Hebrew triliteral roots whose final consonant is y have an anomalous conjugation in most templates. Why is this?
- It is only in the passive participle template CaCuC, as far as I can see, that the y actually appears as predicted: e.g. from the root t-l-y "hang", taluy "hung".
- In some templates the y disappears completely: e.g. past 3sg. m. CaCaC, tala "he hung"; past 3pl. m. CaCCu, talu "they hung".
- In other templates an expected ay or oy turns into e: e.g. fut. 3sg. m. yiCCaC/yiCCoC, yitle "he will hang".
- Strangest of all, in some templates a t appears instead of the y: e.g. inf. liCCoC, litlot "to hang"; past 3sg. f. CaCCa, talta "she hung".
The first two changes (loss of y and change to e) seem phonetically understandable, though still apparently irregular. But the appearance of t has no plausible phonetic basis that I can see. What are the reasons for these anomalies?