This paper investigates the ability of neural network architectures to effectively learn diachronic phonological generalizations in a multilingual setting. We employ models using three different types of language embedding (dense, sigmoid, and straight-through). We find that the Straight-Through model outperforms the other two in terms of accuracy, but the Sigmoid model's language embeddings show the strongest agreement with the traditional subgrouping of the Slavic languages. We find that the Straight-Through model has learned coherent, semi-interpretable information about sound change, and outline directions for future research.

Slavic and Baltic show a curious way of overt definiteness marking by means of a specific type of adjective inflection. It arose when the adjective merged with a definiteness marker of pronominal origin. This paper aims at a comprehensive explanation of the rise of the definite adjective morphology in the former branch. First of all, it is argued that contrary to what has been claimed, there is no need to posit loss of trapped morphology to plausibly account for the definite adjective inflection. It can be explained by a series of well-established processes, i.e., sound change, reanalysis, and analogy. An interplay of these processes led to an increase in the phonological fusion of the adjective and the definiteness marker and ultimately resulted in their merger. In discussing the involved changes and in identifying factors that may have motivated them, the paper contributes to the general study of how new inflectional morphology emerges from the coalescence of two formerly independent elements.

No abstract

It has long been noticed that the Slavic superlative prefix nai‐/naj‐ comprises two components: *na + *i. The former can be identified with the preposition Sl na ‘on(to)’ which developed an intensifying meaning when used as a prefix. The origin of the second component, on the other hand, has not been determined satisfactorily so far. This paper argues that it can be identified with the Slavic definiteness marker. Drawing upon a typologically common pattern, it is supposed that at a pre‐stage of Slavic superlative was derived from the comparative with the aid of a definiteness marker. Definiteness of a noun phrase in earlier Slavic and its neighbouring branch Baltic was denoted by adding a definiteness marker of pronominal origin to the right edge of the adjective. As data from Old Lithuanian show, this morpheme was originally not bound to the position following the adjective. When an adjective/participle was preceded by a prefix/preverb, it could be placed between these two elements. I argue that this state of affairs is also reflected in the Slavic superlative prefix naj‐/nai‐. Based on the proposed scenario, the article discusses certain general issues related to the externalization of trapped inflectional morphology and the phenomenon of endoclisis.

No abstract