CC BY
47
Nikolai N. Kazansky
"LINGUISTIC THOUGHT" OF MYCENAEAN SCRIBES
By this paper I would like to draw attention to a problem often overlooked by mycenologists, that of a special study of the perception of a syllabic written text. It has been a commonly accepted practice to rely on a certain set of writing rules and to apply them automatically to reading as well1. Meanwhile, the problem of reading, i.e., of the perception of a written text, has remained outside the mainstream of mycenological research, despite the indisputable significance of reading as an institutionalized component of the palace culture for our general comprehension of Mycenaean culture.
Among the Modern history's popular myths, one of the most naive and lovely is that of the Antiquity as the age of childhood - a golden childhood - of European culture. While the Greek archaic and classic periods are described as the cradle of European culture, the Mycenaean civilization may, at the best, claim to be regarded as a stage almost embryonic and, as such, is excluded from a cultural interpretation. It is striking though that, for all this, nobody has disputed the complexity of Mycenaean civilization. Yet, simplistic myths of this kind continue, partly because they have been long-lasting and widely spread, partly because their eradication turns out to be difficult due to the narrowness of the circle of scholars familiar with Mycenaean materials.
Henry Hankey's beautiful picture which J. Chadwick placed on the cover of his popular book2 (a knelt boy manufacturing tablets, a man writing with obvious difficulty and another man supervising them both with a club in his hand) gives a clear and vivid idea of the Mycenaeans as semi-literate people. No less characteristic are remarks on «poor adaptation of the Mycenaean graphic system to rendering Greek words» or statements that «the incompatibility of the Mycenaean graphic system with the phonetic system is the main factor which complicates, or sometimes renders utterly impossible, the interpretation of the Mycenaean lexicon» (e.g. S. Sharypkin in: Molchanov, Neroznak,
heading rules, which are, in fact, rather rules for writing, were formulated by M. Ventris and J. Chadwick in their first publication as early as 1953. Cf. also: Chadwick 1967, where Chadwick dwells on the peculiarities of Mycenaean scribes' activity.
2«I am much indebted for it (an attempt to reconstruct the scene in the archive room at Pylos - NK) to the Hon. Henry Hankey, who very kindly painted the picture following a brief I drew up» (Chadwick 1976:19).
Sharypkin 1988: 81). Solomo Luria has put it even more clearly and sharply, noting (Luria 1957) that the Mycenaean form e-ke may have 16 different readings! Indeed, Mycenaean graphic rules do allow coding 16 sound sequences by means of these two signs3. This, however, applies to the writing rules; rules for reading might have been less ambiguous. In Russian, the sequence of characters Доде can scarcely lead the reader so far as to confusion of /dod'e/ - the dative case of a diminutive form of a personal name stressed on the first syllable - and the name of a writer (Daudet). Why should we think, then, that a Mycenaean scribe, a literate person trained at a scribal school (how else can we explain the uniformity of both the orthography and the format of documents in the entire area of spread of Mycenaean culture from Thebes in Beothia to Pylos in Peloponnesos, and from Cnossos or Chania on Crete to Mycenae and Tiryns?), spelled a text syllable by syllable, selecting one of some 16 options for each word? Wouldn't it make more sense to assume that the Mycenaean literate man simply read instead of going through the options resulting from imperfection of the graphic system? Isn't it more possible to suppose that this imperfection itself is illusory and is connected solely with writing, but by no means with decoding the written fixation of a speech act?
A Mycenaean scribe did not write for himself. We know this definitely, as texts were often corrected, amended, specified and supplemented by another hand. How could such an imperfect writing system exist, moreover, get spread over half a dozen of state centers so remote from one another if it was indeed so bad? Reading syllable by syllable, so persistently and unanimously postulated for the Mycenaean scribes, could scarcely be justified by the administrative practices of the Mycenaean epoch. Let us imagine a reference to a document4 at an actual trial. Could a scribe, in the presence of the King (wa-na-ka) and the Council (ke-ro-si-ja), or facing the people (da-mo), allow reading in the manner of a modern dyslexic? This seems extremely doubtful. Yet, the belief in the Mycenaean semi-literacy has established itself so firmly that the work of Mycenaean scribes sometimes appears to be best described in terms borrowed from medical surgery rather than linguistics.
This belief probably arises from the notion of superiority of the ancient Oriental culture for which nobody has ever dared to postulate semi-literacy. A poor adaptation of the Mycenaean graphics and impossibility of its usage for the fixation of texts have become one of the
3 «At any rate, this graphic principle is very inconvenient for Greek, which has been the cause of its further alterations» - Tronsky 1973: 84.
4 E.g., a purchase contract, whose existence as a self-contained genre of legal documents, indirectly reflected in a separate group of Cnossos tablets, has been proved by J.-P. Olivier (Olivier 1987: 479-498).
axioms not only in mycenology, but also in the history of Greek culture in general.
In my opinion, the sole cause of a wide proliferation of these concepts is to be sought in the absence of any attempts at typological interpretation of Mycenaean texts from the point of view of the perception of information, i.e., of reading. The contrast between the Mycenaean syllabic and the Greek alphabetical systems is, naturally, immense. They differ in their very approach to the text: the segmentation into words is common in Mycenaean texts vs. the fixation of speech flow in early Greek alphabetic inscriptions, e.g., those written in boustrophedon. What strikes the eye here is the difference in the concept of a basic text (or speech) unit. The strict singling out of a word is the foundation of the Mycenaean linguistic analysis. The recognition of a word is the basis of reading in Mycenaean; and the significance of the word as a language unit for the Mycenaean mode of writing has been already stated by scholars. Yet, the next step - the conclusion that words were recognizable for the reader - has not been made. To put it differently, if we regard the word as the basic unit of a Mycenaean text, isn't it natural to suppose that the reader at once recognized a word as an integral sign with all its distinctive features? The text was written - syllable by syllable or word by word - with a certain degree of approximation to the actual pronunciation, but was read word by word (not sign by sign, as is commonly assumed nowadays) and was voiced in the manner common for standard official oral speech. We can, therefore, assume a fundamental difference in the principles of the fixation and perception of text between the Mycenaean (as based on the recognition of words) and the alphabetical (based on the reproduction of a sound sequence) system of writing.
This difference has an important implication. In alphabetical Greek texts morphological elements are written out with an extreme accuracy, while in the Mycenaean syllabic script they are marked off in a rather slap-dash way. We can speak of a principal neglect of inflection in the Mycenaean graphic system, despite the fact that the Greek language of the Mycenaean epoch as well as of the following centuries is a classical example of a language based on flexion (fusion). To have an example, let us list the paradigm of the word (w)oikos 'house' as it would have looked
in Mycenaean and alphabetic Greek:
Case Mycenaean script Greek
Nom. sg. *wo-(i)-ko (^)oiko"
Acc. sg *wo-(i)-ko (^)olkov
Gen. sg. *wo-(i)-ko-jo (^)o'lkolo
Dat. sg. *wo-(i)-ko (^)olkul
Allat. sg. wo-(i)-ko-de (f)oiKov8e
Loc. sg. *wo-(i)-ko (^)o'lkol
It is not difficult to notice that Mycenaean writing yields no evidence reliable enough for us to draw conclusions about the grammar (genitive and allative excluded). This neglect of grammar is obvious, which does not at all mean that Mycenaean scribes somehow managed without the full stock of grammatical markers, so typical of Greek, in their everyday colloquial speech. The fact that the same Mycenaean scribes had, however, almost no need of grammar marking in written speech is an interesting contradiction only in part reducible to the opposition orality: literacy as formulated for the Greek and Roman cultures. It appears wiser at this point to avoid lapsing into a straightforward evaluation of these facts and to look for some typological parallels for the Mycenaean features of the fixation and decoding of a text at a later cultural stage, both in and outside Greece. Thus, we would like to turn to abbreviations in ancient Greek papyri and medieval Greek manuscripts as parallels within the domain of the Greek language.
Papyri have been treated in a comparatively recent monograph by Kathleen McNamee, who has provided a detailed account of the cases of violation of the linearity principle in the Greek writing (McNamee 1981: 118-119). It would be reasonable to quote here the classification she has proposed, especially since it is minuter than those of her predecessors are.
1. A non-final character is superscript, the word reads from top to bottom, e.g. P = hp(epan)
2. Several non-final characters are superscript, the word reads from bottom to top: ol
K = kol(wt|)-
Cf. also character sequences read in the same direction:
p
o o
p = po(Lpa). c = Xop(O")
3. One (several) character(s) is/are inscript into another character:
A = Xe(ipeL).
4. Three characters are written in a vertical string which reads from middle to bottom, the top character being added at the end; thus the word
paXaLO"w ould be written by means of three characters as shown: p
a
5. Combination of a superscript character with an abbreviation sign.
6. Monograms of the type 0 = OLon, IH = piq(xU?).
7. The first character differs from the rest in shape and size:
ctl= ctkto?).
8. Other features, chiefly involving subscript characters like
ehp = espíalo?).
To a certain extent, Mycenaean abbreviations are comparable with the papyrus data quoted above5. Leaving ideograms (pictures) aside, let us survey instances of acrophonic abbreviation. These are particularly frequent in enumeration of spices (Y SA = sa-sa-ma 'sesame', ¥ MA = ma-ra-tu-wo 'fennel', etc.). Especially striking is coincident abbreviation
bM
of the word 8ea|iai 'sheaf (of cane)' in Mycenaean (abbreviated: X de) and in Greek household papyri from Hellenistic Egypt (abbreviated: 8e) (Ventris, Chadwick 1973: 593).
Apart from symbols with a purely syllabic value, the Mycenaean script makes use of ideograms that designate a certain concept. In scholarly research, these ideograms are usually treated from a purely functional viewpoint: as has been stated, they never assume the function of the determinative, as is common in cuneiform texts; they are placed at the end of a line as a brief summary of the information preceding them; they may render concepts which are either general (*110-*118 designate the concepts of volume, measure and weight) or extremely concrete (*200-219, *226-229, *250 designate various types of vessels ranging from a bullhead-shaped rhyton (*227) to an amphora *209).
According to the manner of designation, ideograms may be divided into the following classes.
I. Ideograms proper, i.e. signs of concepts, include:
1) numbers
| 1 || 2 ||| 3 |||| 4 ||||| 5 ....., 6 T 9
- 10 = 20 etc.
o 100 oo 200 ooo 300 etc. o 1000 oo 2000 ooo 3000 etc.
2) designations of measures:
a) of weight and volume:
*116 N # * 111 V <1 *110Z^ * 113 S *115P2
*114Q í *112T I *117 M * *118M ^
b) presumably also of time:
5 Claire Préaux was the first to systematically compare the Mycenaean and Hellenistic modes of writing (Préaux 1959: 79-85). Since this publication, there have been undertaken practically no systematical comparisons. Moreover, a feature significant for our research has not been noticed at all: abbreviations appear rather late in the Greek tradition, although the earliest inventory lists date back to the early Classic.
* 173 LUN i «a lunar month»
3) ostensive ideograms (pictograms):
a) of people, including the ideograms for «man» *101 VIR
and «woman» *102 ML^
n
*155 *204 *210
b ) of animals: *104CERV *105EQU i-'^0 *1060VIS '» * 107 CAP A *108 SUS ^ * 109 BOS ^
c) of vessels:
*200vas *201vas \T/ *202vas ^ *203vas
vas
VAS V *206vas V *207vas *208vasX7" *209vas ^
vas
*216
vas
*212
vas
*213
vas
*217V
5 ¿11ovas
*218V
*214 *219
vas\J
*215
vas
V
vas
H
*222
vas
*227
vas
*228
vas
*229
vas
including such complex signs as *225 ALV d) of arms:
Q
*226
vas
* 162 TUN
* 191 GAL
>
*231 SAG ■»— *254 JAC
e) of chariots:
f) of plants:
* 121 HORD 1 *122 OLIV ^
g ) of metals:
$ N
* 141 AUR A * 140 AF.S I
*230 HAS
*233 PUG
*243 ROTA
*125 CYP
= 140 AES
h) of food:
'123 AROM
£
* 130 OLE 1 * 131 VIN H>
i) of fabric and clothes:
* 159 TELA D
4) unidentified ideograms: *19OA *142 ^ *146 M *150 *152 W *154J?*158<^ * 160 El *164 *165 *170^ * 171 J *174(+we) *177 ^
fl ,245 ¥ .257 PI *256 ^
tr
* 183
!
= 177
* * *
Among these ideograms we may distinguish pictograms and ideograms with a phonetic value. An ideographic fUnction may be assumed either by pictures that have nothing in common with phonetic
signs (*225 ALV *226VAS A> *162 TUN M *191 GAL Q *230 HAS ^^ *231 SAG -»- *254 JAC-> *233 PUG G
*240 BIG *241 CUR *243 ROTA ® ), or by
signs which, although having an ideographic value, are arranged acrophonically. In the former case we are simply dealing with an element of pictorial writing, whereas in the latter case we are confronted with an ideogram, which is an abbreviation in no way (except for its syllabic form) different from those found in an alphabetic script (e.g. abbreviations in encyclopedias).
II. Abbreviation of a word to its first syllable. Acrophonic principle of ideogram composition
This usage of the ideogram as a syllabic sign was found by G. Neumann for a fig-tree (Y /ni/) whose designation is built after the acrophonic principle: according to Hesychios of Alexandria, the Cretans used for a fig-tree as late as during the Hellenistic period: niKuXea- auica (Neumann 1958: 156-158). Cases like this are well-known, cf. Y SA = ri-no, where the initial syllable stands for an unattested non-Greek word used as an ideogram according the acrophonic principle. Such writing is
especially frequent in enumeration of species: Y SA = sa-sa-ma «sesame», ¥ MA = ma-ra-tu-wo «fennel», T KO = ko-ri-ja-do-no «coriander». KU = ku-mi-no «caraway». 1/ MI = mi-ta «mint». The
Trtt
ideographic use of the syllabic sign *65/ju/= * 129 FAR "" may also belong to this category.
III. Combination of an ideogram and signs based on the acrophonic principle
In some cases, the acrophonic principle is combined with an ideogram:
a) identifiable: *211VAS+PO or
w
b) non-identified: * 153 <í> *189 HU *174 ( +we ) U *246
A *180
* 178 H—I^ *180 U *182 H—P *179 *253 W IV. Logograms
A special group of compound symbols is formed by logograms, i.e., by two or more syllabic signs in a ligature constituting a word and thus rendering a concept. Logograms may only be used in the position characteristic of ideograms; they are often read from top to bottom in compliance with the spirit and general principles of Mycenaean script. Yet, an equally great number of logograms read from bottom to top. Logograms can be subdivided as follows:
1) Logograms which can be read horizontally: there are no safe
examples, but it is not improbable that the ideogram * 161 may originally represent the ligature D/pe/ + ? /pu2/.
2) Logograms which can be read vertically: a) from top to bottom.
Ideogram * 156 TURO2 ^ is obviously a ligature of signs'■/ /tu/ + 't /ro2/. The meaning «cheese» appears when pronouncing the sequence of syllable signs /tu+ryo/read down from the top. Similarly, ideogram * 135
MERI ^ reads as a word, also down from the top: T /me/ + l lúl and
* 120 GRA ^ has been read as a logogram by in Cornelis J. Ruijgh ( M /si/ + T /to/) (W. F. Wyatt // Kadmos 1968. Bd 7, 100; C. J. Ruijgh //
Kadmos 1970. Bd 9. 173). Ideogram * 157 ^ (we+wi vel wi+we) might also belong to this category.
b) from bottom to top. The ideogram for «wool» which has been
m
conventionally rendered as * 145 LANA - is in reality a ligature of syllabic signs /ma/+/ru/ MA+RU ¥ T behind which lurks a word cognate to |idAAuices" ' Tpixe? attested in Hesychius. Similarly the
logogram * 127 KAPO ^ is a ligature of signs © /ka/ + ' /po/ = KA+PO /karpos/ «fruit». The ideogram for «saffron» is also a ligature, this time of three signs, and is read from bottom to top: © /ka/ + Y /na/ +
T /ko/ = KA+NA+KO /knakos/. Similarly logogram * 133 AREPA ^ consists of syllabic signs Y /a/ + Y /re/ + ^ /pa/ disguising the word /aleiphar/.
c) sign inscript in another sign. There are no reliable examples.
Possibly, 247 DIPTE ^ might read inside out as i /di/+IJ/pte/. This method of designation is often utilized in combination of ideograms with
a phonetic determinative, e. g., *258 V. Ideograms with a phonetic determinative
Just like logograms and compound ideograms, these can be read in different directions: the usual reading is from top to bottom, but reading from bottom to top and from outside inside also occurs. These signs are used to note the whole word and suit the perception of a word as a unity. Due to this fact, a text acquires a second dimension, whereby a sign sequence is read as a linear text from left to right, while the isolated ideograms are read from top to bottom.
It is of interest how animal gender is designated - the difference between the ideogram BOS «bull, cow» differs from BOSm by lacking the sign ^ /pa/ in the first ideogram. This sign does not, however, become a logogram, i.e., it is not pronounced as F /mu/ +
+ /pa/, as is the case
with sign * 127 KAPO ® /karpoi/, but the ideogram is understood as a single and indivisible sign for the word «ox». A counterpart to this indivisible sign is furnished by a variant of the ideogram with a subscript syllabic sign resembling Ä /wo/ which denotes a «cow». This opposition of male and female genders is characteristic of all other animals with the
exception of fallow deer (* 104 CERV ) and donkey for which the ideogram is unknown.
VI. Ligatures of two signs rendering one word by an ideogram strengthened by a phonetic sign used acrophonically
a) a phonetic sign is subscript:
Such usage does not render the whole concept but only specifies an ideogram and occurs in many compound signs: *108+KA (SUS+KA) = /kaper/. In all probability, on the semantic level the subscript syllable /KA/ under the ideogram for «swine» only specifies the generic meaning «swine», whereas on the linguistic (phonetic) level it causes substitution of the reading (pronouncing) /kaper/ «hog» for /sus/ in the same way as the syllable /SI/ subscript under the ideogram for «swine» in the compound sign *108 + SI (SUS+SI) = /sialos/ only specifies the generic meaning «swine», turning it into the word /sialos/ «hog». A similar semantic supposedly determines the SI under the ideogram for «bull»: *109 + SI (BOS+SI) = /sialos/, as well as the combination *106+TA (OVIS+TA) - the meaning of the second element is unknown. A combination of signs also stands for a single word in cases when a
phonetic sign is superscript, e.g. *212vas+U (W) = u-do-ro /hudroi/ -two signs stand for only one word. The same may hold for combination *146+PE with undiscovered meaning. In all these cases the combination of an ideogram and a phonetic sign serves to designate one word, and thus this mode of notation is typologically close to approaches what we observe in the cuneiform script (cf. Hittite designation of vessels DUGhapuwai- n., DUGhariuli- n., etc.), where a sign designates a generic concept. In Mycenaean texts, there is only a hint at the phonetic part: the indication of the word-initial syllable.
b) a phonetic sign inscribed in an ideogram.
*159+TE (TELA+TE) = /te-pa/ «carpet fabric» was probably used as a single word, unlike *159+PA (TELA+PA) = /puktalia/ «cloak fabric», *159+PU (TELA+PU) = /puktalia/ «fabric folded twice». The meanings of *159+KU (= TELA+KU) and *159+ZO (=TELA+ZO) are unknown; it, therefore, appears impossible to define whether they rendered one word or two. The signs *162+KI (TUN+KI) and *162+QE (TUN+QE) also defy definition, whereas in the case of *162+RI (TUN+RI) = /kithon litinos/, Mycenaean scribes rendered a word group by an ideogram.
It is beyond doubt that only one word is disguised by the notations ^
*209vas+A = a-pi-po-re-u /amphiporeus/ and W *210VAS+KA = ka-ra-re-we /khlarewes/; the proof thereof is furnished by Hesychius' gloss
xXapov eXanqpos- ko'jOoj. And the word Sena? probably lurks behind
one of the ideograms with an inscript determinative DI: 0 *202vas+DI,
The ideogram *211VAS (+PO) = po-ti-ni[, po-ti-[ ]-we (meaning unknown) does not allow to say definitely how many words lurk behind
it, as well as m *258 (+WI) (meaning unknown). The same is true of *166+WE and *167+PE, both with unknown meanings, where syllabic signs are written across the respective ideograms (the last sign includes WE).
VII. Ligatures of signs rendering two words: the first word ideographically, the second one acrophonically
a) The phonetic sign is superscript: *244 ROTA+TE = /armo termidwen/ «rimmed wheel». The undoubtedly compound sign *244 ROTA+TE stands for two words, whereby syllabic sign *04 5I5 /te/ is
added on top of the ideogram *243 ROTA vF . The resulting ideogram should read as /armo termidwen/, «rimmed wheel», i.e. one (compound!) sign contains two words, of which the first one is rendered ideographically, the second one - acrophonically, by the initial syllable of the word. Something similar has to be postulated for compound signs *159+PA (TELA+PA) = /TELA pharweha/, *159+PU (TELA+PU) = /TELA puktalia/ which, in all probability, contain a word group, as the epithet «cloth folded twice» could hardly be in usage as a single word for a special concept (unlike *159+TE (TELA+TE) = /te-pa/ «carpet fabric», which is likely to conceal one word). Meanings of other ideograms denoting types of fabric are not known to us: *159+KU (=TELA+KU), 159+ZO (=TELA+ZO); it is, therefore, impossible to decide whether they render one word or two. Undoubtedly two words are rendered by sign *162+RI (TUN+RI) = /kithon litinos/, but signs *162+KI (TUN+KI) and *162+QE (TUN+QE) defy analysis. It seems more likely that they represented a word group - a compound technical term.
Two vessels of different shapes have one and the same determinative
DI inscript in a respective ideogram:
0*202vas+DI,
XT *214vas+
DI. It may be assumed that in almost all these cases the phonetic sign represents the word di-pa = /depas/, as in epic this word is used
terminologically, at least Sena? aiicpiKUTTf Xov. which corresponds in
number of handles to what is shown on the ideogram in text KN K 770. It is almost sure that one part of the signs disguise a single word 8epa?, the other part represents a word group resembling Homer's formula cited above.
2) A phonetic sign is subscript: the undoubtedly compound sign * 120+PE ( GRA+PE) where the syllabic sign *72 D
/pe/ is subscript to the
ideogram *120 = /sito/. The resulting ideogram should read as /sitoio spermo/ - «seed of wheat (?)», i.e. one sign contains two words of which the first is rendered ideographically and the second outlined acrophonically, i.e. by the initial syllable of a phonetic word. Likewise, two words are contained in the compound signs *122+A (OLIV+A) = /elaiwa agria/, *122+TI (OLIV+TI) = /elaiwa tithasos/, *130+PA (OLE+PA) = /elaiwon sphakowen/, where a phonetic sign is subscript. Something similar should be postulated for the compound signs *120+O (GRA+O), *124+QA (PYC+QA), *125+KU (CYP+KU), *125+O (CYP+O), *125+PA (CYP+PA), *130+A (OLE+A), * 130+WE (OLE+WE), where the phonetic sign is also subscript.
3) A sign is inscript in a sign.
The undoubtedly compound sign * 123 AROM + KO T (AROM + KO) = /aroma koriandnoio/, where a syllabic sign is inscript in an ideographic one, represents a word group.
As we have seen, the principles underlying the composition of ideograms/logograms with phonetic determinatives are very diverse and include all possible options: a phonetic sign can be placed above, below, by side and across, or inside an ideogram. All this however does not hinder comprehension of an ideogram. These miscellaneous and contradictory principles must reflect several stages (probably non-simultaneous) in the evolution of ideographic signs. The Mycenaean script, although based on principles, which are often mutually exclusive, was a single system for the Mycenaeans, each sign having only one reading. Thus, we may see, in the discord of principles of ideogram composition, several chronological stages of script development; but we should - from the synchronic and systemic standpoint - regard all ideograms of the Linear B script as clearly and unambiguously recognizable. The order of syllabograms in a linear horizontal text sequence coincides with that of signs in a logogram. Let us take
logogram * 133 AREPA ¥ as an example. It consists of syllabograms 'l /a/ + Y /re/ + ^ /pa/, rendering the word /aleiphar/ «ointment». This
logogram reads from bottom to top. Mycenaean, however, allows another
direction of reading (from top to bottom): ^ /pa/ + Y /re/ + V /a/ which results in the intelligible /phaleha/, a word attested in Homer's compound /(tru)phaleia/ «helmet». Obviously, the order of signs in a logogram was not determined by the convenience of reading based on some unified principles; but it suited the perception of a group of syllabic signs within the limits of a word felt as a single indivisible sign. It is evident that one of the important components in teaching scribes was cultivation of skills necessary for unambiguous recognition of words and logograms perceived as a «word-sign» written in the manner required by an established tradition.
The first step in education however must have been the teaching of the alphabet. Let us now proceed to the RECONSTRUCTION OF AN ALPHABETIC SEQUENCE IN THE MYCENAEAN SCRIPT. As material, we will use the mistakes of Mycenaean scribes.
First of all, we have to point out that, though we have at our disposal several palimpsests representing two texts simultaneously, they play virtually no role in the linguistic study of mistakes, and neither do a few curious cases, when a scribe falters, erasing and then rewriting the same sign6. There are also some cases, such as PY Ub 13187, in which we are incapable to clarify the reasons for corrections. Such materials cannot constitute an object for linguistic study either. However, a general classification of mistakes corrected by Mycenaean scribes seems pertinent.
The omission of a word is very frequent, although it is not interesting from the linguistic point of view. For instance, it is quite evident that the scribe PY En 695.4 omitted the first word in the formula to-so-de pe-mo, which led to a rather unexpected sentence we-te-re-u, i-e-re-u, o-na-to, eke, pe-mo GRA T 1 «the priest W. has rented grain» instead of «the priest W. has rented a plot of land; the total of grain is T 1»8. In the same manner, another scribe had already started writingpa-ra-wa-jo, when he noticed that he had omitted e-pi-ko-ru-si-jo 2 pa-ra-wa-jo9. The same can be observed in PY Ta 708.3 ta-ra-nu , ku-te-se-jo , a-ja-me-no , a-di-ri-ja-pi that was later corrected to ta-ra-nu , ku-te-se-jo , a-ja-me-no , e-re-pa-te-jo , a-di-ri-ja-pi , re-wo-pi-qe10. An omitted word would usually be
6 Cf. e.g. KN Vc 191 wa-ke-i-jo (jo over jo).
7 PY Ub 1318.6 a-pe-i-ja , u-po , ka-ro , we-[ ]-ja 1 u-po , we-e-wi-ja , e-ra-pe-ja E 1 (6 second u-po possibly over [[ka-ro]]).
8 PY En 659.4 we-te-re-u , i-e-re-u , o-na-to , e-ke , to-so-de pe-mo GRA T 1 (.4to-so- over erasure, probably [[pe ]] perhaps [[pe-mo ?]] ).
9 KN Sk 789.B to-]ra / e-pi-ko-ru-si-jo 2 pa-ra-wa-jo[ (e-pi-ko-ru-si-jo perhaps over pa).
0 PY Ta 708.3 ta-ra-nu , ku-te-se-jo , a-ja-me-no , e-re-pa-te-jo ,
inserted above the line. The only conclusion that can be drawn from this common practice of correction is that scribes normally reread their texts.
In some cases a scribe starts writing a wrong word and then corrects himself. This type of correction is especially frequent in lists of personal names. Sometimes changes are minimal, as for example in PY Eb 895 [+] 906.B ko-to-no-o-ko, where -no- is written above the erased [na], that is, the scribe started writing a more frequent word ko-to-na and then realized that he was mistaken. A similar case can be found in PY Fr 1219.2 wa-na-so-i , po-se-da-o-ne OLE+A V 2, where the scribe started writing po-se-da-i-jo , «to the temple of Poseidon» and then changed it to «to Poseidon». Very close to it is the omission of signs -se-wi- corrected from [[wi-ja]] only after the word ]pe-se-wi-ja[ was written to the end (PY Xa 1385). Both in this text and elsewhere, the length of the erased portion indicates a correction of a considerable part of the word. For instance, e-ni-qe e-ra-pe-me-na was changed from e-qe e-ra-pe-me-na11. To illustrate this practice we can recall the omission of the first syllable, which makes the word unrecognizable. In one of such cases (PY Sa 794), the scribe omitted both the first syllable and the negation before it: the original [pe-re-e] was changed to no-pe-re-e, ROTA ZE 1 [ («useless») in the ultimate version of the text. Often this kind of correction comprises the whole phrase, for instance, se-re-mo-ka-ra-o-re-qe , ku-ru-so[[ ]] instead of [[ku-wa-no[ se-]re-mo , ka-ra-o-re]]12. KN Sd 4401 + 8718 + fr. (Hand128) between .a ]a-ra-ru-ja, a-ni-ja-pi, wi-ri-ni-jo, o-po-qo, ke-ra-ja-pi, o-pi-i-ja-pi CUR[ and .b i-qi-jo , / a-ja-me-no , e-re-pa-te , a-ra-ro-mo-te-me-no po-ni-ki[-jo, it was a line with the text o-U-qe , pte-no , po-si.
Such evidence is certainly very interesting, but again linguistically rather irrelevant. A true linguistic classification should be based on more frequent phenomena. One of the regular types of correction is caused by the fact that a scribe finds himself to be one sign ahead of his text, which urges him to erase the sign that he has just written and to put the omitted one instead.
In this case we can trace how scribes dictated words to themselves, dividing them by syllables and checking the correctness of their spelling:
a-di-ri-ja-pi , re-wo-pi-qe (.3 e-re-pa-te-jo over [[a-di-ri-ja-pi]] ).
11 KN L 647 + 2012 + 5943 + 5974 .B ]ra , / e-ni-qe e-ra-pe-me-na 'nu-wa-
ja'
.B Traces of erasure: -ni- over qe, and -qe e-ra- perhaps over e-ra-pe (note that e- now follows immediately after -qe, suggesting that the scribe may have originally written e-qe e-ra-pe).
12 PY Ta 714.2 a-ja-me-na , ku-ru-so , a-di-ri-ja-pi , se-re-mo-ka-ra-o-re-qe , ku-ru-so[[ ]] , ku-ru-so-qe , po-ni-ki-pi 1 (se-re-mo-ka-ra-o-re-qe , ku-ru-so , over [[ku-wa-no[ se-]re-mo , ka-ra-o-re]]).
one of the scribes writes mo- instead of re-mo[ (KN Xe 691.1 scribe 103)13, another one starts writing di-ka-so, but realizes that he has made a mistake and changes it to di-ta-ka-so (KN Ga(2) 427+8102 scribe 136)14. In other instances, we-da-ne-we is changed to we-u-da-ne-we (PY An 519)15, a-na-ta is changed to a-na-i-ta1 , the adjective e-ka-te-re-ta is spelled e-te-re-ta11, po-]ni-ja18 is corrected to po-]ni-ki-ja, and ki-ri-te-wi-ja at first looks like ki-ri-wi-ja19. In another curious case, se-ri-no-te is changed to se-ri-no-wo-te20, but the correction remains incomplete and the word still lacks the last sign, e-pi-ki-to-ni-ja is written e-pi-ki-to-ja21, wi-je-so beginning with the sign je-,22 the adjective we-re-ne-ja was at first written as re-ne-ja23, e-re-pa-te-ja-pi as re-pa-te-ja-pi24. The scribe often omits a syllable in the end of the word being impatient to proceed to the final sign, e.g., a2-zo-qi-jo had been corrected from a2-zo-jo25, a3-ki-a2-ri-jo first written as a3-ki-a2-jo26 or a-pi-no-e-wi-jo where the scribe first omitted the syllable -wi-: a-pi-no-e-jo . Somewhat different reasons may have produced such mistakes as ko-na instead of ko-to-na, repeated twice (PY En 659 and PY Ep 21228).
Mistakes resulting from the fact that the scribe pronounced his text to himself before writing it down seem to be indicative of the manner in
13 KN Xe 691.1 o-pi , re-mo[ (re- over mo), cf. KN X 7502 to-so / i-ka[, with i-written over ka.
14 KN Ga 427+8102.1 da-wi-jo / a-pu-do-si po-ni-ki-jo di-ta-ka-so M 8 N[ ]o[ (-ta- over ka).
15 PY An 519.1 pa-ro , we-u-da-ne-we (-u- over erasure, perhaps da).
16 KN Sf 4419 + 5119 + 5814 ]i-qi-ja / a-na-i-ta CAPS 80 (a-na-i-ta: -i- over ta).
17 KN Se 891+1006+1042.B pte-re-wa / e-ka-te-re-ta , a3-ki-no-o 2(-ka- over
te).
18 KN Se 882 po-]ni-ki-ja CUR[ (-ki- overja).
19 PY Eb 321 [+] 327 [+] 1153.A ki-ri-te-wi-ja , e-ko-si , o[-]na[-to ke-ke-]me-na-o (-te-wi- over wi-ja).
20 PY Qa 1290 i-je-re-u , se-ri-no-wo [ *189 (se-ri-no-wo-te -te after -no-).
21 KN L693 .2 sa-pa P 2 Q 1 e-pi-ki-to-ni-ja AES M 1[ ' (-ni- over ja).
22 KN Da 1163 + 1400.A we-we-si-jo OVIS 100
.B wi-je-so , / pa-i-to ki-ri-jo-te (.B wi- perhaps over je).
23 PY Ub 1318.7 mu-te-we , we-re-ne-ja , ku[ (we- over re).
24 PY Ta 722.3 ta-ra-nu , a-ja-me-no , e-re-pa-te-ja-pi , ka-ru-pi *220 1 ta-ra-nu , a-ja-me-no , e-re-pa-te-ja-pi , ka-ru-pi *220 1 (e- of second e-re-pa-te-ja-pi over re).
25 PY Un 1193.4 ]nu-we-we DA 10 te-so-qe DA 5 a2-zo-qi-jo (-qi- over jo ).
26 PY Fn 79.15 a3'-ki-a2-ri-jo HORD T 1 V 4 (-ri- over jo).
27 PY Mb 1396 a-pi-no-e-wi-jo *146 2(-wi- possibly over erasure, perhaps jo).
28PY Ep 212.3 e-ko-to , te-o-jo , do-e-ro , o-na-to , e-ke , ke-ke-me-na , ko-to-na , ko-na , pa-ro , da-mo , to-so pe-mo GRA T 1 V 3 (ko-to-na , {ko-na ,} ; possibly corresponds to Eb 913, 935; cf. Ep 705.8;
which such preliminary oral preparation was done. Sometimes scribes omit a sign without even noticing that they have transgressed the word's limits. The text PY Sa 797 seems to be especially remarkable: it consists of only one line that includes two sentences - the title where a chariot is mentioned (a-te-wo-jo , wo-ka) and the main part that characterizes a pair of wheels belonging to this chariot (a-te-wo-jo , wo-ka , we-je-ke-e ROTA+TE ZE 1 [). Dictating the text to himself the scribe made a mistake not just on the border between two words but on the border between two sense units, having omitted -ka and started writing wo-we-je-ke-e.
An unusual case that indicates the lack of attention caused by the correction according to the number of syllables, which does not make the text any more reasonable than it was before, can be found in PY Ea 30529 .a qo-qo-ta-o
ke-re-te-u e-ke-na-to , to , ke-ke-me-na ko-to-na GRA T 1.
In a number of cases we can suspect that omission of a certain sign was triggered by the phonetic proximity of neighboring signs, as, for instance, in the name ku-ta-to that was first written as ku-to30, or in the sequence o-pi, /po-po[ that was mistakenly written as o-po-po31. In one case the scribe presented the word pe-mo /spermo/ by the single sign mo, but realized it in the process of writing32. In a few cases, scribes started writing to-so-pe instead of the final formula to-so-de pe-mo33. Despite the fact that this mistake occurs on the word boundary, it does not differ from omission of a sign within a single word, as it is highly probable that the whole expression could be united by the common accent and therefore could be regarded as a single phonetic word.
Some cases, in which a scribe changed his mind and spelled a word both in a more economical and precise way using rarer signs that reflected the phonetic reality more closely, seem to be especially interesting for us. For example, a scribe 101 in the text KN As(2) 1516.11 si-ja-pu2-ro VIR 1 used the sign pu2- instead of the usualpu-. In the same way, the sign -ta2- appears instead of the usual ta- in the text KN X 9338.1 ]-ta2-mo[34. In different texts we can find both the variants pe-te/pte- used equally and the sequence pe-te changed to the single
29 to , over he; possibly to be interpreted as e-ke <o>-na-to , {to, } .
30 KN Da 1123 + 7178 a-qi-ro / ku-ta-to OVIS 100 [ (-ta- over to).
31 KN L(1) 648 .a ri[-ta pa-we-a .b o-pi , / po-po[ (-pi , / po- over po-po).
PY Eb 294 .2 to-so-de , pe-mo, GRA 2 T 5(pe- over erasure, probably mo).
33 PY En 609.18 ]to , e-ke , to-so-de pe-mo GRA T 3 (-de overpe).' PY Eq 146.2 ko-ro , to-so-de , pe-mo [[GRA]] GRA [ (-de over erasure,
32
perhaps pe).
4 J.-P.Olivier: .1 -ta2- over , ta.
symbol pte-, for example KN So(1) 4429 + 5790 + 6019 + frr. b ]a-mo-ta , / pte-re-wa , te-mi-dwe-ta ROTA ZE 23? ROTA 1 [35, or PY Un 219.6 di-pte-ra-po-ro , RA 1 O 3 ko-ro[ ] l36. In both cases, for syllables starting with a stop, that graphic variant is preferred, which better corresponds to the sound. The same striving for phonetic precision pertains to the reflection of the groups formed from two sonorants, for instance, ro2- = /ryo/ instead of ri-jo in the names e-ro2-qo ra-pte (PY Ea 29)37, ko-tu-ro2 (PY Jn 431)38, PY Un 219.4 te-qi-jo-ne (=te-qi-<ri->jo-ne : cf. Fn 187.12).
There is also a certain tendency to represent the diphthongs ai and au by special signs, which urges the scribe of PY Eb 866. B to change ra- to ra3 (PY Eb 866 .B ma-ra3-wa, PY Ta 709.1 pi-je-ra339). The sign au written over a can be found in the personal names in MY Au 657.2 au-wi-ja-to VIR 1 [ and PY Ta 711.1 au40-ke-wa , da-mo-ko-ro.
We can clearly observe here the significance of the internal unity of a sign series based upon the common initial consonant that a Mycenaean scribe was certainly aware of. The series as a certain unity of signs was evidently also supported morphologically by different gender endings of nouns and adjectives: -ra/-ro , -wo /-wa 2, -jo/-ja43. The same series also served to distinguish nouns by grammatical number. Corrections related to the dual number seem to be especially important in this respect: i-qi-jo (du.) / i-qi-ja (sg.)44, te-mi-dwe-te (du.) / te-mi-dwe-ta (pl.n.)45.
35 J.-P. Olivier wrote: .b pte-re-wa over erasure (pte- changed frompe- without erasure; -re-wa over te-re[).
36 -pte- over [[pe]]. For more examples see Perpillou 1977: 241-242.
37 PY Ea 29 e-ro2-qo ra-pte , e-ke , o-na-to , pa-ro , i-ma-di-jo GRA T 1 (under -ro2- possibly ri ).
38 PY Jn 431.2 wi-ja-ni-jo AES M 5 ka-ra-*82[ ] AES M 5[ ] ko-tu-ro2 AES M 5 (ro after -ro2). For more examples see Perpillou 1977: 240-241.
39 PY Eb 866 .A ke-ke-me-na , ko-to-na , pa-ro , da-mo
.B ma-ra3-wa , te-o-jo , do-e-ra , e-ke-qe , o-na-to GRA [ T 2 (-ra3- over ra).
PY Ta 709.1 pi-je-ra3 , to-qi-de-ja *200 VAS 3 pa-ko-to, a-pe-te-me-ne *214 VAS 2 po-ro-e-ke-te-ri-ja *228 AS1 ko-te-ri-ja 6 (-ra3 over erasure, perhaps ra). For more examples see Perpillou 1977: 240.
40 au- over a.
41 PY Eo 276.5 ma-re-ku-na , te-o-jo , do-e-ro , e-ke-qe , o-na-to , pa-ro ru-*83-e GRA T 1 (-ra over -ro ), cf. En 74.6, where do-e-ro;
42 KN Ak 614 + fr. (ko-wo corrected from ko-wa).
43 PY Ta 714.1 to-no , we-a2-re-jo , a-ja-me-no , ku-za-no , pa-ra-ku-we-qe , ku-ru-so-qe , o-pi-ke-re-mi-ni-ja (-jo over erasure, perhaps ja).
44 KN Sd 4401 + 8718 + fr. .b i-qi-jo , / a-ja-me-no , e-re-pa-te , a-ra-ro-mo-te-me-no po-ni-ki[-jo (i-qi-jo corrected from -ja).
45 KN So 4433 + 4444.a ] te-mi-dwe-te , [
A Mycenaean syllabary sign representing an open syllable includes both the consonant series indication and the timbre of the vowel. Change of vowel within a single series can be taken as evidence for a phonetic analysis previous to the usage of a certain syllabic sign, which can lead us to the assumption that the Mycenaean syllabic signs could have been organized in a regular alphabetical order. Changes made by scribes in case of such mistakes clearly demonstrate that that the first sign that came to the scribe' mind was the one containing the vowel a, afterwards they realized that they were mistaken and changed this sign to the one with a different vowel. Change of vowel proves to be a very frequent type of correction: pe is written above pa46, qo above qa47, te above ta 8(?), wo above wa49, ti above ta50, etc. All examples of vowel change within a single series can be best explained not by some phonetic or morphologic reasons, or by a casual error, but by the influence of some kind of superimposed system. Most probably, it was an alphabetic regularity of signs that could serve as such a system.
The vowel a seems to be primary for the sign series above, which can be also supported by the regular confusion of a/o (wa-na-ka, ke-ni-qa etc. -k(w)s), and unexpected spelling variants, such as wa-na-ka-te instead of wa-na-ke-te. There are some rarer kinds of corrections reflecting the confusion of u- and o-signs, for instance the sign no- can be changed to nu-51, and the sign wa- to u-. We cannot totally exclude morphological reasons for the latter correction: a scribe might have mixed up two different case forms and then corrected the nominative to the dative (as in PY Cn 131.5, which contained the nominative form po-ro-u-te-u before the correction) or the genitive to the nominative (as in PY Ea 780 o-ko-me-ne-u , o-na-to GRA 2 [, where the genitive ending [[wo ]] was changed to the nominative one -u).
A correction explained partly by stylistic reasons and partly by striving for precision in juridical terminology can be observed in PY Eo 24752. Having written .1 a3-ti-jo-qo , ki-ti-me-na , ko-to-na in the first
.b ] , a-re-ki-si-to , wo-zo-me-no , ROTA ZE 1 [ (.a Final -te over ta).
46 KN Df 5260 + 5348.B ] tu-ni-ja , pe OVIS 10 [ (pe over pa)
47 KN Vc(1) 290 wo-ro-to-qo 1 [ (-qo probably over qa?).
48 KN As 609 + 5866 + 8589 + fr. .3 qe-te-se-u VIR [1] no-da-ro VIR 1 a-ku-ri-jo[ (-te- of qe-te-se-u over ta ?)
49 ko-wo corrected from ko-wa: KN Ak 614 + fr.; cf. KN Se 880 + 1017.2 ] , po-ni-ke-a , wo-ra-we-sa CUR 1 (wo- perhaps over wa).
50 KN L(1) 5949.b ]a3-tu-ti-ja[ (-ti- over ta).
51 KN Db 1464 + 7070.B a-nu-ko / pa-i-to ki-.ri-jo-te (-nu- probably over no).
52 PY Eo 247.2 e-ko-to , te-o-jo , do-e-ro , e-ke-qe , o-na-to , pa-ro , a3-ti-jo-qe , ko-to-no-o-ko , GRA T[ 1 (-qe of a3-ti-jo-qe probably over qo).
line, the scribe put this name in the genitive in the second line as well, but then corrected it to the dative pa-ro , a3-ti-jo-qe , ko-to-no-o-ko.
In the last two examples, it would be tempting to ascribe the mistake to a certain pressure exerted by morphology, but a number of examples analyzed above completely excludes any morphologic influence. On the other hand, a do/de53 confusion can be almost totally explained by an influence from the alphabet as a system. It is highly probable that a zo-/ze-54 confusion constitutes a similar case, and if this observation is correct, the sequence of Ca- Ce- Co- series within the Mycenaean alphabet appears to be quite plausible. In any case, a regulation of syllabic signs according to the vowel seems to be supported by numerous instances of the initial consonant omission, when, for example, a scribe writes e-mo instead of pe-mo55, or o-so instead of to-so56.
A different kind of mistakes is constituted by some cases of incorrect choice of a consonant: some scribes use qa instead of ka (or qe?)57, to
58 59 60
instead of do , wa instead of ja(?) , wo instead of jo , jo instead of wo61, a instead of wa62, ma instead of na63(?), and na instead of ma64. This group of mistakes probably also includes the usage of se instead of su65 and *56 instead of fu66(??). Mistakes of such kind obviously demonstrate that scribes clearly felt a certain similarity between the
53 PY Jo 438.12 ka-ra-do-ro , ko-re-te[ (-do- over erasure, probably de) PY Na 848 .B ]ke-ki-do [ (-do over de).
54 PY Ed 236.1 ka-ma-e-we , o-na-ta , e-ko-te , ke-ke-me-na-o , ko-to-na-o , .2 wo-zo-te , to-so , pe-mo GRA 30 T 2 V 3 (.2-zo- over erasure, possibly ze).
55 PY En 609.17 ]o-na-to , e-ke , to-so-de , pe-mo GRA T 1 V 3 (pe- over e).
56 KN Od 666.a ] to-so o LANA 14 (to- over erasure, perhaps o).
57 KN V(3) 429.1 ]...qa-ra-jo 1[ (qa- over qe or ka).
58 PY Ea 803 ko-do , e-ke , da-mi-jo GRA T 6 V 2[ (-do over erasure, perhaps to).
59 KN Od 666 v. ]ke-re-wa LANA 9 M 1 P 6 (-wa over ja?).
60 PY Cn 599.1a pa-ro
.1 wa-no-jo , wo-wo , ne-ti-ja-no a-ke-o-jo CAPm 100(.1-jo of wa-no-jo over wo).
61 KN Xd 119 ]-ne-ri-jo , e-[ (-jo over erasure, perhaps wo).
62 PY Eo 211.1 wa-na-ta-jo-jo , ko-to-na , ki-ti-me-na GRA 2[ V 1
.2 a-tu-ko , e-te-do-mo, te-o-jo , e-ke-qe , o-na-to , pa-ro , wa-na-ta[-jo] GRA [ T V 1 (.2 a- possibly over wa).
63 KN X 7565 ma-ze-to-[ (Over erasure, first erased sign being na-.-twe-difficult. Possibly -ma[ at right)
64 MY 659 .4 e-ri-tu-pi-na , te-o-do-ra-'qe' 2 (.4 e-ri- su [[ [.]-ma ]]).
65 PY Ea 822.a ra-wa-ke-si-jo
e-u-me-ne , e-ke , o-na-to , pa-ro su-qo-ta GRA T 5 (su- over se).
66 KN Da 1135 + 7182.B ke-to , / *56-ko-we (.B perhaps *56- over u).
labiovelar kw and the guttural k, between the voiced and voiceless dental t/d, and between the sonorant m and n.
A certain number of mistakes has inevitably to remain without any reasonable explanation, as for example the occasional confusion between
o- tl and I ha-61, nu- N and ka- © or qe- ©68, ro- \ and qa- ? 69, po- ^
and ta- H or su- E70 etc.71
Sometimes there is reason to believe that we deal with a mechanical repetition of a sign, for instance in the case of e-wi-su-[su]-ko that was changed to e-wi-su-zo-ko72. There is no visual similarity between these
signs (-zo- ^ vs. E -su-), but we have to keep in mind that the sign for *zu- has not been found in any of the available Mycenaean texts, which makes it quite plausible that it did not exist. Therefore, we might be dealing here with an attempt to transmit in script a syllable for which there was no sign. The scribe first tried to use the vowel /u/, but after rereading the text he realized that the word he had written contained the second element -su-ko-, which was usual in compound words, but did not
suit the context. Then the scribe made another attempt using the sign ^ -zo- containing the consonant that made the syllable as a whole sound closer to the phonetic original, in spite of its vocalic imprecision.
We cannot exclude some phonetic proximity between the signs *56
H and u f73, although it cannot be proven by the material at our disposal; in the same way, there is no certainty about identifying the sign
*65 Titt as /ju/. In such situation it would not be justified, however tempting it may seem at a first glance ( especially considering KN U( 1)
49, to discuss the phonetic correlation between *65 lltl and jo T74.
The confusion between i- and e-series evidently differs from everything discussed above, as it can be easily explained by phonetic reasons alone. This confusion occurs in many different words both in
67 KN Dd 1218.B o-wi-ro / *56-ko-we pa OVIS 6 ([.B o- perhaps over a2)
68 KN V(4) 5913 + 5921 + 8738 + 8818 + frr. .3 nu-to 1 [ (.3 nu- over erasure. Perhaps ka[ or qe [).
69 KN Xd 148.1 ro-a , ku[ (.1 ro- over qa).
70 KN C(4) 912 + 5027.11 ko-to , / po-ku[ ] OVIS 100[ (.11 po- perhaps over ta or su).
71 Cf. PY Ep 212.4 ko-ri-si-ja te-o-jo , do-e-ra , o-na-to , e-ke , ke-ke-me-na , ko-to-na , pa-ro , da-mo , to-so pe-mo GRA 1 (.4-na of ko-to-na over -ko ).
72 KN Se 965 + 1008.A e-wi-su-zo-ko , ka[ (-zo- perhaps over su).
73 KN Da 1135 + 7182.B ke-to , / *56-ko-we , (.B perhaps *56- over u).
74 Cf. KN U(1) 49 ri-*65-no / a-pe-re QO 1 (-*65- over erasure, possibly jo).
Mycenaean and in the Greek dialects of the first millennium BC: e- is changed to i-75, ti- is changed to te-76, and di- is changed to de-77. A direct phonetic interpretation has to be discarded as well78, because a
79 80
number of different signs were substituted by the sign si-: a- , pa- , to-81. The reason for such popularity of the sign si- remains without any explanation.
It seems that the mistakes of Mycenaean scribes show explicitly that the «alphabetic» order of syllabic signs was not occasional, and that
within the series of signs the sequence presumably started with the Ca.
* * *
The question of the understanding of a Mycenaean text by the reader has not been raised as an independent problem: instead, Ventris' rules have been evoked which apply to the writing of Greek words rather than to their reading, i.e., which are in fact the rules for coding rather than decoding. In this case it is possible for us to penetrate directly into the process of writing. As shown by analysis of errors corrected by the scribe himself or an editor, the scribe dictated the signs to himself, often skipping a sign in anticipation of a following one.
We have no such direct evidence for reading, except, perhaps, for those lines in which the scribe kept the task of reading in mind and took care of his reader. Before we examine a concrete example, let us consider some observations drawn from the facts of modern languages. It can hardly be doubted that Mycenaean texts were read aloud rather than silently82. I. M. Schlesinger's fundamental work, dedicated to the
75 PY Eo 247.4 i-pa-sa-na-ti , te-o-jo , do-e-ra , e-ke-qe , o-na-to , pa-ro , a3-ti-jo-qe , GRA T 2 (.4 i- over [[e]], cf. En 74.13, where e-pa-sa-na-ti). 6 PY Es 650.5 a3-ki-wa-ro , a-te-mi-to , do-e-ro e-ke to-so-de pe-mo GRA 1
(.5-te- over erasure, possibly [[ti]]).
77 KN Ak(1) 614 + fr. .A ] TA 2 MUL 30 pa di 4[ (di over erased sign, perhaps de).
78 There were two cases where instead of the expected te- we encounter si-: PY Ep 704.5 e-ri-ta , i-je-re-ja , e-ke , e-u-ke-to-qe , e-to-ni-jo , e-ke-e , te-o, (.5 te-over erasure, probably [[si]]) and KN Dn 5668 + fr. ]qo-te-jo OVIS 3300[ (te- perhaps over si. The interpretation of PY 704.5, which seemed to be strikingly similar to a later Laconic (Alcman) ctlol, is now rejected. It is difficult to confuse the signs te- sb and si- A because of their visual unsimilarity.
79 KN Sc 237 a-e-da-do-ro probably over ersure (a- certainly over si).
80 KN Ce 50.1b te-pa-ra , pe-re-qo-ta (.1b te-pa-ra: -pa- over si).
81 PY Jn 601.7 ko-to-wa-[ ]AES M 8 ,to[-so-]de , e-pi-da-to AES M 7 .7 e-pi-da-to: -to over erasure, perhaps [[si ]] ;
82 The question of silent reading probably may be responded positively for the Antiquity, cf. an exhaustive collection of materials in: Gavrilov 1994/1995: 17-
syntactic structure of a sentence and its guessing / decoding in the process of reading (Schlesinger 1968: 27), contains a number of observations of immediate interest for our problem. Thus, Schlesinger assumes that reading aloud is not reading word by word; instead, the reader grasps with the eye several words at once while voicing only one of them at the same moment. Meanwhile, already Fairbanks' works (Fairbanks 1937: 78-107) have established that the perceptive ability decreases as the reader approaches the limit of the unit grasped by his eye. The average length of a perceivable text span does not exceed four words. Bearing in mind that in Indo-European languages, Greek in particular, the word length varies from 1 to 4-5 syllables, we can easily calculate that the length of a perceivable text span ranges from the minimum of 4 syllables to the maximum of 20 syllables. It should be noted that prosody has paid special attention to the correlation of verse length and performer's breath. Knowledge of limitations imposed on human breath allows specialists in meter to distinguish, depending on the number of syllables, between two basic units: verses requiring one resp. two breath pauses. Indeed, Mycenaean documents do contain longer lines as, e.g., text PY Eq 213.1 o-wi-de, a-ko-so-ta, to-ro-qe-jo-me-no, a-ro-u-ra, a2-ri-sa; but syntactic and intonational division of this title is no less evident than that of Schlesinger's example, The woman teacher, who had taught him Latin, was very pleased, where, as he notes, the words 'teacher', 'Latin', and 'pleased' are the boundaries of syntactic units. Syntactic unit boundaries can be discovered in the Mycenaean text in a similar way; thus, o-wi-de, a-ko-so-ta 'that is what Axotas saw' and to-ro-qe-jo-me-no, a-ro-u-ra, a2-ri-sa 'traveling through the seaside arable lands' hold intonational and syntactic information evident from the first word for a Mycenaean reader, but providing a chain of riddles allowing various solutions for a modern researcher who analyzes each form syllable by syllable. The consideration of modern experimental materials demonstrates the significance of an actual division of a sentence. Thus, Schlesinger examined the process of reading Hebrew sentences with regard to syntactic arrangement. In practice, he dealt with an actual division of a sentence, the matter still largely overlooked by mycenologists.
Word order, syntactic position in a sentence, alongside with rare inflexion marks allow to read a text correctly by dividing it into syntactic and intonational syntagmas and finding the words that serve as boundaries between them. Sometimes the arrangement of a text into lines is also guided by the actual division of the sentence. Among such cases are, first of all, several texts where the scribe erases a word only to
33. However, I would not dare postulate a wide spread of silent reading in the Mycenaean time.
rewrite it at the beginning of the next line, i.e., where arrangement of a text into lines may have been a tool for marking semantic groups and syntagmas, e.g.:
PY Ep 704 .7 ka-pa-ti-ja , ka-ra-wi-po-ro , e-ke , ke-ke-me-no , o-pe-ro-sa , du-wo-u-pi , wo-ze-e , o--wo-ze , [[to-so]] .8 to-so[ pe-mo GRA ]4
«Karpatia, a priestess with a key, has a kekemeno field, having obliged herself to cultivate twice, does not cultivate. All-in-all grain... » We can see that the scribe took the trouble of erasing the word 'all-in-all' in line 8 only to write it again at the beginning of the next line, thus separating two parts of a period.
Text PY Un 267 contains two amendments related to actual division of a sentence: .1 o-do-ke , a-ko-so-ta .2 tu-we-ta , a-re-pa-zo-o .3 tu-we-a , a-re-pa-te [[ , ze-so-me ]] .4 ze-so-me-no [[ ko ]] .5 ko-ri-a2-da-na ArOm 6 .6 ku-pa-ro2 AROM 6 *1e»5e»7e» 1e»6 .7 KAPO 2 T 5 VINa 20 ME 2 .8 LANA 2 VINb 2 .9-11 vacant
Text PY Un 267 (ho-doke aksotas tuwestai aleiphazoi tuweha aleiphatei zesomenoi) may be understood as «That is what Axotas gave to the brewer of ointments who was going to brew for himself aromas for ointments: coriander 6 measures, rush 6 measures, etc.83« Evident in this text is the scribe's striving for its arrangement into syntagmas by correctly dividing it into lines. In order to secure correct reading of the text, the scribe erases words twice only to rewrite them in the onset of the following line. Obviously the scribe is afraid that line .3 tu-we-a , a-re-pa-te [[ , ze-so-me ]] may be misunderstood or not understood at all because of a wrong intonational division; line .4 ze-so-me-no [[ ko ]] was amended for fear that a «brewed coriander» might result instead of the correct «brewed ointments».
The scribe acts in absolutely the same way when describing objects, cf. PY Ta 713, where the beginning of a sentence (= beginning of a line), at first misplaced, was rewritten at the beginning of the next line:
83 For more detail cf. Kazansky 1997: 42-43.
.1 to-pe-za , ra-e-ja , ku-te-se-jo , e-ka-ma-pi , e-re-pa-te-jo-qe ,
a-pi-qo-to , e-ne-wo-pe-za , qe-qi-no-me-na , to-qi-de 1 .2 to-pe-za , e-re-pa-te-ja , po-ro-e-ke , pi-ti-ro2-we-sa , we-pe-za ,
qe-qi-no-me-na , to-qi-de 1 [[to-pe-za]] .3 to-pe-za , ku-te-se-ja , e-re-pa-te-jo , e-ka-ma-pi , a-pi-qo-to , e-ne-wo-pe-za , ko-ki-re-ja
Let us now pay attention to the word order, which is, in accordance with the household genre, extremely rigid. This rigidity is undoubtedly somehow related to the marking of keywords, each followed by attributes: «a table (to-pe-za) (made of) stone, (made of) ebony, with supports, round, with nine legs ...»; others postpositive attributes follow in agreement/apposition to the word defined.
Such textual amendments are the only evidence available of the actual division of the sentence in Mycenaean. Undoubtedly, the texts also contain a syntactic break attested in a number of texts; cf., e.g., PY Eq 146 where violations occur in the title which is followed by three phrases, each in a separate line: «and that's how much Proteos (Leprios, Admeos) has; all-in-all grain so much «, but later we again observe a transition to another syntagmatic division: «and that's how X, a physician, hired//has grain so much», where the intonational division would have required a different arrangement into three syntagmas: o-da-]a2, [ ]-me-no, i-ja-te (end of syntagma), o-na-to, e-]ke (end of syntagma), to-so-de pe-mo GRA 1. In text PY Eq 146 quoted below such inconsistencies are marked in bold:
.1 o-da-a2 te-re-ta , e-[
.2 ko-ro , to-so-de , pe-mo [[GRA ]] GRA [
.3 o-da-a2 , po-ro-te-u , e-ke , to-so-de , pe-mo GRA 1
.4 o-da-a2 , re-pi-ri-jo , e-ke , to-so-de , pe-mo GRA 1
.5 o-da-a2 , a-de-me-we , e-ke , to-so-de , pe-mo GRA 1 V 3
.6-8 vacant
.9 o-da-]a2 , [ ]-me-no , i-ja-te , o-na-to ,
.10e-]ke ,. to-so-de pe-mo GRA 1
.11o-da-a2 , a-si-wi-jo , i-qo-na-to-mo ,
.12o-na-to , e-ke[ to-]so-de pe-mo GRA T 7
A Mycenaean scribe who read word by word (not like us, reading sign by sign!) could hardly be troubled by ambiguity of reading. Thus, a group of signs, i.e., a word, denoting a geographical concept, signals that the sentence opens with Nomen Loci (loc.) (where NL(loc) stands for a separate syntagma most frequently consisting of a single word, although combinations of two words, e.g., ze-i-ja ka-ra-na, also occur) which is, in its turn, followed by another syntagma beginning the main text which
may consist of several syntactic structures.
Returning to the word, which was undoubtedly thought of as the basic language unit in the Mycenaean time, it should be noted that Mycenaean knows no fundamental difference between a word written as a linear sign sequence and as an ideo- or logogram.
Comparison with materials from later epochs shows that a tendency towards marking off the syntagma boundaries emerges rather early in Greek alphabetical inscriptions and that it is undoubtedly linked with the convenience of reading84. We have seen that the Mycenaean scribes, with the same purpose, resorted to rewriting a word in the next line instead of interpunction. In both cases syntagmas are not consistently marked.
I have attempted to compare abbreviations of the Mycenaean time with those of the Hellenistic papyri. The feature that both of them have turned out to have in common is the tendency towards abbreviation at the expense of inflection or by writing only the initial syllable (sign) of a word. The latter is frequent in modern alphabetical texts as well, e.g., in encyclopedia articles.
What conclusions can be drawn from these comparisons? First of all: the continuity of the Mycenaean tradition. Abbreviations are unusual in archaic Greek inscriptions. They become common only in the Hellenistic epoch (abbreviations on Greek ostraca are excluded; this is a special case, as the first character of a name sufficed to correctly identify the person to be ostracized). Abbreviations in Hellenistic and medieval texts are a fruit born by literacy, not lack of literacy. The state of affairs could hardly be different in the Mycenaean time. In the analysis of Mycenaean texts, the level of scribes' literacy is, I am afraid, significantly underestimated. Abbreviations in Latin epigraphic texts grow in number hand in hand with the spread of literacy, reaching their maximum in the epoch of the Empire85. A similar situation is demonstrated by the modes of abbreviation in early Latin texts. Defining the notion of suspensio, O. A. Dobiash-Rozhdestvenskaya wrote: «The simplest and most natural abbreviation system (especially if the author uses his native language, of which both he and his reader have a good command) consists in not writing a word up to the end, omitting its end; [a device] which in old
84 «punctuation does not appear in any of the earliest inscriptions except that of Pithekoussai» (Jeffery 1961: 50).
85 As an explanation of this phenomenon, the continuity of the uninterrupted existence of the Roman official tradition in writing documents is often pointed out to, e.g., E. V. Fedorova's wording; «The existence of a unified political power on a vast territory in the course of many centuries naturally led to the appearance, for texts similar in contents, of uniform expressions which could be abbreviated because they had established themselves in usage and become commonly known» (Fedorova 1982: 17).
times was termed «suspension» (suspensio) in the Latin world. In an extreme case, it reduces a word to a single (initial) character, littera singularis, which was long ago termed sigla, sigle» (Dobiash-Rozhdestvenskaya 1936: 166). This principle of suspense, which flourished while Latin was a living language and flexions of words could be easily guessed, was later superseded by the principle of contraction, which consists in contracting the word by dropping its medial, less characteristic, parts. In this case, the beginning and the end of a word, «sometimes even the whole flexion» (Dobiash-Rozhdestvenskaya 1936: 167), are invariably preserved. Significant for the conception proposed here is the chronological order of these two contrary abbreviation principles. It may be stated that, to use Dobiash-Rozhdestvenskaya's formulation, «a natural explanation for preponderance of suspense in the epochs of Antiquity and Caroling Renaissance and contraction during the Middle Ages <is>... that a writer could afford abbreviation by dropping the flexion only in a social environment in which a free use of the language he wrote in was presupposed, and in whcih a firm knowledge of agreement rules and of the correct usage of inflections had been fostered since childhood or had been thoroughly studied at school» (Dobiash-Rozhdestvenskaya 1936: 169). On the other hand, the frequency and admissible degree of abbreviation may vary depending on the genre to which a text belongs. Household documents, which generally allow an easier formalization of material, are usually permeated with abbreviations to a greater degree than texts of other genres.
Thus, we may assume that Mycenaean texts were meant for reading by another person, moreover, for an unambiguous reading. This may have involved the use of certain devices in addition to the signs of syllabic script, such as the correct division into syntagmas in case of household texts.
References
Chadwick 1967 - J. Chadwick. The decipherment of Linear B. Second ed., Cambridge.
Chadwick 1976 - J. Chadwick. The Mycenaean World. Cambridge. Dobiash-Rozhdestvenskaya 1936 - O. A. Dobiash-Rozhdestvenskaya. Istoriya
pis'ma v sredniye veka. Moscow; Leningrad. Fairbanks 1937 - G. Fairbanks. The relation between eye movement and voice in oral reading of good and silent readers // Psychol. Monographs. Vol. 48 (215).
Fedorova 1982 - E. V. Fedorova, Vvedenie v latinskuyu epigrafiku. M., Gavrilov 1994/1995 - A. K. Gavrilov. Chtenie pro sebya v drevnosti (Obzor
antichnyh svidetelstv) // Hyperboreus. Studia Classica. Vol. I, fasc.2. Jeffery 1961 - L. Jeffery. The local scripts of Archaic Greece. Oxford.
Kazansky 1997 - N. N. Kazansky. Funktsionalnye kharakteristiki mediya v yazyke krito-mikenskih nadpisey // MOYSEION. Professoru A. I. Zaitcevu k semidesyatiletiyu, St.-Petersburg.
Luria 1957 - S. Ya. Luria. Yazyk i kultura mikenskoy Grecii. Moscow; Leningrad.
McNamee 1981 - K. McNamee. Abbreviations in Greek Literary Papyri and Ostraca. Scholar Press (= Bulletin of the American Society of Papyrologists. Supplement 3).
Molchanov, Neroznak, Sharypkin 1988 - A. A. Molchanov, V. P. Neroznak, S. Ya. Sharypkin. Pamyatniki drevneyshey grecheskoy pismennosti. Vvedenie v mikenologiyu. Moscow.
Neumann 1958 - G. Neumann. Zur Sprache der kretischen Linearschrift A // Glotta. Bd 36. H. 1/2.
Olivier 1987 - J.-P. Olivier. Des extraits de contracts de vente d' ésclaves dans les tablettes de Cnossos // Studies in Mycenaean and Classical Greek Presented to John Chawick. Salamanca (= Minos, 1987. Vol. XX-XXII).
Perpillou 1977 - J.-P. Perpillou. Répantires de scribes // Revue de philologie 51,2.
Préaux 1959 - C. Préaux. Du «Linéaire B» créto-mycénien aux ostraca grecs d' Égypte // Chronique d' Égypte. N 67.
Schlesinger 1968 - I. M. Schlesinger. Sentence structure and the reading process. The Hague; Paris, (=Janua linguarum. Series Minor, 69)
Tronsky 1973 - I. M. Tronsky. Voprosy yazykovogo razvitiya v antichnom obschestve. Leningrad.
Ventris, Chadwick 1973 - M. Ventris, J. Chadwick. Documents in Mycenaean Greek. 2-nd ed. by J. Chadwick. Cambridge.
