deriv LSK ETT STT aSTA ALPH OLDHOMEPAGE NEWHOMEPAGE
The inria reader is a segmenter made to help college students.
It works like this:
You go to The Sanskrit Reader Companion and set the "Input convention" dropdown to KH (meaning HK).
Then you type a sentence in there, such as
rAmo 'patat रामो ऽपतत्
or
apatad rAmaH अपतद्रामः
(Do not forget the /avagraha (') before patat!)
Then click the READ button.
The machine then tries to figure out what words made up that sentence and if they are nouns or verbs. To do that it uses a list of nounbases, a list of verb roots, and a list of grammar rules that it aplies in reverse to try to get in as many ways as possible from the roots it has to the sentence you typed.
Unlike a translator like google translator, this machine is smart enough to know that it is not smart enough to always get the meanings that make sense.
If a translator program finds out that a sentence might have been formed in 369 different ways, it chooses the one that it thinks "most likely", and that often results in throwing the baby out with the bathwater. After that it turns that "most likely" analysis into English or whatever.
This machine is far more useful. Because it is a segmenter, not a translator. When it finds that its rules and dictionaries allow it to analyze your sentence in 369 possible ways, it shows you the 369 possible analyses so that you can you discard the bad ones yourself, and it does not even attempt to translate.
Back to our example of "rAma 'patat".
After you click "Read", the machine shows color boxes. In this case the machine tells us that we have six possibilities, and we have fuchsia blue and yellow boxes. Every box is a possible word. Below every box you have a V and a X. Click X if you think that the machine screwed up, and click V if you think it is the good one. After clicking ONCE, wait for the page to reload.
The blue rAmaH box means that the machine thinks this might be the **rAmas रामस् noun meaning "rAma", formed from a nounbase **rAma- रामॱ plus a noun ending /su.
The fuchsia rAmaH box means that the machine thinks this might be the **rAmas रामस् verb meaning "we give".
If we found that sentence in a passage of the rAmAyaNa that explains how lakSmaNa hunted a deer then skinned it and then "rAmo 'patat", we can discard the possibility "we gave". So we click the X under the RED verb rAmaH, OR, we click the V under the BLUE noun rAmaH. Then we wait for reload.
The thing then tells us that three possibilities remain. If the fuchsia apatat is good, apatat is a verb and our sentence means "rAma cooked". If the yellow a together with the blue apatat is good, then a-patat is a /samAsa. And if the green is good, then the yellow "a" together with the green "patat" make a neuter /samAsa /Amantrita.
When you have say the level of a freshman, discarding the yellow needs no thinking. You can feel certain that only the fuchsia apatat is good at this stage. Otherwise, you click the fuchsia, the blue and the yellow and the green, trying to figure out which of them makes sense.
When you click the fuchsia, you get —
(impft. (1) ac. sg. 3 )(pat)
This means that apacat अपचत् is made by joining root pac पच् and tense /laT, and replacing /laT with /tip —
pac पच् + /laG /tip → !**apatat अपतत्
The (1) means that pat पत् belongs to verb class one.
How did you figure out that "impft. (1) ac. sg. 3" means /laT and /tip?
See inria abbreviations.