The linguistic encoding of landscape in Lokono

This thesis investigates how landscape elements are expressed linguistically in Lokono (Arawakan). As a background to the analysis, an account of language vitality and a description of the grammar of space are given. The linguistic analysis of landscape starts with landform terms, which form a system of non-lexicalized phrases sharing a generic term 'horhorho' (landform). The findings reveal that landscape vocabulary is organized on partonymic and spatial relations. Second, I discuss vegetation terms, which distinguish terms for dry and wet areas of vegetation. The results point to the body of environmental knowledge encoded in such vocabulary and the importance of socio-cultural practices in shaping the system of landscape classification. Third, I focus on place names, discussing their morphosyntactic, sociolinguistic, referential, and semantic properties. Lokono place names form a language-internally definable class, identified by the proper/generic distinction and the what/where distinction—a type of noun categorization discussed in the remaining chapters. What-nouns encode entities that are prototypical Figures, while the where-nouns encode entities that are prototypical Grounds. The distinction is found in the cognitively universal directionality component of the spatial expression and is based on the ontological properties of the referents. A comparative study of the distinctions in three languages identifies a hierarchy of nouns illustrating the likelihood of a noun being categorized as a what- or where-noun. The cline is analyzed in the light of the cognitive geography theory of objects and places, shedding light on which ontological properties of landscape are grammaticalized in language structure