If you use a South Asian script like Devanāgarī, then you will follow its conventions for hyphenation. If not, you might find the following rules helpful:
First of all note that writing is different than reading.
- The Roman alphabet is not completely appropriate for Sanskrit. Please remember to consider signs including an aspirate (like kh-, gh-, ch-, jh-) or representing a diphthong (like ai or au) as INSEPARABLE. E.g. ai-śva-ryau and not *a-i-śva-rya-u
- the Devanāgarī (short for all other Brahmī-based Indian scripts) writing system has its own conventions, which do not apply outside it. Thus, refer to syllables, not to Devanāgarī graphemes. E.g., dhar-ma-tat-tva and not *dha-rma-ta-ttva
- You do not know what is an admissable syllable in Sanskrit (syllables are not entirely “objective” entities and they vary a lot from language to language, although all languages agree in needing a vowel-sound in each syllable, see the figure for a glance into the different English syllables) and you do not want to dwell into Prātiśākhya discussions about it (or to read Giovanni Ciotti’s article here)? Use this very simple rule of thumb: A syllable is one with which a word can start. Thus, tva– is an admissible Sanskrit syllable, whereas *-rva is not.
- Thus, only leave a consonant with the preceding vowel if it cannot go with the following one. E.g., Ma-dhva instead of *Madh-va, because there are words beginning with dhv+V[owel], e.g., dhvaja. Similarly, Pañ-ca-rā-tra instead of *Pa-ñca-rāt-ra.
Which criteria do you apply?
I see some problems here:
To distinquish are the “phonetic aksharas” and/from the “graphic aksharas” — we spoke about these problems either in the Arbeitskreis with Prof. Steinkellner and during the editorial seminars of Prof. Preisendanz.
ad 2.) in the case of “dharmatattva”: graphic aksharas will be exactly: dha-rma-ta-ttva; the correct phonetic aksharas are: dhar-ma-tat-tva.
ad 4.) in this case “Madh-va”, which you designated as non correct (*), will be the normal “phonetic” transliteration into aksharas according to Sanskrit prosody.
To sum up: when we say “syllable” in English or “Silbe” in German, we should make it clear what instance/unity exactly do we mean: the phonetic or the graphic one, for both can and ARE desingnated “aksharas” in Sanskrit.
Thanks for the comments, Edgar.
Let us leave for a future time the discussion about the meaning of “akṣara” (and of “varṇa” etc., perhaps you have followed a past discussion about it on Aleix Ruiz Falqués’ blog between me and Giovanni Ciotti).
As for “syllable”, I use it in a purely phonetic or phonemic sense (since its meaning changes according to the language one is discussing). The whole point of the post was exactly to show that phonetics does not go together with graphic representation and that one needs to distinguish them while transliterating (for further comments about it, see here: http://elisafreschi.blogspot.co.at/2013/03/sanskrit-punctuation-and-related-matters.html).
Regarding your point “ad 2)”: graphic akṣaras only make sense while using Devanāgarī (or a Devanāgarī-similar) script. We do not ought to follow them while using a completely different system, such as the Roman alphabet. Thus, I insist on following what you call “phonetic akṣaras” while transliterating. I think you will agree.
Regarding your point “ad 4)”, I am really surprised. Please provide some reference for this hyphenation, e.g., in verses. I cannot recall any caesura like the one you indicate.
Addendum: Of course you are absolutely right when you say of impossibility (or more correctly — of a strict prohibition) to hyphen such single consonants as “ph”, “bh” (as “p-h”, “b-h”) etc., which are digraphs only in Latin transliteration., or such types of vowels as diphtongs, conventionally expressed in Latin characters also as digraphs: ai, au, etc.