Deborah Beck, Homeric Conversation
Part I. One-on-one Conversations. Chapter 1. One-on-One Conversations (Odyssey)
Chapter 2. One-on-one Conversations (Odysseus and Penelope)
Chapter 3. One-on-one Conversations (Iliad)
Part II. Single Speeches and Group Conversations. Chapter 4. Single Speeches and Variations on the Battlefield
Chapter 5. Group Contexts I—Assemblies
Chapter 6. Group Contexts II—Athletic Games, Laments
Appendix I. Breakdown of Direct Speeches in the Iliad and the Odyssey
Appendix II. All Participles that Appear in Reply Formulas
Appendix III. Full-verse Context-specific Introductory Formulas
Appendix IV. Full-verse Speech Concluding Formulas
“And what is the use of a book,” thought Alice, “without pictures or conversations?”
Lewis Carroll, Alice’s Adventures in Wonderland
When some other good singer takes five hours to sing a tale, I need ten. I don’t know whether that’s good or bad, but that’s how it is.
The Iliad and the Odyssey, as readers since ancient times have realized, derive much of their force and power from the many speeches of their characters. These speeches represent a high proportion of the poems in comparison to other narrative poetry: between one-half (Iliad) and two-thirds (Odyssey) of the Homeric epics consists of direct speech, while Apollonius’ Argonautica contains just 29% direct speech.  Because of the quantity and quality of direct speech in the Iliad and the Odyssey, and because the distinction between narrative and speech is one of the most basic in literature, a great deal of ink has been spilled on speech and speeches in Homeric poetry. However, virtually all of this ink has been used to write about “speech” rather than the joining of several speeches into a sequence to form a “conversation.”  This is surprising, in that conversation is an inherently important phenomenon in the Iliad and the Odyssey simply because of how frequently it appears: something over three-quarters of the direct speeches in the poems either elicit a response or are themselves responding to a prior speech.  Therefore, we do not truly understand how direct speech works in the Homeric poems if we consider it simply as speech and not in its broader conversational contexts. Moreover, some kind of conversation appears in virtually every kind of scene in both the Iliad and the Odyssey. It therefore follows that understanding Homeric conversation will provide a broad and powerful kind of insight into the Homeric epics themselves.
My goals for this book are as follows: to establish that conversation should be considered a type; to describe the regular patterns of speech sequence and formula associated with the type; and to analyze from an aesthetic standpoint conversations that vary from typical patterns in order to better understand how types work within the poems overall, how conversation shapes the themes of the poems, and how the Iliad and the Odyssey work as traditional oral art. Finally, in the conclusion, I will explore the ramifications of these ideas for larger questions of composition and reception, although these conclusions are necessarily speculative.
The remainder of the first section of the introduction gives a very brief overview of the two major areas that underlie my perspective on conversation, type scenes and formulas, and linguistics. It also lays out in more detail my approach to aesthetics. The second section elaborates on the reasons we should consider conversation to be a type, what scholars have done when studying other kinds of type scenes, and how formulas and type scenes relate to one another (II, Types and Homeric Conversation). The third section (III, Linguistic Perspectives on Conversation) looks at the way the Homeric poems talk about conversation. It then provides an overview of the different areas of linguistics that can best complement the notion of conversation we find in the poems themselves. The fourth lays out the different sequences in which speeches occur in conversation and the kinds of full-verse formulas that link these speeches together into a conversation (IV, Repeated Speech Sequences and Formulas in Conversation).
Conversation as a Homeric Type
One key aim of this book is to describe for the first time the variety of sequences and contexts in which a series of individual speeches may be linked together to form a conversation. Essentially, this descriptive process aims to establish conversation as a Homeric type alongside more well-known types such as arrival, sacrifice, and so forth. Although conversational type scenes do not contain a series of different elements, like other types do (e.g. killing the sacrificial animal, spitting it, roasting it, eating it), they do contain regular sequences of events—speeches, sometimes accompanied by specific context-dependent actions—which are joined together into a conversation by the narrator  through recognizable formulas of speech introduction and conclusion associated specifically with conversation. Accordingly, they should be considered types, and they can be studied using the same approaches (both descriptive and aesthetic) that have been applied to other kinds of type scenes. This process will show us the common patterns that govern conversation in Homeric poetry. In addition, it will demonstrate the underlying connections that link together apparently dissimilar activities like athletic games and laments.
Like other type scenes, conversation can provide stimulating and fruitful material for thinking about Homeric aesthetics.  When we understand the common patterns for formulaic speech frames  and for the sequences of individual speeches in a conversation, we can profitably study the departures from these patterns from an aesthetic standpoint. A coherent and useful system of oral aesthetics for the Homeric poems has long been a desideratum of Homeric studies. Recently, a consensus about the general outlines of oral aesthetics has begun to emerge among researchers working on Homeric poetry from an oral perspective. This consensus identifies the interplay between repeated elements (which may apply to the formulas, the individual narrative elements, and the sequence of these elements that appear in a type scene) and variations on these repeated patterns as a central feature of oral aesthetics. Well-established repeated patterns characterize both the order of speeches within Homeric conversations and the formulaic speech frames that join them together. Hence, Homeric conversations provide a very powerful and illuminating basis for a study of oral aesthetics in Homeric poetry.
Aesthetics and Homeric Poetry
What exactly do we mean by “aesthetics”? A recent survey article on the subject published in a cross-disciplinary journal devoted to aesthetics defines the problems that aesthetics is concerned with as follows:For Homeric studies, the nature and functioning of oral aesthetics has been a central question at least since the 1960s, when Nagler stated that “no coherent aesthetic theory has as yet emerged which would equip us to understand or appreciate the special nature of oral poetry as poetry.”  Since then, many scholars have developed useful ideas about what this aesthetic theory might look like. Indeed, it may now be said that a general consensus about oral aesthetics has emerged.
The nature and defining characteristics of art, the meaning works of art are said to have, how they may be judged, valued, or interpreted, the nature of imagination and of creativity, the kinds of experience offered by art, &C [sic]. 
Russo has put forward a particularly clear and concise statement of one view of oral aesthetics that is held by a number of scholars interested in the oral aspects of Homeric poetry:So, two elements of oral aesthetics are repetition and variation, and fullness of expression. Fullness of expression, stated another way, means that length constitutes emphasis within an oral poetic system.  In fact, the length of the speeches in the Homeric poems is in and of itself one of the key components of the epics’ fullness of expression.  Current notions of Homeric oral aesthetics have moved away from the idea that original language is necessary in order to achieve an aesthetic effect. Instead, many scholars now find the unique artistic merit of the Homeric poems in their “individual and innovative use of traditional material”  rather than in their invention of entirely new or original material for particularly dramatic or important moments.
[Homer] composes in a style guided by the ear and meant to be heard, a style that pleases through verbal play based on an aesthetic of repetition and variation, and of relaxed fullness of expression whenever the context allows it. 
In sum, “oral aesthetics” as used in this work refers to a way of evaluating and appreciating the Homeric poems as works of art by using the following criteria: i) artistic excellence comes primarily through a gifted use of traditional materials rather than through inventing new things; ii) interplay between repeated language or story patterns and variations from these repeated patterns has a key role to play in creating the aesthetic pleasure of the poems and in emphasizing important points in the story; iii) unusual length creates emphasis. Although this consensus on the nature of oral aesthetics is now well established, there are relatively few studies that explore in depth how this actually works in the Homeric epics.  More studies of this kind are needed in order to work out the implications of these general ideas about aesthetics for our understanding of the specific building blocks of the poems—the words, phrases, formulas, scenes, stories, characters and so on that make up the Iliad and the Odyssey. 
Conversation provides a particularly apt and powerful basis for an aesthetic study of Homeric poetry for several reasons. First, the conversational type contains repeated elements of several different kinds: formulaic language of individual verses (speech introductions and conclusions); specific actions or events within a type scene (e.g. the fact that a speaker in an assembly rises before he begins speaking and sits down when he finishes); and the overall sequence of speeches in, and social context of, the conversation. Any or all of these elements may depart from common patterns in order to create aesthetic effects. Moreover, conversation appears in both poems and is ubiquitous throughout each poem. This means that any insights we gain about the structure and aesthetic functioning of conversation, in contrast to less widespread types, are broadly applicable to Homeric epic overall.
In connection with this exploration of the aesthetics of the variations in formulaic speech frames and in the common sequences and contexts of conversation, it may be useful to briefly survey various scholarly approaches to the notion of “traditional” language, characters, and so on in the Homeric poems. Although the opposition between traditional and original has generally been abandoned as a tool of aesthetic analysis, the concept of “traditional” features remains a useful one for understanding Homeric poetry on a number of levels as long as “traditional” is not considered to be the opposite (and artistic inferior) of “original.”  Repeated patterns of language and narrative sequence within the Iliad and the Odyssey give us one kind of baseline against which to study variations in the poems. Traditional patterns that can be reconstructed in various ways as lying behind or before the Iliad and the Odyssey give us another kind of repetition to which the poems respond. Sometimes we can tell that repeated patterns in the poems are also traditional, but these two categories are not necessarily the same thing.
Since Milman Parry proposed in his French thèses that the epithets and formulas in the Homeric poems were traditional in nature,  scholars interested in oral theory have been exploring the range of uses of “traditional” language in formulas in the Homeric poems.  During the 1960s in particular, this interest often found expression in an adversarial model of the relationship between an oral poet and the traditional language that Parry argued he had inherited from his poetic forebears.  An influential recent approach to the notion of “traditional” as it relates to language in oral poetry has been that of Foley. In a series of publications over the last fifteen years or so, he has developed the idea that each use of traditional language (such as an epithet) evokes the whole mass of traditional stories, phrases, and so forth that are connected to a particular character.  This is part of a general trend in more recent studies of Homeric poetry that views the traditional language of oral poetry and its conventions not as a regrettable limitation on a gifted artist, but as the tool and the vehicle of the success of such an artist. This conception is closely linked to the aesthetic framework that oral theorists have developed in recent years for appreciating Homeric poetry specifically as oral poetry.
A different strain of Homeric scholarship, neoanalysis, has focused on the use and transformation of traditional materials on the level of characters and story elements. Neoanalysts often focus on surprising or inconsistent aspects of the Homeric poems, and they explore other early sources for the stories that the Homeric poems tell (such as the Epic Cycle). By doing this, neoanalysis aims to discover the traditional story motifs that lie behind the Homeric poems—in particular the Iliad—and to describe how the Iliad changes or adapts these motifs.  It has been pointed out from time to time by both neoanalysts and oral theorists  that both of these approaches to Homeric epic are based on examining traditional elements of the poems and considering how the poems change or adapt their traditional materials of formula, narrative structure, and story motif. Moreover, both neoanalysts and oral theorists tend to be unitarian in their attitude toward the Homeric epics. 
Linguistic Perspectives on Conversation
The concept of the type scene gives us a powerful tool for understanding the structure and aesthetics of conversation as an important but previously under-appreciated building block of the narration in the Iliad and the Odyssey. At the same time, this approach is limited because it tells us fairly general things that apply to a wide range of activities that characters in the Homeric poems regularly do. Linguistic theory gives us the basis for describing the elements of the type and integrating them into a cohesive whole.
A number of different areas of linguistics focus on speech exchanges between people.  Conversation analysis describes conversation as basically a turn-taking activity that the participants themselves understand and control. Moreover, the particular social context in which a conversation takes place affects the specific tendencies or rules that govern turn taking in that conversation.  Discourse analysis—a broad discipline of which conversation analysis is sometimes considered a sub-field—focuses on how speech is understood in context, or language in use.  Historical pragmatics applies principles developed originally for studying current or modern spoken language to texts, which in addition to being written rather than oral may also be fictional and/or in a dead language. This approach—unlike conversation analysis, which is based almost entirely on transcripts of actual conversations—assumes that it is possible to describe and understand language use even if the language in question is no longer spoken and the evidence available is entirely literary or even fictional.  Linguistic perspectives will appear in this book primarily to provide a context for the way in which the conversational type is described, but they also appear from time to time in the analysis of particular conversations, particularly those in group contexts (Chapters 5 and 6).
Type Scenes and Homeric Conversation
Generally speaking, scholarship on type scenes has dealt with speeches as one component of various other types (such as arrival or assembly) rather than treating speech or conversation in and of itself as a type. Insofar as speech has been considered a type, the salient feature of the speech has tended to be the context of the speech (lament, monologue) or the intention of the person giving the speech (persuasion, consolation). Speech per se has not been considered a type, nor does most literature on type scenes distinguish between “speech” and “conversation.”  Although conversation has not generally been considered a type scene, in fact, it meets the definition put forward by Edwards in his recent survey of scholarly literature about type scenes. At the outset of this review, he defines a type scene as “a recurrent block of narrative with an identifiable structure.” Edwards goes on to say that “in narratological terms, an amplified type-scene is not necessary to the ‘story’.”  That is to say, Edwards’ definition fits Homeric depictions of conversation from both the descriptive and the aesthetic standpoints. He is perfectly right to question whether “speech” meets the definition of a type,  given that multiple instances of speech do not necessarily have an identifiable structure or consistent sequence. However, conversation—as distinct from speech—does consist of recurrent blocks of narrative with an identifiable structure, namely speeches joined together in a series by formulas specific to conversation. At the same time, the particular narrative blocks that recur, and the identifiable structures that order them, may vary somewhat depending on the number of speakers and the social context.
All the varieties of conversation discussed in this book share a basic similarity: their main components are an exchange of speeches and formulas linking these speeches together that appear either exclusively or predominantly between two speeches in an ongoing conversation. At the same time, a conversation in a given social context has specific characteristics that do not occur in other contexts: for example, assemblies have particular features that are not found in either informal conversations or other formal group contexts such as laments. While the content of the speeches in a conversation are generally not repeated between one conversation and another, the formulas that link the speeches together are among the most repetitive and regular parts of the Homeric epics. Conversations thus have the identifiable structure of a type scene throughout, although the lexical repetition in conversations is found much more in the formulas linking the speeches together than in the speeches themselves. In fact, as will be discussed further below, these formulaic verses are found specifically in sequences of speeches: they tend not to appear as introductions to speeches that either begin or fall outside of conversational exchanges. This makes it legitimate to see conversation as a repeated sequence or block of narrative even though not all of a conversational sequence is repeated at the lexical level from one example to the next.
Previous Scholarship on Type Scenes
Arend first pointed out that the repetition in type scenes was not to be attributed to awkwardness or incompetence, but was part of the poetic vocabulary of the Homeric poems.  His work primarily described different types and the variations they might include without the then-usual attempt to isolate the “original” example of a Homeric repetition from a herd of imitations and interpolations.  The repetition in type scenes thus became an aspect of Homeric poetry worth studying, instead of something that needed to be explained away by anointing one example as the original and relegating all the others to an inferior or derivative role. Arend, not incorporating Parry’s nearly contemporary work, did not discuss the structural or aesthetic implications of type scenes for specifically oral poetry, but Parry and especially Lord  considered type scenes to be a key component of the repeated patterns that characterize Homeric epic and other oral poetic traditions.
Lord describes an analogy between the theme (as he calls type scenes) at the level of narrative and the formula at the level of individual words.  In theory, then, type scenes and formulas should have received equal attention from scholars wishing to understand the Homeric epics as oral poems, but in practice, this has not been the case.  Fenik 1968 is deservedly the most influential of the few studies that used type scenes to explore broadly the structure and aesthetics of Homeric epic. Although, as he says, “the chief aim of this study is merely to identify and describe the recurrent, formulaic elements of the battle narrative,”  Fenik also shows very persuasively how time and again, the most powerful and significant battle scenes in the Iliad achieve their effects “not by inventing a new action or new details, but by the larger than average accumulation of familiar details.”  Scholars have benefited a great deal from Fenik’s work,  but surprisingly few people have extended his method to other types.
Fenik’s approach is essentially also the approach of this book: to describe the components of a familiar and common type in the Homeric poems in order to understand both the construction and permutations of the type better and the way in which the elements of the type may be used—rather than abandoned in favor of different or unique material—to shape the important moments in the poem. One may reasonably assume that the insights of Fenik and others about the aesthetic functions of type scenes apply broadly to the poems and not simply to the particular types or poems on which their individual studies focus, but this is an inference rather than a clear result of these studies. No study has yet analyzed a type that is sufficiently widespread in both the Iliad and the Odyssey to cogently apply its conclusions to both poems overall, or to allow comparison of the two poems. 
Conversation, which not only appears in both poems but in basically every scene in each poem, is an example of such a type. Indeed, conversation occurs in as wide a variety of episodes as the poems themselves contain: it is a common denominator not only between the two Homeric poems, but between individuals and groups, between gods and mortals, between men and women, and between friends and enemies. This means that insights about how conversation functions aesthetically apply broadly to the Homeric poems, rather than to only one of the poems or to one particular type of scene in the poems. Indeed, any general conclusions about the aesthetics of conversation should apply to the aesthetics of the poems overall, given the frequency and ubiquity of conversation within the poems.
Formulas and Conversation Type Scenes
The most regular and repetitive part of conversation type scenes is the speech frames that link the individual speeches together. The vast majority of these formulaic verses follow the general pattern [accusative pronoun] / [verb of speaking/answering] / [nominative name/epithet] and mean something like “X answered him.” These are among the most common and easily recognizable expressions in the Homeric poems. Although these are generally referred to as “speech” introductions, in fact, they are reply introductions: they are both the most common kind of speech introductory formula found in conversations and they are restricted almost entirely to replies, particularly those that occur in one-on-one conversations. The noun-epithet formulas found at the end of the verse in full-verse reply formulas provided the backbone for Milman Parry’s seminal work on the traditional, formulaic nature of Homeric language. This is because these noun-epithet formulas demonstrate to a particularly high degree the qualities of economy (one expression of a given metrical shape available for a particular idea)  and extension (expressions of various metrical shapes available for a commonly occurring idea).  Parry argued in later work that built on his thèses that these features could only be explained by an oral model of composition for the Homeric poems that presupposed a body of traditional formulaic language from which an individual poet drew in composing his poem. 
At length, Parry’s idea that this traditional language significantly diminished the role of the individual poet in creating the language of his poem  resulted in an intellectual position known as “hard Parryism,” which saw artistry and formulas as mutually exclusive.  Hard Parryism, in turn, created a backlash in the 1960s and 1970s. During these decades, many scholars were engaged in refuting hard Parryism by exploring in different ways the flexibility and artistic functioning of formulas.  In addition, the term “formula” itself underwent careful scrutiny, with the result that various ideas of the term gained currency alongside, or sometimes instead of, Parry’s widely used definition of a formula as “an expression regularly used, under the same metrical conditions, to express an essential idea. What is essential in an idea is what remains after all stylistic superfluity has been taken from it.”  Recently a persuasive argument has been made that the failure of any single definition of the term to gain the ascendancy should be seen as a productive multiplicity rather than a vitiation of the entire idea of “formula.” 
Introductory Formulas  For Direct Speech: Most other types of formulas have been rehabilitated at least somewhat from the hard Parryan idea that meter in Homeric poetry determines all. Speech introductions, however, perhaps because they formed the backbone of Parry’s original research, are still dismissed almost universally as lacking in meaning or artistic significance. A prominent scholar of oral poetry, who in general is a strong proponent of meaning and artistry within an oral framework, has recently called speech introductions “workaday phrases”  and “boiler-plate.” 
Several article-length studies have collected examples of speech-related formulas. These generally strive to catalogue rather than analyze the formulas they discuss.  Edwards 1970, the most important study of Homeric speech introductions, provides some of the same data as the current study. Edwards’ approach differs from mine in two respects. First, he is interested primarily in examining speech introductions as a Parryan formulaic system, and so he focuses throughout on economy and the metrical shapes of the components of a given speech introductory verse. At the same time, he is not interested in, and so does not discuss, how the conversational context of a particular speech relates to the speech introduction that introduces it. This means that differences in meaning between different formulas tend to be discussed only if the formulas in question also have different metrical shapes.
In addition, Edwards (at least in 1970) had an aesthetic stance toward Homeric epic that is fundamentally at odds with the idea of interplay of repetition and variation that underlies this book. For the most part, Edwards associates aesthetic force or interest with unique phraseology. He is interested in the “range of tone possible within the formulaic expressions”, which he describes as “considerable.”  His main interest, however, is in speech introductions that use unique or unusual language. The discussion of such verses takes up over half of the article. Indeed, Edwards concludes that the emotional peaks in the poem “draw their effectiveness not from colourful language but from the setting and content of the speeches, which are introduced by the most regular and familiar of verses.”  In fact, this is not the case, in that many emotional scenes derive their power partly from the use of unexpected (although not necessarily nonformulaic or untraditional) speech introductory language. Moreover, this formulation does not allow for the possibility of creating emotional effects with traditional language, which also occurs in a number of emotionally heightened scenes. This study begins from the same point that Edwards began: what are the different formulas for introducing speeches in Homeric poetry? However, in this study, that question provides the basis for answering larger questions about the construction and aesthetics of Homeric poetry rather than an end in itself.
A number of articles in the last twenty years have considered verb formulas within speech introductory formulas that are doublets in meaning, metrical shape, or both; the consensus on these has been that they are equivalent in meaning.  More recent work has pointed out fruitful avenues for exploration,  but these individual studies do not add up to a cohesive line of thought about speech introductions or any kind of systematic advancement of our understanding of how speech introductions work. Indeed, there has never been a full-length study of Homeric speech introductions, in spite of their frequency in the Homeric epics and their importance in structuring the narrative. 
Notwithstanding the pervasive scholarly neglect of speech introductions, these expressions have a number of features that richly repay systematic study. Although the noun-epithet formulas within single-verse speech introductions have been generally abandoned as carriers of meaning, at the same time, the existence of several different metrically equivalent participles with obviously significant and contextually determined meanings that characterize the emotions or behavior of the speaker is widely admitted.  It is odd that meaning is granted to one component of speech introductory formulas and so universally denied to speech introductions in general. Indeed, speech introductions and conclusions are so repetitive, and so frequent, that it is tacitly assumed that they have no aesthetic contribution to make to either the construction or the understanding of the Homeric epics.
I take the opposite position: a strongly defined and frequently occurring pattern—like formulaic speech frames—presents an ideal opportunity for a poet to create aesthetic effects by playing off this pattern. That is, where the “normal” way to say something is easily identifiable, it is equally easy for the audience to identify instances where a different kind of language has been used and to ascribe importance to such departures from the norm. When expectations are as clearly defined as they are for speech introductions and conclusions, a poet would have an unparalleled opportunity to create effects by failing to meet these expectations. This need not mean that the poet uses untraditional or even nonformulaic language, although sometimes this is what happens at such moments: language that is traditional or formulaic in one context can create a moving effect if it is used in another context. 
Within conversation, then, speech framing formulas provide the bulk of the material for aesthetic studies based on the interplay of repetition and variation. This is because speech frames are very repetitive while the contents of the speeches themselves are much less so. This is not to say, of course, that the speeches are therefore devoid of interest, either aesthetically or otherwise. The conversation type consists of both speeches and the formulas that link them together, and the relationship between the two displays certain patterns that can either be followed or not followed in order to create aesthetic effects. Moreover, in some contexts, a conversational sequence may vary from the typical by including a speech at a particular spot where usually no speech would occur, and in these instances, the speech becomes noteworthy because it is a variation from the usual pattern.
The foregoing discussions have treated formulas and type scenes, aspects of Homeric poetry that are generally agreed to originate in an oral poetic tradition. Many readers of the poems who are quite ready to recognize that techniques of oral poetry are ubiquitous in the Iliad and the Odyssey at the level of individual verses and scenes have balked at seeing the entire poems as they have come down to us as likely or even conceivable productions of an oral poet. In fact, most of the major objections to the idea that the poems as we have them reflect oral poems that existed in substantially the same form as our texts can be countered with parallels from oral traditions of other cultures or from aspects of Greek culture itself. For example, the influence of our Iliad and Odyssey is extremely murky and unclear until sometime in the fifth century B.C.E.  This has often been used to suggest that the poems in their current form did not exist until the fifth century or even later. More generally, it is unclear how or why written texts of the poems would be seen as authoritative in an essentially oral culture, and thus it is hard to figure out how the poems would have come to be written down in the first place during the archaic period.
The Domesday Book, a survey of land and land owners made in early medieval England under the impetus of William the Conqueror, provides an excellent example of a written document that eventually came to have a major impact, but which lay essentially fallow and unnoticed for two hundred years after it was written down.  Once writing became an ingrained part of the culture of the English people, the Domesday Book was regularly consulted for legal precedents, but until then, it was used only rarely. This may have been the situation with the Homeric poems too. The assumption has been that if the poems were committed to writing, they must have immediately been a) used and b) authoritative in their written form, but the Domesday Book shows us that these assumptions need not be true. Indeed, the act of writing the poems down and the authoritative status of these written texts may well be two separate phenomena. The Domesday Book also provides a useful template for how the poems might have come to be written down in the first place, in that it was made by an outsider and not by a member of the oral culture that produced the contents.
The other problems with the idea that the poems came into being in substantially their current form at an early date and in an essentially oral milieu are primarily technical or logistical and have to do with the great length of the poems. From a poetic standpoint, the length of the poems, while noteworthy, is by no means unprecedented.  From the production standpoint, where did the writing ability and technical know-how in general come from that made it possible for poems of such length to be committed to writing? We don’t know the answer to this question. This, however, should not be taken to mean that there is no possible answer: we know very little about either the state of the technology of writing in Greece in the archaic period or the extent to which members of other cultures with possibly more developed writing capabilities might have been present in Greece. We do not know that committing the Iliad and Odyssey to writing would have been definitely impossible in the archaic period.
Aristophanes’ allusions to Euripides provide a useful perspective on the problems of arguing about how poems were disseminated based on the technical aspects of writing and/or book production.  It is patently obvious from Aristophanes’ plays that he and probably his audience were very familiar with the plays of Euripides, not only those produced at the same time as Aristophanic comedies but those produced earlier. We know a lot more about Aristophanes, Euripides, the dramatic uses that Aristophanes made of Euripides, and the fifth century in general than we do about Homeric poetry. Nevertheless, our ideas about how either Aristophanes or his audience came by their specific and detailed knowledge of Euripides remain almost entirely speculative. Even so, no one would assert that, because we do not know how (or if) texts of Euripides were circulating at this time, the allusions did not occur. That being the case, it seems to me impossible to use writing technology as cogent proof either for or against any particular view of the transmission of the Homeric poems. Concerns about the length of the Homeric epics being incompatible with the technology of writing or the limits of oral composition cannot be either supported or disproved by our knowledge about archaic Greece or about oral poetry.
More positively, Avdo and other living oral poets who have been studied by experts share a set of characteristics that seem to fit the Iliad and the Odyssey as we have them. Avdo, quoted in the epigraph of this book, told the same stories as other poets, but he took much more time to do so and dwelled at length on the emotions and thoughts of the characters in the story. This produced a song recognized by both field workers and the members of Avdo’s community for its outstanding quality. Field work in Africa produced a particularly long, detailed, and engrossing song from a well known singer who was stimulated by an engaged audience and a field worker recording his song.  A number of outstanding singers in modern Scotland responded to the presence of a worker recording their songs by lengthening and elaborating on the usual patterns for that particular tale without decreasing the quality of the song.  Niles, indeed, highlights the role of the individual in developing and maintaining traditional poetry based on his experiences in documenting the traditional songs and singers of Scotland: “it is through active, self-conscious, intelligent tradition-bearers like Međedović and Williamson that an oral culture realizes its full potential.”  This book does not aim to prove one way or the other that the Iliad and Odyssey were orally composed in archaic Greece in essentially their current form. At the same time, the overall picture of the construction and aesthetics that emerges from these pages is consistent with such an origin for the poems, and as the previous discussion suggests, the usual objections to the Homeric poems as oral compositions of the archaic period should be strongly questioned.
Linguistic Perspectives on Conversation
The Homeric Perspective
The basic vocabulary used to refer to conversation tells us something important about what kind of activity the Homeric poems imagine conversation to be. The most common verb root used for “answer, reply” by both the main narrator and the characters who report the conversations of other characters is ἀμειβ-, which means “exchange.”  Nearly one-half of all full-verse reply formulas used by the main narrator that mean something like “he answered” contain a form of ἀμειβομαι.  A number of other reply introductions that are not full-verse formulas also use this root. Similarly, the most common formulas that Odysseus and Menelaus use when they report conversations in the course of their storytelling all contain a form of ἀμείβομαι.  Occasionally, we see individual characters in the Homeric epics explicitly referring to this notion of conversation. For example, the disguised Odysseus—as one might expect from such a skilled and self-aware conversationalist—excuses himself when he speaks to Telemachus without having been spoken to first by saying “ὦ φίλ’, ἐπεί θήν μοι καὶ ἀμείψασθαι θέμις ἐστίν”  (“Dear friend, since in truth it is right for me to take an answering turn as well”, my translation of Odyssey 16.91). In fact, these are the first words in the poem that Odysseus addresses to his son. Similarly, Nestor refers to two speakers as ἀμειβομένω (Odyssey 3.148). So, ἀμείβομαι and its compound are used for the idea “answer, reply” at a number of different levels of Homeric discourse: the main narrator and the characters both use it when quoting the speech of other characters, and characters occasionally use it when talking about conversation.
It seems, therefore, that the Homeric poems consistently view conversation as one manifestation of a more general notion of societal exchange of commodities of value. This conception of exchange also includes the reciprocal giving of gifts.  The middle voice, which is used for “reply”, shows the strongly reciprocal nature of conversational exchange: conversation involves essentially a simultaneous, interdependent exchange of turns at speaking.  Gift-giving, in contrast, which uses the active voice of ἀμείβω, entails a more discrete and separable pair of actions between the parties involved. One cannot have a conversation without exchange, whereas one can give a gift without gift exchange.  The distribution of ἀμειβ- words for “answer” reinforces the idea that exchange is associated specifically with conversation, rather than with speech. ἀμειβ- is the only root commonly found in formulas that introduce replies that is restricted to replies: unlike other common verbs of answering,  there is only one example in over 200 instances of ἀμειβ- in speech introductions (in both simple and compound forms) where the verb appears to introduce a speech that is not a reply.  We may understand the verb as meaning “take one’s turn [by speaking]” rather than “reply in turn”, as it is often translated: a reply is a turn, one piece of a give-and-take structure between people that is based on speaking rather than on physical, material goods.
Modern Linguistic Perspectives
The linguistic discipline of conversation analysis takes essentially the same view of conversation (sometimes called “talk in interaction” by linguists working on conversation analysis). The most influential statement of the principles of conversation analysis  starts out by locating conversation within a broader spectrum of “speech-exchange systems;” it also notes that a system of exchange regulated by turn-taking “suggests an economy, with turns for something being valued.”  Thus, conversation analysis shares with the etymology of Homeric vocabulary a basic conception of conversation as a medium in which a commodity of value is exchanged in a social context. Conversation analysis tells us that conversation is governed by rules for taking turns among the participants. These rules, which are followed by speakers in the Homeric poems, include most basically the notions that one speaker speaks at a time, and that change of speaker is a recurring phenomenon.  Conversely, although only one person should speak at once (and not more than one), if no one speaks for any extended period of time, this may be perceived by the people in the conversation as some kind of mistake or problem in the conversation that needs to be rectified or otherwise addressed.  Moreover, the particular social context in which a conversation takes place affects the specific tendencies or rules that govern turn taking in that conversation.  Methodologically, conversation analysis stipulates that if a given rule or principle is to be meaningful as a tool of analysis, one must be able to demonstrate that speakers are actively oriented toward the rule in their conversational behavior.  As we will see, Homeric characters do behave in regular ways when they talk that are dependent on—and may be said to define—the specific context in which the conversation occurs, and they sometimes violate or adapt these rules. Moreover, they speak one at a time, and if one speaker interrupts another or no one responds after someone has finished speaking, that is cause for surprise or even chagrin.
This similarity of outlook between conversation analysis and the Homeric epics suggests that conversation analysis may have ideas about how to conceptualize and study conversation that can be usefully applied to Homeric poetry, at least at the level of the broad, context-independent insights it has reached about how conversation works. At the same time, the research methods of conversation analysis depend almost entirely on analyzing transcripts of tapes of actual conversations. Clearly, this data is very different from what we find in the Homeric epics, and the methods and conclusions that depend closely on specific conversations, rather than on the aggregate observations about many different specific conversations,  will not be applicable to literary texts.
Discourse analysis is a wide-ranging umbrella term for a number of different approaches to studying language use in context. In contrast to conversation analysis, which is a very specific methodological approach to studying conversation, a recent overview of discourse analysis explicitly disavows the idea of discourse analysis as a research method.  So, the study of language use in context may include conversation analysis, but it also includes other approaches that are less closely tied to live utterances as data. Some scholars working in discourse analysis focus on how language use (or, more generally, social relations) may be produced or represented in one context for an audience who is somehow separated or distant from the original production. This approach offers us two useful insights: first, one may legitimately apply linguistic principles to an utterance or representation that is produced in one context to be received or understood in another (in contrast to a conversation recorded by a researcher in conversation analysis, whose participants are all present in the same place and time);  and second, it is reasonable to assume that, in such a situation, “whether or not we identify with the way we are addressed, we do understand how we are addressed, because we do understand the way images represent social interactions and social relations.”  That is, a representation of a conversation draws on the same kinds of ideas about how conversations work as a live conversation does. This is true for both the person creating the representation and the audience to whom the representation is addressed. Although the specific research methods that conversation analysis uses are not appropriate for Homeric poetry, work in discourse analysis suggests that we can nevertheless apply the basic ideas that conversation analysis has developed to representations of conversations as well as to actual recorded conversations.
When these representations are found in literary texts recorded in a dead language, our interpretive problems multiply, because we lack a clearly identifiable social and/or cultural context within which to place either the Homeric poems’ representation of conversation or the way in which an audience might have responded to these representations. No scholarly discipline has yet come up with a scheme that will resolve the myriad problems that arise because we lack definitive information about the context of performance, the manner of composition, the time of composition, the manner of transmission, the nature of the audience(s), and the time and manner in which the Homeric poems were committed to writing. However, historical pragmatics does give us a basis on which to apply linguistic ideas not simply to a representation of a conversation rather than an actual conversation, but to a representation contained in a text in a dead language, a literary text, or both.
Historical pragmatics combines the disciplines of historical linguistics, which studies language change over time, and pragmatics, which studies certain features of language use that affect how language is perceived by the people who hear it.  Some scholars working in this field emphasize the historical over the pragmatic, and aim to understand language evolution by tracing specific pragmatic features like politeness markers in a given language over time. Others emphasize the pragmatic more than the historical, in order to describe and understand “conventions of language use in communities that once existed and that are no longer accessible for direct observation.”  In other words, historical pragmatics gives us a basis for studying language products of vanished cultures as though they are amenable to the same kinds of analysis and explanation as we use to explain language use in cultures that still exist. 
In sum, then, this study draws on conversation analysis for its basic model of conversation as a turn-taking activity whose structure both responds to and depends on the social context in which the conversation takes place. Discourse analysis shows us that we can apply the basic insights of conversation analysis about how conversation works to representations of conversation as well as to oral conversations. Historical pragmatics offers a parallel methodological transfer of a linguistics discipline originally developed for use in analyzing oral speech to texts from the past. Indeed, a recent overview of historical analysis includes conversation analysis as one area of linguistics that may be used to analyze texts from the past. 
Perhaps the most significant difference between a represented conversation and an oral conversation is the presence of the representer as a mediator between the represented conversation and the audience to whom the conversation is represented. A Homeric conversation essentially has three participants where an oral conversation would have two: besides the two speaking characters, the narrator also plays a role in the conversation. Indeed, it is the narrator rather than the character who performs many of the conversational activities that conversation analysis focuses on. Accordingly, we must look to the information contained in speech frames at least as much as to the speeches themselves for information about how (for instance) repair is effected in a Homeric conversation when no speaker takes the next turn after one speaker has finished his turn. When silence falls in the middle of a conversation, the characters themselves generally do not comment explicitly on the fact that no one has spoken. They may indirectly allude to it in their response, but it is the narrator who reports this silence and is the source of the audience’s knowledge of and attitude toward it. Similarly, we usually do not see characters speaking about the social factors that affect the way they behave when they talk and the order in which they talk. Nevertheless, we can see from the comments contained in speech frames that these things do matter, and how they matter, in Homeric conversation.
From a linguistic perspective, then, the speech frames provide us with some of the information we need in order to understand how Homeric conversation works. Speech frames convey many of the non-verbal aspects of conversational interaction that conversation analysis studies. From a narratological standpoint, the directly quoted speeches in a conversation exist in both the story and the text.  The speech frames, on the other hand, represent verbally for the external audience of the text things that are not verbal for the characters in the story. And, from an aesthetic point of view, their extremely regular and formulaic language offers an effective basis for creating artistic effects through variation on repeated language. It is for all of these reasons that speech frames take precedence in this analysis over the speeches themselves. The next section gives an in-depth explanation of the methodology that was used to describe and classify the data on conversational sequences and formulaic speech frames that underlies this book.
Repeated Speech Sequences and Formulas in Conversation
In the remainder of the introduction, the conversational terminology and data that underlie this study will be discussed. First, I will describe the different ways that individual speeches can relate to other speeches within a conversational sequence. The bulk of this section lists the formulas for introducing and concluding speeches, organized with reference to two features: where in a conversation the speech being introduced falls; and whether it is addressed to one person or many people.
How to Have a Homeric Conversation
A key component of the research that underlies this study is a catalog of each speech in the Iliad and Odyssey with respect both to its position relative to adjacent speeches (if any) and the type of formula (if any) that introduces the speech.  Any given speech may take a conversational “turn” in one of three basic ways: it may not participate in a conversation at all; it may be the first turn in a conversation; or it may respond in various ways to a previous turn in an ongoing conversation. Depending on several factors like the number of speakers involved, the social context for the speech, and others to be discussed further below, many different sequences exist for joining together a series of turns to form a conversation. Let us now turn to an elaboration of these basic categories and the various combinations in which they occur to form conversations.
Methodology: Before giving the results of this data collection, I will describe the criteria that I used to classify these speeches. My primary goals for this data collection were, first, to explore and describe the range of organizational turn-taking structures that can govern Homeric conversation, and second, to evaluate the preponderance and types of formulas for organizing different varieties of conversational turn-taking. For both of these goals, it is preferable to overestimate rather than underestimate the proportion of Homeric speech that occurs in conversation.
From the standpoint of describing turn-taking, overestimating the preponderance of conversation is preferable because describing a limited and very regular group of conversational types is easier than describing a more wide-ranging and varied group of conversations. Hence, a model that can successfully describe a broad range of conversations is more powerful and more useful than one that applies to a more limited set of data. Similarly, from the standpoint of the preponderance of formulas that introduce and conclude speech, a more wide-ranging group of conversational types yields a more heterogeneous group of speeches; these in turn are likely to be framed by a more rather than less heterogeneous group of framing expressions. So, overestimating the proportion of speeches in Homer that appear in conversation will result in a conservative conclusion rather than an excessive one of how common and how regular formulas are for introducing and concluding speeches in conversation.
Types of Turns in Homeric Conversation: The most basic distinction in this data set is between speeches that appear alone (“single” speeches)  and those that appear as part of a conversation (those that are either preceded or followed by another speech). A speech like Iliad 5.714-718, in which Hera speaks briefly to Athena as a result of her distress at events on the battlefield before Troy, demonstrates many common features of a single speech. First of all, it is separate from the speeches on either side of it. The last speech before this one is clearly unconnected to it for several reasons: first, it appears thirty verses earlier (5.684-688) and events involving many different characters as well as a lapse of time intervene between the two; no speaker is common between the two speeches (at 684-688, Sarpedon addresses Hector); and the location is different (the battlefield in the former, as distinct from Olympus in the latter). The speech that follows most closely after Hera’s address of Athena occurs some forty verses later (757-763) after Hera and Athena have equipped their chariot and driven it some distance away from where they started.
The speech itself also displays several features common to single speeches. First of all, Hera is giving Athena a command (couched as a hortatory subjunctive: ἀλλ’ ἄγε δὴ καὶ νῶι μεδώμεθα θούριδος ἀλκῆς [Come then, let us rather think of our own stark courage], 5.718). Commands, instructions to messengers, prayers, vaunts, and other kinds of boasts are among the most common kinds of non-conversational speeches. All of these forms of speech are inherently hierarchical in some way, and involve an imbalance of power between the speaker and the addressee. While these types of speeches by definition do not lead to replies, they may lead to non-verbal responses of various kinds. At the end of Hera’s speech, the narrator says, ὣς ἔφατ’, οὐδ’ ἀπίθησε θεὰ γλαυκῶπις Ἀθήνη (So she spoke, nor did the goddess grey-eyed Athene disobey her, 719). This full-verse formula frequently occurs in the Homeric poems and it generally indicates that a particular speech will not receive a reply, whether the speech itself is a single speech or the last one in a conversation. 
Clearly, then, Hera’s speech to Athena has no conversational links to the speeches immediately preceding or following it. Some or all of the following criteria apply to the speeches that I have classified as single speeches: in relation to preceding and following speeches, single speeches display a change of subject or theme; a change of both participants (speaker and addressee); a change of physical location; a significant lapse of story time; and/or single speeches are separated by passages of narrative from adjacent direct speeches. Not all of these criteria may apply in any given case. For instance, in Iliad 2, Agamemnon prays to Zeus (412-418); in the next speech, Nestor addresses Agamemnon (434-440). Both of these speeches are single speeches, although Agamemnon participates in both, insofar as 1) the first speech is a prayer and the second a command (to order the heralds to assemble the Greeks), both of which are types of speech that usually lack a reply and 2) a meal and a lapse of time occur between the two speeches. The two speeches are not about the same topic, nor does Nestor’s reply respond to anything that Agamemnon said in his prayer.
Occasionally, a particular speech occurs while a conversation is in progress, but does not involve the main participants in the conversation. Such speeches are classified as “simultaneous single” speeches. These often involve anonymous speakers, as speeches introduced with the formula ὧδε δέ τις εἴπεσκεν ἰδὼν ἐς πλησίον ἄλλον (and thus a man would speak, looking at another fellow next to him [my translation], Iliad 2.271 and eight other places in the Homeric poems). One function of such anonymous speeches, indeed, is to dramatize the feelings or views of a multitude by using one speech as a representative sample of what the various members of the group were saying.  Such speeches are separate from conversational turn structures, although they accompany such a structure. Hence they are considered single speeches even though they accompany a conversation.
At the other end of the conversational spectrum we find an episode like the conversation of Penelope and the disguised Odysseus in Odyssey 19, which extends for over 250 verses (103-360) and includes 11 speeches in an unbroken alternating turn sequence between the two characters. This turn-taking sequence pauses briefly on two occasions to give details of Penelope’s emotional response to things that Odysseus has been saying (after Odysseus’ second and third speeches Penelope bursts into tears), but with the exception of these details, no elaboration, event, or passage of time occurs between the speeches. Moreover, a regular and consistent turn-taking structure governs the order of speeches, with Penelope and Odysseus speaking alternately. Just as these speeches follow one another, the content of each speech responds to or develops the content of the previous speech. One of the concerns of this book is the analysis of narrative digressions like “Penelope wept” that occur between one speech and another within an extended turn-taking sequence that is strongly oriented toward an alternative turn-taking organizational model. These digressions represent a variation of the normal pattern in the one-on-one conversational type of minimal narrative presence between speeches, and as such, merit our attention.
Conversational turn-taking structures vary in their length, the way they are organized, how closely one speech follows another, and the extent to which events as well as speeches form part of the turn-taking structure. To accommodate this variety, I have adopted a general, negative definition of the connections between speeches that constitute a “conversation”: any speech that is not a single speech is considered to be part of a conversation. In practical terms, although the degree of linkage between speeches in different conversations varies significantly, single speeches tend to stand out clearly from their contexts because they display features of single speech described above. Thus, when one is in doubt about the nature of the linkage between a speech and its neighbor, the question is usually how strong the link is rather than whether it exists at all. In the few cases when I was unsure whether a particular speech was a single speech or was loosely connected to an ongoing conversation, I took unity of theme as the deciding consideration: if the speech was responding to an ongoing theme of the conversation and was addressed to or spoken by one of the participants in the conversation, I included it in the conversation. If it was not about the same topic as the rest of the conversation, I classified it as a single speech, even if it was spoken by one of the participants in the conversation or was spoken at the same time as the conversation.
Within the general category of “speeches that appear in conversation,” I have distinguished three different kinds of conversational speeches: initial speeches (those that appear first in a conversation), replies (those preceded by another speech spoken by a different person), and successive speeches (like replies, preceded by another speech, but a speech spoken by the same person as the immediately preceding speech). Of these types, the reply is the most common in Homeric epic: 299 of 678 speeches in the Iliad are replies (44% of speeches) and 340 of 545 speeches in the Odyssey are replies (62% of speeches). Initial speeches can only be distinguished retroactively from single speeches. When a speech begins that is not preceded by another speech, the audience does not know until the speech is over and a reply either does or doesn’t follow whether a conversation is in progress. Thus, initial speeches resemble single speeches in that they themselves are not responding to any other speech, but they also resemble replies insofar as they do participate in turn-taking conversational structures. Both the Iliad and the Odyssey contain few successive speeches, which violate normal conversational turn-taking structures at a very basic level because one person takes two speaking turns in a row: 37 of 678 speeches (5%) in the Iliad are successive speeches and 17 of 545 speeches (3%) in the Odyssey. Most successive speeches occur in group contexts, where the turn-taking structures are more fluid than they are for conversations between two people. Alternatively, an entire sequence may consist of one character giving two successive speeches without receiving a response from any of the people to whom he is speaking. Such successive speeches are basically two single speeches back to back rather than a conversation proper: with only one speaker, there is no real sense of turn-taking.
Formulaic Speech Frames for Different Kinds of Conversational Turns
Having examined the various sequences of turn-taking that organize the speeches in Homeric conversations, we now turn to the formulaic speech frames that both establish and display these forms of organization. This section gives an overview of the different types of full-verse speech introductions and conclusions, examining both their structure and content and their relationship to the conversational turn sequences in which they occur. The overall goal here is to introduce the reader to the range of full-verse formulaic speech frames that appear at various points in conversational types, with reference to the kinds of information they tend to include, the language they use to convey this information, and the types of contexts in which they generally occur. This will provide the reader with an understanding of the general characteristics of commonly found speech introductions and conclusions that will be necessary in order to appreciate the variations that these formulas may display. For the most part, more detailed discussions of how these verses are distributed throughout the Homeric poems and how they function in particular contexts will occur in the chapters of the book. I will begin with a brief discussion of the term “formula.” The bulk of this section gives examples of the most common formulaic speech frames. It starts with those that precede a speech replying to another speech, goes on to the formulas that precede speeches of other sorts, and ends with an overview of speech concluding formulas.
Definition of “Formula”: For the purposes of this study, I will be using a fairly narrow definition of “formula”: an expression of at least two words that is repeated at least three times with regard to the words that are used,  the order and syntactical relationship in which the individual words stand to one another, and the spot in the verse in which the expression is located. Not all of the words in a formula need be adjacent to one another: many speech framing formulas consist of several words that always occur in the same parts of the verse, with a gap or gaps left in between for a proper name or title. For instance, within verses that have the form καὶ τότ(ε) [accusative, addressee] προσέφη [nominative, subject], the fixed elements of the formula alternate with flexible slots for the name of the addressee and of the speaker. That is to say, καὶ τότ’ ἄρ’ Ἰδαῖον προσέφη κρείων Ἀγαμέμνων (then powerful Agamemnon spoke to Idaios,  Iliad 7.405) and καὶ τότε κήρυκα προσέφη μένος Ἀλκινόοιο (then the hallowed prince Alkinoös spoke to his herald, Odyssey 13.49) are considered to be examples of the same formula. τότε may be elided or followed by an additional particle. 
I have excluded repetitions that are only attested twice,  with one exception, because there are a number of repeated expressions that appear only twice in contexts that are thematically linked. It seems to me an open question whether such repetitions should be considered “formulas” or not, and so I am excluding them from my definition. As many scholars have observed (including Parry), not all repetitions are formulas.  I have considered one expression that only appears twice, λιποῦσ’ ἀνδροτῆτα καὶ ἥβην (leaving youth and manhood behind), to be a formula because linguistic evidence shows that it is an extremely old expression. Therefore, it may be considered traditional as well as repeated, and for this reason is unlikely to be a new invention. Although my definition of “formula” specifies a minimum length of two words, most of the formulaic expressions I will discuss are a full verse in length. This is because the vast majority of formulaic reply introductions are a full verse long. Therefore, six feet of dactylic hexameter was adopted as a uniform standard of length for all the formulas to be compared to one another in order to avoid the possibility that a difference in metrical shape affects the distribution or usage of two different expressions. Partial-verse formulas do appear in this study, but primarily as components of full-verse formulas.
My definition of the term “formula” should not be taken to mean that I believe that no language which does not meet it is therefore “nonformulaic”—quite the contrary. I have adopted this definition because it is one of my goals in the following pages to demonstrate that even highly formulaic language is compatible with aesthetic significance and meaning. The assumption that originality and artistry must go hand in hand in Homeric poetry, although less true than formerly, is not as dead as one might think. A recent statement on the subject of Homeric style and oral poetics asserts: “it must be remembered that Homer always has (and often avails himself of) the option of not taking advantage of the opportunities offered by the formulaic diction but of substituting something fitting the context and quite untraditional.”  This statement implies that “untraditional” language is required in order to fit the context, when in fact this is not the case, as Fenik and others have demonstrated. This is not to say that nonformulaic language has no role in creating aesthetic effects, or that this study will not have anything to say about aesthetic effects of nonformulaic language. Rather, the focus of most research on the artistic possibilities of nonformulaic language has obscured the complementary importance of formulas as aesthetic devices.
Alongside my definition of “formula,” I will be using the term “traditional.” This expression will refer to language, characters, themes, and so forth which can reasonably be supposed to have had an independent existence before their use in the Iliad or the Odyssey.  Now, since we are distressingly lacking in any kind of context for the production of the Homeric epics, identifying traditional features is to some extent a matter of guesswork. Nevertheless, it seems safe to assume (for instance) that Achilles is a traditional character, and that the noun-epithet formula πόδας ὠκὺς Ἀχιλλεύς (swift footed Achilles) is also traditional. Such identifications will be made where they seem indicated by linguistic features or by frequency. Although “traditional” and “formulaic” are not equivalent expressions, the definition of “formula” adopted for this study is intended to maximize the likelihood that formulaic expressions are also traditional. Consequently, if it can be demonstrated that formulaic speech frames have aesthetic functions and that such formulas are likely to be traditional, this will to some extent uncouple the ideas of “invention” and “aesthetics” in the context of Homeric poetry. It will mean that traditional language—and thus the traditional art form of oral epic poetry—is compatible with the aesthetic system I will propose.
Speech Introductory Formulas: The proportion of a particular type of speech (reply, initial, single, or successive) that is introduced by a full-verse formula roughly corresponds to the level of involvement that type of speech has in a turn-taking conversational structure. Successive speeches, which most markedly violate basic principles of turn-taking, are not only uncommon in Homeric poetry but are also the least likely to be introduced with a full-verse formula. There are 37 successive speeches in the Iliad, of which only seven (19%) are preceded by a full-verse formulaic speech introduction. In the Odyssey, there are 17 successive speeches and five of these are introduced by a full-verse formulaic speech introduction (29%). At the other end of the spectrum, 232 of 299 replies in the Iliad are introduced by a full-verse formula (78%); in the Odyssey, 288 of 340 replies have a full-verse formula introducing them (85%). Initial speeches and single speeches are similar to each other in the percentage of full-verse formulas that introduce such speeches: in the Iliad, 67 of 155 initial speeches are preceded by a full-verse formula (43%) and 82 of 187 single speeches have a full-verse formulaic introduction (44%); in the Odyssey, we find 61 of 115 initial speeches introduced by a full-verse formula (53%) and 35 of 73 single speeches so introduced (48%). 
What do these numbers tell us? Replies exemplify conversational turn-taking and they are almost always preceded by full-verse formulas. Successive speeches, in contrast, violate turn-taking organizational principles of conversation and are seldom introduced by full-verse formulas. Both initial speeches and single speeches are neither participating in an established turn-taking conversational structure nor violating such a structure; these two types of speeches are preceded by a full-verse speech introductory formula more often than successive speeches and less often than replies. Full-verse formulas are very likely to introduce a speech that is highly involved in a regular turn-taking conversational structure, and correspondingly less likely to introduce a speech that is uninvolved in or violating such a structure. Put another way, the most typical and regular way to talk in Homeric poetry is to have a conversation.
Introductory Formulas—Replies:  Reply formulas, which are from the standpoint of sheer numbers by far the most common type of formulaic speech introduction, state in one self-contained verse that the next turn in a conversational sequence occurred. There are not very many different reply formulas. They all share the essential meaning and structure “X [name, often with some kind of descriptive epithet] answered Y [pronoun]”, and most occur over 50 times in the Homeric epics. The following formulas are listed according to the main verb of speaking, from the most common verb to the least common.
I. reply formulas with προσέφη
[accusative pronoun, addressee] δ’ ἀπαμειβόμενος προσέφη [nominative name/epithet, speaker]
“Then in answer to [him/her] spoke [speaker],” 106 times in the Homeric poems (36x Iliad, 70x Odyssey) 
Example: τὸν δ’ ἀπαμειβόμενος προσέφη πόδας ὠκὺς Ἀχιλλεύς, Iliad 1.84
Then in answer to him spoke Achilleus of the swift feet 
Various other participles may occur instead of ἀπαμειβόμενος, of which the most frequently occurring is ὑπόδρα ἰδών, “looking darkly, glaring” (20 times). 
Example: τὸν δ’ ἄρ’ ὑπόδρα ἰδὼν προσέφη πόδας ὠκὺς Ἀχιλλεύς, Iliad 1.148
Then looking darkly at him Achilleus of the swift feet spoke
Sometimes the name of the speaker is split up within a reply formula whose main verb of speaking is προσέφη, as in
[accusative pronoun, addressee] δ’ αὖτ’ [nominative, speaker] προσέφη [additional nominative(s) referring to the speaker, usually a patronymic]
“Then [speaker, son of so-and-so] addressed [him/her],” once in the Iliad and five times in the Odyssey
Example: τὴν δ’ αὖτ’ Ἀντίνοος προσέφη, Εὐπείθεος υἱός, Odyssey 1.383
Then Antinoös the son of Eupeithes addressed her
II. reply formulas with προσέειπε(ν)
[accusative pronoun, addressee] δ’ αὖτε προσέειπε(ν) [nominative name/epithet, subject]
[speaker] spoke to [addressee], 95 times in the Homeric epics (42x Iliad, 53x Odyssey)
Example: τὸν δ’ αὖτε προσέειπε θεὰ γλαυκῶπις Ἀθήνη, Iliad 1.206
Then in answer the goddess grey-eyed Athene spoke to him
[nominative, speaker] δέ μιν οἶος ἀμειβόμενος προσέειπε
[speaker] alone, taking an answering turn, spoke to [him/her], three times in Odyssey
Example: Ἀντίνοος δέ μιν οἶος ἀμειβόμενος προσέειπε, Odyssey 2.84
Antinoös alone, taking an answering turn, spoke to him
III. reply formulas with ἠμείβετ’
[accusative pronoun, addressee] δ’ ἠμείβετ’ ἔπειτα [nominative name/epithet, speaker]
Then [speaker] took an answering turn respecting [addressee], 68 times in the Homeric epics (47x Iliad, 21x Odyssey)
Example: τὸν δ’ ἠμείβετ’ ἔπειτα ἄναξ ἀνδρῶν Ἀγαμέμνων, Iliad 1.172
Then the lord of men Agamemnon took an answering turn respecting him
IV. reply formulas with ἀντίον ηὔδα
[accusative pronoun, addressee] δ’ αὖ [nominative name/epithet, speaker] ἀντίον ηὔδα
Then [speaker] said to [addressee] in answer, 68 times in the Homeric epics (14x Iliad, 54x Odyssey  )
Example: τὴν δ’ αὖ Τηλέμαχος πεπνυμένος ἀντίον ηὔδα, Odyssey 1.213
Then the thoughtful Telemachos said to her in answer
V. reply formulas with ἀπαμείβετο φώνησέν
[accusative pronoun, addressee] δ’ αὖ(τ’) [nominative proper name, speaker] ἀπαμείβετο φώνησέν τε
Then [speaker] took an answering turn and addressed [addressee], once in the Iliad and ten times in the Odyssey 
Example: τὸν δ’ αὖτ’ Αἰνείας ἀπαμείβετο φώνησέν τε, Iliad 20.199
Then in turn Aineias took an answering turn and addressed him
VI. reply formulas with προσεφώνεε
[accusative pronoun, addressee] δ’ ἀπαμειβόμενος προσεφώνεε [nominative name/epithet, subject]
Then in answer to [him/her] spoke [speaker], twice in the Iliad and twice in the Odyssey
Example: τὸν δ’ ἀπαμειβόμενος προσεφώνεε Τεῦκρος ἀμύμων, Iliad 8.292
Then in answer to him spoke Teucer the blameless
VII. reply formulas with πρὸς μῦθον ἔειπεν
[accusative pronoun, addressee] δ’ αὖτε [nominative, subject] πρὸς μῦθον ἔειπεν,
Then [speaker] spoke a word to [addressee], once in the Iliad and twice in the Odyssey
Example: τὴν δ’ αὖτ’ Εὐρυνόμη ταμίη πρὸς μῦθον ἔειπεν, Odyssey 17.495
Then the housekeeper, Eurynome, spoke a word to her
Introductory Formulas—Group Replies: The same basic elements found in all reply formulas, “[name] addressed [object pronoun]”, also characterizes formulas that introduce speeches made to or in a group (which I have called “group reply” formulas) as well as those that introduce a speech explicitly identified as the first in a series (“initial” formulas). Group reply formulas have basically the same form as simple reply formulas, except that the object pronoun is dative plural instead of accusative singular and the main verb of speaking has a prefix μετ- “[spoke] among” instead of προσ- “[spoke] to, addressed.” Group reply formulas occur much less frequently than simple reply formulas, and there are fewer of them.
I. group reply formulas with μετέειπε
τοῖσι δὲ καὶ μετέειπε [nominative name/epithet, speaker]
And [speaker] spoke to/among them, 19 times in the Homeric poems (8x Iliad, 11x Odyssey)
Example: τοῖσι δὲ καὶ μετέειπε Γερήνιος ἱππότα Νέστωρ, Iliad 2.336
And among them spoke the Gerenian horseman, Nestor
τοῖς αὖτις μετέειπε [nominative name/epithet, speaker]
Now [speaker] spoke to/among them, twice in the Iliad and twice in the Odyssey
Example: τοῖς αὖτις μετέειφ’ ἱερὴ ἲς Τηλεμάχοιο, Odyssey 18.60
The hallowed prince Telemachus now spoke among them
ὀψὲ δὲ δὴ μετέειπε [nominative name/epithet, speaker]
At long last, [speaker] spoke to/among them, 7 times in the Homeric poems (5x Iliad, 2x Odyssey)
Example: ὀψὲ δὲ δὴ μετέειπε βοὴν ἀγαθὸς Διομήδης, Iliad 7.399
At long last, Diomedes of the great war cry spoke to them
This formula demonstrates indirectly that the normal expectation is that speakers talk one right after another, without gaps. This, in fact, reflects one of the principles of conversation analysis, which has documented that if silence rather than talk occurs at a point when talk is expected, silence will be perceived by the participants in the conversation as a lapse.  Thus, this formula says not simply that a silence intervenes between one speaker and the next, but—more importantly—it implies that this is neither the usual nor the appropriate state of affairs. ὀψὲ δὲ δὴ μετέειπε [name/epithet], by itself, does not have all three of the basic features usually found in a reply formula, namely a pronoun indicating the addressee, a verb of speaking, and a nominative form of the speaker’s name. So, it cannot stand alone to introduce a group reply. However, it is always found after the formulaic speech conclusion ὣς ἔφαθ’, οἳ δ’ ἄρα πάντες ἀκὴν ἐγένοντο σιωπῇ (So he spoke, and all of them stayed quiet in silence, 15 times in the Homeric poems including Iliad 7.398). Hence, these two verses together convey all the information that customarily occurs in formulaic group reply formulas. However, the group being addressed, instead of being represented by a dative pronoun within a single-verse formula, appears in a separate verse as a silent audience for the belated words of the speaker.
The close relationship between these two full-verse formulas points to a feature specific to group speech contexts: group reply formulas occur with accompanying verses of various kinds much more regularly than simple reply formulas do. Often two (or more) verses make the transition between one speech and the next in a group context. For example, person A speaks; then the reaction of the group is described; then a formula introduces the response of B; then B speaks in response to A. In a speech between two individuals, in contrast, the response of the audience is identical to the response of the next speaker, and so narrating the audience’s response separately is not necessary. Indeed, in some kinds of group conversational contexts, describing the audience’s behavior after one speech is as necessary to the structure of the conversation as the formulaic introduction to the next speech. 
II. group reply formulas with μετέφη
τοῖσι δὲ [participle] μετέφη [nominative name/epithet, speaker]
[Having done X], [speaker] spoke to/among them, six times in the Homeric poems (4x Iliad, 2x Odyssey)
Example: τοῖσι δ’ ἀνιστάμενος μετέφη πόδας ὠκὺς Ἀχιλλεύς, Iliad 1.58
Achilles of the swift feet stood up among them and spoke
This verse is analogous to the individual reply formulas grouped together in I (b) pp. 33–34, although there is no group reply formula that uses a participle that simply means “answered.” The participle in such a verse always conveys some additional sense besides “spoke among them.” 
τοῖσι δὲ [name in nominative, speaker] μετέφη [nominative patronymic], “[speaker, son of so-and-so] spoke to/among them”, six times in the Odyssey
Example: τοῖσιν δ’ Ἀντίνοος μετέφη, Εὐπείθεος υἱός, Odyssey 4.660
Antinous the son of Eupeithes spoke to them
Introductory Formulas—Initial Speech to a Group: A small group of formulas exist that show this same basic pattern of [pronoun referring to addressee] [verb of speaking] [nominative name/epithet, speaker] but with some kind of language that specifies that the speech is the first in a series. I have termed these “initial” formulas. As with reply formulas, different initial formulas exist depending on whether the audience of an initial speech is a group or an individual. Initial formulas, however, differ from reply formulas in that there are more different full-verse formulas for initial speeches in a group than for initial speeches to just one person. Group initial formulas include the following:
I. group initial formulas with μύθων ἦρχε
τοῖσι δὲ μύθων ἦρχε (or ἄρχε) [nominative name/epithet, speaker]
And [speaker] began the talk to/among them, 16 times in the Homeric poems (7x Iliad, 9x Odyssey)
Example: τοῖσι δὲ μύθων ἦρχε θεὰ γλαυκῶπις Ἀθήνη, Iliad 5.420
And the goddess grey-eyed Athene began the talk among them
τοῖς ἄρα μύθων ἦρχε (or ἄρχε) [nominative name/epithet, speaker]
[speaker] then began the talk to/among them, five times in the Homeric poems (1x Iliad, 4x Odyssey)
Example: τοῖς ἄρα μύθων ἦρχε Γερήνιος ἱππότα Νέστωρ, Iliad 2.433
Gerenian horseman Nestor then began the talk among them
τοῖσι(ν) δ(ὲ) [nominative name/epithet, speaker] ἤρχετο (or ἄρχετο, as printed by Allen against the manuscripts) μύθων
[speaker] began the talk to them, five times in the Odyssey
Example: τοῖσι δὲ Τηλέμαχος πεπνυμένος ἤρχετο μύθων, Odyssey 1.367
Then the thoughtful Telemachus began the talk among them
These verses appear for speeches that begin a sequence of talk that occurs in a group, whether or not the particular speech being introduced is to an individual member of the group or to the group as a whole.  This demonstrates that in a group context, both the addressee and the audience of a particular speech are seen as the “objects” of that speech.
II. group initial formulas with ἦρχ’ ἀγορεύειν
τοῖσιν δ’ [nominative name/epithet, speaker] ἦρχ’ ἀγορεύειν
[speaker] began speaking publicly to them, eight times in the Homeric epics (3x Iliad, 5x Odyssey)
Example: τοῖσιν δ’ Ἀντήνωρ πεπνυμένος ἦρχ’ ἀγορεύειν, Iliad 7.347
Antenor the thoughtful began speaking publicly to them 
III. group initial formulas with ἀγορήσατο καὶ μετέειπε
τοῖσιν δὲ [name, speaker] ἀγορήσατο καὶ μετέειπε
[speaker] spoke publicly and addressed them, seven times in the Odyssey
Example: τοῖσιν δ’ Ἀντίνοος ἀγορήσατο καὶ μετέειπε, Odyssey 4.773
Antinous spoke publicly and addressed them 
All of the initial formulas that contain the verb ἀγορεύειν, “speak in public”, introduce a speech to a group rather than to an individual in the presence of a group. Most of the time they introduce a speech that occurs in the context of a formal assembly as well. Although there is no character whose name appears in both II and III, the names that are used in each are of the same metrical shape: –⏑⏑–. However, names that appear in (II) above are modified with an epithet, so that the name plus the epithet of the speaker has the metrical shape –⏑⏑–⏑⏑–⏑⏑. In contrast, names that appear in (III) do not have such an epithet available, and the name appears alone. Hence, it is not clear whether these two verses should be considered alternate formulas for the same idea “spoke to a group” where the use of an epithet with the speaker’s name is the distinguishing feature; or whether there is a subtle difference in meaning between the verb portions of these two formulas and the use of an epithet is secondary to this verbal difference.
Introductory Formulas—Initial Speech by One Individual to Another: The formulas that introduce an initial speech by one individual to another individual both imply that the speaker took the initiative in speaking in relation to his addressee.
I. individual initial formulas with πρότερος/η προσέειπε
[accusative pronoun, addressee] πρότερος/η προσέειπε [nominative name/epithet, speaker]
[speaker] spoke first to [him], eleven times in the Homeric poems (10x Iliad, 1x Odyssey)
Example: τὸν πρότερος προσέειπε βοὴν ἀγαθὸς Διομήδης, Iliad 6.122
Diomedes of the great war cry spoke first to him
II. individual initial formulas with πρότερος πρὸς μῦθον ἔειπεBeing the first to speak often has a competitive connotation in Iliadic battle scenes. 
[accusative pronoun, addressee] καὶ [name, speaker] πρότερος πρὸς μῦθον ἔειπε
And [speaker] first spoke a word to [him], four times in the Homeric poems (2x Iliad, 2x Odyssey)
Example: τὸν καὶ Τληπόλεμος πρότερος πρὸς μῦθον ἔειπε, Iliad 5.632
Tlepolemus first spoke a word to him 
Introductory Formulas—Context-Specific: Reply, group reply, and initial speech introductory formulas all share the same basic structure. They prepare for a speech by mentioning only three essential features that all speeches have (name of speaker, sometimes with epithet; action of speaking; pronominal reference to addressee). The main verb of speaking in such formulas generally identifies the type of speech (what kind of turn is it? is the audience one person or a group?). Another group of full-verse speech introductory formulas exists that describes some aspect of the conversational context other than the kind of turn and the name of the speaker. Such verses often omit one or more of the three basic pieces of information that occur in the different kinds of reply and initial formulas (most often the name of the addressee). I have termed this group “context-specific” formulas.
One example of such a formula is ὅ σφιν ἐϋφρονέων ἀγορήσατο καὶ μετέειπεν (he in kind intention toward all spoke out and addressed them), which appears 15 times in the Homeric poems to introduce a speech by an individual to an assembled group of his allies or comrades. From the structural standpoint, this verse cannot stand alone to introduce a speech, because it does not name the speaker. Many context-specific speech introductions require an additional verse or verses in order to supply enough information to introduce the following speech clearly. Such verses do not appear as often as reply formulas, since their content restricts them to a particular context. For the same reason, there are many more different context-specific full-verse formulas than there are reply, group reply, or initial formulas. In addition, they have less in common with one another in terms of both structure and content than reply or initial formulas do. Context-specific formulas are found regularly with all types of speeches, including successive ones. In light of the rather eclectic nature of this group of formulas, a complete list of context-specific formulas is provided in Appendix III rather than here in the introduction. What follows is an overview of the kinds of contextual information that may be provided in a context-specific speech introductory formula.
Some context-specific speech introductory formulas identify by name not the speaker, as is commonly done in other kinds of formulaic speech introductions, but either the addressee or both the addressee and the speaker. Others specify the place that the speaker is standing (near the addressee, by the head of the addressee) or the mental state of the speaker (reproaching the addressee, grief-stricken). On the battlefield, several different full-verse formulas may characterize a speech as a shout. Occasionally, characters speak to themselves, and context-specific formulas introduce speeches “to one’s own heart.” A formula even exists to introduce a speech by an anonymous speaker. Speeches that occur in particular social contexts—laments, assembly speeches, battlefield vaunts, and announcements in athletic games—may be introduced by context-specific speech introductions related to the genre of the speech rather than the identity of the participants or the mental state or location of the speaker. Patterns that govern individual examples of this varied group will be discussed in more detail as relevant for specific conversational types.
An Anomalous Speech Introductory Formula—ἔπεα πτερόεντα: The verse καί μιν φωνήσας (or φωνήσασ’, for a feminine subject) ἔπεα πτερόεντα προσηύδα is a unique anomaly among these various categories of full-verse speech introductions (s/he uttered winged words and addressed him, 37 times in the Homeric epics).  This verse has a unique distribution in that it appears approximately equally often before replies, initial speeches, and single speeches. Only this formula appears to lack an association with a particular type of turn in conversation (i.e. reply or initial speech). In the Iliad, there are 8 examples of this verse in reply introductions (3% of replies), 5 examples in initial introductions (3% of initial speeches) and 4 examples in single introductions (2% of single speeches). For the Odyssey, the figures are 8 reply introductions (2% of replies), 9 initial introductions (8% of initial speeches) and 2 single introductions (3% of single introductions).  All other types of speech introductions are specific to one type of turn and appear by far most frequently with this turn type, although they can and do appear with other turn types.
Moreover, the construction of this verse resembles a context-specific formula in that, because it does not name the speaker, it cannot stand alone to introduce a speech. And yet, it neither conveys context-specific information nor is limited to an identifiable kind of conversational or social context. καί μιν φωνήσας ἔπεα πτερόεντα προσηύδα is constructed like a context-specific formula but lacks clear links to any identifiable context. This lack of a clear contextual reference is emphasized by the existence of several context-specific formulas that end in the half-verse ἔπεα πτερόεντα προσηύδα (he spoke winged words) but have clearly context-related beginnings.  In fact, the formulaic half-verse ἔπεα πτερόεντα προσηύδα is itself unique: no other partial verse speech introductory formula appears with a range of different context-specific initial half-verses to make various full-verse context-specific speech introductions, in addition to a full-verse formula that is not contextually specific. In addition to these peculiar features, σφεας (them) appears four times instead of μιν (him/her) to introduce a speech to a group. There is no other formulaic speech introduction for a speech to an individual that can be transformed into a group reply formula without changing both the verb of speaking and the object pronoun to reflect a group addressee. This is also the only speech introductory language that is used by both the primary narrator of the Odyssey and by Odysseus in Books 9-12.
In sum, we can see that the half-verse ἔπεα πτερόεντα προσηύδα appears in a uniquely broad range of speech introductory contexts. Similarly, the most commonly found full verse formula in which it is found does not behave like any other full verse speech introductory formula. The verse καί μιν φωνήσας ἔπεα πτερόεντα προσηύδα, indeed, essentially constitutes its own category of speech introductory formulas, a category that might be called “variable.” Taken together, these facts argue against any one meaning of the much-belabored phrase ἔπεα πτερόεντα  or any attempt to link together the verses in which this phrase appears. Rather, in order to evaluate the nature and context of a speech introduced by this phrase, it is most important to look at the rest of the language in the verse containing the phrase ἔπεα πτερόεντα. 
When the full-verse formula καί μιν φωνήσας (or -ασ’) ἔπεα πτερόεντα προσηύδα precedes a reply, approximately half the time, the immediate context contains some description of the emotional state of the person who is about to speak. Of the 38 instances of this full-verse formula, 18 precede a reply. 11 of these reply contexts also include a verse or verses describing the emotions of the person about to speak, such as the following formulaic couplet: 
ὣς φάτο, ῥίγησεν δὲ πολύτλας δῖος Ὀδυσσεύς,An additional five instances of this full-verse formula are associated with a description of the speaker’s emotion but do not precede a reply. We can say that the association of the full-verse formula καί μιν φωνήσας ἔπεα πτερόεντα προσηύδα with describing the speaker’s emotions is particularly strong in the case of replies, but that it also exists for speeches introduced with this verse that are not replies.
καί μιν φωνήσας ἔπεα πτερόεντα προσηύδα
So she [Calypso] spoke to him, but long-suffering great Odysseus
shuddered, and spoke again in winged words and addressed her
καί μιν φωνήσας ἔπεα πτερόεντα προσηύδα
So she [Calypso] spoke to him, but long-suffering great Odysseus
shuddered, and spoke again in winged words and addressed her
Speech Conclusions: Speech conclusions, separate verses that say something like “thus s/he spoke, and [something else happened next]”, generally occur at the end of a conversation or after a single speech. They are almost never found during one-on-one conversations. Although they do sometimes appear in the middle of group conversations, there are particular speech concluding formulas that tend to be used during conversations and these usually show up in specific kinds of contexts.
The six formulaic speech conclusions that appear more than ten times in the Homeric poems provide a representative overview of this kind of formulaic speech frame. One of these verses, ὣς ἔφαθ’, οἳ δ’ ἄρα πάντες ἀκὴν ἐγένοντο σιωπῇ (So he spoke, and all of them stayed stricken to silence, 15 times in the Homeric epics), appears almost exclusively in the middle of group conversations. Sometimes a speech introduction immediately follows; sometimes a longer passage describes the emotions of the audience as a result of what they have heard. It emphasizes the amazement or consternation of the group at what person A has said before person B at length replies, and it reminds us indirectly that normally, emotions do not run high enough to intervene between speeches and no gaps intervene between one speech and the next. This emphasizes the startling effect of the speech of person A. This formula marks a high point of emotion in a particular conversation: we do not find more than one of these in any one conversation. Formal group contexts are distinguished from other kinds of conversational situations by (among other things) the consistent or repeated appearance of formulaic speech conclusions in between one speech and the next during an ongoing conversation. 
The other five most common speech concluding formulas, which appear a total of 81 times, appear only seven times in the middle of a conversation rather than after speech is finished. The verse ὣς οἳ μὲν τοιαῦτα πρὸς ἀλλήλους ἀγόρευον (Now as these were speaking things like this to each other), which generally concludes a conversation, is responsible for five of these mid-conversation speech conclusions. This is not surprising, since this verse—unlike most speech conclusions—is associated specifically with conversation rather than with speech in general. Moreover, speech conclusions appear more often in the Iliad, which has a greater proportion of single speech, than they do in the Odyssey: of our six most common speech conclusions, 64 of 96 examples (67%) occur in the Iliad.
The six chapters are grouped into two parts of three chapters each. Part I focuses on one-on-one conversations (Chapters 1 and 2 on the Odyssey and Chapter 3 in the Iliad). In both poems, conversation consistently dramatizes a particular view of human relationships that is central to the poem’s overall themes and concerns. Part II considers other kinds of conversations. Chapter 4 discusses single genres of speech that commonly appear on the battlefield. These single genres have a central role in depicting the characters of the main characters in the Iliad; by relying heavily on speech genres and contexts that are usually found in battle, the poem implicitly underlines its view that human relationships are essentially hostile and adversarial. Chapters 5 (assemblies) and 6 (games and laments) focus on various kinds of formal group contexts. These genres have a number of characteristics in common with each other because of the common features that occur in formal contexts of various kinds. Moreover, the specific formal genres that the Iliad emphasizes both shape and depict its vision of the events it narrates, particularly the quarrel between Agamemnon and Achilles and the death of Hector.
Alongside this grouping into two parts of three chapters each, the book also falls into three parts of two chapters each: Chapters 1 and 2 focus on the Odyssey; Chapters 3 and 4 on one-on-one conversations in the Iliad; and Chapters 5 and 6 on formal group conversations in the Iliad. These pairs of chapters complement each other and usually come to similar conclusions; this arrangement also calls attention to the significant differences between the Iliad and the Odyssey in terms of which specific genres of conversation most prominently appear. Despite these differences, as we will see, the poems display broad and comprehensive similarities in the way that conversations vary from typical patterns and in the aesthetic and poetic significance of these variations.
[ back ] 1. Laird 1999:154n3 for the Apollonius figure; the Homeric figures are mine.
[ back ] 2. See recently Martin 1989 and Bakker 1997. Martin uses the speech act theory of J. L. Austin to explore in depth individual speakers and speeches in the Homeric poems, particularly Achilles. Bakker explores the language of the Homeric poems as a poetic version of oral speech rather than as an oral version of poetic speech. Kirk 1990:28-35 provides a general overview of speech as distinct from narrative.
Mackie 1996 forms a partial exception to this general tendency, in that one aspect of her inquiry into the differences between the Greeks and the Trojans as speakers focuses on their different behavior in face-to-face conversations on the battlefield (e.g. 6).
Mackie 1996 forms a partial exception to this general tendency, in that one aspect of her inquiry into the differences between the Greeks and the Trojans as speakers focuses on their different behavior in face-to-face conversations on the battlefield (e.g. 6).
[ back ] 3. In Virgil, in contrast, a recent study states that only one-quarter of the speeches in the poem that are represented in direct speech receive an answer (Laird 1999:183).
[ back ] 4. Throughout this study I will be using the word “narrator” to refer to the controlling and arranging force in the poems. I have chosen this terminology, rather than “poet”, because most readers agree that the Homeric poems do have a narrator.
[ back ] 5. Fenik 1968 is a groundbreaking study on aesthetics and type scenes (focusing on battle scenes). More recent full-length examples include Reece 1993 (hospitality scenes) and Louden 1999 (on the repeating narrative patterns in the Odyssey).
[ back ] 6. The term “speech frames” is used collectively to comprise both speech introductions and speech conclusions; this expression is also used for specific passages that contain both speech introductory and speech concluding language. Miller 1996 also uses the term “frame” for this kind of language in talking about the Hebrew Bible.
[ back ] 7. Diffey 1995:61.
[ back ] 8. Nagler 1967:273.
[ back ] 9. Russo 1994:372.
[ back ] 10. Austin 1966, Martin 1989:206-230, Nagy 1996:76-77.
[ back ] 11. Compare the length of the speeches that Demodocus quotes to those quoted by the main narrator of the poems. The expansionist effect of the speeches in the Homeric epics becomes even clearer if they are compared to transcripts of actual conversations.
[ back ] 12. Edwards 1990:323.
[ back ] 13. In addition to the studies already cited, Taplin 1992 puts forward an extremely articulate and appealing vision of the aesthetics of the Iliad along these lines (see especially 9-43), but his specific analyses are not fully persuasive in support of it.
[ back ] 14. Scholars who have called for such studies include Foley 1985:18, Edwards 1988:57, and Wright and Jones 1997:23, which specifically points to repetitive language as a fruitful area for research.
[ back ] 15. See Nagy 1979:1-5 for an important statement of the interrelationship of tradition, theme, and language in Homeric epic.
[ back ] 16. Found in his collected works, Parry 1987.
[ back ] 17. On this question, see also Lord (most influentially 1960), Nagy 1979, and the various publications of Foley during the 1990s (most recently 1999). Edwards’ exhaustive bibliographies (1986a, 1988, 1992) give basically all the scholarship on formulas and type scenes that was done up to the dates of publication.
[ back ] 18. An influential article written at this time has the illuminating title “Homer Against His Tradition” (Russo 1968). Segal, writing at almost the same time (1971), departs from this general tendency in an excellent study of the different ways that traditional formulas can be used that does not fall victim to the idea that “departure from traditional usage” has to mean “invent original usage.”
[ back ] 19. E.g. Foley 1991 passim.
[ back ] 20. Kullmann 1984 is a useful comparison of neoanalysis and oral theory as approaches to the Homeric poems. Kullmann 1960, an exhaustive examination of the Iliad, is one of the most influential works of neoanalysis; see also Kakridis 1949. It is noteworthy that neoanalytic scholarship has much more helpful and interesting things to say about the Iliad, for the most part, than it does about the Odyssey.
[ back ] 21. Holoka 1991 is a recent summary of the relations between these two schools of thought. Clark 1986 is a bibliographic review of neoanalysis which, like Holoka and Kullmann 1984, points out the similarities between neoanalysis and oral theory and uses these similarities to argue for more interchange between the two schools of thought. Reece 1994 is an example of a study that uses “neoanalysis . . . informed by the principles of oral theory” (158).
[ back ] 22. That is, they generally believe that each poem is a successful and effective unity, rather than the work of several different composers (in contrast to their scholarly predecessors, the Analysts). This is not uniformly true: Schadewaldt, the teacher of Kullmann, believed that the Odyssey was the work of several hands (see e.g. 1966), but the majority of more recent neoanalyst scholarship is unitarian (see e.g. Erbse 1972 for a thorough and painstaking unitarian refutation of various theories about interpolations and inconsistencies in the Odyssey, and in English, see the bibliographic review of Clark 1986).
[ back ] 23. These areas, mentioned briefly here, will be discussed in more detail in Section III of the introduction.
[ back ] 24. Hutchby and Wooffitt 1998:1.
[ back ] 25. Jaworski and Coupland 1999:xi.
[ back ] 26. Jucker 1995:3-6.
[ back ] 27. See e.g. the categories in the bibliographic review of Edwards 1992 and the specific entries under “typical scenes” in Russo et al. 1992.
[ back ] 28. Edwards 1992:285.
[ back ] 29. Edwards 1992:316.
[ back ] 30. Arend 1933:6.
[ back ] 31. So for example Wilamowitz 1916:75-77 on the relationship between the death scenes of Hector and Patroclus.
[ back ] 32. 1960, especially 68-98.
[ back ] 33. Lord 1960:69; see also Foley 1985:32.
[ back ] 34. A rough and ready demonstration of this is available from comparing the total length of Edwards’ review bibliographies on formulas (1986a:171-230 and 1988:11-60) with the one on type scenes (1992:285-330).
[ back ] 35. Fenik 1968:165.
[ back ] 36. Fenik 1968:204.
[ back ] 37. Edwards 1992:293 describes it as “a work of the highest importance for the understanding of the battle scenes of the Iliad;” while true as far as it goes, this represents the general and (in my opinion) unnecessarily limiting view of the book as a study of battle scenes in particular rather than of type scenes in general.
[ back ] 38. But see Gunn 1971 for a demonstration that the Iliad and the Odyssey use the supernatural visit type in the same way. Based on this, Gunn asserts that both poems are the work of the same poet. His brief studies (1970 focuses on types in order to argue for an oral poet who dictated the Iliad and the Odyssey) do not allow for the kind of broad or overall view of the poems that I am arguing for here.
[ back ] 39. Indeed, it has recently been argued that economy is confined to such expressions (Foley 1999:210-211).
[ back ] 40. For Parry’s ideas about economy and extension, see Parry 1987:7 and passim.
[ back ] 41. Parry 1987:328-329.
[ back ] 42. See e.g. Parry 1987:22: “use of the fixed epithet, that is, of the ornamental as opposed to the particularized epithet, is entirely dependent on its convenience in versification [and not on the poet’s choosing to use it].”
[ back ] 43. Combellack 1959 is the most influential and also one of the strongest statements of this view.
[ back ] 44. On flexibility, see in particular Hoekstra 1965 and Hainsworth 1968. For artistry and formulas, Holoka’s Homer bibliography from 1973 gives a good overview both of the contemporary work being done and also of the assumptions and ideas about artistry that condition the debate about artistry at that time: the bibliography is entitled “Homeric Originality: A Survey.”
[ back ] 45. Parry 1987:13. For a recent useful survey of definitions of “formula”, see Russo 1997, especially 245. Nagy 1979:272 puts forward the fascinating idea that formulas are “the selfsame words spoken by the Muses themselves.” See also Nagy 1996:22-25 on what formulas are (and aren’t).
[ back ] 46. Russo 1997:242.
[ back ] 47. Speech conclusions have thus far received almost no scholarly attention, so in this section I will be talking about “speech introductions” rather than “speech frames.”
[ back ] 48. Foley 1999:104.
[ back ] 49. Ibid. 221.
[ back ] 50. Krarup 1941; Fournier 1946; Edwards 1969 and 1970.
[ back ] 51. Edwards 1970:1.
[ back ] 52. Ibid. 35.
[ back ] 53. Janko 1981, Riggsby 1992, Olson 1994.
[ back ] 54. Machacek 1994, anticipated to some extent by sections of Austin 1975 (in particular Chapter 1); and Beck 1999. The appendix to Pope 1960 points out instances of lack of economy in speech introductions for Athena, but does not explore the aesthetic possibilities of this lack of economy.
[ back ] 55. Books on direct speech do not discuss the role of speech introductions in connecting speech to narrative. Lohmann’s classic work on the structure of speeches (1970) has little to say about the interaction of speech and speech introduction; Martin 1989, which is centrally concerned with different ways of characterizing Homeric speech, does not discuss speech introductions per se and has no entry in its index for these expressions.
[ back ] 56. As is demonstrated, for instance, by the extensive bibliography on the meaning and linguistic origin of the participial expression ὑπόδρα ἰδών, on which see Holoka 1983 with bibliography.
[ back ] 57. For a wonderfully sensitive analysis of such usage, see Segal 1971 on formulas for dying warriors applied to Andromache when she sees Hector killed before Troy.
[ back ] 58. See recently Burgess 2001.
[ back ] 59. Clanchy 1979:19.
[ back ] 60. Besides the comparatively well known poetry of Avdo Međedović in Yugoslavia (in Lord and Bynum 1974), lengthy oral compositions are known from (e.g.) the Philippines, where a poem of over 10,000 verses requiring 18 hours to chant has been documented (Maquiso 1977:38 and 47).
[ back ] 61. See Harriott 1962 on the nature and possible mechanisms of allusions to Euripides in Aristophanes.
[ back ] 62. West 1990:46-47.
[ back ] 63. Niles 1999:104-107. He suggests that oral performance that is recorded, rather than being an adulterated version of a “true” oral performance, should be studied and valued as a distinct type of performance in its own right. He goes on to say (224n33) that the theory of oral dictation and Nagy’s evolutionary theory can be reconciled if the date of oral dictation is moved later than it is posited by e.g. Janko.
[ back ] 64. Ibid. 128.
[ back ] 65. Chantraine 1990.
[ back ] 66. 85 full-verse reply formulas with ἀμειβ- in them out of 188 reply formulas in the Iliad; 99 of 235 in the Odyssey.
[ back ] 67. These two characters are the only ones who regularly quote directly the conversation of other characters. Their formulaic repertoire includes ὣς ἔφατ’, αὐτὰρ ἐγώ μιν ἀμειβόμενος προσέειπον (So s/he spoke, and I in turn spoke up and made answer; 14x Odyssey); ὣς ἐφάμην, ἡ δ’ αὐτίκ’ ἀμείβετο δῖα θεάων (So I spoke, and she, shining among the goddesses, answered; 5x Odyssey); ὣς ἐφάμην, ὁ δέ μ’ αὐτίκ’ ἀμειβόμενος προσέειπεν (So I spoke, and he in turn spoke up and made answer; 3x Odyssey).
[ back ] 68. Greek quotations are from Allen unless otherwise noted. Translations are generally from Lattimore 1951 and 1965, but some are my own where Lattimore’s rendering obscures some aspect of the Greek that is important for my point. I have retained Lattimore’s spelling, even where it is inconsistent with mine, to remind the reader that translation is an interpretation.
[ back ] 69. See Viechnicki 1994 for a linguistic and anthropological discussion of the etymology of ἀμείβω in terms of its connection to gift-exchange and its cognates in other Indo-European languages. He proposes to define ἀμείβω as “social reciprocity” (120) or “balanced reciprocity” (122) rather than simply “exchange”, which fits well with its conversational use.
[ back ] 70. The middle voice is also used in a set of compounds that denote movement from one place to another, a sense which Viechnicki links to the original Indo-European stem’s intransitive meaning. Over time, this meaning was supplanted in Greek by the “answer” meaning (130-131).
[ back ] 71. Viechnicki interprets this differently: he attributes the difference in voice to a supposed intransitive sense of “answer” (124) and connects this to the tendency of Indo-European verbs to have transitive meanings in the active forms and intransitive ones in the middle or passive. I think that explanation is unsatisfactory, given that forms of ἀμείβομαι are almost always found with accusative pronouns.
[ back ] 72. Verbs commonly found in formulas to introduce replies that also occur in non-reply contexts include the participle φωνήσας (spoke) as well as the finite verb forms προσέειπε(ν), προσηύδα or ἀντίον ηὔδα, and προσεφώνεε(ν) (all of which essentially mean “addressed”). Most of these occur predominantly in replies, but all occur regularly to introduce speeches that are not replies.
[ back ] 73. Odyssey 5.96: καὶ τότε δή μιν ἔπεσσιν ἀμειβόμενος προσέειπεν (then he began to speak, answering what she had asked him). [ back ] In this speech, Hermes is addressing Calypso at the beginning of a conversation. Before this point, he has arrived at her island to give her the message that she must let Odysseus go (75-80) and she asks him what his business with her is (87-91). Hermes answers this question (97-115) but not until after Calypso has served him a meal (92-95). As a result, his speech both begins a conversation and answers a previous question. In fact, this single example of ἀμείβεσθαι to introduce an initial speech is the exception that proves the rule.
[ back ] 74. Sacks et al. 1974.
[ back ] 75. Sacks et al. 1974:696.
[ back ] 76. Sacks 1992 v. 2:32. 1992 v. 2:32-43 provides an extremely helpful overview of the question of “what is a conversation?” from the standpoint of conversation analysis.
[ back ] 77. Sacks et al. 1974:714-715.
[ back ] 78. See e.g. Drew and Heritage 1992 on features of conversations that take place in an institutional environment.
[ back ] 79. E.g. Silverman 1998:30; Searle 1992 (with response in the same volume by Goffman) critiques the idea that conversationalists are following a rule.
[ back ] 80. Sacks et al. 1974:699-700 note the simultaneous context-specificity and context-independence of conversation analysis.
[ back ] 81. Jaworski 1999:xi.
[ back ] 82. Kress, G. and T. van Leeuwen 1999:379.
[ back ] 83. Ibid. 380.
[ back ] 84. For a recent survey of the language features studied in pragmatics, see Levinson 1983. Miller 1996, although she does not use the term “historical pragmatics”, is an example of a study that applies linguistic principles that are primarily pragmatic in orientation to an ancient literary text (the Hebrew Bible).
[ back ] 85. Jucker 1995:6.
[ back ] 86. Lloyd 2004 performs such an analysis of the root κερτομ- in Homer.
[ back ] 87. Jucker 1995:10.
[ back ] 88. Using the word “story” in the narratological sense to refer to the “signified or narrative content” (Genette 1980:27) or “the narrated events and participants in abstraction from the text” (Rimmon-Kenan 1983:6) in contrast to the specific version of these contained in the text.
[ back ] 89. These results are given in table form in Appendix I.
[ back ] 90. Goffman 1981:79n2 uses the word “single” to mean “a party of one present among other parties”, in contrast to a solitary person who is all alone. This is usually the situation of a speaker who makes a single speech in Homeric poetry.
[ back ] 91. 23 times in the Homeric epics, 22 of which occur after a single speech or at the end of a conversation.
[ back ] 92. See de Jong 1987b: 80-84 on the function of τις-speeches.
[ back ] 93. Although not the exact form of words: elision and nu-movable are not considered to change a formula to a different formula when such phenomena occur in order to accommodate proper names that begin with consonants versus vowels.
[ back ] 94. Slightly adapted from Lattimore. One of the areas in which even Lattimore, the most exact translator of Homer, departs from the Greek is in the different translations he gives for repeated expressions like καὶ τότε, which at different times he renders as “at this” (Iliad 1.92), “then” (Iliad 4.444), and “now” (Odyssey 2.389), to name just three examples.
[ back ] 95. Edwards 1970 uses the same criteria in relation to the flexibility of particles and the names of addressees for classifying speech introductions (4).
[ back ] 96. Cf. Fenik 1968:5, which defines language that appears twice or more in the Iliad as “typical.”
[ back ] 97. Some repetitions, as Strasser 1984:45 points out, are not even intentional.
[ back ] 98. Edwards 1997:274.
[ back ] 99. For the importance of “pre-existing” as a criterion for discussing formulaic language, see Hainsworth 1993, especially 16.
[ back ] 100. In these figures and throughout this study, Odysseus’ retelling of his adventures to the Phaeacians is considered one long speech insofar as it represents one turn at speaking by a character within a conversation organized according to recognizable turn-taking structures. Hence, formulas and conversational structures that Odysseus uses in his narration in Books 9-12 are not counted in my data. Similarly, direct speech in the second song of Demodocus is not counted, because it too is reported by a character and not by the primary narrator of the poem. I am currently at work on a study of speech representation by characters in the Homeric epics, in which I will deal with this language and related phenomena.
[ back ] 101. The data in the pages that follow and in the Appendices was obtained through Pandora scans of the TLG Disk D. These scans were checked and corrected against concordances and against the texts of the poems (Allen’s Oxford Classical Texts, primarily, with reference to van Thiel, Leaf, and West where conflicts arose among the readings of Pandora, concordances, and Allen). The verses to be scanned were identified through multiple readings of the entire text of both poems, readings during which each speech introduction and speech conclusion and its main verb (or expression) of speaking were separately scanned with Pandora to determine how frequently the language appeared.
[ back ] 102. Includes two instances with ἀπαμειβόμενον (neuter participle instead of masculine) and 13 with προσέφης (“you spoke”, vocative addresses to Eumaeus). The expression ἀπαμειβομένη προσέφη (participle of feminine gender) never appears in Homeric epic. In the compound form ἀπαμείβομαι, the reciprocity already inherent in the verb ἀμείβομαι is enhanced further by the reciprocal prefix ἀπ-, “back again” (LSJ def. D4).
[ back ] 103. Adapted from Lattimore; nearly all of the translations in this section are adapted in order to bring out as literally as possible the small differences in meaning that distinguish one formula from another. Translations that depart significantly from Lattimore’s are marked as my own.
[ back ] 104. A complete list of participles that occur in this verse is given in Appendix II.
[ back ] 105. This appears much more often in the Odyssey than it does in the Iliad (14x Iliad, 54x Odyssey) due almost entirely to the frequent appearance in the Odyssey of the Telemachus verse quoted here.
[ back ] 106. Several MSS. have this verse for Odyssey 17.405, but it is not included here (although it is the reading preferred in Dunbar’s concordance) because it is printed by neither Allen nor van Thiel.
[ back ] 107. Sacks et al. 1974:714.
[ back ] 108. These group contexts are the subject of Chapters 5 and 6.
[ back ] 109. In addition to Iliad 1.58, also 2.411 (εὐχόμενος “praying”), 4.153 (βαρὺ στενάχων “groaning heavily”) and 19.55 (=1.59); Odyssey 18.51 and 21.274 (both of these are τοῖς δὲ δολοφρονέων μετέφη πολύμητις Ὀδυσσεύς [But now resourceful Odysseus, with crafty thoughts, said to them]).
[ back ] 110. On four puzzling occasions, however, a group initial verse introduces a speech to Odysseus when no one but he and the speaker are present: Odyssey 5.202 (Calypso addresses him), 7.47 (Athena), and 13.374 (also Athena), and 17.184 (Eumaeus). Edwards 1970:8n16 includes two occurrences in the conversation between Penelope and Odysseus in Odyssey 19 in his list of non-group contexts (103 and 508). In fact, this conversation takes place in the palace hall in the presence of other people, a fact that contributes at times to the drama of the scene.
[ back ] 111. My translation.
[ back ] 112. My translation.
[ back ] 113. My translation.
[ back ] 114. See Chapter 4 on the relationship between speaking and attacking in battle scenes.
[ back ] 115. The total figures in Appendix I for “variable formula” vary slightly from these because they include variations like καί σφεας (them; occurs four times) for καί μιν (him/her) and ἀμειβόμενος (taking a turn; occurs twice) for φωνήσας (addressing). Also, the figures I have given above do not include the few instances of this formula used by characters rather than the primary narrator.
[ back ] 116. One such verse in the Odyssey introduces a successive speech.
[ back ] 117. The most common of these is ἀγχοῦ δ’ ἱστάμενος/η ἔπεα πτερόεντα προσηύδα (he came and stood close beside him and addressed him in winged words, 13 times); a complete list of context-specific formulas that end with ἔπεα πτερόεντα προσηύδα can be found in Appendix IV.
[ back ] 118. See Létoublon 1999 for a recent overview of the various interpretations and interpreters of this formula.
[ back ] 119. Here I differ from Martin 1989, who sees ἔπεα πτερόεντα as introducing a “highly marked” speech (30). Rather, I agree with Latacz 1968, who argues that πτερόεντα is a generalized epithet referring via an implied comparison with arrows to the fact that words, once spoken, cannot be unspoken.
[ back ] 120. Three times in the Homeric poems, with various names at the end of the first verse.
[ back ] 121. See Chapters 5 and 6.