Featured Artifical Intelligence Translation?

One thing I do as both a Bible college professor (Greek, Church History, missions, theology, etc.) and Bible translator is mentor. AI will never be able to mentor a real live student who is struggling.

I have mentored students who were having trouble in class (dyslexic, slow learning, etc.), students whose theology differed from the college, who wanted to quit, who were just afraid of Greek, who thought they knew more than the prof, etc., etc. AI will never be able to do any of these, because they are human responsibilities depending on human abilities. Many times they are spiritual activities, and an AI program will never have a spirit, or know how to pray with a student.

In the area of Bible translation, I have mentored African translators there, and American and German translators here. I have advised a Bible translator in the Middle East over the phone how to translate from Hebrew into his language. What AI program could do this?

For the record our college and seminary do not even have online learning options because that would mean we could not mentor very well. I've taught by Zoom during Covid, and it was tough. I wouldn't want to do it again. Very hard to mentor online.

Some years ago when Paige Patterson was president of Southwestern BTS, I did some research there and had the privilege of fellowshipping with Dr. Patterson. He told me, "There is one department here that is doing quite well, but I'm disappointed in that." When I asked him why, he said, "It's the online program, and we can't mentor students online." Likewise, an AI "prof" could not mentor students.

Our president likes to say, "You can't produce special forces soldiers online." And AI is never going to mentor and produce superior graduates.

 

No, it is an extremely difficult and costly human effort. There are various Bible translation tools out there, but none yet that can do anything like what a translator does in an unreached tribe: live with the tribe in very primitive conditions, learn the language from scratch, invent an alphabet, learn the nuances from the native speakers, mentor tribal helpers, communicate the source languages to those helpers, etc., etc. Nor will there ever be an AI program that can do that.

 

This post is confusing. How am I "blending" missionary Bible translation with some other kind of translation? Do you not include pioneer Bible translation with new (unneeded) translations into English or another Indo-European language? Is missionary Bible translation somehow not Bible translation??

To me the cutting edge of Bible translation is missionary Bible translation. I'm always disappointed when I hear of another unneeded English translation or revision and all the money spent on it when there are still 3000 languages in the world without any Bible in their language.

Concerning AI translation, an AI program cannot do that on its own, but must have access to translation software. The early programs were lousy. I had a Japanese-English program and soon gave up on it. It used to be that Google Translate came up with bizarre renderings, but it has greatly improved (though I would not use it to do a Bible translation).

The catch is that there are many, many languages that Google cannot access. For AI to translate into such languages humans would have to do linguistic analysis, something far beyond an AI translation program.

 

I know someone will say that proofreading could easily be done by an AI program, and that may be true. However, proofreading a Bible translation is a much more complicated process than most people realize. At the present time, I know of no programs capable of proofreading a Japanese translation, although there are some functions in Microsoft Word that are helpful.

I'm going to attach to this post a document we had to use to mentor and train our Japanese proofreaders in the hope that this complicated process will become more clear. You'll be interested to read the comments of two of our proofers at the beginning of the document.

 

I believe the public is losing confidence in so-called AI. Various news sites are reporting that ChatGPT lost ten per cent of its users in one month. I remain skeptical of the whole AI thing, myself. As I've pointed out, it cannot mentor.

Another thing AI would have trouble with is non-digital research. Our faculty is mildly worried that students might use AI websites like ChatGPT to write a research paper. Van referenced this idea obliquely:

Personally, I'm not worried. Think of a paper for "Survey of Church History." I require at least five pages, with at least four sources in addition to the textbook, one of which must be from a peer-reviewed journal. (Most of them are behind a pay wall. Does ChatGPT have money for those subscriptions? ) Any Internet source used must be by a genuine scholar, and may not be anonymous (so much for Wiki!).

One option is to write a paper about "the history and beliefs of a non-Catholic group up to AD 1,500. How would an AI program handle that? I don't see how it possibly could. The mental processes needed are way beyond the capabilities of AI as it exists today.

 

Irrelevant. This is a Christian website. Bible translation and teaching Greek (or whatever else you referenced) are ministries, not jobs.

I did address this, but apparently not in the way you wanted. So please tell us, what do you mean by "cottage industry"?

Is this what you mean: "A cottage industry is a small business that is run from someone's home, especially one that involves a craft such as knitting or pottery." (Cottage industry definition in American English | Collins English Dictionary)

If that is what you mean, nobody gets rich by translating the Bible. It's not really an industry or a business.

Of course they can't. I've backed up what I said. So that refutes you.

That's my expert opinion. But since there has never been an AI Bible translation, it's a moot point.

Anyway, you never addressed my detailed points that professors, missionaries, and Bible translators must mentor, and AI cannot mentor. That to me is the most cogent point I've made, but you ignored it.

I don't believe that AI "can replace the need for many of the sit in their cubicle translators." But I'm not even sure what "sit in their cubicle" means. I've never known one of that kind of translator. Please explain what you mean.

You think AI can replace translators. Please prove it.

 

Here are some experts that agree with me that AI cannot actually translate the Bible, but can be a help:
How AI is helping to translate the Bible into rare languages

From the website:

This the Catholic News Agency, but they are referencing a ministry effort supported by Wycliffe. Follow the link to the "Greek Room" website.

 

Here are other experts that agree with me: Artificial Intelligence Bible Translation: Will it Prove Harmful or Helpful? - CBN Israel

Note this statement from the article:

So, Van, please find an article by Bible translators or experts in the area that agree with you.

 

Why AI still doesn’t have consciousness?
https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cit2.12035

 

I haven't "disparaged" AI--just doubted the hype.

And my way of life as a professor and Bible translator is not threatened in the slightest. Not worried at all!!

Yep.

Again, it's not a job, it's a ministry. I've never heard of a translator living by the royalties of a Bible translation. Doesn't happen. I certainly have not received a single penny or 円 (yen) for my Bible translating, though I have invested many thousands of hours in that ministry.

Who's ranting and railing? I'm certainly not, as any unbiased witness can see. I'm simply discussing the issue with you.

And I'm not opposed to AI. Have not said so, and it's not true. It will probably be a useful tool for translators, but will not replace them. I simply don't believe all of the hype.

I'm simply trying to discuss the issue. And you still have not answered my point about professors and translators being mentors, something

Thank you. Now please explain how Bible translation can be a cottage industry--and how it can be considered an actual industry for that matter, rather than a ministry as I have said.

Also, please explain how an AI program can be a mentor, something professors and Bible translators must do, but I believe AI cannot do.

 

Folks, I'm going to post the style sheet of our NT translation effort. I have no idea how an IT program can put together a translation style sheet. Perhaps Van can tell me.

Footnotes are at the bottom of each post. (BB will not provide links, though the numbers are in blue.)


Lifeline Japanese New Testament Style Sheet

INTRODUCTION: We are using a literal, word-for-word method to produce a Japanese NT that is true to the original while having good readability in Japanese. If need be, though, we will sacrifice readability for accuracy. Our Greek New Testament for the source document of the translation is Scrivener's edition of the Textus Receptus Greek New Testament (伝承本文ギリシャ語新約聖書), with reference to the KJV. This will be the first time this Greek text has been used for a modern colloquial Japanese New Testament, though the Nagai Version (永井訳, or 新契約聖書) used it for a Classical Japanese version.

I. Translation Theory
A. The theoretical basis for our translation is Optimal Equivalence as delineated by Hebrew scholar Dr. James Price: “a theory of translation that focuses on the equivalence of words, kernel clauses, transformations, and literary form.” [1]
B. Linguistically, this means that transformational grammar and relevance theory have influenced us.
C. We also are influenced by the theories of translation studies scholar Lawrence Venuti, as delineated in his books.[2] He teaches that normally a translation will be either foreignizing or domesticating, with foreignizing being better: “a TT [target text] is produced which deliberately breaks target conventions by retaining something of the foreignness of the original.”[3]
D. Therefore, though we strive for good readability in the target language of Japanese, we are not afraid for our renderings to seem a little awkward if that is necessary for faithfulness to the original languages. Faithfulness to the source text (authorial intent) is more important than specific readability issues.
E. Our Skopos (translation goal) is to produce a good, readable Japanese Bible which faithfully represents the received texts of the original languages: the Textus Receptus (伝承本文) Greek New Testament edited by Scrivener, and the Masoretic Text of the Hebrew Old Testament.

ΙΙ. Syntax
A. Wherever possible, we translate according to the form in the original language: verb for verb, noun for noun, adjective for adjective, etc. This is often not possible, such as when the target language does not have a definite article like Greek does. Also, certain particles have no semantic equivalent, but influence syntax.
B. We do not feel it necessary to reproduce the Greek word order in most cases.
C. We aim first of all for accuracy, but we do want readability. In cases where one must choose accuracy or readability, we choose accuracy.
D. Due to the nature of the two languages, Greek and Japanese, we realize it is not always possible to match one Greek sentence with one Japanese sentence. For example, Japanese can have a causal statement in two sentences, whereas Greek usually has cause and effect in one sentence. Again, sometimes in the writings of Paul, one sentence may last for several verses. In such a case we may break up the Greek sentence into two or more Japanese sentences.

[1] James Price, A Theory for Biblical Translation: An Optimal Equivalence Model (Lewiston, NY: Edwin Mellen Press, 2008), 336.
[2] Especially The Translator’s Invisibility and The Scandals of Translation.
[3] Mark Shuttleworth and Moira Cowrie, Dictionary of Translation Studies (New York: Routledge, 1997), 59.

 

ΙΙΙ. Semantics
A. We do not add words just for clarity that are not in the original text unless absolutely necessary. That would be dynamic equivalence, so-called "meaning based translation," where the response of the reader trumps the meaning given by the divine and human authors. Example: We do not add 偶像 to バーアルfor clarification in Rom. 11:4, as some translations do. If need be here, 偶像 can be put in a footnote.
B. We often translate by concordance (every occurrence in the target language the same rendering for one Greek word). However context overrules simple concordance in many cases. So for example, where the KJV has “church” in Acts 19, we translate 集会 and 会議 for ἐκκλησία (ekklesia, "assembly") instead of 教会.
C. We aim for a literary style rather than a colloquial style. However, on occasion we may use a colloquialism when the original warrants it for meaning, or to produce the right impact.
D. For clarity in Japanese, we often use proper names in place of pronouns in the Greek.
E. We keep the original language weights and measures, with modern meanings in footnotes. Putting modern weights and measures into a translation is called anachronism. Keeping the original weights and measures is another example of foreignization, and helps the reader remember the cultural context of the original. Christ did not live in the 20th century!
F. Biblical Greek very often uses a conjunction at the beginning of a sentence, especially και (“and”), and δε (“but” or “and”). We retain these in the target language to avoid losing data from the source text.

 

IV. General Principles of Style
A. In the narrative books (Gospels, Acts, Revelation) we use the literary form, である.
B. In the epistles we use です、ーます.
C. Quotes from the OT are given in である form.
D. We use たち for the plural unless another form is clearly indicated, such as 人々 instead of 人たち. This is because it appears to be a more common, modern usage in colloquial Japanese.
E. We translate what is specific in the original or easily inferred. We do not use italics except for phrases, since that is not a native Japanese convention and can be misunderstood. When we do use italics we have a footnote to explain.
F. If the original is ambiguous, we try to preserve the ambiguity in the target language.
G. In translating idioms, we first aim for the same idiom in Japanese. If Japanese does not have the same idiom, we translate literally. If a literal translation does not have the same meaning in Japanese as the original idiom had, we look as much as possible for an exact equivalent idiom in Japanese.
H. If the literal meaning sounds a little strange in Japanese, but to change it would alter the nuance of the original, we keep the odd wording when at all possible. This process is in accordance with the method called "foreignization" by noted secular translation theorist Lawrence Venuti.

 

V. Honorifics Style
A. We use plain language for New Testament quotes of the Old Testament: だ, etc.
B. We have Christ normally speaking to others with polite language: です and ―ます. We use the polite forms for the words of Christ to show that He spoke with love.
1. For verbs referencing Jesus, we use the passive verbal honorific in Japanese.
2. We use honorifics for every mention of God.
C. We use plain language for when Christ speaks to demons, Satan, and sometimes human opponents: だ, etc. We do not believe Christ was polite to Satan and demons, or to the Pharisees when he called them serpents and vipers (Matt. 23:33).
D. We refer to God (including Christ) with honorifics.
E. In our version, some people do not use polite language when speaking to Christ.
F. We use normal polite language for usual conversations.

VI. Pronouns
A. Japanese often avoids the use of pronouns. If the pronoun is in the verb form in Greek, but the target rendering is not ambiguous, we may avoid the use of pronouns in our translation. In Greek, pronouns are represented in the verb endings, but there are also non-verbal pronouns.
B. We used Japanese pronouns in our translation if there is a non-verbal pronoun in the Greek; that particular Greek construction indicates emphasis.
C. Sometimes pronouns can be avoided by using その人, その方, etc.

 

VII. Grammar
A. We stay as close as possible to the tenses of the original languages. Exceptions include the historical present tense of Greek, which should usually be translated with a past tense in Japanese. However, Greek has several past tenses with differing verbal aspects: aorist, imperfect, perfect, pluperfect. We try to represent those aspects when possible, but it is often not possible, especially with the Greek perfect tense.
B. When possible, we use the Japanese passive for Greek passives. However, this is not always an option because of special usages of the Japanese passive form for honorific, negative emphasis, etc.