IN SEARCH OF PRONOUN EQUIVALENCE: CHALLENGES IN TEACHING INTERPRETING

There are some challenges in Interpreting; in this study, we focus more on the interpreting of pronouns from English as Source Language (SL) to Indonesian as Target Language (TL). The data was collected within simulations in classroom setting. The results showed many unnatural translation of pronoun ‘it’ and ‘this’. References of these pronouns are anaphoric and may take different linguistic units (clause, phrase, word). There were some strategies employed to achieve natural interpreting in TL. Some students repeated the reference(s); but they need to recover the whole references (might be very challenging when the reference(s) are complex). Other strategies are partial recovery, or lexical addition that complies with TL structure. There are some suggestions for an interpreter to deal with the challenges; one of them is to be familiar with the topic of what they are interpreting. In this way, they do not have to rely on short-term memory skills to decide which strategy to employ.


Introduction
Translation is the conversion of text or speech from source language (SL) to target language (TT). As for Interpreting, it is a kind of oral translation, instead of textual translation. The difference is well discussed by Tirkkonen-Condit and Jääskeläinen (2000).
As this study discusses the application of translation/interpreting in a classroom setting, it is important to review 'translation' class here before we proceed to 'interpreting'. The discussion is restricted to language-related program. We need here to distinguish translation as a practical skill, and translation as a study, although both are aimed at delivering meaning to speakers of target language (Robinson, 2004). As a practical skill, translation is closer to the industry. Therefore, a translator must be able to satisfy the demand of the industry. Fair and acceptable translation is appreciated financially by the clients. Often, the key success is the mastery of domain specific terms, such as legal, chemistry and religion (cf. Torrejón & Rico, 2002). Therefore, many translators do not have language-specific degree.
As for translation study, it studies translation scientifically; translation product is reviewed scientifically and often discussed within linguistic/crosslinguistic frames, such as ideology, social structure and gender (cf. Munday, 2016). While translation classes at the university level offer both (practical and translation studies), the content in Interpreting classes deals with translation as practical skills. The aim of the class is usually to allow students to simulate interpreting events.

Methods
In this study, the data is obtained from observations during interpreting class and exams. The class is offered to undergraduate students and Diploma-3/vocational program. The material, however, is equally vocational (even to undergraduate students). Interpreting is considered more challenging than (text) translation as it requires 'faster' response.
Prerequisite to this class is ideally very strict (students have to be excellent in listening, speaking, and translation class). However, should that be the case, this should be an optional class. It is then decided to reduce the level of difficulty so that it is more accessible to various students. The aim is focused on interpreting in introductory level (allow students to simulate interpreting).
The material is designed from English to Indonesian, as it will be easier to deliver the oral translation. The audio files are (in monologue, obtained from (http://learningenglish.voanews.com/), given without subtitle. There are two types of interpreting; consecutive and simultaneous. As its name suggests, simultaneous requires the interpreter to do the interpreting task while the client is talking. As for consecutive, it waits for the client to cease, then the interpreter delivers the interpreting. The simulation in this introductory class is restricted to the level of consecutive interpreting, as much better comprehension and production is required in simultaneous interpreting (Bajo, 2000).

Findings and Discussion
Challenges in teaching interpreting are actually many (cf. Garzone & Viezzi (2002) and Braun (2007) for discussion in detail). Bogucki (2010) discusses the challenges within classroom setting. However, challenges usually lie on three stages; input, processing, and output. Input is the reception stage. As for interpreting, input is in the form of speech in SL. Some challenges are; the client is speaking too fast, the speech is long and complex, there are numbers and proper names, and the use of domain specific terms. Should the interpreters succeed at this stage; they may proceed to the next stage, which is processing.
In the processing stage, it is assumed that an interpreter is already aware of the client speech in SL. This is the stage where the SL has to be converted to TL. There are times when an interpreter knows what it means in SL, but cannot find the equivalence in TL. Another time, the SL is understood, but it is too complex and it requires to be restructured.
This stage is considered less vital than the previous ones. Problems in this stage are usually related to physical factors (ex: the interpreter speaks too slowly, too fast, too softly or too loudly). However, the interpreter's accent is sometimes a problem. Say, in Indonesian or English, saying [r] with thrill or flap does not change meaning. However, in some TLs, they do distinguish meaning. Pronoun reference is a problem that lies in input and processing stage. The central role of pronoun in linguistics is studied by Gordon (1993).
Students are actually familiar with pronoun reference exercise in many language testing simulations, but they are all receptive (like questions in TOEFL, 'to whom does 'she' refer to', 'it in line 5 refers to …'). The reference is also not hard to discover as it is usually anaphoric (located in the previous speech/sentence). The skill it requires is merely identification. As for English, grammatical and semantic features often help (ex: they is plural, she is a woman, it is a non-human entity, etc). There have been several efforts to computationally deal with this challenge. See LeNagard & Koehn (2010) and Nakaiwa & Shirai (1996). Identification is not the problem, when they read. But when they hear speech sounds, which will be a different case, as they are required to recover the reference very quickly. This of course will affect the production. If they fail the identification stage, then how will they be able to give output? Figure 1. Unnatural Translation to TL across Pronouns Figure 1 shows the results of the research, where errors related to pronoun are dominated by 'this' and 'it'. Errors here are not necessarily absolute, but they do make the TL sound unnatural. There are some reasons why errors such pronouns are dominating over the others. One reason is their frequency in the data. 'This' and 'it' dominates over the other pronouns. Failure to deliver proper interpreting on such pronouns might also be driven by the location of the reference, which is usually in the previous sentence. Consider In table 1, two different ways of translating pronouns are identified. The first one is literal translation, where 'it' is translated into ini. However, in the second one, the translation is enhanced with hal 'matter'. We understand that the reference of this, is 'understanding', which actually the head noun. In this case, it will refer to the whole noun phrase. By using hal ini 'this matter', it will refer to the whole phrase. While using 'it' is enough in English to achieve natural coherence, using hal ini makes the equivalence in Indonesian more natural. Let us have a look on another data, which is present in table 2: The reference of 'this', is on the clause level 'breeding has …'. The first translation refers to the literal translation. The second translation is more natural. Unlike table 1, the second translation in example 2 takes different approach. Instead of complementing hal 'matter' with ini 'this', the reference is repeated with sentence structure modification. For another approach, let us have a look at table 3: The strategy employed in the translation as shown by table 3 is a little bit different. In the first translation, literal translation strategy is repeated twice, in two different sentences. In the second example, a construction of clause combination (in one sentence) is employed. Instead of following SL structure, the student used similar to (but not the same) 'not only-but also' construction to TL 'selain …, juga…'. As specifically for 'it', it is translated into buku ini 'the book'. The word it is not translated to its literal equivalence, however the insertion of ini may equally show definiteness.  Tables 4 and 5 will be described in unison. The reference of the two tables is the same as table 3. In the English version in table 4 and 5, pronoun 'it' is used. In the first translation, literal translation strategies are used. Better translations are provided in the second examples, although meaning shift actually takes place. Pronoun 'it' in these examples actually refers to the whole book. But in reality, 'it' is only parts of the book; therefore, sebagian 'part' is employed. Here, the translation is not faithful, but still correct and natural although they do shift the meaning. Unlike previous examples, the addition here (as shown by the second example), is bound form/particle -lah is added. The function of -lah are many. One of them is to mark definiteness. However, why such definite marker is required while the head is already there (making it already definite)? We resort to another meaning, which is the emphasis. So, the translation, besides natural, also emphasize on the focus of the reference. The presence of -lah makes the TL stay coherent. In table 7, one student employed literal translation, while another student negotiated the meaning in the TL. In the TL, student 2 translated it (singular) into hal-hal (plural reduplication). The reduplication changes the noun reference into plural. However, this seems very well tolerated by the ears of native speakers of TL as sang songs and like singing are the references as well. These two references are well correlated so that the shift is still acceptable. Audience does not seem to bother ini and itu as the shift is still fairly acceptable.

Why IT and THIS: Some Suggestions
The two pronouns (It and this) are empirically frequent in the SL. What makes these two pronouns dominate the interpreting errors? Some pronouns are relatively less challenging, such as she, he, but this may pose problems too, and although not many, errors in translating these pronouns also took place (see figure 1). There are also other pronouns that are empirically frequent but the reference is not anaphoric. The references are located still in the same phrase, therefore they do not have to trace the reference back. As for pronouns that are not correctly translated, most of them are bare pronoun. So there are two main problems here. First, the target pronouns are empirically frequent. And the second, is that the reference is located anaphorically. Unlike translating texts, students cannot go back to the previous sentence to trace the reference(s).
What some students may not realize, those who successfully translated the reference with natural translations may not necessarily know the reference either! It is true that in some strategies when reference(s) are mentioned, students need to recover the reference fully and quickly. But in some other strategies, errors are overt because of only the choice of style. For instance, translating SL 'this' literally into ini without any lexical addition may seem strange. However, adding hal into hal ini might result in more natural translation. I argue that understanding natural structure in TL is also a must for an interpreter so that the conversion process can be executed in no time.

Conclusion
In the findings section, the statistic has shown that the errors are dominated when students are translating pronoun this and it in the initial position. Translating the pronoun literally does not comply with the natural structure of TL. Therefore, some strategies are employed. The first one is pronoun recovery, either full or partial. This, however, requires the interpreter to recall the pronoun reference(s). It might not be a problem when the reference is short, but for long complex reference(s), this may cause problem. Another strategy is by lexical addition (hal, tersebut, etc). The last strategy is deletion, where the pronoun is deleted in the TL. However, this strategy has to be employed very carefully as it requires additional strategy such as unification, in which the interpreter needs to reorganize the whole structure.
There are few ways to generate better translation in TL. The first one is by improving recall skills/memory skills. But this might be difficult for beginner as memorizing skill is gained by experience. There are some useful papers to read in order to understand the link between memory and interpreting skill such as Lambert (1991), Hulme (1991), Liu (2004). Another useful tip is to focus on potential reference. In this way, reference is anticipated before the pronoun(s) surface. Reference is usually a noun. But, it can also be a phrase or the whole sentence. Pronoun also mostly refers to given instead of new information. That is why it is located anaphorically, and seldom cataphorically. In some cases, we can generalize the reference, especially when the reference is in sentence level.
In trainings, of course an instructor can slow down the rate of speech in the recording. Many tools can be used to manipulate the rate. But this must not happen continuously as simulation is conducted to give experience interpreting as closely to realities as possible. A better suggestion, for an interpreter is to gather a lot of information about the topic that s/he is going to deal with. It will be a good idea to spend a few minutes with the client and talk about the topic before the interpreting process is delivered on public.