• Front Page
   
  
• July ‘01 Issue
   
  
• April ‘01 Issue
   
  
• January ‘01 Issue
   
  
• October ‘00 Issue
   
  
• July ‘00 Issue
   
  
• April ‘00 Issue
   
  
• January ‘00 Issue
   
  
• October ‘99 Issue
   
  
• July ‘99 Issue
   
  
• April ‘99 Issue
   
  
• January ‘99 Issue
   
  
• October ‘98 Issue
   
  
• July ‘98 Issue
   
  
• April ‘98 Issue
   
  
• January ‘98 Issue
   
  
• October ‘97 Issue
   
  
• July ‘97 Issue
   
  

Constructing a Model for Shift Analysis in Translation

by

1.1. Introduction

n his attempt to transfer meaning from one language (SL) to another (TL) by means of the universally known practice of translation, the translator faces a plethora of linguistic, stylistic and even cultural problems. In this regard, Popovič (1970: 79) confirms that "this transfer is not performed directly and is not without its difficulties." This means that the act of translation can be analyzed along a range of possibilities, which brings about a number of shifts in the linguistic, aesthetic and intellectual values of the source text (ST).

Shifts are all the mandatory actions of the translator (those dictated by the structural discrepancies between the two language systems involved in this process) and the optional ones (those dictated by the his personal and stylistic preferences) to which he resorts consciously for the purpose of natural and communicative rendition of an SL text into another language.

This process of rendition should be carried out in accordance with the norms and principles of translation science in addition to those inherent to the language systems involved in this process.

In accordance with the above statement, the model incorporates various linguistic approaches and methodologies that may have some bearing on the process of translation.

1.2. The Model Constructed

The proposed model of shift analysis is illustrated by Diagram 1 to be followed by a full description of its components:

1.2.1. A Horizontal Description Generally speaking, the model is product-oriented in the sense that it applies to two texts involved in the translation. It is also obvious that the model comprises two levels of analysis, i.e., micro and macro levels with two basic dimensions: the semantic dimension, represented by the message shared by source (ST) and target (TT) texts, which are supposed to convey—roughly speaking—the same message, and the syntactic dimension, where each of the two texts is assigned a different syntactic description, since these texts represent two different embodiments of the same message. This message constitutes the core of the translation task as a whole.

Furthermore, the existence of this message, which is shared between the two texts, provides us with a criterion to formulate the tertium comparationis required compare the two texts (cf. James, 1980).

Another important point to raise here is the fact that this central position of the 'Message' serves a dual purpose. On the one hand, it provides the model with a good means to form a qualitative balance between the ST and the TT. On the other hand, this position represents the maximum balance between form and content of the message, a case which rarely happens. Based on the above discussion, and to account for the interrelationship between form and content as two essential extremes in any translation act, the position of the 'Message' can be slightly changed in four different ways. The first two are motivated by the orientation of translation whereas the other two are motivated by the relationship between form and content in both texts. The adoption of any of these four versions on the part of the translator will result in certain types of shift which can only be explained by referring to the translator's priorities, his style, purpose, the type of the text to be translated, and some other communicative and stylistic norms. The operation of each level of the model is given below.

At this level, the analysis is carried out within the morpho-syntactic component of the model based on Systemic Grammar (SG) and Transformational Grammar (TG). A description of the operation of this level is given below.

This major component is considered a potential area of microstructural shifts, the analysis of which is one of the main objectives of this model. This component comprises two parts. The first represents some sort of a 'categorial grammar,' and builds on Halliday's (1961) Categories of the Theory of Grammar.

The second part borrows much from TG, and it is intended to supplement the first part, which is concerned with the surface structure (SS) analysis. Furthermore, this deep-level part is operated to explore the underlying semantic and syntactic relations existing in each text. The analysis of shifts is then done in the light of these relations which represent the starting point for such an analysis.

Two further remarks need to be made concerning the scope and operation of this component. First, the component is basically, but not necessarily exclusively, sentential in nature, i.e., operates at the level of the sentence and its lower constituents. Second, in addition to the structural orientation of this component, it has a functional one as well. In other words, in analyzing the structure of the ST and TT units, the following questions must guide the analysis of each text: (a) What is that unit?, i.e., its grammatical membership, (b) What does that unit do within the given text?, i.e., the function of that unit, and (c) Where does that unit occur?, i.e., the location of the unit. Thus, it is obvious that we are interested in the grammatical units of the two texts from two angles: the units as independent entities and the units as members of other units in the same text.

As for the operation, we do not consider it a condition for the simultaneous operation of the whole component that fragments of either text be analyzed. To put it differently, the, model makes it possible to use any of its parameters found suitable for performing the task of shift identification. Hence, one can select any sub-level—if found relevant—as a starting point to carry out this analysis.

Finally, one should not overlook the need to operate this component in both texts in the same way, i.e., the sub-level chosen to analyze the ST should be the same as that chosen to analyze the TT.

So far, we have given an overall description of the morpho-syntactic component. A detailed description of its parts is presented below.

1.2.2.1.1. The Categorial Part (Surface-level Analysis)

This part accounts for shifts happening at the SS level in terms of four theoretical categories, namely unit, structure, class, and system, which provide the framework for the syntactic and functional description of the data.

In addition to these categories, the categorial part of the syntactic component makes use of some other language-specific categories called 'descriptive' categories. These are used "to talk about the grammar of any particular language" (Halliday et al, 1964:31) and are considered to be instances of the theoretical ones. Hence, the abstract theoretical category of 'unit' may comprise in a particular language such instances as 'sentence,' 'clause,' 'group,' etc. The description of these categories in one language should be made in terms specific to that language and not in terms of any other universal construct. This is a very significant procedural precaution, which the analyst must keep in mind. However, one might think of the problem of comparison at a later stage while making such descriptions.

One solution to this problem comes from the fact that these categories are only components of some other universal ones which can be used as the constant entity in the comparison. Another equally powerful solution comes from the hypothesis of language universals: despite their apparent differences, human languages exhibit some syntactic and semantic similarities at various levels of abstraction.

The relations among the descriptive categories themselves and their relations with the data are expressed by means of three interrelated scales, namely 'rank,' 'exponence,' and 'delicacy.' From these, the first and the second scales are utilized in the present model. Interlingually, the scale of rank accounts for shifts caused by substituting an SL unit in a particular position on the rank scale by a TL unit in a lower or higher position, e.g., an SL group by a TL clause or word. The scale of exponence, on the other hand, is the scale of realization. It refers to the relation between systems and structures on one hand and to the relation between structures and the formal items of grammar on the other. Hence, terms from systems are realized by structures and the elements of these structures are realized by formal items. In translation, some shifts occur in the realization of SL systems and structures in the TL, e.g., the realization of an English interrogative in Arabic.

The following diagram serves as an illustration of overt rank and covert exponence shifts taking place in the category of unit.

1. Vertically, overt micro shifts may occur either up or down the scale of rank, hence the vertical arrow; upward shifts take place when an SL unit is substituted by a higher-in-rank TL unit whereas downwards shifts take place when an SL unit is substituted by a lower-in-rank TL unit. In English into Arabic translation, for example, upwards shifts represent the marked type as they seem less common due to the fact that the translator usually does not need to go upwards the TL rank scale in search for a substitution for an SL unit. On the other hand, downwards shifts represent the unmarked type; shifts in this direction are more likely than those in the reverse direction. Usually, when looking for the appropriate substitution, the translator moves, optionally or obligatorily, down the TL rank scale until he reaches the lowest-in-rank unit. Only when these lower-in-rank units fall short, will the translator look upwards.

2. Unless dictated by any structural factor, the overt micro unit shifts must be considered optional. In other words, the translator has to decide—in the light of his own translation standards—either to keep the same SL unit rank, or move up or down this rank in the TL. When all factors are the same, translation to the same unit rank provides the maximum degree of structural accuracy; otherwise, the translator has to look for the TL unit nearest in rank to the SL one.
3. Overt micro unit shifts are by no means restricted to the sentence rank; they may occur at any point on the scale, hence, the shadow square is given opposite each move in Diagram 2.

4. Horizontally, Diag. 2 accounts for overt micro shifts motivated by different TL realizations of an SL unit of the same rank. These shifts are more likely to be obligatory rather than optional. This is a typical case in translating between English and Arabic, as the internal systems of these languages are largely different.

Before going into the discussion of the microanalysis of shifts in other categories, it is important to emphasize the functional nature of micro unit analysis of shifts. In addition to the structural character of these units, their grammatical meaning, i.e., their functions should also be taken into consideration when micro unit analysis of shifts is carried out.

1.2.2.1.2. Structure

The descriptive units of the grammar of any language are arranged into meaningful stretches or patterns. One single instance of these patterns is called 'structure.' This abstract category which applies to all units in the grammar of a language (except the one lowest in rank), accounts for the various ways in which one unit may be realized by the unit next below it. Sometimes, however, a unit may be realized by a unit above it. This phenomenon is known as rank shifting.

Languages exhibit a considerable amount of differences both in the realization of similar structures existing in these languages and in the type of structures existing in each language. It is worth mentioning here that the distinction between deep representation of linguistic relations and their surface realizations constitutes an important phase for the analysis of structural shifts.

It is essential to state that there are two ways for describing every single structure. The first relates to the sequence of elements realizing it, i.e., their order. The second relates to the class of these elements. These two methods can be adopted for the sake of a more delicate analysis of microstructural shifts.

Furthermore, we need to distinguish between two types of choices implied in each element of a structure. On the one hand, one is free to choose, for example, between singular nominal group and plural nominal group in English to realize the function 'subject.' On the other hand, such freedom is lacking in verbal groups with past and future tense forms. This distinction can only be made by referring to the categories 'class' and 'system' (cf. below). This constitutes another potential area of interlingual microstructural shifts.

Following Muir (1972:4), we need to distinguish between optional and obligatory elements entering in the realization of a structure. In English, for example, the 'root' is an obligatory element in the structure of any word while affixes are optional ones. When considered interlingually, this distinction leads to another potential area of shifts. Arabic, for instance, exhibits more variations than English with regard to optional and obligatory elements in the structure of the unit 'sentence.' Indeed, interlingual micro structural shifts are likely to occur within each unit that exhibits a structure, i.e., all units except the lowest in rank.

1.2.2.1.3. Class

By a 'class' it is meant the grouping of the constituents of a unit according to the way they operate in the structure of another unit next higher in rank. In other words, a class refers to any set of items having the same possibilities of operation in the structure of a particular unit (Halliday et al, 1964:29).

The need to refer to this category in the analysis of interlingual micro shifts comes from the fact that languages differ in the restrictions they place on the occurrence of some units in the structure of some other higher units. In English, for instance, not all the members of the unit 'phrase' can operate as 'predicate' in clause structure, and those which can do so cannot operate in another place (Muir, 1972:3). The items of each unit are assigned a class name according to their potential capability of operating in the structure of units next above the one they refer to. Hence, the more delicate the class is, the wider are the differences between the languages involved in the comparison and the greater the number of shifts will be.

1.2.2.1.4. System

By a 'system' it is meant the closed number of elements among which a choice must be made, e.g., the system of number in English and Arabic. In fact, the terms available in each system in one language can show fundamental differences from the terms of the same system in another language. This can be considered a major source of obligatory micro shifts at this level of language description. It is worth noting here that the translator is compelled to be bound by the SL writer's choice; otherwise his performance is destined to be erroneous. In cases where compatible terms with the source system are taking place in the target system, the translator has to bridge the gap by using some other means, e.g., the use of a lexical marker of number 'two' to express duality when translating from Arabic into English.

The occurrence of shifts here can be accounted for by means of terms existing in the system of individual languages. What increases the possibility of such occurrence is the fact that all the descriptive units required for the description of a language are systemic in nature; they are realized by means of specific choices of the particular systems of that language. Actually, these choices are language-specific and their applicability is governed by three criteria. The first is specified "in terms of rank of unit to which the system is applicable; the second is specified in terms of the part the unit is playing in the structure of a higher unit"; the third is specified "in terms of the other options which must be chosen before the options of the given system become available." (Berry, 1977:13)

Now, we should emphasize the mutual integration of the individual categories. In this regard, two types of structural surface relations, namely paradigmatic and syntagmatic relations, can be stated.

Interlingually, different dependency relations are realized by different syntactic means. One important point to talk about is word order, as in the case of the unilateral dependency relation between the head and the adjective in English nominal groups. The order of these elements is Adjective + Head while the same relation is realized by the reverse order in Arabic, i.e., Head + Adjective. The same is true for unilateral dependency relation of possession in both languages, e.g.,

English		Arabic
Robert's possessor	book possessed	/Robert/ possessor	/kitaabu/ possessed

In these structures, where one element is typically obligatory while other elements are optional, an agreement between the head and its modifiers is usually observed in some languages, e.g., number and gender agreement between 'noun' and 'adjective' in the Arabic nominal group. However, languages vary so widely in the restrictions they assign to this agreement. In the English nominal group, for instance, this agreement is observed between 'articles' and 'nouns' but overlooked between 'nouns' and 'adjectives.' By contrast, Arabic seeks such agreement in both cases. Again, this is another potential area of obligatory structural shifts in translation.

As for bilateral dependent structures, the distribution of either constituent elements is different from that of the structure as a whole as in the prepositional groups in English and Arabic. Following Brown and Miller (1980:255), the majority of the syntactic relations in all languages are of this type. Furthermore, the variety of functional labels used to refer to these relations reflects the variation of these relations in different languages.

The importance of bilateral dependency relations to the analysis of structural shifts in translation can be appreciated by examining the following sentences:

-John beats the dog.

-Fido is a dog.

-John went home.

Although one single string of elements could be assigned to the above sentences, namely NP + V + NP, the relation of V with the NP following it in each sentence is different. In order to account for this difference, various functional labels are used, e.g., 'predicate,' 'complement,' 'object,' etc.

Many bi-directional relations presume that one constituent element requires other constituents to be in a particular case, e.g., in a prepositional group with a personal pronoun as a realization of the NP, the preposition requires an oblique case in English, and an accusative one in Arabic. This indicates that languages use different ways for implementing these relations, which adds to the likelihood of structural shifts occurrences in this area. The degree of this likelihood increases with the fact that none of the constituent elements in these structures is optional.

In the third type of dependency relations, i.e., co-ordinate dependency, neither constituent depends syntactically on the other and the distribution of each constituent is the same as that of the structure as a whole. In terms of symbols, the description of these structures is:

A A' (+) A' (+)...Aⁿ, where A is any co-ordinate dependent structure and A', A', Aⁿ are constituent elements of the same distribution. Theoretically, no limit is assigned to the number of these elements in any given structure. Yet languages may exhibit certain restrictions on the order of these elements. The order of adjectives in the English nominal group is a good example of these restrictions which are mostly language-specific. The optional (+) in the above formula of these structures is meant to emphasize the possibility of having a co-ordination marker such as 'and,' 'or,' etc. for some structures. Up to this point, the need for such markers varies interlingually. Hence, another area of shifts can be manifested.

The last type of dependency relations is that of the exclusion relation which is useful for defining some grammatical classes such as the verbs of state in English which do not agree with auxiliaries for the progressive aspect, and proper nouns which do not take the definite article 'the.'

However, it must be mentioned that "dependency relations cannot always be captured in a straightforward fashion in constituent grammar" (Brown and Miller, 1980:259). What we also need to know is an interpretive machinery to interpret these relations. In the present case, such a machinery can be borrowed from TG which enters as a supplementary part in the syntactic component of this model of analysis (see Diagram 1). The incorporation of this type of grammar is accounted for in the following section.

1.2.2.1.5. Deep-level Analysis

The transformational sub-component is considered to bridge the gap in the categorial sub-component.

The need for this injunction refers to the fact that in any translation task, the translator needs to employ more or less four transformational syntactic processes, namely, deletion, insertion, permutation, and/or substitution. Each of these processes is binary in nature, i.e., optional (its adoption depends on the translator's own preference) or obligatory (the translator is compelled to apply it in order to produce well-formed TL sentences). It goes without saying that languages exhibit substantial differences in the application of TRs and allow different means for the application of the optional ones. It is these qualitative and quantitative differences which allow us to amplify obligatory and optional syntactic shifts in translation. In other words, we would call the shifts motivated by the application of obligatory TRs as obligatory syntactic shifts and those motivated by the application of the optional ones as optional syntactic shifts.

Following Nida (1964:65), two practical advantages can be derived from the adoption of this procedure. First, the equivalence of different formal (syntactic) structures possessing the same meaningful relation can be seen even interlinguistically. Second, the equivalence of formally similar structures possessing different meanings can also be readily seen. And in the same way we would like to add a third one, namely, complex structures can be easily plotted by reference to their kernel, i.e., deep, structures.

So far, the description of the micro level of the present model is complete. The rest of the present paper is dedicated to describe in detail the second major level of this model, i.e., the macro level of analysis.

1.3. Macro-Level Analysis

At this level, the model is switched to analyze a considerable amount of obligatory and optional shifts which take place at a level higher than the micro level. In order to account for this requirement, the present model presupposes an independent broad level of analysis called the macro-level of analysis.

The main difference between this level and the previous one, i.e., the micro level, comes from the direction of analysis. On the one hand, the micro level moves within the domain of the sentence as the maximum unit of the syntactic description. The macro level, on the other hand, moves within the domain of the text. In this sense, the macro level accounts for all variables of texture, culture, style and rhetoric, which contribute to the occurrence of shifts at levels other than the syntactic level. Hence, it is broken down into a number of components, each of which accounts for a particular variable of the above ones. Diagrams 3 and 4 are presented to illustrate this difference between the micro and macro levels of analysis.

However, this difference should not be exaggerated. Instead, the two levels should be considered unitary since both would accept the traditional view that the sentence is the locus of structural and stylistic variation, "though with the proviso that it entails spans wider than sentence" (Hendricks 1976:40-41).

What follows is a description of the individual components within the macro level of analysis along with their scope of inclusion in relation to the possibility of shifts within each component.

1.3.1. The Semantic Component

Meaning should be the main preoccupation of all translation. However, the amount of this interest varies according to the type of meaning conveyed by the lexical items of a given text. As far as translation is concerned, the translator has to do his best to transfer as much of the original meaning as he can into the TL. But since we know that the process of meaning transfer is not a straightforward process, the translator, therefore, is often called upon to make some semantic adjustments in order to accomplish this task. In our case, such semantic adjustments are analyzed as semantic shifts, which can be obligatory or optional. The former are dictated by the unavoidable semantic gaps between the SL and TL. Such gaps are mainly caused by some cultural and conceptual differences between the two languages. The latter in turn arise when the translator attempts to maintain the gist of the original meaning while practicing some means of semantic polishing.

The analysis of both types of shifts has to be carried out by extracting the semantic relations within the lexical items of the ST then examining the possibility of conveying similar relations into the TL by similar or different formal devices. It should be mentioned here that meaning extraction should be made in the light of the immediate situation in which the ST functions; otherwise, the analysis is destined to be vague. This relation is discussed below.

1.7.1.1. The Relation between Meaning and Situation

Language is performed in order to serve a variety of functions over its 'ideational' function (cf. Halliday, 1976). In performing all these functions, language is determined situationally, i.e., the selection of linguistic elements to convey a particular meaning is determined by the elements of the situation in which these elements are used. This same relation is described as one of inclusion; the former includes or presupposes the latter.

As far as the present model is concerned, the relation of inclusion between meaning and situation brings about significant implications, the first of which is the necessity of taking the situational variables into account in defining the meaning of the ST. The second implication goes as follows. In addition to the impact of situation on the realization of meaning, part of this meaning is mapped by the linguistic organization of the language in which this meaning is encoded. Hence, one can safely generalize that if the context of situation is changed "changes will inevitably take place in the linguistic texture. Conversely, if a shift is carried out on the linguistic level, this context of situation will also change." (Wilss, 1982:71).

The above generalization necessitates the adoption of a broader view of the concept of meaning. Such an extended view of meaning applies to all text types in general and those having figurative semantic relations in particular, i.e., literary texts. In this regard, Nida (1985:119) states:

Of course, our model does not go as far as Nida does. His words are cited to emphasize that meaning in its widest sense would serve the purpose of the present paper. To put it differently, the analysis of semantic shifts will be carried out in terms of the situation in which language is used.

Hence, only the paradigmatic relations could be considered within the semantic component of the macro level of analysis. This is so because such relations are semantics proper. The other types of relations, due to their textual and stylistic values, will be accounted for within independent components at the same level.

1.3.1.1. Paradigmatic Relations: Synonymy and Semantic Fields

The significance of synonymy as a paradigmatic semantic relation to translation is stated by Baldinger (1980:251) as follows: "Translation is nothing than a problem of synonymy." It is evident, then, that Baldinger perceives synonymy in its widest sense to mean, in translation, the search for equivalent meaning on all linguistic levels. However, translation, strictly speaking, cannot be perceived as a simple task of haphazard matching of SL lexical items with their TL counterparts. Any individual can do this by relying on a bilingual dictionary. By contrast, the translator needs to analyze the meaning of the SL lexical items before attempting to find TL equivalents for these items. In his search for efficient lexical equivalents in the TL, the translator has to play the role of a competent proxy on behalf of his readers; he must identify the areas of cultural overlap and linguistic interference between the two languages. His suffering starts at this stage: identical symbols in the two languages do not necessarily convey the same meaning. Much worse is the difference in people's experiences and the variation of conceptual boundaries from one language to another "in a way that defies principled explanation" (Leech, 1974:3). Knowing that lexical items are the vehicles by which people's experience is encoded and their concepts are expressed leads to the conclusion that shifts in interlingual synonymy are inevitable phenomena in translation.

Apart from the problem of denotation in the study of synonymy, Nida (1964:89) captures the structural specification of words as another source for semantic shifts in this area. In this regard, he states:

In our opinion, this process of finding semantically equivalent lexical items is carried out by performing a variety of shifts in the central and/or peripheral components of the ST lexical items. By means of careful contextual conditioning, the translator may remove or insert some componential values associated with the ST lexical items. According to Nida (1969:107), "in many instances, shifts of components involve only a shift from a literal etymological meaning to one which is functionally more relevant." Nida's example for this case is the translation of the word 'devil' whose etymological meaning is 'Satan.' If translated, say, into Arabic, the word would mean nothing unless an etymological shift is used, i.e., the translator has to refer to its etymological origin then transfer it into the TL. Another type of componential shift goes from generic to specific meaning or vice versa.

As for the relation between the lexical items and their referents, which is the core of their referential meaning, the translator is likely to face three situations. The first one is "the existence of a term (and its corresponding referent) in the receptor language, but with an equivalent function being performed by another referent" (ibid. 44). A good example for such a situation arises when translating from English into a language which has no word for 'snow.' The translator has to replace the word 'snow' in the phrase 'as white as snow' by another word, which refers to a white-colored object.

The second situation is "the existence of the referent in the receptor language, but with a different function from what it has in the source language."(ibid.). The English word 'owl' and its Arabic equivalent 'buum' represent a good example for such a problem. In English, it refers to a class of birds with positive connotation, i.e., wisdom and good omen. Arabic has exactly the opposite connotation for the same referent, the fact which necessitates finding another word referring to an object with similar connotations.

The third situation is "the non-existence of the referent in the receptor language and no other referent with a parallel function" (ibid.45). The translation of any lexical item denoting technological inventions from English into Arabic provides a good example for such a problem. Here, the translator is compelled to force foreign words into the TT or to use descriptive phrases to explain the meaning of individual lexical items.

In all the situations discussed above, the translator finds himself obliged to adopt some strategies so as to bridge these semantic gaps. In this regard, Jacobson's (1959:234) words would serve to conclude the discussion:

1.4. The Textual Component

To be described as such, a text should exhibit two kinds of structural and cohesive relations: local and global (Hendricks, 1976:41). The first can be accounted for by 'Sentence Grammar' on which enough has already been said. The second kind includes the relations which cannot be accounted for "without reference to inter-sentence features and to portions of the text beyond the sentence under consideration" (Enkvist, 1973:111). Consequently, the inclusion of such a component would inevitably presume reliance on text linguistics. As a matter of fact, the incorporation of this linguistic approach is an essential procedural condition for the macro level of analysis. In other words, the analysis cannot be carried out on randomly chosen sentences without taking into account that these sentences should exhibit the property of global cohesion in addition to their local cohesion. This particular requirement is satisfied by inserting a textual component within the macro level. The analysis within this component will focus on the elements within individual sentences which, in addition to playing a role in the structure of the sentence itself, also contribute to its integration into the textual whole, by making it dependent in some way on other sentences within the same text. These dependent sentences convey information about one another, which makes them constitute a cohesive whole. There are two other sources of this interdependence, namely, textual and discoursal. The former refers to the variety of global and local cohesive markers within the portions of the text while the latter refers to the functional dependency among these portions. The latter dimension imposes the inclusion of another macro discipline of language description, namely, 'discourse analysis.' This means that the question to be asked about any linguistic unit is what the user hopes to achieve with this particular bit of language, i.e., its use,in addition to its form. This issue is accounted for by the pragmatic component within the macro level. The description of this component will be given after the description of the present component.

By now, it is obvious that the textual component of shift analysis views the data in their broad scope. Hence, this analysis will be carried out in terms of the textual well-formedness which entails such variables as collocation, reiteration, ellipsis, references, substitution and the like.

1.5.The Pragmatic Component

Stalinker (1973:38) defines pragmatics as "the study of purposes for which sentences are used, of the real-world conditions under which a sentence may be appropriately used and alternate." In this sense the meaning of a single expression may vary in accordance with the purposes behind it and the conditions surrounding the communicative act.

This view of meaning, which necessitates the inclusion of a pragmatic component in any proper semantic analysis, corresponds to what Widdowson (1973:69) refers to as "the communicative use of sentences in the performing of social actions." Leech (1974:141) uses the term 'connotative meaning' to refer to the same type of meaning. To him, the connotation of an expression is the "communicative value an expression has...over and above its purely conceptual meaning."

In contrast to the linguistic meaning, which can be extracted from the grammatical relations within a given text, pragmatic meaning can only be analyzed by referring to the cultural and/or linguistic context of that text. Accordingly, the analysis of pragmatic shifts in translation can only be carried out by attending to the immediate cultural context of situation of the ST and matching it with that in the TL so as to put a finger on the possible areas of shifts when the translator tries to convey the same message into the TL. In this connection, reference is to be made to speech acts theory as initiated by some pragmatists, e.g., Austin (1965 ). This means that the analysis should account for variables such as the intentions of the writer or speaker, his expectation, the time of utterance, the truth value of the propositions expressed, other speech acts being performed in the same situation, and so on. In other words, the analysis will take into consideration the major functions of language as a means of communication in a social setting. What is important here is that the realization of these functions varies greatly from one language to another, which adds to the necessity of using a pragmatic component in the present model. In fact, this variation in the realization of the pragmatic functions of language goes is expressed in two phases. First, languages employ different formal devices for realizing similar speech acts. These formal variations include all lexical and syntactic means allowed by the grammar of each language. Second, the contextual spectrum, which imposes the performing of particular acts, differs considerably from one culture to another. Furthermore, at a higher level of delicacy, "cultures may also differ in the rules for when certain speech acts can be appropriately performed." (Benthalia and Davies, 1989:102). In this regard, one may refer to the considerable differences between English and Arabic in the kind of formulas commonly used to perform the acts of greeting, leave-taking, thanking, apologizing and so on. The content of these formulas and the rules of their use frequently reflect the particular values and beliefs of their....users. To take only one example of these differences, Arab speakers frequently use formulas containing religious references for greeting and thanking, e.g., / baraka Allahu fiik/ (lit. 'blessing of God upon you'); / Allah ykhaliik/ (lit. 'may God preserve you'), etc., while functionally corresponding English formulas do not contain such references. Similar differences between the two languages can be recognized in the formal realization of 'imperatives.' There are many situations in which an Arab speaker uses an imperative construction while intending to convey a polite request or invitation. In such a case he may use some forms of invocation or good wish for the addressee.

The above discussion entails that the categorization of speech acts into, greeting, thinking, request, etc., is a universal phenomenon, the linguistic realization of these acts and the rules of their performance in one language "do not necessarily have exact equivalents in another language, and raise a lot of questions related to the theory of translation." (Enkvist, 1973:57). The implication of this statement on the phenomenon of shifts is self-evident: one needs to account for all these differences in order to point out the possible shifts within this particular area. Another requirement for this analysis is to identify in each language which formal devices are used for particular speech acts. Both formal realizations then will be compared so as to point the obligatory and optional shifts between them.

The inclusion of the textual component in the present model entails the inclusion of another related, through distinct, component, namely the rhetorical on which a brief account is given below.

1.6. The Rhetorical Component

The native speakers of any language are capable sometimes of maintaining the logical relationship that exists between a group of linguistic units even with the absence of explicit maker for this relation. An English native speaker, for example, can give the same interpretation for the following sentences, where the first contains an overt linkage marker, which is lacking in the second:

- Medicines can kill and therefore should be kept out of children's reach.

- Medicines can kill; they should be kept out of children's reach.

Kaplan (1972:ix) attributes this ability to the fact that this is how speakers of English organize their thought by means of culture-specific devices known as 'rhetorical devices.' The cultural restrictedness of these devices implies the inevitable occurrence of shifts in this particular area for which this particular component is dedicated. In his attempt to characterize the rhetorical structure of a number of languages, Kaplan (ibid. 61) views English as 'direct' whereas much oriental writings are 'indirect' or 'circulocutionary.' According to him, the speakers of Semitic languages tend to transfer a complex series of parallel constructions to English. Therefore, this gives evidence for the likelihood of shifts in this particular area.

Languages may also exhibit many differences in other phases of rhetoric such as foregrounding, irony, allegory, metaphor, simile, metonym, etc., The reason behind these differences is self-evident: these phases are associated with people's conceptual experiences and ideologies. In this regard, the variations between English and Arabic in the area of simile and metaphor represent a good example. In consequence, this particular component has been included so as to account for the possible shifts that may arise with some major rhetorical devices, e.g., metaphor, idiomatic expressions, foregrounding and metonym.

1.7. The Stylistic Component

Style is the last area to be dealt with at the macro level of analysis. Hence, we shall consider certain overall features of style which contribute to the occurrence of shifts of various levels of the TL text. Furthermore, we do not intend to restrict the term 'stylistic' to its literary conception. Instead, following Fowler (1966:15), I hold the view that style is "a property of all texts," without, however, going as far as to overlook the fact that literary texts exhibit some stylistic features more clearly than non-literary ones. With this clear precaution in mind, we assume that every language has its own stylistic conventions which may differ from those of other languages, which may cause stylistic shifts to arise when two languages are involved in terms of the function of these conventions and their formal carriers, i.e., their linguistic realizations. When two or more TL expressions are available at the translator's disposal to express the same SL meaning, stylistic shifts become possible. Obviously, the role of content here is to serve as the starting point for shift analysis. The other issue relevant to the analysis of stylistic shifts is the problem of style definition. The need for a satisfactory definition of the term 'style' would help us get rid of the complexities of literary criticism. In other words, we need to restrict this term so as to account for the measurement of stylistic shifts in the TT regardless of their type. This means that this term should be defined in purely linguistic terms rather than defining it as a literary concept.

Here, it is necessary to emphasize the overwhelmingly optional nature of stylistic shifts. In other words, I perceive these shifts as TL structural alternative means of expressing a single SL message at various levels of language use.

Interlingually, stylistic shifts can be explained with reference to the same distinction between obligatory and optional application of language rules. An obligatory rule in one language could be optional in another. Accordingly, the analyst's task is to analyze the original writer's typical strategies in utilizing optional transformations and his use of different kinds of transformational operations to compare them with those of the translator.

The second important point presumed by Galperin's statement refers to the variety of implications conjoined with the term 'style.' The point can put more concretely as follows: the definition of style "implies that words [and other linguistic units] on a page might been different, or differently arranged, without a corresponding difference in substance" (Ohman, 1964:430). One significant implication of this statement is that a distinction should be made between the form of the message and its content. The following section is a discussion of this issue.

1.7.1. Dichotomy of Form and Content:

Nida and Taber's (1969:105-6) statement in favor of this dichotomy seems the best start ing point for this section. It reads as follows:

The implication of the above statement is evident: stylistic shifts are expected with the translator's effort to preserve the balance between form and content of the message on the one hand and his tendency to reflect his character on the other.

Although some scholars tend to restrict the criterion of form to literary texts, our position here is that "there is probably no absolute formal distinction between literature and non-literature: neither of these two categories is formally homogenous." (Fowler, 1966:16). However, this generalization should not be misunderstood as to deny the existence of literature. Instead, it is meant as being a working hypothesis necessary for the task of analyzing stylistic shifts within a linguistic framework. To put it more clearly, we assume that all examples of language use exhibit a linguistic form susceptible for empirical investigation (ibid.). Furthermore, it makes no difference if the designation 'literature' is used for a certain class of constructions, since members of this class exhibit formal differences among themselves as well as compared to other members outside this class. In short, "there is no constant, or a set of constants, which differentiates all members of the class 'literature' from the members of the class ''non-literature.'" (ibid.11). Even when we agree on the importance of form to literature, this does not trivialize the fact that linguistic forms exist and should be taken as an essential area of investigation in all other examples of language use. The inseparability of form and content goes with the view that form has a function and the translator has to discover and transfer it to the TL (cf. Crystal and Davy, 1969; Leech and Short, 1981; Hatim and Mason, 1990). In this sense, the translator's task is not only to transfer the content of the message but also to transfer its form as far as possible. However, following Nida (1985:24) "languages clearly do not differ primarily in what they can communicate, but in how they do it." This is an overt reference to the occurrence of stylistic shifts in translation at two levels. On the one hand, there is the intrasentential level where languages differ in their optional and obligatory rules of sentence formation. On the other hand, stylistic shifts are also likely to occur at the inter-sentential macro level where language may exhibit substantial differences in the rules of text formation and message organization. Consequently, the analysis of these shifts will be carried out on both levels in parallel with the axis of obligatory and optional shifts.

Now the conclusion to be drawn is that 'stylistic shifts' is a cover term used to refer to the variety of macro formal modifications of the ST when transferred into the TL. The occurrence of these shifts, moreover, can only be predicted by referring to the rhetorical and stylistic conventions of each language in question in addition to the translator's preference, choice, and ability.

Before moving to the framework of analyzing stylistic shifts, it should be mentioned here that the contribution of form to the meaning of a text varies according to the text type. The amount of stylistic shifts varies accordingly. In some genres, e.g., prose, poetry, religious texts, etc., form has a cohesive and an aesthetic function which conveys "the creative will of the writer and lend the text an outward shape" (Wilss, 1982:76).

1.7.2. The Role of the Translator

Among all factors affecting the occurrence of stylistic shifts, the role of the translator stands as the most recognizable factor. The majority of optional shifts taking place in translation can be attributed to the differences between the original writer and the translator as two text-producers. However, the impacts of these differences are usually suppressed by the literary norms of the TL and the norms of the translation activity itself. More important is the translator's relation to the text given. This relation is neatly described by Popovič (1970:80) as follows:

Popovič's statement reminds us of many factors, which affect the translator's adoption of a particular style in rendering a particular text into another language. One of these factors is the literary norms that may differ in the SL and TL, the case which leaves the translator with three choices: to imitate the original style, to rely on the TL stylistic norms, or to compromise the two by practicing his own stylistic prejudice. The last two options would naturally result in a great deal of stylistic shifts.

The other point is that some languages may have much more highly developed aesthetic and rhetorical patterns than other languages, which gives the translator more freedom to choose the way he likes in expressing the original message. Moreover, the range and refinement of some literary genres could be more developed in one language than in another. Both cases are typically applicable to the translation of elevated literature such as poems, epics, religious texts, etc.

The third factor relevant to the role of the translator in stylistic shifts relates to the 'national features' of the ST. In this regard Zora Jesenka (quoted by Popovič, 1970:81) has the following to say:

Thus, it is the aim of making such literary works accessible to the TL reader that encourages the translator to use stylistic shifts. Following Popovič (ibid.), such shifts are expected as a rule "because the identity and difference in relation to the original cannot be solved without some residue." Up to this point, the translator's dilemma becomes evident: he would never strive to preserve all the singularities of the original but rather he would try to reflect his own identity while preserving the gist of the original message. Furthermore, he will try to make use of contemporary equivalents and comprehensible by his perceptive reader. Doing all these tasks, the translator will display much of his translation skill and literary taste. Skill and literary taste are two prerequisites to produce a 'natural' translation because the act of substituting the SL norms by TL ones is a highly subjective issue that demands creative intuition on the part of the translator. Again, this is so because direct transfer of specific stylistic features from the SL into the TL is hindered by both the organic character of the ST components and the divergence between the two stylistic norms of both languages, on the other. This transfer becomes possible "only by means of an equivalent function, namely by appropriate shifts." (Popovič, ibid.83).

To sum up, our perception of the role of the translator is that he is a performer of a dual task. On the one hand, he has to adhere as much as he can to the content of the message, including its form (if it is part of this content); on the other hand, he tries to reflect his identity and tends to produce a 'natural' text. This tendency, we believe, can best be achieved by means of a set of stylistic shifts.

Conclusions

The first noteworthy conclusion of this paper is that the phenomenon of 'shift' should be redefined positively as the consequence of the translator's effort to establish translation equivalence (TE) between two different language-systems: that of the SL and that of the TL. Psychologically, the occurrence of these shifts reflects the translator's awareness of the linguistic and non-linguistic discrepancies between the SL and TL. In this sense, shifts can be defined as problem-solving strategies adopted consciously to minimize the inevitable loss of meaning when rendering a text from one language into another.

Second, since translation proper is concerned with the transfer of meaning, the analysis of shifts in translation should take into account the non-linguistic factors—in addition to the linguistic ones—so as to achieve a comprehensive analysis of these shifts.

Third, shifts in translation constitute a counterclaim to language universals ; therefore, these shifts can be better examined within the domain of 'difference' in translation. This conclusion is based on the assumption that languages do not differ primarily in what they communicate but in how they do so. Consequently, CA has been proved as a powerful diagnostic tool for shift analysis.

Fourth, the distinction between various types of shifts at various levels necessitates the distinction between various types of equivalence in translation, e.g., functional, pragmatic, textual, collocational, rhetorical, etc.

Fifth, the distinction between micro-level and macro-level shifts is compatible with the distinction between various types of translation, e.g., literal, free, etc. The same distinction is also compatible with the distinction between various units of translation, e.g., word, sentence, paragraph, etc.

Sixth, the postulation of the terms, 'optional' and 'obligatory' shifts satisfies the need to account for linguistic and non-linguistic differences between the languages involved in this process.

Finally, we assume for our model, as it is described above, a universal operation regardless of the languages involved in this task.

References

Austin, J. L. (1965) How to Do Things with Words. New York: Macmilan.

Baldinger, K. (1980) Semantic Theory. New York: St.Martin's Press.

Benthahila, A. and Davies, E. (1989) 'Culture and Language Use: A Problem for Foreign Language Teaching.' IRAL, Vol xxvii/2, 1989, p.102.

Berry, H. M. (1977) Introduction to Systemic Linguistics 2. and Links. Batsford and Sons.

Brown, E. K., and Miller, J. E. (1980) Syntax: A Levels Linguistic Introduction to Sentence Structure. London: Hutchinson & Co. (publishers) Ltd.

Crystal, D. and Davy, D. (1969) Investigating English Style. Cambridge: Cambridge University Press.

Enkvist, N. E. (1973) Linguistic Stylistics. Mouton, The Hague.

Fowler, R (1966) Linguistic Theory and the study of literature.' In N.G.Leech (ed.) Linguistics and the Figures of Rhetoric. pp. 135-56.

Halliday, M. A. K. (1961) 'Categories of the Theory of Grammar,' Word 1961 17.3.241-92. (Also in G. R. Kress (1976) (ed.) 52-72.

Halliday, M. A. K., A. McIntosh and P. Strevens (1964). The Linguistic Sciences and Language Teaching. London: Longman.

Halliday, M. A. K., and Hasan, R. (1976) Cohesion in English. London, Longman.

Hatim, B., and Mason, I (1990). Discourse and the Translator. London: Longman.

Hendricks, W. (1976) Grammars of Style and Styles of Grammar. Amsterdam: North Holland Publishing Company.

Jacobson, R. (1959) 'On Linguistic Aspects of Translation.' In R. A. Brower (ed.) On Translation Cambridge, Mass: Harvard University Press, 1959, pp. 232-9.

James, C. (1980) Contrastive Analysis, London: Oxford University Press.

Kaplan, F. (1972) -, London: Oxford University Press.

Leech, J. (1974) Semantics. London: Hazell Watson & Vinery Ltd.

Leech, G. N., and Short, M. H., (1981) Style in Fiction: A Linguistic Introduction to English Fictional Prose. New York: Longman Group Ltd.

Muir, J. (1972) A Modern Approach to English Grammar. Batsford and Sons.

Nida, E, (1964). Toward a Science of Translation. Leiden, E. J. Brill.

_______, (1969). 'Science of Translation,' Language, Vol.45, No.3, 1969, pp. 495-497.

Nida, E., and Taber, C., (1969). The Theory and Practice of Translation. Leiden: E. J. Brill.

Nida, E., (1985) ' Translating means Translating Meaning,' Publication of FIT, UNISCO, PP. 119-25.

Ohman, Richard (1964) 'Modes of order,' In Donald C. Freeman (ed.) Linguistics and Literary Style. Holt, Rinehart and Winston, Inc. 1970, pp. 209-42.

Popovič, A (1970) "The concept 'shift of expression' in Translation," in Holmes, J. (ed.) The Nature of Translation. Mouton: The Hague.

Stalinker, R. C. (1973) 'Pragmatics,' In D. Davidson and G. Harman (eds.) Semantics of Natural Languages.Dordrecht: Foris Publications.

Widdowson, H. G. (1973) An Applied Linguistic Approach to Discourse Analysis. Unpublished Ph. D thesis, University of Edinburgh.

Wills, W. (1982) The Science of Translation: Problems and Methods. Tubingen: Gunter Narr Verlag.