Journal of Uralic Linguistics - Volume 4, Issue 1, 2025

In this paper I examine the empirical properties of so-called factive verbs in Hungarian, with an emphasis on various syntactic, morphological and pragmatic factors that induce factivity alternations with the same predicates. On the basis of these facts I argue that factivity is not a single property, but a set of properties: a speaker-related evidential veridical inference, a subject-related evidential veridical inference and the projection of the speaker-related veridical inference from the scope of entailment-cancelling operators. Although a default pragmatic principle favours the co-occurrence of these properties, morpho-syntactic and contextual cues can signal non-speaker-veridical or non-projective readings. The lexical meaning of so-called factives does not entail the truth of the complement, only suggesting that the attitude is based on a certain type of evidence.

A domain exhibits selective opacity if it is transparent to some operations but opaque to others. The Williams Cycle (Williams 1974, 2003, 2013) makes a connection between opacity and the structural size of embedded clauses: Syntactically smaller embedded clauses are transparent to more sub-extraction types than larger ones. This paper argues that various restrictions on cross-clausal movement in Hungarian can be accounted for in terms of the Williams Cycle. This novel support for the Williams Cycle is welcome since Hungarian is known for exhibiting many movement types landing at various heights (e.g. Horvath 1981; É. Kiss 1987; Puskás 2000; Surányi 2008; Egedi 2021). Bounding can be overt: a long movement type landing lower in the matrix left periphery is blocked by constituents occurring high in the embedded left periphery. But bounding can also be covert: Even if clause-types (e.g. subjunctive vs. tensed) are not distinguished in the left-periphery, assuming these are of different sizes, their opacity-differences can be traced back to size-differences.

Hungarian reflexives and 1sg and 2sg object pronouns have a possessive origin, and, together with 1sg and 2sg possessive noun phrases, they exhibit accusative case drop to varying degrees in object positions in a language that is otherwise consistently accusative marking. This paper argues that there is a systematic correlation between the frequency of accusative drop and the degree of grammaticalization in the possessive domain: the more grammaticalized possessive structure a reflexive or an object pronoun has, the less likely accusative case is to be spelled out. Following É. Kiss & Mus (2022), accusative drop is analysed as an instance of exaptation, that is, a case of repurposing possessive morphology as accusative marking. The paper presents evidence from diachronic and synchronic corpora to argue that this analysis makes the right predictions both for the historical changes that have taken place in this domain, and for the fine-grained variation in accusative use on reflexives in the synchronic system.

In both nominal and verbal paradigms of Hungarian a vowel may occur between the stem and the suffix. The presence and the quality of this vowel is determined by a wide array of phonological, morphological, lexical, even syntactic and semantic factors. In this paper we argue that this vowel functions as a paradigmatic class marker and that this status is responsible for the differences in the behaviour of linking vowels and other vowels in suffixes.