This 'facsimile' style of transliteration is not one that commends
itself for development of a standard, and the usual transliteration
schemes will not support it.'Facsimile' ___ Deep structure ____ Transcription-like ___ Mixed
As an extreme case, it would sometimes be possible to make a
transliteration which was then processed to produce a facsimile of the
original source text (if the same font were used). For this purpose
one would have to distinguish between allographs, which would make for
an extremely elaborate scheme.
This 'facsimile' style of transliteration is not one that commends
itself for development of a standard, and the usual transliteration
schemes will not support it.
The DEV allographs of "kka" are all transliterated as kka, reflecting the deep structure k+k+a. The following table shows some examples:
Surface structure Deep structure Transliteration -------------------------------------------------------Hence transliteration usually works from the deep structure of the scripts. Usually, a scripteme has more than one transliteration, while allographs have the same transliteration. Are we not always transliterating the deep structure in this way? For a few of the elements things are not so clear (see the next two documents).DEV k + aa kaa
DEV (none alone) (none alone)
DEV k + k + a kka
ORI t + m + a tma
ORI t + m + a tma
This would be a tightly restricted modification of a reversible
transliteration, showing words in a form nearer to their
pronunciation. The most one would expect in the case of Indic
scripts would, I think, be:
(i) omission of the inherent vowel where it is standardly
not sounded in North Indian languages. This occurs in
Assamese and Bengali (not when final after a consonant
cluster; I forget whether it is ever omitted medially);
also in Hindi (finally and in some medial positions);
are there any others?
(ii) phonetic representation of the inherent vowel in Assamese,
Bengali, and Oriya. This would be 'o', long and short,
instead of 'a'. The fact that there are these two
pronunciations of this 'o' is an added difficulty.
A serious difficulty here is that a word which has the same spelling
in two (or more) languages because it is the 'same' word, would lose
its obvious equivalence.
This style of transcription is not reversible, and could not be the
basic standard. The Hindi word pronounced [matlab] is written
matalaba in DEV and in a reversible transliteration. Why shouldn't
it be equally easy to learn how to pronounce it in either
representation? Why should the way it looks in English be important?
To try to cover all possibilities: a combination of styles 2, 3 might seem
to be attractive in some situations. These will have to be judged on
their individual merits, but it seems desirable to keep to the deep
structure as far as possible.