Carrying Across or Pulling down? Understanding Translation through its Metaphors: A Cross-linguistic Perspective

: e paper investigates how translation is conceptualised through metaphors employed in academic texts in English and Lithuanian focusing on translation problems. As established by previous research, metaphors are tools of rendering abstract thought in terms of more concrete experiences. e methodology of this investigation is based on the Conceptual Metaphor eory and further development in metaphor research, the main principles of Metaphor Identiﬁcation Procedure and metaphorical patterns. e results suggest that English tends to more frequently conceptualise translation as human and also as a dynamic activity, whereas Lithuanian opts for more static conceptualisation of translation in terms of object and material. Such tendencies might be linked, among other factors, to very diﬀerent etymologies of the verb ‘translate’ and its derivatives in English and Lithuanian as well as other senses of the word.


Introduction. Metaphor and translation: four intersecting points
Metaphor and translation have been studied from many different perspectives; they can be overviewed as four intersecting points.One was identified more than 40 years ago (Shuttleworth, 2014, p. 57) and is concerned with investigating metaphors in the source language and in translation in search of different strategies and techniques of rendering metaphors across languages.During the last three decades, aer modifying the understanding of metaphor and introducing Conceptual Metaphor eory (CMT;Lakoff, Johnson, 2003), research has expanded to include parallel contrastive studies of metaphors (Deignan, Potter, 2004;Stefanowitsch, 2004) and strategies of translating specific conceptual metaphors (Al-Hasnawi, 2007), metaphors in different genres and registers (Semino, 2011;Abdullah, Shuttleworth, 2013), metaphor universality and cultural specificity (Schäffner, 2004;Kövecses, 2014;Ibarretxe-Antuñano, 2013).
e second point of intersection is concerned with the role of metaphor in translation theories.Like in science, metaphors can help mould and formulate theories and thus play a theory-constitutive role (Knudsen, 2003).In science, for example, we can speak of genetic maps.e metaphor has eventually become a scientific term (Temmerman, 2001).Translation theories, as pointed out by Martin de León (2010), employ metaphors of transfer, footsteps, target, assimilation and reincarnation.Presumably, they may influence the translator in how he/ she renders the text in another language (Fuertes-Olivera, Velasco-Sacristán, 2001).Metaphors may change over time with changing theories (Martin de León 2010; Guldin, 2010).ese ideas have been instigated by CMT where the key claim is that metaphor is a matter of reasoning about more abstract things in terms of more concrete (Lakoff, Johnson, 1999, 2003, etc.), also that in human cognition human body and experience are paramount (Johnson, 2007).
e third point of intersection is concerned with the relationship of metaphor studies and translation studies with linguistics.Notably, contemporary metaphor studies and translation studies share interdisciplinary nature, as neither of them is confined strictly to language; they balance between language and cognition as a universal human parameter, also between language and culture as a more specific aspect of human life (Shuttleworth, 2014).e cognitive trend in linguistics reconciled translation and linguistics; their relationship until the arrival of cognitive sciences was aptly described as love-hate (Rojo, Ibarretxe-Antuñano, 2013, pp. 3-4).Now we speak not only about a cognitive metaphor theory, but also about a cognitive translation theory (ibid.).e cognitive trend has integrated language into human cognition thus erasing firmly set boundaries between them.Language and other forms of cognition are motivated by human experience, first of all, by bodily experience; also language is shaped by actual usage (Johnson, 2007;Rojo, Ibarretxe-Antuñano, 2013).e latter, metaphorically speaking, has a tightly interwoven thread of culture.In the process of translation, it is important to adjust the text to the target language and the target culture; therefore, a translator is oen not only bilingual but also bicultural (Al-Hasnawi, 2007;Shuttleworth, 2014, p. 60).e cognitive trend has given an opportunity for translation studies to move away from rigid prescriptivism and to account for various, previously considered deviant, translation strategies (Fernández, 2013).As a result, as noted by Guldin (2010, p. 187), translation theorists give preference to the treatment of translation as intercultural communication rather than as a purely linguistic process.
e fourth point of intersection between metaphor and translation is an interesting etymological overlap (Al-Hasnawi, 2007;Shuttleworth 2014;Guldin, 2010), In English (also EN), the etymology of the words translation and metaphor is concerned with meaning transfer, carrying across, moving from one place to another.us metaphor is a transfer between two domains and could be treated as a kind of translation.e etymology of the word translation may be one of the reasons why the theory of transfer in translation is so firmly established (see Schäffner 2004; Martin de León 2010).e parallel is extended to the terminology employed by both CMT and translation.e former employs the source (more concrete) and target (more abstract) domains in its understanding of metaphor, and translation makes use of the terms source and target languages.
Some of the ideas articulated under the second (metaphor is theoryconstitutive) and the fourth (etymological and terminological overlap) points are worth discussing in more detail.e theory-constitutive transfer metaphor, as claimed by Martin de León (2010, p. 80), is linguistically realised by the terms source and target languages.It is not quite clear why the author adheres to such interpretation.Another methodology (Pragglejaz, 2007;Steen et al., 2010) would treat the word target as manifestation of the metaphor of war or fight rather than transfer, since the word's basic meaning refers to 'an object that people practice shooting at' (ODE 2010).Source, in turn, is primarily concerned with the place where a river starts (ODE 2010), which has hardly anything to do with transfer either.
e etymological overlap between metaphor and translation, which has possibly given rise to the theory-constitutive metaphor of transfer, by some authors is described in reference to its Greek origin and claimed to be EN-specific (Guldin, 2010, pp. 178-180).is overlap is not found in, for example, Romance languages where the word traducio only preserves the meaning of displacement of material objects.Translation in other languages, such as Polish or Czech, preserves the meaning of transportation but also implies transformation and change.In Hungarian the word for translation is linked to the notion of turning and rotating.is metaphor (Somló, 2007, discussed in Guldin, 2010, p. 180) in translation means that the original is turned upside down.e Lithuanian (also LT) word for translate, translation 'versti, vertimas' is also based on the notion of turning round the axis of the hinges, as in opening the door (Knobloch 1970), trees pulled down during a storm or turning over the pages of a book (LKŽ, 2005;DLKŽ, 2011).e meaning of turning round the axis in contemporary LT has largely been lost in the word versti 'translate' and only kept in the word verti 'open by turning', which is considered an etymological cognate with versti (Knobloch 1970).In contemporary LT the meanings of pulling down and turning over have been preserved in the senses of the word versti (LKŽ, 2005;DLKŽ, 2011).

Translation-specific metaphors in academic discourse. Research question
Professional discourse such as legal, philosophical or political due to highly abstract content relies on metaphors to a very large extent (Lakoff, Johnson, 1999;Šeškauskienė, Stepančuk, 2014).As claimed by Zinken (2008quoted in Semino, 2011), each discourse is framed by a discourse-specific metaphorical projection.Academic discourse, inevitably involving abstract reasoning, is also highly metaphorical.As confirmed by empirical research, of four registers, news, fiction, conversation and academic, the last is the most metaphorical (Steen et al., 2010).Academic discourse is specific in that it consists mainly of two types of discourse: metadiscourse and subject-specific discourse.Presumably, both layers are conceptualised through their own metaphors, with metadiscoursal metaphors shared by all academic texts and subjectspecific discourses characterised by their own metaphors (Herrmann, 2013, pp. 125-126).Following this line of reasoning, academic discourse focusing on translation has its own metaphors, possibly language-and culture-specific.
is paper has chosen to explore how translation is conceptualised in two rather distant cultures -Lithuanian and English.e research question focuses on whether the way we write about translation tells us anything about how we reason about it considering vast encyclopaedic knowledge including the etymology of the word for translation and its adjacent senses and taking into account the role of metaphor in reasoning.A smaller size of the LT corpus is due, first of all, to shorter papers, with 4,058 words per article on average.In EN, the average number of words per article is 6,192.e difference could be also accounted for by the nature of LT, which has no articles, is highly inflected and tends to express relations between words through case forms rather than prepositions.Moreover, LT has a well-developed derivational morphology, which makes its words longer than in EN.

Data and methodological framework
e methodology of research employs the key principles of CMT.It relies on cross-domain mapping, or conceptualising "one thing in terms of another" (Lakoff, Johnson, 2003, p. 5).Metaphorical expressions (MEs) realise underlying metaphors.Later developments in metaphor research have demonstrated ample reliance on corpus based methodologies (Semino, 2006;Stefanowitsch, 2004 among others), which are also employed in this research.
e identification of MEs was based on the key word of the target domain of translation: translate and its derivatives in EN and versti and its derivatives in LT. e keywords were identified in the texts of both languages.To speed up the process of search for the words in question, the AntConc soware (Anthony 2014) was employed.In the EN corpus, the search focused on the lemma translat* and in the LT on the lemma vert*/verst*, followed by manual selection, since automatic selection in (numerous) cases where inflections modify the stems was not possible.
Further procedure relied on two methodologies: Metaphor Identification Procedure (MIP;Pragglejaz, 2007;Steen et al., 2010), and the notion of a metaphorical pattern (Stefanowitsch, 2004).MIP was defined by Pragglejaz group as consisting of four main steps (Pragglejaz, 2007, p. 3), which include very detailed procedure from identifying lexical units to establishing their (more basic) meanings and metaphoricity.
Metaphorical patterns are defined as frames consisting of a word from a source domain (also SD) and a word from a target domain (also TD) (Stefanowitsch, 2004, pp. 138-139).In this investigation, all collocates with the TD words translat* and vert*/verst* had to be collected and identified if they were metaphorical or not depending on SD words in the context surrounding the TD keyword.
To demonstrate, let us analyse one example.e expression the text is impossible to translate is non-metaphorical, whereas the translation requires a full understanding is metaphorical due to a clash between the meaning of the word translation (mainly interpreted as a process or result of changing of something that is written or spoken into another language (ODE, 2010)), which is not alive, and requires, which in its basic meaning is related to a human capability to express will or need; hence the conceptualisation of translation as a human being.

Results and discussion. Major tendencies
Following the definitions of translation given in dictionaries, it is understood as changing something that is written or spoken into another language (ODE, 2010;LKŽ, 2005).Some researchers (Abdulah, Shuttleworth, 2013) note that translation has two senses: the process of changing and the result of changing.e two senses could be interpreted in the framework of attention phenomena (Talmy, 2007); i.e. in some cases the focus of attention is on the mental activity and the result is backgrounded, in some other cases the result is more salient, foregrounded, whereas the mental process is less relevant.When the second sense is understood as a text, it could be interpreted in the framework of metonymy, with mappings occurring in the same domain where the process of mentally changing words of one language into those of another coexists with the result of such change: a written or spoken text.Arguably, a spoken or written text and translation referring to such a text are different in that the latter, unlike the former, is perceived as resulting from the process of changing words of one language into those of another.Both, the processual and the resultative, can be treated as equally viable senses of the word translation.e verb to translate in its multiple forms is much less ambiguous.
Below Table 1 gives raw and normalised frequencies of the selected words.e LT texts demonstrate higher frequency in the usage of the selected words.It might be related to the fact that in LT, both oral and written translation are subsumed under the same word, whereas in EN translation is mainly confined to written translation and interpretation refers to oral translation.As seen in Table 1, in LT, the lemma variation is greater than in EN.Aer retrieving all translat* lemmas in both languages, metaphorical patterns were identified based on contextual clues and establishing a mapping between a more concrete SD and a more abstract TD of translation.
e analysis into the metaphoricity of the patterns has revealed that 64.5% of all cases in EN and 47% in LT, were metaphorical (see Table 2 below).However, due to a slightly larger EN corpus, the parameter of normalised frequency is lower for EN than for LT, accounting for 97 MEs per 10,000 words in EN and 113 in LT.

Table 2
MEs with translat* in EN and LT Further analysis into SDs has shown that in EN their distribution is fairly even (Fig. 1), with object/ material, container, person and journey SDs covering about 70% of the data.In LT, the most numerously represented SD is that of object/ material (Fig. 2), accounting for almost half of the cases.Despite that in EN this SD also features quite prominently accounting for about a quarter of all the data, it is significantly less frequent than in LT.

Fig. 1
SDs of conceptualising translation in EN Notably, the container SD is also among those featuring prominently in both languages -it accounts for about one fih in both sub-corpora.In some cases container and object can be treated as overlapping SDs, since containers are oen concrete objects.However, objects are oen described as graspable, they can be put, taken, held, thrown, etc., which is not necessarily the case when we speak about containers, which are mostly identifiable through depth or inside.In this investigation container and object/material are treated as two distinct SDs.

Fig. 2 SDs of conceptualising translation in LT
As seen in Fig. 1 and Fig. 2, there are some domains which could be treated as LT-or EN-specific.Interestingly, LT is prone to more static conceptualisation and mainly sees translation as an object/material or container.In EN, in addition to the object/material and container, there are SDs such as person and journey, which are rather frequent in EN academic discourse in general.is is especially true of personification (Low, 1999;also see Master, 2001).e fact that it is much less frequent in LT has been confirmed by previous research (Šeškauskienė, 2010).e journey metaphor also features more prominently in EN than in LT (ibid.).is might be linked to personification, as we usually think of people involved in journeys rather than animals or artefacts.
Further major SDs in metaphorising translation will be described in more detail.e most prominent SDs will be given more attention.

e SDs of object/ material and container
In EN, when translation is understood as an object, contextual clues are usually verbs of giving, keeping, losing, or binding, also adding; translation may be described as transparent, etc.All of these discourse elements in their basic meaning are linked to material objects or substances, e.g.: (1) A verbatim translation will be given to the defense later (EN_5) (2) [...] translation can adhere closely to the literal wording of those elements (EN_6) Of particular interest is the notion of translation bound to a relevant culture, hence the term culture-bound translation.No such conceptualisation was found in the LT data.
In LT, metaphorical patterns pointing at conceptualising translation as an object are realised in collocations with the adjectives sklandus 'smooth', sunkus 'heavy, difficult', tikslus 'precise' or rišlus 'cohesive', which in their primary meaning point at the quality of hard surface (smooth), substance sticking together (cohesive), or a precisely measured amount of some material.Object or material is also signalled by verbs referring to adding or leaving out some information or parts of the text, twisting, bending.In LT, it is natural to discuss translation as something twisted, especially if it is wrong, e.g.: (3) […] buvo padaryta vertimo prasmę iškraipanti klaida.(LT_1) '[t]here was an error made twisting the meaning of the translation' Twisting seems to be LT-specific conceptualisation, apparently linked to the etymology of the LT word for translate.No such conceptualisation has been identified in the EN data.
Translation is frequently discussed as smooth, without lumps (sklandus).Smoothness, a characteristic of touching, when mapped onto translation, does not only define it but also adds positive evaluation.
A frequent image employed in discussing translation in LT is that of reflection.It is concerned with material objects reflected on a surface such as water, glass or mirror.Oen translation is discussed as reflecting culture, e.g.: (4) […] dažniau [buvo] analizuoti svetimų kultūrų atspindžiai vertimuose.(LT_3) 'Reflections of foreign cultures in translations have been analysed more frequently' In the EN data, such collocates were not found; however, the image of reflection is not excluded.EN is also prone to conceptualise translation as transparent, which is not characteristic of LT.
Conceptualising translation as a container is identifiable in LT in such collocations as vertimo turinys 'content of translation' or užpildytos pauzės vertime 'filled in pauses in translation'.In EN, the most indicative collocate words were full or fully, content and the prepositions in and into, which point at the interior, e.g.: (5) It should be noted that a qualifier must be added to fully translate chez l'Arabe.(EN_2) So content seems to be a cross-culturally acceptable collocate; full translation or fully translate are only found in EN. e locative case in LT, roughly corresponding to the EN preposition in, seems to be one of the strongest indicators of containers.

e SDs of person and journey
e two SDs seem particularly relevant for EN.Personification is oen concerned with patterns when translation takes the subject position in a sentence.Some authors refer to such subjects as inanimate (Master 2001), their combination with animate verbs is very EN-specific (ibid.).In academic discourse such patterns are oen seen as a strategy of hedging and persuasion, showing either a lack of commitment on the part of the author or avoidance of direct criticism towards his/her opponents.In LT academic discourse such patterns are less frequent (Šeškauskienė, 2010).
Such patterns can also be interpreted as metaphors (Low, 1999).Contextual indicators of such metaphors are verbs like take, give, require, consider, accentuate, etc. used in the predicate position with translation employed in the subject position.e verbs in their basic meaning are concerned with actions and activities characteristic of people or, less frequently, animals.Many verbs refer to mental activity, which help support the claim about personification, e.g.: (6) e translation in Example 4 accentuates an important point.(EN_4) Interestingly, translation can also be seen as faithful, even though the word in the example below is signalled by inverted commas, which usually marks the awareness of metaphor on the author's part (Deignan et al., 2013, pp. 20-23), e.g.: Human features of translation in LT are mostly limited to helping someone to create (kurti), perceive (suvokti), foresee or imply (numanyti).As suggested by the data, in LT translation is much less frequently personified.
Journey metaphor is also more numerously represented in EN.One of its contextual indicators is the word approach, with its basic meaning of coming closer to someone in space.Journey is also indicated by the words direct, translators are described as deviating om the text, they are unable to keep pace, translation is guided by research, genre conventions function as signposts, e.g.: (9) e translation was not very literal as it was guided by research.(EN_3) (10) (…) the Danish genre conventions functioning as signposts when translating statutes into English.(EN_4) In LT, the realisation of the metaphor is limited to translator encountering difficulties, translators following closely the original or moving further away from it, or translators lagging behind the speakers in oral translation.

e SDs of physical activity and war
e physical activity SD is rather general.It refers to metaphors resulting from conceptualising time in terms of space, especially when in EN translation is seen as a long procedure or process.Notably, length is primarily spatial.
In LT, in addition to what was said about EN, the most frequent indicator of space mapped onto time was the word sinchroninis 'synchronous'.It is of Greek origin where the element syn-indicates togetherness, referring to location and then to temporal relations.Also LT makes use of cutting or interrupting; translation is cut in time.In one case in LT translation is treated as cultural transfer, which is apparently derived from physical, or spatial, transfer, e.g.: (11) Vertimas visada yra tarpkultūrinis perkėlimo veiksmas.(LT_4) War metaphors are oen signalled in both languages by the strategy of translation.In EN, another frequent indicator is the term target language or, less frequently, target text/ audience.e primary meaning of target is concerned with shooting and attacking, which is why the word is treated as an element of the domain of war.Moreover, in EN there have been cases when translation was seen as a battlefield or an enemy, which was resisted; in court, defendants may fall victims of a standard of translation, e.g.: (12) Some Japanese have argued that the defendants at the trial were victims of a standard of translation that fell short of Nuremberg's […].(EN_5) e above scenario suggests a very negative attitude towards translation, possibly concerned with gross errors leading to misinterpretation of the text.Translation as an enemy can be inferred from the text suggesting resistance, e.g.: (13) It is the very resistance to translation that constitutes the unique identity of these languages.(EN_1) In LT, war metaphors are mostly realised by employing strategies and the scenario of suffering.e word strategy has eventually become conventionalised and is treated as a term (Deignan et al., 2013, p. 18).A scenario where victims or sufferers are involved is identifiable in many types of academic and public discourses in LT.In translation studies, it is usually the text which suffers from bad translators, e.g.: (14) Dėl greito kalbos tempo labiausiai nukenčia sinchroninio vertimo turinys.(LT_2) Notably, the suffering scenario is not always as clear-cut.In (12), the victims were people who participated in the trial; in ( 14), the victim is the translated text itself (its content), since some information of the original is missing in the translation.What suffer most are form, content, quality, efficiency, parameters, relationship, and they are mostly found in nonfiction texts, as attested by a simple search in the Corpus of Contemporary Lithuanian (CCL).

Building, product, communication and other SDs
e building SD is usually signalled by such contextual indicators as structure, construction, based on, build/built, etc.In the EN data, we come across such expressions as built-in commentary or construction of a new reader or a text.
e conceptualisation of verbal activity in terms of building is rather well-established (Semino, 2006).e focus in most cases is on the lower part of a building or construction.We hardly ever speak about windows or roofs of theories, but bases or foundations are frequent (Grady, Johnson, 1997).
e product metaphor is realised through patterns involving input, output, by-product, quality, etc. e two languages do not manifest any specific differences.
Metaphors of communication and mediation were only found in EN.At the level of realisation, there were several cases where senders and receivers of translated texts were identifiable; translators were conceptualised as intermediaries between languages, e.g.: (15) (…) translation is an endeavour deemed successful as long as its results were accurate […]and therefore "transparent"; translators were treated as if they were "invisible" intermediaries between one language and another.(EN_5) Visibility in this case is related to negative rather than positive evaluation.e 'invisibility' of a translator means that the translator has done his/her job without interfering in the text too much.On the other hand, a transparent text, a metaphorical quality also concerned with vision, is positive and closely linked to the metaphors light (also easy to see through. is good and dark .noteasy to see through. is bad as well as understanding is seeing.What is transparent is easy to understand, what is not, is difficult, hence unacceptable.Some minor metaphors include cooperation, when translators are seen as those who have to cooperate with readers, mechanism, since translators operate with several points: terminology, linguistic structures, and textual conventions, food, e.g.well-seasoned translation.Several cases involved understanding translation as foreign as opposed to domestic, own.is metaphor also carries evaluative load; foreign is associated with distant, hence negative. Finally, it is worth drawing attention to a frequent metaphorical term of source (e.g.source language, text, etc.) employed in many texts under investigation.In its primary meaning source is concerned with nature and refers to a place where a river or a stream starts.e natural phenomena SD is not very detailed or elaborate either, as there are very few clues to it in the data.LT in this case sticks to an explicit, non-metaphorical term originalo kalba/ tekstas 'language/text of the original'.Some cases can be interpreted as mixed metaphors, when in the same situation several SDs are identifiable sometimes rendering rather incoherent images.In the example below, translation is conceptualised as a human being or force which is resisted; at the same time, it is also conceptualised as a bridge, cf.: (16) [...] it [the German text] resists straightforward translation and it is oen far from being the bridge between cultures.(EN_6) e two images are hardly compatible: [text as] person is capable of resistance and at the same time is perceived as a bridge, probably related to the journey metaphor.When the two SDs are employed in the same utterance, its meaning may be difficult to process.A tendency to objectify rather than personify translation in the LT data might be linked to the etymology and other senses of the LT translation words.e basic, and historically older, sense of bending, twisting or rotating is preserved in other, metaphorical, senses of the word, including translation.e physical sense focuses on an object physically manipulated by an agent.No such senses are associated with the EN translation.Its etymology is linked to more dynamic situations involving a scenario of carrying something across.e idea is preserved in the first element of the word trans-, in contemporary EN employed in many derivatives, e.g.transatlantic.

Summary and concluding remarks
e results of this investigation have demonstrated that academic texts employ metaphors which perform a theory-constitutive role and oen carry evaluative load.Cohesive, smooth, transparent translation is usually very good; so is the one where the translator is 'invisible'.What is foreign and unknown is negative, so is the translator's presence.e latter is probably concerned with changing or distorting information, dispreferred by the reader.
Despite a fairly small corpus of this investigation, it seems to point at some important and sometimes overlooked aspect of metaphor research concerned with the etymology and other senses of the word.In further research, the hypothesis could be verified on other languages and larger corpora.
e LT and EN data for the present investigation, 65,558 words in total, has been collected from journals publishing research on translation.e LT corpus (28,408 words) has been collected from two research journals, Vertimo studijos and Kalbų studijos.e corpus consists of seven articles written between 2010 and 2013.e EN corpus is slightly larger (37,150 words) and consists of six articles collected from different specialised journals between 2003 and 2013.e journals include the following titles: International Journal of Translation and Interpreting, Kalbų studijos, Translation Journal, Vertimo studijos, Target, Translation Studies.e full list of articles is given at the end of the paper.

Fig. 3
Fig. 3 below summarises the most frequent SDs in EN and LT.According to individual categories, the object/ material SD is preferred in LT, whereas container and physical activity are similarly represented in both languages.Personification and journey are more prominent in the EN data.

Fig. 3
Fig. 3 Most frequent source domains in EN and LT, raw data Preference of the object/material SD in the LT data signals a more static picture of academic translation discourse.e EN data refers to more dynamic scenarios and foregrounds personification, which demonstrates a more prominently featuring parameter of embodiment.

Table 1
Translat* lemmas in EN and LT 1