Abstract
This project addresses the problem of Bible translations that appear wooden as a result of following a source language text too closely. It examines what effect a discourse workshop focusing on receptor language structures had on translators and their revisions to selected Scripture texts in three translation projects in Liberia. The article evaluates the process, material, and outcomes of the workshop and gives recommendations for improvement and further training.
Keywords
Introduction
This article is condensed from my DMin thesis entitled “Employing Natural Discourse Forms in the Translation of the Bible: Assessing a Discourse Workshop” (Grossmann 2019). The project addresses the problem of Bible translations that appear wooden as a result of following a source language text too closely. It examines what effect a discourse training workshop focusing on receptor language (RL) structures had on both translators and their revisions to selected Scripture texts in three translation projects in Liberia, West Africa. The article evaluates the design, process, material, and outcomes of the workshop, and gives recommendations for improvement and further training. The main content and format for the workshop were designed by Carla Bartsch, SIL Asia Area Linguistics Consultant, serving internationally. The entirety of the project can be obtained at https://gordonconwell.on.worldcat.org/oclc/1101443669. Bartsch’s material is in the process of being published by SIL’s Global Publishing Services.
Much has been written about the importance of using natural structures in Bible translation—forms that are found on both micro and macro levels. However, in my experience, I have found that teams tend to follow source texts closely, resulting in somewhat wooden translations. Given the amount of writing on the topic, I suspect this is true of others’ experience as well. In Liberia, where I have done all my work, English is the national language. Most teams use English versions as their base texts. Many people have much of their English version memorized. As I was working with teams, I noted that I could follow the English text closely. I was bothered by the fact that I could tell which English version they were using as their base text. Clearly, natural forms were being abandoned in order to follow the base text as closely as possible. Numerous reasons are behind this common problem: the natural translation tendency to copy source language (SL) structures, respect for the Scriptures and fear of changing too much, the influence of the prestige SL, and the division of the text into verses, reinforced by the translation software. But added to these is a lack of awareness of RL patterns. One of the hardest translation skills to learn is using a different structure from the source text to convey the meaning of the passage. It is harder to produce the more natural form when it is not present in the source.
Ideally, teams would be trained in discourse from the outset. In reality, many get trained in a bottom-up approach, initially learning how to deal with the lower-level features of language but never getting to an understanding of the structures at the discourse level. It is not enough to rely on the intuitions of the translator. Training in translation theory and in analyzing RL structures gives the translators a greater capacity to translate in a more natural and meaningful way as they know at conscious levels what tools they have, what they are used for, and how to use them at both lower and higher levels of grammar. No translator learns all the skills at once, so having a variety of training opportunities will equip a team to grow in their abilities. For this project, I set out to address the following two questions: What effect will a discourse workshop have on how teams translate biblical material? How will identification of discourse markers and features in the RL influence the translation to reflect natural text structures?
The project aimed to assess what capacity building takes place when established teams are given training in discourse features. The purpose of the study was to evaluate the effectiveness of an RL discourse workshop on translated biblical material. The thesis argued that training translators to recognize discourse features in their language would make them better equipped to produce a text that more closely resembles other natural texts in the RL. With such training, translators would more readily recognize and use features and structures that they could consciously apply to their translation in places where intuition alone is not sufficient, as evidenced in a comparison of before and after workshop texts. Thus, as a result of the workshop training, translators would produce a more usable and acceptable natural translation which is clearer and possibly more accurate.
Project design
The project was carried out through a two-week discourse workshop with three teams active in Liberia at the time. Later, a follow-up session was held with each team to assist with comprehension, assimilation, and application of the discoveries made during the workshop. These teams were previously trained in an introductory workshop using material from Katy Barnwell’s book Bible Translation: An Introductory Course in Translation Principles (SIL, 3rd ed. rev., 2002). Little was done at that time to address discourse structures that tie a text together, such as genre, boundary markers, referent identification, and peak-marking structures. Those introductory workshops were part of a translator selection process in projects with no resident linguist. Obtaining texts for analysis required more than the participants or workshop time frames could address.
Bartsch’s material is written for local translators who may not have a lot of formal linguistic training. It emphasizes self-discovery and provides instruction that shows examples. She also designed exercises to apply the new knowledge to translated text. Ideally, participants discover one or more features in the natural text and rewrite their translated text, incorporating such features where appropriate. Another positive aspect of our workshop was that the various translators met together as they learned. Given the time, they could share with one another what they had discovered, both within the teams and across teams. Two of the languages at the workshop are related and possibly share structures.
The workshop design gave each team the opportunity to analyze various discourse features of their language from texts gathered prior to the workshop, and followed by applying what was learned to select portions of already drafted Scripture. For the sake of simplicity and comparison, each team used the same four selections from the Gospel of Mark. These were selected because all the teams had an end-stage (approved for printing) base translation of the book of Mark. The Scripture portions selected were: “Jesus calms the storm,” Mark 4.35-41; “Jesus heals a man with evil spirits,” Mark 5.1-20; “Jesus blesses little children,” Mark 10.13-16; and “The parable of the tenants in the vineyard,” Mark 12.1-12.
A comparison of pre- and post-workshop Scripture texts as recorded in an individual Paratext project set up just for the workshop was the tool to reveal changes to the Scripture texts as translators applied new knowledge from their natural texts. The hypothesis was that the revised texts would reflect more of the features common in the natural texts, showing the benefits of such a workshop. The Paratext program allows one to compare the wording and features of the original text with the texts as they appeared after the workshop. The changes were assessed by me in comparison with the pre-workshop texts and with features discovered in the natural texts. The participants were to insert an explanatory note at each change, specifying the reasons for the change and, in so doing, giving another opportunity to assess what was comprehended from the workshop and what motivated the change in the text. In addition, some tools and charts on paper were used to count features in the natural texts and those in both the pre- and post-workshop translated texts for comparison. Analysis and assignments took more time than was expected. Consequently, the Paratext portion of the workshop was not completed until after a post-workshop review. Following the post-workshop review (described below), I collated the revisions from the Paratext files to determine the effect the workshop had on both the translators and the translated texts.
Implementation
Prior to the workshop, all participants were asked to gather some natural narrative texts. The criteria for each text were that it be a true story told initially in the local language and not a translation. The texts were to be 15–35 sentences, contain quotations, describe a problem solved, and be told in third person. These criteria help obtain texts that have features similar to narratives in the Bible. Such features include, but are not limited to, an initiating event, background information, participant reference, increased tension, problem-solving resolution, and closing. Covering all the possibilities of narrative discourse features was not the goal, neither was a comparative study of various genres or subgenres. Ideally, these texts were to be entered in the computer with a word-for-word translation as well as a free translation. Most of the texts were given to me on paper. I then transferred each story into the FLEX program for interlinearization and exported a copy which I edited in MS Word and formatted into two charts designed by Bartsch. These charts are exactly the same, except that the heading of the fifth column changes from “Speech, Word Order” to “Type of Boundary Marker, Connector” in order to focus the analysis on different discourse features (Figure 1).
Charts for story analysis, with different fifth columns (from Grossman 2019, 133, 141).
These charts contained only the text and the word-for-word translations. Having them already charted was a huge time saver and also helped because many of the members of these teams are still new computer users. They do not have the technical skills to manipulate the material into charts. We wanted the workshop to focus on their analysis of texts and not on computer skills. All charts were printed out so that analysis took place on paper. (Figure 1 shows charts with some preliminary analysis done by workshop presenters in columns 1, 5, and 6. These columns were blank in participants’ copies.)
Bartsch and I also did some analysis beforehand on at least one story from each language group so that we had prior knowledge of some of the patterns we might find. A system of color-coding helped make various features stand out, such as tracking of participants and quotations and their formulas. The translated Scripture passages were also charted and the storyline laid out on the basis of an analysis by Bartsch.
The workshop consisted of lectures on a series of topics, review of examples in various texts and in some of the stories the teams brought, as well as time to look in their story charts for the features being taught. Topics covered during the workshop included the communication situation, pattern of a story, participant identification, reported speech, literary devices, peak and theme marking, and connectors and other boundary markers such as sentences, paragraphs, storyline components, and episodes.
For the topic communication situation, the translators had been instructed to record when, where, and by whom the story was told and to seek permission to use the story. The purpose for this was to create an awareness of the importance of the context surrounding a story. Some of the stories used personal names, with the assumption that the audience knew who the participants were. This knowledge made us mindful that unknown participants might require a different sort of introduction, as we see in the Mark 5 passage when the man from the graves has to be introduced.
In order to analyze the storyline, participants were to record in column one where various storyline patterns started (see figures). The storyline would become useful later as the location of quotations and the identification of boundary markers were noted. Peak and theme marking would also be analyzed in relation to the storyline, so it was important to do this early in the analysis process. The storyline is based on content, and marks the setting, initiating event, increasing tension, major problems, problem solved, and a closing or summary statement. These storyline events might happen more than once in episodes throughout the text.
Attention was drawn to oral versus written text structures. The style in their written stories might vary from a story told orally. The stories being analyzed were told orally and then written down. The style used in the translation of Scripture could more closely follow an oral style or could follow a more formalized, written style. Participants were encouraged to think about their audiences and the fact that, in their context, most people will be hearing the Scriptures read, so a more oral style might make the Scriptures easier to understand.
As story participants were identified, a series of questions was answered to determine main characters, secondary participants, and props. The translators color-coded the participants using a different color for each referent. Any word that referred to that character was to be colored the same. This allowed for an easier analysis of how often a referent was mentioned, how they were mentioned (name, noun, or pronoun), whether they were given direct speech, etc. Correlations could then be made about how main and secondary participants were introduced, how they were kept track of, and how they might participate in quotations. Any special formulas were to be noted, such as a participant being introduced before or after a verb, who was named, or who was introduced with the most words. Comparisons were made of how a participant was referenced—for example, whether the participant was in control or a victim and when pronouns or null references were used.
Passive and active verb structures were discussed, as well as issues of showing respect and how that might affect translation. African cultures require respect of those who are teachers or who are older. This is often shown in nouns of direct address and is a feature that might not be present in SL texts of Scripture. Passive structures are rare or nonexistent in Liberian languages, so where SL texts use them, RL texts may have to find a different structure. To analyze reported speech, workshop participants were to color-code and label all direct speech. Distinctions were made for indirect and semi-direct speech. Quote formulas were color-coded. All quote formulas were labeled by what was included, such as speaker, listener, and speech verb. Places where no quote formula occurred were labeled “drama.” Reported speech was analyzed for both type and location in the storyline.
Figure 2 shows two pages from a chart that was pre-printed for one of the workshop participants, who, in the course of the workshop, color-coded the quote formulas and quotations and also the story participants. 1 The storyline is marked in the left column, “Pattern of the story.” Connectors are analyzed as well as peak markers. Similar charts and analysis were used for the Mark passages for comparison. It can be seen in the chart of the original-language story how few connectors occur between sentences. In this particular language, translators often used an abundance of connectors in their translation of Mark, particularly the connector glossed “then” (see 15a, column 3, in Figure 2). In this story and one other, it was clear that this connector is limited most often to the “problem solved” section of the storyline. It occurred as a translation of the English word “then” multiple times in the Mark translation. The translators readily removed the many occurrences of this connector, as well as another overused connector they often translated as “and,” when they realized the particular uses of those connectors in their natural stories.
Examples of a story analysis chart (from Grossmann 2019, 154, 156).
The color-coded story participants were easy to spot and were analyzed in tables for the types of structures used to refer to them: name, noun, pronoun, or null. These could be counted for clues about who the main characters were and how and when referents were referred to by noun, name, or pronoun.
Several analysis assignments focused on the peak of a story. The peak is where the most tension is or where the tension is resolved or both. The storyline around the “major problems”/climax or “problems solved”/denouement is where the peak will be found. Bartsch illustrated various features that might be found at the peak and how to identify them. The discourse structure should contain something special to highlight the peak. The same features might also be found elsewhere in the story for different reasons, such as highlighting a theme. Peak-marking features can vary greatly from language to language.
This part of the workshop pointed out many unfamiliar features and some that are difficult to spot. They include parallelism, long–short or short–long descriptions, chiasm or X structure, burger structure (clauses or phrases acting like bookends), and tail–head, tail–tail, and head–head constructions as part of the repetition that slows down the story or marks paragraph boundaries. The head refers to the beginning of a sentence. The tail refers to the end. In tail–head structures the beginning of one sentence repeats something from the end of a sentence that comes before it. In tail–tail structures, the end of one sentence is similar to the end of another. For head–head, the beginnings of several sentences are similar.
Vivid language and “super” words were discussed. Vivid language draws the attention and expresses ideas and events in a way that people will remember. Super words express something as a superlative. African languages contain a type of vivid language called ideophones, which are not present in SL texts but can make an RL text more interesting. Other things could also mark a peak, such as fewer connectors or the presence of reported speech. Peak features were listed in the chart analysis column and were also listed and compared in tables designed for this purpose, as in Table 1.
Comparison of peak-marking features (Grossmann 2019, 309).
In Table 1, we can see that some of the natural peak-marking features are being used in the translation of Mark. This particular team applied parallelism where it was not previously found in their translation of the Mark 4 passage where Jesus commands the wind and waves. They were delighted with the results in that not only were they able to apply their new knowledge in a concrete way, but they liked how the story now flowed, cohered better than before, and was more impactful.
After the first chart was completed, teams moved on to the second chart and focused on boundaries and connectors. Participants were to look for where in the storyline the theme was brought out and what features similar to peak-marking features were used to indicate it. Titles and section heads were discussed here, as they play an important role in the translations.
Boundary markers at the beginning of paragraphs and connectors that occur mid-paragraph were marked and distinguished. They were labeled in the boundaries column as to whether there was no marker, a connecting word, a time word or marker, a location marker, a tail–head construction or other construction that pointed back, a burger structure, a shift in speaker, or a summary statement. The boundary markers were counted and compared with the Mark 5 passage. (Participants did not always maintain the intended distinction between the charts, as Figure 2 shows.)
As the workshop progressed and various features were discovered, the preselected Scripture passages were examined. For example, concerning participant identification, the translators examined how new characters were introduced in their Scripture texts and whether it was done in a similar way to the natural stories. They also investigated whether Jesus was being referred to in a similar manner as the main character in their natural texts, including any necessary markers of respect. To do this, the translated Mark texts also had to be color-coded for participant identification. The same features that were analyzed in the natural stories were also analyzed in their current Mark text charts. Though the Paratext files were ready to use during the workshop, the revisions to the Mark texts mostly happened after the workshop.
Post-workshop review
Because participants were not able to fully revise their Scripture passages during the workshop, a final assignment was given to review the features studied and then to complete their revisions in the Paratext projects designed for the workshop. Several months after the workshop, I met with two of the teams separately to review topics and analysis from the workshop, as most of the participants had not completed the final assignment. Prior to meeting with them, I reviewed their charts and synthesized some of their analysis. Because of time, we focused on boundary markers and connectors, quote formulas, and peak-marking features. I went through their natural texts with them pointing out features that they found in the natural texts and how they compared with the translated texts. They were then given the opportunity to work through their final assignment in Paratext. Only two of the teams are featured in the evaluation and outcomes. The fact that there is little to report on for the third team emphasizes the need for such a follow-up or more time in the workshop to apply new knowledge and revise the translation texts.
Following the post-workshop review, I took each Paratext project and compared the pre- and post-workshop versions. I noted all changes and the rationale the translators reported. In addition to analyzing text changes, I had the opportunity to watch how the teams applied what they learned in actual translation-checking sessions.
Evaluation and outcomes
The story collection process was begun almost a year prior to the workshop. Enough time must be allowed to get stories that meet the criteria. One problem arose because some of the participants viewed me, an outsider to the culture, as the audience. Thus, stories were simplified, summarized, or told differently with more explanation than if they had been told to a fellow member of their culture and language group. Even though I was not present when the stories were told or written down, I had made the request for the stories, so I was the assumed audience, according to some of the participants. Workshop participants could tell their own stories as long as they were told in the third person. Stories that did not meet the criteria were not useable. Having time to evaluate the stories before the workshop proved helpful in weeding out stories that were unsuitable. Without appropriate stories, grammatical features of the language will be absent or skewed. Keeping in mind the differences between oral and written literature is also important. All the stories used here were told orally before being written down. One caution in that regard is that a story told extemporaneously might not use the best language forms, and so conclusions should preferably not be drawn from it for application to a Bible translation. A more formalized oral story may be better. One must also keep in mind that the few texts employed are not going to reveal all there is to know about a language’s literary devices and structures.
In such a workshop, it is helpful to have someone with linguistic training and familiarity with the language(s) present to help look for patterns. An awareness of the basic grammar of each language is also useful. Some of the word-for-word translations were not as accurate as we would have liked, and led to errors in the preliminary analyses of the stories. These had to be resolved in the workshop and the post-workshop meetings.
A further consequence of the stories being written down for the first time and of translators’ low level of previous written literary experience with their own language stories was that the stories came all in one big paragraph. Participants needed more time to process the storyline and determine where paragraphing made sense for their stories before moving into the boundary and connections portion of the workshop. Some features such as connectors and the storyline interact with each other to determine good paragraph boundaries. With most of the information being so new, digesting and synthesizing it all to make a good analysis proved difficult. Knowing the culture also helped determine what part of the story was actually the peak as our outsider determination proved wrong in at least one story. Correcting the misperception also cleared up some questions on the use of several connectors.
Bartsch’s materials are geared for those without much linguistic knowledge. Workshop presenters will need to be aware of the capacities of their students. Knowing the educational level plus an informal inquiry with consultants or team leaders should give the presenters enough understanding to adapt the material to the workshop participants. The process was to teach a topic, apply it to their stories, and then apply the new knowledge to the Bible translation. The charts are simple and were preprinted to save more time for analysis. However, if participants are technologically advanced, more of the preparation of text and charts could be given to them, so they will have even more capacity to do further study. All the charting and analyzing can be done with paper and pencil as well.
The workshop self-evaluation tools revealed two things. First, most participants expressed extreme satisfaction with the learning experience and promised to make use of their new knowledge, and second, the concepts about the topics covered were understood, as demonstrated in their thorough summaries of course materials. The amount of new knowledge that the teams acquired was significant. Prior to this workshop, they only had formal language study in English about English. This was the first time they were given the opportunity to apply literary categories and features to their own languages and stories. It is evident from their evaluations that the workshop was highly effective in increasing awareness of linguistic and literary structures and in increasing their capacity to recognize them. Additional workshops can add to the knowledge base and reinforce what has been learned.
The capacity and confidence of the team members grew throughout the workshop. Prior to this workshop, none of the translators had any experience with this type of analysis. Their abilities to look through the charts and discover features improved as the workshop progressed.
This workshop was concerned with the impact of this discovery process on texts that were already translated and then revised as a result of the new knowledge of RL structures. This made sense for the sake of time and the fact that these particular passages were already translated. Another strategy would have been to have each translator translate their own version of the text prior to the workshop, followed by a post-workshop translation of the same text to compare with the first. This latter version would not be a revision but a new version altogether.
Several key translation issues were resolved as a result of the analysis. As pointed out briefly earlier, in one of the languages it was discovered that connectors are used sparingly throughout natural narrative text. However, in the Bible translation, connectors abounded. When asked why, the translator’s response was that it was in the source text. One connector used abundantly occurred whenever sentence-initial kai occurred in the Greek, often translated “then” in English versions. The translator felt bound to have something. However, the study of the vernacular connector in natural contexts revealed it does not even mean “then” as a time marker or marker that moves a story forward, but is used at the “problem-solved” spot in the storyline. When given permission to remove the excess connectors and place the problem-solving connector in the right place, the translators were both happy and relieved. They went ahead and revised the Mark passages accordingly, as well as other books that were still being drafted. One translator, who previously had only three changes in all four passages, introduced closer to twenty revisions after our discussion and clarification of natural text features.
Other changes made were: reducing the number of nouns for referents, replacing them with appropriate pronouns; replacing a borrowed word with a local term; adding some implicit information for correct collocations; and adding specific tail–head connectors rather than general ones in places where there was a change in time or space, as was found in the natural stories. Vivid language, nouns of direct address, and some particles that do not exist in the SL text were added as translators recognized their necessity and as their confidence increased. Some of the findings from the workshop confirmed areas where the translators already got it right, such as subtitles introducing themes and how new or important characters are introduced in a story.
Overall, such a workshop is useful for producing more natural translations, as evidenced above through the numerous changes made in the select Scriptures and those made during further consultant checking. The benefit to the translators’ skills and confidence cannot be overstated as their knowledge of the literary features of their own RL was hugely expanded. As teams gain capacity, my prediction is that they will continue to be more aware that they do not need to be such slaves to source text structures but will be freed up to use the structures that sound “sweet” to their ears. My hope is that they will continue to expand their repertoire of natural text features as they continue to pay attention to them since they have now been made aware that such features exist and that it is acceptable for their translations to look different from source text structures.
More discovery workshops on other genres will be important. The features discovered here may or may not transfer to another genre type. The analysis skills of the teams should increase as other genres are examined. The same general methods should prove beneficial.
The norm in translation team training for Bible translation is to focus on the structures of the source text. Equal time and effort should go into studying the RL structures. This study should be done sooner in the training process and not after a large corpus of translation drafts has accumulated. We do translators a disservice if they are not trained in RL structures. Or, put positively, we bolster the translation process and translator skills and confidence by training teams in methods to analyze the discourse structures of the RL while learning to incorporate this knowledge into the translation process.
