Word splitting & hyphenation conventions in roman transliter

[Dermot Killingley]

Dear colleagues,

I too am interested in word division in romanized Sanskrit, so I
welcome Harry Spier's query. My practice is:

(1). Mark word division with a space wherever possible, i.e. wherever
two vowels are not run together in sandhi. E.g. tad avadat; tad
vadati; so 'vadat; tac chrutam; but: tenoktam; tadAvadat; tadarSiH.

(2). Within compounds, mark division between stems wherever possible,
i.e on the same condition as above. E.g. vRddha-vyAdhaH; mat-putraH;
mac-chiSyaH; but mahendraH; kSIraudanam; maharSiH.

(2a). I sometimes leave a compound undivided if it is not an ad hoc
phrase but an institutionalized unit with a meaning of its own
(there's a grey area here). E.g. tatpuruSaH if it means 'endocentric
compound' (but tat-puruSaH if it means 'his man'); hRcchayaH 'love'
(but hRc-chokaH 'pain in the heart').

(1) is common practice, except that many writers don't put a space
before an avagraha (e.g. so'vadat). Nevertheless there is an
alternative practice of not putting spaces where they would not be
put in Devanagari (e.g. tadavadat; tadvadati; tacchrutam). This is
sometimes said to be because of sandhi, but it seems rather to
result from reverence for the supposed principles of Devanagari,
disregarding the fact that putting spaces in Devanagari at all is a
relatively modern practice.

(2) is less common, but by no means rare.

(3). The practice in Devanagari is again to mark word division with
a space wherever possible, except that what is possible in roman may
not be possible in Devanagari. So word division is marked wherever
the first word ends with a vowel, anusvAra or visarga, but not where
it ends with a consonant. This is why, as Spier notes, the
spacing in a romanized text does not match the spacing in a
Devanagari text. Again, many books don't put a space before avagraha,
but I don't see why not.

It should be noted that spacing, either in romanization or
Devanagari, is a matter of editorial interpretation, not an integral
part of the text. E.g. an editor who prints anAdi matparaM brahma in
Bhagavad-gItA 13, 13 is reading the same text as one who prints
anAdimat paraM brahma, but signalling his intention to construe it
as 'beginningless brahman intent on me' (or something like that) and
not 'beginningless supreme brahman' (the same applies in Devanagari,
except that there would be no space in mat paraM).

There is an old but interesting discussion by C. R. Lanman in
J.Hertel, _The Panchatantra...of the Jain monk, Purnabhadra_,
Cambridge Mass., Harvard University Press, 1908, pp. xix-xlviii.
Dr Dermot Killingley (Reader in Hindu Studies)
Department of Religious Studies,
University of Newcastle upon Tyne,
Newcastle upon Tyne NE1 7RU
Tel: (0191) 222 6730. Fax: (0191) 222 5185