Google Translator v. Lame-Duck ICEL – COLLECT 27th Sunday

Google Translator has added Latin.  I am skeptical.

That said.

A reader sent this:

Noting your skepticism about Google Translate’s addition of Latin, I have a suggestion: What if we tried a head-to-head comparison of Google Translate to the old ICEL translation? Which would win?

One example:
Ecce Agnus Dei, ecce qui tollit peccata mundi. Beati qui ad cenam Agni vocati sunt.

old ICEL
This is the Lamb of God, who takes away the sins of the world. Happy are those who are called to his supper.

Behold, the Lamb of God, behold he that taketh away the sins of the world. Blessed are they that unto the supper of the Lamb are invited.

Google’s training set for this translator likely involved older, public domain translation, which would explain some of its style. Clearly it’s not perfect or always mellifluous, but still… compared with old ICEL…

Here is what they did with the Collect for the upcoming 27th Sunday of Ordinary Time.

Omnipotens sempiterne Deus, qui abundantia pietatis tuae et merita supplicum excedis et vota, effunde super nos misericordiam tuam, ut dimittas quae conscientia metuit, et adicias quod oratio non praesumit.

Almighty and everlasting God,
who in the abundance of Thy goodness
and exceed the proper and just punishment and vows,
pour it out upon us mercy,
‘so that the consciousness of forgive
what they fear him and will add to what we do not dare to ask.

After removing line break from the Latin:

Almighty and everlasting God, who in the abundance of Thy loving kindness than either we desire or deserve, pour down upon us Thy mercy, that Thou mayest forgive things which our conscience is afraid, and apply what we are not worthy.

Almighty and everlasting God, who in the abundance of Your goodness
surpass both the merits and the prayerful vows of suppliants,
pour forth Your mercy upon us,
so that You set aside those things which our conscience fears,
and apply what our prayer dares not.

your love for us
surpasses all our hopes and desires.
Forgive our failings,
keep us in your peace
and lead us in the way of salvation.


About Fr. John Zuhlsdorf

Fr. Z is the guy who runs this blog. o{]:¬)
This entry was posted in Lighter fare, WDTPRS and tagged , , . Bookmark the permalink.


  1. Fr Matthew says:

    The Google version needs work, but at least it contains most of the important concepts, and it has that lovely “Thy” (like the “taketh” in the Agnus Dei). The ICEL version removes God’s power and everlastingness, and strips us of our consciences, our vows, and our timidity born of awareness of our unworthiness.

    I’d almost rather have a garbled approximation of a translation than a warm-and-fuzzy watered down re-writing.

    Long live the corrected translation! I can’t wait until we can use it!

  2. De Tribulis says:

    Wow. It comes to something when the Google Translate version (with all its faults) has a better claim on the word “translation” than the official rendering, doesn’t it?

  3. kab63 says:

    I, whose knowledge of Latin consists of recognizing “Agnus Dei,” would much prefer moving a few misplaced nouns around. How in the world did they go from “Omnipotens sempiterne Deus” to “Father”?? [Dynamic equivalence, I guess.]

  4. dans0622 says:

    Yes, the google translation is sometimes wrong and at other times is garbled. But, at least it tried and worked from the original. The lame duck “translation”–where did that come from?


  5. Bryan Boyle says:

    In order:

    1. Original
    2. Fr. Z’s slavishly faithful translation
    3. Google
    6. ICEL 197x paraphrase…

  6. Hieronymus says:

    Latin translation is part science, part art.

    Sentences are somewhat like mathematical equations, but due to the fact that word order is not set in stone, and often one has to differentiate between possibilities for forms that look alike, there is a definite art to it. This latter part, a computer program cannot do, but at least it gets the science down (with some confusion in those areas that require human discernment).

    ICEL doesn’t seem to have either the art or the science. We could look at this metaphorically: someone wants a beautiful painting to hang in his living room, and because he wouldn’t hire a true artist with talent, he must either buy a print (Google Translator) or have one of his buddies do it for free because he likes to doodle in his notebook sometimes (ICEL). At least the print looks more like the real thing, even if it is rough around the edges and artificially mass produced. The “real life painting” that our buddy ICEL whipped up looks like the work of a first grader — you can’t tell if he was trying to create a modern art masterpiece or just lacks talent. The only thing that is certain is that his rendering doesn’t look like the real object.

  7. Titus says:

    I think folks are right: Google has better word choice than old-edition ICEL.

    It seems to have quite the time with articles and pronouns. I’m not terribly surprised at that, though. Those are highly contextual in Latin and there aren’t generally (or in this bit) independent words for articles. It would take a startlingly complex translation program to match up those sort of subtleties. That being said, it ought to be possible to get less-garbled translations from Latin, given the full inflection. But I suppose, again, that the computer has no practical way to distinguish between, e.g., the dative and the ablative where they have the same endings.

    Any clues on how Google got “punishment” out of “supplicum”?

  8. pfreddys says:

    Google is definately clumsy but is more reverent than the ICEL.

  9. jatucker says:

    Oh that is just too funny! The lame-duck translation bested by a machine!

    This year of waiting for the new Missal will be very painful – and when you consider that we’ve had forty years of this, much of the rest of our sufferings begin to make sense.

  10. The good thing and the bad thing about Google Translate is that a good deal of it is about word-matching. It ignores case and gender and tense most of the time, even when that could easily be included. People entering “better translations” of a whole block of text, instead of word-to-word, can have Very Bad Consequences, at least in the short run.

    Probably one of the major reasons Google took so long to implement Latin was that Whitaker of Whitaker’s Words, and most of the other Latin translator programmers, were not real willing to give over all their hard work for nothing. I think Google finally stopped cheaping out, but they seem to have gone with even more of the “large block of text supposedly says this” method. Given the (sometimes necessary) liberties that translators take with Latin, and the fact that somebody actually input “Credo” as “We believe”, I suspect this is going to be a Big Pain in the Butt.

    And yet, possibly handier than the alternatives, for many people.

  11. Gregorius says:

    Actually, when I first saw that translations would be based off of existing texts, I feared that people would just upload the old ICEL texts. Thank goodness somebody at Google tried to do their homework.

  12. accat88 says:

    This reminds me of a quote from Jimmy Akin: “The problem with the ICEL is that half of them don’t know English, and the other half don’t know Latin.”

  13. Salvatore_Giuseppe says:

    It never ceases to amaze me that the ICEL could take a prayer, and not only completely change it, but change it into a sentence on the level of a third grader.

    The sentence “Father, your love for us surpasses all our hopes and desires” sounds so childish. Especially compared to “Almighty and everlasting God, who in the abundance of Your goodness
    surpass both the merits and the prayerful vows of suppliants”

  14. Jacob says:

    When I was a child back in the second half of the 1980s, on Saturday mornings there was a TV show called ‘Captain Power and the Soldiers of the Future’ where the title characters fought a machine-dominated civilization out to enslave humanity. It was a live action show meant to sell toys to the little kiddies, but it was also well written and didn’t take its audience for idiots.

    In a special video, the hero declared at the end, “No machine will ever out-think a human.”

    ICEL: 0
    Google: 1
    Humanity: ??? :D

  15. ghp95134 says:

    Jacob: When I was a child back in the second half of the 1980s, on Saturday mornings there was a TV show called ‘Captain Power and the Soldiers of the Future’ …
    I was an infantry captain then and knew NOTHING about this cartoon … until one blissful day, while stationed in Japan, I answered my phone in the usual prescribed military manner: “S3, Captain Power speaking…” The caller broke up laughing; he then explained what was so funny.

    Guy Power

  16. WBBritton says:

    It’s almost as good as a prayer translated by Cranmer. :-)

    Would that we would rediscover “thy, thee, thou, thine” when referring to the Almighty.

  17. Konichiwa says:

    Back in high school I could have done a better translation of the Latin above. Back at that time, I also was a firm believer in that too much smarts made ya stupid. I’m still convinced that it is so. Stupid Google FTW.

  18. shadowlands says:

    ‘so that You set aside those things which our conscience fears,
    and apply what our prayer dares not.’


  19. ar_danziger says:

    It’s always astonishing and mind-boggling to me that the lame-duck translation seems so purposefully anemic compared to a literal translation. Even more so now that it’s plain that even an algorithm can do better. Thanks for bringing such clarity to the language, Fr. Z.

  20. dahveed says:

    Your slavishly accurate translation was best, of course. That the Google translation was so much more accurate than the old ICEL speaks volumes.

  21. James Locke says:

    I went to play with the translation software last night and was pleasantly surprised by the relatively high degree of accuracy that the software had. In fact, it seems better than say, the Spanish translator. I threw some poetry at it and it gave me good prose.

  22. ChadS says:

    I haven’t played with the Google translator yet, but I was wondering what would happen if you put the ICEL renderings we are using and translated it back to Latin. How much different would that Latin look from what it was translated from all those many years ago.

  23. Jacob says:


    I was an infantry captain then and knew NOTHING about this cartoon … until one blissful day, while stationed in Japan, I answered my phone in the usual prescribed military manner: “S3, Captain Power speaking…” The caller broke up laughing; he then explained what was so funny.

    I’ve considered what it would be like to have that as a surname in the past. Having the same rank too would be interesting.

  24. marniebcn says:

    I grew up in the late 70’s early 80’s. Never heard a word about God’s mercy or that I should ask for it. Now I know it was due in great part to the “I’m OK, you’re OK” translation.
    Thankfully He had mercy on me and made me walk into a TLM one day.

  25. Brother Paul Mary says:

    I wonder when they will retranslate some of the breviary as well as the mass.

    Br. Paul

  26. asophist says:

    Makes you wonder if the ICEL folks intended to do something malicious to the liturgical text. Whether or no, the results are the same as if they did. An agnostic machine can express the faith more faithfully than they! – and how many of them were bishops? As was said above, this little comparison speaks volumes. Pray for the old ICEL participants in case any of them have real culpability in the disaster they wrought.

  27. RichR says:

    In these glaring examples of poor translation, I am always reminded of the
    humorous quote from Fr. Peter Stravinskas that got him in some hot water:

    “There are only two problems with ICEL: half of them know no Latin and the other half know no English.”

  28. maskaggs says:

    There HAS to be a lolcats photo that Vincenzo can work up out of this… :)

  29. lofstrr says:

    So… Google is smarter than liberals. :)

  30. Josephus Muris Saliensis says:

    Interesting. At least, to the enquiring mind, it conveys the depth of the meaning, even if the syntax is a bit obtuse (wrong!).

    Amusingly, it did not occur to me to get it to do the translation this way, when I was playing with it last night. I translated a page of our blog into Latin, and that was singularly amusing! I think in web-speak the work is “Fail” !

  31. matt1618 says:

    Folks, we should be charitable at least concerning the intelligence of the old ICEL committee. I’m sure most of them could translate that text as well or better than Google did. But asophist is on to them… ideology was at play – that’s what made the difference. It’s not a question of intelligence, it’s a question of hermeneutics and epistemology.

  32. archambt says:

    Perhaps at the 4 spot above, we should put the Divine Office (i.e. English liturgy of the hours translation). I’m taking a break from the ’62, if only because I’m enjoying the Office of Readings, and have been praying the English version, which is quite nice (and very reverently bound, at least). Anyway, the Collect:

    Almighty, ever-living God,
    whose love surpasses all that we ask or deserve,
    open up for us the treasures of your mercy,
    Forgive us all that weighs on our conscience,
    and grant us more even than we dare to ask.

    I think that’s quite nice. Not slavish, but not anemic.

  33. dcs says:

    Any clues on how Google got “punishment” out of “supplicum”?

    Probably mistaking it for “supplicium”.

  34. Gail F says:

    “and apply what our prayer dares not.” That’s beautiful.

  35. Agnes says:

    So it lead me to ponder…will Google acquire ICEL and we will very soon have Google Mass?

  36. Ben Yanke says:

    Ha! That’s good! Google Mass: Dynamic, traditional translations of both the EF and the OF into all languages! Follow along anywhere in the world!!

  37. Ben Yanke says:

    I almost forgot. Do we have the new translation of the propers of the mass yet? or just the ordinary?

Comments are closed.