The concept of false dichotomy in Homeric textual criticism

By Alejandro Abritta

Presented originally as “El concepto de falsa dicotomía en la crítica textual homérica”, XXVII Simposio Nacional de Estudios Clásicos. III Congreso Internacional sobre el Mundo Clásico, Santa Rosa, septiembre de 2022.


As with any discipline of considerable tradition and size, Homeric studies often function as a series of watertight compartments. Those who approach the poems from a literary perspective do not know or are not interested in those who approach them from a linguistic perspective, who do not know or are not interested in those who approach them from a textual perspective, who do not know or are not interested in those who approach them from an oral studies perspective, who do not know or are not interested in those who approach them from a literary perspective. Of course, the reality is infinitely complex, and there is no lack of authors with a vast knowledge of the area as a whole, nor of those who successfully combine different perspectives in their analyses. In practice, however, there are methodological impediments that sometimes seem insurmountable: in the literary analysis of a passage, it is necessary to assume the meaning of certain words or the syntax of certain phrases, even when these present difficulties from a linguistic point of view.

In itself, this is not a profound problem, nor specific to Homeric studies. Again, it is something that happens in any area or discipline of any size, where the density of scholarly output is greater than an individual researcher can encompass. But there are cases where such methodological incompatibilities or mutual mismatches do constitute a major shortcoming, namely those where the object of study itself is affected by the drawback.

This phenomenon is, in a sense, common to any discipline. The data that are compiled are compiled based on a predefined methodology, generally with an already established objective. Observations are made with the tunnel vision of one who is looking for something specific. However, in the case of literary studies and work with pre-modern and especially ancient texts, the intervention on the object of study is more significant, insofar as that object of study is not observed or studied for a specific experiment or project, but to a large extent produced as an objective in itself. I am referring, of course, to editorial work.

Indeed, the constitution of a text is a fundamental work in philological studies, not only because it is one of the key objectives of the discipline, but also because this constitution is the basis for all subsequent studies. When we analyze a text on the basis of an edition, even if we exercise all the necessary precautions, we are inevitably traversed by the work of the person who constituted that edition.[1]

Which, again, is not too significant a problem in most cases.[2] Part of the training in classical studies is to understand how to work with the texts and the reality that they have only come to us through a very long history of mediation. In the case of Homeric texts, however, this is the least of the problems, because these texts are oral texts.

This orality has a fundamental consequence for its constitution in a critical edition: an oral text has no original version but is reworked on each occasion of performance. In the specific case of the Homeric text, this is a problem in itself, because the coexistence of theories about the process of fixation makes it difficult even to answer the question of whether this holds for the Iliad and the Odyssey. In other words, even accepting that there is no such thing as an “original” in an oral tradition, perhaps there is an “original” in an oral tradition that becomes written.

This is essentially the difference between those who defend the dictation theory and those who support the evolutionary theory.[3] For the former, there is an original text, which is the one dictated by the rhapsode, and this text is equivalent to any text written by the hand of an author. For the latter, the gradual process of fixation of the text makes it impossible to speak of an “original”.[4] This difference in approach is perhaps responsible for the fact that the debate has not permeated the main specialized editions, nor, for that matter, much of the rest of the scholarship.

But this is, if I may say so, too comfortable. It is clear that there is no “original” from the evolutionary point of view, but it is also clear that not even the strictest advocates of this model deny the importance of a textual methodology in Homer. However the text was fixed, we have two thousand years of written transmission between then and now. Much more important than this, most advocates of dictation theory today assume that, discounting the exceptional nature of this process, there is nothing special about the performance that fixed the text. Put another way, even if the “original” exists, there is very little in that original that distinguishes it (methodologically, of course) from the version of the same story that the rhapsode would sing somewhere else a week, a month, or a year before or after participating in the dictation process.

The most significant thing to note here is that this is particularly true in the case of the textual problems we face in the constitution of texts. Evolutionary theory and dictation theory propose fundamental macro-level distinctions,[5] but neither could ever deny that the chances of determining which specific variant is the “correct” one are often minuscule. Even assuming (I maintain this view) that there was one instance of dictation of the poem, knowing whether, for example, verse 5 of Iliad began οἰωνοῖσί τε πᾶσι or οἰωνοῖσί τε δαῖτα is not only impossible, but it is also impossible to determine the origin of these variants. Could they not both be part of the rhapsodic tradition, formulaic alternatives for the context? That this was the case there can be no doubt: in Il. 4.115 and 195, 205 the variants Ἀτρέος υἱόν and ἀρχὸν Ἀχαιῶν alternate at verse-end, the former a well-attested formula, the latter an expression appearing only in this context, but most likely formulaic.[6] Similarly, in 5.856 we find the variants ἐπέρεισε δὲ Παλλὰς Ἀθήνη [and Pallas Athena impelled her] and ἶν’ ἀπέλεθρον [and an immeasurable impulse impelled her], both formulaic expressions and both appropriate to the passage and the context. How to determine which was the one sung by the rhapsode on the day of the poem’s dictation?[7]

What is interesting about this is that the argument is not affected by, for example, the proposal of “variations by concordance.”[8] Even if one were to propose that Παλλλὰς Ἀθήνη was inserted to specify the action of the goddess, there is no criterion according to which it can be denied that such an insertion was made only in the written tradition and never before in oral performance.[9]

This means, therefore, that a multitextual approach is feasible regardless of one’s preferred theoretical approach to textual fixation. Now, is it possible to accomplish a multitextual edition of the poem? Critics of evolutionary theory have argued so at least since the work of Nagy (1996) and indeed have advanced this perspective in the Homer Multitext project and, more specifically, in the production of a multitextual edition of Iliad 10 in Dué and Ebbot (2010). The authors (p. 57), however, acknowledge the limitations of the physical format for such a project, and it is inevitable, upon studying the outcome of their work, to feel it is disappointing. Despite considerable effort to make something more than a papyri edition, the reality is that Dué and Ebbot’s “multitextual edition” of Iliad 10 is too much like West’s (1967 – ! -) Iliadic papyri edition to be considered a significant move forward.

Now, it is obvious that this is an unfair assessment. The physical book is totally inadequate to represent the reality of epic performance (a fact well known in the study of oral literature), regardless of the methodological precautions and efforts made. The real question is whether it is possible to produce a multi-textual digital edition of the Iliad, and I understand that the answer is “no,” because of similar deficiencies.

It is possible to demonstrate this with a reductio. What would an ideal multi-textual edition of the Iliad look like? It should fulfill at least three functions: first, it should preserve as valid the variants that it is possible to attribute to the rhapsodic variation; second, it should not assign priority to any of those variants, eliminating or restricting the idea of an “original” version superior to the others; and third, and this is the key point, it should be modified in different readings, as oral performances are different from each other.

From a technical point of view, this is not difficult to realize: we should generate a digital text that is reconstituted with each access, randomly introducing oral variants.[10] Thus, for example, this multitextual edition accessed once would yield a verse 5 of Iliad οἰωνοῖσί τε πᾶσι, accessed a second time would yield οἰωνοῖσί τε δαῖτι, and accessed a third time again οἰωνοῖσί τε δαῖτι, but the fourth would again οἰωνοῖσί τε πᾶσι, and also the sixth, seventh, and so on. Multiplied by the several dozen potentially oral variants preserved, this would imply a true multi-textual edition, which instead of one Homeric text would give us thousands.[11]*

As an exercise, such an edition would be invaluable. I think it is evident, however, that there would not be a single philologist on the planet, much less those engaged in literary criticism, who would use it. How does one comment on or analyze a text that changes every time one tries to read it? The question is fascinating (in fact, it is very much the question for the study of oral literature), but more from a theoretical than a methodological point of view, particularly when working with texts of such stable transmission as the Homeric poems. We can play with the problem of oral variants in our analyses, but I imagine that almost all philologists will draw the line at the real possibility of these variants intervening in their work.

Much more important than that, even that edition does not solve the problem, because the user would always find a variant the first time they enter, and that variant would be the default variant. This would create a bias in favor of that variant, and the perception that the different text entered the second time is a “worse” text.[12] And if this is true for the hypothetical variable multitextual edition, it is certainly much more so for the fixed editions that have been proposed so far.[13] If the reader will always consider the first thing he encounters in the text to be superior (and he is guaranteed to do so), any subsequent attempt to modify that impression has an additional processing cost that makes it impossible to produce a true “multitext”.

And this result, it cannot be forgotten, is only in the context of a work intended for philologists. If we broaden the perspective to translators, specialists from other disciplines and the general public, the impossibility of a multitextual approach becomes more apparent. The only way to, for example, translate a multitext is for the translation to be multitextual, which, in itself, is not too complex (particularly with the limited variations of the Homeric text), but in practice would be unacceptable to 99% of the audience for such translations.

We are thus at a difficult crossroads. On the one hand, we can persevere in generating multitexts, digital or not, with a very restricted audience and no real impact on the study of the poems. On the other hand, we can return to the traditional methodology of textual criticism, analyzing the variants in the Homeric text as we would do with the variants in the Platonic text. Neither option seems admissible, and certainly, neither is too satisfactory.

My proposal to avoid them – there is no doubt about it – shares both characteristics, but at least it has the advantage of incorporating the reality that we are working with an oral poem in the production of a traditional written edition. It is, as the title of this paper announces, the concept of false dichotomy. It is possible to define a false dichotomy as the coexistence, in the written transmission, of possible variants in the oral performance of a poem, about which literary, linguistic, or paleographic arguments are not admissible, and only the preferential variant can be determined quantitatively. As can be seen, the concept has two parts, one theoretical and the other methodological. The first is nothing more than the explanation of what I have been talking about since the beginning of this work: in the Homeric textual tradition, both ancient and medieval, we preserve variants that can be attributed to the variation between different performances. This, as we have seen, is hardly open to debate. The second part is the real key to the concept: if a variant is recognized as oral, then it must be considered a false dichotomy, and none of the usual arguments for determining which variant should be printed apply. The only way to determine this is preference in transmission, in analogy to what would happen in the oral tradition, where usages are fixed by quantitative preference. Put more simply, in false dichotomies, it is always appropriate to print the majority variant.

A few examples illustrate the methodology, which, strictly speaking, is too simple to require illustration.[14] The simplest type of false dichotomy is purely orthographic, the alternations of representation in the written language of identical or nearly identical phonetic sequences. The sequence /duriklytos/ may, for example, be printed as one word (δουρικλυτὸς) or as two (δουρὶ κλυτὸς), but the difference would be imperceptible in orality.[15] Some preserved manuscripts transmit two words, but most print only one, which is, therefore, to be preferred.

This type of false dichotomy, however, is the least interesting. Much more valuable are the (exceptional) cases in which there is a majority variant that the majority of publishers choose not to print, favoring for various reasons a minority one. There are examples of this in Il. 5.854 and 15.240, but perhaps the most peculiar is in 22.468, where most of the manuscripts bring τῆλε δ’ ἀπὸ κρατὸς χέε δέσματα σιγαλόεντα [and far from his head the radiant ribbons were scattered], but a minority, τῆλε δ’ ἀπὸ κρατὸς βάλε δέσματα σιγαλόεντα [and away from his head threw the radiant bonds]. Although the sequence has given critics headaches, once it is considered that the τῆλε may be very relative and refer to the way the ribbons slide or roll on the ground with the veil, it is clear that χέε is as acceptable as βάλε. We should not discuss which is the more appropriate to print: both verbs are possible, and the only argument that can be admitted is quantitative.

There are also cases where there are very valid reasons to defend a minority variant, but this is irrelevant to the admissibility of the two preserved variants. In 22.374, for example, three manuscripts (the Venetus as a supralinear variant) bring Ἕκτωρ ἢ ὅτε νῆας ἐνέπρηθεν πυρὶ κηλέῳ [<how much softer is to palpate> Hector than when he burned the ships with blazing fire], while most convey the aorist ἐνέπρησεν. The imperfect is more appropriate to the context and even seems to reinforce the irony in the anonymous speaker’s voice. West and Leaf, in fact, print it. But none of the variants can be rejected, so, even against our preference, it is appropriate to print the most common in the sources.

It should be noted, finally, that false dichotomies are not restricted to textual variants within a verse, but also include cases where entire verses are omitted by some sources.[16] An interesting example of this kind is found in 5.900-901, where the majority tradition brings τῷ δ’ ἐπὶ Παιήων ὀδυνήφατα φάρμακα πάσσσεν (with omission of 901) [Paian on him applied potions that soothe pains], but a minority brings τῷ δ’ ἐπὶ Παιήων ὀδυνήφατα φάρμακα πάσσων | ἠκέσατ’- οὐ μὲν γάρ τι καταταθνητός γ’ ἐτέτυκτο [Paian, upon him applying pain-relieving potions, | cured him, for he was not mortal at all]. That both versions are valid is relatively obvious, but it is reinforced because the minority version is identical to 5.400-401.[17] The peculiarity of this case is, first, that the majority variant omits the verse (in the vast majority of false dichotomies of this type, the converse is true), and, second, that the variation encompasses the preceding verse as well, a very unusual phenomenon.

The concept of false dichotomy for the editing of the Homeric text does not eliminate the problems of editing an oral poem, but it implies a recognition that the traditional criteria of textual criticism are not applicable in this one without precautions. The first question we must ask ourselves when approaching a work of this type (the Homeric poems, the Hesiodic poems, the Homeric hymns) is whether the variants we are studying are the product of the written or oral tradition. In the former case, the usual tools come into play; in the latter, we are left only to suspend judgment and accept the majority transmission. The alternative is to imply that we have a way of identifying which variant is more “original” than the other, and that is a question that, in the study of an oral text, is meaningless.

  • Abritta, A. et al. (2021) Ilíada: Canto 1. Traducción comentada, tercera edición, ampliada y corregida, Buenos Aires:
  • Bird, G. D. (2010) Multitextuality in the Homeric Iliad: The Witness of the Ptolemaic Papyri, Washington, DC: Center for Hellenic Studies.
  • Dinghra, N., Gorn, Z., Kener, A. and Dana, J. (2012) “The default pull: An experimental demonstration of subtle default effects on preferences”, Judgment and Decision Making, 7, 69-76.
  • Dinner, I., Johnson, E. J., Goldstein, D. G. and Liu, K. (2011) “Partitioning Default Effects: Why People Choose Not to Choose,” Journal of Experimental Psychology: Applied 17, 332-341.
  • Dué, C. and Ebbott, M. (2010) Iliad 10 and the Poetics of Ambush: A Multitext Edition with Essays and Commentary, Washington, DC: Center for Hellenic Studies.
  • Finkelberg, M. (2000) “The Cypria, the Iliad and the Problem of Multiformity in Oral and Written Tradition”, CPh 95, 1-11.
  • Foley, J. M. (1999) Homer’s Traditional Art, University Park, PA: The Pennsylvania State University Press.
  • Janko, R. (2000) “West’s Ilias,” CR 50, 1-4.
  • Jensen, M. S. (2011) Writing Homer. A study based on results from modern fieldwork, Copenhagen: The Royal Danish Academy of Sciences and Letters.
  • Nagy, G. (1996) Poetry as Performance: Homer and Beyond, Washington, DC: Cambridge University Press.
  • Nagy, G. (2001) “Homeric Poetry and Problems of Multiformity: The ‘Panathenaic Bottleneck'”, CPh 96, 109-119.
  • Reece, S. (2011) “Toward an Ethnopoetically Grounded Edition of Homer’s Odyssey,” Oral Tradition 26, 1-26.
  • West, S. (1967) The Ptolemaic Papyri of Homer, Wiesbaden: Springer.

[1] Particularly important here is the default bias, to which I will return below.

[2] And where it is (say, for example, much of the Shearer corpus), researchers are usually fully aware that it is.

[3] Cf. Abritta et al. (2021: xli-xlviii), with additional references.

[4] This is why they are the ones who have proposed to advance in multitextual editions, as will be seen below.

[5] Which, in fact, allows us to reject the former relatively forcefully (cf. Finkelberg, 2000, and the failed attempt to defend the approach in Nagy, 2001; cf. also Jensen, 2011: 214-247).

[6] Cf. Janko (2000: 3), who makes the same suggestion.

[7] Cf. additional examples and the same reasoning in Nagy (1996: 147-149) and Bird (2010).

[8] Cf. West (2001: 13-14, 146).

[9] The question must necessarily be simplified because it is presupposing the almost total persistence of the context of the verse between different performances of the poem, which is at least debatable. The point, however, stands, because the possibility of further variation reinforces it, not diminishes it.

[10] Recorded or not, depending on the objective. If we are talking about an “edition”, of course, the former is advisable, but the latter has some validity as a didactic exercise to understand the functioning of an oral tradition.

[11]* Huilén Abed Moure has developed a sample of this system. The result can be consulted here:, and at this address: the files used in its production.

[12] Product of the well-studied bias in favor of the default option (cf. e.g., Dinghra, Gorn, Kener, & Dana, 2012; Dinner, Johnson, Goldstein, & Liu, 2011).

[13] Including, naturally, the non-multitextual “oral” editions suggested by Foley (1999) or Reece (2011).

[14] The examples I will include have been commented on in the bilingual texts published in For the sake of brevity, I do not add references in the bibliography.

[15] Although the grave in δουρὶ could generate a small difference in pitch. It is precisely the possibility of these minute differences that allows us to include this group within the false dichotomies.

[16] One could even argue that these are the most basic cases of false dichotomies, given that the most significant inter-performance variation is the expansion or contraction in the narration of episodes.

[17] Which, of course, has led to the suggestion that 901 is an interpolation by concordance, a concept that is inadmissible in the context of an oral poem.

Deja una respuesta

Tu dirección de correo electrónico no será publicada. Los campos obligatorios están marcados con *

Scroll hacia arriba