If someone has gone through and transliterated it and done a word-for-word gloss. But I have worked through the grammar enough that I can at least follow that.

Let’s work through an example. We start, as Westerners have for more than a century, with the Hitopadeśa, a medieval book of sagely advice told through animal stories. I start with Max Müller’s 1864 edition.  Here’s a sample line.

राजोवाच । भो भोः पंडिताः श्रूयतां ।

râjâ   -jan, N.sg.  The King
uvâcha:  vach, 3.sg.Perf.Par.  said:
bho  Ind.  O
bhos  Ind.  ye
paṇḍitâs  -ta, V.pl.m.  wise,
śrûyatâm  śru, 3 sg. Imp. Pass.   be it heard

Now, Devanāgarī is not hard to read. It’s an abugida, meaning that the basic grapheme is a single consonant with an inherent vowel. E.g. it starts with क = ka. Diacritics modify it to change the vowel: कि ki, कु ku, का , and so on. If you really want a naked k, perhaps at the end of a word, you write क्.

If you actually transliterate Müller’s Devanāgarī, syllable by syllable, you get this:

rā-jo-vā-ca bho bhoḥ paṃ-ḍi-tāḥ śrū-ya-tāṃ

Which, if you look carefully, isn’t what Müller provides.  What happened?

Sandhi happened. All languages have processes of assimilation and relaxation that happen as words are uttered in context. Occasionally these become noticeable to people and they attempt to write them down— e.g. someone is represented as saying “I hafta go” for “I have to go”.  Sometimes the assimilations are lexicalized, which is why we write assimilation and not adsimilation.

Well, in Sanskrit there are a lot of such adaptations, and you have to write them all. So for instance the vowels ā + u combine into o: rājā uvāca > rājovāca. (Müller’s â / ch are older transliterations; we now use ā / c.)  The –s at the end of paṇḍitās changes to ḥ before the following  ś, while the final in the last word changes to ṃ, which in this case indicates nasalization. Before a stop, it’s pronounced as a homorganic stop, which is why paṇ- changed to paṃ-.

There are special diacritics for these last two letters: e.g. kaṃ would be कं, and kaḥ would be कः.

So, Müller is providing the pre-sandhi versions of the words, which makes them easier to look up in a dictionary.

(A complication for the actual book I’m writing: It turns out that Word and Illustrator don’t properly handle Devanāgarī. They can’t do the combinations– e.g. nra should be written न्र, but they turn that into न् र, like barbarians. So I won’t be able to use a lot of Devanāgarī except as, shudder, bitmaps.)

Next we need to translate his glosses to a briefer and more modern convention:

king-s.nom say-perf.part.-3s oh wise-p.voc.m hear-imper.pass.-3s

Müller glosses bho bhos as “O ye”, but this is a bit confusing— bhos is not a pronoun. An online dictionary suggests that it’s an interjection often used in addressing people: oh! hello!  indeed!   And it seems that we’re actually dealing with a reduplicated form here, bhobhos.

Finally we can provide the translation:

The king said: O wise men, let it be heard…

That’s enough for today, but on request I’ll tell you what the king wanted heard. And you should request it, because then I can talk about Sanskrit’s insane mega-compounds.

By the way, classical Sanskrit wasn’t written in Devanāgarī— it was written in the local, contemporary script. All modern Indian scripts, and Southeast Asian ones as well, ultimately derive from Brāhmī, which is what Aśoka knew. If you write your vernacular in Devanāgarī, as of course Hindi speakers do, then you write your Sanskrit in Devanāgarī; but if you speak Tamil you use Tamil script, and so on.

How, you may wonder, does this compare to learning wényán for my China book? The script is way easier, of course. But sandhi is a nightmare, and the grammar is far less accessible. You can boldly translate wényán poems knowing little but the glosses, but I don’t think I’ll be doing my own translations of Sanskrit poetry.