Text-related and scribe-related epigrams

An important distinction to be made when one studies book epigrams is the one between text-related and scribe-related epigrams.Footnote ⁸ Each of these two types of epigram has a different context and must therefore be read and understood in a slightly different way.

Text-related epigrams are, in a broad sense, comments on the main text of the manuscript. This may be a metrical paraphrase, a clarifying note to the text, or simply a comment by the scribe regarding the content or the author of the text. Text-related epigrams are therefore very much bound to the specific manuscript they are written in and to the specific text that they accompany, in that they only make sense next to that text. They are sometimes even so closely connected to the main text that they were felt by later scribes to be part of it and copied as such. In this respect, text-related epigrams are quite fixed and not open to changes by the scribe, who generally copied them as faithfully as possible.

An entirely different and, for us, more interesting type of book epigram is the scribe-related epigram. The main topic of these texts is scribal activity, such as a prayer by the scribe before he starts copying, an expression of joy as he finishes the manuscript, or a metrical colophon. Scribe-related epigrams are much more fluid than text-related epigrams because they have very little connection to the main text and can therefore easily ‘migrate’ from one manuscript to another. They were not felt to be fixed, which caused scribes to adopt and adapt the texts to their own needs. When Christine Thomas defines ‘fluid texts’ she asserts that they behave ‘similarly to oral tradition, with each manuscript representing a new “performance” of the work in another context. Yet this occurs on the level of written text.’Footnote ⁹ This is certainly applicable to scribe-related book epigrams. Despite their fundamentally written character, they share certain characteristics with oral texts, since they were cited from memory rather than copied from paper,Footnote ¹⁰ which in turn leads to a very changeable type of text, with each variant being a performance in its own right and with its own value.

The ‘ὥσπερ ξένοι’ epigrams

The popular scribal ‘ὥσπερ ξένοι’ epigrams,Footnote ¹¹ which are the subject of the present article, are an example of scribe-related epigrams. The different occurrences of these epigrams abound in variation, including metrical variation on the dodecasyllabic type, which will prove to be important for our understanding of the pragmatics of Byzantine metre. Treu, Brock and Lemay have collected numerous occurrences of the ‘ὥσπερ ξένοι’ epigrams and they focus on lexical and grammatical variation.Footnote ¹² However, none of these studies had access to the collection of the DBBE, in which more than one hundred and fifty occurrences of this type have so far been collected at the time of writing,Footnote ¹³ nor did they focus on the metrical variation in the different occurrences. The following example, metrically an accentual dodecasyllable, with strong caesura after the seventh syllable, has by far the most occurrences (24):Footnote ¹⁴

The following, longer, variant with the same metrical structure has an impressive amount of occurrences as well (19):Footnote ¹⁷

The orthographical, lexical and metrical variation on these two examples is virtually endless, as authors added lines and words with apparently little regard for the pattern of the dodecasyllabic metre. The epigram was so popular – with over one hundred and fifty occurrences, no doubt we only slightly gain a glimpse of the real extent of its popularityFootnote ¹⁹ – that it must have been widely known, and therefore scribes must have had its blueprint in the back of their minds as they produced their own version of it.Footnote ²⁰ This may to some extent explain the vast variation in the various occurrences, as every scribe could freely add to these epigrams as they pleased. However, it is interesting to note that there are discernible patterns in the recurring metrical deviations, which suggests that something more is going on here than simple idiosyncrasies or irregularities.

Metrical variants

What we call ‘variants’ are often simply dismissed as mistakes. At first sight, that is indeed what they appear to be: mistakes by a scribe who was incapable of composing correct dodecasyllables. However, more seems to be at work here. The sheer number of deviations is staggering and might make one question whether any of these scribes knew what a dodecasyllable looked like at all. Of the one hundred and fifty nine ‘ὥσπερ ξένοι’ epigrams collected to date, only a mere seventy-nine are written entirely in correct dodecasyllables, while the remaining eighty all contain a metrical irregularity of some sort. It hardly seems likely that none of these eighty scribes knew how to count up to twelve syllables.

What is even more interesting is that the metrical variants of the ‘ὥσπερ ξένοι’ epigrams are not random, but instead the same types of ‘mistakes’ keep on recurring. One can as a matter of fact neatly subdivide the variants into three groups: decasyllables, decatetrasyllables and decapentasyllables.

We start with the least common variant, which is the decasyllable. The following example contains an irregular decasyllabic line in the third verse, whereas the rest of the epigram is composed in correct dodecasyllables:

This is yet another argument against the assumed ignorance of the scribes, who were supposedly unable to compose correct dodecasyllables. It is clear that this scribe knew perfectly well how to do it, as he did it correctly in three out of four lines. Indeed, it hardly seems likely that he copied these lines from another manuscript, but made an error whilst copying line 3 (resulting in a decasyllable). As mentioned earlier, these popular formulas were most often not copied from parchment, but instead reproduced from memory. This means that our scribe must have read the extended version of the ‘ὥσπερ ξένοι’ formula somewhere else, reproduced it, but thought it necessary to change line 3 into a decasyllable. He knew what a dodecasyllable should look like (cf. lines 1, 2 and 4), but seemingly did not mind deviating from it. Should we then consider line 3 to be a mistake, or simply a variant? Was this felt by the Byzantine public, who read and/or heard this epigram, to be wrong or not ?

It is important to notice that the deviating line exhibits a medial caesura, dividing the verse into two metrical cola of five syllables. The following, definitely less elegant, yet very similar, variant has a decasyllabic first line with the same colometrical division:

Thus we find a pentasyllabic colon on either side of the strong caesura at B5,Footnote ²⁴ which is of course metrically irregular within the framework of the dodecasyllable, but nonetheless understandable because the pentasyllabic colon can occur both in pre- and post-caesural positions in a regular dodecasyllable.

The second largest group of metrical variants in the ‘ὥσπερ ξένοι’ epigrams consists of the decapentasyllable. Two examples are the following:

In both epigrams, the decapentasyllabic line is the last one (if read with synizesis: καὶ οἱ = κι͜ οἱ) and both times it is the odd one out in an otherwise dodecasyllabic epigram. We may assume in these two cases that the last line is meant to be a political verse and therefore belongs to another metre entirely. However, the question still remains as to why the scribe thought it suitable to suddenly switch from the dodecasyllable, which was a very familiar metre for book epigrams and had a rather archaic and therefore educated ring to it,Footnote ²⁷ to the political verse, which was felt to be ‘unmetrical’ (μέτρον ἄμετρον),Footnote ²⁸ rather low-brow, and much more suited to didactic poetry and longer narratives, such as epics, romances, and verse chronicles. Is it a mistake again or simply variation?

The last and by far the largest group of metrical deviations is the group of the decatetrasyllables. No fewer than thirty-nine occurrences display one or more decatetrasyllabic lines. The following occurrence is typical:

The two extra syllables in the last line are caused by the repetition of ‘ἰδεῖν’ in the second line, ‘gegen das Metrum’, as Treu puts it.Footnote ³⁰ This seems to suggest that the repetition here is due to inadvertence on the part of the scribe, who may have copied the word unwittingly in order to create a stylistic parallelism with the first line, at the expense of the metrical regularity of the verse. It looks as if this hypothesis is further corroborated by DBBE 22, where the second ‘ἰδεῖν’ in the last verse seems to have been erased, possibly because the scribe realized he had written a decatetrasyllable instead of a dodecasyllable.Footnote ³¹

However, stylistic parallelism does not account for the strikingly large number of occurrences with a decatetrasyllable in the last line (31), in which ‘ἰδεῖν’ is sometimes replaced by ‘εὑρεῖν’,Footnote ³² nor for occurrences such as the following, where the decatetrasyllable occurs in the first line instead:

In this case, the two extra syllables are caused by the addition of ‘γένους’ at the end of the first line, without any analogy to explain the metrical irregularity. The following occurrence even contains two decatetrasyllables, in the second and third lines respectively:

There are eight more occurrences of this type with a decatetrasyllabic line somewhere other than in the last line,Footnote ³⁵ which suggests that it really was not that uncommon for scribes to produce decatetrasyllabic lines – with or without an analogy to spur this on. Moreover, parallel to the decasyllable, the decatetrasyllable invariably has a strong medial caesura at B7, which divides the verse into two heptasyllabic cola. An interesting variation on this pattern is the following:

The last two lines of this occurrence, if read without synizesis, consist of an octasyllabic colon before, and a heptasyllabic colon after, the caesura. In other words, they would result in two decapentasyllabic verses with strong caesurae after the eighth syllable. However, if read with synizesis (‘καὶ ὥσπερ’ = ‘κι͜ ὥσπερ’ and ‘τοιούτῳ’ = ‘το͜ιούτῳ’), the same lines would again scan like decatetrasyllables, with a heptasyllabic colon on either side of the caesura at B7. Given the prevalent stress pattern of the first colon (X́XX́XX́XX), it makes more sense to read both lines as decatetra- instead of decapentasyllables.

In this context, it is interesting to note that Lauxtermann mentions a potential predecessor of these decatetrasyllables, when he cites a ninth-century hymn from Barb. Gr. 310 that was presented in paired heptasyllables.Footnote ³⁷ Nevertheless, we must certainly not assume that we have stumbled upon a new Byzantine metrical pattern here, since decatetrasyllables were, despite everything, still considered to be irregular: there are no examples of epigrams written entirely in decatetrasyllables, but instead they are always the odd ones out in a dodecasyllabic epigram. In this regard, they were irregular, but still not so irregular that it was felt to be problematic to produce them on a relatively frequent basis.

A cognitive-linguistic analysis

The occurrence of deca-, decatetra- and decapentasyllabic lines in what is essentially a dodecasyllabic epigram is usually explained as being a metrical irregularity or a scribal error. As we have shown, however, simply dismissing these irregularities as mistakes is not entirely satisfactory. Why do the same types of mistakes keep on recurring? And why do they arise so very frequently? How do such ‘errors’ occur exactly? To understand this phenomenon, we have to take into account the cognitive mechanisms underlying the process of versification. The production (and interpretation) of verse is in important respects comparable to the production (and interpretation) of speech. This is particularly evident in the case of (conceptually) oral poetry, such as Homer,Footnote ³⁸ but it is also true for fluid texts, such as the ‘ὥσπερ ξένοι’ epigrams. As was already mentioned earlier, fluid texts – such as scribe-related book epigrams – are texts with numerous changes and variations in their different attestations, as they could often be rewritten and adapted to new needs.Footnote ³⁹ They therefore behave ‘similarly to oral tradition’,Footnote ⁴⁰ because they were cited from memory rather than copied from paper. In a similar vein, Bakker calls the (conceptually) oral poetry of Homer ‘special speech’.Footnote ⁴¹ Like speech, poetry is not composed of long, continuous stretches, but of shorter units, called ‘idea units’ or ‘intonation units’ (IUs) by the American linguist Wallace Chafe.Footnote ⁴² In recent years, Chafe's theory of IUs has been consistently and successfully applied to the analysis of the Byzantine πολιτικὸς στίχος poetry by Soltic, who has shown convincingly that the cola of the Byzantine metres are the metrical equivalents of IUs.Footnote ⁴³ Cola are in essence cognitive units, both conceptually (qua ‘idea’) and formally (qua ‘intonation’), which function as the building blocks of the verse, an insight anticipated by Mackridge and Lauxtermann.Footnote ⁴⁴ The latter's ‘principle of pairing’Footnote ⁴⁵ explains the origin of the decapentasyllable as the pairing of an octa- and a heptasyllable, two independent metres which were often paired to form decahexa- and decatetrasyllabic verses respectively.Footnote ⁴⁶ Lauxtermann very appropriately observes that ‘the juxtaposition of two metrical segments rather rudimentarily corresponds to certain cognitive processes of the human mind.’Footnote ⁴⁷

It is interesting to note that this principle of pairing is not something that is only limited to Byzantine metrics. This very same concept of pairing together shorter units is applied in a similar way in many Byzantine rhetorical and liturgical texts, mostly in order to emphasize an antithesis or a parallelism. This is what is called the ‘commatic style’, which is used most often in Asianic rhetorical texts (as opposed to Attic oratory).Footnote ⁴⁸ It is not very surprising that Byzantine oratory and Byzantine accentual poetry exhibit striking similarities, since the Byzantines did not consider these two categories to be separate from one another, but instead conceived them to be two ends of one continuum. As Lauxtermann states: ‘In the Byzantine world . . . prose and poetry dance to the same tune and respond to the same rhythmical rules.’Footnote ⁴⁹

The principle of pairing suggests that the composition of a verse – any verse – is a cognitive process which not only helps poets to produce their verses by stringing together cola as cognitive building blocks, that is to say, IUs, but which also assists the poets’ audiences to process these IUs by tying them together, one after the other. The principle of pairing also explains the occurrence of ‘irregular’ verses within what is otherwise a standard dodecasyllable. The scribes write, or rather compose, their verses by pairing cola which, if paired improperly, may result in decasyllables (5+5), decatetrasyllables (7+7) or even decapentasyllables (8+7), since octasyllabic cola were as common as penta- and heptasyllabic cola during the period under scrutiny here. A good example is the following occurrence, which combines a dodecasyllable, a decatetrasyllable and a decapentasyllable:

The cognitive independence of the colon as an IU is further illustrated by the fact that it resembles the formula of epic poetry, as in the following occurrence:

The extension of the standard book epigram by several lines – in which seafarers, soldiers, sick people, merchants, and sometimes fishermen or prisoners of war underline the parallelism with the writers – is quite common and different scribes often opt for different similes, sometimes only including the seafarers or sometimes inventing even more comparisons. However, in this particular occurrence, something seems to have gone astray in the third line. The simile with the soldiers is usually the following one: ‘καὶ οἱ στρατευόμενοι | _B7 ἰδεῖν τὸ νῖκος,’Footnote ⁵² while the idea of profit commonly occurs in a simile about merchants: ‘καὶ οἱ πραγματτεύοντες | _B7 ἰδεῖν τὸ κέρδος,’Footnote ⁵³ both of which make more sense in terms of content. What seems to have happened here is that the scribe mixed up these two very well-known verses in his head and merged them into one.Footnote ⁵⁴ More precisely, he paired cola belonging to two different verses, which confirms the idea of metrical cola as being cognitive units and building blocks of the verse.