65$ K. B. WIKLUND.
g. jiggde, tjappgan, etc.), partly between the first and second vowels of the word (e. g. Kurruk, Tikkenlik, etc.). In my sources I failed to find practically any traces whatever of long consonants, except in the case of words like tallik, where -lik is a derivative suffix, added to a root ending with an l sound. Consequently I was driven to conclude, that the duplication of the consonant, which occurred very frequently indeed, must, at any rate for the greater part, be due to faulty apprehension. This was a case too in which a Swede would be more particularly liable to err, owing to the fact that in his native language a consonant at the end of a syllable, following a short stressed vowel, is always long, while Swedish can show us no instance of short, stressed vowel + short consonant + vowel, this being replaced by the combination short vowel + long consonant + vowel. At the same time it is true that in Turkish words the accent falls upon the last syllable; but according to Radloff, Phonetik der nördlichen Türksprachen, § 136, the first syllable of the word has also a secondary accent, and, as I suppose, it was this which misled Dr. Hedin's Swedish ear. However that may be, I believed I had good reason to simplify the first of two succeeding consonants as well as the intervocalic consonant whenever I was able to get confirmation of the form of the word in the literary language. Nevertheless, as I discovered during the progress of the work, I bad sometimes been needlessly afraid of simplifying the intervocalic consonant; but it was then, as I have already observed, too late to make any alteration, and it was only exceptionally that in the text which was printed later I was able to employ a slightly modified form of the name which appeared on the map; it would of course have been exceedingly inappropriate to have employed different name-forms in the text and in the map. The thought which dictated my scruples with regard to the simplification of double consonants was this: it might be just possible that in this or the other instance the pronounciation might preserve a long consonant of great historical interest, which I had no right to obscure by a simplification of spelling. Certain duplications which I noticed in Shaw confirmed my scruples with regard to this, and when in November 1904 a phonetically written Kaschgar text was for the first time printed, namely in the journal Keleti Szemle (Budapest), my surmise was converted into certainty. The only way to obtain complete certainty on such points as this is to make investigations in the locality where the dialect is actually spoken; sitting at my study table I have no means of determining which are the words in which an intervocalic consonant really ought to be written long, any more than a physicist can without cause alter a single figure in a result which he has reached by a process of rigid investigation or experiment.
Another of the principles which I endeavoured to carry through was, instead of using g in both the low-vowel and high-vowel words, as Dr. Hedin had done, to reserve g for the high vowel words only, and to employ gh for the low vowel words, following in this respect the spoken language. In the great majority of cases it was quite easy to carry through this principle, for it was evident on the face of the word itself to which of these two categories it belonged. Sometimes however very considerable doubt arose, owing to the fact that in the same word there appeared to be both a (at all events ex oculis) low and a high vowel, as for example in Ugendarja, »also pronounced Ögen-darja». Had any other but a Swede written down
IIIFカラー高解像度
