I tried to convert each word on the Unix wordlist (/usr/share/dict/words) into a pattern that shows how long it is and which letters repeat where. Examples: trek → ABCD, eerie → AABCA.
The most common pattern with n repeating letters:
n≥0: ABCDEF (3545 words): e.g. abduct, abrupt, absent, ...
n≥1: ABCCDEF (694 words): e.g. battery, buffalo, burrito, ...
n≥2: ABCCBDE (131 words): e.g. ballast, billion, cabbage, ...
n≥3: ABCDEFFEF (15 words): e.g. actresses, conferrer, confesses, ...
---
I also found a typo on the wordlist: "analyticalally".