A new use for the parallel Latin translations in the Patrologia Graeca

Now that we have a very effective Latin translation in Google translate, it occurs to me that we can also use this to read a great deal of patristic Greek.  For as we all know, the Greek fathers were all translated into Latin at the renaissance and after, and were nearly always printed with parallel Latin translation, right the way down to the 19th century.

The obvious example of this is Migne’s Patrologia Graeca, our standard reference collection of texts.  It’s never been worth transcribing the Latin side.  But maybe now it is, just as a reading aid for those of us without fluent Greek?

This isn’t a new situation, in a way.  Indeed the reason why all these Latin translations even exist at all, is that knowledge of Greek was always rarer than fluency in Latin.  The translations are not always reliable; but something is better than nothing.

On the other hand it won’t be all that easy to OCR the Latin of Migne…

An excerpt from PG volume 78, column 226, a letter of Isidore of Pelusium in the Migne edition.

The low quality of Migne’s printing is something that we have all struggled with.

But there are workarounds.  The last time that I needed to OCR the Latin of Migne, I went and found the edition that he was reprinting on Google Books.  This, needless to say, was far better printed, and created many fewer errors in Finereader 15.

So it is possible, and it’s worth bearing in mind if we need to work with a large patristic text for which no modern translation exists.  Spend some time creating an electronic text of the Latin translation, and push it through Google Translate!

Update (5 Aug 2023): Note that it is actually possible to copy the OCR’d text from Google books itself, for both the Greek and Latin sides in the PG.  Go to the page in question.  Hit the cut-and-paste icon so it goes dark grey, then drag a rectangle over the area that you want to copy the text from. As you release the mouse, a dialogue will pop up, and the text is in the top box. It looks as if its monotonic for Greek. The results are quite respectable.


17 thoughts on “A new use for the parallel Latin translations in the Patrologia Graeca

  1. Obviously, it is good to read Greek. But Migne in Greek is also difficult to make out, as my squinting eyes could tell you.

    And when you are looking for info about topics, or trying hard to find the meaning of a weird Greek word, or making sure that splotch is the same Greek letter you think it is, the Latin certainly helps. Also, if you know Bible Latin well, a “different” phrase is like a beacon.

    This example is talking about “sine modo,” which is not a phrase in the Vulgate. For whatever reason, Matthew 5:22 in Latin just says “if you get angry at your brother,” but the Greek adds “eike,” which means roughly “without a reason” or “without reasonable measure.” (Or maybe “over nothing” or “for nothing” would be better.)

    That is a whole different place to go.

  2. This raises an interesting question for me. If you look at this work of Marius Mercator in Migne (with my very limited Latin) here:

    I see that there is what appears to be the text of Marius and underneath much larger sections with the heading “Castigationes et Notae”. Is the top part the text of Marius and the bottom part some kind of annotations/commentary by Migne’s staff? I know this is a very elementary question but I just don’t know Latin well enough to know for sure :-D.

  3. The link takes us to a random page which is not the text of Marius Mercator but some thesis. I’m on a mobile so can’t scroll easily. But yes, it sounds like it. “Objections and notes”.

  4. I apologize, I was looking at column 63 in that volume. The text I was referring to starts towards the bottom of that page. Thank you for your help with figuring out what “Castigationes et Notae” meant!

  5. I’ve probably been living under a rock or something, but I just realized several months ago that the googlebooks pdfs of PG can be viewed in google in plain text mode. It takes some work, with copy and pasting, and cleaning things up, but both the Greek and Latin texts are there in transcription …

  6. Yes, hit the cut-and-paste icon so it goes dark grey, then drag a rectangle over the area that you want to copy the text from. As you release the mouse, a dialogue will pop up, and the text is in the top box. It looks as if its monotonic for Greek. But very much better than I recall!


  7. The cut-and-paste button produces plain text, apparently OCR’d on the fly, and with no punctuation as far as I recall. But there’s a much better option using the “Plain text” option in the Settings dropdown button (gear icon), which takes you to:
    These pages often have format (italics) and actual polytonic Greek. That depends on the copy: in some all the text is interpreted as Latin (or Greek, sometimes modern), but if you check different copies there’s always one or two with the right combination. For example:
    Although as you see it thinks that some parts of the Greek text are images. Sometimes it fails to separate columns correctly. But with Migne scans you can’t expect it to be perfect.

  8. This post sent me down a rabbit-hole. Today’s LLM-based translations are more accurate than Google Translate, thus I was wondering if anybody had thought of translating the PG/PL with it.

    Here’s an early example of Dionysius’ 6th Letter to Sopatros (with zero fine-tuning or prompt engineering)

    *Original Greek (https://archive.org/stream/Patrologia_Graeca_vol_003/PG003_djvu.txt):*

    Μὴ τοῦτο οἵου νίκην, ἱερξ Σωαίπατρε, τὸ εἰς θρη- Noli hoc putare victoriam , venerande Soslpater, σκείαν # δόξαν ὑδρίσαι, μὴ ἀγαθὴν φαινομένην οὐδὲ — quod in eam religionem vel opinionem inveheris, γὰρ οὐδ᾽ εἰ κεκριµένως αὐτὴν ἐξελέγξεις, ἤδη τὰ — qum non bona videatur; neque enim, si illam recte Σωσιπάτρου καλά” δυνατὸν γὰρ καὶ cb xai ἄλλους — coarguas, continuo Sosipairi res bene habent; ἐν πολλοῖς τοῖς ψεύδεσι καὶ φαινομένοις, ἕν ὃν xal — potest enim lam te quam alios, inter multa falsa et πρύφιον τὸ ἀληθὲς λανθάνειν. Οὐδὲ γὰρ, εἴ τι μὴ apparentia unum illud et occultum et verum 18- ἐρυθρὸν, ἤδη λευχόν” οὐδὲ, εἴ τις μὴ ἵππος, ἐξ ἀνάγ- — tere, Neque enim ut quid rutilum non sit, jam xn ἄνθρωπος. Οὕτω δὲ ποιήσεις, ἐμοὶ πειθόμενος” — continuo album est ; neque, si quis non equus, “8- ἀφέξῃ μὲν τοῦ καθ᾽ ἑτέρων λέγειν, ὑπὲρ ἀληθείας cessario est homo. Sic itaque, si me audies, facito ; δὲ οὕτως ἐρεῖς “5, ὡς πάντη εἶναι ἀνεξέλεγχτα τὰ desistes quidem adversus alios dicere, sed omnino λεγόμενα. B ita pro veritate dices, ut εἰ ea que. dicuntur refelli

    *GPT4 Translation:*

    “Do not consider this a victory, venerable Sosipater, to attack a religion or opinion that does not seem good. For even if you refute it correctly, it does not necessarily mean that things are going well for Sosipater. It’s possible for you and others to miss the one hidden truth among many falsehoods and appearances. Just because something isn’t red doesn’t mean it’s white, and just because something isn’t a horse doesn’t mean it’s a human. So, if you follow my advice, you will cease speaking against others, but you will speak for the truth in such a way that all your statements are irrefutable. Speak for the truth in such a way that the things you say cannot be refuted.”

    *CCEL Translation (https://www.ccel.org/ccel/dionysius/works.html)*

    “Do not imagine this a victory, holy Sopatros, to have denounced73 a devotion, or an opinion, which apparently is not good. For neither—even if you should have convicted it accurately—are the (teachings) of Sopatros consequently good. For it is possible, both that you and others, whilst occupied in many things that are false and apparent, should overlook the true, which is One and hidden. For neither, if anything is not red, is it therefore white, nor if something is not a horse, is it necessarily a man. But thus will you do, if you follow my advice, you will cease indeed to speak
    against others, but will so speak on behalf of truth, that every thing said is altogether unquestionable.”

    The open questions I’m curious about now are:
    1. What the quality would be like after fine-tuning on all the extant English translations?

    2. I only found good OCR’d versions of PG Vol 1-3 so far, other versions on archive.org had too much noise. I’m curious if there are high quality OCR or clean plain-text of more of the PG/PL.

  9. Very interesting, isn’t it? Thank you so much for posting that.

    I’m wary of AI, because it seems to be basically predictive text. But that is a very nice result!

    I don’t know of OCR’d PG/PL. The original print was terrible, you know?

  10. Is there a sense of which parts of the PG/PL have not been translated yet, but have the highest interest?

Leave a Reply