Answers are grounded in this paper and cite their sources.
Historical linguists observe that many fusional (unsegmentable) morphological structures developed from agglutinative (segmentable) predecessors. Such changes may result when learners fail to acquire a phonological alternation, and instead, “chunk” the altered versions of morphemes and memorize them as underlying representations. We present a Bayesian model of this process, which learns which morphosyntactic properties are chunked together, what their underlying representations are, and what phonological processes apply to them. In simulations using artificial data, we provide quantitative support to two claims about agglutinative and fusional structures: that optional morphological markers discourage fusion from developing, but that stress-based vowel reduction encourages it.
No takes yet. Share an insight, caveat, or question.
Elsner et al. (2020) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: