fusionalsetest

Fusionalsetest is a statistical metric and accompanying methodology proposed for evaluating the fusional character of a language’s morphology. It aims to quantify how much a single morpheme encodes multiple grammatical categories, a hallmark of fusional languages. The method operates on a corpus with annotated inflected forms or on outputs from a reliable morphological analyser. By identifying morphemes and aligning them across paradigms, it computes a fusionality score that reflects the degree of morphologically fused information in the language.

Two core indices are used: the Fusion Rate (FR), the average number of grammatical features encoded per morpheme across the analyzed forms, and the Paradigmatic Fusion Index (PFI), the proportion of forms in which a single affix marks more than one feature. These are combined into a Fusionality Index (FI), normalized to the range [0,1], where higher values indicate greater fusion. The exact aggregation can be tuned to specific research goals, such as emphasizing bound morphemes or cross-paradigm consistency.

Applications of fusionalsetest include supporting typological classification, cross-linguistic comparison of morphological systems, and diachronic studies of morphological change. In computational linguistics, the metric can inform the design of morphological analyzers and influence language modeling, tokenization, and feature selection for languages with complex inflectional systems.

Limitations include sensitivity to data quality and morpheme segmentation, potential confounds from derivational morphology, and challenges in ensuring cross-language comparability. Proper normalization and transparent methodological choices are essential for reliable comparisons.