Various points

[michaelnmmeyer]

To help identify problems, I created four additional files. malten_full.txt, iast_full.txt and ori_full.txt hold the full text of the respective versions, concatenated into a single file, and without markers. missing_full.txt is a new version of missing_alpha.txt.

There are about 40 less lines in malten_full than in iast_full and ori_full, probably because some markers from malten_full where incorrectly transliterated. In malten_full itself, there are quite a few weird character sequences, like "coÔøΩkhaÔøΩ", which are preserved in iast_full but obviously should have been transliterated to something more meaningful.

Concernins scans, they take up a lot of space, so I suggest we host them somewhere else than in a git repository. I will make PDFs out of them and send you the files.

Now some specific points:

(1) OK for ẏ and ḷ.

(2) We have both "b" and "w" transformed to "ବ". Is this normal? Should I also modify the "iast" version to use a single of these characters?

(3) For "ṛ", there are indeed several cases where it is preceded by a vowel. I found the following:

jṛ
tṛ
hṛ
nṛ
pṛ

Should I replace these with jr̥, etc. in the IAST version, globally? For other instances of "ṛ", where it represents ଡ଼, should this symbol be treated as a consonant viz. if a vowel follows, should it be represented as an initial vowel or as a vowel mark in Oriya script?

(4) The transliteration issues you spotted in Text_files_problems.txt are harder to tackle, I need more specific transformation rules. For instance we have pUboGge > pUrbe but also eboGg > ebaM, I do not know the criteria to decide whether boGg > rb or boGg > aM.

[arlogriffiths]

Thanks. Can you update the README in the light of the of you say in the first few paragraphs of the above issue?

(2) Odia script makes no distinction between b and v and there is no graphic distinction between w and v either. I would personally have transliterated all cases of w and b in the source data as v. Please convert all cases of w to v but leave cases of b to be looked at later. (I don't know whether any pattern can be discerned in the typist's choices.)

(3) jṛ — only one case. It may be a typo for jra but for now let's make it jr̥ tṛ — only one case. tṛtīẏa >> tr̥tīẏa hṛ — two cases. replace both by hr̥ nṛ — three cases. replace all by nr̥ pṛ — two cases. replace both by pr̥. in the second word, correct the pṛthrabīki to pr̥thibīki

(4) could you try to execute the steps I have formulated? I don't see any other way than to eliminate the typist's confusions more or less on a case-by-case basis.

An additional point from my side: for compliance with ISO15919/DHARMA, please change all cases of ṃ to ṁ in the source txt files.

[michaelnmmeyer]

OK for everything, except the treatment of ṛ and of clusters of vowels. Can you please correct the following assertions? They are probably incorrect.

Whenever "ṛ" is followed by a vowel, convert it to "ଡ଼" in the Oriya output. But keep "ṛ" in the iso transliteration.

Whenever we have a consonant followed by several vowels (except for consonant + ai/au + something that is not a vowel), all but the first vowel must take their initial form in the Oriya output. We assume that vowels after the first one each hold one character in transliteration: for instance, diai must be interpreted as d, i, a, i, not d, i, ai. In transliteration, each vowel after the first one must be represented in capitals.

[arlogriffiths]

1. "Whenever "ṛ" is followed by a vowel, convert it to "ଡ଼" in the Oriya output. But keep "ṛ" in the iso transliteration." I think this is correct. Only the above cases jṛ etc. are to be converted to r̥ in DHARMA-ISO. And whenever ṛV occurs, ṛ must ve a consonant akṣara and hence be converted to ଡ଼ in Oriya.

2. I would restate the rest as follows: a. all cases of ai and au are ambiguous (whether ai/au or aI/aU is intended) and need to be cleaned up progressively by identifying frequently occurring words and make the necessary replacements per word, then manually correcting less frequent cases b. all other C+vowel+vowel combinations have to be interpreted as lower-case vowel plus upper-case vowel in DHARMA-ISO (e..g., nei > neI, basāile > basāIle) c. all vowel+vowel+vowel combinations must be interpreted as lower-case vowel plus lower-case vowel plus upper-case vowel, because (DHARMA-ISO independent vowels) Ai and Au, anyhow presumably very rare, I expect only to occur at word-beginning and always to be followed by a consonant, if they occur at all. by this logicl, diai has to be interpreted as diAI = ଦିଅଇ

I hope this restatement is helpful. As I am not very familiar with medieval Oriya, some of my assumptions may be wrong. Before making batch replacements, it may be safest to supply me a list of all cases of space+a+i and space+a+u so I can try to confirm my assumptions under 2c.