Abstract
The Parallel Passages tool in the Bible translation software Paratext is a powerful tool. As with all powerful tools, it can have both positive and negative impacts on a project. Consultants need to be aware of some of the known “issues” that will sometimes result in false positives or false negatives in the highlighting of which parts of the vernacular verse are parallel. Correct usage of the tool will help a team improve the quality of the translation, but incorrectly used it can lead to a lack of naturalness and accuracy because the grammatical features of the vernacular will not allow perfect parallels in certain contexts. In this article, the positive and negative features of the tool will be illustrated with examples from the Apal language in Papua New Guinea. The article will conclude with a list of suggestions for consultants with respect to the Parallel Passages tool.
The Parallel Passages tool in the Bible translation software Paratext is powerful. In 2019, I was looking at a project that has no regular external advisor. I had been asked to make an estimate about how much work it would take for the team to finish the Gospel of Matthew. I did not think that they had even started on the book, but when I opened the project I noticed that most of Matthew had suddenly appeared in the project. There were odd things about the draft, however, in that many of the verses were not under the correct verse number. The material was there, but it needed to be properly formatted. These oddities led me to investigate even further. When I clicked on the Parallel Passages tool, I found that all the parallel passages in Matthew had been copied over from either Mark or Luke and then some of them had been modified where there were major differences. As a consultant who “grew up” in the days before the Parallel Passages tool was developed, I realized that making parallel passages correctly parallel after translating each book independently is a difficult task that requires repeatedly going back through the passages to correct and update things. So, I can empathize with a national translator wanting to just start out with all the parallel passages in the correct form and then fill in the holes, but my first reaction was more like, “What in the world has he done?” My second reaction was, “OK, it is done and I cannot change that. How do I as a consultant evaluate the validity of this approach?”
The Parallel Passages tool can be helpful for a number of different purposes, such as: finding translation errors and inconsistencies between parallels; helping the vernacular translation to reflect parallels in a similar way to those shown by the Language of Wider Communication; helping the vernacular translation to reflect the ways in which parallel texts have influenced each other; and helping the vernacular to reflect quotations and allusions to other biblical books, including, for example, the New Testament’s use of the Old Testament. Which of these is given priority will depend to some extent on the priorities set up for the project when it begins.
In this article, I give a brief review of some of the positive and negative impacts of the Parallel Passages tool on translation projects. Next, I give examples of the types of issues that I have encountered while using the tool. In section 3, I share a few examples that illustrate how the tool has improved the quality of the Apal translation project in Papua New Guinea. Finally, I conclude with a summary of some of the advice I intend to give about the Parallel Passages tool to consultant trainees that I am mentoring.
1. Positive and negative impact
The Parallel Passages tool for earlier versions of Paratext was described briefly by Stephen Pattemore (2011, 89–90, 97–98), and I am fairly certain that almost all consultants and translation project advisors are familiar with the tool and use it. Before going to specific positive and negative impacts of the tool, I would like to raise one red flag concerning an issue about which many consultants may not be aware.
When the Parallel Passages tool moved from one version of Paratext to the next, occasionally parallel passages were dropped. One passage that was missing in Paratext 8 was Exod 21.24/Lev 24.20/Deut 19.21/Matt 5.38—the famous “eye for eye and tooth for tooth” passage. It obviously should be in the Parallel Passages tool. I do not think that anyone would disagree with that. In Paratext 7.5, the parallel passages were all present when one started from Matt 5.38 (Matt 5.38/Exod 21.24/Lev 24.20/Deut 19.21), but the passage was missing in Paratext 8. When Paratext 9 was released in 2020 these passages were finally returned to the list.
1.1 Positive impact
First, here are some of the positive features that are in the Paratext 9 version of the tool. I am sure that most readers are quite familiar with these nice features of the current version of the tool.
The current version of Paratext 9 allows you to filter the passages in a variety of ways so that you can easily focus on the passages that have changed or still need to be corrected/made more parallel.
The current version of Paratext 9 indicates when the translation of a passage has changed and allows you to see what has changed.
The current version of Paratext 9 allows for easier editing of the verse that needs to be corrected by providing a separate editing window.
The current version of Paratext 9 allows for parallels to be approved either by sets or individually, or by a combination of both approaches. Based on my experience, approving by sets could produce less accurate results, but there are situations where it could be useful.
I appreciate greatly the fact that the Paratext team has continued to improve the tool and has obviously listened to translators to find out what would make using Paratext easier.
1.2 Negative impact
Here are a few of the general negative aspects of working with the tool. More specific issues will be discussed in the next section.
It is important to remember that the tool is not perfect yet, i.e., just because a passage is not in the list does not mean that it should not be listed. Users should make good use of the “Give feedback” option to suggest additional parallels.
Teams quickly become tired of going over and over the passages every time they make a change. To prevent frustration, it would be good to have one member of the team preview all the changes and again put check marks in the boxes of the passages where the change does not really impact the parallel nature of the passages, i.e., the change is outside of the scope of the part that is parallel.
Teams sometimes bow to the “pressure” of the computer’s “harassment” and change the passage just so they do not have to look at the passage again, even if it was acceptable in its current form. This can adversely affect naturalness because teams often correct a newly drafted gospel to bring it into agreement with a consultant-checked gospel. Unfortunately, this consultant-checked gospel is often the first book that was translated. All consultants know that the first gospel translated is rarely the most natural.
If the team does not have an exegete who is monitoring the original language (Greek/Hebrew), then the team might easily fall prey to making the translation parallel to the model text translation that they are using. If it is a good model text, then that could be fine, but if it is not, then the benefits of the Parallel Passages tool quickly disappear.
The Parallel Passages tool is having a strong impact on Paratext projects, but sometimes the overall impact is not 100% positive. Despite this, I believe that the tool is worth using and will help teams to produce more accurate translations as long as teams look carefully at the “green” (completely identical) and “yellow” (slight grammatical differences or synonymous expressions) highlighting in the Greek/Hebrew text.
2. Specific issues about which to be aware
Every team that has spent large amounts of time working with the Parallel Passages tool is always looking for the “all green” window in which the Greek is perfectly parallel and the vernacular is perfectly parallel as in Luke 11.23 and Matt 12.30. However, this is rarely the case. Below is a list of things that I have seen happen as the Apal team has used the Parallel Passages tool.
2.1 Problems in the tool itself
2.1.1 Errors in the marking up of the Greek
On very, very rare occasions there are errors in the green and yellow highlighting or in the display of the Greek. One small example is found in Luke 11.10, where square brackets in the middle of a Greek word have dropped out of the display (Figure 1). Exegetes (and translators) should not assume that Paratext is perfect.
Error in display of Greek text.
2.1.2 Parallel Passages tool icon does not have a green check mark
If the vernacular passage is a combined verse and the parallel verse is not the first verse in the span of verses, the Parallel Passages tool icon will not have a green check mark showing that there is a parallel, because the nonparallel first verse is what is listed in the chapter/verse list at the top of the Paratext window. For instance, if vv. 36 and 37 are combined in John 19, the Parallel Passages tool icon will not have a green check mark because the parallel verse is actually John 19.37 (see Figure 2). If you know that the parallel passage exists, you can find it by manually opening the Parallel Passages tool or by manually changing the reference at the top to John 19.37 before clicking on the Parallel Passages icon.
Combined verses (here John 19.36-37) without and with a green check indicating a parallel passage, depending on which verse is entered in the reference boxes.
2.1.3 Viewing changes in verse spans is sometimes a problem
If the vernacular passage is a combined verse and you click to see what has changed since the last time the passage was marked as parallel, you will sometimes get an interesting window in which the program becomes confused and reduplicates most of the verse, highlighting it in red (Figure 3). The differences are highlighted properly, but there is a lot of garbage surrounding the differences and only people who read the language will be able to discern what has happened.
Viewing changes in combined verses.
2.1.4 Confusing choices by the program
Many Paratext users may not be aware of a statement in the Help file about the highlighting of parallels: “Green highlighting of all other versions [besides Hebrew and Greek] is done by a computer algorithm and only approximately indicates identical words or phrases between the parallel passages.” This mechanized markup can produce some puzzling results. Sometimes the program seems to require everything in a phrase/clause to be exactly the same in order for it to be marked green in the vernacular. For instance, in the Apal translation of Matt 22.18-20, I would have expected everything except the object pronouns to be marked with green since the two object pronouns were the only differences in these two sentences, but instead both sentences in both gospels remained unmarked (white; see Figure 4).
Matt 22.18-20, highlighting of vernacular.
In contrast, in the Apal translation of Col 1.20, the program marked several long strings of words as identical even though they most obviously are not identical in the word order (compare the text in the boxes of Figure 5). These complex noun phrases in the less natural ordering that was probably based on the model text, Tok Pisin, are one of the many features of the translation that the national translators have been trying to identify and correct, but the Parallel Passages tool has obviously failed to identify these kinds of issues. However, in this Colossians passage the definite nominative markers (agɨlaŋ) in the Apal translation are left white on both of the complex noun phrases, even though they are identical, so perhaps in that way the program is trying to draw attention to the fact that something, somewhere is not identical. I have seen this happen on multiple occasions.
Col 1.20, highlighting of vernacular.
2.2 False positives and negatives
Many times in Papua New Guinea we know that someone has malaria but the test gives a false negative because they have not tested for the disease at the appropriate time. In a similar way, there are many times that a consultant might look at the Parallel Passages tool and be distracted by false positives and false negatives.
2.2.1 False negatives
If I did not know the Apal language, I as a consultant would have been raising questions such as, “Why aren’t these two obviously parallel verses not completely parallel?” But, since I know the language, I realize that the commands in Matt 8.9 are spoken by the centurion himself, whereas in Luke 7.7-8 the friends of the centurion are speaking on his behalf. In Apal these verses may not look like they are parallel, but they are exactly as parallel as they should be in light of the quote structure in the Apal language. This is a false negative.
2.2.2 False positives
Some of the false positives can be easily explained, but there are others that just leave me confused.
2.2.2.1 Unexplainable false positive
The parallel in Greek between Matt 6.30 and Luke 12.28 includes some grammatical differences that do not have an impact on the way that Apal expresses the same meaning. Apal generally does not use the kinds of rhetorical questions that are seen at the end of the verse. So, it is correct that these two verses should be all green, but when I actually went through checking them word by word, I found to my dismay that there was an extra word in the Apal text, a definite marker, in Luke 12.28 that was not present in Matt 6.30. This definite marker needed to be deleted, but the program was misleading in that a quick glance led me to believe that everything was perfectly parallel. It was not, but I have no explanation for why Paratext did not identify this. So, just because everything is all green does not necessarily mean that everything is exactly parallel. This is a false positive.
2.2.2.2 Explainable false positive
In Apal, the term for “sign” had been revised to a longer term that included the older/shorter term. In the final revisions of the Apal translation of Luke 11.29, the parallel passages Matt 12.39 and Matt 16.4 in the main dialect were completely green. Those two passages had the correct newly revised term that is being used to translate the word “sign.” As a result, both of them were green. If there had been only one newer/longer translation, then the missing phrase in the older/shorter translation of the term for “sign” would have been correctly identified because part of the longer phrase for the word “sign” would have been in white and the contrast would have been easy to see. Instead, the difference was hidden due to the fact that two passages with the newer/longer phrase were identical and thus were correctly completely green, but this meant that the error (old/shorter term used for translating “sign”) was not identified by Paratext (see underlined phrases in Figure 6). This is an example of an explainable false positive.
Highlighting hides omission errors (see underlining).
2.3 The tool is fine, but …
Sometimes the tool is perfectly fine, but when the team works with the tool they can make the vernacular more parallel than the Greek (i.e., intentionally or unintentionally harmonizing the text), and sometimes, even though it is parallel in the Greek, there are linguistic features in the language that mean that the vernacular would be inaccurate if it were as parallel as the Greek.
2.3.1 Vernacular is more parallel than technically required
Sometimes the passages become more parallel than technically required, as can be seen in Matt 23.4/Luke 11.46. The major difference between the two is a change from third plural to second plural, and that will show up in the final verbs in the quotations in Apal not being highlighted in green. There are lots of other smaller differences in Greek, but the concepts are basically the same. As a result, if the translation is parallel except for the person markers, and the passage communicates accurately in Apal, I am quite willing to allow that translation to stand rather than seeking more variety that may not communicate as accurately or naturally. These are the kind of issues about which each team will have to decide and make note of in their translation briefs.
2.3.2 Parallel in Greek, but should not be parallel in the vernacular
2.3.2.1 Deictics
The parallel passages for Matt 26.26 (Mark 14.22; Luke 22.19; 1 Cor 11.24) use the same basic phrase in Greek for “This is my body,” with just a minor change of word order in the 1 Cor 11 passage. All of these are in quotations, so one would think that they should also be parallel in the vernacular, and they were until someone raised the question, “Where was the bread when he said this? Was it in his hand or had it already been handed over to the disciples?” Greek uses “this” for all of the passages, but in Apal, the deictics would be inaccurate if they had been made all the same, which had happened at one point. They finally concluded that if it was in Jesus’ hand it should be the near deictic (1 Cor 11) and if it had already been given to the disciples it should be the mid-deictic, “That is my body” (gospels).
2.3.2.2 Pronouns vs nouns
Col 1.20 and Eph 1.10 used to have very parallel translations of “things in heaven and things on earth,” but in inverted order. Thankfully, in the checking process we finally caught the fact that Apal had introduced a quote structure to express God’s thoughts. As a result, the material in Col 1.20 was part of an embedded quote in which God was speaking about heaven, which is “God’s village” in Apal. God, however, would not refer to himself as God, so that had to be changed to “my,” and since God is located in heaven, the things in heaven needed to be marked with the near deictic rather than the simple definite marker, though the program still, for some unexplainable reason, continues to assert that the definite is parallel to something/somewhere in the verse. Two important changes were necessary to make the Apal accurate and natural.
2.3.3 Parallels in the Western text have been added to the Parallel Passages tool
The Apal translation follows the language of wider communication (Tok Pisin) and often notes the Western text 1 verse in footnotes. These verses used to have to be manually checked because parallels that are only in the Western text were not listed in the Parallel Passages tool in Paratext 8. Normally I would applaud this decision because I generally do not follow the Western text, but I consult with projects that do use the Western text. Many of these verse references have been added to Parallel Passages in Paratext 9 when the verse is not found in the critical text (GRK in Paratext). This means that verses such as Matt 18.11 can be more easily made parallel to Luke 19.10, which only differs by one major term. This will be a major help to projects that are based on the Western text. However, the Western text itself (BYZ in Paratext) does not appear where the Greek normally is in the Parallel Passages tool. So, for these verses there is no Greek text listed (Figure 7; a way to get around this is to display BYZ as a Comparative Text). For projects that are not even aware that there is a Western Text, this new feature could be confusing when users see a blank in the Greek.
Western text (represented by BYZ in Paratext) is listed but not displayed in Parallel Passages.
2.3.4 Multiple dialects in one Bible
As consultants, you will probably never or rarely encounter a project like the Apal Scriptures in which several books are in a different dialect from the rest, but if you do, the Parallel Passages tool will not help you as much as in a normal project. In fact, the Parallel Passages tool might encourage the minority dialect to be more like the majority dialect in discourse structure and vocabulary choices than it should be. Because of this, I have had to be very careful to always emphasize the right of the minority dialect to say things in a different way when the team is looking at the Parallel Passages tool.
2.3.5 When consistency is not a priority
I have encountered projects in which the translation team has insisted that since the gospels are written by different authors, the styles of the books should be completely unique and so they do not need to bother looking at the Parallel Passages tool, because consistency is not a value for that team. While working with teams like that, I still glance at the Parallel Passages tool in the hopes of showing them that the tool can help them catch some glaring error that they might not have caught otherwise. When that happens, I find that the teams are more willing to listen and look at the Parallel Passages tool, though they will never use the tool in the same way that the Apal team has used the tool. For teams that do not put a priority on consistency, I encourage them to add a statement in their translation brief that clearly explains their position on consistency. When they do this, the translation consultant enters the session knowing that the consistency issue cannot be focused upon with this team.
3. Examples of problems found and corrected using Paratext
The Parallel Passages tool has helped the Apal translation team to find and correct errors and make the translation more natural in the process. Below are two examples of parallel passages that have been changed.
3.1 “Eye for eye, tooth for tooth”
In the trial publication of Exod 21.24, the Apal translation could be literally translated as follows:
“‘If/when he puts bad his eye or tooth or hand or leg, you (pl.) should just like that also put bad in exchange the man’s (the one who was the hitter) eye or tooth or hand or leg,’ you should say,” he said.
By the time we translated Matthew, which was the last book the team translated, we were translating many things in a slightly different way. The Parallel Passages tool reminded us that Exod 21.24 and Matt 5.38 should be more parallel than they were in the translation. The problem in the Exodus version was that the exact reciprocal nature of the body parts was not clear. It was such that if you hurt any body part you could in exchange hurt any body part that you chose. Second, while both the LXX and NT Greek lack verbs for the action that affected the body parts, Apal would be more natural if it used a term that was appropriate for the body part rather than using the generic “put bad.” As a result, the current back-translation of that part of Matt 5.38 reads,
“One man hits-splits the eye of another man. You (pl.) should also hit-split his eye in exchange. He knocks loose a tooth and you also should knock loose his tooth in exchange.”
After we saw the more natural translation in Matthew using the Parallel Passages tool, Exodus was revised and is now much more accurate and natural.
3.2 Between the altar and the temple
When the Apal team looked at our translation of Matt 23.35-36 and Luke 11.50-51, we immediately noticed the extra piece of information that had somehow slipped into the translation of the Matthew passage after long discussions of the exact location of Zechariah’s death—the doors of the temple. While the information is probably referentially correct, it was an unnecessary addition. That, however, drew our attention to the reference to the temple in Luke, the first book translated into Apal. Technically it was accurate in that the temple is just called a house in this verse in the Greek, but a quick glance at the translation of this verse in major English translations showed that almost all of them had made this passage “more parallel” than technically necessary. When we investigated more, we found that leaving it as “God’s house” rather than using the term for temple meant that some people were left wondering if this was God’s personal house rather than a spirit house where God is worshipped. Since the Apal translation is for a language group without access to much formal education, the decision was made long ago that we would make things clear rather than leaving them ambiguous. So, the revised versions of these verses are now much more parallel than the Greek is and much more likely to be understood.
4. Suggestions for consultants-in-training
Here is my list of suggestions about the Parallel Passages tool for consultants-in-training that I am mentoring:
If the project is in the early stages, do not encourage translators to use the Parallel Passages tool until after two of the three Synoptic Gospels have been translated independently. By doing this, you allow the team to develop their skills and, in the process, to find better ways of translating passages.
If a team has another consultant-checked gospel, do look at it, but always remember to find out when and by whom it was checked. Also find out whether or not the published gospel has been accepted and used well. If it was not accepted, then you do not want to simply copy a problem translation from one gospel to the next.
If parallel passages are different when they are identical in Greek (all green), have the team read both passages in the vernacular and decide which or which part(s) of each passage is better. In my experience, there are occasions when one gospel is significantly better than the other, but often there are good and better (or bad and worse) choices in each of the gospels. It takes time to sort these out, but having a good, natural translation is worth the time.
Do not assume that advisors and national translators know how to use the tool just because they can put check marks in the proper boxes. Using the tool properly involves identifying the differences, confirming that they should or should not be different based on the source text, and checking the larger context to make sure there was not a reason for the difference.
Encourage teams to use the Parallel Passages tool well before the final read-through. It is not simple to use. In my experience, it can take 1–2 months of solid hard work to go through all the parallel passages just in the gospels, and in the process the team will sometimes create problems rather than solve problems. Some teams may need the help of a consultant if they do not have a team member who can appropriately use the Greek and Hebrew parallels to analyze and correct problems in their translation.
If the translation is one of those rare hybrid forms in which there are multiple dialects in one publishable Paratext project, then the Parallel Passages tool is less helpful and can negatively impact naturalness in the minority dialect.
