Within any language, there are rules that govern the addition of inflectional morphemes to words in order to create new forms. These rules are acquired early in child language acquisition and include very general rules (such as the rule ʺadd '-ed' to the verb to form the English past tenseʺ) as well as rules that apply to smaller subsets of items (ʺchange /i/ to /ae/ internal to a root to form the past tense in English, as in sing/sang, ring/rang, etc.ʺ). All rules, however, have exceptions, even with common words (most speakers say ʺbring/brought,ʺ not ʺbring/*brangʺ) and when exceptions are many 'paradigm gaps' can arise in which speakers are uncertain about what the proper form of a word should be, especially when the lexeme is infrequent. Take, as an example from English, the verb “to forsake” – many otherwise proficient speakers of English would encounter difficulty in forming the past tense of this word (forsaked? forsoke? forsooken?). This, along with the fact that even common words can be rule exceptions (consider the ungrammatical forms *haved, *goed, *getted), has led some to pursue an explanation for inflectional morphology that isn’t rule-based at all, but rather item-based. Tomasello, in his 1992 case study First Verbs, describes what he calls a “verb island”. In his research, he noted that in studying the acquisition of English by a child, out of “162 verbs and predicate terms used, almost half were used in one and only one construction type, and over twothirds were used in either one or two construction types,” and as a result, their vocabulary of verbs consists “totally of verb-speciﬁc constructions with open nominal slots”. To resolve the controversy between the two opposing schools of thought, more research is required on how paradigmatic gaps arise in a language, and how speakers treat them. The language that will be the focus of this project is Chimwiini, an endangered Bantu language spoken in Barawa, Somalia. The language is interesting because it has several very regular and productive rules for forming the past tense which sometimes compete with each other. A result is that one often cannot tell based on the form of the verb root which past tense rule the verb will follow (or indeed if it will be an exception to all the rules). This study will have several goals: • Complete a basic statistical analysis fo Chimwiini verbs based on Kisseberth and Abasheikh's 2003 lexicon of the language to determine the nature of past tense rule formation and the frequency of exceptions. • Design an online survey for native speakers that will test for paradigm gaps in the past tense of infrequent and nonce (invented) verbs in the language. The design will follow Albright (2003) where the similarity of output between speakers is synthesized with their confidence in their responses.• Results will be compared to ongoing work, such as that of Yang et al. (2010) who suggests that there is a mathematically specific threshold for the number of exceptions a productive rule will tolerate before it is no longer productive.