Editor’s note: This is part 2 of a 3-part series. In part 1, we examined the difference between translation quality control for majority languages versus minority languages, and asked whether editorial control over translations should be given to outsiders or to native speakers. In this part, we will explore further evidence that suggests the need for native speakers to have increased control over translation decisions.

In English, only approximately one-third of the grammar and spelling follow commonly understood rules. The rest of the English language requires practice and an almost musical sense of genius to become a well-published author. Over the 31 years I have observed minority Bible translation teams, a consistent observation is that the nationals who are considered excellent translators are often musically-talented individuals, like Martin Luther. In English, best-selling authors often frustrate English professors by bending and breaking grammar rules in order to communicate meaning better. 

Another longitudinal observation I have made of roughly 220 published New Testament language translations across Papua New Guinea, is a high correlation in the success/failure results between two cultural groups of expatriate (expat) translators. The first cultural group of missionary translators, who grew up in Europe or in multilingual settings, and who have been involved in New Testament (NT) translation projects, have had a higher rate of NT purchases and long-term use, and externally observable use, than the expat translators who were from monolingual cultures or families prior to their language classes in junior high and high school. Generally, the translations where the expat came from a monolingual cultural background have been less successfully purchased and used in the minority communities than those where the expat lead translator had a multilingual cultural background prior to both types attending the very same linguistic training classes. Why would that be? Further research is needed. One of my observational hypotheses from 31 years of oral conversation observations with translators and their co-translators is that the translators from monolingual settings or backgrounds are quicker to tell the minority language speakers, “Look, this is how your language works. Here is the rule I have found.” The translators from a multilingual background seem to be more likely to ask their co-translators, “What is a natural way to say that in your language?” and then accept the answer without arguing.

Some New Testaments produced by missionary translators who led the translation project have been well-received and used, which nearly always has led to the request and strong efforts by the people group to produce complete Bibles. Unfortunately, even with the high excitement and insistence by the expat missionaries at Bible dedication celebrations to use the book, after the dedication day, a significant portion of the Scriptures have been poorly purchased and used. Notable is that at some dedications with thousands of minority speakers in attendance, as few as 1-20 Bibles have been sold to the nationals. (In 1982, in PNG, only one Kamano-Kafe mini-Bible sold; Umbu-Ungu sold five in 2005, and Agarabi sold 20 NTs in 2012.) In the Wayne Dye 2002 PNG section of the Scripture-Use Research study, the results were clear that both with the hymnbooks and Bibles, where the editorial control was given to the native speakers, the use and acceptance of the Bibles and hymnbooks showed a remarkable increase in acceptance, sales and use. For example, the research found that in Kamano-Kafe (where one Bible sold) during the same 1975–1985 time frame, hundreds of hymnbooks and story books, all written and mostly edited by the mother tongue speakers, sold. Was it the genre? In 2014, after a re-translation of their NT by a Kamano-Kafe national team who had the editorial authority, 394 NTs sold in 3 days (compared to one copy in 1982), with sales actually increasing on the second and third days. When an expatriate is involved who defers to the mother-tongue speaker, or where the nationals have been given editorial control, a clear finding is that the volume of sales and use increases.

One easy way to increase translation quality and speed is to replace the expatriate typing at the keyboard with a mother-tongue speaker during translation sessions. I have trained dozens of national translators to type, and their initial painfully-slow typing speed in the first weeks and months is comparable to the pain I have felt watching experienced missionary translators sitting behind the keyboard slowing down translation sessions with grammar and language learning discussions. Removing the missionary allows the mother-tongue speakers who desire to give suggestions to improve the text to no longer need to teach and explain every change to the expatriate language learner. Many missionaries greatly enjoy translation sessions because they double as language and linguistic learning opportunities. When a fellow mother-tongue speaker controls the keyboard, the frequency of discussion and time length of each interchange is reduced, along with the resulting crosscultural stress for the nationals, who no longer have to explain to the expatriate their reasoning for each change. 

Corbett and Fikkert, in Table 1 (Participation Continuum by Corbett and Fikkert, 2012, p. 140), have challenged us in missions to consider the continuum in the chart presented below. In my observation, the Bible translation community of missionaries and Bible translation consultants have been both more reluctant and slower than missionaries engaged in other ministries I have observed to apply this chart to our work. If we worked to apply this, could we speed up Bible translation and increase our discipleship efforts by moving from a minority language translation approach to a majority translation approach in many of the translation projects?

Table 1 Participation Continuum by Corbett and Fikkert (2012), p. 140.2 

Who should control the quality check?

Initially, in the 1930s, a quality control process was developed and applied largely by Eugene Nida (Noss) to improve translations in minority languages where there were few or no Christians. This checking process is not well understood by those who have not actually observed or participated in it. What follows is a brief explanation of the process.

Every verse is first translated into the minority language, and then for quality checking (QC) purposes, each verse is translated in writing back into a majority language (e.g. English, Russian, German, Spanish, etc.) to be checked for accuracy, clarity and naturalness. This consultant checking process is commonly known as Vernacular to English (VE) or Back to English (BTE). During this checking and publication approval step, a “Bible Translation Consultant,” who in the vast majority of cases does not know the language, first spends weeks looking over every verse using the VE/BTE. Then the consultant does either a live consultation (face-to-face or Zoom) with the translation team members or corresponds by email. The consultant often also uses the same translation computer program (i.e. Paratext, with its tools like the interlinearizer that makes educated guesses, which are of some help to the non-native speakers). When I have sat with mother-tongue speakers working with tools like an interlinearizer, I observe much head-shaking and smiles at the wrong guesses the computers make. These reactions are similar to my reactions to nonnative English speakers or when Google translates written assembly directions in English. We will do well to remember that these translation guess rendering tools need a full dictionary to increase accuracy before we can just Google translate. In the email check, the Bible translation consultant writes questions and suggested changes to the minority language text. The in-person consultant checks cover between 3-6 chapters each day (the Bible has 1,189 chapters). The consultant may or may not listen to the minority speakers read their language text. He/she then asks questions based on his/her study of the VE to check if the same thing is communicated in the minority language text.

To be very clear, this paper makes no argument that this VE/BTE checking process does not uncover some translation errors, and sometimes many errors, in the text. This uncovering of needed changes is a fulfilling feeling for many missionary Bible Translation consultants who see much value in improving the text. However, for those in the Bible translation movement who are concerned about producing quality translations in minority languages, a more important question we need to examine carefully is: Are there undetected errors still left in the text after using the VE/BTE method, the process which the majority of the Forum of Bible Agencies International (FOBAI) use to approve the text for publication? And if errors are left in the text, what is the quantity and significance of missed errors? More studies like the report in this paper and double blind studies need to be conducted before the FOBAI agency leaders and funders can be assured that the current methods are still appropriate for any languages they sponsor where educated minority church leaders are now present. Notable in answering this question is that the BTE or VE process is never used to improve translation quality of any majority language Bible translation. There is no need, because educated speakers in the majority language can and do the QC directly in their own language text. However, in a minority language needing a Scripture portion, the assumption is too often quickly made that there are not educated speakers, so the QC process needs to be accomplished by the mission personnel, not the native speakers. Bible translation is a people intensive process that takes many years to complete. Are we circumventing God’s call to make disciples by doing the consultant checking ourselves? Finding these mother-tongue Christian leaders usually takes time, work and, above all, humility.

At one point, a missionary from another agency told me that he knew of no real Christians in the language group where we both worked. He was the same missionary that spent a day arguing with a visiting village language informant and local Christian pastor over how to write the language. The informant was paid and excused at the end of that day. I met this same mother-tongue village language informant some months later. At the writing of this paper, he is now a full FOBAI Bible Translation Consultant. He also was the lead translator of a hymnal (with 332 hymns), helped his language translation team by consultant-checking parts of their entire Bible, and he teaches Greek and Hebrew at a local Bible College, and pastors a church as well. 

When I asked a mission leader, “When will the native speakers be allowed to move from the requirement of doing a VE/BTE minority language check method to a majority language check method?” I was told, “When the minority language group can pay for the printing of the book.” The problem is that most of these languages that need Scripture are small, economically disadvantaged people groups that will never have the book sales volume to fully cover the language team of speakers to spend many years to translate and pay for the printing. Others have said, “The person/agency paying should get to test quality and dictate the conditions.” But is there a better way for these minority languages? 

Who should produce the VE that the quality checking is based on? Unfortunately, many VEs are still produced by non-native speakers (the expatriates)! I ran a quick survey on a mission center because my national team members were struggling as native speakers to back-translate back into English or their trade language (Tok Pisin), which are not their native languages. They were struggling to get 25 to 30 verses back-translated in an eight hour time frame. So I started asking fellow expatriate (expat) missionaries how they did their language VE/BTEs. I asked one expat (fluent in English), whose family had served as translators, how many verses a day she could back-translate from the vernacular back into English. She said, “I average about 300 verses back-translated each 8-hour work day.” When I did a survey of 12 expat translation teams, I asked, “Who does your VEs?” Nine out of 12 answered with some form of, “I or my husband do it.” When I asked why and how was that choice made, their answers were often: “Because when we had a national (native speaker) do it, the checking sessions with the expat consultant were slowed down because the nationals on our team don’t know how to write English. When we do it [because we are fluent in English], the consultants can check faster and get more chapters checked each day.” The other three expats expressed similar frustration. One said he had caught his national co-translator just copying from the NIV Bible onto the paper for the VE.

But a question that needs to be asked is: If someone is writing the VE/BTE who is not a ‘fluent speaker’ and who does not dream in the language, but who already knows what the storyline in the VE/BTE language, will they use that knowledge to write their interpretation of what the minority languages text should say for the quality check? Is that an accurate enough quality check? The reason to ask and do further research is that some of the VEs I have checked in several languages produced by expats clearly followed the English storyline just like the national who copied from the NIV rather than the minority language they were supposed to reflect. When I checked with careful language speakers who back-translated them again, even with careful follow up questioning, we found as much as one third of some ideas and concepts in the translated and published local language Bible verses were missing. The written VE verses were not an accurate reflection of the published local language Bible verses. These missing concepts were problems in VEs produced by expats. How widespread has this practice been? 

Every time the BTE/VE process is tested in minority language groups, the research and feedback shows that many glaring errors are still missed in this consultant-checking process after the Bible translation consultant finishes checking and approves the text for publication. This happens even after the changes suggested in the checking session are made. The errors missed in the QC process in the texts tested have been even more extensive when a non-native speaker (expat) produced the VE/BTE document on which the quality check was based. For example, when I asked a long-term local Bible translator about the quality of the published translation that he had helped an expatriate produce (where the wife of that expat produced the VE/BTE), he quietly said, “The pastors and people who try to read our own language tell me, ‘It feels like we are lost in our own forest.’” After further, gentle questioning, it was clear that the people group wanted their Scriptures (which had sold only five copies at their NT dedication) to be re-translated so that it would read like their language, not a made-up form of it. They are waiting for the expat translator to retire (and now die) so it can be retranslated, because they do not want to offend the expat translator (personal communication).

In another minority language translation, I found the following errors, as well as many missing key points and an unnatural translation. The first error brought to my attention was through a polite question from a native speaker, “Is it supposed to say, ‘What shall a man give Satan for his soul?’” (See Mark 8:37: “What shall a man give in exchange for his soul?”) The VE had been written by an expat and looked just fine. After further investigation, hundreds of errors were found. Some examples follow: in the Ten Commandments, their translation stated, “Do not get divorced” rather than “Do not commit adultery.” When Lazarus died, the text said that Jesus died rather than Lazarus; in the instruction to circumcise, it said, “Circumcise your boys 8 times”; and in 1 Samuel, David called Saul “the God” instead of “my lord.”

Since I was tasked to figure out why the people in this language group were not reading their Bibles, I read the original English VEs, which were used in the original checking sessions. I saw that even though nothing was missing in the English backtranslations (which had been checked by expat senior Bible translation consultants), the published vernacular translation had at times an average of one-third of the information contained in the good-looking VEs missing in the minority language verses. The VEs being used to check the translation were an inaccurate interpretation of the actual published text! Furthermore, all of these VEs were written by the expat translator, not by a native language speaker.

In 2018, 200 verses of that original 1982 translation were checked by two consultants, a national church leader who was also a native speaker of the language and an expat senior Bible translation consultant. The native speaker checker had become a full consultant, and had been trained by the mission with its ever-increasing requirements, thus this native speaker had more education than the average expat missionary as a Bible Translation Consultant (who had made different, more stringent rules for nationals to qualify than for the expats to be approved to become full consultants). The national (native-speaker) consultant found 25 significant and heretical errors in those 200 verses, and many missing points. He said, “I would not ever approve or allow my people to read this text.”

This same text had been approved by expat consultants for his people group previously. The mission agency had assumed that their quality checks by different consultants were satisfactory. The people group was labeled by the mission agency as “not spiritually hungry.” The truth was that the translation quality was poor. This national full-consultant could not understand how the old text passed consultant checks done by multiple expat consultants. He came to the conclusion from this examination of the old VEs that only mother-tongue speakers should be allowed to produce VEs for consultant checks. The expats, while not copying from the Bible like some nationals have done, are assuming words and phrases mean things they don’t, and thus the expats are clearly producing inaccurate VEs. This national full-consultant has seen expats mistakenly back-translating meaning in his current checks in other languages as well. He has made comments like, “The [expat] missionary forced me to approve the books for publication. Yet it is clear that the nationals helping him are not happy and will not read their Bible after my check is finished.”

Furthermore, he felt that the expat translators should not be the ones reading aloud during the consultant checking sessions. He thought that the minority speakers stumbled in their reading in poorly translated texts, while the expat decoded without needing the meaning to pronounce the words. Thus, a community listening to someone read who is not from their language does not promote healthy quality checks. 

The two consultants (one a native speaker and one an expat senior consultant) sat together and checked the same 200 verses that had been re-translated and re-checked using a new method based on how majority languages are checked and approved. This method is currently called the Own Language Checking Method, using not just one, but two, national church leaders from separate denominations, who have had some Bible school and consultant training, and who are fluent in their own language, to check and approve each verse without a VE/BTE. Instead, they study the minority language text just as is done in any majority language directly in the language to be published, while consulting Bible resources in a regional majority language. For this test, a VE was produced for the non-native expat senior consultant to do a quality check (a double check). He found one error in 200 verses, and wrote:

I asked many questions and I was very satisfied with the responses. Either the question proved to be irrelevant, being an issue with the VE back translation, or it was a translation issue the team was well aware of and had already discussed a lot during drafting. Only once did it seem there was a minor issue. In 1 Kings 8:55, Solomon blesses the people with a ‘loud’ voice. The word ‘loud’ was missing.

After the expat senior Bible translation consultant left, one of the national translators said, “We don’t need the word ‘loud’ in the text because in our language, when someone is speaking to ‘all the people’, it is already understood that they will speak loudly. It will make our text unnatural to add the word ‘loud’.” 

Many have told me that the consultant-checking of today is superior to that done 15 and 30 years ago, so even revisions of old Bibles would benefit from new consultant checks. The following two examples show that the consultant checking process used today is still inadequate.

This Own Language Checking Method, or majority language checking process, was first investigated and tried out after I, as an advisor, had worked with the Kamano-Kafe team to retranslate books in their New Testament. We had done several consultant checks with traditional VEs, yet I continued to have pastors who were speakers of Kamano-Kafe returning to me after reading our work and showing us errors that they were finding, and they felt that improvement in the text was still needed. It was after that happened a half dozen times that I came to the conclusion that we needed to find a way to transfer the checks on this minority language to a majority language checking approach. That process became known as the “Own Language Checkers only” approach. A total of 72% of the retranslated New Testament was re-consultant checked with VE/BTEs before having the remaining 28% checked and approved by two trained fluent Kamano-Kafe speakers. We later rechecked the other 72% because of the higher quality of checks by the own language checkers, who found numerous errors left by the expat consultant-checkers for whom we had previously produced the best VEs we could. The national team said, “Never make us go back to the old way of checking!” Just one example of the difference is that one of the weaker native-speaker consultants checked and noted, “This glottal here in this position gives the meaning that Jesus’ parents climbed up the stairs that were inside the temple, like a house. Remove the glottal so they will go up to the temple.” That problem was totally missed with a back-translation approach.

In another language, the original translation was done by an expat with national language informants as co-translators. That translation was not used by the people. After the translator retired, a mission-led team thought that the orthography was difficult, and applied a simplified orthography to the text using an Adapt-it program. That process allowed more nationals to read the text. Once they could read the text, it was discovered that the text had translation errors and was in need of a re-translation. A national team was formed who drafted multiple Old Testament books, and then a Gospel, and had some of them consultant-checked in 2017. In 2019, the team was struggling because their lead national translator was recruited away by another mission to supervise their work. To jump-start the team after months of almost no progress, as a test case, in late 2019, we added an expat English advisor checker to their team and trained national pastors to be their own-language consultants.

After weeks of team-checking each verse with the expat advisor, we then gave the verses to the three trained own-language consultants. At the end of team checking and revising Matthew, I was sitting in the room as we finished Matthew. The man then leading the team said, “Let’s check Mark now.” I stopped and said, “Hasn’t Mark already been consultant-checked with a VE by a consultant last year? Aren’t we going to skip Mark and do Luke next?” The lead national translator said, “No, we need you to advisor-check Mark for us again.”

What ensued was the discovery of mistake after mistake in the whole book of Mark, which had already been consultant-checked by the VE method. Because I was busy, two other expats who were serving on other projects stepped in and helped check chapters. At one point I asked the three of us to each pick at random 10 verses and unobtrusively count the errors that we asked the team to change, either because something was missing, added or unclear. In 30 verses checked we changed 32 things that were missing or had the wrong meaning from the text which was already consultant-checked text in 2017. The English VE and the former consultant check had found items to improve, and that consultant had felt the text was ready for publication, but it clearly needed much more improvement, given that at least 32 more errors were left in just 30 verses of text after the consultant check. These corrections were applied in our advisor check and then it was sent to the church leader checkers for more checking and improvement, rather than those 32 errors going to publication.

A red flag for agencies and funders depending on a VE/BTE check is that a fully trained consultant had missed 32 translation mistakes in just 30 verses. The lead translator had realized the VE check had missed items and wanted help. These examples demonstrate that the VE checking process is a significant reason why many Bibles are rejected in minority languages. Therefore, accepting the VE checking process as a standard of excellence or it being upheld as the only or even primary method to check Scripture should be further researched in multiple languages or abandoned.

Should the VE check only be used in minority languages for the first few hundred Bible verses while we find, recruit, and train native speakers to check accuracy and naturalness?

An area needing more research is that there seem to be two different types of errors that are more common for a national-led team and expat-led team to make. The errors generated in translations led by expats who serve as the translation editor, lead translator, and/or back-translator, are the most severe, with issues like our team found (e.g. circumcising your boys 8 times), while translation errors introduced into the text by nationals who are not fluent in the source text language(s) more often miss the meaning of different words such as “like” and “as.” (For example, “they heard the sound like marching in the treetops” is translated as “they heard marching in the treetops” in 2 Samuel 5:24.) National-led teams are more likely to miss or leave out a concept due to inability to understand the source language, while an expat-led team will add it in an unnatural way. One clear finding based on 31 years of observation in many languages is that the need for full re-translation is much greater for Bibles that were expat-led translations, whereas those led by national translators can more often be revised with minor corrections.

This is part 2 of a 3-part series. The final part will conclude with further suggestions on how to empower native speakers in Bible translation.