Uab Meto subject-agreement prefixes and the meaning of “adjacency” in allomorphic conditioning

Speaker: Tyler Lemon (https://sites.google.com/view/tyler-lemon)

Abstract: Nearly all verbs in Uab Meto (Austronesian; Indonesia/Timor-Leste) have subject-agreement prefixes from one of two sets: "asyllabic" C- prefixes and "syllabic" CV- prefixes.  The choice between the two is conditioned by a variety of factors.  Building on previous work (Edwards 2020, Tan 2023) on the language and data from my own fieldwork, I affirm the descriptive generalization that phonological factors like the shape of the stem (V-, CV-, or CCV-initial and number of syllables) are the primary determinant of the choice of allomorph, but morphosyntactic factors like causativization also play a role.  Then I provide a theoretical analysis of the allomorphy within the framework of Distributed Morphology (Halle & Marantz 1993) and discuss a couple of aspects of this allomorphy that lead to unconventional conclusions about possibilities for allomorphic conditioning.  Firstly, non-null causative v is a suffix, separated from the agreement prefixes by at least the root, but it conditions the syllabic agreement prefixes despite the absence of linear adjacency.  However, with Pruning (Embick 2015), causative v can be made structurally adjacent to the agreement prefixes on Agr, suggesting that this sort of adjacency is sufficient for allomorphic conditioning.  I will also show that Agr is higher in the structure than causative v, which demonstrates the possibility of inward-sensitive morphosyntactically conditioned allomorphy (Harizanov & Gribanova 2014).  Secondly, verb stems that are three syllables or longer condition the asyllabic agreement prefixes.  In many cases no individual morpheme in the stem is 3+ syllables on its own, but the span (Svenonius 2012, Merchant 2015) of Voice, v, and the root can collectively create the relevant conditioning environment for Agr.  In this case, Agr is structurally adjacent to the node containing Voice, v, and the root.  Ultimately, these allomorphy patterns suggest that structural adjacency, rather than linear adjacency, is what matters for allomorphic conditioning.