A systemic functional analysis of conjunction in Au Revoir les Enfants

This paper involves an analysis of conjunctive cohesive markers in the subtitled version of the film Au Revoir les Enfants (Malle 1989) from a systemic functional perspective. The underlying hypothesis is that in the translation of this film from French into English some conjunctive cohesive markers are omitted because of time and space constraints (Gottlieb 1992) as well as the principle of relevance (Bogucki 2004) imposed on the subtitles. Thus, four types of coordinators (i.e. et, mais, car and donc) and five types of subordinators (i.e. que, si, quand, comme and parce que/ puisque) are analysed at the clause and clause-complex level, respectively. The omission of these conjunctions is accompanied by shifts and transformations at the lexical, phonological and syntactic levels. On the basis of these shifts and transformations, twelve strategies are developed to serve as models or teaching mechanisms in subtitler training.


Introduction
Conjunction is one type of cohesive marker that establishes semantic relationships in a spoken or written text. In this study, conjunction is analysed in Au Revoir les Enfants (Malle 1989), a film which won the Venice Film Festival's Golden Lion Award in 1987. The film was translated into English by Anselm Hollo in 1988 and subtitled in 1989. My interest in this film lies in that it depicts the scenes of contemporary social life based on Malle's experience. The dominant scenes are, for example, the courage of the Catholic monks and teachers to defy the authority of the Gestapo in German-occupied France, a mother's love for her children, the devotion of teachers to educate young French boys in the winter, the exaggerated discipline of the German soldiers, and both the arrest and death of Father Jean as well as the refugee children in concentration camps in Auschwitz and Mauthausen. More importantly, the film clearly demonstrates how subtitlers manipulate both syntax and lexis when some conjunctions are omitted in the translation of the film dialogue from French into English. The analysis of conjunction in the film requires both a systematic and contrastive investigation to ascertain those conjunctive cohesive markers which have been omitted. This systematic and contrastive investigation uses systemic functional grammar (SFG) as a tool which, to my knowledge, has not been carried out in previous research. It is based on a corpus of 94 French clauses and their English translation equivalents (see Appendix) in the film and in the book (Malle and Hollo 1988). These items will serve as examples.
In Linguistics, conjunction has been discussed in some detail by scholars such as De Beaugrande and Dressler (1981) and Halliday and Hasan (1976). This paper focuses on the systemic model of conjunction developed by Halliday and Hasan (1976) and formalised by Halliday (1985Halliday ( /2004) and his followers (e.g. Bloor and Bloor 1995, Eggins 1994, Thompson 2004. According to this model, conjunction may be defined as a cohesive resource, the devices of which signal the relationships that exist between clauses or clause complexes in a text. As suggested by Halliday and Hasan (1976:233-235), this definition excludes coordinate relations that are established between syntactic elements (such as nouns, pronouns, adjectives, and adverbs) within the clause.
In Translation Studies (TS), conjunction has been researched by, for example, Baker (1992) and Hatim and Mason (1990). Following Halliday and Hasan (1976), Baker (1992:190-191) considers conjunction as a set of formal markers that are used to relate clauses, sentences and paragraphs to each other in a text. These markers signal the way in which the writer wants the reader to relate the subsequent chunk of information to the previous one. They include five main categories: additive (e.g. and, or, furthermore), adversative (e.g. but, yet, however), causal (e.g. so, consequently, because), temporal (e.g. then, next, finally), and continuative (e.g. now, of course, well). The continuative items are discourse markers which, according to Halliday and Hasan (1976:267), have a cohesive force in the text but do not express any particular one of the other four conjunctive relations.
However, for translation purposes, Baker (1992:191-192) departs from Halliday and Hasan's theory of cohesion and thus broadens her conception of conjunction which comprises both coordinating and subordinating conjunctions. As she rightly observes, "it makes more sense to take a broader view of cohesion and to consider any element cohesive as long as it signals a conjunctive-type relation between parts of a text, whether these parts are sentences, clauses (dependent or independent), or paragraphs" (Baker 1992:192). Baker's view is also adopted in this study to simplify the analysis of conjunction in film translation. Hatim and Mason (1990) discuss conjunction along the same lines as De Beaugrande and Dressler (1981), and focus on junction and inter-propositional coherence. Junction refers to "surface signals of relations among events or situations in a text world" (Hatim and Mason 1990:205-206). These relations include those developed by Halliday and Hasan (1976) and adopted by Baker (1992), as well as the general semantic relations elaborated by Crombie (1985) in terms of binary values such as Cause-Effect or Condition-Consequence. Hatim and Mason (1990:206) note that these relational categories are always inferable when they are not explicitly signalled in natural discourse.
Inter-propositional coherence involves explicit and implicit relations in both spoken and written discourse. Explicit relations are considered as internal relations holding between propositions in the text world, whereas implicit relations are considered as external relations www.spilplus.journals.ac.za holding between events and processes in the real world. When inter-propositional relations are not explicitly signalled, Hatim and Mason (1990:208) suggest that readers are constantly involved in using the inferencing strategy to make the text coherent. In brief, while junction is seen as a cover term for overt signals that lead to cohesion, inter-propositional relations are viewed as perceived intentions that should always be preserved in translation in order to maintain the coherence of a text.
In interlingual subtitling, conjunction is discussed by De Linde and Kay (1999) while dealing with textual cohesion. Following Halliday andHasan (1976), De Linde andKay (1999:29) argue that conjunction is different from reference, substitution and ellipsis in that its cohesive devices "signal relationships between parts of a text [and thus] are not anaphoric in nature". Like Halliday (1985Halliday ( /2004, the authors distinguish between four main types of conjunction: additive (e.g. and, or), adversative (e.g. but, however), causal (e.g. so, consequently), and temporal (e.g. then, finally). It is interesting to note that in this taxonomy De Linde and Kay have ignored the continuative items that were included in Baker's (1992) classification and which were investigated by Chaume (2004).
However, in line with De Linde and Kay (1999:29-30), it may be argued that the role of these conjunctions is to make the relationships explicit in the text and not to create new ones. This implies that when two clauses are not conjoined, they may be juxtaposed for the continuity of the message in spoken or written discourse. The juxtaposition of the two clauses will thus serve to signal the relationship through binary values (Crombie 1985). This occurs in different languages and may be one of the reasons why subtitlers are tempted to omit some cohesive markers which are inferable from the context.
At this stage, it may be argued that the notion of 'conjunction', as it is discussed by different scholars surveyed earlier, raises a number of issues regarding its status, its taxonomy, and the juxtaposition of clauses which leads to binary values. Thus, the key questions that arise before discussing the different types of conjunction in both French and English are: How is conjunction conceived of in this article? Furthermore, is it appropriate to consider discourse markers and binary values as conjunctions or as transitions?
Following Bloor and Bloor (1995:24, 56), it must be said that there is some ambiguity in the use of the term "conjunction" which, in Halliday and Hasan's (1976:242-243) classification as well as that of Halliday (1985Halliday ( /2004, includes both conjunctions and adverbials. This confusion may be clarified by referring to research on connectives and connection in French linguistics (cf. Chaurand 1987, Corblin 1987, and by adopting a pragmatic approach to conjunction in systemic functional linguistics (SFL). In this approach, the CONJUNCTION system consists of connectives that are cohesive resources establishing connections between clauses, clause complexes, and paragraphs in a given text. These connectives comprise two types of resources, namely conjunctions and transitions (Chaurand 1987:216).
Conjunctions are those cohesive resources which connect ideas between clauses of equal status through coordination, and between clauses of unequal importance through subordination. Item 57 from the data is a good example of the coordinating conjunction et ('and'), whereas item 60 clearly illustrates the subordinating conjunction si ('if'). In these two items, both conjunctions have cohesive power because they join two semantic entities encoded in the clauses. These clauses are the minimum grammatical units which correspond to speech acts as basic semantic units of communication. This means that for a conjunction to have cohesive power it should join two or more clauses at the grammatical level which correspond to two or more speech acts at the semantic level. As Corblin (1987:153) and Tesnière (1959:44) argue, there is never a structural or grammatical connection without a semantic connection.
By contrast, transitions are cohesive resources which connect ideas between separate clauses and/or clause complexes to clarify relationships between them. These transitions are established through discourse markers (Chaume 2004, Ducrot and Bourcier 1980, Schiffrin 1987 or continuatives (Halliday and Hasan 1976:228, 267), and through binary values (Crombie 1985:3). These binary values establish both general semantic relations in any type of discourse, and interactive semantic relations in conversational discourse. Whereas discourse markers are overt signals, binary values may be signalled explicitly or implicitly between juxtaposed clauses. When they are implicit, they are inferable from the context. Items 10 and 51 illustrate different types of transition.

51.
F: Et les allemands, en fait, c'est vrai qu'ils ont tiré? E: What about the Germans? Did they really fire at you? H: And the Germans? Did they fire at you?
In 10a, the transition is open since there is no linking signal to establish the relationship between the two clauses. In this case, the binary value of Result-Reason that establishes the semantic relation is assigned to the two clauses by inference. In contrast, this binary value is explicitly signalled in (10b) and (10c) by the subordinator parce que and because, respectively.
In 51, however, the transition is closed since the discourse marker et, translated as 'what about', fills the clause-initial position. This discourse marker indicates both the transition from one idea to another and the continuity of the message at the clause level.
It is important to note here that open transitions are more frequent in conversational discourse as hearers are constantly involved in making inferences in order to establish semantic connections between juxtaposed clauses. These semantic connections lead hearers to interpret the clauses which are thus juxtaposed as pieces of cohesive and coherent discourse. Figure 1 is www.spilplus.journals.ac.za a summary of the CONJUNCTION system, or the system of connectives that is involved in establishing semantic relations in cohesive and coherent discourse. However, there is no obvious reason to limit the CONJUNCTION system to the above categories or parts of speech. Chaurand (1987) suggests that there are other potential connectives, such as the verb, which could be included in the system. He argues quite strongly that (English translation to follow): L'étude des connecteurs s'est surtout orientée jusqu'ici vers les conjonctifs et les adverbiaux. Les autres parties du discours, le verbe en particulier, ont été, à quelques exceptions près, négligées, ou fondues à d'autres catégories qui paraissaient à cet égard prédominantes ou même exclusives. Je me propose […] de montrer à quel point certains verbes sont à la base de connexions multiples et variées. (Chaurand 1987:216) ['Up to now, the study of connectors has been especially oriented towards conjunctions and adverbials. The other parts of speech, the verb in particular, have been, with a few exceptions, neglected, or fused into other categories which, in this regard, seemed to be predominant or even exclusive. I propose to show […] to some degree how some verbs are at the root of multiple and various connexions.'] Thus, following Chaurand (1987) and Corblin (1987:152-156), conjunction as a system should be replaced by the term "connexion". The reason for this is that the word "conjunction" is restrictive on grammatical grounds, whereas the term "connexion" is neutral and has the potential to include, characterise and describe the cohesive resources of conjunctions, transitions and what Chaurand (1987:223-225) refers to as "verbes connecteurs" or connective verbs. The latter are verbs and verbal expressions which are involved in establishing semantic connections between clauses and/or clause complexes in interactive discourse. It is important to note that the term "connexion", denoting the system, will be spelt with an 'x' in this article to differentiate it from its counterpart "connection", which should be reserved for semantic, structural or any other connections. Since it is common practice in SFL to use capital letters for systems and lexicogrammatical terms (e.g. Adjunct, Mood, Theme, etc.), "connexion", as In this article, the focus will be on the system of conjunctions, leaving the transitions and connective verbs for further research. The systems of conjunctions in French and English are described in sections 2 and 3, respectively.

The system of conjunctions in French
Conjunctions in French have already been dealt with in traditional grammar by scholars such as Cayrou, Laurent and Lods (1960), Dubois and Lagane (1973), Grevisse (1957), and Price (2003). This section describes conjunctions in French from a systemic functional perspective.
Following Caffarel (2006), conjunctions are investigated in both the logical and the textual metafunctions. The logical metafunction is one of the two components which make up the grammar of ideation, the other component being the experiential metafunction. The logical metafunction is mainly concerned with the provision of commonsense logic which contributes to establishing chains of functional and semantic relations between clauses. The textual metafunction, on the other hand, is a component of what Caffarel (2006:165) calls the "enabling grammar", and what is referred to in this article as the "grammar of text creation". This metafunction is mainly concerned with "the use of language to form coherent text as realised through THEME and INFORMATION" (Martin 1981:311) and with reference to its context of situation and culture.
In the logical metafunction, conjunctions relate to the CLAUSE COMPLEX system. This system distinguishes between two types of interdependency relations, namely parataxis and hypotaxis. Parataxis takes place between clauses of equal importance, and comprises coordination and juxtaposition as well as direct speech and thought. Hypotaxis occurs between clauses of unequal status, and includes subordination (excluding embedding) and indirect speech and thought. The PARATAXIS and HYPOTAXIS systems are known together as the INTERDEPENDENCY system or TAXIS. This system interacts with the LOGICO-SEMANTIC RELATION system that consists of two types of relations, namely expansion and projection.
www.spilplus.journals.ac.za Expansion comprises circumstantial relations introduced by subordinators as well as the relations of addition and apposition. Following Caffarel (2006:23), circumstantial relations refer to the relations of enhancement, the relations of addition to extension, and those of apposition to elaboration. These relations may combine with TAXIS. Thus, for example, the combination of parataxis and elaboration produces juxtaposition and addition, and the combination of parataxis and extension and that of parataxis and enhancement leads to coordination. In contrast, the combination of hypotaxis and elaboration yields non-defining relative clauses; the combination of hypotaxis and extension and that of hypotaxis and enhancement results in adverbial clauses via circumstantial subordination.
Projection comprises the notions of 'direct speech' and 'indirect speech'. As with expansion relations, projection relations can combine with TAXIS. For instance, the combination of projection and parataxis yields direct quoted speech (locution) or thought (idea). In contrast, the combination of projection and hypotaxis produces indirect quoted speech or thought. In brief, projection and parataxis are characterised by the juxtaposition of clauses in direct speech, and projection and hypotaxis by the subordination of clauses in indirect speech. Figure 3 demonstrates the composition of the CLAUSE COMPLEX system and Figure 4 the interaction of TAXIS with LOGICO-SEMANTIC relations in French. These figures have been adapted from Caffarel (2006:24, 26) to which the reader is referred for further details.  Since this article mainly deals with the analysis of conjunctions, the focus is on the interaction of projection with hypotaxis, the interaction of parataxis with extension and enhancement, as well as the interaction of hypotaxis with extension and enhancement. The systems of paratactic expansion and hypotactic expansion appear in Figures 5 and 6   In the textual metafunction, conjunctions are, according to Caffarel (2006:171), included in textual Themes together with discourse markers or continuatives. When they are used in the same clause, these textual Themes precede the interpersonal Themes (i.e. Vocatives, modal Adjuncts and Mood markers) and topical or experiential Themes (i.e. what the clause is about and not necessarily the Subject). I shall return to textual Themes in detail when analysing conjunctions in section 4. However, we turn now to the description of conjunction in English. www.spilplus.journals.ac.za

The system of conjunctions in English
As in French, conjunctions have already been discussed in English grammar by scholars such as Huddleston (1984), Leech and Svartvik (1975), Quirk, Greenbaum, Leech and Svartvik (1972), and Thomson and Martinet (1986). The following description is from a systemic functional perspective.
In SFG, conjunctions have been mixed with sentence adverbials or discourse Adjuncts by Halliday (1985Halliday ( /2004. This mixing, also carried out in French by Caffarel (2006:34) in her markers of paratactic expansion, is due to the bivalence of some connectors, such as however, nevertheless and therefore which are used as both conjunctions and adverbials as suggested by Thomson and Martinet (1986:288).
Following Halliday (1985Halliday ( /2004, these discourse Adjuncts are non-structural and are therefore used only cohesively. In the same vein, it is suggested that some universal conjunctions, like et ('and') or mais ('but'), are used as discourse Adjuncts when they occur in the clause-initial position, and have some pragmatic value to strengthen the continuity of the message. Item 51 (presented earlier) is a good case in point; since it has a pragmatic value, the conjunction et is translated either as 'what about' in the subtitle or as 'and' in Hollo's translation (Malle and Hollo 1988).
As in French, these conjunctions can be dealt with in both the logical and the textual metafunctions. In the logical metafunction, conjunctions relate to two basic relations that characterise the CLAUSE COMPLEX system. These relations are the LOGICAL DEPENDENCY relations or TAXIS and the LOGICO-SEMANTIC relations (Halliday 1985:373, Thompson 2004. The LOGICAL DEPENDENCY relations consist of hypotaxis and parataxis. The latter comprises coordination, juxtaposition, and both direct speech and thought. The former includes subordination minus embedding, and both indirect speech and thought. Once again, as in French, the LOGICO-SEMANTIC relations consist of expansion and projection. In the relation of expansion, one clause expands on the meaning of another through the process of elaboration, extension and enhancement. In the relation of projection, one clause projects another through both direct and indirect speech. These LOGICO-SEMANTIC relations of expansion and projection interact with the LOGICAL DEPENDENCY relations of hypotaxis and parataxis. This interaction yields the same results as in French. For example, hypotactic enhancement produces subordination in terms of adverbial clauses, whereas paratactic enhancement produces coordination. Likewise, hypotactic extension leads to subordination while paratactic extension results in coordination. Finally, hypotactic elaboration produces apposition and juxtaposition, whereas paratactic elaboration yields non-defining relative clauses as well as non-finite clauses. Further details on these relations and their interaction can be found in Thompson (2004:198-214).
In the textual metafunction, conjunctions are considered as one type of textual Theme, the others being continuatives and conjunctive Adjuncts, according to Halliday's (1985Halliday's ( /2004 taxonomy. The distinction between these three types of textual Themes is that conjunctions are either paratactic linkers or hypotactic binders of the clause in which they occur, continuatives are markers which signal a move in the discourse, and conjunctive Adjuncts are adverbials and www.spilplus.journals.ac.za prepositional phrases relating the clause to the preceding chunk of information or text (Bloor and Bloor 1995:24, 98-99;Halliday 1985Halliday /2004. However, as mentioned earlier, some universal conjunctions, such as and or but, may be used as continuatives in the opening phase of a conversation. In this case, they simply "serve in the system of continuity [and] are a characteristic feature of dialogic text" (Halliday 1985(Halliday /2004. Item 51 (presented earlier) is a good example of the conjunction and used as a continuative in the clauseinitial position in Hollo's translation.
Thus, Figures 7 and 8 summarise the system of paratactic expansion and that of hypotactic expansion, respectively. These two systems are adapted from Halliday (1985Halliday ( /2004.
Although they seem to be more elaborated than their French counterparts, they contain the same types of conjunctions at the paratactic and hypotactic levels. They are thus helpful for the contrastive analysis of conjunctions in French and English that will be carried out in section 4.

Analysis of conjunctions in the film
The aim of this section is to analyse the data collected from the film Au Revoir les Enfants (Malle 1989). All the data are listed in the Appendix, and consist of 94 items of French utterances, English subtitles and Hollo's English translations (Malle and Hollo 1988). The French utterances (F) constitute the source text (ST) based on Malle's (1987) film script. The English subtitles (E) constitute the target text (TT) based on the ST. Hollo's English translations (H) are also based directly on the ST, and are used to serve as the frame of reference for the shifts that occur from the ST.
The conjunctions are written in bold and italics in the ST, the TT and the H translations. In describing the conjunctions in the French utterances and in the English subtitles, reference will be made to Hollo's translations to see whether the same conjunctions are maintained or omitted by the subtitler. This will assist in drawing conclusions on the impact of time and space constraints as well as the effect of the principle of relevance on the subtitles. On this point, let us turn to the analysis of conjunction proper, following Martin's (1981:310-311) clear-cut proposal that conjunction is located at the message-group or clause-complex level, and continuity at the message or clause level.

The conjunction et and its translation equivalent
Et is one of the universal and most versatile conjunctions which speakers and writers use to coordinate clauses of equal importance or to show the continuity of messages in French. It is a positive additive connector establishing the relation of extension in the paratactic expansion.
In the film, the conjunction et appears in 4, 5, 6, 7, 11, 12, 18, 35, 43, 44, 57, 66, 73, 89 and 90. It is translated into its equivalent 'and' in 4, 5, 6, 11, 12, 18, 44, 57  In all these cases, the H translations are not only structurally and semantically very close to the utterances in F, they are also longer than the English subtitles. This suggests that sentence length may have militated against their being used as appropriate subtitles because of time and space constraints. In the English subtitles, the omission of et has led the subtitler to operate some syntactic and lexical transformations. These transformations may be considered as strategies which help subtitlers to reduce the text to a manageable size in order to accommodate it for the screen and the speed at which the film's plot progresses.

The conjunction mais and its translation equivalents
Mais is another universal conjunction that is used to coordinate clauses of equal importance in French. It is an adversative conjunction establishing the relation of extension in the paratactic expansion. In these two items, both instances of the elliptical clause oui is omitted in E to comply with the principle of relevance, whereas it is maintained in the H translations. This shows that the H translations are very close to the French utterances from a syntactic point of view.

The conjunction car and its translation equivalent
Car is a purposive conjunction establishing the relation of causal enhancement in the paratactic expansion. Its peculiarity is that the clause in which it appears never precedes another one to which it is attached. This conjunction occurs twice in the film and twice it is omitted in the translations, as shown in items 6 and 11. Here again, the H translations are very close to the ST in F as opposed to the subtitles in E. In 6, car is omitted in E, and the coordination of clauses is replaced by juxtaposition which establishes the Effect-Cause relation between the two clauses. In addition, there are changes in the syntax of the juxtaposed clause: the main changes are the omission of the adverb vraiment ('indeed'), the replacement of the indefinite article with the definite one, the addition of the adjective "real" as a modifier of 'food' and 'drink', and the replacement of a comma with a full stop. These changes are purely stylistic and do not affect meaning, yet they do make the subtitle in E sound more informal.
In 11, however, the translation equivalent of car and the clause in which it appears are simply omitted in E on the assumption that readers know the verse. This confirms that spatio-temporal constraints and the principle of relevance are effectively at work in interlingual subtitling. Thus, any piece of information that is assumed to be mutually known by subtitlers and readers, or any constituent that can be inferred immediately from the context, is likely to be omitted from the subtitles.

The conjunction donc and its translation equivalent
Donc is a consequential connector that establishes the relation of causal enhancement in the paratactic expansion. It generally expresses consequence (Cayrou et al. 1960:271) and thus "comes first in the clause" (Price 2003:483). This conjunction occurs in 16 where it indicates logical consequence in the clause-initial position. It is rendered into its equivalent 'therefore' in E and H, and stays in the same position in these two contexts. Since donc is not omitted in both translations, there is no need to list item 16 here, and the reader is referred to the Appendix for details. We now turn to the analysis of the conjunctions of subordination.

The conjunction que and its translation equivalents
Que is a connective that introduces various types of subordinate clauses in French (Dubois and Lagane 1973:146). The most notable of these clauses are conjunctive clauses (Cayrou et al. 1960:324, 328, 330) or que-clauses (Price 2003:541), as well as the subordinate clauses of comparison, concession and purpose. Concerning the conjunctive clauses, que sets up the relation of hypotactic projection between the projecting and the projected clauses (Thompson 2004:210-213). In these cases, it is translated into its equivalent 'that' which, in informal English, is often omitted when the "that"-clause (Leech and Svartvik 1975:249) functions as a complement, an object or a postponed subject.
In the clauses of comparison, que is used in 17, 21, 24, 30, 42, 56, 78 and 93. In each of these items, que is translated into its equivalent 'than', which is maintained in the English subtitles.
In the clauses of purpose, que appears only in 8 where it is omitted together with the projected clause.
www.spilplus.journals.ac.za It can be argued that there is a loss of information in 8 occasioned by the omission of the clause of purpose in the English subtitle. The reason for deleting this information is that it is made redundant, since what the speaker is doing is visible in the moving picture. This means that the verbal information is complemented by the visual information. Thus, it becomes irrelevant and is removed from the subtitle to comply with the constraint of time and space.

The conjunction si and its translation equivalents
Si is a connective that introduces conditional clauses and establishes the relation of enhancing condition in the hypotactic expansion in French. It is translated as 'if' when it has a positive meaning (i.e. when it is used alone), and as 'unless' when it has a negative meaning (i.e. when it is followed by a negation). These two translation equivalents establish the same relation in English. In the film, si is used in 2, 4,5,15,19,23,37,54,60,67,68,69,74,79 and 92. It is rendered as 'if' in 19,60,68,69 and 79,and as 'unless' in 5. In the remaining items, si is omitted and different syntactic strategies are used to translate F into E. In 2, the si-clause is turned into a clause simplex and coordinates with the appositive clause, while the main clause is turned into a clause simplex. In 4, the si-clause and its main clause are combined and replaced with an imperative clause. This is then joined to the preceding clause simplex. In 15, the projecting clause 'I wish' is used in replacement of si to introduce the projected clause and to express the speaker's regret at not having another job. In 23, 37 and 92, the si-clause and the main clause are combined to form a clause simplex. In 54, the si-clause is turned into an imperative clause. This is then joined to the main clause by the alternative conjunction 'or', thereby changing the condition to a face-threatening act (Brown and Levinson 1987:247). In 67, the si-clause is simply omitted. Finally, in 74, the si-clause is turned into an interrogative one asking for permission, while the main clause is turned into a clause simplex.
Thus, clause deletion, sentence combining, clause replacement, transforming a si-clause into an imperative or an interrogative clause, and turning conditional and/or main clauses into clause simplexes are the main syntactic strategies used by the subtitler. The reason for doing so is to reduce the text and accommodate it to the speed of the film plot. However, text reduction is not only syntactic but also lexical and phonological.
From a lexical point of view, the subtitler uses short words or expressions whereas Hollo uses longer expressions. For example, in 2, the verb déguisais is rendered as 'to dress up' in E and as 'to disguise myself' in H. Likewise, in 74, the verb rentrais is rendered as 'come' in E and as 'went back' in H, and the noun papa is rendered as 'Dad' in E and as 'Papa' in H. Finally, in 92, the clause il m'aurait jamais foutu à la porte is rendered as 'I got fired' in E, and as 'I wouldn't have been fired' in H.
From a phonological point of view, there is some preference for using contractions since these shorten words and therefore aid in text reduction. For instance, in item 2, 'I'd' appears in E and 'I would/could' in H. Likewise, in item 4, 'I'm' appears in E and 'my name' in H. Thus, using contractions is a phonological strategy that subtitlers use not only to reduce the text but also to give the "flavour" of the spoken language to the subtitles. www.spilplus.journals.ac.za

The conjunction quand and its translation equivalent
The conjunction quand introduces a subordinate clause of time, and establishes the enhancing relation of time between clauses in the hypotactic expansion in French. The same holds true for its English equivalent 'when'. In the film, quand is used in 11, 25, 58 and 84. In these items, quand is translated as 'when' in 11, 58 and 84. In 25, quand and the clause in which it appears is formulated with the genitive case. Unlike the adverbial clause in the H translation in 25, the genitive case in E has reduced the text to comply with the spatio-temporal constraints. Thus, formulating a subordinate clause with the genitive case is another strategy for text reduction in interlingual subtitling.

The conjunction comme and its translation equivalent
Comme is a connector that introduces a subordinate clause of comparison and establishes the enhancing relation of manner between clauses in the hypotactic expansion in French. Its translation equivalent 'as' also sets up the same relation in English. In the film, comme is translated as 'as' in 20 but is omitted in 58.

58.
F: Comme je comprends la colère de ceux qui n'ont rien, quand les riches banquettent avec arrogance. E: I understand the anger of the poor, when the rich feast so arrogantly. H: How well I understand the anger of those who have nothing, while the rich feast so arrogantly.
In this item, not only is comme omitted but the pronominal group ceux qui n'ont rien also becomes the nominal group 'the poor'. Furthermore, the omission of comme has led the subtitler to translate the adverb clause of comparison in F as a main clause in E. Thus, replacing a pronominal group with a nominal one and turning an adverb clause of comparison into a main clause are other syntactic strategies used to reduce the text in interlingual subtitling.

The conjunctions parce que/ puisque and their translation equivalents
The conjunctions parce que and puisque are two connectives which introduce subordinate clauses of cause and establish the enhancing relation of cause between clauses in the hypotactic expansion in French. Apart from their difference in spelling, they have similar meanings (Dubois and Lagane 1973:192) and are translated as 'because' and 'since', both of which establish the enhancing relation of cause and reason in the hypotactic expansion in English. In the film, parce que is used in 29, 52, 53, 55 and 88, and puisque in 39 only. Parce que is translated as 'because' in 52, 53 and 88, whereas its counterpart puisque is rendered as 'since'.
In the remaining items, parce que is omitted in 29, and in 55 it is replaced with 'so'. Mubenga www.spilplus.journals.ac.za 29. F: Les types qui se cachent parce qu'ils ne veulent pas aller faire leur travail obligatoire en Allemagne. E: Guys avoiding forced labor in Germany. H: Guys who hide out because they don't want to go do their forced labor in Germany.

55.
F: Parce qu'on vous a donné beaucoup, il vous sera beaucoup demandé. E: You've been given much, so, much will be asked of you. H: Because much has been given to you, much will be asked of you.
In 29, the omission of parce que is also accompanied by the omission of the relative clause qui se cachent; the latter has been replaced with the present participle 'avoiding', and the subordinate clause of cause has been replaced with a nominal group to reduce the text in E. In contrast, parce que is omitted from E in 55 and replaced with the consequential connector 'so', thus changing the enhancing relation of cause to that of consequence. This replacement of parce que with 'so' is accompanied by a change in the thematic structure of the causal clause in the sense that the Indirect Object Complement vous in F becomes the topical Theme 'you' in E. Thus, the subtitler uses these syntactic transformations as strategies to reduce the text, while the H translations are closer to the ST. At this stage, the question is: What can be learned from this systemic functional analysis of conjunction in Au Revoir les Enfants?

Concluding remarks and recommendations for further research
In an attempt to answer the question concluding the previous section, it may be said that this analysis is important for various reasons. The first and most important reason is to confirm the initial hypothesis that some conjunctions are omitted in the translation of the film from French into English because of the effects of time and space constraints as well as the principle of relevance imposed on the subtitles. This paper provides sufficient evidence in this respect. The analysis detailed in this paper involved four types of coordinators (i.e. et, mais, car and donc) which occur at the clause level, and five types of subordinators (i.e. que, si, quand, comme and parce que/ puisque) which occur at the clause-complex level. The findings reveal that these coordinators and subordinators are subject to omission in the subtitles in order to comply with the spatio-temporal constraints and the relevance principle.
Concerning the coordinators, all except donc have been omitted in some contexts in the subtitles. The conjunction et has a higher percentage of omission than the conjunctions mais and car. Regarding the subordinators, all except puisque have been omitted in some contexts in the subtitles, which is understandable since puisque was used only once in the film. The conjunction que in the "that"-clause has a higher percentage of omission than other subordinators and the highest percentage of omission of all the conjunctions in the film. These omissions are characterised by a wide range of transformations in the TT and thus lend full support to Chaume's (2004) hypothesis that discourse markers are omitted in audiovisual translation. Table 5 shows the statistics for the omission of the different conjunctions. The second important reason for conducting this analysis is that it allows for the establishment of the system of French and English conjunctions in the film. The system of French conjunctions is summarised in Figure 9 and that of English conjunctions appears in Figure 10.
The systems in these figures consist only of those conjunctions found in the film. The barred zero Ø indicates that the conjunction is not found in the ST or is omitted in the TT. These systems may thus serve as taxonomies of the conjunctions in the film. The conjunctions or idiomatic expressions used in the H translations are not included in these taxonomies.  The third related reason for conducting this analysis is that the omission of conjunctions in the subtitles is accompanied by shifts and/or transformations at the lexical, phonological and syntactic levels. These shifts and transformations have allowed for the determination of a wide range of strategies used by the subtitler. The following twelve strategies seem to be the most striking ones, and could well be applied when teaching students interlingual subtitling: (i) Deleting elliptical clauses and some adverbs; (ii) Changing a compound clause to a clause simplex and a nominal/prepositional group; (iii) Changing a compound clause into juxtaposed clauses; (iv) Changing a prepositional group into a clause simplex; (v) Changing a pronominal group into a subordinate clause or a nominal group; (vi) Changing a relative clause into the present participle and a subordinate clause to the genitive case; (vii) Formulating an if-clause as I wish plus a projected clause; (viii) Replacing a conjunction with another one; (ix) Replacing an indefinite article with the definite one; (x) Replacing the topical Theme with another one; (xi) Using contractions rather than full forms, and (xii) Using short informal words and/or expressions The above strategies are not exhaustive, nor are they intended to be complete; they are used here only for illustrative purposes. Further research, however, needs to be conducted to establish their status and the impact of the omission of conjunctions in interlingual subtitling.
The final reason for conducting this analysis is that the systemic functional analysis of conjunction in the film, the taxonomy of conjunctions and the discovery of lexical, phonological and syntactic strategies have pedagogical implications for subtitler training. Subtitlers are likely to do their job well if they are well-trained. The systemic functional analysis of conjunction demonstrates both the paratactic and the hypotactic relations that conjunctions establish within clauses. Knowledge or mastery of these relations is a prerequisite for using conjunctions effectively and efficiently in translation as an act of communication. www.spilplus.journals.ac.za The taxonomy of conjunctions is important as an aspect of language awareness in the source language and the target language. As such, it contributes greatly to developing the subtitler's linguistic and communicative competence as well as his/her contrastive and cultural competence. These types of competence are much needed in translation to know when a given conjunction should be omitted and which syntactic structures and vocabulary should be used instead so as to preserve the meaning of the ST in the TT.
Finally, a set of lexical, phonological and syntactic strategies may well serve as models of subtitling strategies that novice subtitlers have to develop and use when learning to write meaningful subtitles. Training in subtitling should focus on these strategies and may in the process discover others which can facilitate the use of conjunctions in interlingual subtitling. Thus, developing subtitling strategies of this sort should be greatly encouraged and fully implemented in all programmes for subtitler training.