BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 47 



yuhi^m. eagle (also yuWm. is dd^- n- ear 



found) 

 guln'm oak hehe^n rushes 



¥u\um. fish (sp.?) ga'TcIan house ladder 



legem- kidney gwit On- wrist 



It should not be concealed that a few words (such as hulUn ocean, 

 t.'aga^m, lake, and yuk!um-a- bones) do not seem to conform to the 

 phonetic law implied by the table; but more exact knowledge of the 

 etymology of these and similar words would doubtless show such 

 disagreement to be but apparent. It is probable that in delga'n- 

 BUTTOCKS, hilga'n- breast, and do'Win-i- anus, the g, {¥) im- 

 mediately following upon the I prevented the expected dissimila- 

 tion of n to m; in le'¥wan- anus the dissimilation was perhaps 

 thwarted by a counter-tendency to dissimilate the two labials (V^ 

 and m) that would thus result. *yalan-an- lose (tr.), dissimilated, 

 as we have seen, to yalal-an-, fails to be further dissimilated to *yalal- 

 am- because, doubtless, there is a feehng against the obscuring of 

 the phonetic form of the causative suffix -an-. The great probability 

 of the existence of a dissimilatory tendency involving the change 

 of 71 to m is clinched by the form do'Wim-i- anus alongside of 

 do^Win-i-. 



A dissimilation of an original lio n (the reverse of the process first 

 described), because of an I in the stem, is found in 



yill^nma'^n I keep asking for it (= original *yillHma'^n [I inserted 

 as repetition of stem -I- in iterative formation from yilima'^n 

 I ask him]) 



le^ha'nxde^ I am carrying (object not specified) (= original *Ze^- 

 la'lxde^) ; cf . identical suffix -al-x-, e. g., gayawa'lxde^ I eat. 



In u^gwa'nxde^ I drink (stem ugw-), it hardly seems plausible that 

 -an-x- is at all morphologically different from the -al {-an) -x- of these 

 words, yet no satisfactory reason can be given here for a change 

 of the I to 71. 



§ 22. CATCH DISSIMILATION 



If to a form with a glottal catch in the last syllable is added a syn- 

 tactic (conjunctive) element, itself containing a catch, the first catch 

 is lost, but without involving a change in the character of the pitch- 

 accent; the loss of the catch is frequently accompanied by a length- 

 ening of the preceding vowel (or rather, in many cases, a restoration 

 of the original length). This phonetic process finds its most frequent 



22 § 



