A Corpus-based Study of Units of Translation in English-Persian Literary Translations

Zohreh Shahrestani Abstract

In the present study, the notion of unit of translation as a challenging issue in Descriptive Translation Studies (DTS) is addressed. Considering this notion from a product-oriented point of view as "the TT unit that can be mapped onto a ST unit" (Baker, 2001: 286), the researcher's main concern is to investigate a hierarchy of units of translation (UTs) proposed by Newmark (1991: 66-68) including word, phrase, clause, sentence, and paragraph in the literary translations. At the preliminary stage, two questions were raised to detect the most frequent UT adopted by the professional literary translators, and to explore the relationship between the UTs and the free-literal dichotomy in terms of the occurrence of unit/rank shifts. To this end, a corpus of three famous English novels (originally written in English by the renowned authors) and two best-selling translations of each (done by professional translators) were chosen to be analyzed. Through a contrastive analysis, two hundred and ten coupled pairs of ST-TT segments were extracted from the first ten pages of each novel and its two translations based on establishing relations of equivalence between the ST-TT segments and adopting sentence as the major unit of analysis. The UTs adopted in the ST-TT segments were then identified. The obtained results of the UT categories demonstrated that the most frequent UT adopted by the professional literary translators was sentence. The unit-shifts applied in the UTs were also signified. The statistical calculation of frequency of unit-shifts in each translator's UTs proved that the more frequent is the occurrence of unitshifts in the UTs of the translator, the more deviated is his translation from the formal correspondence, the more different the size of his UTs is, and finally the freer his translation will be. Key Words: Descriptive Translation Studies (DTS), units of translation, free-literal .dichotomy, unit/rank shifts, equivalence, formal correspondence

1. Introduction
Translation Studies is a new discipline which is concerned with the study of theory and phenomena of translation. A classical concern for translation theory which is frequently mentioned in older literature on the subject is the level at which equivalence should be established, i.e. what units of translation one should choose during the translation process. Catford (1965:21) suggests that the goal of translation theory is to define the nature of translation equivalence. To him,
The central problem of translation practice is that of finding TL translation equivalents. A central task of translation theory is that of finding the nature and conditions of translation equivalents.

In translation studies, much discussion in the translation literature has focused on identifying what should be equivalent in a translation. For example, with regard to the linguistic form, discussion in translation literature has focused on whether equivalence is to be pursued at the level of words, clauses, phrases, sentences, paragraphs, or the entire text. Accordingly, this has given rise to the emergence of the concept of Translation Units which is one of the key concepts in translation theory that has exercised translation theorists over a very long period. In the field of translation, from a product-oriented approach, a translation unit is a segment of a target text which the translator treats as a single cognitive unit. The translation unit may be a single word, or it may be a phrase, a clause, a sentence, or even a larger unit like a paragraph. In translation studies, the issue of UT is frequently raised in conjunction with that of translation equivalence. As Sager (1994: 222) puts it, both lie at the heart of any theoretical or practical discussion about translation. This is because theorists, consciously or unconsciously, take the UT as a compartment in which what they believe to be translation equivalence materializes. There is a point in establishing equivalence, Toury believes, only insofar as it can serve as a stepping stone to uncovering the overall concept of translation underlying the corpus it has been found to pertain to; besides, the notion of equivalence may also facilitate the explanation of the entire network of translational relationship and the individual coupled pairs as representing actual translation units under the dominant norm of translation equivalence (1995: 86). In this regard, one of the tasks of the researcher wishing to probe into the translation units is to establish the

equivalent relationships between the coupled pairs of ST and TT segments which can pave the way for the identification and classification of units of translation at different levels. In other words, to investigate unit(s) of translation that the translator chooses during the translation process, one needs to establish a relation of equivalence between the ST and the TT. In earlier work on translation equivalence, Catford (1965: 20) defines translation as "the replacement of textual material in one language (SL) by equivalent textual material in another language (TL)". He distinguishes textual equivalence from formal correspondence, which are respectively called by Nida as dynamic equivalence and formal equivalence. A formal correspondent is "any TL category (unit, class, structure, element of structure, etc.) which can be said to occupy, as nearly as possible, the "same" place in the "economy" of the TL as the given SL category occupies in the SL". A textual equivalent is "any TL text or portion of text which is observed on a particular occasion to be the equivalent of a given SL text or portion of text" (ibid: 27). It is worth mentioning, however, that departures from formal correspondence between the source and target texts denote Translation Shifts (ibid: 73), the investigation of which has a long-standing tradition in translation studies. In other words, shifts are deviations or changes that occur at every level during the translation process as a result of the systemic differences between the source and target languages. There has been a great argument among theorists about the length (size) of unit of translation. For most of them, the length of translation units is an indication of proficiency, with professional translators working with larger units (sentence, discourse, or text) and moving more comfortably between different unit levels. This controversial argument about the length of unit of translation is, according to Newmark (1988: 54), a concrete reflection of an age-old conflict between free and literal translation: The freer the translation the longer the UT, the more literal the translation; the shorter the UT, the closer to the word. Therefore, despite major shifts of viewpoint on translation, one of the oldest as well as the most decried conflicts in translation has been the concept of literal versus free translation, or the distinction

between word-for-word translation and sense-for-sense translation. The controversy over literal versus free translation has a long history, with convincing supporters on each side. In this research, the issue of units of translation is approached from a productoriented viewpoint to seek answers for the two following two questions: RQ1: What is the most frequent UT among the professional translators of the famous English novels? RQ2: What is the relationship between the UTs and the kinds of translation, i.e. free vs. literal, adopted by the professional literary translators in terms of the occurrence of unit-shifts?

2. Theoretical Discussions
2.1Descriptive Translation Studies (DTS) A branch of Translation Studies, developed in most detail by Toury (1995), that involves the empirical, non-prescriptive analysis of STs and TTs with the aim of identifying general characteristics and laws of translation (Hatim and Munday, 2004: 338). According to Munday (2001: 10-11), DTS is a branch of 'pure' research in Holmes's map of Translation Studies and has three possible foci: examination of the product, the function, and the process. 2.2 Translation Units According to Baker (2001: 286), the term 'unit of translation', considered from a product-oriented approach, is defined as "the TT unit that can be mapped onto a ST unit". Newmark (1991: 66-68) assumes the main translation units to be a hierarchy: text, paragraph, sentence, clause, group, word, and morpheme. 2.3 Equivalence Baker (2001: 77) defines equivalence as the relationship between a ST and a TT that allows the TT to be considered as a translation of the ST in the first place. Vinay and Darbelnet view equivalence-oriented translation as a procedure which "replicates the

same situation as in the original, whilst using completely different wording" (cited in Shuttleworth and Cowie, 1997: 51). 2.4 Dynamic/Textual equivalence vs. Formal equivalence Defined by Nida (1964, cited in Bassnett, 1980: 33), the former (also known as functional equivalence) is "the closest natural equivalent to the source-language message" (ibid: 166) and attempts to convey the thought expressed in a source text (at the expense of literalness, original word order, the source text's grammatical voice, etc., if necessary); while the latter (also known as formal correspondence) attempts to render the text word-for-word (at the expense of natural expression in the target language, if necessary). Also, defined by Catford (1965: 27), the former (also known as textual equivalence) is "any TL text or portion of text which is observed on a particular occasion to be the equivalent of a given SL text or portion of text" and the latter is "any TL category (unit, class, structure, element of structure, etc.) which can be said to occupy, as nearly as possible, the same place in the economy of the TL as the given SL category occupies in the SL". 2.5 Shift As far as translation shifts are concerned, Catford (1965: 73) defines them as "departures from formal correspondence in the process of going from the SL to the TL", i.e. if translational equivalents are not formal correspondent. According to AlZoubi and Al-Hassnawi (2001: 2), shifts should be defined positively as the consequence of the translator's effort to establish translation equivalence (TE) between two different language systems. To them, shifts are all the mandatory and optional actions of the translator to which s/he resorts consciously for the purpose of natural and communicative rendition of an SL text into another language (ibid).

2.6 Unit/rank shift Catford (1965: 79) defines unit/rank shifts as those departures from formal correspondence in which "the translation equivalent of a unit at one rank in the SL is a unit at a different rank in the TL". 2.7 Literal Translation

Literal or word-for-word translation is defined by Robinson as "the segmentation of the SL text into individual words and TL rendering of those word-segments one at a time" (1998, cited in Baker, 2001: 125). A literal translation can be defined in linguistic terms as a translation "made on a level lower than is sufficient to convey the content unchanged while observing TL norms" (Barkhudarov, 1969, cited in Shuttleworth and Cowie, 1997: 95). In a similar vein, Catford also offers a definition based on the notion of the UT: literal translation takes word-for-word translation as its starting point, although because of the necessity of conforming to TL grammar, the final TT may also display group-group or clause-clause equivalence (1965: 25). 2.8 Free Translation Also known as sense-for-sense translation, it is a type of translation in which more attention is paid to producing a naturally reading TT than to preserving the ST wording intact (Shuttleworth and Cowie, 1997: 62). Linguistically, it can be defined as a translation "made on a higher level than is necessary to convey the content unchanged while observing TL norms" (Barkhudarov, 1969: 11, translated, cited in ibid). In other words, the UT in a free translation might be anything up to a sentence (or more) even if the content of the ST in question could be reproduced satisfactorily by translating on the word or group level (ibid). Besides, according to Catford (1965: 25), it is a prerequisite of free translations that they should also be unbounded as regards the rank (or level) on which they are performed. Free translations are thus generally more TL-oriented than literal translations.

3. Methodology
Through conducting this research, an attempt has been made to investigate the argument about the problematic nature of units of translation in relation to free and literal translations adopted in English-Persian literary translations regarding the unitshifts. Put it in another way, the present research seeks to study translation units that the professional literary translators adopt in the process of translating famous novels from English into Persian, and it is carried out by establishing a relation (of equivalence) between the coupled pairs of ST and TT segments (that is to say, to ascertain whether the translated literary texts are the closest natural equivalent to the original message (Nida's definition of translation, 1964: 166), i.e. dynamically

equivalent, or formally equivalent), while taking into account the dichotomy of freeliteral approach to translation in terms of the occurrence of unit-shifts in the UTs. So the approach is limited inasmuch as the researcher has looked at Units of Translation only from the angle of the product of translation. As a consequence, this research is placed within the framework of Pure Translation Studies-in Holmes's map of translation studies (Toury, 1995: 10, cited in Munday, 2001: 10-12) -which actually has Descriptive Translation Studies (DTS) as one of its major branches. In fact, DTS embarks upon examination of the product, the function and the process as three focal points among which the first one is highlighted in the course of this research. Since this study is concerned with the product of translation and is a comparative analysis of several TTs of the same ST, it is a "Descriptive" research. Stated by Farhady, "Through descriptive method, researchers attempt to describe and interpret the current status of phenomena" (2001: 144). Descriptive research is defined by Birjandi and Mosallanejad (2002: 184-86) as the basis for qualitative research that deals with what is happening now. So the design of this research is "descriptive" content analysis. Moreover, this research goes under the heading of "Qualitative". A qualitative research explains how all parts work together to form a whole. Patten defines qualitative research as "an effort to understand situations in their uniqueness as part of a particular context. It is not attempting to predict what may happen in the future, but to understand the nature of the setting" (cited in Birjandi and Mosallanejad, 2002: 76-7). Moreover, through several subcategories Farhady (2001: 144, 154) represents for descriptive method of research, "Casual-Comparative" method which is, in turn, a subcategory of "interrelational" methods seemed the most appropriate to the researcher to conduct this research. The research is by nature comparative in that it is aimed at comparing and contrasting pairs of ST and TT segments so as to find the most frequent UT among the professional literary translators and to trace and discover the relationship between their UTs and the kinds of translation, i.e. free vs. literal, applied by them in terms of the occurrence of unit-shifts in UTs in the move from the ST to the TT. Thus, it can be found out that this study falls under a comparative category for research method.

3.1 Corpus Selection Procedure In order for the samples of this research to meet the representativeness criterion, i.e. to be representative of the whole population, the selection of materials was based on a non-random sampling criterion which is described by Farhady (2001: 212) as a process of choosing research population when random sampling is not possible. For the sake of choosing certain English-Persian literary works, both the source texts and the target texts were selected based on a purposive sampling which is, according to Farhady (ibid: 212), a procedure for selecting a non-random sampling, and defined by him as "the procedure directed toward obtaining a certain type of members with predetermined characteristics" (ibid). Taking all these criteria into account, the novels and the translations of each were meticulously selected. These were then supposed to be segmented, compared and contrasted from the viewpoint of units of translation. Indeed, the corpus used in this study is a parallel corpus, that is to say, original English source texts and their translations in Persian. A parallel corpus is defined by Olohan (2004: 24) as "a corpus consisting of a set of texts in one language and their translations in another language". The English novels were selected based on purposive sampling to fulfill the following selection criteria, Originally written in English, Being regarded as masterpieces, Closely related to each other in terms of genre, and Written by renowned authors. Persian translations were also selected based on purposive sampling to include those consistent with the following certain criteria: Best-selling Persian translations, Being considered as the pick of the numerous existing translations, and Done by professional translators. The final samples are presented in Tables 1 & 2. Table 1 The list of English novels .No Novel Title Author Year of the First Selected

1 2 3

Heart of Darkness Lord of the Flies Cry, the Beloved Country

Joseph Conrad William Golding Alan Paton

1899 1954 1948

1-10 1-10 1-10

The list of Persian Translations

Table 2

Persian Translations of Heart of Darkness Selected Pages 1-10 1-10 Year of Publication 2001 1986 Year of the First Edition 1985 1986

Translator

Title

.No

1 2

Persian Translations of Lord of the Flies Selected Pages 1-10 1-10 Year of Publication 1984 1984 Year of the First Edition 1984 1984
( ). ()

Translator

Title

.No

1 2

Persian Translations of Cry, the Beloved Country Selected Pages 1-10 1-10 Year of Publication 1982 2004 Year of the First Edition 1972 1983

Translator

Title

.No

1 2

3.2 Data Collection Procedure In order to manage the process of data collection, the first ten pages of each novel and their Persian translations were selected. Then, to make a thorough comparison

between the STs and their selected TTs possible, the first two hundred and ten sentences from those ten pages of each novel were extracted. The extracted sentences of each novel were then matched with their two translations. In this way, the ST-TT segments were specified for each novel based on the established equivalent relations. The ST-TT segments extracted from each novel and its two translations were then included in the separate tables related to each novel. Here, a point to mention is that the researcher had to adopt a unit of analysis to make it possible to specify ST-TT segments and later to make it feasible to identify the UTs applied in each segment and, hence, to discover the occurrence of unit-shifts in those UTs. 1) So, the first stage was to specify the ST segments. For that matter, sentence was basically adopted as the major unit of analysis. Because it is mainly regarded as a meaningful unit that conveys the message completely. Besides, among the language levels the sentence is where sentence linguistics and text linguistics overlap, and decisions made at any other language levels will be duly reflected within the contour of the sentence, the primary building block for TL text construction (Hewson and Martin, 1991: 86). However, the researcher encountered some rare cases in each ST (novel) where a complete message was conveyed through a word or phrase, so she considered word or phrase as the minor units of analysis. Moreover, in order to specify the ST segments the researcher had to stick to a punctuation mark to separate the units of analysis; therefore, she essentially used full stops to separate the ST sentences. Because among punctuation signs that operate to (con)textualize, full stops are the most significant marks since they signal the full sentential independence of a language segment (Zhu, 1996: 438). 2) Yet, after specification of the ST segments as mentioned above, the two translations of each ST segment were specified in the next stage. Since the translations were supposed to be specified based on the established equivalent relation between the ST and the TT, the translation column in the tables is entitled equivalent translation, which is to Catford (1965: 27) an empirical phenomenon, discovered by comparing SL and TL texts. Also, it was important to the researcher whether the translation was formally equivalent, i.e. directed more towards the form of the ST or formal correspondence, or dynamically equivalent which is described as "the closest natural equivalent to the source-language message" (Nida, 1964: 166).

10

The researcher actually regarded it as a basis to later enable identification of the occurrence of unit-shifts in specified UTs.

3.3 Data Analysis Procedure After specifying the ST-TT segments, they had to be analyzed to see what UT(s) were applied in them by each translator. One source of inspiration for choosing the units of translation was Newmark (1991: 66-68)'s statement that assumes the main translation units to be a hierarchy: text, paragraph, sentence, clause, phrase/group, word, and morpheme. Yet, in order to increase the degree of manageability of the research, an attempt was made to select those UTs which are frequently preferred as basic working UTs by the translators. Therefore, in ascending order, word, phrase, clause, sentence and paragraph were selected as categories of UT. 3.3.1 Investigating Units of Translation 3.3.1.1 Word as UT: It is clear that, despite its apparent convenience, the word on its own is unsuitable for consideration as the basis for a unit of translation. Further, although the researcher has been mostly concerned with the sentence as unit of analysis, there were in fact some rare cases in each story where the researcher had to regard word as UT, because the translator could have successfully conveyed the message to the reader through one word in TT, as in the following cases: Heavens!
That's right. Tomorrow, she said.

! . .

3.3.1.2 Phrase as UT: Hatim and Mason (1990: 180) maintain that there is no doubt that translators work with phrases as their raw material, and equivalence cannot truly be established at these levels. Phrase is considered as "two or more words that function together as a group" (Swan, 2005: xxii) and conveys a thorough message per se, as in the following cases:
Old knitter of black wool. "Sucks to your ass-mar!" This letter, Stephen. ! . .

11

3.3.1.3 Clause as UT: Syntactically clause forms a part of a sentence and has a subject-predicate structure which is not complete by itself and is semantically dependent (Richards and Platt, 1992: 52-53); therefore, it is not a meaningful unit and should be completed by another sentence. So this UT has not been separately observed. In fact, the clauses were taken into account in the form of sentences incorporating them, i.e. complex sentences- which contain one or more dependent (or subordinate) clauses and an independent (or main) clause- and compound-complex sentences- which contain two or more independent clauses and one or more dependent clauses (Frank, 1972: 1). In the present study, the clauses have been taken into consideration under two broader constituent categories, i.e. complex sentence or compound-complex sentence. Also, the number of both complex sentences and compound-complex sentences is considered as indicative of clause as UT. 3.3.1.3.1 Clause as UT: Complex Sentences: As defined by Frank (1972: 1), complex sentence contains one or more dependent (or subordinate) clauses and an independent (or main) clause. For example:
They were men enough to face the darkness. "I expect there's a lot more of us scattered about.
. .

There is a lovely road that runs from Ixopo into the hills. .

3.3.1.3.2 Clause as UT: Compound-Complex Sentences: Defined by Frank (1972: 1), compound-complex sentence contains two or more independent (or main) clauses and one or more dependent (or subordinate) clauses. For example:
It was the biggest thing in the town, and everybody I met was full of it.
.

When he gets leave he'll come and rescue us.

For there there is a multitude of buses, and only one bus in ten, one bus in twenty maybe, is the right bus.

3.3.1.4 Sentence as UT: According to Richards and Platt (1992: 330), sentence is the largest unit of grammatical organization within which parts of speech (e.g. nouns, verbs, adverbs) and grammatical classes (e.g. word, phrase, clause) function,

12

and a sentence normally consists of one independent clause with a finite verb. Also, according to Frank (1993: 220), a sentence is a full, independent prediction containing a subject plus a predicate in the form of independent clause. Elsewhere he defines the independent clause as a full prediction that may stand alone as a sentence [222]. Based on the independent clause(s) consisting sentences, the sentences can be generally classified into two types: simple and compound, both of which contain independent clause as their only building block. So this UT was treated in simple sentences and compound sentences, and the number of both simple sentences and compound sentences is reckoned as indicative of UT as sentence. 3.3.1.4.1 Sentence as UT: Simple Sentences: To Frank (1972: 1), simple sentence contains one full subject and predicate and can take the form of a statement, a question, a request, or an exclamation. Such a sentence has only one full prediction in the form of an independent clause (Frank, 1993: 222). For example:
His remark did not seem at all surprising. Piggy bore this with a sort of humble patience. It is not an easy letter. .
. .

3.3.1.4.2 Sentence as UT: Compound Sentence: As stated by Frank (1972: 1), compound sentence contains two or more sentences joined into one by punctuation alone, punctuation and a conjunctive adverb, or a coordinate conjunction; when such sentences are joined coordinately, they are each called independent clause. Such sentences have two or more full predictions in the form of independent clauses (ibid, 1993: 222). For example:
I gave my name, and looked about. . Ralph giggled into the sand. She took the letter and she felt it.
.

3.3.1.5 Paragraph as UT: Defined by Richards and Platt (1992: 262), paragraph is a unit of organization of written language, which serves to indicate how the main ideas in a written text are grouped. In text linguistics, paragraphs are considered as macrostructure of a text and they group sentences which belong together and deal with the same topic. Consequently, a paragraph, as a macro-structure, usually consists of a group of related sentences such as simple, compound, complex, or compound-

13

complex which together incorporate a whole unit. Yet, in this study, paragraph as UT was found to be exclusively implemented in the both translations of Heart of Darkness by the same number, and no cases of such UT were found in the both translations of the two other stories. For example:
And at last, in its curved and imperceptible fall, the sun sank low, and from glowing white changed to a dull red without rays and without heat, as if about to go out suddenly, stricken to death by the touch of that gloom brooding over a crowd of men.
.

3.3.2 Investigating Unit-shifts in the UTs Applied by the Translators Based on the categories mentioned above, the UTs applied in the ST-TT segments were identified. Concurrently, while identifying the UTs in the ST-TT segments, unit/rank shifts or those departures from formal correspondence in which "the translation equivalent of a unit at one rank in the SL is a unit at a different rank in the TL" (Catford, 1965: 79) were also sought after. The unit-shifts were specified to later gauge the relationship between the UT and the free-literal dichotomy. Apparently, according to Catford, shift is not formally equivalent. In fact, if the SL is imitated exactly in the TL, the result is called formally equivalent translation which is awkward or unnatural, more directed towards the form of the ST, and basically source-oriented. However, to avoid such a translation, the translator may deviate from the ST and move away from close linguistic equivalence, so a shift occurs and the resulting translation distancing from formal correspondence (equivalence) is called dynamically (textually) equivalent translation which is described as "the closest natural equivalent to the source-language message" (Nida, 1964: 166). The kind of shift which is taken into account in the current study is unit/rank shift that is a subdivision of category shift and is defined by Catford (1965, cited in Munday, 2001: 61) as the shift "where the translation equivalent in the TL is at a different rank to the SL", as in the following cases:

14

4. Discussion of Findings and Conclusions

While analyzing the collected data, it seemed logical to calculate the frequency and percentage of units of translation applied in the three novels as well as the frequency and percentage of unit-shifts in the UTs adopted by the professional translators of those novels. Based on the findings of the analysis, the results of the statistical analysis are presented in the following tables: Table 3 Frequency and Percentage of Units of Translation in Heart of Darkness, Lord of the Flies, and Cry, the Beloved Country Units of Translation Subcategories of Units of Translation Frequenc y Total Frequenc y 24 44 580 184 419 691 272 4 4 20.25 0.29 0.29 13.70 31.19 51.45 Percentage Total Percentage % 1.81 3.27 43.18

Word Phrase Clause Complex Sentence Compoundcomplex sentence Simple Sentence Compound Sentence

24 44 396

1.81 3.27 29.48

Sentence

Paragraph

As Table 3 reveals, the most frequently applied unit of translation among the literary translators is the sentence which remarkably includes the majority of samples, the highest frequency as well as the highest percentage which ranks sentence as the top list category and the foremost adopted unit of translation. In addition, clause covering a wide range of samples and having an approximately high frequency and

15

percentage occupies the second prominent position among the applied units of translation. Lastly, phrase, word and paragraph are respectively other applied units of translation whose frequency and percentage are not highly significant. A summary of the statistical findings obtained in this section is presented in the following chart:

Units O Translation f

43.18

sentence Paragraph W ord 51.45 Phrase C lause 3.27 1.81 0.29

This leads to the conclusion that successful literary translators are mostly concerned with the sentence as their unit of translation to find the closest natural equivalent to the source-language message and to best convey the message to the TL reader. Table 4 Frequency and Percentage of shifts in the UTs in Heart of Darkness, Lord of the Flies, and Cry, the Beloved Country Novels Heart of Darkness Lord of the Flies Translators Hosseini's Translation Hajati's Translation Azad's Translation Ardekani's Translation Daneshvar's Translation Cry, the Beloved Country Hafezipoor's Translation Frequency 88 97 75 98 77 88 % Percentage 41.90 46.19 35.71 46.66 36.66 41.90

16

Since the occurrence of unit-shifts, as departures from formal correspondence in the UTs in the move from SL to TL, is the focus of study in this section, here the frequency and percentage of shifts occurred in the UTs of each translator have been calculated separately to make the comparison possible. As indicated in Table 4, unitshift has occurred most frequently in Ardekani's translation of Lord of the Flies, so it contains the highest percentage. Also, in Hajati's translation of Heart of Darkness a nearly similar number of unit-shifts has occurred. It can be representative of the fact that these two translators are highly oriented towards deviating from the ST, applying translation units of a size different from the ST, and, thus, their translations tend to be freer. The obtained results have been displayed in the following graph:
occurrence of units-shift in translations
50 45 40 35 Percentage 30 25 20 15 10 5 0 Hosseini's Translat ion Hajat i's Translat ion Azad's Translat ion Ardekani's Daneshvar's Hafezipoor's Translat ion Translat ion Translat ion 41.9 35.71 36.66 46.19 46.66 41.9

Translators

It can be inferred that, as far as the product-oriented view of the UTs is concerned, the more frequent is the occurrence of shifts in the UTs of the translator, the more deviated is his/her translation from the formal correspondence, the more different the size of his/her UTs is, and finally the freer his translation will be. Thus, there is a direct relationship between the number of occurrence of shifts in the units of translation (i.e. unit-shifts) and free translation. Besides, although frequency of the occurrence of unit-shifts is closely related to a free translation being produced and it

17

may make a translation freer, it may change the size of the UTs to a longer or shorter UT; so for the UTs it is the matter of either/or.

5. Final Words
The findings, theoretical discussions, as well as practical evidences of this research can provide guidelines for the novice translators who need to gain the initial knowledge to take the preliminary steps. Also, the results of this study may introduce some usable hints on the application of the most appropriate UT in the literary translation for university students majoring in translation and translation courses. Since the most frequently applied UT among the literary translators proved to be the 'sentence', grammar exercises and translation tasks on grammatical structures can be used in translation classes. For fulfilling such a purpose, teachers had better use a grammar-oriented approach in their translation classes, especially in courses such as translation principles and methodology, as well as translation of simple texts in general and literary texts in particular. This is due to the fact that the ST segments can have a deep structure and a surface structure whose identification can help apply the UT that is true equivalence of the ST and best fits the translation of literary texts. Furthermore, based upon the relationship found in this research between the UTs and the free-literal dichotomy in terms of the unit-shifts, the translation trainees can be instructed that application of unit-shifts in the process of going from the ST to the TT helps them to achieve a free translation and that the literary translation needs to undergo deviations from the formal correspondence to meet this requirement. At the end, given the importance of application of the most appropriate UT in the literary translations, a need is felt for fulfilling further researches into the domain of UT and it is hoped that this study paves the way for other studies in this area.

Works Cited
Al-Zoubi, M. Q. and A. R. Al-Hassnawi (2001). Constructing a Model for Shift Analysis in Translation. Translation Journal. Retrieved December 10, 2006 from the World Wide Web: http://accurapid.com/journal/18theory.htm

18

Baker, M. (2001). The Routledge Encyclooedia of Translation Studies. London: Routledge. Bassnett, S. (1980/1991). Translation Studies. London and New York: Routledge Birjandi, P. and P. Mosallanejad (2002). Research Methods and Principles. Tehran, Iran: Shahid Mahdavi Publications. Catford, J. C. (1965): A Linguistic Theory of Translation. London: Oxford University Press. Farhady, H. (1995). Research methods in applied linguistics. Tehran: Payame Noor University. Frank, M. (1972). Modern English: Exercises for non-native speakers, Part II. United States of America: Prentice-Hall. ------- (1993). Modern English: A Practical Reference Guide. United States of America: Prentice-Hall. Hatim, B. and I. Mason (1990). Discourse and the Translator. London and New York: Longman. Hatim, B. and J. Munday (2004). Translation: An advanced resource book. Routledge: New York. Hewson, L. and J. Martin (1991). Redefining Translation: The Variational Approach. London and New York: Routledge. Munday, J. (2001). Introducing translation studies: theories and applications. London & New York: Routledge. Newmark, P. (1988). A Textbook of Translation. New York, London: Prentice Hall. ------ (1991). About Translation. Clevedon: Multilingual Matters. Nida, E. A. (1964). Toward a Science of Translating. Leiden: E. J. Brill.

19

Olohan, M. (2004). Introducing Corpora in Translation Studies. London and New York, Routledge. Richards, J. C., J. Platt and H. Platt (1992). Dictionary of Language Teaching and Applied Linguistics. Great Britain: Longman. Sager, J. (1994). Language Engineering and Translation Consequences of Automation. Amsterdam and Philadelphia: John Benjamins. Shuttleworth, M. and M. Cowie (1997). Dictionary of Translation Studies. Manchester: St. Jerome. Swan, M. (2005). Practical English Usage. 3rd ed. New York: Oxford University Press. Toury, G. (1995). Descriptive Translation Studies and Beyond. Amsterdam and Philadelphia: John Benjamins. Zhu, C. (1996). Climb Up and Look Down: On sentences as the key functional UT (Unit Of Translation) in text translation, Proceedings of the 14th World Congress of the Fdration Internationale des Traducteurs (FIT), February 1996, Vol. 1, Melbourne, AUSIT, pp. 322-343.

20

It can be representative of the fact that the deviations from the ST have been considerably high in these two translations, UTs of a different size from the ST have been applied, and they tend to be freer. This study sets out, amid the points already mentioned, a method for the comparison of ST and TT pairs: identifying the relationships between the coupled pairs of ST and TT segments and establishing equivalence and attempting generalizations about the underlying concept of unit of translation to explore what UT is most frequently adopted by the professional literary translators and to argue the relationship between the UTs and the free-literal dichotomy in terms of the occurrence of unit/rank shifts or changes in the UTs in the move from the ST to the TT. Hence, in line with Newmark's hypothesis, this product-oriented descriptive translation research intends to enrich the theory of describing the phenomenon of English-Persian literary translation from the viewpoint (perspective) of units of translation: to investigate the most frequent unit of translation applied by the professional literary translators, and to inquire into the relationship between the UT and the kind of translation, i.e. free vs. literal, adopted by the literary translators in terms of unit shift.

21

This paper tries to draw a clear picture of the notion of Unit of Translation as a key concept in translation studies. In so doing, it is argued by the researcher that the development of the issue of UT inevitably involves a theory of equivalence, which can be said to be the central issue in translation. It also sheds light on the distinction between literal translation and free translation, which is an important and ever debated point in the literature of translation, in terms of the occurrence of unit/rank shifts or changes in the UTs in the move from the ST to the TT, and in a corpus of literary translations from English into Persian (outstanding novels originally written in English and their successful translations done by professional translators). To this end, the researcher has been mainly concerned with comparing the SL and TL texts (certain literary translations with their original texts) so as to highlight the differences between professional literary translators concerning the two points noted above. The comparison of texts in two different languages for the purpose of specifying units of translation inevitably involves a focus on equivalence which is the central issue in translation. So this study also aims to ascertain whether the translated literary texts are formally equivalent, i.e. directed more towards the form of the ST or formal correspondence, or dynamically equivalent which is described as "the closest natural equivalent to the source-language message" (Nida, 1964: 166). To achieve the above-mentioned goals, the researcher aims to look into three famous English novels considered as masterpieces and to compare and contrast ten pages of each novel with two of its successful translations done by professional translators. Through conclusions which will be drawn at the end of the analysis and implications which will be provided for further studies, it is hoped that new insights can be offered to the literary translators

22

Individual translators, with different foci of attention, may prefer different units as their basic working UTs.

It is widely agreed to be the case that translation and translation studies have never had it so good. Over the last two or three decades, translation has become a more prolific, more visible and more respectable activity than perhaps ever before. And alongside translation itself, a new field of academic study has come into existence, initially called Translatology (but not for long) which is now changed into Translation Studies, and it has gathered remarkable academic momentum. The concept of UT (unit of translation) has been an essential issue not only in translation theory over the last years, but also in modern translation studies In this light, Vinay and Darbelnet (1958/95) maintain that for any science, one of the essential and often the most controversial preliminary steps is defining the units with which to operate (cited in Hatim and Munday, 2004: 137). This is equally true of translation. It is axiomatic that, despite its apparent convenience, the word on its own is unsuitable for consideration as the basis for a unit of translation. According to Newmark (1988: 30-31), normally you translate sentence by sentence (not breath-group by breath-group), running the risk of not paying enough attention to the sentence-join. If the translation of a sentence has no problems, it is based firmly on literal translation, plus virtually automatic and spontaneous transpositions and shifts, changes in word order, etc. He further argues that "since the sentence is the basic unit of thought, presenting an object and what it does, is, or is affected by, so the sentence is, in the first instance, your unit of translation, even though you may later find many SL and TL correspondences within that sentence" (31). 23

. Unit of translation is an issue that has exercised translation theorists over a very long period, indeed. The issue of unit of translation is of a paramount importance in the study of translation in general and the translation of literary works in particular: It might illustrate the kind of translation adopted in the process of translating, i.e. from the dichotomy of literal-free translation which has dominated translation theory for a very long time; it helps to explore the nature of the notion of equivalence, a key concept which is basic to any linguistically oriented translation theory, in that every translation has to stand in some kind of equivalence relation to the original and this equivalence relation, which is anything but clear-cut and predictable, is the outcome of the workings on Units of Translation; it can also be indicative of the translation shifts or changes that take place in the move from ST to TT and the perception of which triggers a sort of adjustment mechanism that ensures the correct interpretation of the message. It is generally defined as the study of the theory and phenomena of translation. It is, according to many researchers in the field, an emerging discipline, yet to gain the status of an independent and distinct discipline in the academia around the world. Translation studies, as an umbrella term, manifests that translation has been practiced for thousands of years, and debates on the nature of translation have been part of translation practice for almost as long. The debates on translation practice go back to the very definition of what translation is. SL emphasis Word-for-word translation Literal translation Faithful translation Semantic translation TL emphasis Adaptation Free translation Idiomatic translation Communicative translation

Figure 2.1. Types of Translation according to Newmark (1988: 45) His model shows that word-for-word translation, for example, is the closest in form to the original structure of the source text; whereas adaptation puts the most emphasis on the fluency of the target text. He further sheds light on each of these translation methods as follows:

24

Stated by Hatim and Munday (2004: 255), it is Peter Newmark's 'semantic translation' that has come closest to what formal equivalence might entail. In semantic translation, the translator attempts, within the bare syntactic and semantic constraints of the TL, to reproduce the precise contextual meaning of the author (1988: 22, cited in ibid). In this light, Munday (2001: 44) maintains that Newmark's description of communicative translation resembles Nida's dynamic equivalence in the effect it is trying to create on the TT reader, while semantic translation has similarities to Nida's formal equivalence. Yet, Newmark keeps himself away from the full principle of equivalent effect, since that effect "is inoperant if the text is out of the TL space and time" (1981: 69, cited in ibid). He further indicates that semantic translation differs from literal translation in that it 'respects context', interprets and even explains; on the other hand, literal translation means word-for-word translation which sticks very closely to ST lexis and syntax (1981: 63, cited in ibid). Nevertheless, he considers literal translation to be the best approach in both semantic and communicative translation: Translation may be defined as the replacement of textual material in one language (SL) by equivalent textual material in another language (TL). The central problem of translation practice is that of finding TL translation equivalents. A central task of translation theory is that of defining the nature and conditions of translation equivalence (Carford, 1965: 21). Chesterman (1989: 99) notes that equivalence is obviously a central concept in translation theory. To Kenny (1997: 77), equivalence is supposed to define translation, and translation, in turn, defines equivalence. As stated by Bassnett (1980: 32-33), Albercht Neubert, who distinguishes between the study of translation as a process and as a product, stresses the need for a theory of equivalence relations as the missing link between both components of a complete theory of translation, while Raymond van den Broeck challenges the extensive use of the term claiming that the precise definition of equivalence in mathematics is a serious obstacle to its use in translation theory! However, according to the concept of this fundamental translation term, which is somehow different from that of the linguistic term of 'correspondence', translating is not merely replacing the words of the ST with their corresponding words in the TL. In this respect, Munday (2001: 49) quotes Bassnett as saying:

25

Translation involves far more than replacement of lexical and grammatical items between languages ... Once the translator moves away from close linguistic equivalence, the problem of determining the exact nature of the level of equivalence aimed for begin to emerge (Bassnett, 1980/91: 25). Catford (1965, cited in Shuttleworth and Cowie, 1997: 51) views equivalence as something essentially quantifiable- and translation as simply a matter of replacing each SL item with the most suitable TL equivalent. This perception has led some scholars to subdivide the notion of equivalence in various ways. Thus, some have distinguished between the equivalence found at the levels of different units of translation, while others have formulated a number of equivalence typologies, such as Nidas (1964) influenced dynamic and formal equivalence, relationship. Pointed out by Catford (1965: 27), and Sarhady (Translation Studies, 7 & 8 (2004 & 2005): 67-68), from a linguistic approach, a distinction must be made between textual equivalence and formal correspondence. A textual equivalence is any TL text or portion of text which is observed on a particular occasion to be the equivalent of a given SL text or portion of text. A formal correspondence, on the other hand, is any TL category which can be said to occupy, as neatly as possible, the same place in the economy of the TL as the given SL category occupies in the SL. Since every language is a unique system, it seems that formal correspondence is approximate. To Catford (ibid), the degree of divergence between textual equivalence and formal correspondence may be used as a measure of typological differences between languages. Eugene Nida (1964) distinguishes two types of equivalence, formal and dynamic, and proposes them as two basic orientations in translating. In formal equivalence, the focus of attention is on the message itself, in both form and content, and one is concerned that the message in the receptor language should match as closely as possible the different elements in the source language. Thus, it is basically source-oriented; that is it is designed to reveal as much as possible of the form and content of the original message. Such a translation in which one is concerned with such correspondences as poetry to poetry, sentence to sentence, and concept to concept is called by Nide a 'gloss translation', which aims to allow the 26 Each of these individual categories of equivalence which will be elaborated through the upcoming paragraphs encapsulates a particular type of ST-TT

reader to understand as much of the SL context as possible (Nida, 1964, cited in Venuti, 1995:136). To have a formal equivalence, one attempts to reproduce several formal elements, including grammatical units, consistency in word usage, and meanings in terms of the source context. Therefore, correspondence in grammatical units, such as translating nouns by nouns, verbs by verbs, etc., keeping all phrases and sentences intact, and preserving lexical concordance are of the characteristics of formal equivalence (ibid). However, in dynamic equivalence, the focus of attention is directed, not so much toward the source message, as toward the receptor response. A dynamic equivalence translation may be described as one concerning which a bilingual and bicultural person can justifiably say, "That is just the way we would say it" One way of defining the dynamic equivalence translation is to describe it as "the closest natural equivalent to the source-language message" (ibid). It is assumed by Hatim and Munday (2004: 44) that: The more form-bound a meaning is, the more formal the equivalence relation will have to be. Alternatively, the more context-bound a meaning is , the more dynamic the equivalence will have to be. FE <<<<<<< FORM- BOUND Meaning >>>>>>>DE CONTEXT- BOUND

Figure 2.6 Formal (FE) vs. dynamic (DE) equivalence (ibid). Accordingly, formal equivalence presupposes matching SL- and TL- content, which is a very reasonable requirement if it is supposed to hold for sufficiently large units, and at the same time matching forms- the most faithful rendering possible of the word order, parts of speech, grammatical constructions and also genre-determined form (meter, etc) of the SL-text (Pederson, 1988: 18). To Hatim and Mason (1990: 7), although formal equivalence is a means of providing some degree of insight into the lexical, grammatical or structural form of a source text; orientation towards dynamic equivalence, on the other hand, is assumed to be the normal strategy. As claimed by Nida (1964: 160, cited in ibid), "the present direction is toward increasing emphasis on dynamic equivalences. Indeed, he defines the goal of dynamic equivalence as seeking the "closest natural equivalent to the source-language message" (Nida, 1964: 166, Nida and Taber, 1969: 12).

27

Viewing the UT as a language level on which translation equivalence is to be established is a misguided conception based on three unwarranted beliefs: (a) UT is a formal unit in nature and can be treated in isolation; (b) language units are automatic UTs; and (c) complete equivalence is achievable (Zhu, 1996: 433). reveals Newmark's apparent vacillation about the range of the UT. Elsewhere he has stressed the importance of the sentence, conspicuously absent in the above definition, as the "natural" or primary UT, while those below the sentence are "subunits of translation" (Newmark, 1988: 31, 65). Importantly, Newmark (1988: 66-7) makes the crucial point that "all lengths of language can, at different moments and also simultaneously, be used as units of translation in the course of the translation activity". While, as mentioned by Hatim and Munday (2004: 25), it may be that the translator most often works at the sentence level, paying specific attention to the problems raised by individual words or groups in that context. However, to Newmark (1988: 69), the longer the unit, the rarer one-to-one translation in which each SL word has a corresponding TL word. As distinctly pointed out by Newmark (1988: 30-31), since sentence is the basic unit of thought, so it is your unit of translation, even though you may later find many SL and TL correspondences within that sentence. Normally you translate sentence by sentence (not breath-group by breath-group), running the risk of not paying enough attention to the sentence-joins. Indeed, primarily you translate by the sentence, and in each sentence, it is the object and what happens to it that you sort out first. Additionally, he (ibid: 65) insists that the sentence is the 'natural' unit of translation, just as it is the natural unit of comprehension and recorded thought. Newmark (1991: 66-68) assumes the main descriptive units to be a hierarchy: text, paragraph, sentence, clause, group, word, and morpheme. Hatim and Mason (1990: 180) maintain that there is no doubt that translators work with words and phrases as their raw material, and equivalence cannot truly be established at these levels. Sometimes the translator is compelled to translate a part of his/her material with a change/shift in the unit of translation such as the time when s/he encounters cases of untranslatability in which a number of items are found in the text for which there is no corresponding equivalence or even the nearest equivalence in the TL and decides to render those parts using a free or communicative method of translation. Texts tend to be identified as translations when shifts from source-language text are perceived. 28

The perception of these shifts triggers in the addressee a sort of adjustment mechanism which ensures the correct interpretation of the message. Actually, Catford was the first scholar to use the term in his A Linguistic Theory of Translation (1965, cited in Hatim and Munday, 2001: 26). In Catford's own words, translation shifts are "departures from formal correspondence in the process of going from the SL to the TL" (1965: 73). According to Munday, Catford (1965: 20) follows the Firthian and Hallidayan linguistic model, which analyzes language as communication, operating functionally in context and on a range of different levels (e.g. phonology, graphology, grammar, lexis) and ranks (sentence, clause, group, word, morpheme, etc.) (Munday, 2001: 60). Baker (2001: 226-227) defines shift as changes that occur in the process of translating. Shifts are deemed as required, inevitable, and indispensable changes which result from attempts to deal with the systemic differences which exist between source and target languages and cultures. Shifts allow the translators to overcome such differences. Also, Venuti (2000: 148) regards shifts as 'deviations' that can occur at such linguistic levels as graphology, phonology, grammar, and lexis. Further, he clarifies that when Anton Popovic asserts that "shifts do not occur because the translator wishes to change a work, but because he strives to reproduce it as faithfully as possible", the kind of "faithfulness" he has in mind is "functional", with the translator locating "suitable equivalents in the milieu of his time and society" (Popovic, 1970: 80, 82, cited in ibid). As far as the product-oriented view of shifts is concerned, Popovic (1970: 79, cited in Baker, 2001: 228) defines shifts from a descriptive point of view: All that appears as new with respect to the original, or fails to appear where it might have been expected, may be interpreted as a shift. Popovic also comments that shifts represent "the relationship between the wording of the original work and that of the translation" (Popovic, 1970: 85, cited in Shuttleworth and Cowie, 1997: 153). As pointed out by Newmark (1988: 85, 285), 'shift' also termed by Vinay and Darbelnet as 'transposition' is a translation procedure involving a change in the grammar from SL to TL.

29

Within the framework of a linguistic theory of translation, Catford (1965: 27) distinguishes between formal correspondence and textual equivalence, which was later to be developed by Koller: A formal correspondent is "any TL category (unit, class, structure, element of structure, etc.) which can be said to occupy, as nearly as possible, the "same" place in the "economy" of the TL as the given SL category occupies in the SL". A textual equivalent is "any TL text or portion of text which is observed on a particular occasion to be the equivalent of a given SL text or portion of text". Munday (2001: 60) maintains that textual equivalence is thus tied to a particular STTT pair and focuses on the relations that exit between elements in a specific ST-TT pair. While formal equivalence has to do with general, non-specific and systembased relationships between a pair of languages or elements in two languages. When the two concepts diverge, a translation shift is deemed to have occurred. To Haim and Munday (2004: 28), a shift occurs if, in a given TT, a translation equivalent other than the formal correspondent occurs for a specific SL element. METHODOLOGY , there is considerable disagreement among translation theorists investigating the notion of units of translation on the level at which equivalence should be established, i.e. what units to choose as Units of Translation. In the field of translation, from a product-oriented approach, a translation unit is a segment of a target text which the translator treats as a single cognitive unit. The translation unit may be a single word, or it may be a phrase, a clause, a sentence, or even a larger unit like a paragraph. This controversial argument about the length of unit of translation is, according to Newmark (1988:45), a concrete reflection of an age-old conflict between free and literal translation. The freer the translation, the longer the UT; the more literal the translation, the shorter the UT, the closer to the word. The present research seeks to study translation units that the professional literary translators adopt in the process of translating famous novels from English into Persian and it is carried out by establishing a relation (of equivalence) between the coupled pairs of ST and TT segments (that is to say, to ascertain whether the translated literary texts are the closest natural equivalent to the original message (Nida's definition of translation, 1969: 19), i.e. dynamically equivalent, or formally 30

equivalent), while taking into account the dichotomy of free-literal approach to translation in terms of the occurrence of unit-shifts in the UTs. So the approach is limited inasmuch as the researcher has looked at Units of Translation only from the angle of the product of translation. As a consequence, this research is placed within the framework of Pure Translation Studies-in Holmes's map of translation studies (Toury, 1995: 10, cited in Munday, 2001: 10-12) -which actually has Descriptive Translation Studies (DTS) as one of its major branches. In fact, DTS embarks upon examination of the product, the function and the process as three focal points among which the first one is highlighted in the :course of this research 1. Product-oriented DTS intends to describe or analyze the existing translations and address a single ST-TT pair or a comparative analysis of several TTs of the same ST.

31