What Happens When No One Can Read It Anymore?

What Happens When No One Can Read It Anymore?

Jeffrey D. Shaffer


History has been handed down to us in the form of oral tradition and written accounts of events, thoughts, and even stories. Around the world there exists a large body of works written in ancient languages. Over the last century or so, many of these great works have been translated into modern languages, but the number of people who are able to read them, let alone translate them, is shrinking rapidly. Upon a time, Western students were taught Latin and French and some went on to learn a plethora of other languages such as Old English, Middle French, etc. Many also learned Hebrew and Koine Greek, at least some of these, in order to read the Bibles in the original.

Now, however, we tend to gravitate towards translations that feel “modern” and are “easy to read”. This can be seen quite clearly in book sales, but also in academic discussion (i.e., articles, research papers, etc.). We discuss things more and more as if they were and always have been written in Modern English*, we quote our favorite works (in translation), and we even sometimes argue over writing styles and word-choice within these translations, errantly assuming we’re arguing over the style and words of the original.

I, like many others, am greatly thankful for modern translations and read most of the works that interest me in translation, though I begin to worry:

What happens when no one is left who can read the originals anymore?

If no one is able to read “Beowulf” or the “Divine Comedy” in the original, we have only the translations to rely upon, and translations, unfortunately, are imperfect. This means that the true impact of the original, the true meaning, thoughts, and feelings incased within the original will be lost, forever.


“Just wait until Machine Translation really takes off,” some might say, “Soon, AI will be able to translate everything, quickly and accurately.” Others might say, “There will always be at least SOMEONE who has studied these languages and can read them.” And most lightly many would say, “So what? Translations are close enough.”

These three positions sound good, and might even appear logical at some level, but they are neither accurate nor true.

Machine Translation
Machine Translation, or “the power of AI” as it is being touted these days, is far from able to simply look at any text and translate it. An AI machine translator needs a lot of input and then it also needs some method of assessing the produced output — some type of feedback that tells the AI what is good and what is inaccurate. For modern languages like English, Chinese, or French (to pick a random few), there exists a huge body of digitized texts, that can be fed into the AI translator, and many modern translations of said works can then be used to provide feedback. This methodology plus some feedback by bilingual users are basically how systems like Google Translate work today. However, what happens when you do not have a large body of input and few, if any, translations or bilinguals are available to provide feedback? The system will then struggle to create accurate output.

Currently, AI translators like Google Translate also still struggle between languages like English and Japanese because they are so fundamentally different, despite having vast numbers of digitized texts to work with. Japanese to English produces some of the worst results as Japanese often omits the subject of a sentence, but English typically requires one. So, Google Translate “guesses” what the appropriate subject might be. Thus, we end up with translations that say things like, “I was hungry. So, she ate an egg.”

Another problem with waiting for the development of Machine Translation is also the availability of digitized texts. All ancient texts were handwritten, and many of the originals (or even very old copies of the originals) have deteriorated to where they are neigh on unintelligible. Scholars in these ancient languages struggle to decipher letters and words based on “similar passages”, “the author’s writing style”, and “similar words that might fit in the given context.” It’s a type of linguistic puzzle that is only solved, if it can be solved, by those deep in the language and literature of the time. Corrupt input, or even unreadable input (as computers still, today, struggle to read handwritten words from a page) immediately put an end to any chance of Machine Translation.

There Will Always Be Someone Who Can Read
Again, some might say that there will always be someone who has taken a personal interest in an old language and old tests and have taught themselves how to read. While this might be true, as the numbers of able-readers decrease and general interest also decreases, the number of available “courses” and “learning resources” begin to disappear. A quick Google Search turned up less than a half-dozen courses available in “Gothic Language”, world-wide, though to be fair, a brief “introduction” to the language appears in several language-related history courses.

Certainly, a motivated learner can pick up an ancient language on their own with the right resources, but for some languages these are still hard to find. More problematic than this is the fact that most people lack the grammatical rigor required to learn a new language on their own, especially from a book. For example, there are many books on how to learn Old English and Old Norse, and even a few on Gothic, though most take the same approach of laying out the pronunciation rules, then giving a short section explaining the grammatical rules, and then a “reader” where the learner practices reading. The student is expected to provide their own dictionary in most cases.

Without a solid foundation in “grammar”, as it is seen used as a fundamental language construct, grammar as it exists in all languages, reading to understand how grammar is used in a new language simply by reading about it, is a hard challenge for most to overcome. For some, that challenge is exciting and enjoyable, though for most, a classroom setting with a knowledgable teacher who is able to help the learner and answer questions is an easier approach.

Translations Are Good Enough
Here we come to the final proposal, the one that is perhaps most commonly held — translations are good enough. Yes, people still enjoy Dante’s “Divine Comedy” after hundreds of years, we still get excited reading about the Norse Gods, and pretty much every Christian reads the Bible in translation. So, isn’t translation good enough?

The short answer is NO. Translation is NOT good enough. A good translation is helpful and is ideal for gaining “facts” and “information”, but translations are typically lacking when they come to accurately communicating the feelings of the original, and they are terrible at maintaining the “style” of the original.

By “style” here I mean the deliberate word-choice, stress patterns, and rhyming patterns the author used. A well-known translation problem is clearly seen in the translated works of Shakespeare. Shakespeare was a master not only of stress patterns (iambic pentameter, etc.), but also a master of double-entendre — words and phrases with double-meaning. When translated into modern English or other languages, the double-meanings are almost always lost.

Another great loss in style is often the deliberate stress patterns the author employed, such as those found in poetry. This is most clearly seen in translations of Japanese haiku where there is not only a required “seasonal word” used to set the mood, but a strict 5-7-5 stress pattern of syllables. Many translators have found it nearly impossible to maintain both, and often elect to maintain the “feeling” and “mood” of the poem while losing the 5-7-5 pattern. This same problem is detrimental to Old English and Old Norse who maintain a strict use of alliteration and pattern-repetitions. Indeed, one of the greatest charms of Old Norse is the way each short phrase carries a small, but hefty punch in feeling and meaning, accentuated by a bounce of alliteration. A few translators, such as JRR Tolkien, have tried to recreate the alliterative and stress patterns within their own translations of these great works. It’s often not realized that Tolkien’s own poetry, such as the elvish poetry featured in “The Lord of the Rings”, often relied on these borrowed, ancient patterns.)

Perhaps, though, the greatest loss due to translation is that of “feelings and thoughts”. A word or passage in the original can rarely be matched with a word or phrase in the target language having the same nuance and meaning. The translator will either chose a word that is “close enough” or turn the original into a longer phrase hoping to at least capture the meaning and feeling of the original. Some translators even appear to focus almost purely on the feeling of the original, considering the author’s choice of words and patterns as being of less importance. For example, in a book I had recently come across, in one passage the “ground exploded” as an enemy appeared from deep within the earth. This was translated as the “ground tore”. They are quite similar in meaning and effect, but the nuance and feeling behind these words differ greatly. The first, “exploded”, is violent and shocking, whereas the second, “tore”, is more surprising and less frightening — a bit tamer, one might say.


And so, I am concerned. As much as I personally like to take small dips into various works in their original language, I am unfortunately not one of the great scholars of the century before. My inability to grasp the “deep” grammar that is found among all languages and my difficulty in memorizing new vocabulary prevent me from reading fluently in the multitude of languages I would like to. However, I worry not so much for myself, but for the future generations who are slowly losing access to these great works of old — the enchanting tales, insightful diaries, and rich histories of times and people of yesteryear. It is sad to realize that one day these people, their thoughts, lives, and dreams will only be accessible, and remembered, through the imperfect mirror of translations.


* This is obviously a problem for all ancient languages, not just Western ones. However, as my own area of experience deals mainly with Western languages such as Old English, Old Norse, and Latin, I am writing from this point of view.