Particles in Ancient Greek Discourse: Exploring Particle Use across Genres

Citation: Bonifazi, Anna, Annemieke Drummen, and Mark de Kreij. 2016. Particles in Ancient Greek Discourse: Exploring Particle Use across Genres. Hellenic Studies Series 79. Washington, DC: Center for Hellenic Studies. http://nrs.harvard.edu/urn-3:hul.ebook:CHS_BonifaziA_DrummenA_deKreijM.Particles_in_Ancient_Greek_Discourse.2016.

II.3 Moves: Particles at discourse transitions

§1. The present chapter builds directly on chapter II.2 and presupposes knowledge of its main points; our understanding of discourse acts is summarized in II.2.1.2. Greek particles reflect the production of discourse in cognitively manageable units – discourse acts – which are the building blocks of epic and lyric compositions. The analysis in II.2 reveals how a performer or author produces his work in small increments to guide his audience through the discourse. In this chapter I am concerned with all kinds of larger discourse units, such as narrative episodes or scenes, and the function of particles with relation to them. [1] First, I introduce the term “move” to describe coherent discourse units consisting of at least one, but generally of multiple discourse acts (§§6-11). An understanding of this phenomenon in discourse then informs my reading of γάρ (§§22-29), καὶ γάρ (§§30-32), and ἦ (§§33-43) in Homer and Pindar, focusing on the introduction of Homeric embedded narratives as a case study. After examining a few other ways in which the poets embed narratives within the larger discourse (§§45-50), I examine how those narratives themselves are articulated. In this section I explore the different functions of δή, considering also the diachronic development of the particle that may emerge from the Homeric corpus (§§53-64). A narrative from Pindar provides the backdrop for a discussion of his consistent use of ἄρα to round off larger units in discourse (§§65-67). Finally, I consider the role of particles in one of Pindar’s compositions (Pythian 2), especially in the many transitions between different kinds of discourse (§§68-76).

3.1 Moves

§2. Discourse acts, the smallest functional subdivisions of discourse, are the building blocks of larger sections called “moves.” [2] The term originated in the analysis of dialogic discourse, where it makes sense to divide conversation into different moves conceived by the interlocutors. [3] An often-cited example is that of the invitation: [4]

(t1)

A: Are you free tonight? | ‘Cause I have an extra ticket for the symphony orchestra. |

B: Well, | I really should work on this paper tonight. | Sorry, | maybe next time? |

In two turns, which could be expanded almost infinitely, speakers A and B go through an exchange that consists of an “invitation” and “rejection.” [5] Each of the two turns consists of multiple acts that are united by the fact that they share a common communicative goal. [6]

§3. In the example above, the move starts with an unannounced question, but in practice the context will provide some kind of embedding. We may expect a prefatory remark like, “Oh, by the way, are you…”, when the invitation comes after another move (such as a greeting), or, if the interlocutors have just met face-to-face, “Hey, I wanted to ask you something: are you…” On the whole, when speakers are initiating a new move, they feel a strong need to mark the transition. At the “end” of moves, conversely, there is typically less explicit marking in the language. [7] Of course, some moves have inherent endings, like the invitation above, but just as often the end of one move is recognizable only because another move begins. [8]

§4. The terminology of moves and acts forms part of a larger framework concerning the subdivision of discourse, and it is most often applied to analyze dialogic discourse. [9] In this framework, researchers are concerned with establishing the structure of an exchange, which they divide into main and subsidiary elements. [10] Since I approach epic and lyric from the perspective of performance, I do not believe such a hierarchical analysis to be the most productive. Rather, I choose to focus on the on-line delivery and processing of the discourse. [11] Whatever structure may emerge from the written text, in performance the cues to structure would have worked mostly on a local level, guiding the audience linearly through the discourse. [12] What one expects to be marked, then, are the transitions to new moves, since these are the places where both performer and audience need a cue for cognitive reorientation. Since epic and lyric performance make reception linear, the most important thing is that a move beginning is recognized as some sort of redirection. [13]

§5. The most recognizable transitions in Homer are the beginnings and endings of direct speech. Metalinguistically, every start of direct speech is marked explicitly with a speech-introduction verse, probably because the transition from one speaking source (the narrator) to another (a character) is crucial to following the narrative. The same holds for the great majority of transitions after direct speech. [14] The performer of the Iliad or the Odyssey had only one voice, but comparative evidence from other epic material suggests that transitions from direct speech to narrator text may have been marked prosodically. [15] In the remainder of Homeric epic, most moves are less obviously recognizable as units than stretches of direct speech. Only by considering both the content and the linguistic form can we come to an informed analysis of transitions in discourse. Such moments are often marked by some kind of linguistic turbulence, [16] in the form of a change in tense, orientation, or source, or through the occurrence of metalanguage, often in the form of particles. Attention to the path of the discourse, with all its bends and sudden turns, may thus reveal certain functions of particles. At the same time, the occurrence of certain particles may serve as a cue that the discourse is taking a new direction.

3.1.1 Move transitions

§6. Because epic and lyric have been transmitted in written form, it is all too easy to consider the genres as ultimately monologic discourse, structured by an author according to a functional hierarchy. As argued in the II.4, however, it is important to keep in mind that both epic and lyric represent interactions between performer and audience. In fact, traces of this ultimately interactive nature are present throughout the discourse, not only in the use of particles that mark its production in acts, but also in more explicit metanarrative comments. Consider the introduction to the Homeric Catalogue of Ships:

(t2)

ἔσπετε νῦν μοι Μοῦσαι Ὀλύμπια δώματ’ ἔχουσαι· | [17]
(…)
οἵ τινες ἡγεμόνες Δαναῶν καὶ κοίρανοι ἦσαν· |
πληθὺν δ’ οὐκ ἂν ἐγὼ μυθήσομαι | οὐδ’ ὀνομήνω, |

Iliad 2.484 and 487-488

Tell me now, Muses who have Olympian houses,
(…)
who were the leaders of the Greeks and the captains.
Their multitude I would not be able to speak nor name,

In a gesture rare in Homeric epic, the (persona of the) narrator breaks the spell of the narrative and foregrounds his role as a performer. Most of the Homeric narrative effaces the presence of the performer and audience as much as possible; in order to maintain the illusion of their absence, the performer divides his discourse by more subtle means. In this instance, however, a special piece of discourse is set apart by a special introduction. [18] The performer professes his inability to complete his intended move without help from the Muses, which he apparently receives, [19] since he concludes:

(t3)

οὗτοι ἄρ’ ἡγεμόνες Δαναῶν καὶ κοίρανοι ἦσαν·

Iliad 2.760

These then were the leaders of the Danaans and the captains.

The listing of the leaders of the Greek army is thus composed as a separate move, explicitly demarcated by metanarrative comments. [20]

§7. When I call the Catalogue a separate move, I mean this in a relative sense. Above the level of the act, different-sized units are hard to distinguish on any absolute basis. [21] In the Catalogue, for example, it would be fully justified to call each geographical subsection a “move” within the larger whole. Rather than assigning different names to moves of different sizes, we use a single term – move – with the understanding that the concept is a relational one: the Catalogue is a move within the Iliad, the subsection on the men from Pylos (2.591-602) is a move within the Catalogue, and the story of the singer Thamyris (2.594-600) is a move within that subsection. [22] With these caveats we may productively approach ancient Greek epic with the terminology employed by the Birmingham and Geneva schools. [23] The performer of epic is engaged in a constant interaction with his audience, the reflections of which must inform our understanding of the language.

§8. For Pindar’s lyric compositions the term move is not only helpful, it is, as we shall see, indispensable. [24] While it is not intuitive to describe narrative scenes or episodes in terms of moves, lyric discourse invites it. The work of Schadewaldt and Bundy has taught us to see Pindar’s Victory Odes as coherent wholes conceived with a single goal in mind: praise. This praise can be multi-faceted, but it is present in every component part of the ode. The priamels, the praise of the family, the reflections upon humanity and the gods, the narrative sections, and of course the praise of the victor are all moves in the larger interaction that is the song, which in turn is part of the festive occasion. The component parts of Pindar’s songs, different yet fitting into a larger whole, represent Pindar’s many moves toward his goal of praising the victor. [25] Just as a speaker might prepare his interlocutor for an actual invitation by starting with “Are you free tonight?” Pindar often primes his audience for the praise of the victor by starting with a priamel. Another clearly observable kind of move are the gnomic, evaluative statements that occur within narratives, or at transitions between myths and direct praise. [26] Just like scenes and episodes in epic narratives, Pindaric moves are distinguished by linguistic markers, and are characterized by a specific linguistic form.

§9. We call larger and smaller subdivisions of epic and lyric discourse “moves,” and we assume that these linguistic units in discourse represent specific aspects of a composer or performer’s discourse strategy. This assumption is implicit in our belief that the discourse units we identify represent a textual reality and are not just abstractions projected upon the text. Our interpretations always go hand in hand with an analysis of the linguistic form at what we believe are move transitions. For Homer, I focus on the transition into and out of self-contained stories, as well as the navigation of transitions within narrative. [27] For Pindar’s lyric, my analysis involves transitions between one kind of discourse and another, as from narrative to gnṓmē, or from an address of a god to praise of the victor’s clan.

§10. A performer has a number of tools at his disposal to mark new beginnings. One basic difference we can establish is between transitions with and those without metalanguage. Non-metalinguistic transitions include instances like the priming discourse acts discussed in the preceding chapter, which contain no metalanguage, or language that “discusses” the transition. Instead, the narrative is presented in such a way as to accommodate the cognitive processing needed to navigate the discontinuity. [28] Metalinguistic transitions, on the other hand, are those where the performer uses language that is relevant not to the propositional content, but to the ongoing interaction between him and the audience. Metanarrative questions or comments (such as the opening of the Catalogue) are the prime example of this kind of transition, but they occupy only one end of the scale. The rest of the spectrum is taken up mostly by particles or combinations of particles. [29]

§11. Awareness of different moves in discourse is indispensable for gaining a full understanding of certain particles. In narrative contexts, as in many others, particles are better understood as relevant to the interaction than to the content or syntax. [30] Herman says the following when discussing cognitive narratology: “At issue is how stories reflexively model cognitive, interactional, and other dimensions of acts of narration along with other forms of communicative practice.” [31] His corpus is written discourse, and in the case of performed narratives like the Homeric epics and Pindar’s songs we may focus not only on “how stories model cognitive and interactional acts,” but also on how the language of Homeric and Pindaric stories reflects and encodes such cognitive and interactional acts. Particles are crucial tools for locating and understanding such metalinguistic actions behind and beyond the texts.

3.2 Particles in narrative

§12. Narrative has received special attention as an object of research ever since the work of the structuralists in the mid-twentieth century, [32] but especially since the advent of narratology. [33] The approach has gained considerable popularity in the field of literary studies and has engendered a significant number of studies in Classics. [34] On the whole, narratology’s focus is on the relation between narrator (as removed from author or performer) and narrative, and it considers the latter more as a product than as an ongoing event. As a result, the linguistic form of a story is relevant mainly when it informs the researcher of the narrator’s manipulation of time or space. [35] In the following, we examine the role of particles in articulating and guiding narrative, approaching the texts as encodings of an interactive process between performer(s) and audience. By treating Homeric and Pindaric stories as acts of narration and studying the ways in which language reflects the narrative process we aim to provide a complement to the traditional narratological analyses. [36]

§13. If a narrative is perceived as an ongoing interaction between a performer and an audience, it will yield linguistic signs of the composer’s cognitive processes as well as his assumptions about the hearer’s cognitive processing. In the production of a discourse, whether fully composed beforehand or in situ, [37] multiple cognitive processes simultaneously affect the linguistic realization. At the most basic level, the composer’s linguistic competence and his training in composing are important determinants. This largely unconscious dimension of language production includes considerations of linguistic limitations, social conventions, and his own cognitive limitations. [38] More conscious is the influence of his expectations about what the audience will be able to process, what relevant pre-existing knowledge they have, and of course what they will appreciate. Such cognitive processes are inherent to all communication, [39] but since epic and lyric are not day-to-day speech, but rather “special discourse” meant for performance before a listening audience, their literary language may more visibly reflect this active consideration of the audience. [40] With this in mind, let us consider how these processes manifest themselves in the performer’s construction of his discourse in acts and moves.

3.2.1 Narrative moves

§14. In the emergent structure of discourse in corpora like Homeric epic and Pindar’s lyric, the component parts of stories are most clearly delineated, especially in Homer. Things do not just happen in stories, but situations are set up, characters are introduced, a complication is presented, and after things come to a head a resolution is achieved. This basic structure of narrative first proposed by Labov and Waletzky has been shown to be inherent in stories across languages, cultures, and media. [41] Their narrative scheme looks as follows (after adaptation in later publications): [42]

(t4)

Abstract (a short preview/overview of the narrative)

Orientation (introduction of time, place, and characters)

Complicating Action (the body of the narrative)

Resolution (the complication is resolved)

Coda (a metanarrative comment to cap the narrative)

Evaluation (the point of telling the story. In the earliest versions of the scheme “evaluation” was placed between “resolution” and “coda,” but in later publications Labov, Waletzky, and others argued that evaluation is generally present throughout the narrative, in the choice of vocabulary, syntax, evaluative meta-comments, gestures and facial expressions, et cetera.) [43]

The construction of narrative along these lines may be regarded as the result of social convention, but this convention in turn probably has a cognitive basis. [44] Whatever form the narrative takes, verbal or not, it is often possible to distinguish the different constituent parts. The omnipresence of this narrative structure suggests either that the storyteller is aware of the differences between sections and presents them differently, or that a certain linguistic form comes automatically with a certain part of the story—it is hard to establish the extent to which some of this linguistic marking is conscious. Regardless, in the Homeric and Pindaric corpora there appears to be a consistent marking of boundaries between Labovian sections of narrative, and they are realized in different linguistic forms. In the following, I first study embedded narratives as moves within the larger discourse, with a focus on Homer, and I then examine different narrative sections as moves within those stories.

§15. Of all the components in the Labovian model, the complicating action and the resolution are the minimal requirements for something to be perceived as a narrative. When either or both of those components are missing, the most we can say is that an expected narrative is lacking, but not that an actual narrative is presented. Most commonly, a narrative will have a complicating action and a resolution, often preceded by at least some kind of orientation. Multiple scholars have proposed to include an additional component between complicating action and resolution, alternatively called “climax” or “peak.” [45] From a linguistic perspective this claim is sound, since the language of climactic scenes commonly differs from that of the surrounding discourse. Therefore I employ the terms climax and peak to denote the pivotal scene between complicating action and resolution. In Labovian terms, the climax may be regarded as the last part of the complicating action, or the start of the resolution. [46]

§16. Consider the following example of a two-sentence narrative in Pindar’s first Pythian Ode. It tells how the Greek heroes on their way to Troy came to get Philoctetes, son of Poias, on Lemnos:

(t5)

Pindar Pythian 1.50-55

50      (…) Yes, just now, following Philoktetes’ way,
          he campaigned, and in need even a proud one
          greeted him as a friend. They say that from Lemnos they came to take him, tired out by his wound, [47]
          the godlike heroes, Poias’ archer son.
          He razed Priam’s city, and ended the Danaans’ troubles.
55      Though walking with flesh infirm, still it was fated.

After a brief orientation (50-51), consisting in the naming of a main character (Philoctetes) as a parallel for the laudandus (Hieron), Pindar presents a famous narrative in an extremely condensed form. The actual start of the complication is marked by φαντὶ δέ, which shows once more that δέ in Pindar regularly marks major discourse boundaries. [48]

§17. The complicating action takes up no more than one sentence (52-53 Λαμνόθεν… τοξόταν) before coming to the climax (ὃς Πριάμοιο πόλιν πέρσεν) and then the resolution (τελεύτασέν τε πόνους Δαναοῖς). Note that the run-up to the most important event is presented in the form of an accusativus cum infinitivo, while the climax – Philoctetes’ role in bringing down Troy – is in the form of a finite clause. This climactic clause, moreover, is formally a relative clause, which is perhaps not expected to drive the narrative forward. [49] It is likely that Pindar is influenced by the use of the relative pronoun in Homer, where its value is regularly closer to that of a demonstrative. Hence, my translation reflects the narrative force of the act (He razed Priam’s city, and ended the Danaans’ troubles) rather than the grammatical form of the clause (…Poias’ archer son / who razed Priam’s city…).

§18. The point of the story is the analogy between Philoctetes, who went to battle wounded, and the laudandus Hieron, who did so as well. Here the evaluation (ἀλλὰ μοιρίδιον ἦν), in Labov’s term, follows the resolution. The punctuation chosen by Snell and Maehler reflects the interpretation that the participial clause ἀσθενεῖ μὲν χρωτὶ βαίνων following the resolution is still part of the narrative. On several linguistic grounds, however, it is more attractive to regard this act as the start of a new move. [50] First of all the act would be superfluous on the level of content if it were part of the preceding narrative (see ἕλκει τειρόμενον, 52). More important, however, is the use of the particle μέν. As I argue in II.2, μέν has a projecting function, carrying a constituent, thought, or action forward. [51] As a result it also often has an asseverative function, starting a new move, here with the pragmatic enrichment that gives it a concessive force. Rather than to the preceding finite verbs, μέν connects its host act to the ἀλλά clause that follows: “Though walking with flesh infirm, still it was fated [that he would win the battle].” If taken in this form, the statement is readily applicable to both Philoctetes and Hieron, the laudandus. This reading is supported by the imperfect ἦν, which follows the aorists in the complicating action (μεταβάσοντας ἐλθεῖν 52), climax (πέρσεν 54), and resolution (τελεύτασεν 54): a change of tense suggests a new move, and the imperfect often occurs in evaluative statements. [52]

§19. This brief example illustrates two things: on the one hand the fact that narrative transitions can be explicitly marked in the language, and on the other hand that different narrative sections may manifest themselves through different linguistic patterns. [53] By analyzing these two factors we can observe how certain particles or strings of particles are relevant to transitions between narrative moves.

3.2.2 Narrative beginnings: γάρ

§20. The Homeric Iliad and Odyssey are practically all narrative, in the narrow sense. So most relevant for the current chapter are the transitions to the many discrete narratives embedded within the larger plots of the two epics. [54]

§21. Literature on self-contained stories in Homeric epic abounds, and the scholarship has established an impressive set of terms to describe the phenomenon from different angles. One kind is what Slater called the lyric narrative, characterized especially by its ring-compositional form. This form is not unconnected from the idea of epic regression, although it has sometimes been treated separately. [55] Slater, following West, notes that the lyric narrative in Homer is often introduced by a relative pronoun + aorist + ποτε + aorist participle. [56] Contrary to what Slater suggests, however, this kind of construction does not hold the monopoly on introducing embedded narratives in epic. Each component of Slater’s construction in combination with other elements can start a self-contained narrative, which is not by nature different from the “lyric narrative” described by Slater. In all such constructions, the pronoun often does not have a relative but rather a demonstrative force: not “…, who” but “…. S/He.” As for ποτε, although it is by nature an ideal marker to use for displacing performer and audience to a moment before the narrative hic et nunc, parallels show that it is by no means the performer’s only or even the most frequently used tool. The particular construction described by Slater is an asyndetic transition into an embedded narrative, which is only one of many ways to embed a narrative in the larger plot. I revisit the construction briefly toward the end of this section. [57]

§22. The transition is more often accompanied by one or more particles, instead of or in addition to the use of ποτε. Very common in Homer is the combination of a pronoun followed by a particle (for which see II.5), but here I focus on the beginnings of embedded narratives marked by (combinations of) particles, sometimes along with other words. One of the particles most at home in the beginning of embedded narratives is γάρ. [58] Consider this passage from Agamemnon’s speech to Diomedes in book 4:

(t6)

Iliad 4.375-377

375    (…) And they say he was beyond others.
          Oh yes, for [59] he came to Mycene in peace
          as a guest, with godlike Polyneices, to collect an army.

Agamemnon is talking to Diomedes about the latter’s father Tydeus, in an attempt to get him back into the fight. Although he has never met Tydeus himself, Agamemnon has heard good things about him (line 375), which leads him to narrate the story about Tydeus’ visit to Mycene (lines 376-398). Traditionally, γάρ is explained in this and other instances as providing the justification for what precedes, which would fit into its generally understood main “causal” or “explanatory” function. [60] Consider this relatively recent description by Sicking of γάρ in Lysias: “The purpose of sentences introduced by γάρ is primarily explanatory: they provide answers to all sorts of questions raised by the speaker’s utterances.” [61]

§23. For ancient Greek in general, Slings argues that “the most typical PUSH particle is γάρ,” which is to say that it is the particle most typically used to mark the displacement to a new “frame of reference.” [62] The PUSH is answered by a POP, which marks the return to the main line of narrative or argumentation. [63] De Jong has approached the use of γάρ in Homeric embedded narratives from a different angle, that of epic regression. [64] It appears to be a typical tool of the epic performer to work his way back from the outcome to the beginning of a story, at which point he tells it from beginning to end in more detail: D-C-B-A-A’-B’-C’-D’ in De Jong’s notation. [65] Since γάρ is the particle used to provide the cause or justification of something just mentioned, it fits quite naturally at the beginning of the receding steps C-B-A: D happened because C happened, C happened because B happened, and so on. Over time, a variant on this pattern emerged, D-A-B-C-D’, where the narrator tells the outcome and then skips immediately to the start of a story, which is thus as a whole introduced by γάρ. [66]

§24. Both Slings and De Jong’s approach to γάρ at the beginning of embedded narratives start from the idea that the essence of γάρ, and its basic function, is to provide explanation or justification. However, scholarship on γάρ is divided on the issue of what the original force of the particle was and – by extension – how that original force led to its different uses in later literature. [67] The communis opinio emerging from the end of the nineteenth century, and set in stone by Denniston, is that the causal function of γάρ is a development from an earlier affirmative function, which reflects the particle’s origins in the combination γε + ἄρ. [68] Denniston, among others, argued that this force had been lost by the time of the earliest extant Greek literature, except in combinations. [69]

§25. One view that recurs throughout scholarhip is that γάρ is relevant to starting new parts of discourse. [70] Hoogeveen initially called it “inchoative,” while in later scholarship it is more generally called asseverative. [71] The focus has been on γάρ in embedded narratives, as in the studies by Slings and De Jong discussed above, and parentheticals. [72] Although asseverative and inchoative are neutral adjectives, since they denote nothing more than a discontinuity, an assumption underlies many of these studies that γάρ marks its host act or move as in some way subordinate to its co-text. [73]

§26. The recent scholarship on γάρ, then, is characterized by these two tendencies, first the tendency to attribute some causal or explanatory value to all instances of γάρ, and second to regard the discourse introduced by γάρ as new, but also somehow backgrounded or subsidiary to the text that precedes it. [74] Both of these ideas may profit from a re-examination of γάρ from the perspective of the division of discourse into acts and moves. Like most particles – and like “because” or “for” in English – γάρ can mark relations both between acts and between moves. [75] Moreover, it can mark relations on a propositional or on a metalinguistic level. When γάρ is used as metalanguage it will most often function on the level of the move, but this is only a tendency, not a rule.

§27. This distinction is relevant to the occurrence of γάρ at the beginning of embedded narratives, since γάρ in those instances should be regarded as metanarrative. The particle says something about the direction of the upcoming discourse, not so much about the content of the upcoming sentence. Slings implies as much in his discussion of γάρ as a PUSH particle, but Bakker’s dicussion of γάρ in terms of the “flow of speech” or “movement in speech” is clearer. [76] Bakker argues that the particle in Homer marks necessary steps in the flow of discourse, to develop something mentioned earlier; [77] this development has nothing to do with foreground or background. [78] Ηe thus redirects attention to another aspect of the use of γάρ in Homer that vindicates the idea of an “explicative” function, but with a more literal understanding of its root “explic-”: “unfold”.

§28. Bakker’s terminology of “developing” captures well what happens in the ongoing epic discourse. Every embedded narrative in Homer represents a choice, since with every character or event he refers to, the epic performer has the freedom to move on or to expand. [79] γάρ at the beginning of embedded narratives marks sections that are associative unfoldings of the collective memory that is the realm of the Homeric performer. [80] It represents the link that the performer perceives between the preceding and subsequent discourse, even if on the surface the two may seem unconnected. γάρ marks the activation by the performer of a narrative that relates to the ongoing narrative in some way: by enriching the story, reflecting upon it, or putting it in a different light. Sometimes the information provided in the γάρ move proves to be nonessential to the narrative, but more often it is indispensable to the development of the ongoing action. In (t6) above, for example, the story invoked by the performer, in the voice of Agamemnon, is a crucial part of Agamemnon’s rhetorical strategy in his effort to persuade Diomedes. [81]

§29. In these contexts, it is thus best not to regard γάρ as explanatory, (“[I say this] because…”), [82] but as a marker of the cognitive act of association. [83] This association, moreover, occurs not on a microlevel but on a macrolevel: it is often not the sentence containing γάρ itself that is particularly relevant to the preceding, but the whole move that follows. That is to say, γάρ at the beginning of embedded-narrative moves marks the theme or idea represented by the move as relevant to the ongoing discourse. [84] The linguistic form of this move may be of any size, but it is more likely to be a multi-act unit than not. [85] A look through Homer and through Homeric scholarship shows that γάρ often occurs at the beginning of embedded narratives, so in the following discussion I focus more specifically on καὶ γάρ and ἤδη (γάρ) in Homer. In Pindaric song the range of functions of γάρ is the same as in Homer; in chapter II.4 I explore one aspect of the particle’s use by Pindar further. [86]

3.2.2.1 καὶ γάρ

§30. About καὶ γάρ De Jong says the following, in a note on Iliad 22.46: “καὶ γάρ typically signals the introduction of an example which must back up a general claim.” [87] More recently, Aftosmis has made the same claim for καὶ γάρ in Pindar. [88] De Jong’s reading is appropriate for Iliad 22.46, and there are a number of parallels, [89] but the characterization of καὶ γάρ as introducing examples or paradeigmata oversimplifies the combination’s workings. This should not be a surprise, since καὶ γάρ is a combination of two particles and thus also a collocation of two spectra of functions, which would make it unlikely for the combination to have only a limited function. [90] In the forty-five instances of καὶ γάρ in Homer, we find combinations of the whole range of functions of both particles. De Jong chooses to focus on καὶ γάρ used to introduce exempla, but her claim that this is “typical” is unfounded. I am more interested in the combination’s use in introducing associative narratives, a group of twelve instances that overlap in part with De Jong’s paradigmatic narratives. [91] The following example of καὶ γάρ from the last book of the Iliad represents De Jong’s idea of the introduction of an example, while it takes the form of an associative narrative: [92]

(t7)

(…) νῦν δὲ μνησώμεθα δόρπου. |
καὶ γάρ τ’ [93] ἠΰκομος Νιόβη ἐμνήσατο σίτου, |
τῇ περ δώδεκα παῖδες ἐνὶ μεγάροισιν ὄλοντο |

Iliad 24.601-603

(…) Now let us remember the meal.
After all even pretty-haired Niobe thought of food,
though for her twelve children died in the halls

As in (t6), the narrative serves as a persuasive device, and again it is triggered by association in the performer’s mind. The example of Niobe’s story is triggered by the parallel with Priam: just as she ate despite her grief, so Priam should too. It is the association between the two episodes that explains why we find γάρ. [94] In this instance, there is no doubt that the story introduced by γάρ represents a paradigmatic narrative meant to persuade the interlocutors to have dinner. As such, it is a particular example that follows after a general claim. However, association is the key element in the relation between the preceding and upcoming move signalled by γάρ. As a result, the new move can also be a general statement following upon a particular one; the inverse of an example backing up a general claim. [95]

§31. The apparently specialized function of the combination καὶ γάρ in these instances to introduce associative narratives in fact follows naturally from the combination of καί and γάρ. The twelve instances where καὶ γάρ introduces an associative narrative involve three different funtions of καί, while the function of γάρ remains constant. The relevant functions of καί, discussed at length in IV.2, are the scalar function, the function of pinning down, and the more common function of marking similarity. In the little narratives, the inset story features either better men, women, or gods doing something that should enlighten the course of action in the current situation, [96] or something that happened to one person or group in particular, [97] or something that was also once the case for someone else. [98]

§32. The scalar force of καὶ X (“even X…”) is especially suited to introducing a paradigmatic story, since if even in more dire situations better men did something, how could we not follow their example? [99] The second group involves the pinning-down function of καί, where one person or group is singled out from an earlier collective. [100] Finally the last four instances, all in the Odyssey, show perhaps the most familiar face of καί marking a perceived parallel, to be translated as “too” or “also.” [101] The function of γάρ in all these instances does not fluctuate as that of καί does, but in all cases it marks an association between the new move and the ongoing narrative. Because of the differences among even these few cases, it is unproductive to think of καὶ γάρ as working in a cluster with one specific function in Homer. In Homer the two particles work separately, combining different functions of καί with γάρ to introduce different kinds of moves. The new move may be a paradigmatic narrative, as in the case of the Niobe story, but it may also be associated to the surrounding discourse in other ways, depending on the function of καί.

3.2.3 ἤδη and ἦ marking beginnings

§33. The function of γάρ to introduce associative narratives illustrates the particle’s importance for the process of producing, unfolding the narrative. Not surprisingly, then, γάρ can introduce narratives in other combinations than only with καί, as in the following passage from Iliad book 3, where Antenor describes Odysseus to Priam. We are on the walls of Troy, and Helen is pointing out the best of the Achaeans at Priam’s request. After her brief introduction of Odysseus, Antenor pitches in with an associative narrative to illustrate where Odysseus’ real talents lie:

(t8)

          Τὴν δ’ αὖτ’ Ἀντήνωρ πεπνυμένος ἀντίον ηὔδα·
          “ὦ γύναι | ἦ μάλα τοῦτο ἔπος νημερτὲς ἔειπες· |
205    ἤδη γὰρ καὶ δεῦρό ποτ’ [102] ἤλυθε | δῖος Ὀδυσσεὺς” |

Iliad 3.203-205

          Her in turn did Antenor the prudent respond:
          “Milady, really you told that story truthfully. [103]
205    For already here too once came noble Odysseus.” [104]

In a valiant attempt to do justice to the string ἤδη γὰρ καὶ δεῦρό ποτ’, Murray translates line 204 as: “for (γάρ) once (ποτ’) before (ἤδη) also (καί) noble Odysseus came here (δεῦρο).” [105] This translation sounds forced because, among other reasons, it stretches the meaning of ἤδη. The presence of ποτε (“once”) means that the most common sense of the temporal adverb, “already”, would be superfluous. Murray here translates it to mean “at some point before the present,” when elsewhere in Homer it always bears a relation to the present, so “now” or “already” in the sense of “by now.” [106] Adverbs like ἤδη are called “mobile,” since they can occur in any position in an act. Thus we find it at the beginning (Odyssey 2.89 ἤδη γὰρ τρίτον ἐστὶν ἔτος, “it is already the third year”), in the middle (Iliad 7.293 νὺξ δ’ ἤδη τελέθει “and night is already here”), or at the end (Odyssey 1.303 αὐτὰρ ἐγὼν ἐπὶ νῆα θοὴν κατελεύσομαι ἤδη “and I will go to my swift ship now”). [107] In (t8) and a number of parallels, however, I believe that we are not looking at ἤδη at the beginning of an embedded narrative, but at ἦ δή. In the following section, I argue that a number of instances of ἤδη γάρ in the Iliad may in fact represent ἦ δὴ γάρ. To back up that argument, I discuss the relevant passages in relation to the instances where our manuscripts do read ἦ δή. Finally, I offer a possible explanation of ἦ in these and similar contexts. Since the relevant instances of ἤδη γάρ are limited to the Iliad, in this section there are no examples from the Odyssey.

§34. Although the manuscripts – especially Venetus A and B – offer valuable readings, strings of letters like ΗΔΗ are often ambiguous, and the choices made by the medieval scribes and more recently by editors are to some extent arbitrary. [108] When found at act beginning, ΗΔΗ may equally resolve as either ἤδη or ἦ δή. [109] The combination ἦ δή is restricted to act-initial position, since ἦ is initial in Homer. The use of ἤδη throughout the Homeric epics shows beyond doubt that it can be a mobile adverb with the meaning of “already” or “(by) now.” However, it is not to be taken for granted that whenever our manuscripts have ἤδη it is the adverb that we should read. The nature of the textual transmission of the Homeric corpus means that it is possible that ΗΔΗ (ΓΑΡ) at verse beginning was simply rendered homogeneously as ἤδη γάρ despite the fact that in some cases it stood for ἦ δὴ γάρ. [110]

§35. Now to return to (t8) above, I propose the following reading: ἦ | δὴ γάρ καὶ δεῦρό ποτ’ ἤλυθε | δῖος Ὀδυσσεὺς | (“Oh yes (ἦ), for actually (δὴ γάρ) he came right here [111] once, god-like Odysseus”). This reading of ἦ as a separate act with interjectional value and a distinct prosodic contour marks a strong discontinuity in the discourse, here the start of an embedded narrative. This use of ἦ at move beginning has many parallels in Homer. We find this asseverative force of ἦ in the string ἦ δή at the start of direct speech (see below), in ἤτοι (or ἦ τοι), [112] ἤδη, the swearing formula ἦ μήν, and possibly even the apparent verb form ἦ (“he spoke”, from ἦμι) in Homeric discourse. [113] Moreover, the string δὴ γάρ at act beginning is uniquely Homeric. In Homer we find a limited number of constructions that allow δή at act beginning: δὴ γάρ (never at verse beginning), δὴ τότε, δή ῥα τότε (both always at verse beginning). In four other separate instances the position of δή is debatable. [114] In later Greek δή becomes mostly restricted to peninitial position in the act. [115] The fact that δὴ γάρ is never found at verse beginning [116] may be another reason why we find the string ἦ δὴ γάρ. [117]

§36. Compare the following introduction of another associative narrative in Iliad 1: [118]

(t9)

          (…) ἀργαλέος γὰρ Ὀλύμπιος ἀντιφέρεσθαι· |
590    ἤδη γάρ με καὶ ἄλλοτ’ | ἀλεξέμεναι μεμαῶτα |
          ῥῖψε ποδὸς τεταγὼν ἀπὸ βηλοῦ θεσπεσίοιο, |

Iliad 1.589-591

          (…) Because the Olympian is dangerous to oppose.
590    For already also another time, as I was trying to save you,
          he threw me, having seized my foot, from the divine threshold.

Line 589 may be read alternatively, with my emendation, as follows: ἦ | δὴ γάρ με (“Yes, for actually (καί) another time, as I was trying to save you, he threw me…”). If we want to read ἤδη with the manuscripts, the same temporal problem holds as in (t8) above. The event has direct relevance to the present, but if ἤδη indeed is a temporal marker here, its function is not to emphasize that it has “already happened,” but that it “happened at some point before the present” which is in fact fully expressed by ἄλλοτε, as it is by ποτε in (t8).

§37. These two examples may suggest that we should always translate ἦ as “(Oh) yes…” but the exact translation is not the point. [119] The force of ἦ is connected to its prosodic contour, and carries no fixed semantic load. Compare this example from Iliad book 14:

(t10)

Ιliad 14.247-249

To Zeus, I would not go any closer to Cronus’ son,
nor would I lull him into sleep, if it were not he himself that ordered me.
For already also another time your command taught me,

If we read ἦ | δὴ γάρ με here, the thought expressed is perhaps best rendered in English as “no” (“No, for actually also another time your command taught me”). A little narrative of that earlier event then follows.

§38. All of the examples above occur in direct speech, where ἦ might at the same time be regarded as a sign of the character’s involvement. [121] In fact, where the manuscripts consistently give ἦ δή rather than ἤδη, the particles always occur at the start of direct speech. [122] These passages provide another argument to read ἦ | δὴ γάρ rather than ἤδη γάρ in the examples discussed above and their parallels. Three instances are especially relevant: Iliad 15.467, 17.538, and 21.583. For Iliad 15.467 the Venetus A manuscript offers an informative reading: “ὦ πόποι· ἦ, δὴ πάγχυ…” [123] The comma after ἦ suggests a pause after this single syllable, which matches my reading of the instances of ἤδη γάρ under discussion, listed in note 110. [124] The other two relevant instances of ἦ δή at the start of direct speech, Iliad 17.538 and 21.583, are similar to ἤδη γάρ in that another second-position word follows δή: it is followed by μάν in Iliad 17.538 and που in Iliad 21.583. Whereas που, being an enclitic, is often pushed back, [125] μάν is accented and is never pushed from its second position by another second-position word in Homer, [126] except in this passage, Iliad 17.538. [127] This exception suggest that here μάν is in fact in its normal second position, and that ἦ, as in the beginnings of embedded narratives in ἤδη γάρ (or ἦ δὴ γάρ), forms its own discourse act. [128] Seeing Hector and Aeneas flee before the Aiantes, Patroclus’ charioteer Automedon rejoices for having killed Aretus:

(t11)

Iliad 17.537-539

He stripped him of his armor he spoke this word in prayer:
“Yes! Surely at least a little, now that Menoetius’ son died,
I have eased my heart, having killed only a lesser one.”

As a marker of discursive discontinuity, ἦ is not limited to introducing embedded narratives, which it does in the cluster ἦ | δὴ γάρ. In (t11) ἦ starts direct speech (with numerous parallels in note 123) and it regularly occurs at the start of oaths (in the cluster ἦ μήν). In both cases one can imagine the effectiveness of a prosodic interruption in the form of ἦ to mark the transition to a new kind of discourse, a new move. [129]

§39. This correlation between the occurrence of ἦ and the start of a new move supports several scholars’ characterization of ἦ as asseverative. [130] More often, however, descriptions of ἦ focus on the function of affirming what the speaker is saying, as a marker of (emotional) involvement. [131] Whether it is actually used in dialogue, as in drama, or not, its force is generally understood as dialogic in nature. For the passages cited above, we have translated ἦ as a reaction to an implicit question, to better render the discontinuity in discourse. [132] It marks the speaker anticipating a likely question from his audience, and answering it with the narrative (or other kind of move) that is relevant to the imagined question. At the start of direct speech, ἦ rather marks a reaction to something that just happened, as Automedon reacts to what he has done in (t11).

§40. The majority of approaches to ἦ have focused on its possible force and function on the sentence level, with little attention to its place within the larger discourse. Despite etymologists’ observations that the particle may derive from an interjection, few scholars have regarded ἦ from this angle. [133] Primary interjections are linguistic renderings of often non-lexical exclamations, and as a result their force lies less in their semantic content (if there is any) [134] and much more in their prosodic contour. [135] Consider the following definition of an interjection by Ameka: “Primary interjections are little words or non-words which (…) can constitute an utterance by themselves. (…) They could be uttered as co-utterances with other units.” [136] And this same author adds that interjections are “always separated by a pause from the other utterances with which they may co-occur.” [137]

§41. The pronunciation of ἦ appears to have been a continuum ranging from a full interjection, [138] reflecting its possible origin, to the disjunctive particle ἤ, which has a rather narrow and clear function. This continuum might be regarded as the synchronic reflection of a diachronic evolution. [139] Likewise, the Homeric corpus shows traces of δή as a mobile adverb but also as a particle restricted to second position, while the combination ἦ + δή = ἤδη became a mobile adverb, possibly under the influence of δή. [140] If ἦ was indeed originally an interjection, it would have constituted a discourse act by itself, as a vocative does, for example. [141] ἦ᾽s origin as an interjection would explain why it developed into a particle restricted to act-initial position. An independent interjection would always be used before (or after) another act, and if it were to be assimilated into one of the surrounding acts, the interjection would most naturally come at the end of the preceding act or the beginning of the following. Since interjections in Greek are often turn-initial, [142] it makes sense that ἦ would become restricted to act-initial position. [143]

§42. Because of its initial position, ἦ will have become collocated with peninitial words, often other particles. The resulting collocations then specialized into combinations with specific functions (such as ἦ μέν/μήν/μάν to introduce oaths, and ἦ γάρ as a tag question requesting confirmation). The existence of these specializations, however, should not blind us to the original force of the component parts, for two reasons. First, the specialized combination could only come to work as it does because of the function of the particles that originally formed it; a cluster does not form arbitrarily. Second, the fact that a certain cluster specializes into a specific function does not mean that every instance of the combination must have this specialized function. [144] For example, ἦ μέν does not always have to introduce an oath: it may also occur where ἦ is used to mark another kind of new move and μέν marks its host act as the beginning of a larger part of discourse. Likewise ΗΔΗ is not always ἤδη, ΗΤΟΙ can be both ἦ τοι and ἤτοι, and so on. [145]

§43. It is important to remember that even the consensus readings of Homeric manuscripts, scholia, and editions are fallible. By exercising due skepticism, we may identify phenomena that orthography has obscured. As becomes clear only when one takes into account the larger discourse, the instances of ἦ (δή) above form a separate group that shows a specific (perhaps older) function of ἦ. On its own and in combination with other particles in Homer, ἦ may serve as a prosodic marker of the beginning of a new move, a paralinguistic means to signal discontinuities in the discourse. [146] However, this does not mean that it does not function at the same time as a marker of speaker involvement, as it has been described in most of the recent publications. [147] The particle can mark a discontinuity or salient moment in the interaction, and at the same time it may reflect the emotional engagement of the speaker. [148]

§44. For different reasons, then, both γάρ and ἦ occur at the beginning of embedded narratives or other moves. The force of γάρ is too often regarded as clearly causal or explanatory, but the causal function does not best describe its force when used in epic or lyric. As for ἦ, I have offered an analysis that considers its function in the flow of discourse, taking into account its possible origin. Whereas the use of γάρ must perhaps be seen as a result of the cognitive activities of composer or performer, the use of ἦ is a direct reflection of an interactional situation, either real or imagined.

3.2.4 Other narrative beginnings

§45. Embedded narratives in epic represent unfoldings on the path through the Homeric storyworld. Beside ἦ and γάρ there are multiple other linguistic means to start embedded narratives, such as adverbs of time and place. Certain locales in the storyworld bring with them longer or shorter stories. These can be told at any point when a place is mentioned in the narrative, and the stories can therefore be activated with locative adverbs. Consider the following excerpt from one of Odysseus’ stories:

(t12)

Θρινακίην δ’ ἐς νῆσον ἀφίξεαι· | ἔνθα δὲ πολλαὶ
βόσκοντ’ Ἠελίοιο βόες | καὶ ἴφια μῆλα.

Odyssey 12.127-128

And you will come to the island of Thrinacia. There in numbers
graze the cattle of Helios and his strong flocks.

ἔνθα is here clearly used in a spatial sense, its core value. In that capacity it serves well to initiate a little narrative associated with a place. The word, however, is used more broadly in Homeric epic—in fact, the majority of instances in Homer marks the start of a new move within a narrative, generally translated with “then.” [149] ἔνθα thus straddles a fuzzy semantic border where it can mean either (or both) “then” and “there”. When ἔνθα means “there” it refers to a geographical place in the storyworld, but when it is most naturally translated as “then,” it may well be a marker of a certain “place” on the unfolding path of the narrative. [150]

§46. Another marker often occurring in move-initial position is ὡς/ὥς. Like ἔνθα, this word is multi-faceted: it can mean “just like…” or the corresponding “that’s how,” typical of the Homeric simile, but it can also start a “when” clause, often preposed to the initial action of a new scene. The function of such constructions in providing a static setting before the dynamic action makes complete sense with regard to the build-up of narratives in general. [151] As such, it is important to consider their discursive or narrative function, and unproductive to focus on their syntactical subordinate status. [152]

§47. There are, then, several ways to mark narrative beginnings linguistically. My analysis of γάρ, ἦ, and other words demonstrates that it is important to be aware of such macro-discursive divisions when looking at particles. Only then can they be seen as what they are: reflections of an ongoing interactive exchange. Particles mark moments in the act of narration rather than syntactical relations in a text, and as such provide invaluable insight into the interaction between performer and audience.

§48. On occasion, however, there is no explicit linguistic marking of important transitions in the discourse. Since we must start from the assumption that the performer would not have wanted to confuse the audience, a lack of linguistic signs would necessitate some kind of non-linguistic or paralinguistic marking. In Pindar, every new strophe or antistrophe is a new beginning, and the performative discontinuity is even stronger with the start of a new epode. Since the Homeric epics were not generally performed in one go, or not by a single performer, [153] we may assume that they too offered possibilities to break off and start again and thus creating very clear boundaries of which very little trace remains in the text.

§49. Whether or not we may find traces of such heavy performative discontinuities in our texts, on a local level there are passages where a strong discontinuity on the level of content or orientation remains without syndetic marking. In the famous battlefield encounter between Glaucus and Diomedes in book six of the Iliad, Glaucus embarks on a long story about the fame of his forefathers at line 152:

(t13)

ἔστι πόλις Ἐφύρη | μυχῷ Ἄργεος ἱπποβότοιο, |
ἔνθα δὲ Σίσυφος ἔσκεν, | ὃ κέρδιστος γένετ’ ἀνδρῶν, |

Iliad 6.152-153

There is a city [154] Ephyre, in a corner of Argos grazed by horses.
And there Sisyphus lived, who was the craftiest of men.

With the story Glaucus starts a new move, but contrary to expectation we do not find any syndetic marking at its beginning. This start of a new move in asyndeton is informative. The lack of metalinguistic marking suggests that the discontinuity must have been marked by the performer in some non-linguistic or paralinguistic manner. [155] This is not to say that the presence of a metalinguistic marker voids the possibility of extralinguistic or paralinguistic marking. The example shows that there are multiple strategies to negotiate discourse transitions, involving metalinguistic marking to a greater or lesser extent, and always possibly marked otherwise in performance.

§50. Another kind of asyndetic start of narratives is represented by the example below. It consists in a relative or demonstrative pronoun in an oblique case, followed by a little narrative about its referent.

(t14)

(…) γλαυκόχροα κόσμον ἐλαίας, | τάν ποτε
Ἴστρου ἀπὸ σκιαρᾶν παγᾶν ἔνεικεν Ἀμφιτρυωνιάδας |

Pindar Olympian 3.13-14

(…) the graycolored adornment of olive, which once
from the shady springs of the Ister Amphitryon’s son brought, [156]

Here we see a narrative of the kind described by Slater and West as a “lyric narrative”, introduced by a pronoun, ποτε, and an aorist. [157] The transition to the narrative falls on a metrical boundary in the song, which would allow for some prosodic marking on τάν ποτε to signal that it is the start of a new move. [158] For the present study it suffices to note that asyndeton, the start of a new sentence without the use of a conjunction, conjunctive adverb, or particle, often occurs at the beginning of new moves. For Homer asyndeta occur mostly at the start of embedded narratives or subsections of narratives, while in Pindar asyndeton can occur at the start of an embedded narrative, at narrative transitions, introducing a gnomic statement, or at the transition to the hic et nunc. [159] The lack of any metalinguistic marking of so strong a discursive transition suggests some kind of prosodic discontinuity such as an extended pause. [160]

3.3 Move transitions in Homeric narrative

§51. The preceding section concerned itself with the range of possible methods of marking embedded narratives linguistically. The next step is to discuss the transition between different components of a certain narrative, taking the Labovian division (§14) into consideration. Let us return to Antenor’s story about Odysseus (t8). After making some initial remarks about Odysseus’ earlier visit at Helen’s summons, Antenor continues:

(t15)

Iliad 3.207-224

The progress of Antenor’s story exemplifies the way Homeric epic habitually signals each narrative step with a temporal marker combined with particles. At the same time, this excerpt is special in that it has the same combination four times within thirteen lines: ἀλλ᾽ ὅτε δή (209, 212, 216, 221). [163] Despite the high frequency of the combination in the Homeric corpus, this four-fold clustering is limited to only three places. [164] The effect in this passage, as in the two parallel passages, is one of crescendo. The audience is kept in suspense as Antenor works toward the climax of his narrative: Odysseus was older than Menelaus, and the latter spoke briefly but very well, and Odysseus did not look impressive; only when he started speaking could one see his true character.

§52. The quadruple ὅτε is answered by ἔπειτα, introducing the climax of the story: “when…and when…and when…and when…then…” [165] In the Labovian model, line 223 is the resolution, the final line of the story. The fact that ἔπειτα is followed by a second temporal marker in 224 (τότε) is marked, suggesting a special addition. This interpretation is supported by the switch to the first person plural (ἀγασσάμεθ’). The last line places the point of view firmly with Antenor (and perhaps the other Trojans), giving it a more explicitly evaluative character. Antenor’s personal involvement in these lines evinces itself in another, more subtle manner. The final two lines contain the only two instances of γε in the entire speech, directing attention first to Odysseus himself, and then to τότε, “at that very moment.” The sudden occurrence of γε here illustrates our belief that a higher frequency of γε may correlate with more personal involvement of the speaker, such as in emotional passages. [166] Not surprisingly, in the Homeric corpus γε occurs more than twice as much in direct speech as in narrator text. [167]

3.3.1 Homeric δή I: marking narrative steps

§53. Using temporal markers is a very natural way of navigating the narrative, especially in the complicating action. In Homer, however, the temporal marker does not typically occur on its own; among other particles, δή tends to gravitate toward such markers in Homer. [168] This tendency is well-known to readers of Homer, and it has led scholars to posit that the original value of δή was temporal. [169] Given its formal likeness to ἤδη, this theory is intuitively attractive. As a result, however, these scholars have attributed a temporal value to all instances of δή in Homer, whether combined with temporal markers or not. In opposition to this approach, there is a theory that δή is in fact etymologically connected to δήλον (clear, visible), [170] which would better explain its generically emphatic function in koinḗ Greek. [171]

§54. However, attempting to establish a general value of δή in Homer passes over an important particularity in the distribution of δή in the corpus. If the value of δή were indeed evidential (or more generally emphatic), one would expect it to occur more frequently in direct speech than in narrative (which it does, by a factor of almost 3 to 1 in the Iliad, and less markedly by 1.2 to 1 in the Odyssey), [172] but one would not expect the use of the particle to be qualitatively different between direct speech and narrative. The same presupposition would hold if δή were to be read everywhere as having some temporal value. Neither of these explanations of the particle can account for the remarkable pattern emergent from the Homeric material, which begs closer analysis. In the entire text, around half of the instances of δή coincide with a temporal marker (-οτε, ἔπει-, νῦν, τῆμος, ἔνθα), [173] but in narrator text the average increases to 75-85%. [174] As a result, the relative frequency of δή with temporal markers in direct speech is quite a bit lower, and in direct speech the particle in fact occurs in a much larger spectrum of cotexts. [175]

§55. These numbers urge us to consider that δή in combination with temporal markers is inherently different from other collocations, which allows it to be used freely in narrator text. In what follows, I present a closer analysis of two groups of instances of δή, which provide a basis for more nuanced descriptions of the particle’s function in epic and lyric: (1) δή in peninitial position with a temporal marker, (2) instances of δή in initial or otherwise marked position. The second category also provides material to understand δή in peninitial position without a temporal marker. Since δή is very rare in Pindar (twenty instances in the Odes), and use of the particle in Pindar matches that in Homer, the discussion below concerns itself only with Homer; I cite Pindaric parallels for each kind in footnotes. [176]

§56. In its most common use, in peninitial position in combination with a temporal marker, δή articulates the progression of discourse in larger steps. For example in the combination ἀλλ᾽ ὅτε δή, where δή is in its expected peninitial position, [177] the particle signals a narrative boundary on the level of the move. The act introduced by δή is often not an especially salient new step in the narrative, nor does the act require any kind of intensification. Likewise, the temporal adverb it follows in such contexts does not appear to need particular stress. Consider the following excerpt from (t16):

(t16)

Iliad 3.207-209

Them I welcomed and hosted in my halls.
Of both of them I learned the nature, and the cunning tricks.
Now when [178] among the assembled Trojans they mingled…

After the introductory two lines (207-208), ἀλλ᾽ ὅτε δή marks the progression to the complicating action. The move introduced by the combination is not a narrative peak, nor does it in any other way seem to be emphasized. That is to say, it does not seem to be the case that the passage reads with small-scope δή as “very much at that moment” (i.e. “just when”). [179] Its function, which is perhaps impossible to render into English, is to mark in concert with ἀλλά that there is some kind of narrative discontinuity, which coincides with the start of a new move. [180] The function of δή along with a temporal marker to mark new moves in the narrative correlates with the findings of Bestgen and Vonk. [181] In a study of the effect of the use of temporal markers versus “and” at narrative transitions, they found that temporal markers (segmentation markers in their terms) reduced the availability of words in the preceding discourse. This suggests that readers regard those temporal markers as some kind of new beginning. In Greek, the temporal adverb marks the progression, while δή marks the discontinuity. [182] It is no coincidence that in this function δή often occurs in a subordinate clause, which syntactically projects an answering main clause, and thus a longer piece of discourse. There is a productive analogy for this effect of a subordinate clause in recent research on aspect in English. In an experiment, readers were asked to predict what was to follow after a clause with imperfective or perfective aspect: “The diver was snorkeling…” versus “The diver had snorkeled…”. The latter of the two led to the inference that the details of the snorkeling event are less relevant for the ongoing interpretation than whatever follows in the discourse. [183] Likewise, a when-clause of the form ἀλλ᾽ ὅτε δή suggests that the information in the subsequent act will be more salient than that of the when-clause itself. [184] This formulation makes it less likely that δή here functions to intensify its host act.

§57. For this function of δή it is productive to retrieve the possible etymological relation between δέ and δή. [185] Most scholars believe that δέ and δή are to be regarded as formally related, regarding δέ as a bleached development from δή. Alternatively, I propose the possibility that δή is a lengthened form of δέ, analogous to the development of ἆρα (with long alpha) from ἄρα. This morphological relation is then, for one function of δή, mirrorred in a functional relation: δέ marks boundaries between discourse acts, while δή divides the discourse into larger steps, consistently occurring at the start of new moves. [186]

§58. There is also a group of instances where δή occurs with temporal markers, in narrator text, but in initial position: δὴ τότε and δή ῥα τότε. The act-initial position of δή as well as its position in the larger discourse suggests that the particle has a different function in this construction. A different interpretation of ὅτε δή on the one hand and δὴ τότε on the other is moreover suggested by the fact that the former introduces a subordinate clause, whereas the latter introduces a main clause. Consider the following passage from the Odyssey:

(t17)

Odyssey 12.401-406

          And we, getting on board quickly, set out on the wide sea,
          setting up the mast and hoisting up the white sail.
But when we left the island behind, and no other part
          appeared of the lands, but only sky and sea,
405    right at that moment Cronos’ son raised a black cloud
          over the hollow ship, and the sea grew dark under it.

In line 403, ἀλλ᾽ ὅτε δή introduces a new step in the narrative, but the crucial event occurs with δὴ τότε. Whereas δή in peninitial position merely marks progression of the narrative, in initial position it marks a salient moment. Here and in a number of parallels, δὴ τότε introduces a peak in the narrative, the moment that has been worked up to until then. [187] The scope of δή in such instances is debatable: it either intensifies τότε or the entire act. The translation shows that I take it as intensifying τότε, but δὴ τότε together introduces the entire move: the intensified temporal adverb functions as what we call a “peak marker.” [188] Compare the same narrative moment in this passage from the Iliad:

(t18)

Iliad 6.174-176

For nine days he hosted him, and nine cows he sacrificed.
175 But when the tenth rosy-fingered Dawn appeared,
right then he questioned him and asked to see the token,

The function of καί to introduce a peak or climax is discussed elsewhere, and the combinations δὴ τότε and καὶ τότε have the same function in Homer: to introduce a narrative peak. [189] In the combination καὶ τότε δή, finally, it is hard to establish whether δή marks a move boundary or intensifies the act, along with καί. The construction may serve as a reminder that boundaries are fuzzy, and that the following conclusions are based on patterns we perceive. Since in the end δή is just one word, its multiple functions may have blended into one another, and more than one of the particle’s aspects may be relevant in a single instance.

3.3.2 Homeric δή II: intensifying constituents or acts

§59. Particle scholarship has disregarded the discourse-articulating function of δή outlined in the previous section, and instead focused on its emphatic function. More recently, Bakker has suggested that δή “is better characterized as a marker of evidentiality”: [190] “[t]he use of this particle draws the hearer into the story by marking the narration as deriving from a shared basis, a common experience that binds the narrator and listeners together as if they were actually jointly witnessing a given scene.” [191] Bakker’s reading of δή in Homer is part of his argument that Homeric language suggests that the performance was a process of bringing the mountain to Mohammed, bringing the narrative into the here and now instead of displacing the audience into the past. [192] He sees a trace of this same function in the omnipresent Homeric δέ, which he regards as a bleached version of δή: “[i]f dé is a weak form of dḗ, then its meaning is similar but weaker.” [193]

§60. As argued above, I do not think that emphasis or evidentiality is necessarily to be sought in those constructions where δή occurs in peninitial position with a temporal marker. Now, in the Homeric corpus δή is not always a peninitial particle, but is also allowed in act-initial position. The instances of δή in this position – limited to a few constructions: δὴ τότε, δή ῥα τότε, δὴ γάρ, and three isolated examples [194] – reveal a different function, closer to the “emphatic” or “evidential” function posited by earlier scholarship. However, a close analysis of δή in apparently divergent position reveals a more detailed picture. Consider first the following two instances:

(t19)

τέκνον ἐμόν | δὴ πάμπαν ἀποίχεαι ἀνδρὸς ἑῆος

Iliad 19.342 (direct speech)

My child, utterly you forsake your own warrior

(t20)

Ἀντίλοχος δ’ ἄρα δὴ λοισθήϊον ἔκφερ’ ἄεθλον [195]

Iliad 23.785 (narrator text)

Antilochus, then, carried off the very last prize.

The instance from direct speech (t19) reveals the act-initial position of δή because it follows a vocative. The instance in Iliad 23 is less straightforward, but I argue that δή is a mobile here, with scope over the following word. [196]

§61. In (t19) δή is arguably act-initial modifying the following word: πάμπαν. [197] The combination of δή and comparatives or superlatives, including “first,” “last,” “all,” and “many,” is widespread enough for me to consider it as a special construction with the particle, where δή functions as an intensifier of an adjective at one end of a scale. If we then consider (t21), δή seems to modify λοισθήϊον (“the very last”), and therefore occurs just before the adjective, in act-medial position. [198] As far as position is concerned, (t19) and (t20) show that δή in its intensifying function is moderately mobile. The Homeric corpus suggests that δή in some functions can precede its host adjective or adverb, which allows it to occur anywhere from act-initial to act-medial position. [199] Already in Homer δή does not always precede the constituent it modifies, like in (t19) and (t20), but can also follow it. [200] Because of this mobility, the collocation of δή directly in front of polar adjectives (adjectives near either end of a scale, e.g. first and last) also occurs with δή in its typical peninitial position. [201] As far as the division between direct speech and narrator text is concerned, the use of δή as intensifier is largely limited to direct speech, most likely because the narrator does not tend to offer intensifications. In order to better understand this intensifying function of δή, a study of the combination δὴ γάρ is informative.

§62. The peculiar combination δὴ γάρ is largely limited to Homer, with an additional three instances in the Homeric Hymns and two in Hesiod. [202] In principle, there are two ways to explain such a combination of particles: either δή and γάρ occur together because they work together, or they occur together because their positions (initial and peninitial) happen to be contiguous. Now, δή in Homer occurs more frequently in peninitial than in initial position, but a pattern emerges from the analysis of the exceptions above. At first glance the δὴ γάρ group might be read like (t19) and (t20) above, as γάρ intervening in a unit of δή X, where δὴ has scope over the constituent that immediately follows γάρ. [203] Keeping in mind the possibility that one aspect of δή is its use to intensify a following adjective, consider this passage:

(t21)

αὐτὰρ ὁ βῆ κατὰ δῶμα | φίλον τετιημένος ἦτορ, |
νευστάζων κεφαλῇ· | δὴ γὰρ κακὸν ὄσσετο θυμῷ |

Odyssey 18.153-154

And he went down the hall | sorrowful in his own heart |
bowing his head. | For his spirit boded a truly bad thing. |

The translation above represents the reading of δή with small scope, intensifying κακόν; Murray appears to give a similar reading, with “for his spirit boded ill indeed,” but the English is ambiguous. This reading, however, only works in this and two other instances of δὴ γάρ, [204] too low a number to sufficiently explain the combination’s function. In the fourteen remaining cases δή cannot be read as intensifying the word immediately following γάρ, but rather appears to modify the entire act. This passage from the Odyssey, where δὴ γάρ is in act-initial position but not followed by an adjective, exemplifies these cases:

(t22)

Odyssey 16.278-280

          Now, I tell you, order them to cease from their follies,
          coaxing them with gentle words. And they, not at all
280    will they obey you. For really, their fated day is at hand.

We have rendered δὴ γάρ as a separate punctuation unit in English, which is how sentence adverbs are generally rendered to signal that they have scope over the entire sentence. [206] Whereas with scope over one adjective δή intensifies the meaning of that word, with a larger scope it rather marks an intensity behind the utterance of the act: not “very X” but “take note that X” or “I insist that X.” This reading of δὴ γάρ is supported by a final example of δή in act-initial position:

(t23)

Τεῦκρε πέπον | δὴ νῶϊν ἀπέκτατο πιστὸς ἑταῖρος

Iliad 15.437 (in direct speech)

Good Teucer, truly a trusted friend of ours was slain.

The context and content strongly suggest that there is no reason to read δή as intensifying νῶϊν here; the focus of the act is on conveying that a close friend has passed away, not that he has passed away particularly for them. The most natural reading is that δή marks the intensity behind the utterance, and does not function to intensify one of the constituents in the act. Therefore, δή has scope over at least its entire act, and its force modifies the act of uttering rather than the content of the utterance. [207] This use of the particle is inextricably linked to the “voice” of a subjective speaker, and as such is more frequent in direct speech than in narrator text. At the same time, when it does occur in narrator text, the particle may reveal something of the “voice” of the narrator. This aspect of δή is explored further in IV.4.5.3-4.5.5 and 4.6.1, considering the particle in direct speech, indirect speech and thought, and authorial statements in Herodotus and Thucydides.

§63. In its intensifying function, then, δή can modify the content conveyed (with small scope) or it can mark the intensity of the speaker in conveying his discourse (with act scope), and the boundary between these two options is inevitably fuzzy. To clarify how δή works as a mobile in Homer, I suggest an analogy with καί, a particle that knows a similarly strong correlation between position in the act and scope. If καί has scope over (at least) its host act, it occurs in act-initial position, while if it has scope over one word (group) it immediately precedes that word. [208] Homer shows remnants of a similar distribution of δή, either directly preceding a constituent that it modifies (t19 and t20), or occurring in act-initial position if it modifies (at least) its host act (t22 and t23).

3.3.3 Homeric δή: conclusions

§64. Recent scholarship on δή has focussed largely on its function of emphasizing, and the tendency has been to see the discourse-articulating function of the particle as a branch of this main function, if it is discussed at all. For δή in the Homeric epics it is important to be aware of the clear discursive differences between the two functions, and for the so-called “emphatic” function we can come to a more nuanced differentiation. When δή marks larger narrative steps, it freely occurs in any kind of discourse, both within and outside direct speech. In this function, δή does not serve to intensify either the content or the act, but along with a temporal marker it signals a progression in the articulation of moves in discourse. In its intensifying function δή modifies either single words or entire discourse acts, originally occurring in act-initial position when it has scope over the act, or directly preceding the word (group) it modifies when it has smaller scope. By Homer already, but especially in later literature, intensifying δή with scope over the entire act gravitates more and more to act-peninitial position. [209] However, δή has small scope in some constructions: preceding the constituent it modifies in fixed constructions like δὴ μάλα, δὴ μάλιστα and following it in constructions like superlative + δή. [210] In its peninitial position, the two functions still discernible in Homer start to flow together, but even in authors like Herodotus both of them may still be identified. [211] Only through a close examination of context and cotext is it possible to gain a deeper understanding of the many faces of δή in Greek literature. [212]

3.4 Move transitions in Pindaric discourse

3.4.1 Particles at move transitions in narrative

§65. The longest narrative in Pindar provides a good counterpoint to Homer’s style. His narrative of Jason and Medea in Pythian 4 is complex and plays with several narrative voices, but it too guides its audience from scene to scene. In the following I pick out a number of transitional passages that illustrate the linguistic and extra-linguistic tools available to Pindar to mark discontinuities. In the first part of the lengthy song (lines 4 – 58), Pindar adduces Medea as the narrator of the first part of the story regarding the foundation of Cyrene. Medea’s words are introduced by a speech formula in 11-12, and start with a story of one of the Argonauts receiving a clod of earth as a guest gift from Triton, which is presented first in regression (lines 20-25) [213] and then recounted from the beginning (line 25 and further):

(t24)

Pindar, Pythian 4.25-35

25      (…) Before, twelve
          days from the Ocean we carried it, across empty ridges of land,
          the spear of the sea, drawing it according to my plans.
          And then the solitary god came to us, with the radiant
          features of a respectable man. With friendly words
30      he began, as when to arriving guests generous men
          first announce dinner.

          But of course the excuse of our sweet return
          kept us from staying. He said that Eurypulos, son of the immortal earthholder and -shaker,
          he was; he noticed that we were pressed. At once having picked up soil,
35      with his right hand he strove to offer it as a makeshift guest-gift.

The narrative progresses at a steady pace, with δέ occurring at every significant narrative step (lines 25, 28, 32, 34, 34). In such narrative passages Pindaric use of δέ may appear to approach that of Homer, but in practice the particle is not as flexible in the Victory Odes. Note that, unlike in editions of Homer, δέ is always preceded by a high dot or a period. In Homer, δέ can introduce practically any kind of act, ranging from prepositional phrases to full main clauses. In Pindar the range is similarly large, but most commonly it serves to separate periods or sentences, as in all 5 examples above. Not only are the discourse segments separated by δέ all syntactic wholes, they also form discrete narrative events. [214] That is to say, δέ in Pindar serves as a boundary marker on a slightly higher level of discourse division than in Homer, closer in fact to the function of δή in the latter (see §61-§62).

§66. After a few lines of reflection, Medea’s speech ends, and it is capped with the intriguing construction ἦ ῥα Μηδείας ἐπέων στίχες (line 57). [215] For the current purpose, the occurrence of the particle ἄρα is most relevant. Like δή, ἄρα occurs much less in Pindar than in Homer, with only 30 instances in the Victory Odes. [216] Of those, seven occur in Pythian 4, which is not so surprising if one keeps in mind that in Homer ἄρα is extremely frequent in narrative, but much less so in direct speech. [217] In Pindar this pattern continues, with the great majority of instances of ἄρα occurring in narrative sections. [218] In this construction in Pythian 4, ἄρα is not only part of a narrative, but also within that context a part of a speech-capping construction, which is an environment where we often find ἄρα. [219] The verse that follows direct speech represents a return to the action, after the special kind of move that direct speech is. In this kind of context, ἄρα marks the expected outcome of what precedes, consider this parallel from later in Pythian 4:

(t25)

Pindar, Pythian 4.132-134

But on the sixth, laying out the entire serious story from the beginning, the man
shared it with his relatives. And they followed him. At once from the couches
he rose with them, and they came to Pelias’ palace.

After telling the story of the homecoming of Jason, who spends five days catching up with his father and family (lines 124-131), Pindar moves on to the peak of the encounter between Jason and Pelias. The peak is introduced with ἀλλ᾽ ἐν ἕκτᾳ (line 132), which follows a typical pattern: “for five days nothing happened, but on the sixth…” [220] As in the ἀλλ᾽ ὅ γε construction, a negative statement is followed by a positive one, introduced by ἀλλά. The negative statement inherently projects that a change will come, often from inactivity to action. [221] Thus, ἀλλ᾽ ἐν ἕκτᾳ in itself promises the peak, and the participial clause that follows (πάντα…ἀρχᾶς, line 132) serves to postpone the crucial events. As the action starts (from οἱ δέ, line 133) the acts become shorter, and we find three finite verbs in three lines. The final act (καί ῥ’ ἦλθον Πελία μέγαρον, line 134) contains ἄρα. Although the act is clearly connected to the preceding two, it represents a shift of frame: they arrive at Pelias’ palace. In fact, the act is the outcome of the preceding scene, and as such functions as a hinge between one scene and the next. [222] This discourse function matches directly the use of ἄρα after direct speech and in the second half of the Homeric simile, [223] and it recurs in its use after pronouns (as Pythian 4.78), especially after unframed discourse. [224]

§67. Pythian 4 is an exceptional ode, but quite representative regarding its narrative transitions. Pindar’s narratives are less linear than Homer’s, proceeding in a more complex manner, but discontinuities are still typically accompanied by particles or particle combinations pointing the way for the audience. Different from epic, Pindar’s odes contain many discursive discontinuities outside of the narrative, which are best illustrated by a close reading of a representative ode.

3.4.2 The discursive flow of lyric song: Pythian 2

§68. In narrative discourse in Pindar, δέ remains the most common marker of progression, though used more sparsely than in Homer. In the same contexts, ἄρα functions to round off scenes or to cap direct speech. Beyond narrative, however, Pindar has a whole array of linguistic and extra-linguistic means at his disposal to steer his song. When considering his navigation of the many (apparent) discontinuities in his discourse, we must remember Bundy’s warning:

(t26)

We forget that this is an oral, public, epideictic literature dedicated to the single purpose of eulogizing men and communities; that these eulogies are concentrated upon athletic achievement; that the environment thus created is hostile to an allusiveness that would strain the powers of a listening audience, hostile to personal, religious, political, philosophical and historical references that might interest the poet but do nothing to enhance the glory of a given patron, hostile to abruptness in transitions, to gross irrelevance, to lengthy sermonizing, to literary scandals and embarrassments…

Bundy 1962:35 [my emphasis]

Pindar’s discourse of praise consists of endless twists and turns, transitions from present to past perhaps reflected in the back-and-forth of a dancing chorus. Despite the decidedly lower frequency of particles in Pindar than in Homer (12.7% of words versus 17.1% in the Iliad and 18% in the Odyssey), they are just as important metalinguistic markers of transitions in the discourse. More so than in Homer, however, asyndeton is the polyfunctional transitional device par excellence and can initiate every imaginable new move. [225] At the same time, the many different kinds of transition have caused several markers to become specialized in certain discourse functions. In the present section I discuss several excerpts from Pindar’s second Pythian, describing its linear progression in acts and moves, along with all the relevant markers occurring at important transitions.

§69. The second Pythian is a good example of the many turns one finds in a Pindaric ode. As often in the corpus, the first strophe and part of the antistrophe are taken up by a complex tour-de-force introducing the laudandus (Hieron), his home, and the gods to whom the ode is directed especially. On the basis of syntax, modern editors take this construction of mounting intricacy and interconnectedness as one sentence, a beautiful piece of language architecture, which is reflected in the recent Loeb translation. [226] Pragmatically, however, the passage consists of a number of separate moves that are themselves articulated in manageable discourse acts. Consider the first strophe:

(t27)

Pindar, Pythian 2.1-8

          Great city—Syracuse! Of deep-in-war
          Ares a sanctuary, of men and steel-clad horses [227] a divine nurse.
          To you, bringing from Thebes-the-Shining this
          song, I come, <with> news of the four-horse chariot that shakes the earth: [228]
5        Hieron of the good chariots, prevailing in that <contest>,
          crowned Ortygia with far-shining garlands,
          seat of the river-goddess Artemis. Not without her
          did he master those pretty-reined mares with his gentle hands.

The intricacy of the syntactic construction of the first strophe belies its performative clarity; consider first the sequence of acts. The first two lines, in three discourse acts, form a tricolon building up to Syracuse’s largest boon: her men and horses. The occasion of this song, victory in a chariot race, is thus suggested, and Pindar moves on: with a second-person plural pronoun he involves the audience, followed by a participial act whose final boundary is determined by the hyperbaton of τόδε and μέλος (ὕμμιν τόδε τᾶν λιπαρᾶν ἀπὸ Θηβᾶν φέρων / μέλος). [229] The “you” implies an “I,” which is realized in line 4 (ἔρχομαι). The apposition to μέλος that closes the line (ἀγγελίαν…) projects a report of the actual news, and the audience’s expectation is fulfilled immediately in line 5, in two acts: the first mention of the laudandus is accompanied by the third reference to the event (εὐάρματος Ἱέρων), which gives Pindar the possibility to refer to the actual victory – already abundantly in focus – with the minimal participial phrase ἐν ᾇ κρατέων. The following act (τηλαυγέσιν ἀνέδησεν Ὀρτυγίαν στεφάνοις) contains the main verb of which Hieron is the subject, but cognitively the act brings the audience from the athletic event to Syracuse, describing how the victor “crowned Ortygia,” which primes the goddess Artemis, named in the following act (ποταμίας ἕδος Ἀρτέμιδος). [230] This then allows the poet to set up the theme of divine aid in the victory, to be elaborated in the antistrophe.

§70. These acts together make up three coherent moves. The first move of the ode consists of three vocatives (Μεγαλοπόλιες ὦ Συράκυσαι, | βαθυπολέμου / τέμενος Ἄρεος, | ἀνδρῶν ἵππων τε σιδαροχαρμᾶν δαιμόνιαι τροφοί, lines 1-2), which establish a marked directionality of the discourse, from chorus or singer to audience. Line 3 marks the transition to the second move with ὕμμιν. [231] Since the pronoun is plural, its referent is potentially ambiguous: it may refer to the city of Syracuse or to its inhabitants. However, deictically ὕμμιν must have been a very strong moment in the song, and the audience cannot but have felt addressed. The second move lays down a foundation for the performance: it establishes the relation between “I” and “you,” it summarizes the nature of the song, and it does all this in the form of a quasi speech-introduction: φέρων μέλος | ἔρχομαι. The third move (lines 5-8) contains the actual ἀγγελία announced at the end of the second move. It follows in asyndeton, and we would expect it to have been performatively marked as different from the preceding. The news is that of Hieron’s victory, probably superfluous to the audience, and it leads the song back to Syracuse, and to Artemis. The hinge act is ποταμίας ἕδος Ἀρτέμιδος (line 7), which describes Ortygia and at the same time serves to introduce the goddess. Lines 7-8, from ἇς οὐκ ἄτερ onward, present Artemis as a divine helper to Hieron, while cohesion with the preceding acts is achieved with yet another reference to the athletic event (ποικιλανίους ἐδάμασσε πώλους, line 8).

§71. It will not have escaped the reader that I have yet to mention particles in this discussion. The simple reason is that apart from τε (line 2), the first strophe of Pythian 2 contains no particles, but its relevance derives precisely from this absence. First, the beginnings of Pindaric odes regularly have very few particles. Second, this strophe illustrates well that move transitions are inherent in any kind of discourse, whether it has particles or not. However, when particles do occur at move beginnings, their function should be understood with respect to those larger movements in the discourse. Consider the first antistrophe of Pythian 2:

(t28)

Pindar, Pythian 2.9-16

          No, on them the virgin archeress, with both hands,
10      and Hermes of the games, place the shining harness, whenever the polished car
          and under the bit-steering chariot he [sc. Hieron] yokes
          the strength of horses, calling on the trident-wielding, wide-ruling God.
          Now, to different people does each man pay tribute,
          a resounding hymn for kings as recompense for their excellence.
15      They often sing about Kinyras,
          the voices of the Cyprians, whom the golden-haired loved willingly, Apollo

γάρ, or more precisely ἐπὶ γάρ, introduces the antistrophe and functions as a join after the performative discontinuity. ἐπί is not followed by a dative (the word following γάρ is in the nominative) which suggests that it refers back to πώλους in line 8. [232] γάρ, meanwhile, points ahead, marking the current act as an elaboration, an unfolding of the claim made earlier (line 8). There is a shift in tense from aorist (ἀνέδησεν 6, εδάμασσε 8) to present (τίθησι), which accompanies a shift from the specific victory to the regular aid that Hieron receives from the gods (ὅταν 10), an extension of the general theme of divine aid. [233]

§72. Race’s translation of γάρ with “because” renders the function of the word well enough here, but the simplicity of the translation hides the complexities of the construction. It is not the case that “Hieron mastered the horses with Artemis’ help, because she puts the harness on.” That is to say, there is no causal relation on the propositional level between the γάρ act and what preceeds. One could say that γάρ is used to mark what Sweetser would call an epistemic relation between the two clauses: “I can say that Hieron had divine help, because Artemis puts the harness on…” [234] More practically, however, γάρ marks its host move (ἐπὶ γάρ…θεόν, 9-12) as one that is triggered by the preceding discourse: the association here is of a general statement with a particular event. [235] Hence my translation of γάρ as “No,…” which marks the current move as an expansion on the preceding, while at the same time clearly signalling the start of something new; an effect that is lost if one translates “because” or “since.” [236]

§73. I do not dwell here on the instances of τε in line 10 and 11, which represent a typical use of the particle after a pronoun to mark a piece of knowledge as shared or to-be-shared. [237] More noteworthy is the occurrence of δέ in line 13, since it illustrates a difference in the way that the particle functions in Homer and Pindar. [238] Pindar uses δέ more frequently in narrative than in other parts of his discourse, but in both contexts he uses the particle to mark significant discontinuities. Here it accompanies the transition from the specific case of Hieron (line 11) to the gnomic (or at least general) statement that different men honor different heroes (ἄλλοις δέ…ἀρετᾶς, 13-14). In discourse terms, both the content and orientation of lines 13 and 14 differ strongly from what precedes and what follows, suggesting that it is a separate move. [239] It is relevant that in this case δέ is not only in act-peninitial position, but also in verse-peninitial position. The fact that this metrical boundary matches the boundary marked by δέ suggests that the discontinuity perceived is stronger than when δέ is in verse-medial position. [240] The next move is marked by μέν (line 15), which projects the theme introduced in the act into the upcoming discourse. Pindar uses the particle more than once to carry a theme or narrative over the metrical boundary between strophe and antistrophe or, as here, between antistrophe and epode. [241]

§74. In the epode that follows, Pindar does not proceed with a narrative about Kinyras, as the audience may have expected, but rather makes a direct comparison between the Cyprian hero and Hieron (lines 17-20). Subsequently he launches into a narrative about Ixion, a paradigmatic narrative of divine punishment for overreaching. [242] Consider the end of the epode and the beginning of the second strophe:

(t29)

25 ἔμαθε δὲ σαφές. | εὐμενέσσι γὰρ παρὰ Κρονίδαις |
γλυκὺν ἑλὼν βίοτον | (…)

Pindar, Pythian 2.21-26

          By the orders of the gods, they tell that Ixion says the following to mortals,
          on his feathered wheel,
          spinning in all directions:
          to go to one’s benefactor and pay him back with good deeds.

25 And he learned it clearly. Indeed, among the kind children of Kronos,
having had a sweet life, (…)

In the epode Pindar introduces the story of Ixion with φαντί (compare the example from Pythian 1, t4), followed by a gnomic thought attributed to Ixion. Since the gnṓmē occurs at the end of the epode, Pindar could have left it at that, but in the following line the third-person aorist form (ἔμαθε) followed by δέ suggests that the audience should assume continuity of grammatical subject at the start of the new strophe: Ixion is still in focus. [243] The actual narrative begins with γάρ (line 25), which introduces a clause that serves as an orientation for the narrative, in the form of a participial construction. The complication is introduced in the apodosis: μακρὸν οὐχ ὑπέμεινεν ὄλβον (line 26), and then the narrative unfolds. Only in lines 34-35 is it interrupted briefly by a gnomic thought, introduced by χρὴ δέ, a cluster that almost invariably starts gnômai in Pindar. [244] After Ixion’s misdeeds and his punishments have been discussed (26-41), the resolution of the story (starting with asyndeton, ἄνευ οἱ Χαρίτων τέκεν, 42) tells of the birth of a son from the union of Ixion and Hera-as-cloud, called Centaur, whose offspring with a mare in turn yields the familiar creatures that are half horse and half man. This rounds off the second epode (41-48).

§75. The third strophe starts with an expanded statement of the principle that gods can put down or raise up any mortal, already brought up in lines 7-11. Pindar then redirects attention to the here and now with a metadiscursive statement in 52 (ἐμὲ δὲ χρεών). This leads to a piece of advice to himself in the form of a gnṓmē at the end of the strophe, urging him to focus on praising positive things rather than blaming (lines 54-56). This advice he immediately applies in the antistrophe to his laudandus: τὺ δὲ σάφα νιν [sc. πλοῦτον] ἔχεις (“But you have it [sc. wealth] with wisdom,” line 57) starts a new expansive move dedicated to praising Hieron’s money and power. Note how once again the explicit mention of a first or second person followed by a particle accompanies move transitions in the songs. The remainder of the song alternates between commenting on Hieron’s prowess and alluding to some people who seem to have slandered either Hieron or Pindar. At the end of the fourth antistrophe these topics are left behind for another gnṓmē, in this case acting as a bridge to the last epode, which stresses once more that gods decide the fate of men.

(t30)

(…) χρὴ δὲ πρὸς θεὸν οὐκ ἐρίζειν |

Pindar, Pythian 2.88-96 (end)

(…) One should not contend with god.

          He raises up, sometimes the fate [lit. “things”] of those people, [245] then again to others does he gives great honor. But not even that
90      warms the mind of jealous men. Pulling some line
          too far, they fix a painful wound into their own heart,
          before they succeed in those things that they plan in their minds.
          To carry lightly the yoke one has taken on one’s neck
95      is helpful. And I tell you, kicking against the goad
          is a slippery course. Pleasing them, may it be granted to me to consort with the good.

Once again χρὴ δέ (line 88) introduces the gnomic thought, but it is followed by a change in direction marked by ἀλλά (line 89). ἀλλά here expresses no semantic contrast but rather marks a redirection of the discourse, and this use of the particle is decidedly more frequent in Homer and Pindar than it is in Attic Greek. In Homer this use appears most commonly in constructions where ἀλλά marks transitions to a move containing an imperative or a wish at the end of direct speech. In Pindar, however, it occurs at all kinds of re-orientations, thus often coinciding with move beginning. Here, the penultimate move of the song (lines 89-96) is concerned with the theme of envious men, but the meaning of the passage is ambiguous. [246] It leads on to the final act and final move of the song, introduced by δέ and containing a first-person pronoun (line 96), in which Pindar expresses a hope for himself. [247] As all Pindaric discourse, even this apparent personal statement implicitly praises the laudandus and the present audience.

§76. The second Pythian ode provides a good case study to support the argument that considering Pindaric discourse as a sequence of moves by the poet offers insight into the use of language in general and of particles in particular. This ode is particularly illustrative since it contains so many different kinds of discursive transitions, but this kind of approach will illuminate the reading of any Pindaric ode. Therefore we apply the terminology of acts and moves throughout the other chapters.

3.5 Conclusions

§77. In this chapter I have shown that in order to understand why certain particles occur in certain places, we need to look at the larger movements of discourse at all levels beyond that of the sentence. Earlier studies on ἦ and δή have focused mainly on their role within the scope of the sentence, or in relation to adjacent words. Moreover, there is a tendency in the scholarship on almost all particles to find a general description that covers all the different aspects of the particle. Both tendencies have obscured certain patterns that can refine our understanding of what particles do.

§78. ἦ, δή, and γάρ all have a facet that is connected to the start of new moves. For ἦ it is its interjection-like value that makes it performatively effective at discourse transitions. δή’s occurence at narrative transitions, often in concert with temporal markers, is not as easily explained. It is clear, however, that these occurrences should not be regarded as representatives of its intensifying function. In fact, the Homeric corpus reveals that δή possesses a whole spectrum of functions, each of which is bound to specific contexts and to δή’s position in the act. Finally, as regards γάρ, apparently simple translations often belie the particle’s more complex functions. Like “for” in English, the particle can signal relations among many different elements in discourse. In narrative, the use of γάρ often marks the start of a new move, which is associatively connected to the surrounding discourse. Understanding this use of γάρ opens up a range of alternative translations that do not foreground the close causal relation that is rendered by “for” or “since”, but rather render the sense of discursive discontinuity between the preceding move and the one introduced by γάρ.

§79. The Attico-centric approach to particles has left its traces everywhere. The focus on the causal/explanatory function of γάρ, the attempt to find a single function for δή, the reading of ἦ as a basically affirmative or emphatic particle, the reading of δέ as adversative, and the assumption that every μέν implies a δέ, are all examples of the orthodoxy that is especially misleading when working with corpora like that of Homer and Pindar. Starting from Attico-centric assumptions prevents us from perceiving the richness of metalinguistic functions that these particles perform, and it leads to the mislabeling of Homeric and Pindaric usage as odd, anacolouthic, or simply metrically convenient. We should rather view their usage as signs of earlier stages in the grammaticalization of certain constructions, or as due to the particularities of performative genres.

§80. A significant part of the chapter considers particles not in isolation, but in clusters and in constructions with other word groups. Particle combinations are not monolithic: the same combination may combine different functions of the particles involved. Hence we should be wary of attributing fixed functions to a certain combination. Furthermore, there is a difference between a combination (where the sum is no more than its parts) and a cluster (where the sum is more than its parts, and the cluster has evolved to become somewhat detached from its origins). Especially for early Greek, but perhaps for ancient Greek in general, it is risky to make generalizing statements about the force of a particle combination. Only when it has completely grammaticalized so as to be regarded as essentially one word can we speak of a coherent function, and even in that case one can never exclude the possibility that in some instances the particles have ended up next to each other by happenstance. To understand particle combinations, one must appreciate the richness in function of all the constituent elements, and search for patterns in the different contexts in which the combination is found.

§81. A final thought to take from this chapter is that editions are not sacred. Especially for vulnerable little words like particles, and even more so for ambiguous lexemes like H, the normalizing practices of the Alexandrians, scribes, and modern editors may have obscured subtle but important differences. Detailed studies of all these different stages of textual transmission combined with an informed scepticism toward the sources at our disposal allow us to propose new readings of particles, where the content and the form of the discourse provide enough material to make an argument. In short, paying attention to the larger discursive context in which individual sentences are situated can enable more nuanced readings of even the smallest details in a text.

Footnotes

[ back ] 1. Compare IV.3 for a similar consideration of differently-sized units in discourse.

[ back ] 2. Sinclair and Coulthard 1975; Roulet 1984, Roulet et al. 1985, Roulet 1997, Roulet et al. 2001; Risselada 1993; Kroon 1995; Langacker 2001; Hannay and Kroon 2005. Move is not an unproblematic term, as it is used in different ways and has been applied in different fields: compare for example Ryan 1991:130, who uses “move” as follows: “I call a ‘move’ an action with a high-priority and a high risk of failure.”

[ back ] 3. In this sense the concept of move has been applied to Homeric epic by Beck in Grethlein and Rengakos 2009, where she says (146) that “[a] ‘move’ is essentially a speech act in a conversational context” (she repeats the definition in Beck 2012:12, with reference to Kroon 1995, Risselada 1993, and Roulet 1984). This is a narrower definition of move than the one we (and most of her sources) employ.

[ back ] 4. This is a constucted example for the sake of illustration.

[ back ] 5. See III.4.1 for the Conversation Analysis approach to interactions. In Conversation Analysis these two “turns” form a “sequence,” and the invitation and rejection are an “adjacency pair.”

[ back ] 6. The establishment of this common goal is inevitably subjective, and it cannot always be ascertained. Consider the final acts by speaker B: | Sorry, | is still part of the rejection move, as an attempt to mitigate the interlocutor’s loss of face. The final act, | maybe next time? | however, could either be regarded as a continuation of this attempt at mitigation, or as an actual question about “next time” in which case it might be regarded as the start of a new move. In practice, it is the reaction of the interlocutor that establishes the “right” interpretation of this act: if speaker A responds “Sure!” the exchange ends without any new moves, but if she responds “OK, | they are also playing next Thursday, | how’s that?” she has clearly taken speaker B’s question as a new move.

[ back ] 7. Roulet’s analyses reveal the same tendency in French: the beginnings of moves are often marked by what he calls “pragmatic markers” (see I.3 for the terminology), while the linguistic form of the ends of moves is never discussed (see especially Roulet 1984:36-39, on a newspaper article). Compare Langacker 2001:178 who remarks that the intonation group “Now Bill,” “serves to announce that structure building is going to start in a new place.” Chafe 2001:674 says the following about “topic”: “Topics generally have clear beginnings, although that is not always the case (…) and their endings are sometimes well defined, sometimes not”; compare Tannen 1984:41-43.

[ back ] 8. Consider Langacker’s insightful comment (2001:177): “at any point in a discourse can we stop working on one structure and start building another.” See IV.3 §§94-98 for the discursive discontinuities that can accompany move starts in historiography.

[ back ] 9. The terminology was established initially by Sinclair and Coulthard 1975 (who founded the so-called Birmingham school). Their work was picked up by Stenström 1994 and by members of the “Geneva school”, represented especially by Roulet 1984, Roulet et. al. 1985, Roulet 1997, Roulet et al. 2001; see II.2 §18. Kroon 1995 applies the Geneva model to written Latin discourse in her discussion of discourse particles.

[ back ] 10. Consider Kroon’s definition (1995:66): a move is “the minimal free unit of discourse that is able to enter into an exchange structure. (…) A move usually consists of a central act (which is the most important act in view of the speaker’s intentions and goals) and one or more subsidiary acts…”

[ back ] 11. Roulet insists upon the hierarchical approach, but he suggests that the hierarchy is not absolute when he argues that moves can contain other moves: in 1981:10-11 the examples suggest that an intervention (move) can itself be part of larger interventions, see Roulet 1984:45 “…moves [can] constitute a larger move and are tied together by interactive functions”; see also Roulet 1997:134 figure 2, and Roulet 2001:53.

[ back ] 12. This matches the idea of “stucturing” instead of the static “structure”; see Sherzer 1982:389, “I use the term structuring rather than structure in order to stress the dynamic process which is involved” [emphasis original]. Kroon 1995 focuses on rhetorical prose and is more optimistic about establishing the structure of her corpus (66-67): “Many particles (both in Latin and in other languages) appear to be involved predominantly or partly in marking the linear or hierarchical structure of a discourse as outlined above. This means that they mark out the separate units of discourse by indicating how these are structurally tied up with other units of the same discourse, both linearly (i.e. involving relations between units of the same rank) and hierarchically (i.e. involving the relationships that units maintain with ‘higher’ or ‘lower’ units). As such these particles can be said to have a mainly organizational function.”

[ back ] 13. Even the speech-capping formulae are generally more forward than backward looking. Many of them mark the end of the direct speech rather by focusing on the upcoming narrative move (such as the announcement of a new speaker) than on rounding off the preceding move of direct speech.

[ back ] 14. See Louviot 2013 on speech-capping in Old-English poetry.

[ back ] 15. See Bonifazi and Elmer 2012:246.

[ back ] 16. I borrow the term from Longacre 1996², discussed in footnote 46 below.

[ back ] 17. The vertical bars mark discourse act boundaries, see II.2.1.2.

[ back ] 18. See II.4 §§11-14 for the comparable “framed” vs. “unframed” discourse, with further examples of this temporary breaking of the spell.

[ back ] 19. The actual interpretation of the passage is contested, and this summary represents only one possible reading, but the discussion is not relevant to my contention that the Catalogue is presented as a separate move.

[ back ] 20. For a discussion of metanarrative passages in the Iliad, with literature, see De Jong 2004b:45-53, and see De Jong 2001a:119-120 for a listing of narratorial interventions in the Odyssey. For this specific line and the cognitive importance of οὗτοι see Bakker 2005:80-81 and 143 (although I would propose a different reading of ἄρα, see for the similar function of ἄρα after the simile II.4.3.2).

[ back ] 21. See IV.3 §100 for moves linked to genre expectations in historiography.

[ back ] 22. We want to avoid any attempt to define “move” in an absolute sense, as some scholars attempt to do. One can imagine that it is very possible to name the different layers of the hierarchy in the Catalogue, which is so clearly divided, but such a system would not easily translate to the rest of the Homeric corpus. If we call the Catalogue an “episode,” the different entries “subsections,” and then the smallest discourse units above the sentence level “moves,” this distinction may be useful on a local level, but it becomes difficult, and perhaps unproductive, outside of the Catalogue. Using a different set of terms, Givón 2005:141 (with reference to Givón [ed.] 1997) does offer a full terminology for what he regards as the different levels of discourse.

Do note that the Catalogue is not representative for the rest of Homeric discourse. It is a very strict, ringed piece of discourse that – not despite but because of its intricacy – allows the performer to reperform a large body of specific knowledge, such as a list of places and their leaders; see for the cognitive aspect of catalogues in Homer Bakker 1997b:60, Minchin 2001:73-99, Tsagalis 2010, and Strauss Clay 2011:117, with note 59.

[ back ] 23. See II.2 §18 for more on the Birmingham and Geneva schools.

[ back ] 24. Bundy describes the Pindaric odes as unities built up out of different pieces all aimed at praising the victor, but defines the separate pieces, which we would call moves, only on a functional level (a gnomic passage, a priamel, a narrative, etc.). The closest he comes to a description in terms of moves is in the conclusion of his second paper: “[T]o follow the movement of the ode is not to follow the development of a thought that has a beginning, a middle, and an end, but to pursue the fulfillment of a single purpose through a complex orchestration of motives and themes that conduce to one end: the glorification, within the considerations of ethical, religious, social, and literary propriety.” See also Felson 2004:387, about Pythian 9: “Despite these ruptures, a set of real-world “moves” can be retrieved by an audience informed about pan-Hellenic contests and the epinician genre.” It is unclear how Felson uses “move” here, but it may be in a deictic sense (compare “shifters”).

[ back ] 25. Wells 2009:61-128 speaks of the different “speech genres” that alternate in Pindaric discourse.

[ back ] 26. See II.4 §§22-23 for more on this view of gnômai.

[ back ] 27. The term “move” is not commonly applied to narrative, but consider Roulet et al. 2001, especially the example on page 328-329.

[ back ] 28. Included in this category are the performative tools that the epic performer has at his disposal to make a transition within the narrative seamless, such as by using sightlines or sounds et cetera.

[ back ] 29. See Maschler 2009, who explores the link between discourse markers and metalanguage in the talk of bilingual speakers of English and Hebrew. Grosz and Sidner 1986:177 talk about “discourse segments” and describe the relation between linguistic form and discourse in the following terms: “There is a two-way interaction between the discourse segment structure and the utterances constituting the discourse: linguistic expressions can be used to convey information about the discourse structure; conversely, the discourse structure constrains the interpretation of expressions.” The first part of their two-way interaction is what I call metadiscursive language.

[ back ] 30. See Grosz and Sidner 1986:177-178 on what they call “discourse segment boundaries”: “The explicit use of certain words and phrases (…) and more subtle cues, such as intonation or changes in tense and aspect, are included in the repertoire of linguistic devices that function (…) to indicate these boundaries.”

[ back ] 31. Herman 2010:140.

[ back ] 32. It started in the study of folktales and myth; the landmark works are Propp [1928] 1958, Lévi-Strauss 1960 (review of Propp), Bremond 1973, Todorov et al. 1979, Detienne 1981, and Vernant and Vidal-Naquet 1986; see also Hansen 2002:1-31 for an overview of the earlier folktale studies and the (limited) application of the methodology in classical studies.

[ back ] 33. See especially Barthes 1957 and Genette 1972; Prince 1997:39-42 gives a very compehensive history of narrative and narratology up to the late nineties, and includes a rich bibliography. A central question that has arisen in the last decades is: what makes a discourse narrative? The discussion is too complex to present here, but studies that address the question include Fludernik 1996, Ryan 2007, and Herman 2009a. A commonly held principle is to regard something as narrative when it contains at least two unique events in a temporal sequence (see e.g. Labov and Waletzky 1967, Couper-Kuhlen 1988:353).

[ back ] 34. De Jong 1987, 2001, 2007 (ed.), 2012 (ed.); Rood 1998; De Jong and Sullivan (eds.) 1994, De Jong, Nünlist, and Bowie (eds.) 2004; Stoddard 2004; Grethlein and Rengakos (eds.) 2009; Köhnken 2006; Grethlein and Rengakos (eds.) 2009.

[ back ] 35. E.g. De Jong (ed.) 2007 and De Jong (ed.) 2012. Prince 1997:40 says that it is “the narrating as opposed to the narrated, (…) the signs in a narrative representing the narrating activity, its origin, its destination, and its context (…) that narratologists have explored most thoroughly.” However, none of his following examples discusses the language of narrative and its relation to the narrating activity, which is exactly what I focus on in the present chapter.

[ back ] 36. See Longacre’s work on the linguistic form of narrative discourse (1985, 1990, and 1995). Pseudo-Longinus, in his work On the Sublime (chapters 23- 29) already discusses the linguistic changes at certain moments in the narrative. His work considers discourse transitions above the sentence level, and suggests that he considers the pragmatic perspective on language; see the summary of Terry 1995:119-120, “Among those changes which he discusses are the expansion of the singular into the plural to convey the idea of multitude (23.2-3), the contraction of the plural into the singular to give an effect of sublimity (chap. 24), the use of the present tense in narrating past time in order to increase vividness (chap. 25), the change of the person addressed from the whole audience to a single individual also to give a vivid effect (chap. 26), the use of the first person for one of the characters to show an outbreak of emotion (chap. 27), and the use of periphrasis or circumlocution to give the work a far richer note (chap. 28-29).”

[ back ] 37. In its original incarnation, I assume that Homeric epic was composed in performance (see e.g. Nagy 1996 and Bakker 1997b), while Pindar’s victory odes were probably composed beforehand. If we assume that a number of odes were performed by a chorus, they must in fact have been finished well before the occasion; see II.1 §§1-5 for more on the topic.

[ back ] 38. There are also relevant non-cognitive processes, such as the limitations imposed by the need to breathe.

[ back ] 39. See Enfield 2006, especially 409-412. This touches upon the topic of common ground, which I explore in terms of discourse memory in II.4.1.

[ back ] 40. See e.g. Reynolds 1995, a comprehensive study about performances of the Sirat Bani Hilal oral epic.

[ back ] 41. See especially Bamberg (ed.) 1997, which contains a re-publication of Labov and Waletzky’s 1967 article; and additions to the scheme offered by Fleischman 1990 (who introduces “peak” between complicating action and resolution), Fludernik 1996, and Klapproth 2004. Minchin 2001:186-196 applies the scheme to Homeric narratives.

[ back ] 42. These are the terms given by Labov and Waletzky, and the brief explanations between parentheses are paraphrases of their longer discussions of the parts.

[ back ] 43. The first explanation of “evaluation” is given in Labov and Waletzky 1967:28-35, with a substantial revision in Labov 1972. In the 1997 anniversary volume, several people expand on the topic: Daiute and Nelson 1997 (evaluation as an emergent point of view) and Fleischman 1997:165-166; see also Fleischman 1986 and Bower 1997 (syntax and word order as a possible marker of evaluation).

[ back ] 44. See for the cognitive basis of narratives: Schank 1995, Talmy 1995, Turner 1996, Herman 2011, and Sanford and Emmott 2012.

[ back ] 45. Chafe 2001:677 notes that the idea of “climax” is missing from the Labovian narrative schema. He himself proposes an alternative schema; for the use of the term peak see Longacre 1996².

[ back ] 46. Longacre 1996² designates peaks in narrative as “zones of turbulence” and shows how they can be marked (1996²:39-48) by: (1) rhetorical underlining, (2) a “crowded stage” i.e. many characters in the frame, (3) heightened vividness, (4) a change of pace, (5) a change of vantage point, and/or (6) the sudden occurrence or disappearace of certain particles as well as use of onomatopoeia.

[ back ] 47. In the translation of lines 51 and 52 I have had to change the word order to obtain understandable English.

[ back ] 48. Note, moreover, that φαντί consistently occurs at the start of (little) narratives in Pindar: Olympian 7.54; Pythian 2.21, 4.88, 4.287, 6.21, 7.19; Isthmian 8.46a.

[ back ] 49. Compare Couper-Kuhlen 1988, who discusses subordinated when-clauses in narrative, which provide foreground rather than background information. See II.5 for more on relative clauses in narrative.

[ back ] 50. Compare the discussion of section beginnings and endings in IV.3.11.

[ back ] 51. See II.2.4.

[ back ] 52. See IV.2 §§ 112-113 for the use of ἐτελεύτα in Thucydides, and Allan 2007:113-115 for the use of the imperfect in evaluative statements in Thucydides.

[ back ] 53. See III.2.1 for the concept of discourse patterns.

[ back ] 54. Other scholars choose to distinguish between the narrative and descriptive mode in epic, but I agree with most current narratologists who argue that description in any form typically performs a narrative function (e.g. Tsagalis 2012: De Jong (ed.) 2012:1-3). By extension, it is unproductive to regard them as two different discourse modes (for the term, see Smith 2003). At the same time, however, the action of a character and the setting of that action may be presented in different linguistic forms, the transition between which we do regard as a narrative transition.

[ back ] 55. See Slater 1983. Lyric narrative is related to the idea of “epic regression”, on which see Schadewaldt 1938:84 who speaks of “zeitlich rückschreitend” (“walking backward in time”); the term “epic regression” was coined by Krischer 1971:136-140.

[ back ] 56. Slater 1983:118, referring to West 1966:161, on Hesiod, Theogony 22; see also Calame 1985.

[ back ] 57. See also II.5 for the different combinations of pronouns + particles and their functions.

[ back ] 58. See Slings 1997a and De Jong 1997. The observation that γάρ often occurs at the beginning of embedded narratives has been noted before, e.g. by Sturz 1801:565 and Slater 1969:99.

[ back ] 59. Although in some of the examples I have chosen to render this force of γάρ with “for”, this practice should not lead the reader to infer that γάρ always has a causal value, but rather that “for” also has a wider range of functions than just the causal.

[ back ] 60. See V.γάρ passim.

[ back ] 61. Sicking 1993:23.

[ back ] 62. Slings 1997a:101: “Embedded sequences are characterised by the fact that they have a different ‘frame of reference’ from the embedding sequence.”

[ back ] 63. See Slings 1997a:101, who takes the PUSH-POP model from Polanyi and Scha 1983.

[ back ] 64. De Jong 1997 “γάρ Introducing Embedded Narratives.”

[ back ] 65. The typical example of this pattern is Iliad 1.8-16 (discussed in De Jong 1997:177), where we are brought from the outcome to the beginning of the story in successive steps, after which the full story is told from beginning to end.

[ back ] 66. De Jong 1997:176-179; this is in essence the same kind of construction that Slater discusses, though using different terminology.

[ back ] 67. Those scholars who regard γάρ as one of the easily understood particles (e.g. Hummel 1993:406) have focused on classical Attic literature in their analysis of the particle, where its use is quite narrow and specialized.

[ back ] 68. Denniston 1950:56, “The derivation of γάρ from γε and ἄρ, though occasionally challenged (…), has been pretty generally accepted by scholars.” Bäumlein 1861:68 exemplifies the approach to the original value of γάρ: “Es wird mithin durch γάρ der ganze Satz als unmittelbar gewiss und unbestreitbar, als eine Thatsache, die nun einmal so ist, nachdrücklich hervorgehoben” (“Through γάρ, the entire sentence is emphatically stressed as immediately certain and incontestable, as a fact that is simply the case”).

[ back ] 69. Denniston 1950:57, referring to Misener 1904:7-10.

[ back ] 70. Bakker 1997b:116-118 notes that γάρ often occurs at the start of the “anecdote” in the structure of the ἀνδροκτασία proposed by Beye 1964.

[ back ] 71. Hoogeveen 1769:187-188; see also Sturz 1801:565-569 (on Xenophon) and Ebeling 1885:160-164 (on Homer). Denniston believes that the primary use of γάρ is asseverative (1950:57), but he argues that this force is retained only in combinations. Compare also Bakker 1997b:114 “…it is not surprising that gár is particularly at home in the vicinity of the starting point of all starting points, the very beginning of the epic tale.”

[ back ] 72. For γάρ-parentheticals in Herodotus see IV.5 §68 and the insightful analysis in Kerschensteiner 1964. Kerschensteiner 1964:36 argues that parentheticals often add crucial, rather than secondary, information (page 36); see also Lang 1984a:6-12, although she does speak of “background digressions.”

[ back ] 73. E.g. Sicking 1993:20 “subordinating the stretch (…) within the scope of the particle,” Slings 1997a:102 “subsidiary,” Wakker 2009b:69 “subsidiary explanation introduced by γάρ.” A similar view of the particle is held in the the paper presented by Luraghi and Celano 2012. De Jong and Bakker have actively opposed this notion.

[ back ] 74. For more on γάρ introducing apparent background information, see II.4.2.

[ back ] 75. Consider again Sicking’s note about “the purpose of sentences introduced by γάρ” [my emphasis]: he regards the force of γάρ as working on the level of the sentence only.

[ back ] 76. Bakker 1997b:112.

[ back ] 77. Bakker 1997b:112, see Hummel 1993:406 (on Pindar) for the same point.

[ back ] 78. See Bakker 1997b:113n50, with reference to Hopper 1979:215-216. Bakker’s description of γάρ is partly anticipated by Reynen 1958:89-90, who describes its use in Iliad 1.55 as making explicit something that has been implicit up to that point. See III.3.3.1.5 for a discussion of γάρ at the beginning of conversational turns, where the contention is likewise that γάρ does not introduce discourse that is in any sense secondary.

[ back ] 79. See II.5 for more on expansions about characters and the crucial role of particles in guiding such passages.

[ back ] 80. The use of γάρ to mark unframed discourse, as discussed in II.4, is not so much an unfolding as it is an associative insertion. In both cases there is a displacement, but here it is from one “frame of reference” to another, while in II.4 I discuss the movement between framed and unframed discourse. Regardless, it is never productive to talk about these displaced pieces of discourse as “background.”

[ back ] 81. Note, however, Kirk’s comment (1985:I.368) that “such a digression was itself attractive from a narrative point of view” [my emphasis]. De Jong 2004²:155-157 rightly notes the function of the story here, anticipated by Austin 1966:300.

[ back ] 82. This would be the equivalent of “epistemic because” in constructions like “John’s out, because the light is off,” discussed by Dancygier and Sweetser 2000:120-122.

[ back ] 83. A combination that I will not discuss extensively is αἲ γάρ/εἰ γάρ (on which see Misener 1908 and Tabachovitz 1951), which introduces wishes. This is the best construction to illustrate that γάρ can not always be “explanatory” (pace Misener 1904 and 1908, who would see a causal relation even in these instances). However, these wishes always arise directly from a preceding thought or utterance, and γάρ signals how the upcoming move arises from the preceding discourse through cognitive association.

[ back ] 84. This function is by its nature of course not limited to embedded narratives, but may also introduce other moves.

[ back ] 85. Compare the discussion of καί in IV.2 §§133-137, concerning the boundaries of the second conjunct. The process described there for establishing the function of καί could similarly be applied to better understand this function of γάρ.

[ back ] 86. A selection of examples of γάρ starting embedded narratives in Pindar: Olympian 1.55 (ἀλλὰ γάρ), 2.48 (μὲν γάρ), 7.27 (καὶ γάρ), 7.48 (καὶ τοὶ γάρ); Pythian 2.25, 3.25, 5.83, 6.32, 9.114; Nemean 6.34 (καὶ γάρ), 8.26, 9.13, 10.60; Isthmian 1.17.

[ back ] 87. De Jong 2012:71, ad 22.46; the passage reads as follows: καὶ γὰρ νῦν δύο παῖδε Λυκάονα καὶ Πολύδωρον / οὐ δύναμαι ἰδέειν. Gesamtkommentar 2009:VIII.2.216 makes the same observation about καὶ γάρ at Iliad 24.602.

[ back ] 88. Aftosmis 2010:244-270, where he also discusses οὐδὲ γάρ (on which see note 96 below). καὶ γάρ occurs eleven times in Pindar: Olympian 7.27, Pythian 1.10, 4.181, 9.42, 10.59, Nemean 1.50, 1.67, 6.34, Isthmian 2.30, 5.4, 5.26; οὐδὲ γάρ once: Olympian 14.8.

[ back ] 89. There are 10 instances out of 45 (22+23) total occurrences of καὶ γάρ in Homer that may be said to introduce particular “examples” in the sense perhaps meant by De Jong 2012 (she does not offer any parallels ad loc.): Iliad 2.292, 2.377 (example in the form of a little narrative), 5.478 (example in the form of a little narrative), 9.502, 16.810, 22.46, 24.602 (example in the form of a little narrative); Odyssey 4.199, 14.70, 17.566.

[ back ] 90. See I.1 on combinations in general, and §42 below on ἦ in combinations.

[ back ] 91. Iliad 2.377, 9.502, 9.533, 11.698, 19.52, 19.95, 24.602; Odyssey 2.17, 17.419, 18.138, 19.75, 19.186.

[ back ] 92. The Niobe narrative is a classical example of a ring-composition; see Richardson 1993:339-340 and Gesamtkommentar 2009:VIII.2.212-215.

[ back ] 93. A possible explanation for the occurrence of τε in instances like this is that it marks a tradition shared between performer and audience, here the famous story of Niobe; I translate “after all.” For an extensive discussion of this aspect of τε see IV.2.3.1 and for the particle’s use in Homer see II.4.3.1 and II.4.4.

[ back ] 94. De Jong 1997 quotes a few more instances of γάρ introducing embedded narratives: Iliad 4.467, 6.38, 6.130 (see below), 14.315; Odyssey 1.260, 3.262, 3.276, 4.677, 14.244, 14.317. These narratives are listed not only because of the use of γάρ, but also because of their regressive structure. As the examples cited here and in the rest of the chapter show, however, it does not look as though the function of γάρ in such contexts should be linked to the recessive nature of the narrative.

[ back ] 95. See for an example of this (t29) below.

[ back ] 96. Zeus in Iliad 19.95 and Niobe in Iliad 24.602. Note also Iliad 6.130-131 οὐδὲ γὰρ οὐδὲ Δρύαντος υἱὸς | κρατερὸς Λυκόοργος / δὴν ἦν and 18.117 οὐδὲ γὰρ οὐδὲ βίη Ἡρακλῆος φύγε κῆρα, which show that the opposite of the scalar function of καί can be rendered with οὐδέ “not even.” These are the only two cases where οὐδὲ γάρ introduces an associative narrative (out of 9 + 3 instances in Homer). On οὐδέ as the “negative scalar particle” see Cooper 2002:4.3069 (sub P), Denniston 1950:196, Ruijgh 1971:190, and Willmott 2011:13-15.

[ back ] 97. The Curetes had been the victim of Artemis’ wrath in Iliad 9.533; Neleus was owed a debt from the Eleians in Iliad 11.698; Aegyptius’ son was killed by the Cyclops in Odyssey 2.17.

[ back ] 98. Agamemnon too had been wounded by Coön in Iliad 19.52; Odysseus too had once lived in a house amongst his men in Odyssey 17.419, 18.138, and 19.75; Odysseus-in-disguise speaks of seeing himself coming to Crete too in Odyssey 19.186.

[ back ] 99. In the words of Richardson 1993:340 on the Niobe narrative (t7), these are arguments “a fortiori.” See for καί as a scalar particle IV.2 §123 and Bakker 1988:75, 84, 113-119, and 205.

[ back ] 100. See IV.2.4.2 for καί used to pin down something specific.

[ back ] 101. See IV.2 §124 for καί rendered as “also”; Denniston 1950:293 calls this function “responsive,” but without clarification of the term.

[ back ] 102. Note that this construction brings to mind the introduction of the “lyric narrative” as discussed by Slater and West (see fn. 56) , including ποτε and the aorist, but without the pronoun.

[ back ] 103. νημερτές is an adjective that goes with ἔπος, but I translate it as an adverb, in order to better render τοῦτο.

[ back ] 104. Here and in the following examples I give the text of the editions (ἤδη) and translate accordingly (“already”).

[ back ] 105. Gesamtkommentar 2009:III.2.84 takes γάρ as explanatory: “γάρ leitet eine Erzählung ein, die als ganze zur Erklärung der vorausgehenden Aussage dient” (“γάρ introduces a narrative, which as a whole serves to explain the preceding utterance”).

[ back ] 106. See Thiersch 1852:427.

[ back ] 107. Thomas 1894:81-83 notes rightly that in Homer in a large number of cases ἤδη is limited to the initial position. This may suggest that ἤδη evolved out of (initial) ἦ + δή, as Thomas believes (84: “In the first part of ἤδη we have plainly nothing but the common circumflexed asseverative particle [ἦ]”). Thiersch 1852:427, conversely, assumes that the manuscripts wrongly render ἤδη in initial position for a certain number of instances and proposes to read ἤδη with a purely temporal sense especially where ἤδη is found in non-initial position (fit hoc imprimis, si ἤδη in media sententia aut post alias particulas infertur).

[ back ] 108. Consider, for example, the reading of Iliad 1.453 in Ven. A and B: ἤδη μέν (ἦ μὲν δή or ἠμὲν δή edd.)

[ back ] 109. ἣ δή and ᾗ δή are also possible, but it is normally easier to establish if a feminine relative pronoun is warranted in a sentence.

[ back ] 110. The instances of ἤδη γάρ at the beginning of narratives are: Iliad 1.260, 1.590, 3.205, 5.188, 14.249; in all of these cases ἤδη γάρ is verse-initial. The same use may be found in Hesiod Theogony 645-646: ὄφρ’ εἴπω τά με θυμὸς ἐνὶ στήθεσσι κελεύει / ἤδη γὰρ μάλα δηρὸν ἐναντίοι ἀλλήλοισι, which is one of only two instances of verse-initial ἤδη γάρ in Hesiod.

Since ἤδη is a mobile, there are also instances where ἤδη in first position does mean “already” or “(by) now,” and as a result ἤδη γάρ does not begin new embedded narrative, namely: Iliad 5.206, 6.361, 14.206 (=14.305), 15.110, 15.139, 15.613, 19.334, 20.306, 23.623, 24.765; In the Odyssey, ἤδη γάρ is never used to introduce an embedded narrative; the instances are: 2.89, 2.211, 3.335, 5.161, 5.223, 6.34, 10.381, 12.451, 13.40, 15.16, 15.66, 17.606, 19.160, 19.222, 20.309. In the majority of instances the adverb is here followed by a present, an imperative, or νῦν.

[ back ] 111. See IV.2.4.2 for this pinning-down function of καί.

[ back ] 112. Thiersch 1852:428 argues that ἤτοι never serves to introduce the first part of an antithesis (against Apollonius Dyscolus and the recent communis opinio, see also Ruijgh 1981), but in fact always serves to start something new (asseverat) with a conclusive force from τοι. Ruijgh 1981 reads ἤτοι in narrative and ἦ τοι in direct speech, but Thiersch 1852:452-453 argues that we should read ἦ τοι everywhere in Homer.

[ back ] 113. Compare the use of interjections after direct speech in the Serbo-Croatian epic song discussed in Bonifazi and Elmer 2012a, with examples from PN 662, line 32“…” Hey, when the serdar’s company heard this on page 302 and line 476 “…” Oh, when Halil understood these words on page 304.

[ back ] 114. Iliad 15.437, 19.342, 23.785, and 24.243, see §§57-63 for a discussion of all four passages. Note that these cases are restricted to the Iliad, just like the instances of ἦ δὴ γάρ starting an embedded narrative. See IV.4 §§101-102 for a discussion of γὰρ δή in Herodotus and IV.4 §147 on ἤδη γάρ and γὰρ ἤδη in Herodotus and Thucydides.

[ back ] 115. Except in Hesiod (8 instances of δή in initial position) and later imitations of the Homeric style. See IV.3.11.3-4 for δή in μὲν δή and other combinations in Herodotus and Thucydides.

[ back ] 116. Metrically it would be quite at home at verse beginning, since the combination often forms a trochae within the verse.

[ back ] 117. For a discussion of further instances of δὴ γάρ, see §62.

[ back ] 118. Gesamtkommentar 2000:I.2.180 calls the narrative a paradigm, which serves to justify (“begründen”) Hephaestus’ warning.

[ back ] 119. The association of ἦ with affirmation is widespread in earlier scholarship, but Ruijgh 1971:192-194 first argues that it was actually the positive antithesis of οὐ (“Or, il nous paraît possible que dans la préhistorie du grec, ἦ ait eu la fonction et la valeur représentées plus tard par ναί ‘oui!’” 192), and his argument is followed by Sicking 1993:55. The problem of paraphrasing ἦ is well illustrated by Stephens 1837:43, when he paraphrases ἦ πολλά (Sophocles, Ajax 1418) as “Much, yes, much” [italics original], but does not put forward “yes” as one of the possible ways to render ἦ. “Yes” + repetition happens to be a way to emphasize a word in English (which is what Stephens wants to render), but this does not mean he regards ἦ and “yes” as equivalent in that passage.

[ back ] 120. See Janko 1992:190-191 for the textual problems in this line, especially regarding ἄλλο τεή.

[ back ] 121. See especially Cuypers 2005 for Homer, who argues that ἦ almost exclusively occurs in direct speech. This is a problematic statistic, however, because ἦ is a typical lexical item that is subject to a lot of editorial interference. For example, there is no consensus about whether the string HTOI should be rendered always as ἤτοι, always as ἦ τοι, or one or the other depending on the context. Ruijgh 1981 reads ἤτοι in narrative and ἦ τοι in direct speech, but consider Thiersch 1852:452-453, who argues we should read ἦ τοι everywhere in Homer. Murray’s Loeb text gives ἦ τοι everywhere, while Van Thiel 1996 and West 1999 give ἤτοι in narrator text in the Iliad; but compare Garvie 1994 and Steiner 2010 (with note on page 96) who consistently print ἦ τοι in their Odyssey texts.

[ back ] 122. The instances of ἦ δή in Homer are: Iliad 1.518, 1.573, 2.272, 2.337, 14.53, 15.467, 17.538 (ἦ δὴ μὰν), 21.583 (ἦ δή που), 24.518, compare moreover 17.629 ἤδη μέν (ἦ δὴ μέν N; West does not note the variant in his apparatus); Odyssey 1.253, 5.182. Compare also ἦ καί at the start of direct speech in Iliad 6.441 and Odyssey 1.158, 21.131; ἦ μάλα δή in Iliad 6.518, 8.102, 11.441, 15.14, 22.229, 22.297, 22.373 and Odyssey 1.384, 4.169, 4.333, 4.770, 5.286, 9.507, 11.436, 12.297, 13.172, 13.383, 15.486, 17.124, 17.264, 22.151, 23.149; ἦ μέγα θαῦμα in Iliad 21.54.

[ back ] 123. Compare the act-initial δὴ πάμπαν in Iliad 19.342, see §60-§61.

[ back ] 124. See IV.3.4 for the relation of prosody to punctuation in ancient and medieval scholarship.

[ back ] 125. Although even the position of που seems to have caused some confusion in 21.583: Herodian notes that some manuscripts wrongly (οὐκ εὖ) read ἤδη here.

[ back ] 126. Only by the first-position word ἀλλά followed by οὐ: Iliad 5.895, 17.41, 17.448, 23.441.

[ back ] 127. There are no instances of μάν δή, but this may be the result of the fact that μάν, μήν, and μέν before consonants were normalized to μέν. There are 59 (33+26) instances of μὲν δή in Homer, of which nine are ἦ μὲν (or ἠμὲν) δή: Iliad 1.453, 2.798, 3.430, 7.97, 9.348, 16.362; Odyssey 4.33, 14.216, 18.257.

See Nägelsbach 1834: 153-176, Friedländer 1859a:820-823, Bäumlein 1861:153-154, and Wackernagel 1916:177-182 on μέν, μάν, and μήν in Homer, and specifically on the interchangeability of μέν and μήν/μάν in Homer see Nägelsbach 1834:159-167, Cobet 1875:365-367, and Leumann 1949:85-89. One of their conclusions is that μέν was probably written before consonants and μάν before vowels.

[ back ] 128. This would mean that there was one other construction which retained δή in act-initial position: δὴ μάν before vowel, as in Iliad 17.538 and ἤδη (ἦ δὴ?) μέν before consonant as in Iliad 17.629, with possible parallels in Iliad 20.187 ἤδη (ἦ δὴ?) μέν and Odyssey 24.506 ἤδη (ἦ δὴ?) μέν. In the other instances of ἤδη μέν in Homer, the adverb ἤδη looks like the best reading.

[ back ] 129. See also Scodel 2012:329, “ἦ (…) most often appear[s] at this boundary where an individual, internal judgment meets the outside world—or the judgment of others.”

[ back ] 130. See Thiersch 1852:424-440, Thomas 1894:81-85, Smyth 1956:649, Slater 1969:223; compare also Kühner 1835:391 where he argues that ἦ μήν introduces an autonomous expression.

[ back ] 131. See V.ἦ passim; most recently, Sicking 1993, Van Erp Taalman Kip 1997, Wakker 1997b, Cuypers 2005, and Caspers 2010 have discussed the particle in terms of the interaction between speaker and hearer. See ΙΙΙ.2.2.10 and III.4 §47 on the function of ἦ in tragic and comic dialogue, IV.4.3.1 for ἦ μήν and IV.4.8 for ἤδη linked to speaker involvement.

[ back ] 132. See Humbert 1960:406, who imagines the speaker talking to himself: “comme qui se dirait: ‘Oui, c’est bien ainsi’” (“as if someone said to himself: ‘Yes, it is really like that.’”).

[ back ] 133. See Schwyzer and Debrunner 1950:II.564, Ruijgh 1971:192, Chantraine 2009²:387, Beekes 2010:I.507.

[ back ] 134. Although Nordgren 2012 attempts to establish the semantics of interjections in ancient Greek drama.

[ back ] 135. See Biraud 2010 for a discussion of the possible prosodic realization of ancient Greek interjections, compare Norrick 2009a:868-869 for the prosodic quality of interjections in English.

[ back ] 136. Ameka 1992, with the definition on page 105; see also Nordgren 2012:8-15 for the discussion of the term as applied to ancient Greek.

[ back ] 137. Ameka 1992:108.

[ back ] 138. As apparently imagined by the scribe of Venetus A at Iliad 15.467, see §38.

[ back ] 139. Given the apparent different linguistic layers in the Homeric corpus, it may even be a direct representation of the diachronic development, with different uses of ἦ reflecting different stages in the creation of the text. However, we insist that it is not problematic to assume that a certain lexical item is used in different ways at one moment in time, with the native speakers often not even conscious of the differences between them, or, conversely, of their common ancestry; see Koier 2013:19-23 for more on this concept.

[ back ] 140. For a discussion of the different functions of δή in Homer see §§53-64.

[ back ] 141. See Fränkel 1965 for more on the vocative and act boundaries (Kolon-boundaries in his terms).

[ back ] 142. Nordgren 2012:52-55.

[ back ] 143. Of course, they share this characteristic with discourse markers or pragmatic markers (see Norrick 2009:870), under which many particles may be subsumed.

[ back ] 144. Compare the discussion of καὶ γάρ in the Homeric epics in §§30-31.

[ back ] 145. This includes the start of an associative narrative in Iliad 6.414 ἦ τοι γάρ, 19.100 ἤτοι ὁ γε, as well as instances where ἦ τοι introduces indirect speech or indirect thought, as in also Iliad 15.699 ὅδ᾽ ἦν νόος | ἦ τοι Ἀχαιοὶ and Odyssey 5.383: αὐτὰρ Ἀθηναίη κούρη Διὸς ἄλλ᾽ ἐνόησεν / ἦ τοι τῶν ἄλλων; compare IV.4.3.1 (ἦ μήν) and IV.4.8 (ἤδη).

[ back ] 146. The corpus of these instances is not complete. There are several other instances of ἤδη at move beginnings, and probably others of ἦ + X as well; consider for instance ἤδη νῦν at Odyssey 10.472, 15.65, 16.168, and perhaps ἤδη τοι in Odyssey 22.101. For now I have focused on ἤδη γάρ and ἦ δή in the manuscripts.

[ back ] 147. See note 130 above.

[ back ] 148. I maintain this point in spite of Norrick’s remark (2009a:868) that in his sizeable English corpus “many primary interjections do not express emotions, as is often maintained of interjections generally, but rather information states.”

[ back ] 149. Note, for example, how often ἔνθα occurs after an indentation in modern editions, and even at the very beginning of book 5 of the Iliad. Bonifazi 2012:283-284 hints at a similar interpretation of ἔνθα as a discourse marker.

[ back ] 150. See Bakker’s separation between “performance time” and “story time” 1997b:68.

[ back ] 151. Compare the function of the priming acts described in II.2.5.

[ back ] 152. For more on such “when clauses” see Bakker 1991.

[ back ] 153. See e.g. Ford 1997.

[ back ] 154. The construction ἦν or ἔστι followed by a geographic location recurs in Homer, both with and without particles, see Gesamtkommentar 2003:II.2.262 ad 2.811 for all parallels and literature. The other asyndetic constructions with ἦν or ἔστι are: Odyssey 7.244, 9.116, 15.403, 19.172. Givón 2005:131-133 discusses the following constructions in English: “There’s this guy…” and finds that the “cataphoric persistence” of those referents in the following discourse is very strong. That is to say that referents introduced through this kind of “existential presentative construction” tend to persist as discourse topics in the following discourse. Recently, Auer and Maschler 2013 have described the narrative function of VS (verb subject) constructions in spoken stories in Hebrew and German, but they do not discuss the construction with existential verbs.

[ back ] 155. A parallel instance is Iliad 9.529 Κουρῆτές τ’ ἐμάχοντο καὶ Αἰτωλοὶ μενεχάρμαι, but there the story is announced more explicitly by the verbum dicendi ἐρέω (I will tell, 528). This verb is in turn a metanarrative comment by the secondary narrator, Phoenix, who is telling this story.

[ back ] 156. Translation Race 1997, order adapted.

[ back ] 157. See Gentili 2013:421. Slater 1983 discusses parallels for this in Homer, with a focus on those narratives introduced by a relative pronoun + ποτε; see above §§21-22. Compare also Pindar Olympian 3.29, 6.75, 7.30, 9.9, 10.104, 13.63; Pythian 1.16, 4.20, 4.107, 4.152, 9.15, 10.31, 12.6; Nemean 4.25.

[ back ] 158. τάν ποτε makes up the final dactyl of the third verse of the epode, Snell/Maehler 1971:11 regard this as a separate metrical colon.

[ back ] 159. See Xanthou 2007 on asyndeton in Pindar, and below §§65-76 for transitions in Pindar involving particles and other markers.

[ back ] 160. Compare IV.3.4 for Nicanor’s contention that asyndetic transitions were accompanied by the longest pause in his system, of four khronoi.

[ back ] 161. This reading of ἦ is again quasi-interactional, in this case relevant to the stance of the narrator: compare ΙΙΙ.2.2.10 and III.4 §47 (ἦ in tragic and comic dialogue), IV.4.3.1 (ἦ μήν), and IV.4.8 (ἤδη).

[ back ] 162. Although generally translated “thus”, ὧδε in fact represents a more complex meaning that often marks particular relevance of this statement for the zero-point of the utterance, the here-and-now (Bonifazi 2012:85). The tensions between τότε “then” and ὧδε “in this way here” represent how the story about Odysseus in the past has direct relevance for how he is now being perceived from the walls of Troy.

[ back ] 163. The repeated ἀλλ᾽ ὅτε δή serves to mark new points in a report or narrative (Gesamtkommentar 2009:III.2.85).

[ back ] 164. Out of 106 instances of ἀλλ᾽ ὅτε δή in the corpus, there are only three passages where four instances are found so closely together: here, in Iliad 6.172-200 (the embedded narrative by Glaucus to Diomedes) and Iliad 10.338-365 (the embedded narrative of Diomedes’ and Odysseus’ pursuit of Dolon); there is a cluster of three instances in Odyssey 14.287-301, but more frequently it occurs in twos: Iliad 5.773-780, 11.170-181, 17.728-732, 23.768-773, Odyssey 3.269-286 (with ἀλλ᾽ ὅτε in line 278), 4.514-519, 10.144-156, 12.329-335, 12.399-403, 14.472-483, 15.457-477.

[ back ] 165. In Iliad 10.366 τότε δή introduces the climax of the story, compare Iliad 5.775 ἔνθ᾽, 11.182 τότε δή ῥα, 23.774 ἔνθ᾽, Odyssey 3.288 τότε δή, 4.520 ἄψ δέ, 10.145 καὶ τότ᾽, 10.157 καὶ τότε, 12.405 δὴ τότε, 14.303 δὴ τότε, 14.484 καὶ τότε, 15.478 μὲν ἔπειτα.

[ back ] 166. In III.3 §§76-83 and III.5 §§51-63 γε is connected to a speaker’s emotional involvement.

[ back ] 167. The frequencies are based on an analysis of four books of the Iliad and four books of the Odyssey (4917 lines). In the Iliad it occurs 2,7 times per 100 lines of narrator text, and 6,4 times per 100 lines of direct speech. In the Odyssey, it occurs 2,9 times per 100 lines of narrator text, and 6,5 times per 100 lines of direct speech.

[ back ] 168. Edwards 2002:38-61 makes a number of important observations considering narrative transitions in Homer. His argument is that the transitional constructions in Homer have the effect of avoiding narrative breaks (see especially 58).

[ back ] 169. See Devarius 1588:63-64, Hoogeveen 1769:276, Hartung 1832:245-246, Nägelsbach 1834:48 and 62, Kühner 1835:386, Ellendt 1835:166-167, Döderlein 1858:362, Wetzell 1879:14, Ebeling 1885:291, Thomas 1894:85, Navarre 1904b:93-94 (but he argues against the idea in 1932:667-679).

[ back ] 170. See Schweighaeuser 1824:150, Heller 1853:277, Döderlein 1858:362, and Bäumlein 1861:98.

[ back ] 171. On the broad application of δή in Hellenistic and later Greek see V.δή passim and I.2 §35.

[ back ] 172. The frequencies for δή are based on an analysis of 12 books of the Iliad (13-24) and 12 books of the Odyssey (1-12), and the numbers are the following: in the Iliad 1,7 instances of δή per 100 lines of narrator text and 5,1 per 100 lines of direct speech, and in the Odyssey 3,8 per 100 lines of narrator text and 4,7 per 100 lines of direct speech.

[ back ] 173. Overall, in the Iliad 39.9% of δή instances are with a temporal marker, while this number is 58.3% in the Odyssey.

[ back ] 174. In narrator text, in the Iliad 75% of δή instances occur with a temporal marker, 25% without; in the Odyssey this is 84.3% with and 15.7% without temporal marking.

[ back ] 175. In direct speech, in the Iliad 23.1% of δή instances occur with a temporal marker, 76.9 % without; in the Odyssey this is 38.4% with and 61.6% without a temporal marker.

[ back ] 176. The exceptions are Olympian 9.9, Pythian 11.17, and Isthmian 2.27, discussed in II.5 §70n194, and Olympian 10.60, Nemean 5.15, Nemean 10.76, where δή occurs in (indirect) questions. In the latter three instances I take δή as intensifying the entire act.

[ back ] 177. For a discussion of the peninitial position see II.2 §12.

[ back ] 178. Murray translates “Now when…,” which has a similar function of moving the discourse along in English, semantically independent of the temporal value of “now.”

[ back ] 179. This is how Denniston 1950:219 reads such instances: δή used with “[r]elative temporal adverbs, ‘precisely when’, ‘just when’.” His reading is followed by the comment on this passage in Gesamtkommentar 2009:III.2.85.

[ back ] 180. “Demetrius” says in his treatise on style (Pseudo-Demetrius of Phaleron Style 56) that in Iliad 21.1 δή is used to mark a new beginning, and that, if the combiner had not been used, the reader might have thought Homer was still talking about the same thing.

[ back ] 181. Bestgen and Vonk 1995, especially 20-21.

[ back ] 182. See also Nemean 8.19 ἵσταμαι δή, where the discursive discontinuity is implicit in the present tense following upon a narrative act in the aorist; in Nemean 10.75 θερμὰ δὴ τέγγων likewise the function of δή is less to intensify than to mark the narrative progression; Isthmian 8.65 ἐνίκασε δή ποτε may represent this function too, or it may represent δήποτε, a result of grammaticalization in which the added value of δή has all but disappeared “once upon a time” (LSJ). See IV.4.5.1 on δή marking narrative steps in Herodotus.

[ back ] 183. Ferretti, Kutas, and McRae 2007.

[ back ] 184. See Bakker 1991 for the narrative function of temporal clauses in Herodotus, Buijs 2005 for temporal and participial clauses in Xenophon, and Muchnová 2003 for the narrative function of ἐπεί clauses in Homer.

[ back ] 185. See II.2.2.1.

[ back ] 186. Compare the three instances of ὅτε δ(έ): Ιliad 16.690, 17.178, 19.134. Especially the last of the three is a good comparandum for ὅτε δή clauses in narrative transitions. In the literature, the relation between the two particles is without exception regarded as a development from δή to a (weaker) δέ. However, the possibility of δή being a prosodically strengthened form of δέ cannot be excluded. This would be a situation comparable to the possible development of ἄρα to ἆρα in second position (different from ἆρα in initial position) as attested in Pindar and drama (see Braswell 1988:173-174 and De Kreij 2015).

[ back ] 187. See also Pindar, Olympian 3.25.

[ back ] 188. The combination τότε δή serves a similar function, and illustrates that in its intensifying function too δή can occur in peninitial position.

[ back ] 189. Bakker 1997b:79 notes that καὶ τότε marks “significant events or breaks in the story”; for this passage, see Graziosi and Haubold 2010:127, “καὶ τότε μιν (…) emphasizes that we have reached a crucial point in the story” and Gesamtkommentar 2008:IV.2.70. See IV.2 §§100-101 on καὶ δή at narrative peaks in Herodotus.

[ back ] 190. Bakker 1997b:75.

[ back ] 191. Earlier scholars have linked δή to perception; see Döderlein 1850:III.362, Thiemann 1881: 530-531, Paley 1881:21, Sicking 1986:133. A similar link between δή and perception is drawn in III.2 §§73-79 for tragedy and comedy and in IV.4.5.2 for historiography. Scholars have also spoken of δή in terms of “evidentiality,” see Bakker 1997b:78-79, Van Ophuijsen 1993:141 and 146 “self-evidential,” Cuypers 2005:38 and 55-59, De Jong 2007:14-15, Van Erp Taalman Kip 2009:114. Wakker 1994, 1995, 1997b argues that it more generally emphasizes the importance of the utterance (1994:351 “δή draws special attention to the (…) proposition”).

[ back ] 192. See further Bakker 1993b:15-25 and 2005:146.

[ back ] 193. Bakker 1997:79.

[ back ] 194. The third example follows below as (t23). There is one other puzzling position of δή, in Iliad 24.243 ῥηΐτεροι γὰρ μᾶλλον ?|? Ἀχαιοῖσιν δὴ ἔσεσθε; this instance is unique and as yet unclear. The scribe of the Venetus A manuscript chose to place the particle between commas, apparently as a parenthetical discourse act: ῥηΐτεροι γὰρ μᾶλλον Ἀχαιοῖσιν, δὴ, ἔσεσθε.

[ back ] 195. This line seems to have been a textually problematic place: British Library add. mss. 17210 reads Ἀντίλοχος δ’ ἄρα οἱ λοισθήϊον ἔκφερ’ ἄεθλον, while a papyrus (P.Lond.Lit 27) has δή added above the line in a second hand.

[ back ] 196. There are seven other instances of the combination ἄρα δή, all in the Iliad, and in most cases I believe that δή has scope over the entire act (the one possible exception is Iliad 17.85 δὴ πρίν).

[ back ] 197. Gesamtkommentar 2009:VI.2.147 notes the remarkable position of δή, and takes the particle to have scope over the entire utterance: “[δή] verleiht der folgenden Aussage bes. Emphase.”

[ back ] 198. Denniston 1950:204 calls this the “emphatic” use, and the intensifying function I posit matches in particular Denniston’s decription of its use to denote “that a thing (…) is very much so.” Moreover, passages in tragedy (Denniston 1950:212) show that small-scope, forward-looking δή could still occur: δὴ μάλιστα. See also Pindar Nemean 8.51 δὴ πάλαι in act-medial position “really long ago.”

[ back ] 199. Denniston 1950:212 regards this positioning as secondary (“Originally, perhaps, δή was regarded as going with the preceding word.”), whereas I believe that the material suggests that the construction where δή precedes its host disappears from Greek quite early on. The formation δὴ ποτέ (sometimes δήποτε) may invite several explanations, but the simplest one is that in the string at act beginning ποτε would naturally follow δή (ποτε δή only occurs twice, in the construction εἰ ποτε δή). When it is the accented interrogative πότε it does shift to initial position, and may be followed by δή.

The following are the relevant instances of the combination ἦ δή where δή intensifies an adjective at one end of a scale: Iliad 1.518 (=1.573) ἦ | δὴ λοίγια, 2.272 ὢ πόποι | ἦ | δὴ μυρί᾽ (compare Euripides Heraclids 331 πόνους δὴ μυρίους), 15.467 ὢ πόποι | ἦ | δὴ πάγχυ, 24.518 ἆ δείλ᾽ | ἦ | δὴ πολλά, Odyssey 1.253 ὢ πόποι | ἦ | δὴ πολλόν, 5.182 ἦ | δὴ ἀλιτρός γ᾽; I believe that ἦ in these instances is at least partly interjectional (§33-§43). The group may be expanded with Pindar Olympian 6.79 πολλὰ δὴ πολλαῖσιν; see also Archilochus fragment 172.3 νῦν δὲ δὴ πολύς.

[ back ] 200. The construction that illustrates this best is εἰ δ᾽ ἐτεὸν δή: Iliad 7.359, 12.233 and Odyssey 23.107. This is the standard order in later Greek; in Thucydides and Herodotus it occurs frequently just after superlatives, e.g. μεγίστη δή (with δή in act-medial position) in Thucydides, see IV.4.6.2.

[ back ] 201. This happens particularly with forms of πρῶτ-, πᾶς*, and πολ-; the loci are Iliad 1.235, 1.545, 2.117, 4.97, 7.207, 9.24, 9.348 (ἦ μὲν δὴ μάλα πολλά), 10.173, 11.219, 11.559, 11.825, 13.275, 14.187, 14.509, 15.291, 15.616, 16.23, 16.113, 16.198, 16.538, 17.427, 18.103, 19.9, 19.54, 19.342, 23.490, 23.607, 24.65, 24.167, 24.713; Odyssey 3.183, 4.414, 5.76, 5.300, 6.227, 7.134, 8.131, 8.282, 13.155, 14.149, 14.289, 15.401 (ὅς τις δὴ μάλα πολλά), 16.340, 16.469, 17.174, 17.217 (νῦν μὲν δὴ μάλα πάγχυ), 22.195 (=17.217), 22.440, 22.457, 23.49, 24.528.

[ back ] 202. Hymn to Demeter 76, 148, and 159; Hesiod, Works and Days 417 and Fragment 204.96 (in both instances δὴ γὰρ τότε).

[ back ] 203. As I argue for καὶ γάρ, §§30-32. Alternatively, one could read it as δή intensifying the force of γάρ, as Denniston 1950:243 does: “The reverse order, δὴ γάρ, which gives an even stronger emphasis [sc. on γάρ], is also frequently found in [Homer].” However, what is intensified is not the relation between the current act and the preceding, but the current act in itself.

[ back ] 204. Iliad 13.122 and 15.400, both have the form δὴ γὰρ μέγα νεῖκος ὄρωρεν. A similar reading could be proposed for Hymn to Demeter 76, 148, and 159.

[ back ] 205. See II.4 §19 for the function of γάρ in δὴ γάρ.

[ back ] 206. Denniston, conversely, insists that δή always modifies one word, which tends to be, but does not have to be, adjacent to the particle. This would mean that in this example δή “modifies” παρίσταται, probably. Although this reading makes sense, it does not recognize the difference in the Greek: it would be very hard to mark that δή is supposed to be construed with a word that occurs much later in the act.

[ back ] 207. Bakker 1997b:75-76 reads this instance of δή as creating involvement through a shared visible reality: “Dḗ conveys that the consciousness verbalized receives its input from the speaker’s immediate environment, from what is perceptually clear and evident.”

[ back ] 208. See I.1 on the link between position and scope.

[ back ] 209. In Homer, δή in peninitial position has an intensifying function with act scope in constructions like εἰ δή (also in Pindar Olympian 1.54) and ἦ μάλα δή (also in Pindar Pythian 4.64).

[ back ] 210. It is in the latter category that we should place Pindar, Olympian 13.99 ἑξηκοντάκι δή (Race translates “full sixty times”), Pythian 4.273 δυσπαλὲς δή “difficult indeed,” Pythian 9.91 τρὶς δή “full three times” (Race), Nemean 1.17 θαμὰ δή “often indeed” (Race), and Nemean 8.48 δὶς δή “full twice.”

[ back ] 211. See the observations in IV.3.11.3-4 and IV.4.5.

[ back ] 212. There is one specific aspect of δή as yet undiscussed, which is the use of the particle after pronouns; see II.5.3.4.

[ back ] 213. Here too the narrative is introduced by τόν ποτε; see §48-§50 for more on such asyndetic narrative beginnings.

[ back ] 214. See also Braswell 1988:284-285 (ad Pythian 4.202-203), but he wants to differentiate between those instances where δέ marks temporally sequential steps, and those where these steps are not temporally sequential. Against the background of the discussion of δέ in II.2 §§31-36, I do not believe such a differentiation is productive.

[ back ] 215. The construction is reminiscent of the Homeric ἦ ῥα, καί after direct speech, but represents an interesting variation on this theme. In my reading (as in Race’s) ἦ is not the verbum dicendi, but the particle ἦ: “Such were the verses of Medea’s speech” (Race). Gentili et al. 1995:444 and Braswell 1988:138-139 take ἦ as the verb of speech, with στίχες as subject in a schema pindaricum (singular verb with plural subject); Segal 1986:154 proposes a similar reading.

[ back ] 216. The edition of Snell/Maehler only has twenty-five, but five instances should be added where Snell/Maehler give ἦρα (ἄρα mss.). I follow Braswell 1988:173-174, who reinstates Boeckh’s reading of ἆρα as a prosodically lengthened form of ἄρα; see also De Kreij 2015. This gives a total of thirty instances of the different forms of ἄρα.

[ back ] 217. In Homer, there are between 8 and 10 instances of ἄρα per 100 lines of narrative, versus between 2 and 3 instances per 100 lines of direct speech. In Pindar, throughout the Odes ἄρα occurs 0.9 times per 100 lines, while in Pythian 4 the seven instances represent 2.3 occurrences per 100 lines.

[ back ] 218. Out of the 30 instances in the Odes, in only three cases does ἄρα clearly occur outside of a narrative context: Isthmian 5.41, Olympian 10.52, Nemean 8.32.

[ back ] 219. For Pindar, 9 out of 30 instances of ἄρα are in speech-capping phrases (in one instance it caps indirect speech: Olympian 6.52); in Pythian 4 it is at line 57, 156, and 232. This makes up more than a third of a total of 22 speech-capping phrases in the Odes.

[ back ] 220. See Braswell 1988:214 for the Homeric origin of this convention; the number of days is not fixed.

[ back ] 221. See II.5 §35 and §46-§47 on ἀλλ᾽ ὅ γε.

[ back ] 222. Compare ἄρα at the end of the arrival scene in line 121 and rounding off the gathering of the Argonauts in 189. Gentili et al. 1995:465 and Braswell 1988:220 rather read ῥα as marking immediate succession, following Denniston 1950:42-43.

[ back ] 223. More on particles in the similes in II.4.3. In that same chapter the higher frequency of the particle in Homer is discussed. This function of ἄρα with regard to discourse articulation is surprisingly absent in the scholarship on the particle, most notably in Denniston.

[ back ] 224. See II.5.3.3.

[ back ] 225. See §48-§50 above.

[ back ] 226. Race uses all of eleven commas in his translation until he comes to the full stop after “trident” (12).

[ back ] 227. Lit. “horses that fight in iron-mail,” see Gildersleeve 1885:256.

[ back ] 228. I am aware that ἀγγελίαν is an apposition of μέλος, but the addition of <with> is the most economical way of translating this line into comprehensible English without interfering too much with the Greek word order.

[ back ] 229. See Lauer 1959:54-58, Race 2002, and Markovic 2006:138-140.

[ back ] 230. Ortygia is the little island on which Syracuse was built, just off the coast of Sicily; see Gentili et al. 1995:366-367.

[ back ] 231. For the pragmatic effect of using the second person see II.2.5.2.1.

[ back ] 232. The argument can be made that we should read it as an anastrophe (or as an adverb), in which case the accent should be written on the first syllable: ἔπι. Regardless of this editorial decision, we may assume that a preposition that is not followed by a noun and a pre-verb in tmesis were pronounced with a different intonational contour to avoid ambiguity.

[ back ] 233. See Gentili et al. 1995:368.

[ back ] 234. Dancygier and Sweetser 2000:120-122; see also Slings 1997:104-106.

[ back ] 235. See §§22-29 above on γάρ in Homer: γάρ can also introduce an example following a general claim.

[ back ] 236. Compare again Sicking’s claim that “sentences introduced by γάρ (…) provide answers to all sorts of questions raised by the speaker’s utterances,” in Sicking 1993:23; if we substitute “moves” for “sentences” in this statement, I would be inclined to agree.

[ back ] 237. This aspect of τε is explored in IV.2.3.1 and II.4.3.1, and particularly for Pindar in II.4.5.

[ back ] 238. See II.2.2.1 for more on δέ in Homer, and §65 above for δέ in Pindar.

[ back ] 239. Compare the discussion of independent thoughts inserted by means of δέ in Herodotus and Thucydides in IV.2 §§26-28.

[ back ] 240. Compare the use of δέ in a narrative section of Pythian 6 (three instances of δέ in a narrative, all mid-verse), discussed in II.2 §43, or the narrative section from Pythian 4 (t24), where in ten lines δέ occurs five times (25, 28, 32, 34, 34), of which four are in verse-medial position, before a stronger transition with a first-person verb followed by δέ in verse-peninitial place (38, πεύθομαι δέ).

[ back ] 241. Compare the discussion in II.2 §55, and further Olympian 6.41, 7.69, 7.88, 8.67; Pythian 2.15, 2.31, 4.53, 4.154, 11.31, 11.46; Isthmian 2.37. A metrical boundary may be compared to what is known as a “Transition Relevance Place” in Conversation Analysis: a point where a switch of speaker may occur. In Pindar, it is a likely place for a change of discourse topic. The use of μέν to signal the continuity of discourse topic across the metrical boundary may be compared to the use of μέν to “hold the floor” across TRPs in tragic and comic dialogue; see III.4 §28-§30.

[ back ] 242. See Burton 1962:116-119 on the story of “one of the great sinners of Greek legend” (116).

[ back ] 243. See II.5 §19 on finite verb + δέ to mark continuity of grammatical subject.

[ back ] 244. Of the eight instances in the Odes, only one (Isthmian 8.16) does not introduce a generic piece wisdom. This generalization may be extended to most instances of clause-initial χρή; see on this instance Gentili et al. 1995:378.

[ back ] 245. See Gildersleeve 1885:267, “the fortunes of those whisperers.”

[ back ] 246. Pindar argues that man should bear the fate he is given, without attempting to overreach. It might be read as advice for Hieron, or as an illustration of Pindar’s own attitude toward life, and toward his host Hieron; see Gentili et al. 1995:405.

[ back ] 247. Gentili et al. 1995:405 calls it a “final prayer” (“preghiera finale”).

About

Fellowships

The Library

Programs