| ™ |
On Translation, Commentary, Evocation
A Rant
An interesting original might yield a whole literature of translation,
commentary, and evocation. I get upset when one masquerades as the
other, because I want to learn what the source said, and have that
teach me how to better read other related sources too. I want
empowerment, not to be kept in a box and made to think about only what
one self-appointed intermediator wants me to think about for their
personal, private, or political purposes (which is far too often!)
Commentary: As the great Paṇini was followed by Patanjali and Bhāskara and others who helpfully explained his terse incomprehensibilities, commentatorial tradition can offer students the context and clarified original meanings of what the original author said. Often we sigh with relief because the original was so incomprehensible without these explanations.
Evocation: On the other hand, folks who never studied the source language but once read or heard someone else's translation or rendering, and who then thought of something yet else that was evoked in their mind, might finally write down their inspired evocation. An evocation gives scholars and students access neither to the original in its actual form nor its basic meaning (as Translation), nor even to its meaning as understood by native speakers who understood the original context of interpretation (as Commentary). Evocations are just evoked. You heard someone say something, and it triggered a thought. It was actually your thought: Your thought was evoked, and if you write it down, it relates to the original by being not a translation but an evocation.
Translation: These should help you access the original itself. Translations are not unique: almost all translations of almost all sentences are different from each other, according to empirical study of many translators translating the same source documents (David Graff, Linguistic Data Consortium, p.c.). Still a translation should try to tell you what the original author said; that is, it offers a path toward, in the direction of, understanding the original. Ideally a translation is a learning resource pointing at an original.
So, separately, please do add your explanatory commentary, which lets you understand the unspoken context, and your inspired evocations, which are interpretations and elaborations which are in dialog with the original. But it's a bitter discovery when I pull down a so-called translation and it gives no access to what was actually said.
I can complain, because I am the worst of sinners. My own Pratyabhijnahrdayam (2004 according to the internet archive "Wayback Machine") long before I took this year of UW Sanskrit classes was more error-full than evocative. At least it was close to the ground, but it was full of errors. Its only virtue was brevity. But since I published it, several, even many, other publications with the same title have emerged, and as I continue to encounter this remotely evoked literature claiming descent from PratyabhijnahRdayam, I feel the verbosity is not as offensive as a claim to be a translation. I would like them to admit they are the original writers of an evocation: a dialog with the original, not a faithful rendering of it.
Now, instead of merely complaining, I would like to offer real help. Since I want us all to do things differently, I hereby offer an easy way to do it.
The linked Translation Graph of Pratyabhijna HRdayam offers support for you to enter into the original Sanskrit words and their meanings, so that you can memorize and understand it directly.
Translation Graphs
The TG idea came from the multi-linear displays that linguists use when presenting materials in another language. In our case the layers run from the original Devanagari, through to the remotest commentaries and evocations. In between, layers include transliteration for the script-learners, then inflectional morphology for those learning grammar, then derivational morphology for those learning vocabulary, then piecewise translation and phrasal translation for those learning syntax, and full-verse translation to make the basic meaning clear, followed also by commentary or evocation where you can say anything you want.Translation Graph technology is at present a PHP library file of a few hundred lines, which allows you to take your verses and transliterate, translate, and comment on them in a structured way. Boundaries like .N0 comprise a period (meaning "boundary"), followed by a capital letter (indicating the first layer in which it was inserted), followed by a number (indicating the ordinal of this particular boundary in that particular layer). .N0 is in this case the start node or boundary for the devaNagari transliteration or original source layer, for example, and by convention .N$ is the end of the verse or fragment.
Any writer or translator, using easy copy, paste, edit, benefits from automatic re-use of existing boundaries in each successive derived layer, simply because in a copy the there-existing boundaries already indicate the correct correspondence points. You can add or remove them without disordering the rest. Editing a complex word in a more-analytic layer amounts to replacing it with its parts separated by additional boundaries: The most natural thing for a linguist, and the most valuable thing for a language learner! tglib will automatically align everything at different layers around the correspondence points which are those boundaries.
In sum: easy to write, and with tglib, easy to show as web-page tables in HTML.
The main work of the tglib.php library is to convert the different layers into a web-page-includeable HTML table that clearly indicates all the layer-by-layer correspondences. See below.
How?(1) A DAG or Directed Acyclic Graph can be used to represent a set of corresponding lines of text containing arbitrary boundaries so long as they stay in a consistent ordering. For example one start and one end boundary might be tagged in all the strings in the form .N0 $line .N$, and below-added boundaries can also be re-used between corresponding substrings in the different layers. "Directed" and "Acyclic" because the content always scrolls forward as in Linguistic Time, never looping back.
(2) A DAG is also a mathematical object which allows linear-time (super-fast) construction of a Topological Sort of its nodes/vertices/boundaries. A topological sort is an exhaustive sequencing of the nodes or boundaries in the DAG such that u < v in the DAG, then u < v in the topological sort. It gives a linear ordering to control searches that have useful properties.
(3) In particular, given a Topological Sort, a column or levels count can be constructed also in linear time using an algorithm I invented for this purpose. (I might not be the first but Google and I don't seem to know anyone else that has done this; so please let me know and I'll give credit. Really it is Very Simple, as evidenced by the fact that it works.
Since one complicated word can be broken down into several parts, tglib makes all the parts visible in aligned rows and columns so the reader can make sense of what corresponds to what, while the writer can very quickly generate the necessary multi-layered content by simply copying and modifying one layer to make another, adjusting the word forms etc., and inserting and removing .N# boundary tags in the derived copies which will show up as divided or merged columns below the previous layer. I'm excited because this makes it so easy to make these complex tables this way. Right away we get 17 boundaries at different places in a simple sentence, and putting the parts into a web page in an HTML table would be hellish detail work to make them line up, without tglib.php.