The submerged genitive in Old Prussian

This paper is devoted to the Old Prussian phrase ʃwaiāʃmu ʃupʃei buttan ‘to his own house’ (Enchiridion, III 876). Far from being simply the result of a syntactic error, the genitive ʃupʃei ‘of oneself’ can be recognized as the reflex of an archaic syntactic pattern, the “submerged genitive”, which has left numerous traces in Baltic and other Indo-European languages (Slavic, Greek, Latin, Old High German).


Introduction
The syntax of Old Prussian is notoriously difficult to analyze in any detail as a result of the fact that its most significant documents are word-for-word translations from German and provide us only limited access to the real use of the language. To overcome this difficulty, scholars generally attach particular importance to all those passages where non-trivial divergences can be observed between the German and the Old Prussian text, with the hope that these divergences may reveal linguistic features genuinely rooted in Old Prussian. Needless to say, this principle of analysis presents serious limitations, because it may happen that the diverging Old Prussian text simply shows a scribal mistake or misunderstanding, without any foundation in the language. This is not a reason for discouragement, however, at least not completely. A close examination of diverging micro-contexts can sometimes give us a glimpse of interesting features of Old Prussian syntax. The aim of this paper is to call attention to one such feature, which can be called "the submerged genitive". It is offered to Axel Holvoet in recognition of his outstanding contribution to Baltic linguistics.
As far as grammatical agreement is concerned, the Old Prussian emphatic adjective sups or subs 'self' generally behaves like the corresponding adjectives of Lithuanian pàts and Latvian pats 'self', both of which agree with the noun or the pronoun to which they are linked. Instances from Old Lithuanian (Mažvydas, Daukša) are the following: • There are two possible explanations. In (2), it is possible that the nominative ʃups refers to the subject of the sentence emnes 'name', not to the reflexive pronoun ʃien. But this explanation cannot work for (3), where the subject is of neuter gender (sta nt 'it') or of mixed gender (kermens m bhe krawia f 'the body and the blood'). An alternative explanation could be that the unexpected nominative subs is an imprecise rendition of the corresponding form in the German original selbs (Modern German selbst), used in both instances as an adverb without clear case marking. A more problematic instance is III 49 16 , where the German adjective ʃelbs, apparently used in reference to a dative (von ihm), is translated in Old Prussian by an obscure form ʃubbai (ex. 4): (4) Old Prussian: The form ʃubbai cannot be analyzed as an accusative (in reference to tennan 'him') nor as a nominative (in reference to the masculine rīks 'kingdom'). The most likely explanation is that it is based on the understanding of German ʃelbs as an adverb, rendered in Old Prussian by a form distinctively marked by the adverbial suffix -ai (cf. labbai 'well').
In the last-mentioned instances (ex. 2-4), an unmarked German form (ʃelbs) is either translated in Old Prussian by a nominative (subs) or understood as adverbial (ʃubbai). This cannot have been the case with ʃupʃei (III 87 6 ), clearly marked as a genitive. Trautmann (1910, 210, § 112) describes its use as a "mixed construction" (Germ. gemischte Konstruktion) and adds (1910, 268 § 223) that "the passage is not necessarily to be recommended, since we expect swaisei = Lith. sawo" (Germ. die Stelle ist nicht unbedingt zu loben, da wir "swaisei" = lit. "sawo" erwarten). This qualification is too imprecise to be really useful to understand the syntax of ʃupʃei. A more accurate explanation is needed.
When dealing with Old Prussian, it is always necessary to start with the German substrate which constitutes the basis for the Old Prussian translation. Old Prussian ʃupʃei renders the German adjective eigen 'own', which, in this context, is not clearly case-marked: this could have been instrumental in the use of an unmarked form in Old Prussian, but can hardly account for the choice of a marked genitive. From a semantic point of view, the specificity of an adjective 'own' (Germ. eigen) is that it can be understood as the possessive form corresponding to the emphatic adjective 'self' (Germ. selbst). 'Own' means 'of oneself', just as 'my' means 'of me', 'your' 'of you', etc. The striking point is that 'own' semantically puts the emphasis on the possessor, but formally agrees with the possessee, exactly in the same way as a possessive adjective like 'my' or 'your' refers to the possessor, but formally agrees with the possessee. These examples show an incorrect use of the adjective sups / subs 'self' (emphasis on the referent) instead of 'own' (emphasis on the possessor of the referent). We are thus confronted with two different solutions to the same problem of translation: German eigen 'own' is rendered either by the genitive ʃupʃei 'of oneself' (ex. 1, 5-6) or erroneously by the adjective ʃups / ʃubs 'self' (ex. 7-8).
What these two solutions have in common is the impossibility of translating Germ. eigen directly: this suggests that Old Prussian did not have an adjective 'own'.

Lithuanian and Latvian
In this respect it is interesting to note that the other Baltic languages encounter the same difficulty. In their oldest existing texts they lack an adjective 'own' and particularly in translated texts have to resort to different strategies to render it, in a way very much similar to Old Prussian. In Lithuanian a sequence like 'my own body' is routinely translated as màno patiẽs kū́nas ('the body of myself'), where the adjective 'own' is rendered by the genitive patiẽs 'self gen.sg ' agreeing with the possessive genitive màno 'of me gen.sg '. Since the possessive meaning is regularly conveyed in Lithuanian by possessive genitives (màno 'of me', tàvo 'of you sg ', jõ 'of him', jõs 'of her', mū́sų 'of us', jū́sų 'of you pl ', jų̃ 'of them'), the genitive patiẽs, fem. pačiõs 'self gen.sg ', pl.
pačių̃ 'self gen.pl ' has an overt head in the genitive to agree with. There is ample evidence for this strategy already from the Old Lithuanian texts (ex. 9-12): In Old Lithuanian, possessive genitives like màno, tàvo 'of me, of you' often compete with possessive adjectives like mãnas, tãvas or manàsis, tavàsis 'my, your'. The striking point is that even there the meaning 'own' can be rendered by the genitive patiẽs, fem. pačiõs 'self gen.sg ', which, like the Old Prussian genitive ʃupʃei, stands alone with no overt genitive to agree with. This type is extremely rare, and I have been able to find only one instance of it in the Old Lithuanian corpus: 4 sword.nom.sg 'And a sword will pierce your own soul' (= Pol. A twoia właſna duſia prʒeniknie miecʒ̇) The genitive pacʒ̇iós 'of oneself' (feminine), translating the Polish emphatic adjective właſna 'own', is linked to the possessive adjective tawa̗ 'your' (acc. sg.). It agrees with a possessive genitive ('of you') which is not overtly expressed, but can be recovered from the deep structure of the possessive adjective ('your'), thus providing an exact parallel to the Old Prussian structure ʃwaiāʃmu ʃupʃei buttan 'to his own house' (III 87 6 ). Interestingly enough, the same formulation is repeated in another passage, but with the possessive genitive: It could be assumed that (14) represents a regular type of agreement of the emphatic adjective pacʒios 'self' with the possessive genitive táwo 'of you' ('of yourself' = 'own') and that (13) has replaced the possessive genitive táwo 'you' by the possessive adjective tawa̗ 'your', resulting in an odd type of agreement ('your of self' = 'of yourself' rendering 'your own'). Or it could be argued the other way around that (13) is the basic structure, just as it is in Old Prussian, whereas (14) has restored an overt form of agreement between the possessor and its emphatic adjective. As we shall see, there is evidence for the antiquity of (13).
As a rule, Latvian uses only possessive adjectives for the first and second person singular (mans, tavs 'my, your') and possessive genitives for the other persons (gen.sg. e.g. viņa 'of him', viņas 'of her', mūsu 'of us', jūsu 'of you pl ', etc.). When possessive genitives are used, the meaning 'own' is regularly rendered by the genitive paša 'of (him)self', pašas 'of (her)self', pašu 'of themselves', agreeing with these possessive genitives: (15) Old Latvian: Tas  When the possessive adjectives mans, tavs 'my, your' (or the reflexive savs) are used, the meaning 'own' is likewise rendered by the genitive paša 'of (him) self', pašas 'of (her)self', exactly in the same way as (13) for Old Lithuanian and (1) for Old Prussian: point replaced by the possessive adjective tavs 'your'. This is very unlikely. The same probably holds true for Old Prussian. In other words, the underlying structure is a pure fiction whose existence seems to be required by the syntactic structure, but has no legal basis in any sector of reality. It would equally be unsatisfactory to claim that the underlying structure represents a cognitive reality referring to what the speaker has in his brain, because, first, this reconstruction cannot be falsified nor verified, and, second, we are unable to explain the transformation of an underlying possessive genitive into a surface possessive adjective. Since I am not willing to admit the existence of ghosts, I prefer to claim that the source of the submerged genitive is based on language-internal analogy. As a matter of fact, in the three Baltic languages, possessive adjectives always show up in a linguistic system that also displays possessive genitives. In Latvian, for example, the structure tavs paša gods 'your own honor' (with the possessive adjective tavs) is supported by the parallel of viņa paša gods 'his own honor' (with the possessive genitive viņa). As a rule, the unusual agreement illustrated by Latvian tavs paša gods presupposes the parallelism of a regular agreement like viņa paša gods. Put another way, whenever we find the submerged genitive, we may expect to find in the language the coexistence of possessive genitives and possessive adjectives.

Other Indo-European Languages
The submerged genitive is not exclusive to the Baltic languages. It also occurs in a number of other Indo-European languages, which leads us to think that it might be fairly ancient and probably inherited from Proto-Indo-European. A necessary pre-condition for its emergence is, of course, that the language under consideration has possessive adjectives, which excludes languages (like Old Irish) that use only possessive genitives. Going further, it is necessary to distinguish languages where the meaning 'own' is rendered by a special adjective (like German eigen) [type 1] and languages where it is rendered by the genitive of 'self' (like Latvian paša) [type 2]. As a rule, we would expect a privative distribution between the two types with the result that a language that has an adjective 'own' does not use the submerged genitive, and vice versa; as we shall see, this rule is not completely mandatory, because there are languages (like Ancient Greek) where we observe a coexistence of the two types.
To begin with, whereas Polish, Czech and Russian have an adjective 'own' (Polish własny, Czech vlastní, Russian собственный) and thus belong to Ancient Greek thus shows that a language can possess both an adjective 'own' and the submerged genitive if they are distinguished by a secondary distribution (here emphasis vs. reflexivity). The use of ἴδιος ídios in the general meaning 'own, pertaining to oneself' was late in Greek (its original meaning, the only attested in Homer, was 'private, personal') and certainly linked, as a drag-chain shift, to the evolution of the submerged genitive construction from the emphatic to the purely reflexive meaning.
In a well-known study based on Sorbian and other Slavic languages, Corbett (1987) has shown that the submerged genitive is a broader phenomenon which does not appear limited to the association of a possessive adjective with an expansion in the genitive. It also occurs with other types of possessive or relational adjectives, which are notoriously productive in Slavic and can likewise be developed by appositional genitives, as shown by (36) This type has caused a lot of ink to flow, both on the Slavic side (e.g. Flier 1974, Huntley 1984, Corbett 1987, Eckhoff 2011 and from an Indo-European perspective (e.g. McCartney 1919, Watkins 1967, Matasović 2011, Mendoza & Álvarez-Pedrosa 2011. For reasons of space, I cannot pursue consideration of this question in this paper, but it is clear that the submerged genitive cannot be simply dismissed as a mere syntactic error in the isolated Old Prussian example in which it survives, almost by accident.

Conclusion
The Old Prussian translation of Martin Luther's Enchiridion (1561) is often regarded as a corrupted text deeply distorted by the translation process and the desperate plight of its philological transmission. In spite of this, the syntactic structures of the Old Prussian language may be revealed in some cases by isolated micro-contexts which prima facie look like mere translation errors,