Bende-Farkas Ágnes

The semantics of Vala-indefinites in Old Hungarian; remarks on the meaning of free choice expressions  

The main focus of the talk is the semantics of expressions of the form vala+pronoun in OH. (MH valaki `someone', valami `something',  valahol `somewhere' a.s.o.) In OH vala+pronoun could co-occur with clausemate negation, with narrow scope (unlike in MH), and  it was not the  preferred means of introducing discourse referents that could be picked up in subsequent discourse. (In MH `valaki' a.s.o. can  be epistemic specific, and it can also introduce a novel discourse referent for subsequent anaphora.) Nevertheless there are a few attested examples of valaki contributing an epistemic-specific (and arguably scopally non-specific) discourse referent (in the Munich and Erdy Codices, and in St Margaret's legend).  

 In addition, vala+pro expressions could be relative pronouns (`Valamit parancsolsz, azt teszem' -- `I'll do what(ever) you tell me to do') or relative-DPs (`valamel' ember' instead of today's `amelyik ember', cf. MH `Amelyik diak elsonek adja le a dolgozatat, plusz pontot kap' --- `Whichever student hands in his paper first, I'll give him a bonus'). Usually, these occurred in free relatives, where the relative pronoun/`DP' is typically analyzed as a maximality operator (as a definite by some researchers, or as a universal quantifier by others, references will be supplied at the talk). Obviously, this is a puzzle and a challenge for a uniform semantic analysis of OH valaki \& co. Furthermore, relative clauses with vala-pronoun could also occur in the restrictor of quantifiers, as in the following paraphrase of an example from the Guary Codex (p.6):  

(G) OH -- normalized form  Minden valaki atyjafiat gyuloli, az ilyen ember gyilkos

     MH Mindenki, aki testveret gyuloli, (az ilyen) gyilkos

     `Everyone who hates his brethren is a murderer'

In examples like (G) the vala-expression is a run-of-the-mill relative pronoun.  

The proposed semantics for OH vala is very weak: The proposal is that it introduced a free variable, without any contextual restrictions. The lack of contextual constraints may explain both the versatility of this expression (its uses cover almost the entire spectrum of indefinites), and its persistence  (in contrast with n\'e- expressions, which in OH were specialized, and today their use is quite restricted).  Maximality effects are said to follow from implicatures accompanying FC indefinites. (Similarly to Jayez and Tovena, implicatures are assumed to be responsible for FC efects). Accordingly, the talk will conclude with some theoretical observations concerning free choice indefinites (e.g. English any, French quelque, un quelconque).