Reduplication and syncope could work: Say that every other vowel backwards from the end of a word is deleted, unless that would result in an initial consonant cluster:احمک ارش-ھجنو wrote:For my 3con language, how do I diachronically come up with patterns that geminate the final consonant of a root and patterns that geminate the first consonant of a root?
/parata/ > [par'ta] vs. /paratata/ > [parat'ta] (reduplication of the final syllable followed by vowel syncope. /a/ in the first syllable retained due to illegal initial [pr] cluster)
/tiluka/ > [til'ka] vs. /atitiluka/ > [attil'ka] (reduplication of the first syllable, with the addition of a prefix, followed by vowel syncope)
IIRC, some believe that reduplication and haplology led to gemination of the second radical in Semitic languages, e.g. -paras- > -parparas- > -parras- (with loss of the second instances of -pa- through haplology). There's a chance syncope may have helped this as well, e.g. -parparas- > -parpras- > -parras- with simplification of the medial cluster -rpr-.