Secondly, that every double letter, though in pronunciation it may be simple, should be transliterated by a double letter, and that a single letter, although its pronunciation be that of a double letter, should be transliterated by a single letter.
If these two principles be strictly observed, everybody will be able to translate in his mind a Canarese book, written with Roman letters, back into Canarese letters, without losing a tittle of the peculiar orthography of Canarese. If we attempted to represent the sounds in transcribing literary languages, we should be unable to tell how, in the original, sounds admitting of several graphic representations were represented. In written languages, therefore, we must rest satisfied with transliterating letters, and not attempt to transcribe sounds.
This will cause certain difficulties, particularly in languages where pronunciation and spelling differ considerably. But even in Greek, if we had to transliterate ἐγγύς, we should, no doubt, have to write ʼeggues, which none but a Greek scholar would know how to pronounce correctly (ʼengues). But if, instead of imitating the letters, we attempted to represent their proper pronunciation at a certain period of history, how should we know, for instance, in transcribing the French of the nineteenth century, whether "su" was meant for "sou," halfpenny, or "sous," under, or "soul," tipsy. In historical languages the system of orthography is too important a point to be lost in transcribing, though it is a mistake to imagine that in living languages all etymological understanding would be lost if phonetic reforms were introduced. The change in the pronunciation of words, though it may seem capricious, is more uniform and regular than we imagine; and if all words were written alike according to a certain system of phonetics, we should lose very little more of etymology than we have lost. Nay, in some cases, the etymology would be re-established by a more consistent phonetic spelling. If we wrote "foreign" "forĕn," and "sovereign" "soverĕn," we should not be led to imagine that either was derived from "reign," regnum, and the analogy of such words as "Africĕn" would point out "foranus" or "foraneus" as the proper etymon of "forĕn." But although every nation has the right to reform the orthography of its language, like everything else where usage has too far receded from original intention, still, as long as a literary language maintains its historical spelling, the prin-
