A Brief Explanation of The Translation Process

© 1995-2001 Untangle Incorporated
Last Updated: Sunday, December 2, 2001

Overview:
Translation is science, and poetry. Let's deal with the science first.

Before you can translate one language into another you must know both languages. Now with many of the very oldest written records, there is no one living who speaks those languages. This is not a minor problem. How words were actually used, what cliches actually meant is not always clear from two versions of the same text in two different languages. The translations we have for anything from Mesopotamia is often a chain of doubtfully understood languages.

For example with the Gilgamesh Epic tablets there is a version in Sumerian, and another in Akkadian, the language of Babylon. We began to understand these two written languages a little bit is because a piece of text that was carved on the face of a cliff in three languages about 2500 years ago. Henry Rawlinson in 1857 was in Afganishtan for the East India Company when he spotted the text high up on the cliff. With rope, paper and charcoal he bravely copied the characters (cuniforms). The third language was Old Persian. What we know about Old Persian comes from commentaries made more recently in Ancient Greek, Latin, medieval Persian and some Sanskrit texts. Also there is an evolution from ancient to modern Persian, as there is between old English and modern English. There is still much room for error.

Errors may arise as the meaning of a word can change over time in any language. English has words that have drastically changed their meaning in less than a 1000 years. ('Nice' four hundred years ago meant the same as 'foolish' or 'wanton' does now). Even the word 'myth' meant 'an inarguable truth' until the last 100 years when it often meant 'a complete lie'. You must know which of the two meanings is being used in a sentence or you could misunderstand the intent of the sentence entirely. Usually we have no idea how a particular Sumerian word evolved over a thousand years or more. The first Sumerian writings we have are about 2600 years before the last ones (3100 B.C. versues 600 B.C.) so there was plenty of time for some of the cuniforms to change meanings as well as shape (an analysis which also has been done).

Another source of error is guessing at the meaning of a word in some text which we have no equivalent for in another language. If it something which makes the unknown word fairly clear then their is little problem, but usually there needs to be a lot of text from different tablets or whatever for scholars to make a reasonable quess at the meaning of a new word.

Idioms are problem in a class by themselves. There is the English idiom, "As right as rain". But what it means depends on context, and who is saying it. It means, when rain is needed, that this is a good thing. When rain would be harmful (say during harvest) then the meaning is sarcastic. Moreover the phrase would mean completely bad by picnickers who are trying to eat outdoors. And the phrase is applied to matters having nothing to do with rain. "Grand kids playing with their grand parents is as right as rain.". There is a Sumerian saying 'seven and seven more' which scholars believes means 'many'. The usual first translation is something like 'dug seven pits and seven pits' which probably means 'dug many pits'.

Finally there is a problem specific to cuniforms. Collections of cuniforms were not likely a whole sentence but just words to remind the scribe what to say orally when reading it. Marie Thomsen put it like this: "The signs were memory aids not complete thoughts." ("The Sumerian Language", pp 23).

Furthermore the signs changed over time.


Transliteration:
The cuniforms look like this: Cuniform Example
from John Hardy's "Sumerian". Until recently cuniforms were hard to reproduce in printed books so instead each cuniform was given a specific code word in Roman letters. You can consider cuniforms as letters (phonetic morphemes is more correct). Transliteration is process of changing from the letters of one 'alphabet' into another. Russian, which uses the Cyrillic alphabet can be changed into English letter by letter (although some Cyrillic letters correspond to two or three English ones together). (Example)

Several different versions of the same document are used to decide which of the symbols is most likely used originally. Once you have several different copies of the same text a sumologist would compare the cuniforms used and see how they correspond and differ. The differences and the similairities are used to reconstruct the most likely orignal (oldest) copy from which all those copies were made. Jacobsen describes the process in detail in the "The King List" (for example on pages 15-27).

Also if a cuniform always or often appears within a group of other ones then if the tablet being studied is damaged but the some of the usual ones in a known group are appearing close to each other then the missing ones are guessed assuming it is the known cuniform 'phrase'. Since Sumerian writing was apparently a 'mnemonic' then the same phrase would be used as a further simplication to aid memory.

The coded words had to be agreed on between scholars over time, just as reasonable quesses as to meaning of each symbol the coded word represents. It often takes decades for scholars using all available evidence on which cuniforms are truly different and which are just changes of a cuniform over time, or are just scribal errors. All evidence available may not however be sufficient to lead to an meaning acceptable to most scholars. When they agree then it can make transliteration and eventual translation easier for new tablets as that code word is found in new tablets. See Tom Jones' "The Sumerian Problem" for a description of all the problems in some detail.

The linguistic scholars thought they knew how the cuniforms were sounded out because Akkadians were apparently, from archeological evidence, Semitic people and there is lots of knowlege about Semitic languages from biblical studies. In fact we have no way of knowing for certain how the language sounded. However the coded letters used represents the assumed sound of each cuniform. Everytime a certain cuniform appears it is assigned a code word in Roman letters, such as 'sha'. Several cuniforms in a row may appear transliterated as 'Sha na la na pa', This gives the translator the starting point to determine what the symbol means in this place as will be described in the translation section below. An example of transliteration is provided in Gilgamesh and the Huluppah Tree translated from Sumerian tablets by Samuel Noah Kramer in 1938.

Translation:
Knowing what code words go with which cuniform is just the simpliest of the tasks. The actual meaning of each code word in context is a much more difficult job. The translator is using the transliteration to determine the grammar of the statement made, and from that, and everything said before and after, a reasonable guess as to what the meaning of each symbol or collection of contiguous symbols means right here. How that is done is basically to guess that words that in Akkadian sounded like most Semitic words. As mentioned the centuries of biblical scholarship was used to assign meanings based on supposed sounds of Akkadian words. The derived meanings gain some reasonable facsimile of certainty when there is many transliterations that can be cross checked to other languages and to each other as well as internally (see The Huluppah Tree for an example at the beginning where the repetition of the "After the...." of several lines seems like a ritualized speaking that would lend power to the language and confidence in the translation. Some repetitive phrasing is seen in the early sources of the Five Book of Moses.

There are problems with taking the transliterated sounds and turning them into meaningful phrases. The first problem is with dialect and sound evolution. In English we write 'the' but it sounds mainly like 'duh'. What this means is that guessing the meaning of a word from the sound of modern versions of Semitic languages is hoping that the dialect of Akkadian sounds the same as current Semitic languages do. As each sound (read coded letters) is surrounded by other sounds then all the meanings that are being thought of for each sound must make sense. The supposed meaning can also be deduced by comparing it to the meaning of the text as written in another and better understood language. So b carefully sifting through many transliterated texts the result will be some sounds which are pretty certain as to meaning, some which have a fair certainty and some which have great uncertainty and then some which have no meaning yet at all. One example is Huluppah Tree itself. Scholars don't really know what Huluppah means. It could correspond to a 'willow' but that is a guess based on what trees likely grew in the area when the story was written. It was not the only type of tree that grew there then. The second problem was a surprise when discovered finally. Semitic languages put words in a certain order usually. In English most sentences go into a (Noun) (VERB) (Noun) order (subject-verb-object or SVO). So English says "We go home". Semitic languages are VSO usually ("go I home"). Akkadian at first was thought to have that structure. But the sound-meaning association was not working well at all. Then it was realized that Akkadian, if it was SOV ("I home go"), the corrleation between meaning and the Semite sounds was much improved. This suggests that the Akkadians spoke a language not in Semitic order . But they were Semites so therefore they must have been subordinate to these other people (eventually realized to be the Sumerians themselves)and were instructed as children by the dominant people at the time in a word order not usually used in people of their language group. (This occurs with people who learn one language, say French as a child, and then, as an adult, learn another, say English. The earlier learned language order dominates the structure of the words in the later language for a long time. French usually puts adjuectives after the noun, English before, yet a native French speaking person learning to speak English often would say 'an angel perfect' because in French it is 'un angel parfait'. )

Just knowing the word order allowed the purpose of a word to be guessed, the last word would likely be a verb, and so if you had another language then you might be able to assign a meaning based on the verb used in that part of the text in the other language.(Assuming you understand the other language well enough to do that with confidence).

Another potentially severe problem is the realisation that the cuniforms were not complete sentences(Thorsen pp 23, "Sumerian Language"). They were mnemonic words to help the scribe remember what was to be said out loud. This would be like rendering the the last sentence into 'write sentences scribe remember to say out loud', or maybe even less and when spoken it is said as a full sentence as originally written.

In spite of these and other problems, after decades of work a dictionary with which most scholars agreed was eventually created. T. Jones' "The Sumerian Problem" gives more examples and explanations. Scholars looking at all the evidence feel the cuniform was actually used until about the Selecucid period (2nd century B.C.), but most of the literature from the Old Babylonian period was not copied after that time. In the later periods the Sumerian texts are incantations, proverbs, and laments, usually with many errors likely because the understanding of the underlying language was gone. So these later texts are not used as a rule because of that.

When Sumerian tablets were found it was realized after a while that they were the formal language (mnemonic) that was used by Akkadian scribes. This process is akin to what happened in Europe where Latin was used to write but was not spoken by very many people from the mid 17th century onward. The meanings of most Akkadian sounds have Sumerian equivalents. In some cases, as I have said there is no meaning yet for the code sound. But at least people know what words need meanings and as new texts are transliterated opportunities to do just that arise. (See J.Hayes 'Sumerian' page 37 for a dictionary of Sumerian sounds).

Literary Translation:
Given a grammar translation with probable meanings (see Huluppah Tree) you can try to form a literary translation that sounds like very entertaining and correct English. N.K. Saunders has done that with the Epic. And so have other writers. Miss Chen did it for MythHome from her grammar translation. And I have done it with the Huluppah Tree grammar translation of S.N. Kramer. See the page on that Sumerian tree to see what how my translation reads.

All that said we hope you enjoy the Epic of Gilgamesh even more knowing the many problems that were (and are still being) tackled in it's translation.
ROC for MythHome.


SOME SOURCES
Hayes, T. "Sumerian",1997
Jacobsen, Thorkild,"The King List", 1939
Jones, T.,"The Sumerian Problem" 1969
Tigay A., "The Evolution of the Gilgamesh Epic",1982
Kramer S.N., many books but for this page we read "Gilgamesh and the Huluppa Tree",1938 Text and Translation chapter 1
Pritchard J.B., ed. "Ancient Near Eastern Texts Relating to the Old Testament", 1969
Saunders, N.K. "Gilgamesh: The Epic of Gilgamesh", 1972 (this is a literary translation)
Thomsen, M. "The Sumerian Language", 1984