The peculiar behavior of vowel hiatus resolution in Biloxi provides a useful test case with which to consider the relative merits of two prominent theories of morphologically conditioned phonology (MCP): Cophonology (Inkelas and Zoll, 2007) and Constraint Indexation (Pater, 2009). Biloxi, an extinct or critically endangered (sources vary) language of the Siouan language family, has received little to no phonological study since the 1974 grammar this project is based on (Einaudi, 1974). Vowel-vowel sequences are generally avoided in Biloxi. When vowel-vowel sequences are formed by affixing, this sequence is repaired by deleting the first vowel, as in (1) /ohi dani-axehe/ → [ohi danaxehe] ‘three sitting on ten’ (=13). When such a sequence is formed by the dative prefix /ki-/ and a vowel-initial verb root, glide epenthesis occurs instead, as in (2) /Ø- ki- ε -tu/ → [kijetu] ‘they said to him’. Yet another exception is observed when /ki-/ specifically precedes the verb root /õ/ ‘to make, do’, resulting in glide fortition and an additional /k/ in the surface form, as in (3) /Ø- ki- õ -tu/ → [kikõtu] ‘they made for him’. This analysis lends support to whichever model of MCP better accounts for the complex vowel hiatus pattern of Biloxi. As an understudied language, a framework’s ability to fit Biloxi, in addition to previously considered languages, gives that model greater explanatory power. Avenues for future research include the consideration of a learnability model for Cophonology as compared to that proposed by Pater (2009) for Constraint Indexation. An updated grammar of Biloxi with more detailed phonological analysis would also extend the scope of modern studies of the language. This work brings new attention to the Biloxi language, and contributes to the ongoing debates over morphologically conditioned phonological processes.

The focus of our research is on digitizing linguistic material from two understudied and endangered Brazilian Amazonian languages, Panãra and Kawaiwete. Panãra (ISO code: kre) is a Northern Jê language with about 600 speakers, and Kawaiwete (ISO code: kyz) is a Tupí-Guaraní language with about 1,000 speakers. Our project focuses on the digital archival of linguistic materials from these communities, the creation of a dictionary, and the study of the morphophonological grammar of these languages. This talk describes the process of data digitization thereby making it usable to a variety of research questions to be answered in future work. Data elicited by Professor Lapierre from field trips in 2018 and 2019 to Kawaiwete speaking communities, and from 2015 to 2019 to Panãra speaking communities forms the foundation for this project. Materials utilized for this research project include elicitation recordings and transcriptions, and field notes obtained through in-situ fieldwork. The initial stage of our research was digitizing handwritten field notes over the course of ten months so the notes could be fully searchable in an online database. We transcribed a total of 16,128 words for the Panãra notebooks and 12,860 words for the Kawaiwete notebooks. The second stage of the project, which is our current stage, is entering lexical items into FieldWorks Language Explorer (FLEx), a software that stores and analyzes lexical and grammatical information. We currently have 351 words entered into the Panãra database, and plan on entering sentences in order to analyze parts of their grammar. The work of this project responds to the lack of an effective writing system in Panãra and the trending language shift towards Portuguese in Kawaiwete. A digital dictionary will help both the creation of written materials by Panãra speakers and the preservation of the current Kawaiwete language.

This presentation discusses a unique grammatical phenomenon surrounding person marking on transitive verbs in Kawaiwete, an endangered Tupi-Guarani language spoken in Mato Grosso, Brazil. Like many related languages, Kawaiwete exhibits complex variation in how it conjugates verbs (i.e. changes verbs to fit grammatical context). Transitive verbs, i.e. verbs that need both a subject and an object, require a prefix on the verb. This prefix denotes the role of the sentence’s component parts, called arguments. Whether the subject prefix or object prefix is selected depends on the grammatical role of the arguments and word order of the sentence, which is flexible. I analyzed data collected by Professor Lapierre by comparing sentences with first and third person arguments (i.e. “I/me” and “he/him”). I surveyed (i) the verbal prefixes used, and (ii) the order of the sentence’s arguments. Consider sentences (1a) /miara jemiwara je a-esat/ ‘I see the jaguar eating’, (1b) /miara jemiwara je w-esat/ ‘I see the jaguar eating’, (2a) /a-esat je miara jemiwara/ ‘I see the jaguar eating’, (2b) */w-esat je miara jemiwara/ (ungrammatical). When /je/ ‘I’ appears before /-esat/ ‘see’, its role as the subject is unambiguous, meaning the verb may take either the prefix /a-/, matching with the subject (1a), or the prefix /w-/, matching with the object (1b). However, when /je/ does not precede /-esat/ (2a), the verb must take /a-/, marking /je/ as the subject (2a). This implies that in situations where the role of the sentence’s arguments is unambiguous due to word order, the verbal prefix may select to match either of the verb’s arguments. Crucially, when word order results in ambiguity, the verbal prefix must match with the subject to disambiguate the sentence. This finding suggests that Kawaiwete verbs mark more than simply the verb’s subject or object, a unique phenomenon rarely observed.