Grammaticalization processes are often construed as the shift from less grammatical to more grammatical (Kuryłowicz 1965), often involving semantic bleaching (Sweetser 1988) and segmental erosion (Heine & Reh 1984). These changes are often illustrated with pieces of language that develop from lexical to grammatical meaning, which may then undergo further grammaticalization processes to develop more discourse-level, pragmatic functions. These forms, in turn, may develop into grammatical markers, suggesting an implicational hierarchy in the semantic changes involved in processes of grammaticalization: from semantic to pragmatic to less semantic-pragmatic (Traugott & Heine 1991: 5). Evidence from an Otomanguean language, Sà’án Sàvǐ ñà Yukúnanǐ (Yucunani Mixtec), indicates that things may be more interesting. The language contains a large inventory of forms historically related to a noun ñà’a ‘thing’. Their uses are examined here in a corpus of unplanned naturalistic data from 14 recordings from 7 different speakers, totaling 1 hour and 53 minutes, covering different genres. The recordings were transcribed using ELAN (ELAN 2022), and Praat (Boersma & Weenink 2022) was used in later stages to segment the texts into Intonation Units (Chafe 1980, 1994; see Himmelmann et al. 2018) and visualize pitch and duration.These forms related to ñà’a ‘thing’ serve six syntactic functions (noun, classifier, demonstrative pronoun, dependent pronoun, relative pronoun, complementizer, and adverbial subordinator) and three discourse functions (hesitation marker, floor-keeping device, and focus marker). This complex web of grammaticalization pathways can be grossly simplified as follows: 1) from lexical to grammatical; 2) from grammatical to further grammatical specialization or to discourse markers; 3) from discourse markers to grammatical markers; and 4) from grammatical markers back to discourse markers. When analyzing the discourse uses of these forms, speakers’ identification of the floor-keeping device and the hesitation marker as article-like—i.e., translating it into Spanish as ‘el’ in transcription sessions— points to a grammaticalization process in which the generic classifiers came to be used cataphorically as a placeholder while the speaker thinks about what to say, similar to the Spanish demonstrative este (Graham 2013). This process would have given rise to two of the three discourse uses of these forms. In addition, their distribution in Intonation Units as well as other prosodic similarities between complementizers, relative pronouns in non-restrictive relative clauses, hesitation markers, and floor-keeping devices suggest that these structures are not that dissimilar from one another. While the discourse markers can be omitted, the complementizer and the relative pronoun cannot. This signals further grammaticalization, moving from the discourse level back to grammar. Finally, the language also contains a focus marker that is the result of a grammaticalization process whereby the copula fused with the complementizer, moving once again from grammar to discourse. The study of these discourse markers in Sà’án Sàvǐ ñà Yukúnanǐ can offer some valuable insights for the development of pedagogical material for the language, as it reveals discourse uses that were previously undescribed. It also illustrates ways in which prosodic structure and discourse strategies may give rise to syntactic structures, underscoring the importance of conducting research on discourse markers.