Page:Morris-Jones Welsh Grammar iv.png

iv book was begun in 1899; but I was dis&shy;satisfied with it, and made a fresh start a year or two later. The progress of the second draft was much hindered by exami&shy;nation work which took up the greater part of my long vacation for some years. In 1907 I had finished the accidence and written more than half of the syntax. As Early Modern lite&shy;rature consists almost wholly of verse in the strict metres, I found myself in the syntax quoting more and more from Medieval prose. At last I was forced to the conclu&shy;sion that the Medieval period would have to be dealt with in the earlier portion, which would therefore have to be entirely re-written. Many Medieval forms had already been quoted in it, in order to show that the Early Modern forms followed the old tradition, espe&shy;cially where the late written form is arti&shy;ficial; in some cases the etymology also was given, in order to show further that the tradi&shy;tional form had developed regularly. In re-casting the first portion I thought it would be well to bring together the laws by which Welsh sounds are derived from Keltic and Primitive Aryan, so that by reference to them any formation or word might be compared with its cognates, and traced to its origin. Thus from a descrip&shy;tive grammar of Modern Welsh the book grew into a Welsh Grammar Histor&shy;ical and Compar&shy;ative.

In its present form the work was commenced early in 1908; and the Phonology and Accidence now published were completed in the Spring of 1912. The volume has taken a year to print; and I have not found the time too long for the final revision of the copy and the correc&shy;tion of proofs.

A few words may here be said of the most important previous works on the subject. The earliest known Welsh grammar is that preserved in the Red Book of Hergest, and printed from a late copy as Dosparth Edeyrn Dafod Aur by Ab Ithel; apart from the treatment of sounds and metres this is little more than a defi&shy;nition of the parts of speech. Simwnt Vychan's grammar (P.Ỻ.) is also of value only for its prosody. The first printed