[Editorial note: [...] indicates use of Coptic text. Original script is available for viewing in the PDF format of this article.]
(CE:A101a-108b)
DIALECTS, MORPHOLOGY OF COPTIC. The existence of quite a number of differential traits in the fields of morphology and morphosyntax may serve to show that the Coptic literary “dialects” comprise not just different pronunciations and spellings, supported by slightly differing vocabularies, of the same linguistic system but, in fact, different normative systems of written communication reflecting more or less directly some of the locally, regionally, or even sometimes nationally balanced spoken idioms. To be sure, these literary dialects cannot be conceived of as mere transcriptional records of the spoken dialects behind them. One may safely assume that each of them had undergone various stages of balance and adjustment—be it through its “natural” usage as a regional or supraregional vernacular or through the exertion of some standardizing force in scribal centers—before it was found worthy of being employed as the literary standard whose specimens have survived. Yet these dialectal varieties still reveal so many diverging traits—phonological, grammatical, and lexical—that only if considered in their sustained proximity to each other can they be identified as varieties of one language. If only the records of, say, Bohairic and Akhmimic had survived as the two extremes of this continuum, one would hardly be able to treat them as “dialects” but would rather classify them as distinct, though closely related, languages.
If, as is usually done, the term “dialect” is taken to cover several more or less closely related varieties of Coptic (i.e. the varieties of Bohairic, Fayyumic, etc.), it seems that on the morphological and morphosyntactic levels, in general, there is greater similarity between dialects, but a lower degree of consistency (i.e., less standardization) within each dialect than there is on the phonological level. Quite a number of morphemic elements that would be typical of dialect D1 may be used more or less regularly in a single text of dialect D2: they are easily understood in this context not so much because they are supposed to be “known” from D1 but because they may represent basic options of the Coptic language as a “diasystem.” The higher a given phenomenon ranks in the system of Coptic grammar (or the closer it is to the fundamentals of Coptic syntax), the more does it seem to be capable of neutralization in terms of dialectal distinction, its remaining variability being influenced by communicative perspective, text type, and individual style.
As for the linguistic value of a given text or variety, what counts is not primarily its degree of accordance with any standards known from other sources (often termed “standardization”) but its degree of internal standardization, which might be more adequately termed “normalization.” If one takes a closer look at the actual dialectal varieties (i.e., grammatically homogenous corpora) of Coptic, it is remarkable to see that even in minor or marginal dialects, the degree of normalization in the morphological field is enormous. Both in terms of morphemic (syntactic) usage and the phonological representation of grammatical morphemes, the greater number of literary manuscripts and groups of manuscripts reveal a degree of normalization that is in no way inferior to their observation of general (morpheme-independent) phonological and orthographic norms. This normalized usage (or état de langue), notwithstanding all the inconsistencies so often deplored by scholars, should be one of the primary subjects of study with both major and minor varieties. External standardization, on the other hand, may be measured in terms of both the amount of manuscripts available for one variety and the amount of neighboring varieties shading into another central “dialect.” Investigations of the latter sort (for a beginning, see Kasser, 1980-1981), which seem to be most promising in particular within the multiple-centered Akhmimic/Subakhmimic southern area and the more clearly triangular Bohairic/Fayyumic/Middle Egyptian northern area, will eventually contribute a great deal to the historical understanding of the dialect situation and development in Coptic Egypt and provide a safer ground for pertinent hypotheses (which they are so much in need of).
Taking into account the large number of “supplementing” dialectal varieties of Coptic that have become known through publications during the last few decades or still await publication, and considering the deficient supply of information about the actual morpheme stock in many of the crucial text editions, a review of the morphological relationship of Coptic dialects at the present time cannot exhaust the whole scale of known varieties. As a matter of fact, comprehensive comparative analysis will have to start by grouping and classifying the smallest discernible units of texts that follow a distinct linguistic norm and joining them gradually together into natural groups (major dialects), with the crucial isoglosses and differential traits being broadly discussed and accurately accounted for. This is one of the tasks of future research. What can be done in an article of the present format, however, is to provide a selection of standard varieties that are more or less typical of the six major literary dialects accepted so far by a greater number of Coptic scholars, A, B, F, L, M, and S, and a list of traits exemplifying their isoglosses, in an attempt to point out the complexity of the task before scholars.
The varieties referred to are, for A, the AKHMIMIC “medial” group of Exodus (Lacau, 1911), Epistula Apostolorum (Schmidt, 1919), and the Strasbourg Codex (Rösch, 1910); for B, biblical BOHAIRIC (in order to facilitate roughly “synchronic” comparabili-ty, only such traits as are in concord with the usage of the “old-Bohairic” manuscripts will be accepted here); for F, some representatives of classical FAYYUMIC proper (F5, second group in Asmus, 1904) such as St. John ap. Zoega, St. Mark (Elanskaya, 1969), Agathonicus (Erichsen, 1932), insofar as they are not discordant with earlier Fayyumic proper (F4, see, e.g., Kahle, 1954, pp. 286-90); for L, the Nag Hammadi type (L6) of Subakhmimic (i.e., LYCOPOLITAN or LYCO-DIOSPOLITAN), exposing its most valuable representative in Codex I, 2, The Apocryphon of James (Malinine et al., 1968; Kirchner, 1977); for M, St. Matthew (Schenke, 1981; see MESOKEMIC); and for S, biblical (in particular, New Testament) SAHIDIC. In order to round off the picture and facilitate taxonomical operations, two other important varieties shall be added: DIALECT P, the idiom of P. Bodmer VI (Kasser, 1960) for its outstanding characteristics, and the Manichaean type of Subakhmimic (or Lyco-Diospolitan), hereafter referred to as L4, that is to say, one variety of the group symbolized formerly as L4 in Kasser (1980a, pp. 68-69, to the exclusion, notably, of Thompson’s Gospel of John, L5), for its abundant corpus, with its most normalized representative being the Homilies (Polotsky, 1934). Of all the minor varieties whose representatives have been published so far, the latter two are without any doubt the most interesting. (Perhaps somewhat “less interesting” for the network of isoglosses are varieties such as that represented by St. John, ed. Husselman, 1962 [the most important member of Kasser’s former V. now W], which does not reveal any single trait that is not shared by either F or M.)
“Morphological” traits, in the sense in which the term will be applied here, fall into three groups: (i) variables in terms of different phonological representations of the same Pan-Coptic, transdialectal morpheme, or “diamorpheme,” which may be called “diamorphemic variables”; (ii) variables in terms of a different handling of allomorphic rules, or “allomorphic variables”; (iii) morphosyntactic variables, including some idiosyncratic grammatical morphs. While the latter two items seem to be fairly conclusive, some words of explanation may be needed with regard to (i).
To begin with, it must be noted that the majority of formal grammatical devices used in given paradigms and/or for given purposes are either phonemically and graphemically invariable for all Coptic dialects (in clear contrast to the majority of lexical morphemes) or their varying phonemic/graphemic representations are conditioned by general phonological rules. Such items cannot be the subject of a special morphological comparison. For instance, the vocalization of the stressed “stem” vowel in some prepositions (status pronominales) and the stressed personal pronouns strictly follows the general rules governing the vocalization of short stressed syllables, dependent on the type of the following consonant. Thus, for example, one finds for o:[...]:
[See PDF version of this article for table.]
But with the suffix of 2nd fem. sing /’/, on finds:
[See PDF version of this article for table.]
Generally, with regard to suffixal pronouns, there is very little dialectal variation except for certain contexts (see the variables quoted below as nos. 6-10). Also, for instance, the different dialectal representations of such forms as [...], one; [...], this one; [...], if; and other full-stress pronouns and particles can be easily reduced to general phonological rules.
The situation is thoroughly different with those morphemes that regularly occur in pretonic syllables, such as those establishing the basic syntactic relations in verbal sentences or connected in some way with the conjugation system. If used in pretonic positions, morphemes occupy slots that, with regard to vowel quality, are very little, if at all, determined by transparent (dialect-)phonological principles. It is rather the morpholexical identity of the form, usually balanced by a pan-Coptic norm, that determines the quality of these syllables. Compare for pretonic e in all dialects, the circumstantial converter; [...], if; [...], because of; [...], ground; [...], sheep; [...], kindle; etc.; and for pretonic [...] in all dialects, the perfect and aorist conjugation bases; [...]; [...], honor; [...], -less; [...], toward; [...] condit. infix; etc.; but on the other hand, [...] A, L6, L4, P versus [...] B, F, M, S in certain t causatives like [...], [...], [...], [...], [...], [...], etc.; or [...] A, L4, P versus [...] B, F, L6, M, S in words like [...], prevail, and [...], treasure. If the distribution, in terms of dialects, of e versus a in a particular grammatical morpheme, say, a conjugation or converter base, is found to follow one of these latter groupings, it may well be classified as determined by some phonological (though less transparent) ratio. But if it shows a grouping of its own, different from any other phonological trait (as is the case with almost all grammatical morphemes unless they are invariable), it may cum grano salis be counted as a “morphological” trait, although it still features not the morphological “system” but its phonological representation (or the plan de l’expression).
This comparison cannot be based on the phonemic system relations within each dialect (especially for the vowel system) or the phonological rules applying for the transition from one dialect to another, since this would not provide a common basis for the comparison. Thus, its validity largely rests on the overall assumption that the phonetic (!) values of the vowel graphemes are approximately the same in all dialectal writing systems, or at least that a Sahidic [...], for example, is remarkably more similar in quality to an Akhmimic or Fayyumic [...] than it is to Akhmimic or Fayyumic [...] or Akhmimic o. This cannot be proved, even after a careful phonemic analysis of the respective graphemic systems; it simply has to be assumed.
i. Diamorphemic Variables
Diamorphemic variables are not classified in respect of the reasons for their variation, e.g., the issue of historical sameness or heterogeneity, unless this reason is synchronically to be seen as a difference in structural principles.
Basic Elements in Connection with the Conjugation.
(1) Operator of negative aorist, [...] B; [...] A, L6, L4, P; [...] F, M, S.
(2) Operator of affirmative perfect, [...] A, B, F, L4, P, S; [...] M; [...], [...] L6 (depending on the suffix chosen; but not fully normalized).
(3) Operator (initial vowel) of energetic future, [...] A; [...] B, F, L6, L4, M, P, S.
(4) Operator of conditional, homonymous with second present and coinciding with vocalization of “imperfect,” [...] A, B, F, M; [...] L6, L4, P, S.
(5) “Causative infinitive,” presence versus absence of [...] ([...]), presence B, F, L6, S; absence A; nonnormalized L4, M, P. Vocalization of Pronominal Suffixes.
(6) First singular with negative energetic future, [...] A, B, M, S; [...] L6, L4; nonnormalized F (?) (unknown for P).
(7) First singular with “causative infinitive,” [...] B, F, L6; [...] A, L4, M, S (unknown for P).
(8) Second feminine singular with possessive article, [...] M, S; [...] A, B, F, L6, L4 (unknown for P).
(9) Third plural with possessive article, [...] F, M, S; [...] A, B, L4, P; nonnormalized L6.
(10) Third plural with causative infinitive (similarly with the negative energetic future), [...] M, S; [...] A, B, F, L6, L4, P.
Formation of Qualitative (Stative) Verb Forms.
What should not be neglected in this connection is the basic morphemic change concerning the verbal lexeme, that is, that between infinitive and qualitative (or more particularly, the formation of the qualitative form: “long” or “short” form; presence or absence of final /-t/, etc.). However, there appears to be but little normalization in this field for quite a number of varieties, so that it seems impossible to give distinct specimens of variables at the present state of research. A special case is the qualitative form of [...]:
(11) Vowel quality /a/ ~ /o/ versus /e/, B, F, L6, M, P, S versus A; nonnormalized L4.
(12) Presence versus absence of /-i(e)/, presence B, F, L6; absence M, P, S; nonnormalized A, L4.
Miscellaneous Grammatical Forms of Transdialectal Identity.
(13) Postdeterminer “each,” [...] A, L6, L4, M, S; [...] B; [...] F; [...] P.
(14) Full-stress pronominal object of second plural, [...] B, F, M; [...] A, L6, L4, P; [...] S.
(15) Possessive pronoun, plural base, [...] A, F, L4; [...] B, L6, P, S.
(16) Prefix negating infinitive, [...] ([...]) A, L6, L4, M, P, S; [...] B, F.
(17) Infinitive connector; interfix of affirmative energetic future; preposition ([...] etc.), [...] B, F, M, S; [...] A, L6, L4, P.
(18) Marker (initial vowel) of “special” imperatives such as [...], [...], [...], [...], etc., [...] L6, P; [...] A, B, F, L4, M, S.
(19) Proclitic particle of epistemic condition ([...] B, [...] S, ... [...] A), second element, [...] A, L6, P; [...] B, F, L4, M, S.
(20) Interrogative adverb of place, presence versus absence of [...], [...] B, F, L6, M, S; [...] A, L4, P.
Miscellaneous “Nonidentical” Forms Filling the Same (or Partially Same) Paradigms, i.e., Lexicogrammatical Traits.
(21) Indefinite pronoun (NP equivalent in nonaffirmative contexts), [...] A, L6, L4, P, S; [...] B; [...] F; [...] (personal), [...] (nonpersonal) M.
(22) Prefix forming negative imperative, [...] A, P; [...] etc. B, F, L6, L4, M, S.
(23) Proclitic relative converter preceding perfect conjugation operator, [...] A, B, F, L4, M, P; [...] L6, S.
(24) Presence versus absence of a special augens form [...] beside the usual [...], etc., presence B, S; absence A, F, L6, L4, M, P.
ii. Allomorphic Variables
Allomorphic Expansion of Conjugation Bases. Allomorphic expansion by [...] in prenominal conjugation forms is handled very much alike in all dialects, allowing for frequent variation in nonnormalized manuscripts, and with striking deviations occurring only in Akhmimic (see Polotsky, 1960, sec. 52-56). Dialect-specific expansion of presuffixal bases may pertain either to the whole paradigm (e.g., the Akhmimic affirmative aorist, see ibid.) or to particular combinations, such as (a) all third-person forms (fluctuations in the affirmative aorist paradigm in L6 and M), (b) the second-person feminine singular, and (c) the second-person plural. The rationale to be recognized behind the outstanding Akhmimic usage seems to be a different “signaling” function of [...]. While in all other dialects it serves to expand a (greater) number of base morphemes so as to make them more “conspicuous” with regard to certain kinds of subject expressions that follow it (in particular, nominal and second feminine singular), in Akhmimic it tends to give up its expanding function and become an invariable part of the base morpheme for a (smaller) number of bases, the rest (in particular, present converted bases) being left without any expansion at all. The result is stronger morphemic uniformity and less submorphemic alternation. The following instances in terms of isoglosses may be typical of the situation:
(25) Circumstantial present, prenominal form, [...] ([...]) B, F, L4, M, 5; [...] A; nonnormalized [...], [...] L6, P.
(26) Second present, prenominal form, [...] ([...]) B, F, L6, L4, M, P, 5; nonnormalized [...], [...] A.
(27) Affirmative aorist, all third-person forms, [...], etc. A, [...] P; [...], etc. B, F, L4, S; nonnormalized L6, M (both seem to prefer [...] to [...]).
(28) Circumstantial present, second present, imrperfect, and conditional, before suffix of second plural (which then appears in different allomorphs accordingly), [...] B, L6; [...] A, F, L4, M, P, S.
Intradialectal Interference of Submorphemic Alternation.
The generalization of an allomorph beyond the contextual scope it is otherwise strictly bound to is a phenomenon very close to grammatical error. In Coptic, as well as in other languages, it is rarely found to be characteristic of a literary standard variety of the language, within the same historical period. A case in point is the combination of the second-person plural pronominal (subject) suffix—that is, its two basic allomorphs [...] and [...]—with the various conjugation bases, which may be divided into “short” and “long” ones (depending on whether they contain [...] or not). The general morphological rule of Coptic that says the short suffix only combines with the long bases and the long suffix only combines with the short bases is invalidated in dialects A and M in opposite directions (cf. Polotsky, 1960, sec. 56; Funk, 1981, sec. 1.4.1). This may be presented in terms of isoglosses as follows:
(29) Second plural suffix, short form: only with long bases A, B, F, L6, L4, P, S; also with short bases M.
(30) Second plural suffix, long form: only with short bases B, F, L6, L4, M, P, S; also with long bases A.
iii. Morphosyntactic Variables
Special “Portmanteau” Morphs.
(31) Special element /er-/ incorporating {rel} + {perf} + {third-person subj. pron.}, presence M; absence A, B, F, L6, L4, P, S.
(32) Special element /-ah-/ (preceded by relative converter [...] [...]) incorporating {perf} + {thirdperson subj. pron.}, presence A, L6, P; absence B, F, L4, M, S.
(33) Special element [...] /-çar-/ (phonologically corresponding in P to Sahidic [...] /[...]/ (preceded by relative converter [...]) incorporating {aorist} + {third-person subj. pron.}, presence P; absence A, B, F, L6, L4, M, S.
Special Conjugation Base.
(34) Presence versus absence of a special “temporalis” clause conjugation [...] (“absence” implies the use of rel. perf. in the same paradigm), presence A, L6, L4, P, S; absence B, F, M (Fayyumic proper is split here; this notation accounts for the usage of biblical manuscripts).
Use of Different (Coexistent) Forms in the Same Paradigm.
(35) Prefix deriving Greek-loaned verb stems, [...] A, B, F; zero S, M; nonnormalized L6, L4, P.
(36) Nuclear element of NP-equivalent relative clauses (not fully normalized), [...], etc. B [...], etc. A, F, L6, L4, M, P, S.
(37) Element forming “instans” verb form (to be used in bipartite conjugation to express future”), stative verb [...] versus infinitive connector [...], in particular after subject pronouns like [...], [...], c, [...] A, B, F, L6, M, P. 5; [...] L4.
(38) Prenominal form of the “causative imperative” operator: {[...]} + {causative infinitive} versus prenominal analogue to [...] causative imperative, [...] P; [...] A, B, F, L6, L4, M, S.
FIGURE 1. GROUPING OF EIGHT COPTIC LITERARY DIALECTS BASED ON MORPHOLOGICAL DATA
[See PDF version of this article for Figure 1.]
These dialectal isoglosses selected for a broad variety of morphological items may be used as a data basis to determine the degrees (and hierarchical order) of relationship between the eight dialectal varieties considered, by means of various “clustering” techniques supplied by modern numerical taxonomy. With a number of such methods having been applied successfully and yielding very similar results, a classification based on morphological traits might be suggested as shown in Figure 1 (neglecting, for the present purpose, the precise hierarchical level for the location of division nodes on the tree).
P A L6 L4 S M F B
The primary division in the set of individual dialects turns out to be that between A, L6, L4, P and B, F, M, S, corresponding to only one trait (which then is the most typical differential trait), namely, the vocalic representation of the Egyptian preposition r- (as Coptic [...] versus [...]) in its various grammatical paradigms (cf. item 14 above). Although nonmorphemic (and quasi-phonological) in nature, this trait seems to symbolize the most profitable division of the whole cluster of dialects into two subsets in terms of morphological isoglosses.
Some further traits of similarity along branch la are (14) [...], (21) [...] (shared with S), (34) “temporalis” (shared with 5), (16) [...] (shared with M and S), and (13) monosyllabic form of [...] (shared with M and S). Along branch 1b are (22) [...], (27) aorist without [...] extension (both shared with L6 and L4), (20) [...], etc. (shared with L6), (19) [...], etc. (shared with L4), and (18) [...], etc. (shared with A and L4).
The differential traits at node 2 are (22) [...] A, P versus [...] L6, L4, plus, perhaps, the presence or absence of [...] extension with the aorist base, (27) [...] A, [...] P versus [...] L4, (L6). Some further traits of similarity within branch 2a are either shared with L4, as in (20) [...], (28) nonextended base (plus F, M, S), (2) perfect [...] (plus B, F, S), and (23) rel. [...] with perfect (plus B, F, M), or shared with L6, as in (19) [...], etc., and (32) [...]. Within branch 2b, some further concurring items are (6) [...], (41) nonnormalized use of [...] (shared with P), (4) e-vocalization (shared with P, plus S), and some other traits shared with either P or A plus B, F, M, S, as in (3), (30), and (38).
The differential traits at node 3 are (35) zero M, S versus [...] B, F, (16) [...] versus [...], (7) [...] versus [...], (8) [...] versus [...], (10) [...] versus [...]. Some further traits of similarity along branch 3a are (13) [...] (plus A, L6, L4), (1) [...] (shared with F), (15) [...] (shared with B, plus L6, P), (6) [...] (shared with B, plus A). Along branch 3b, most of the further concurring items are shared with M: (14) [...], (34) rel. perf. for “temporalis,” (23) rel. [...] with perf. (plus A, L4, P), (4) a-vocalization (plus A). Shared with none is (13) disyllabicity in [...]; some traits are shared with S plus A, (L6), L4, P, as in (2) and (29), or S plus L6, as in (5).
This classification based on morphological traits, perhaps in a more fully elaborated form, may be used to supplement and reinterpret the results of a classification based on purely phonological data (the more so, if any such classifications should be used as a guide to the geographical allocation of dialect centers). It is interesting to note that in terms of both serial and hierarchical order, the two sets of criteria lead to considerably different results. This is easily seen by comparing Figure 1 with what may be the result of a phonological classification. Depending on how much additional emphasis is put on the “natural” vocalization of stressed syllables (as compared with that of unstressed syllables and with consonant traits), either of the classifications shown in Figures 2 and 3 may be preferred. Since it has not yet been determined whether a binary division of the whole set of individuals is really appropriate here, even a third alternative might be worth considering, namely, that shown in Figure 4 (cf. also Hintze, 1984).
Irrespective of the actual hierarchy preferred, the difference between these classifications and the one based on morphological data is quite obvious. The most striking (though least surprising) detail is the differing allocation of P, not only changing its “nearest neighbor” affiliation but rather shuttling between poles (cf. earlier statements to similar effects in Polotsky, 1970, p. 561, n. 11; Kasser, 1960, pp. xxviii ff.). A satisfactory explanation of this phenomenon is not known to have been proposed so far. Of much greater bearing, however, taking into account the historical role of the various dialects and dialectal varieties, is the differing degree of relationship between Sahidic and Bohairic, on one hand (being remarkably stronger in the phonological than in the morphological field), and between either of these and Fayyumic or Mesokemic, on the other. Also, in terms of serial order, it is obvious that Sahidic is much closer to the southern dialects (A, L6, L4) in the morphological field than in the phonological, while the situation of F and M is the reverse. As far as Sahidic is concerned, it may well be the social nature and prehistory of this dialect as a supraregional vernacular rather than its geographical homeland that provides the clue to a greater part of its characteristics and its overall neutralizing behavior.
FIGURE 2. GROUPING BASED ON PHONO-LOGICAL DATA, WITH EQUAL WEIGHTING FOR ALL VARIABLES.
[See PDF version of this article for Figure 2.]
FIGURE 3. GROUPING BASED ON PHONO-LOGICAL DATA, WITH HIGHER WEIGHTING FOR STRESSED VOWELS (BINARY SOLUTION)
[See PDF version of this article for Figure 3.]
FIGURE 4. GROUPING BASED ON PHONOLOGICAL DATA, WITH HIGHER WEIGHTING FOR STRESSED VOWELS (NONBINARY SOLUTION)
[See PDF version of this article for Figure 4.]
WOLF-PETER FUNK