The devil take whom? Interpretation of implicite objects in Hungarian, midway between grammar and context Hungarian displays a two-dimensional agreement system with inflections indicating either only the person and number of the subject or also the person of a definite object which is acted upon by a higher-ranking subject. This means that objects in the subject/object relations S(ubject)2(nd person) -O(bject) 1, S3-01, S3-02 remain 'morphologically implicite' (cf. Abondolo 1988: 88-94 for a detailed account). Contrary to what one might expect, 1st and 2nd person objects in these relations must not be expressed syntactically either, but in a vast majority of the cases, they are ornitted from their clauses. Experience shows also, that in many cases, the referent or person of a missing object of this sort can be identified by the addressee in contextfree clauses, with the choice dependent of ei ther grammatical, or pragmatical considerations, chief among which is whether the referent of the missing object can or cannot be coindexed with one of the non-missing NPs involved in the sentence, without regard for the obligatoriness or optionality of that NP or the syntactic unit it belongs to. In other words, the more details the description of an action or state contains, the more likely it is for both ways of interpretation to be successful. The grammatical conditions for an NP to be or not to be coindexed with another modifier of the same verb, are either incorporated by the vocabulary (e.g. selectional restrictions on verbs, constraints on complement structures of particular lexical items etc.), or can be predicted by the virtue of the 'disreference rule', i.e. a universal constraint against using pronouns (other than reflexive or reciprocal ones) and NPs for referring to the same entity twice within a single clause. Thus, in sentences like János rávett arra, hogy türelmesen várjak 'John gat me to wait patiently' (literaIly 'John gat [01 or 02] (into the state) that 1 wait patiently') or János elvisz tőled 'John is going to take me away from you' (literaIly 'John takes [01 or 02] away from you'), the choice between o1 and 02 is determjned by the grammar. By pragmatic interpretation I mean search for motivation why in S3-0l/02 relations one of the two object referents can be thought of as affected by the change described in the sentence, whereas the other one can not. In this strategy, a number of factors must be considered, coindexation being only one of them. Other main factors include: the semantics of the predicate verb; the actual semantic roles played by it's modifiers; the reference of the (non-missing) NPs belonging to or representing these modifiers; the characteristics of the possessive relations exhibited in the relevant modifier NPs (whether the possessor has a unique access to his/her own property); the characteristics of the relation between the object and the causing participiant represented by the subject (whether one of the possible object referents is dependent on the subject referent in terms of the lexical content of the predicate). Informations of this kind serve to allow the addressee to identify easily the object implied by the sentence, provided that explanation or precondition for the occurence of the actual change appears overtly in the sentence and is associated with one of the rival referents. Thus, in sentences like János utál, mert megmondtam az igazat 'John hates me (lit. hates [Ol or 02]), because I told the truth'; János kárpótolt a szenvedésmért 'John compensated me (lit. compensated rO 1 or 02]) for my suffering' ; Meglepett a viselkedésem 'My behaviour surprised you (lit. surprised [Ol or 02])'; János addig kerget, mígnem összeesem 'John will pursue me (lit. pursues [Ol or 02J until I collapse' and János kirúgott az állásomból 'John kicked me (lit. kicked [Ol or 02] out of my job', the choice is determined by pragmatic factors. The frequency of omitted non-third person objects in Hungarian can be accounted for by a combination of grammatical, pragmatical and other means by which ambiguity is avoided, and can be considered to be one of the reasons for the different distribution of possessive suffixes and complement types in Hungarian and Finnish (see examples in (19) and (20)).