Although the regular nature of language acquisition is often assumed to be a driving force for language change, such changes can introduce apparent irregularities. In this paper, we use a well-known formula for rule productivity, the Tolerance Principle (Yang 2016), to study two inflectional classes in Old Icelandic, masculine a-stem nouns and u-stem nouns, which underwent irregular inflectional changes in the history of Icelandic. We use a tagged corpus of Old Icelandic narrative texts to assess the potential input that children acquiring the language would have been exposed to, and we use the Tolerance Principle to evaluate the productivity of the underlying rules. Our results show that, despite initial appearances, both inflectional classes are in fact predicted to be unstable and thus likely to change. This paper thus shows that apparently irregular changes in the history of Icelandic can be understood from the perspective of language acquisition. In doing so, the paper provides evidence that the Tolerance Principle can be productively applied to pre-modern languages (as previously argued in Kodner 2019, 2020).
Snorrason et al. (Fri,) studied this question.