This talk deals with some of the challenges field linguists face during the grammar writing process, particularly in the context of working with little-known, underdescribed languages. In addition to collecting and processing substantial amounts of data, there are many decisions one must take. Who is the intended audience? If the language is endangered, the challenges multiply, as there might be expectations from both the speech community and the academic community. Should the grammar include information with enough detail to be useful for language development projects? Should it easily allow for comparative work between languages of the family? Should the analysis be placed in the context of crosslinguistic patterns and existing typologies? Should the grammar address current questions on linguistic theory? It seems that there is certain consensus that grammar writers should allow a language to tell its own story, rather than trying to satisfy a particular theory (see, for instance, Rice 2005, 2006; Dixon 2009). To that extent, it would be difficult for the internal organization of a grammar to follow a general, fixed, and predetermined outline. In sum, “each language demands its own strategy of presentation” (Rice 2005: 400). Yet the pedagogical uses of grammars cannot be overlooked. Reference grammars are often used as tools and sources of data to develop an understanding of the diversity of grammatical constructions in the world’s languages. As every grammar is organized differently, and frequently contains idiosyncratic terminology (Haspelmath 2010; Croft 2016), obtaining information from them is an acquired skill.

This talk takes the domain of possession in Kukama-Kukamiria (Peruvian Amazon) as a case in point for discussing methodological and theoretical issues when presenting empirical facts that reveal multiple connections. Kukama-Kukamiria does not have a grammatical construction specifically dedicated to encoding possession. Possession is inferred from several other constructions providing further evidence of a conceptual link between categorization, location, existence, and possession. We will discuss the strategies employed in Vallejos (2016) to present these empirical facts in a cohesive way to the widest possible audience.