Sale notes that Parry's original work with Homer and oral tradition has largely been discarded, thus calling Parry, Homeric scholarship, and scholarship about oral tradition into question. Parry may well have been excessively formulaic, but Sale's opinion is that the theory, in concept, was sound (Sale 1996, 374-375). To analyze this anew, Sale sets out to evaluate what Parry's conclusions were. It is not at all unreasonable that an oral poet would have eventually adopted writing and revised some original ideas and passages. What Parry demonstrated is that Homer, in fact, "preserved the signature of the oral poet far too clearly for us to speak of a literate poet merely infleunced by, or even steeped in, oral poetry" (Sale 1996, 376). Sale defends the idea of Homer as a consumate artist in performance composition.
Parry's simple thesis is that Homer's diction is "formulaic." Such formulae are useful in oral composition, and can be categorized in some way (Sale 1996, 376). Sale sees Parry as continuing to think of Homer as a traditional poet, working with a relatively formulaic tradition, through the remainder of his career (Sale 1996, 377). Sale grants that at times there are formulae in Homer which seem to exist primarily for metrical convenience. In fact, it may be that Homer created some of these formulae and used some of them from knowledge of source material.
Counter to Parry, Sale thinks the audience would be unlikely to be indifferent to the use of the formulae, but would find them important in understanding the characteristics to which they point (Sale 1996, 377).
Sale reviews ten arguments, all but one directly taken from Parry's work.The arguments can be categorized into four general patterns, quantity of formulae, qualitative nature of formulae, irregularities in meter, and comparison with other works in the body of epic (Sale 1996, 378). Some of these arguments lead to "more modest claims, and make a point that Parry did not reckon with" (Sale 1996, 378). While the extrapolations may have had faults, the theory seems to stand.
To evaluate the claims, Sale frequently refers to Quintus os Smyrna, to Avdo Medjedovic, and, of course, Homer (Sale 1996, 378-379). "Quintus has about 8,800 lines, the Zenidba 12,300, the Odyssey about the same, the Iliad nearly 16,000" (Sale 1996, 379). Comparisons can be made, in general, with some adjustment for the length of the different works.
The first argument is based on numbers of formulae (Sale 1996, 380). From an intuitive point of view, we expect that use of formulae can be to the great advantage of the oral poet. Yet the identification of a formula may be a challenge. Sale observes that Parry himself definds "formula" in multiple different ways. Sale describes seven different ways Parry identifies formulae (Sale 1996, 380-382). Sale then discusses the difficuulties of the different categories of formulae. The requirement of identity in wording or identity in location within a metrical line can lead to debate about whether we are looking at one formula or more than one (Sale 1996, 383).
Comparing a selection of formulae, the Odyssey shows a great number of the categories Sale chooses to inspect. Yet Quintus has approximately the same density, or even greater, and he was clearly a writing poet (Sale 1996, 384). The analysis of particular formulae does not necessarily point to a particular compositional method, though it does point to a style, which may be considered a formulaic style. When Sale compares Homer and Quintus to Avdo, it becomes clear that Avdo, an oral poet, uses a lower density of identifiable formulae (Sale 1996, 386).
The second argument is based on formulae which are regularly used (Sale 1996, 386). The question of what "regularly" means is, not surprisingly, a difficulty. Sale creates a frequency graph which indicates some formulae occurring more regularly than others. Some formulae can serve many purposes, so are more likely to appear frequently (Sale 1996, 388). Quintus has a much lower rate of regular-formula occurrence. Avdo has fewer than Homer, but about four times as many as Quintus. Sale suggests that the oral composition of Homer and Avdo would be the likely cause. Quintus, as a literary poet writing in an oral style may not have captured the formulaic pattern as adequately (Sale 1996, 389).
The third through fifth arguments deal with the quality of formulae rather than the frequency. Argument number three, Sale calls the "multi-purpose-formula argument" (Sale 1996, 390). The formulae normally point to likely referents, fit the major parts of the hexameter line, serve as sense units in the line, and evoke important traditions. The formulae can be used in various places and ways (Sale 1996, 392). These formulae occur over 2600 times in the Odyssey (Sale 1996, 393). Sale asks whether it is necessary to be an oral poet to use these multi-purpose formulae. In fact, Quintus uses them, though not as frequently as Homer (Sale 1996, 394).
Sale calls the fourth argument the "argument from extension" (Sale 1996, 394). Here, if this reader understands correctly, the formulae can be used in multiple ways. They are used in ways which might be expected as well as in ways which might not be expected (Sale 1996, 397). Sale's question here is whether literate poets can use formulae in extension. In fact, Sale finds that Virgil does use these systems, though Quintus does not (Sale 1996, 397).
Sale's fifth argument is the "argument from infrequent formulae" (Sale 1996, 398). Here a formula is created according to a predictable pattern but using a noun which is not part of that typical pattern. (Sale 1996, 399). These formulae are very useful for rapid composition. Sale again asks whether a literate poet would be able to use this system, either purposely or intuitively. Quintus does use the infrequent formulae, quite a lot. For this reason, Sale does not consider the use of the infrequent formulae as a clear marker of orality or literacy.
The sixth argument is an argument from economy (Sale 1996, 400). This argument says that the oral poety will generally not have two formulae with the same meaning and the same metrical structure. Sale notes several difficulties in Parry's argument. Most notably, he tends to blend several different uses of formulae under this category. However, the concept may, in fact, be worthy of exploration. The economy is likely well identified based on the referent, not only the meaning of the formula. When viewed that way, the poet is using a metrically appropriate formula to refer to the right referent in a way which can avoid difficulties in the overall sense of the passage (Sale 1996, 402). Sale does ask whether this is a purely oral compositional feature. Quintus has only five nouns that occur in formulae which could demonstrate or violate economy. "All five nouns behave themselves" (Sale 1996, 404). However, Quintus does violate economy frequently, particularly with the "infrequent formulae." The instances strike Sale as existing purely for the sake of creating variety, which is not a characteristic of oral composition (Sale 1996, 405).
The seventh argument is based on the location of a word in the line. Sale observes that this is not an argument made by Parry (Sale 1996, 405). It is relatively common to find a word appearing at a predictable place in a hexameter line. Homer does this about 80% of the time, while Quintus does only about 60% of the time (Sale 1996, 406). Avdo uses localization approximately as often as Homer. Sale suggests that the literary poet knows about localization but does not see a particular use for it, so will use the tool less frequently (Sale 1996, 407).
Sale's eighth argument is the "argument from metrical irregularity." Here because of the need to compose at the speed of the performance, the composer may have a metrical error. (Sale 1996, 407).
The ninth argument suggests that oral poets in any culture and language will have similarities. Sale calls it "the argument from external analogy" (Sale 1996, 408). At issue here is whether formulae are all traditional in some way. The repetition of phrases may carry over to different cultures as well. However, South Slavic oral poetry does not seem to have the level of oral repetition tha we find in Homer. While there is a good deal of repetition within individual authors, the different authors have different patterns (Sale 1996, 408). Sale's opinion is that the correlation among different oral authors is actually quite high, more than one would expect when crossing the bounds of culture and language. Quintus, as a literate poet, makes a good attempt at the oral style, but his correlation is not so high as that of Homer and Avdo (Sale 1996, 409).
The tenth and final argument is based on "Parry's statement that 'we know surely that Homer's poetry is governed by factors unknown to later Greek poetry'" (Sale 1996, 410). This is a thesis that Parry apparently never followed up in detail. Sale asks what an oral poet might know or do that a good imitator would not. In fact, Quintus has some differences from Homer, while Avdo does not have the same differences, and the argument has been made that Avdo is much closer to Homer. Yet the "why" is never answered in this equation, nor is the possibility that, at least at some points, Homer worked or was worked with in a literary manner.
Sale returns to the original question, though, and asks what defending Milman Parry might inform us about the Homeric art. It is possible to analyze Homer in terms that Parry did, and to find that Homer's use of formulae is consistent with that of other oral poets (Sale 1996, 411). Yet it remains unclear what Homer could or could not do. It does remain clear that Homer used most or all the tools typically used by oral poets.