Comparative Study of Metaphor in Literary Texts and their Translations

Since Aristotle metaphor was relegated to the domain of literature, until the revolution instantiated by Lakoff and Johnson in the 1980s showed its pervasiveness in language and thought, but paradoxically, it alienated metaphor research from poetics. The latter has slowly been finding its feet in conceptual metaphor studies and with the new development of corpus research, obtained ample material for cross-cultural analysis especially with the help of parallel texts studies – an efficient way to delve into linguistic and culturally-defined differences. The aim of the research is by identifying the conceptual metaphor behind the metaphorical linguistic expressions in key texts from five of the greatest stylists of the English language, and by comparing them to their translations into Bulgarian, to check whether metaphor is lost or transformed in any way. Conclusions are made regarding literature in translation, suggesting cases in which it is comparable to original writing in terms of metaphor type and density.


The metaphor phenomenon
The phenomenon of metaphor has been puzzling philosophers, psychologists and linguists alike for many centuries. In recent times several major theories explaining its intricacies were proposed. Drawing upon Aristotle's treatment of the term and following Richards' ideas of metaphoric structure (Richards, 1981), modern scholars have tried to explain metaphor by putting forward the substitution theory, the comparison theory, the interaction theory, etc. On the one hand, there is the traditional view in which metaphor functions only at the level of language. The second view, pioneered by Lakoff and Johnson (1980), advanced over the last thirty years or so, holds that metaphor is a conceptual device relating to thought and has an elaborate relationship with language. This change of perspective had reverberations in multiple areas of scientific inquiry associated with language, culture, translation, and literature. Metaphor studies developed into an important area of research. From a cognitive point of view, it generated interdisciplinary research with translation studies (van den Broeck, 1981;Dagut, 1987;Mandelblit, 1995;Schäffner, 2004;Dickins, 2005), discourse analysis (Musolff, 2004;Charteris-Black, 2004), education (Cameron 2003), and more recently, cognitive poetics (Tsur, 1992;Stockwell 2002).
• Conceptual metaphors are identified in five key texts in English literature.
• The metaphors are compared to their translations into Bulgarian.
• The research did not find many cases of explication, paraphrase and loss of metaphor.
• The number of lexicalized metaphor is greater than that of original metaphors.
• Creative metaphors translate readily in the target texts.

Metaphor and translation
Both structurally and etymologically, translation and metaphor happen to be very close,: translation comes from Latin "transferre", trans -"across", ferre -"carry". Metaphor, similarly, derives from the Greek "meta" -"change" and "pherein" -"carry" (etymonline.com). The metalanguage of both contains the concepts of source and target domains, languages, cultures. The Bulgarian word превод contains the same connotations. Tymoczko (2007: 68-77) concludes that in most Indo-European languages the words translation, metaphor, and transfer are conceptually related.
Whilst the central problem of translation is the overall choice of a translation method for a text, the most important particular problem is the translation of metaphor (Newmark, 1988). The complex character of metaphor has preconditioned its translation as being problematic, to say the least. The first to mention this was Kloepfer (1967) and since then a lot of attention has been showered on this issue. Many attempts have been made to study various aspects of metaphor in translation, linking the two areas of research. Fernandez (2005) sums up exhaustively the different approaches based on cultural, textual and cognitive characteristics. According to her classification there are four distinct opinions on metaphor translatability within the linguistically oriented views. The first is based on the presumption that metaphors are by virtue of their nature unpredictable, hence virtually untranslatable, as they present a particularly searching test of the translator's ability, because it involves cultural experiences and semantic associations, a view with proponents such as Vinay and Darbelnet (1958), Nida (1964) and Dagut (1976Dagut ( , 1987. Mason (1982) takes a more moderate stance, claiming that metaphors are translatable with a degree of non-equivalence and their translatability depends on the cultural and semantic associations it creates in the source language and on the extent these can be rendered into the target language. As the overlap is unlikely to be complete, any attempt at an exact rendering is doomed to failure, so all we can establish is a degree of translatability. A second view holds that metaphors are fully translatable and pose no special problem -a position supported mainly by Kloepfer (1967) and Reiss (2000). A still third group of translation researchers, such as Van Den Broeck (1981), Toury (1995) and Newmark (1985Newmark ( , 1988, subscribe to the standpoint that translation of metaphor is possible, although problematic. In line mostly with the substitution theory of metaphor (Martin & Harré, 1982:90), Translation Studies analysts use terms like "image" or "vehicle" for the conventional referent, "object" or "topic" for the actual unconventional referent, and "sense", "ground", or "tenor" for the similarities involved (Schäffner, 2004(Schäffner, : 1255. They suggest elaborate lists of types of metaphor and procedures of metaphor translatability. Newmark (1988: 106-113), for example, asserts that the only fully translatable metaphors are "dead" ones, as they show the greatest proximity of the two polysystems involved and proposes (Newmark, 1981: 87-91) a classification based on seven options, focusing on linguistic systems. In his turn, Van Den Broeck's opts for three possible outcomes: "translation 'sensu stricto', substitution and paraphrase". For Van Den Broeck (1981: 73-84) the rejection of the idea of translatability goes against the grain of Translation Studies. Both Van Den Broeck (1981) and R. Alvarez (1993) see lexicalized metaphors as the "most translatable" ones, while considering novel metaphors to be extremely difficult to translate and stock metaphors fully translatable if the systems involved are culturally close (Alvarez, 1993: 137). The fourth approach, proposed by Snell-Hornby (1988), is called by Fernandez "conciliatory", as it claims "that the range of renderings will depend on the type of text we are dealing with and on ad hoc factors", which in essence focuses on both intralinguisric and extralinguistic factors affecting a translator's process of metaphor rendition. Recently, equivalence also entails "anomalous equivalence" (Toury, 1985: 25) such as 'zero solutions' or even creating metaphor where there exists none in the ST.

Metaphor translation theories in cognitive science
The radical break from the purely linguistic understanding of metaphor, associated with the appearance of the Conceptual Metaphor Theory (CMT), as well as with the more discursive and culturally oriented translation theories proposed over the last thirty years, has changed drastically not simply the view on metaphor but also the sway it holds on translation practice and translation studies. Mandelblit (1995) presents the Cognitive Translation Hypothesis and considers two schemes for the translation of metaphors:  Similar mapping conditions (SMC will obtain if no conceptual shift occurs between the metaphors of the two languages);  Different mapping conditions (DMC occurs when a conceptual shift takes place).
For Schäffner (2004) conceptual metaphors can be identical in the source text (ST) and target text (TT) at the macro-level. Structural components make entailments explicit. A metaphor is more elaborate in the TT, while ST and TT employ different metaphorical expressions, which can be combined under a more abstract conceptual metaphor. The expression in the TT reflects a different aspect of the conceptual metaphor.
Worthy of note is Müller's (2008) approach placed on a firm cognitive stand. Her line of argument refutes the mutually exclusive distinction between "dead" and "live" metaphors. Metaphors, she argues, operate on the level of language use and not on the one of language system. Accordingly, metaphoricity is a dynamic part of a cognitive activation process in an individual person at a given moment in time. Her claims are substantiated by empirical studies of multimodal metaphors that unite language, gestures, pictures, etc.

Metaphor in literary discourse
Conceptual Metaphor Theory made us reevaluate the role of metaphor in everyday language (Semino & Steen, 2008), but also introduced a new viewpoint regarding metaphor in literature, as well. There are two approaches to metaphor in literary discourse. Some scholars consider it no different from metaphor in the other types of discourse, yet others find no points of contact because they consider it quite superior to metaphor in the other types of discourse, which makes their comparison impossible because of the way metaphors in literary discourse interact with each other, and with other aspects of the texts they appear in. For Semino and Steen (2008), metaphor in literary discourse is superior to metaphor in other types of discourse.
In More than Cool Reason Lakoff and Turner (1989) view poetic metaphor as a new reformulation of conceptual metaphors that we use in our daily life. Poets challenge and reuse creatively everyday metaphor. In other words, cognitive linguists claim that most poetic language is based on conventional, ordinary conceptual metaphors. Creative / original metaphors are nothing more than a creative reformulation of conventional conceptual metaphor.

Metaphor in literature and translation
Cultural Linguistics cannot bypass the highest manifestation of both culture and language, namely, literature and especially the transformations that occur during translation between languages and cultures of the highest form of the language of literature that is taught and analyzed in schools and the academia. Cognitive linguistics can be a great tool in literature classes, throwing light on literary analysis by allowing our conceptual knowledge of conventional metaphor to serve as a stepping stone for explaining the hidden meaning of the work veiled in lexicalized, sleeping, conventional or novel linguistic metaphorical expressions. Cognitive grammar can also illuminate the successful or lame translation of metaphor which will convey or ruin the artistic effect of a work of art.

What does parallel text analysis reveal?
Individuals tend to use language differently. Both author and translator are specialists in conveying meaning through language. Indeed, ideally, the meaning in the original and in the translation needs to be identical. Given the professionalism of both, any differences in the conceptualization of metaphor between the two parallel texts are bound to be either culturally or linguistically motivated. In addition, other translation-related transformations, such as generalization and concretization, as well as domestication and foreignization, tend to occur in the process. This further alienates the source text from the target text. Most importantly, metaphors in ST and TT have to fulfill the same functions, expressed by Goatly (1997: 148): to fill language gaps; create meaning and memorability; express attitude and ideology. Metaphors also have aesthetic value, simultaneously functioning as a powerful cohesive device of the literary work, linking themes and ideas within the text and intertextually, between the work and the other texts. Thus, metaphor and translation studies have crossing points with literature and ever more scholars venture into its realms in their attempt to study it in all its manifestations. The study of literature at university level invariably involves working with original texts, very often with select excerpts from works of fiction, to which an analysis is made in the form of close reading, more in line with the empirical study of literature, based on Lakoff and Turner (1989), cognitive stylistics (Semino & Steen, 2008;Tsur, 1992) and cognitive poetics, with its foundational principles of embodiment, prototypicality and naturalness (Stockwell, 2007), than with literary historiography.
Parallel texts provide an opportunity for the study of metaphor as manifest in different languages and cultural environments. The content is a constant, though the languages are different, and the metaphors typical of each language, have evolved differently as people have coined similar or different expressions in the attempt to conceptualize and make sense of their particular surroundings. Parallel texts provide an opportunity to study metaphor universality and variation in culture in similar linguistic context and to analyze the different metaphors used in the two languages that are used to convey one and the same message. In Descriptive Translation studies by default the translational problems are reconstructed through target-source comparison.

Setup of the study
The corpus consists of five crucial texts from five highly acclaimed authors writing in English and their translations. They include a passage from Joseph Conrad's multilayered symbolic novel Heart of Darkness, Virginia Woolf's Mrs. Dalloway, Orwell's 1984, Richard Russo's award-winning novel Empire Falls and James Joyce's The Dead, and their translations into Bulgarian, which are analyzed by identifying metaphors in both texts, using Steen's MIP VU approach (Steen et al., 2010).
Conclusions are drawn regarding metaphor translatability in literary discourse, paying particular attention to different mapping conditions during translation and the cases in which the metaphor has been omitted.

Choice of corpora
The selected texts belong to the recognized literary canon and hence have been translated by the most prestigious translators in the country who have a flair and considerable knowledge of language and literature, though are totally unaware of CMT, since, more often than not, it was not even formulated at the time of their translations. Their extensive knowledge of language and literature, and their innate intuition however, helps them in dealing with the most intricate metaphorical nuances. The texts are such as are frequently analyzed in literature classes at the university level, using the methods of close reading. It involved a close reading of the text, identification of literary devices, such as metaphor, which refer to some aspect, or idea, or mega metaphor evident in the work in general.

Metaphor typology in the study according to use (transparency, conventionalization, novelty)
For the sake of the study we have focused closely on three types of metaphor:  Sleeping metaphors (lexicalized-non-transparent) -are the lexicalized metaphoric linguistic expressions that may be co-activated under certain circumstances in the text;  Linguistically expressed conceptual metaphors (entrenched-conventional and activated in the text) are those linguistic expressions in the literary text that are grounded in experience and that provide structural frames for the interpretation of the text;  Creative metaphors (novel) -whose metaphorical meaning is induced and relevant only in the specific context.

Initial hypothesis
The initial hypothesis involves the idea that the target texts will be longer than the source texts (because of explication, paraphrase, or redundancy) and that the type/token ratio will be greater in the former because of the normalization and simplification (Olohan, 2002) and Toury's law of growing standardization (Toury, 1995). Secondly, we presumed that there would be many cases of explication, paraphrase and loss of metaphor (Baker, 1996). Thirdly, that creative metaphors would predominate and would translate readily in the target texts (traditional expectation in literary analysis). Finally we believed that the target texts would be much inferior to the source texts in terms of metaphor depth and variety, and would not be suitable for close reading and analysis (traditional view). It is quite obvious that in all texts the average length of the TT is longer than the ST. In order to find the reason for this, whether it is because the translators need more words to convey the meaning of the ST, or because of the differences between the languages, we counted the characters, deducted the number of spaces and the punctuation marks and divided by the number of words. In this way we calculated the average length of word in both ST and TT (see Table 1) Although the number of characters in ST and TT were almost the same, these calculations gave us the insight that the increased length of the TT is not caused by redundant words, but by the longer words in the target language, due to the suffixed article and grammatical gender of nouns and adjectives. This result ruled out the translator's role in increasing the length of the translation and relegates it to the characteristic features of the languages involved (Table 2).  The type/token ration (Table 3) is a reliable gauge of the lexical density of the text. The closer the ratio is to 1, the more varied the vocabulary of the writer/translator. Again the results are consistently straightforward: the median ratio ST/TT is 0.482/0.576: a significantly richer vocabulary in the TT. This too, may be caused by the differences in the makeup of the two languages. Generally the higher number of types, compared to the tokens is a sign of the richness of the language.

Results
Our results suggest that the TT contains a higher number of types than the ST. This is in conflict with the expectation that the translator normalizes and standardizes the language. However, it is also a matter of language: the existence of allomorphs in the TL due to inflection (Bulgarian is a language with a predominance of mutable parts of speech, grammatical gender, verb conjugation combined with grammatical gender, all of which compromises the results. Therefore, although the type/token ratio in the target language is greater, this is accounted for by differences in the linguistic setup of the languages and is no proof of the opposite thesis.
In the case of metaphor, which allows compression of the meaning, however, the greater number of words in the TT is a sign of one of the translation universals, namely normalization and standardization. Target text James Joyce "The Dead" Translation into Bulgarian forms were near. His soul had approached that region where dwell the vast hosts of the dead. He was conscious of, but could not apprehend, their wayward and flickering existence. His own identity was fading out into a grey impalpable world: the solid world itself which these dead had one time reared and lived in was dissolving and dwindling.

Conceptual metaphors like LIFE IS A JOURNEY and A LIFETIME IS A DAY and UP
IS MORE, which lie at the bottom of the sentence from The Dead "The time had come for him to set out on his journey westward" have a universal appeal, while others, like "It lay thickly drifted on the crooked crosses and headstones, on the spears of the little gate, on the barren thorns" are culturally limited to readers who have experienced the limitations imposed by religious institutions. These constitute metaphors that work on a macro level and may be even missed by the translator. For example "crooked" in English may have spatial and moral dimensions, whereas the translation in Bulgarian, a country with a much more liberal attitude to religion, suggests only the spatial dimension of crooked, hence the loss of an important metaphor on a macro level.
The colors and fonts used to highlight the translated metaphorical expressions have the following significance: lexicalized metaphor in italic, Conceptual metaphor in bold and creative metaphor is underlined. On the right is the TL, where the same classification is made, and the different colours mark those translations that apply different mapping conditions (red), concretization (green), generalization (blue), formulaic language (pink), ʘ metaphor (metaphor lost) or + metaphor (metaphor added) ʘ + . Contrary to many expectations, the study of parallel literary texts shows that metaphors in the TT often exceed the number in the ST. In spite of linguistic and cultural differences and limitations, the number of metaphors in the source and target texts is comparable. In order to get a detailed picture of which group of metaphors yield to translation more than the rest, we have broken down the metaphors into lexicalized, conceptual, and original and counted them in the ST and TT. As expected, resulting from the different evolution of the languages, their different cultural, geographical and historical development, the lexicalized metaphors (Table 6) proved most difficult to translate. Set phrases and collocations, as well as function words work differently in the SL and the TL and the translators, fit the meaning in the existing linguistic and cultural moulds, sacrificing lexicalized metaphors, which, at the time of the translation (70's and 80's) were not perceived as metaphors at all. The linguistic expressions based on conceptual metaphors (Table 7) are deeply embedded in our thought and reveal much about the way we perceive the world. These metaphors have extra force because they hinge on our worldview and are easy to relate to. They translate readily, though some differences in conceptualization do exist. Finally, the traditional original, known in the past as literary metaphors are hard to miss and translators usually go out of their way to render them in the best possible way. This is evident in the results shown in Table 8: not a single original metaphor has been omitted. One of the reasons why the TT is longer than the ST is paraphrase of metaphor (Table  9). This occurs when the TL does not offer a suitable corresponding metaphorical expression. Another phenomenon observed in the TT is the existence of a metaphorical linguistic expression, where nothing of the sort is found in the ST. This is often the case with unintended lexicalized metaphor.   Having analyzed the relevant characteristics of the parallel texts in terms of length, richness of vocabulary, number of metaphors in ST and TT and their type, we were interested in obtaining information regarding the translation techniques applied to the translation of the three overarching categories of metaphor: lexicalized, conceptual, and original. Table 11 shows the transformations that the metaphors undergo in translation in terms of changing the mapping conditions, generalization and concretization. Of the 63 metaphors with different mapping condition identified in the texts, half of the lexicalized and conceptual metaphors had changed mapping conditions in the translation. This is due to the different conceptualization of the world in the different cultures and its reflection on language. In only a quarter of the original metaphors, however, the mapping conditions were different. These are cases in which the translator, on the basis of his/her personal judgment has changed the mapping. As our results show, this happens less frequently in original metaphors.
The other transformations, generalization and concretization, on the other hand are mostly language motivated, selected by the translator because of existing collocations in the target language. Concretization occurs three times as frequently as generalization because languages differ in the concrete conceptualizations, not the general ones.

Discussion
Though there are considerable dynamics and transformation of metaphor during translation, in the end of the day, the final counts by category are largely similar. A large number of metaphors in translation have the same mapping conditions in English and Bulgarian, due to common cultural background and globalization. This is in line with other studies such as the one by Burmakova and Marugina (2014), who investigate metaphor translation in literary discourse, Chervenkova (2015), who applies a similar to our analysis but confined to one text, and Park (2009), focusing on the analysis of metaphor translation in the short story genre.
Also, terminology, with which one of the texts (Heart of darkness) abounds in, largely made up of metaphor, is standardized and substituted by common words and phrases in translation, because, as a language of a sea-faring nation, English has more sea-related words than Bulgarian.
In literary discourse, due to the clustering of metaphors in nodes, in spite of loss of metaphor, due to linguistic, and culture-related differences, or translator-related preferences, the author's message still gets across in translation, mainly because of the metaphor network that spreads throughout the text, and the possibility of the translator to compensate for the 'zero solutions', or the loss of a metaphor by introducing another metaphor in a place where no such metaphor exists in the original. Similar results are obtained by Swain (2011) in her research into the intertextual perspectives of metaphor translation of literary texts, firmly established on Lemke's semantically-based theory of intertextuality.
Toury's law of growing standardization (1995) is not applicable to the translation of highbrow literature by well-established and experienced literary translators. The study described here reveals that the shifts to concretization are more frequent than the shifts toward generalization.
Lexicalized metaphors suffer more transformations that the other two types because they are more culture-specific and entrenched in language. The creative metaphors suffer the fewest transformations.
Foregrounded metaphors, grammatical metaphors and culturally bound ones turned out to be the ones most difficult to translate.

Conclusions
In an answer to our initial hypothesis, judging from the data we can convincingly state that it was proven wrong. The answer to the statement that the TT will be longer than the ST and that the type/token ratio will be greater in the TT (Olohan, 2002) is definitely negative as is the answer to the second claim that there will be many cases of explication, paraphrase and loss of metaphor. Another claim has also been found negative -that creative metaphors will predominate. Even in literary texts the number of lexicalized, sleeping metaphor is greater than the number of original metaphors. Our insight, though, that creative metaphor will translate readily in the TT, was found to be correct. Finally, our assumption that the TT will be much inferior to the ST and will not be suitable for close reading and analysis because of metaphor loss, was found to be completely wrong.