This paper describes efforts to annotate a dataset of verbs in the Iroquoian language Kanien’kéha (a.k.a. Mohawk) using the UniMorph schema (Batsuren et al. 2022a). It is based on the output of a symbolic model - a hand-built verb conjugator. Morphological constituents of each verb are automatically annotated with UniMorph tags. Overall the process was smooth but some central features of the language did not fall neatly into the schema which resulted in a large number of custom tags and a somewhat ad hoc mapping process. We think the same difficulties are likely to arise for other Iroquoian languages and perhaps other North American language families. This paper describes our decision making process with respect to Kanien’kéha and reports preliminary results of morphological induction experiments using the dataset.

Recent work using word embeddings to model semantic categorization have indicated that static models outperform the more recent contextual class of models (Majewska et al, 2021). In this paper, we consider polysemy as a possible confounding factor, comparing sense-level embeddings with previously studied static embeddings on both coarse- and fine-grained categorization tasks. We find that the effect of polysemy depends on how one defines semantic categorization; while sense-level embeddings dramatically outperform static embeddings in predicting coarse-grained categories derived from a word sorting task, they perform approximately equally in predicting fine-grained categories derived from context-free similarity judgments. Our findings highlight the different processes underlying human behavior on different types of semantic tasks.