Change and conservation

In Late Egyptian, with the rise of a full-fledged article-system, definiteness in possessive construction is open to be tested at last. By that time, however, the analytic indirect pattern is the only productive operation. At the same time, the genitival adjective show no more (number and gender) agreement with the possessed having grammaticalized in a single form. The basic pattern of Late Egyptian possessive construction is therefore pA A n pA B, where pA stands for the whole class of Late Egyptian determiners. In what follows, the pattern will be listed and examined that preserved in some way the signs of an earlier construct state formation.

I. Pattern pA A B

In this pattern a definite article precedes the whole construction. These cases, however, are supposed to be compounds rather than structure produced by a syntactic operation.

(104) pA wHa Apd [LRL 20,8. Černý – Groll (1978: 75) Ex 232.]

DEF.SG.M catcher bird

‘the fowler’

This pattern is very similar to the later Coptic compounds of the xoumise ‘birthday’ type, whose morphological make-up has been claimed to conserve the original construct state formation (with xou- corresponding to the absolute use of the word xoou ‘day’). These compounds are obviously lexicalized, and constitute new words in the lexicon. On occasion, the first member in such compounds survives only as a mere nominal prefix, e.g.

mdt > mNt-. Articles may be freely attached to these lexemes, whose definiteness depends on the wider syntactic context rather than on the internal structure of the word. (For similar lexicalized compounds, see examples in: Till 1961: §120, §§123-24, §130, §§133-140;

Vergote 1973: §87, §101, §103; See also Layton 2000: §109, §112 and Reintges 2004:

§3.1.2) Of course, the exact time when the lexicalization of the individual cases took place is unknown.

II. The pattern A pA B

This is the pattern that resembles the most a productive construct state formation, as both the linking element n- and the definite article on the head noun are missing.

(105) a.wy pA nTr [LRL 1,8. Černý – Groll (1978: 75) Ex 229.]

hand.DUDEF.SG.M god

‘God’s hands’

In these constructions, however, only a well-defined closed set of nouns occurs, which are practically the same nouns that still co-occur with pronominal suffixes, contrary to the standard Late Egyptian grammatical standard that makes use of possessive articles. These nouns are characterized in Late Egyptian text-books as nouns that cannot take an article.

In Friedrich Junge’s description (Junge 1996: §2.1.3(2)), these are designations for inalienable objects (“Bezeichnungen unveräußerlicher Gegenstände”), such as body parts, terms related to persons (e.g. name, condition), kinship terms, certain topographic designations, etc. A shrinking subset of these nouns behaves likewise in Demotic and Coptic (Simpson 1996, 81-82; Till 1961: §188; Layton 2000: §138.). So these nouns are unable to take an article, but they are probably better to be characterized as lexically marked for forming status constructus (or pronominalis) with their possessor, instead of undergoing the productive analytic operation.

If the Egyptian construct state was similar to the Semitic one the absence of article may be accounted for in a natural way. The false impression that it is the head noun itself that does not tolerate the article rather than the whole construction as a whole follows from the fact that these nouns never appear independently outside the possessive. They are relational nouns, i.e. they usually require an additional argument, a possessor expression, to be related to. It is this semantic (and consequently syntactic) boundness that made them be conserved and become irregular remnant in later Egyptian. Leo Depuydt also pointed out (1999: 281-282) that the resistance to the analytic constructions on behalf of nouns denoting body parts and a few other inalienables may be due to fact that nouns denoting body parts are frequently used, and, at the same time, they are often used with suffix pronouns. He also examined the strange “split” genitives in Demotic and Coptic, in which synthetic and analytic designs co-occur, and explained the data by the conflicting forces of the analytic shift and the resistance by the surviving synthetic forms to analysis. As a result, when construct state formation definitively disappeared as a productive mechanism, the absolute forms of this range of resistant nouns were no longer available, which gave

rise to the emergence of these irregular split genitive types. I am in full agreement with his line of reasoning as well as with his proposal that the n- in Coptic compounds such as rMNkhme ‘Egyptian’ might have the same origin.

Examples comparable to (105) can be found in Demotic as well: in (106) only the second member displays an article, but the whole noun phrase is interpreted as definite.

The noun in this type of pattern is again a noun denoting a body part.

(106) a pA nTr [‘Onchsheshonqy 11:23; Johnson (1987: 44) E6a]

hand DEF.SG.M god

‘the hand of God’

The convincing syntactic tests provided by Robert Simpson in his book about the grammar of Ptolemaic decrees (1996: 80) demonstrate that these nouns, although incompatible with determiners indeed, do occur in positions where zero determination is ungrammatical, e.g.

in durative subject position (cf. the test contexts for definiteness in section 2.4). As far as I see, this may only happen because these noun phrases are construct state formations, and accordingly definite by their inner structure.¹²⁴

III. The pattern A n pA B

This pattern is discussed by Leo Depuydt (1999: 294-295) and characterized by him as a hybrid formation: the appearance of the linking element (n-) between the members is an analytic feature, while the absence of the article on the first noun is the survival of a synthetic form. Depuydt himself is doubtful of the existence of an A pA B type (Pattern II above) because of the frequent omission of the grapheme for n in Demotic writing (1999:

292), which causes a great deal of difficulty in many other parts of Demotic grammatical investigations indeed.

To sum it up, it is rather reasonable to assume that a construction of the so-called construct state type did exist in the Egyptian language. This is justified by syntactic factors such as the survival of the Late Egyptian and Demotic A pA B and A n pA B patterns as well as by its morphological traces in Coptic lexicalized compounds. The productivity of this construct state-like direct genitive was somehow limited as early as in Old and Middle

124 From a typological point of view, it is not uncommon that if a language has more than one possessive construction, their distribution is conditioned by an alienability split. In Maltese, for example, only

inalienable nouns form a construct state and the possessor is introduced by a preposition elsewhere. For the alienability split in general, see: Koptjevskaja-Tamm (2001: 965). As it will be shown in the next section, in certain varieties of Coptic (e.g. Bohairic) considerably differ from Sahidic since alienability seems to play an important role in syntax, in the distributional rules of possessives.

Egyptian, and the construction was gradually replaced by the analytic pattern of the indirect genitive. What is really interesting to see is that the Coptic pattern A, which is the successor of the analytic type formally, functionally got reduced to the syntactic environments in which the possessed noun must be definite and strictly adjacent to its possessor. This distribution is strikingly similar to the supposed distribution of the earlier direct genitive construction. For the cases where these requirements are not met, an alternative structure emerged involving the prepositional phrase Nte-/Nta- (Pattern B).

Again, we are extremely fortunate to know Egyptian, i.e. a language documented through millennia, and thus able to follow such an interesting linguistic cycle: a formal and functional opposition first neutralized, quasi disappeared, and later functionally re-emerged in the distribution of the two Coptic possessive patterns. Furthermore, in the northern dialectal varieties of the language, the equivalent of pattern A began to decline again, at least became far more restricted in its usage, bringing about an alienability split within the possessive system along the same lines with what was observed in the earlier stages.¹²⁵

4.3 Possessive constructions in the early Coptic dialects