later the reader will discover that foreign words sound so charming for Denis that he may be confused by their form and use them in the wrong meaning. Thus, the word enamoured, besides its denotative meaning, renders the author’s ironical attitude to the character and satire on contemporary novels. Both Ukrainian and Russian translators manage to render the denotative meaning of the word:

Він був закоханий у красу слів. . Хакслі. Жовтий Кром. Перекл. В.Вишневий)

Он был опьянен красотой слов. (О. Хаксли. Желтый Кром.Перев. РДубровкин)

In Russian, the translator chooses a metaphor which partially compensates the loss of the original expressiveness and adds to the ‘informational density’ of the text, because the metaphor, being double-sensed, increases the text entropy and involves the reader’s imagination into perception of the text. In Ukrainian translation the strategy of rendering content is evidently preferred, and the result is impoverishment of the connotative content of the text. The translator into Russian is following the strategy of rendering the poetic form and looking for a way to at least partially compensate the impossibility to render all the connotations inherent in the source text; he introduces a trite metaphor which is more expressive than a non-metaphoric expression.

Another factor leading to the loss of entropy is that “translator tends to explain the logical relations between ideas even when they are not expressed in the original text…everything is explained and nothing left to instigate the reader’s imagination and intellectual cooperation” [13, p. 65 – 66]. In terms of informational properties of the text this means using more words to render the same information, thus decreasing informational value of each word. Such examples are so common that they hardly ever attract special attention:

In due time it backed up to the door with much gong-clanging, and the capable young medico, in his white linen coat, ready, active, confident, with his smooth face half debonair, half grim, danced up the steps.

"Ambulance call to 49," he said briefly. "What's the trouble?" (O.Henry. The Skylight Room)

В означений час карета із страшним дзвоном розвернулась біля ґанку, і з неї вискочив енергійний молодий медик у білому халаті, впевнений, готовий діяти, спокійне лице його було напівжиттерадісне, напівпохмуре.

- В будинок сорок дев'ять викликали карету,- сказав він коротко.- Що трапилось? (О. Генрі Кімната на горищі. Перекл. Ю. Іванов)

According to one of the fundamental tenets of the theory of information, informative value of a message is inversely proportional to the degree of its organization. Ambiguity and vagueness of various kinds are fraught with many potential meanings and interpretations. Definiteness, on the other hand, sacrifices semantic richness of in favor of accuracy and clarity [4]. Thus, the strategy of content, which may be equitable for ‘informative’ texts, in artistic translation results in decrease in entropy, i.e. potential information of the utterance and lesser reader’s involvement in the interpretation of the text.

Summarizing the above, the informational heterogeneity of the source text influences the translator’s treatment of the certain parts of the message: only the fragments that are ambiguous and unpredictable require the translator’s choice which is determined by the local translation strategy. The choice of foreignizing strategy may result in creation of a target text with a similar or close level of entropy, which means similar effort of the audience in understanding the text, similar possibilities of individual interpretation, the same information potential and often artistic effect. The choice of domestication strategy or strategy of content results in ‘easier understanding’, which means narrower interpretational potential and lower informational value; it may also result in the loss or artistic devices and general depletion of the text informational content. Such ‘fluency’ in translation is often praised by English and American literary critics, although viewed as a flaw by translation theorists [15].

The prospect for further research may be the unveiling of cognitive mechanisms involved in the development and implementation of translation strategies in the texts of different functional styles


  1. АрнольдИ.В. Стилистика. Современный английский язык: учебник для вузов / Ирина Владимировна Арнольд.– 4-е изд., испр. и доп. - М.: Флинта, Наука, 2002, - 384 с.
  2. Довганчина Р. Г. Відтворення ідіостилю Ернеста Гемінґвея в українських та російських перекладах: автореф. дис. на здобуття наук. ступеня канд. філол наук : спец. 10.02.16 «Перекладознавство» / Р. Г.Довганчина. – К., 2011. – 23 с.101.
  3. Кутузов А.Б. Частотные характеристики лексики как инвариант перевода (на материале перевода романа К. Воннегута «Бойня номер пять, или Крестовый поход детей») / А.Б. Кутузов // Вестник Тюменского государственного университета, №1. – 2010. – с. 175–183.
  4. ЛеутаО.Н. Ю.М.Лотман о трех функциях текста / О. Н. Леута // Юрий Михайлович Лотман / Под. ред. В.К. Кантора. – М.: РОССПЭН, 2009, с. 294-309
  5. Лотман Ю. Структура художественного текста / Юрий Лотман [Електронний ресурс] – Режим доступу до джерела: http://www.gumer.info/bibliotek_Buks/Literat/Lotman/_01.php
  6. Михайлов М.Н. Чем длиннее, тем лучше? Как сравнить длины исходного текста и перевода? / М. Н. Михайлов. // Математическая морфология. Смоленск: СГМА, 2003. Том 5. Вып. 1. http://www.smolensk.ru/user/sgma/ MMORPH/N-9-html/mihailov/mihailov.htm
  7. Монин А. С. Дороги в Комаровку // Колмогоров в воспоминаниях учеников / Редактор-составитель А. Н. Ширяев. — М.: МЦНМО, 2006. - с. 174–185.
  8. Ревзин И. И. Совещание в г. Горьком, посвящённое применению математических методов к изучению языка художественной итературы // Структурно-типологические исследования: Сб. статей/ Отв. ред. Т. Н. Молошная; Академия наук СССР: Институт славяноведения. — М.: Изд-во АН СССР, 1962. с. 285–293.
  9. Успенский В. Предварение для читателей «Нового литературного обозрения» к семиотическим посланиям Андрея Николаевича Колмогорова/ Владимир Успенский // «НЛО» 1997, №24 с. 123–215
  10. Цвигун Т.В. О преодолении энтропии в поэтическом тексте (К семиотической структуре текста в русском авангарде 10–30-х гг.) // Общие вопросы авангардоведения [Електронний ресурс : режим доступу до джерела – http://avantgarde.narod.ru/beitraege/ov/c_entropiya.htm]
  11. Швейцер А.Д. Теория перевода: Статус, проблемы, аспекты / Александр Давидович Швейцер. – М.: Наука, 1988. – 215 с.
  12. Levý J. Translation as a decision process / Jirí Levý// The Translation Studies Reader [Ed. by Lawrence Venuti]. – London and New York: Routledge, 2004. – p. 148–159.
  13. Levý J. Will translation theory be of use to translators? / Jirí Levý// First printed in: Rolf Italiaander (ed.) Übersetzen: Vorträge und Beiträge vom Internationalen Kongress literarischer Übersetzer in Hamburg 1965, Frankfurt am Main: Athenäum Verlag, 77–82. – Reprinted in: Scientia Traductionis, No11, 2012, p. 60–71.
  14. Narashimha R.N. Linguistic Entropy in Othello of Shakespeare / N. Narashimha Ramayya. – New Delhi: M D Publications Ltd, 1994.  – 101 p.
  15. Venuti L. The translator’s invisibility: A history of translation / Lawrence Venuti. – London and New York: Routledge, 1995. – 353 р.
  16. Zare-Behtash E. A Diachronic Study of Domestication and Foreignization Strategies of Culture-Specific Items: in English-Persian Translations of Six of Hemingway’s Works / Esmail Zare-Behtash, Sepideh Firoozkoohi // World Applied Sciences Journal 7 (12): 1576-1582, 2009. – р. 1576–1582.
Log in

Log in