IV.4 Tracking voice and stance

4.1 Introduction

§1. Characters’ speeches and deliberations are an indispensable part of Herodotus’ and Thucydides’ historiography; without these elements much of the Histories’ essence would be lost. Regarding Thucydides, Dover (1988:74) characterizes most of his work as “a relation of what people did, said and thought.” He goes on to say, “Delete the extensive and elaborate presentation of what they said, and the most striking of all the distinctive features of the work would be gone. Delete all reference to their thoughts, wishes, fears and the like, and the character of the narrative would be significantly changed.” Herodotus shows a kind of subtlety about his characters’ introspection that is different from but not inferior to Thucydides’ method.
§2. Secondary literature acknowledges not only the importance but also the artistry of both historians’ representation of speeches and thoughts. The present chapter addresses a neglected part of this artistry, namely the way in which the two historians use particles to represent their own and other people’s voices and stances. A fundamental preface concerns commonalities and differences between the key notions selected for this purpose, that is, “voice” and “stance.”
§3. “Voice” has been variously interpreted in linguistics and anthropology: it may refer to anything from persona loquens to the ideas of agency, self-representation, ethos, or identity. [1] In this chapter I consider voice across a discursive continuum ranging from fully reproduced speeches to summaries reporting the contents of an individual’s speech. Behind the latter is the authorial voice, which needs to be taken into account as much as the characters’ voice. I distinguish between explicit and implicit authorial statements, the former being characterized by “I” markers where the referent can only be the author.
§4. “Stance” is generally taken as a heading that encompasses evaluation, judgment, attitude, point of view, and perspective. The notion implies, on the one hand, agents who express attitudes and evaluations, and on the other, objects of regard. Interactional linguistics, conversation analysis, corpus linguistics, discourse analysis, and sociocultural linguistics use the term “stancetaking” alongside “stance” because it conveys that stance results from an action that is situated and has an interactional character. [2] Throughout the chapter I refer to stance as the overall phenomenon of stancetaking; for the analyses I adopt the definition of stance by the linguist John Du Bois.
§5. The two concepts of voice and stance are decidedly linked, but also distinct. Voice encompasses all the expressions of a person’s individuality, stance included. But what is voiced is not necessarily an act of stancetaking. At the same time stancetaking acts are not necessarily (represented as) uttered aloud. However, attributing stance to someone hints at the presence of at least an inner voice connected to the person in question, which is further connected to our auditory imagination of his/her words. [3] I assume that any expression of stance presupposes the existence of someone’s voice—whether re-performed, echoed, or in the form of inner voice—whereas not all the manifestations of voice deal with stancetaking. Therefore, I take voice as a superordinate concept.
§6. Voice and stance are both indispensable for understanding the historiographical methods of Herodotus and Thucydides. Their narratives are driven by a concern with managing access to the voice and stance of political figures, groups of citizens, and prominent individuals. Not only does this approach enhance the vividness of the events being narrated and heighten the reader’s sense of involvement in them, but it also allows the authors to detail the actors’ points of view and personal styles, to illustrate their decisions and motives, [4] and to signpost crucial moments in the progress of the narrative.
§7. I focus on stance instead of “focalization” because stancetaking is a social activity. Speaking subjects take stance not in a vacuum but in a social environment, in front of or in relation to other subjects. The concept naturally leads to considerations of multiple subjects involved, and multiple vectors linking those subjects. Just as for the analysis of the source of and the relations between voices, the analysis of the source of and the relations between stancetakings is not an easy task. Stance directly involves personal attitudes, perceptions, calculations, and feelings, but it does not always manifest in a direct way. Unless explicit verbs of thinking or the like appear, attributing thoughts and considerations to characters or to the creators of the characters can be challenging. In addition, literary texts play with the further level of interaction (real or imaginary) between authors and readers/listeners, which amplifies the possibilities for mirroring, juxtaposing, and manipulating the representation of stance. I choose stance because it indexes this complexity.
§8. A fundamental complexity regards “voice” too, if we acknowledge that multiple voices can be “heard” through the same utterance. Polyphony has wide and deep implications in literary texts. Rather than “polyphony” in the sense of interweaving several opinions and accounts (Herodotus’ λόγοι) into one historiographical work, here I refer to the more specific sense of polyphony as delineated first by Voloshinov 1973 and Bakhtin 1981. The core idea is that a specific stretch of text can be designed to allow for multiple voices to utter it at the same time. [5] The blending of stancetakings by multiple subjects that I consider in this chapter relies on the possibility that expressions of multiple voices can overlap.
§9. In Herodotus and Thucydides, a wide spectrum of linguistic features can signal voice and stance. They range from obvious forms such as first person markers, deictic demonstratives and adverbs (τόδε, νῦν), evaluative expressions such as δοκεῖ μοι, the semantic cues of verbs of saying, thinking, considering, wondering, etc., the construction ὡς + participle, to less obvious forms, such as reflexives and indirect reflexive pronouns (αὑτός and σφ- + αὐτός forms), [6] comparatives, superlatives, and other scalar adjectives, the verb μέλλειν, [7] mood (e.g. oblique optative), and tense (e.g. imperfect vs. present). [8] Beyond lexical and syntactic cues, Ι include more large-scale cues given by narrative technique, such as the inclusion of direct and indirect speech, temporal manipulation, and even the amount of detail used in accounts of events. [9]
§10. In line with a major methodological point of our work, [10] particles signal voice and stance together with other features. Therefore, the best way to understand the contribution of particles to the representation of speech and thought is to observe constructions and patterns that co-occur and co-signify with them, including all the features in the list given above.
§11. As Weydt points out about German particles (1969:60-66), there are particles that lend themselves very well to the expression of communicative intentions and subjective attitudes. Along these lines, Wakker 1997b explores the functions of “modal particles” in Herodotus and Thucydides—that is, particles signaling the speaker’s or narrator’s “attitude towards the proposition he is presenting” (Wakker 1997b:215). The current chapter develops themes touched upon in this article, as it discusses the same particles (ἄρα, δή, τοι, and ἦ μήν) in connection with direct and indirect speech, and investigates whose point of view is conveyed behind the use of ἄρα and δή. I complement and extend findings that Wakker acknowledges as “preliminary“ (1997b:248), chiefly by establishing links with scholarly accounts of stance, by refining the question of who conveys stance, and by extending the analysis to include free indirect thoughts (that is, people’s thoughts that are not introduced by reporting clauses).
§12. As in other chapters, I preface textual analyses with theoretical, methodological, and literary observations that guide the heuristics as well as the conclusions to be drawn. In the first part of the chapter (4.2, 4.3) I focus on “voice.” 4.2 recalls general thoughts about casting others’ voices in the two masterpieces, and about the authors’ strategies to let their own voice emerge. Then, a few textual analyses (4.3) devote attention to particles and particle combinations that mark voice. τοι and γε in authorial statements are the core part of this analysis, together with ἦ μήν in indirect speech.
§13. In the second part (4.4) I introduce stance, the other core idea of the chapter, and apply its main features to study the language of Herodotus and Thucydides. I also explain why I adopt this framework instead of “focalization,” and I add considerations about the stance of readers and the issue of Herodotean irony.
§14. The third part (4.5 to 4.9) includes analyses of multiple passages where particles play a role in marking explicit or implicit expressions of stance. In particular, I attempt to disentangle the complexity underlying several uses of δή in both authors (4.5, 4.6), in connection to people’s sense of certainty. Further sections relate δῆθεν (4.7) and ἤδη (4.8) to stancetaking, through their links with δή. Finally (4.9), I argue that the use of ἄρα in the two Histories only partially relates to stance.

4.2 Tracking voice

§15. As I said in §5, I take voice as a superordinate concept. Voice covers the representation of speech and thought in all its possible variants. Before discussing the role of particles in tracking someone’s voice, let us recall first labels that are commonly used for the main variants, and then general considerations by recent scholarship on the representation of speech and thought in Herodotus and Thucydides.
§16. According to standard accounts, modern fiction employs “direct,” “indirect,” and “free indirect” ways to encode someone’s speech and thought. [11] Direct forms are signaled by quotation marks; reporting clauses may precede as well (“she said,” “she thought”). Direct thought differs from direct speech in that the reported words are not meant to represent words uttered aloud. Besides the lack of quotation marks, what makes speech and thought indirect is the occurrence of reporting clauses (this time explicitly present), and of backshift forms that replace the character’s perspective with the narrator’s (for example, from “Do I have to come now?” to “she wondered if she had to go at that very moment”). Free indirect forms, conversely, lack quotation marks and reporting clauses, and may show no backshifting. [12]
§17. Pascal (1977:8-9) sees “free indirect style” (after the original label “style indirect libre” in Bally 1912) as a hybrid form: it includes a mix of linguistic features that relate to the character as well as the narrator. An example is “He stopped. Was that the car he had seen here yesterday?” (Pascal 1977:8), where the question is formulated in such a way that it captures the immediacy usually associated with direct discourse, but at the same time retains the narrator’s standpoint.
§18. Sanford and Emmott (2012:189) recall the two main readings of this hybrid status. On the one hand free indirect discourse can be interpreted as a variety of “single voice” (that is, the narrator’s); on the other, it can be interpreted as a manifestation of “dual voice,” that is, the mingling of the character’s with the narrator’s. [13] The “single/dual” debate is central to the topics of this chapter. I will consider indirect speech and indirect thought “dual voice” devices. Some particles, in fact, seem to be used polyphonically, to mark the interplay between two voices. [14]

4.2.1 Speech and thought: A figured stage of voices

§19. Within classics, recent scholarship has had much to say about direct speech, indirect speech and thought, and explicit and implicit authorial statements in Herodotus and Thucydides. In this and the next subsections I bring the research done in classics in dialogue with the stylistic analyses that have been performed on modern texts.
§20. Dewald 1999 focuses on the “figured stage” that the initial narratives of the two Histories display by delving into the management of voices. In her view the historiographical discourse of both Herodotus and Thucydides is dialogic (Dewald 1999:247), but the two authors incorporate voices in different ways. From the beginning Herodotus reports the λόγοι of others and gives voice to their sources, in order to show how information has accumulated over time, and to warn his audience that accounts may be self-interested. The first chapters in Thucydides, conversely, focus on general processes that override individuality. It is not until later that Thucydides voices, with particular emphasis, the statements, fears, and thoughts of key persons and key groups, which he does as a way to embody political strategies and human tendencies.
§21. Dewald further finds that while Herodotus plays the role of the one who “gets told things and evaluates them from a distance” (1999:248), Thucydides provides an internal focus for political views that determine decisions and shape the world he wants to account for. Therefore in Thucydides the distance between author and character may collapse, which, as we will see, can be seen especially in the representation of thoughts.
§22. One of the devices employed to represent others’ voices is direct speech, which can perform a number of historiographical functions in the two authors. [15] Speeches represent the most straightforward manifestation of someone’s personality and opinion “there and then,” [16] whether they reconstruct the original words or are largely invented by the author. [17] As Tompkins states, Thucydidean speeches show “characters’ discursive choices” that “reflect different styles of thought” (2013:463). [18] However, as several scholars point out, through the voice of others both historians manage to convey their own views about the complexity of certain situations, or else to mark aspects of a situation that deserve closer attention from the reader. Therefore, to the extent that direct speeches (or parts of them) represent implicit tokens of the authorial voice, they instantiate the “dual voice” modality. After all, access to the voice of each character or group comes from the voice of the author. What matters in terms of particle use in direct speech, as we shall see, is that particles can help implicitly stress either the author’s disalignment with the speaking character’s point of view, or his alignment with that character.
§23. Indirect speech is a more explicit “hybrid form.” [19] The language employed may reflect different degrees of indirectness, from tenses, moods and pronouns that create the effect of direct speech, to backshifted forms and especially infinitive constructions, which underscore the distance from the reported voice. [20] The level of detail may vary as well: indirect speech is a “versatile mode of speech reporting” (Coulmas ed. 1986:5), which can range from faithful adaptations to summarizing paraphrases. Finally, and more crucially, using indirect speech allows the author to blur the boundaries between the thoughts of characters and his own thoughts. [21]
§24. Like indirect speech, which in both Histories is regularly introduced by reporting clauses, indirect thoughts of characters are regularly introduced by verbs of thinking, considering, pondering, etc. [22] The linguistic features included in the clauses attached to these verbs are of particular interest to me, as they can be securely attributed to the representation of others’ thoughts—and therefore to the representation of the imagined voice potentially uttering them. In general, in my analyses I will assume that verbs denoting thinking and the like signal indirect thoughts just as reporting clauses signal indirect speech.
§25. While the two historians do not seem to me to use free indirect speech (unlike modern fiction writers), they do employ the device of free indirect thought, the embedding of characters’ considerations, descriptions, assessments in third-person narrative without a specific linguistic signal (such as “he thought”). [23] In other words, the change of source is not necessarily marked by reporting clauses or participles. This is the difference in form from indirect thought. The features that become relevant in order to detect free indirect thought are pronouns, tense, and particles just as in the case of indirect thought. [24] In her discussion of “free indirect speech” in Herodotus and Thucydides, Wakker (1997b:224-227) also flags μέλλειν and indirect uses of reflexive pronouns to report the thoughts or words (she does not distinguish) of characters. In terms of voice, the “free” component does not change the fact that we can still “hear” the voice potentially uttering those indirect thoughts—especially if certain particles co-occur.

4.2.2 Authorial statements

§26. The most direct way for Herodotus and Thucydides to let their voice come to the fore is by inserting authorial statements explicitly. Such statements range from programmatic statements, where the two authors position themselves with respect to their predecessors or genre expectations, to ad hoc narratorial comments. In the latter case the speaking “I” evaluates beliefs or decisions, and aligns or, more typically, disaligns with sources of information or political exponents; in this sense, the author’s voice expresses the author’s stance. [25] The overt presence of the speaking “I” emerges more often in Herodotus than in Thucydides. [26] This is consistent with Thucydides’ preference for using covert means to give voice to himself, by embedding it within his characters’ thoughts. [27]
§27. Implicit authorial statements are utterances that, notwithstanding the absence of an explicit “I”-marker, are easily attributable to the author as the only possible source, due to their omniscient character. For example, De Jong (2004a:106) regards Herodotean prolepses about the end of a character’s success (e.g. Herodotus 7.213 about Ephialtes) as implicit authorial statements. The author’s evaluation of Ephialtes’ lack of success is expressed in terms of omniscient knowledge about his future. In Thucydides, there is a section where Pericles’ advice to the Athenians is reported: as they approach war they should be confident in their resources and the income they receive from their allies. An inserted γάρ clause provides an external (“unframed”) piece of information about the annual amount of money the Athenians would gather from the allies. [28] This information cannot be part of Pericles’ advice; as Hornblower (1994:137) comments, “the focalizer is really Thucydides himself.”
§28. However, especially in Thucydides the ways in which the authorial stance is conveyed in the two Histories exceed the modalities listed so far. Scholars find authorial stance implicitly revealed through different discourse strategies, ranging from local devices, such as the use of vicarious voices, [29] to global or large-scale devices, such as manipulation of time and space in narration, or artful juxtaposition of accounts across large blocks of text. [30] In the case of large-scale devices there is no inner voice to be recognized, as the author’s stance is expressed through global strategies rather than in local instantiations (that is, actual identifiable words).
§29. The discussion outlined in the last subsections (4.2.1 and 4.2.2) invite a few conclusive considerations. The speech and thought of the people is a significant subject of interpretation in literary criticism. Also, Dewald’s remarks connect the management of voices to the expression of points of view, epistemic and emotional states, and rational judgments. In other words, voice is presumed to be a vehicle for expressing stance. This is why in scholarship the analysis of voices mingles with the analysis of stance. Furthermore, the spots for the author’s voice in the two Histories coincide with the moments in which the authors take stance. Nothwithstanding these convergences, it is important to keep in mind that tracking voice does not conceptually coincide with tracking stance. The next section shows that particles can help detect somone’s voice, regardless of whether the voice in question is used to take stance about something or not.

4.3 The contribution of particles to marking voice

§30. Particles are borderline words, hugging the fence between explicit and implicit communication. Their communicative meaning often relies more on what they imply than on what they directly say. [31] In a number of cases their most direct equivalent in English seems to be a certain intonation or gesture. As with intonation and gestures, their presence or absence makes a difference: the use of particles in certain contexts, and the avoidance of particles in other contexts, can reflect linguistic and literary habits that are worth examining.
§31. I expect the distribution of particles in narrative sections of Herodotus and Thucydides to be more interesting than their distribution in direct speech: narrative sections indeed feature a complex range of ways in which multiple voices and thoughts are represented. This is why in the following paragraphs I discuss a few particles and particle combinations that occur outside of direct speech while direct speech is their usual environment.

4.3.1 ἦ μήν in indirect speech

§32. The works of Scardino (2007; 2012) and De Bakker (2007) have shown that indirect speech in both Histories varies greatly depending on the author’s intentions. [32] Despite the variation, a few features are generally constant: the occurrence of reporting clauses preceding indirect speech, and of reporting strategies within the speeches themselves, such as infinitive constructions. [33]
§33. ἦ μήν is one of the features that break the rules of backshift. Αs it is a formulaic combination that in Classical Greek occurs almost exclusively at the beginning of oaths, we would expect it to occur in direct speech. However, in Thucydides we find ἦ μήν in indirect speech as well, eight times. [34] Here is an example:
οἱ δὲ Ἀκάνθιοι, πολλῶν λεχθέντων πρότερον ἐπ’ ἀμφότερα, κρύφα διαψηφισάμενοι, διά τε τὸ ἐπαγωγὰ εἰπεῖν τὸν Βρασίδαν καὶ περὶ τοῦ καρποῦ φόβῳ ἔγνωσαν οἱ πλείους ἀφίστασθαι Ἀθηναίων, καὶ πιστώσαντες αὐτὸν τοῖς ὅρκοις οὓς τὰ τέλη τῶν Λακεδαιμονίων ὀμόσαντα αὐτὸν ἐξέπεμψαν, ἦ μὴν ἔσεσθαι ξυμμάχους αὐτονόμους οὓς ἂν προσαγάγηται, οὕτω δέχονται τὸν στρατόν.
Thucydides 4.88.1
The Acanthians debated long, with much said on either side, and then took a secret vote. Influenced both by the seduction of Brasidas’ offer and by fear for their crop, they decided by a majority to secede from Athens. They made Brasidas pledge fidelity to the oaths sworn by the Spartan authorities when they sent him out, guaranteeing the autonomy of any people he brought over as allies, and with that pledge given they admitted his army. (tr. Hammond)
This is the only occurrence of ἦ μήν in indirect speech where Thucydides also records the original words, previously spoken by Brasidas during his speech to the Acanthians: αὐτός τε οὐκ ἐπὶ κακῷ, ἐπ’ ἐλευθερώσει δὲ τῶν Ἑλλήνων παρελήλυθα, ὅρκοις τε Λακεδαιμονίων καταλαβὼν τὰ τέλη τοῖς μεγίστοις ἦ μὴν οὓς ἂν ἔγωγε προσαγάγωμαι ξυμμάχους ἔσεσθαι αὐτονόμους, … (4.86.1). Brasidas’ original words ἦ μὴν οὓς ἂν ἔγωγε προσαγάγωμαι ξυμμάχους ἔσεσθαι αὐτονόμους become ἦ μὴν ἔσεσθαι ξυμμάχους αὐτονόμους οὓς ἂν προσαγάγηται in a quite faithful repetition at the moment of the Acanthians’ decision. ἦ μήν suggests that the Acanthians fully re-perform, rather than simply repeat or recall, the content of the oath. Through this touch of vividness Thucydides prepares the way for the judgment he later makes on the general attitude of the Acanthians (4.108.4), that they were moved by “blind wishes” rather than “sound forethought” to follow Brasidas. [35] In a recent article on oaths in the two historians, Lateiner (2012:157) records that Thucydides cites far more oaths than Herodotus (269 to 58), as he “has a penchant for narrating artfully evaded personal agreements, the pathology of personal oath.”

4.3.2 τοι in Herodotus, in and beyond direct speech

§34. τοι occurs almost exclusively in direct speech, and grosso modo only in Herodotus. [36] The vast majority of them are datives of the second person pronoun, just as in Homer, and the act position can greatly vary. In nine cases τοι does not work as a second person pronoun. These instances always occur in act-peninitial position, together with other particles, and with εἰ or ἤν. Interestingly, all the co-texts and contexts suggest that the host discourse acts are part of persuasive assertions; the speakers’ intent while uttering τοι seems to be to persuade their dialogic partners about something. [37] My general conclusion about τοι in Herodotus and Thucydides is that, whether it grammatically works as a “you” form or as a discourse particle, it retains a strong interactional connotation, in that it links the host act to an appeal to the dialogic partner (“I tell you” is one of the translations proposed [38] ).
§35. One of the reasons why τοι occurs more often in Herodotean than in Thucydidean speeches may be that speeches in Herodotus are more dialogic. That is to say, they are arranged in dialogic sets, namely speeches that constitute dialogic turns, more than in Thucydides; [39] additionally, they involve more individuals talking to individuals, rather than collectives talking to collectives as in Thucydides. [40]
§36. In line with the dialogic nature of the single particle τοι, several particle combinations involving τοι in Herodotus’ and Thucydides’ works tend to occur in direct speeches. The only τοιγάρτοι (Thucydides 6.38.3) and two out of three τοιγάρ in Herodotus (3.3.3 and 8.114.2) are in direct speech; all five instances of τοίνυν in Thucydides, and 16 of 22 instances in Herodotus are in direct speech; [41] the same holds for 22 of 24 instances of καίτοι in Thucydides, and for 8 of 18 in Herodotus; all three instances of ἤτοι in Thucydides, and four of seven in Herodotus, are in direct speech. [42] Thus, a dialogic component in the use of τοι seems to affect the use of ἤτοι, τοίνυν, καίτοι (in Thucydides), and τοιγάρ. The sense of appeal to the dialogic partner in these combinations should complement the standard accounts of their function in the two Histories, which pay attention just to the disjunctive (ἤτοι) and the contrastive (καίτοι) relations between the constituents they link. By analogy, the tendency of μέντοι to be used much more frequently in non-direct speech (115 in non-direct speech vs. 28 in direct speech, as for Herodotus, and 69 vs. 11 in Thucydides) should invite scholars to explore the potential dialogism of the third-person utterances hosting μέντοι besides the concessive contribution of the particle (a cluster, in fact). [43]
§37. Let us now turn to τοι occurring in Herodotus’ narrative sections, a fact which is unexpected in view of the particle’s highly dialogic function. There are not many instances of this—three to be precise. For two of them both Legrand (TLG online) and Hude (OCT) give the same manuscript reading. [44] In 2.120.3, while expressing alignment with the Egyptian account about Helen, Herodotus argues the following: even if initially the Trojans wanted to risk their life to keep Helen (εἰ δέ τοι καὶ ἐν τοῖσι πρώτοισι χρόνοισι ταῦτα ἐγίνωσκον), in the end the disasters of war would have made them give her back to the Achaeans. In 3.33.1, while commenting on Cambyses’ madness, the author argues that it would not be improbable that a physical illness would affect one’s brain (οὔ νύν τοι ἀεικὲς οὐδὲν ἦν τοῦ σώματος νοῦσον μεγάλην νοσέοντος μηδὲ τὰς φρένας ὑγιαίνειν). In both contexts Herodotus provides persuasive arguments to defend other arguments he has heard. It is as if by means of τοι he marks a contact (potential or real) with his audience, as he is engaged in the process of reasoning things through. [45]
§38. Herodotus, moreover, uses τοίνυν four times and καίτοι twice in accounts where the authorial “I” is explicitly present. At 1.57.3 τοίνυν occurs in a passage that delineates the contacts between Pelasgians and Athenians (1.57-58), which includes the first-person statements οὐκ ἔχω ἀτρεκέως εἰπεῖν (1.57.1), and ὡς ἐμοὶ καταφαίνεται εἶναι + ἔμοιγε δοκέει (1.58). At 2.15.2 τοίνυν not only appears a few words after ἐμοὶ δοκέει, but it starts a rare authorial question, which has a strong dialogic potential (“If, then, they [the Egyptians] once had no place to live in, why did they make such a business of the theory that they are the oldest race in the world?” [De Sélincourt and Marincola]). Also at 2.22.4 τοίνυν appears in an argument marked by an authorial question (“[If the Nile comes from melting snow] as it flows … from a very hot into a cooler climate, how could it possibly originate in snow?”)—see the “I” markers ἔγωγε οἶδα, and δοκέω (2.23). At 7.139 τοίνυν occurs in a famous argument that Herodotus makes, acknowledging the merits of Athens in the Persian war; the argument starts with a strong “I” deixis (see ἐξέργομαι γνώμην; μοι φαίνεται; οὐκ ἐπισχήσω). At 4.77.1 καίτοι introduces an alternative account introduced by “I (Herodotus) heard” (ἤκουσα); at 8.112.3 καίτοι continues Herodotus’ insights on the people forced to give money to Themistocles, and is preceded by οὐχ ἔχω εἰπεῖν and δοκέω (8.112.2). [46]
§39. To recap, we have seen thus far that τοι has a dialogic nature in general; when it appears in authorial statements, τοι suggests a potential dialogic interaction with an imaginary (or real) audience. [47] Other explicit “I” markers (first person verbs and first person pronouns, both singular and plural [48]),  which draw attention to the author’s presence, also show signs of making possible appeals to the audience. The point of the connection is the emergence of the “I” that communicates the accounts, and therefore signals the voice of the author. [49] Herodotus is more generous than Thucydides in potential appeals to the audience as well as in the use of overt “I” deixis. [50]

4.3.3 γε in authorial statements

§40. The relation between the explicit presence of the author and the particle γε is what I am going to discuss next. In general, γε’s function is to reflect the speaker’s personal involvement by emphasizing a certain element of the discourse. The emphasis provided by γε is comparable to English prosodic prominence attached to phrases with special informational status. [51] Therefore it causes no surprise to see that in both Histories γε appears more often times in direct speech than in the main narrative. [52] When it occurs in the main narrative, and in particular in explicitly authorial utterances, γε serves a parallel function, marking the author’s voice and conveying a sense of his personal involvement in the statement being made. This is true even if γε’s scope is limited to a noun phrase; the main thing to keep in mind is that the emphasis per se brings the author’s presence to the foreground.
§41. The majority of occurrences is found in explicit authorial statements by Herodotus. [53] Let us focus here on an example of γε in an implicit authorial assertion:
Τιμῶσι δὲ ἐκ πάντων τοὺς ἄγχιστα ἑωυτῶν οἰκέοντας μετά γε ἑωυτούς, δεύτερα δὲ τοὺς δευτέρους, …
Herodotus 1.134.2
After themselves, of all people they [the Persians] honor those who live nearest them, [then they honor] second the second [people] nearest, … (tr. AB)
No linguistic evidence confirms that Herodotus-the-historian is the source of this utterance; nevertheless the general and encyclopedic character of the content makes a different source unlikely. The roles of γε in this assertion are the following: γε marks the phrase μετὰ ἑωυτούς as recalling the immediately previous information (about the Persians honoring themselves by rank, in 1.134.1). It also possibly isolates the phrase as a separate discourse act and as a discrete intonation unit (|μετά γε ἑωυτούς |). [54] Finally, it indicates that the speaker narrows down the conditions under which the present claim (i.e. the Persians honor most of all the people closest to them) holds true. [55]
§42. In Thucydides 7 out of 39 (18%) first person authorial statements stress an element of discourse with γε. [56] Let us consider one of them, where γε has scope over an entire move rather than over just a phrase: [57]
οἱ γὰρ ἄνθρωποι πρὸς ἃ ἔπασχον τὴν μνήμην ἐποιοῦντο. ἢν δέ γε οἶμαί ποτε ἄλλος πόλεμος καταλάβῃ Δωρικὸς τοῦδε ὕστερος καὶ ξυμβῇ γενέσθαι λιμόν, κατὰ τὸ εἰκὸς οὕτως ᾄσονται.
Thucydides 2.54.3
Men accommodate their memories to their current experience. I imagine that if at some time another “Dorian war” comes after this one, with famine coinciding, the verse will in all likelihood be recited with that meaning. (tr. Hammond)
Thucydides makes a point about human psychology. Because memory is driven by experience, humans adapt verses to fit what they personally undergo; thus, it is probable that they would sing “The Dorian war will bring famine” if they were to undergo famine, as much as they had sung “The Dorian war will bring pestilence” after undergoing pestilence. δέ + γε + οἶμαι in the passage mark the upcoming move (made of four clauses: ἢν … καταλάβῃ; ξυμβῇ; γενέσθαι λιμόν; ᾄσονται) as an authorial and authoritative supposition. δέ simply marks a discourse boundary after the gnomic sentence; οἶμαι deictically anchors the upcoming move in the voice of the author, and semantically qualifies the move as an opinion. Finally, γε draws attention to the upcoming content of the opinion as something worth remarking, for reasons that are contextually inferable.
§43. Within the account of the plague two more γε besides the one in (t3) accompany observations implicitly attributable by Thucydides (see 2.47.3 οὐ μέντοι τοσοῦτός γε λοιμὸς οὐδὲ φθορὰ οὕτως ἀνθρώπων οὐδαμοῦ ἐμνημονεύετο γενέσθαι, “but nowhere else was there recorded such virulence or so great a loss of life,” and 2.49.8 τῶν γε ἀκρωτηρίων ἀντίληψις αὐτοῦ ἐπεσήμαινεν, “[symptoms] appeared when the disease took hold in their [the bodies’] extremities” [tr. Hammond]). In both cases the information stressed by γε is subjectively perceived as remarkable, and therefore indicates the speaker’s (i.e. the author’s) involvement—epistemic and emotional at the same time. These passages show that the marking of the author’s voice, his epistemic and emotional involvement, and the expression of his opinion can coexist. [58] οἶμαι in (t3) makes the act of expressing an opinion explicit. In the tragedies and comedies under discussion in this work, γε is regularly found in acts of stancetaking about something. [59] On the whole, the stress on some piece of information evokes someone’s voice, both in authorial as well as in a character’s utterance, and whether or not an act of stancetaking is attached to it.
§44. To sum up: γε in explicit authorial statements reflects the presence of the author as communicator, a human being with ideas and opinions, and puts his voice in the foreground. [60] γε can do the same in passages where the authorial voice is implicit, that is when first person markers are absent but the passage still seems to reflect the author’s voice. Furthermore, the link that I see between γε and the speaker’s voice is consistent with, and complements well, the numerous accounts in the scholarly literature that see γε as an equivalent of intonational prominence, especially exclamations (see n54). Therefore, its occurrence in indirect speech in Thucydides can be regarded as evoking the oral vividness of speech delivery. [61] Finally, γε marks someone’s voice also when the stress on some element of discourse can be read as someone’s opinion about something. Opinion and stance introduces us to the next section.

4.4 Tracking stance

§45. Since someone’s voice can be used, reported, echoed, or hinted at in order to convey rational activities such as judging, planning, and clarifying strategic intentions together with less rational activities such as having a certain feeling, and perceiving the surroundings in a certain way, then a more specific term than “voice” is needed. This term is stance. In this section I will draw ideas not just from a general/common meaning of stance, but from a specific study of stance presented by the linguist John Du Bois in 2007.

4.4.1 The stance triangle

§46. Du Bois defines stance as follows:
Stance is a public act by a social actor, achieved dialogically through overt communicative means, of simultaneously evaluating objects, positioning subjects (self and others), and aligning with other subjects, with respect to any salient dimension of the sociocultural field.
Du Bois 2007:163
Stance is considered an act, that is, a dynamic attainment involving people and objects/states of affairs. Such an act is achieved through overt communication: people deliberately convey x about y to someone else, in linguistic, nonlinguistic, or both linguistic and nonlinguistic ways. The general act of taking stance for Du Bois consists of three components: evaluating, positioning, and aligning. Fundamental to structuring the act of stancetaking are the questions whose stance is conveyed, towards which object it is expressed, and whether it aligns or disaligns with others’ stance.
§47. The following drawing illustrates the different components of stancetaking, and the dynamic relation between each of them, which Du Bois (2007:164) calls “vectors”:
Figure 1: The stance triangle (Du Bois 2007:163)
Subject 1 takes stance by evaluating an object or state of affairs (evaluating vector). The same vector in the opposite direction is that of positioning: given a certain object or state of affairs, Subject 1 positions him- or herself with respect to it. The same bidirectional vectors work when another subject, Subject 2, takes stance about the same object or state of affairs—which may imply evaluating, positioning, or both. This is the second side of the stance triangle. Finally, someone’s stance explicitly or implicitly relates to someone else’s stance. The vector in this case (third side of the triangle) is a bidirectional relation of alignment (ranging from full alignment to full disalignment) between Subject 1 and Subject 2.
§48. What interests Du Bois and me in particular are the linguistic ways by which stancetaking is realized. Let us consider three utterances that exemplify the linguistic encoding of evaluating, positioning, and aligning. “I like this” instantiates the positioning of Subject 1 (the speaking “I”) with respect to the object (the referent of the pronoun “this”). “That’s nice” instantiates Subject 1’s evaluating of the object identified by means of “that.” Finally, “I agree” instantiates the alignment of Subject 1 with somebody else (Subject 2) about some matter (the latter two components remaining implicit).
§49. In his definition Du Bois characterizes these components as co-occurring (“a[n] … act … of simultaneously evaluating, … positioning … and aligning,” see (t4) above). What does this mean? How can a subject simultaneously position herself, evaluate an object, and align with other people? The answer is provided in this next remark:
The stance triangle shows how a stance utterance that specifies only one of the three vectors can allow participants to draw inferences about the others.
Du Bois 2007:164
Du Bois’ point is that linguistic expressions of stance have both explicit and implicit aspects, and that receivers of these expressions are able to and do interpret on both levels at once. For example, if at a conference I judge an academic talk positively, and I take stance by explicitly evaluating the talk (“That was a great talk”), I also allow my colleagues to draw inferences about my alignment with the speaker as well as my own position with respect to the contents or the delivery (or both).
§50. In written communication language can be very shaded, as to the vectors of stance, depending on the degree of polysemy or implicitness that is intended or needed. Let us mention an example showing how complex it can be to track stance in language. The sign “Dem lebendigen Geist” (“To the lively mind/spirit”) above the portal of the “New University” building of the University of Heidelberg can be interpreted as a public act of stance whose explicit linguistic component is just the dative case of “lively mind.” The location of the dedication—the entrance of the University building—represents the nonlinguistic component of the meaning(s). The linguistic component has a little history of its own. The social actors who state “To the lively mind” are to be traced back to the 1920s. Later on further social actors change the adjective of the phrase: during the Nazi period the sign turns into “Dem deutschen Geist” (“To the German mind”). After the Second World War new actors decide to reinstate the previous dedication. The reinstated “To the lively mind,” then, has complex and mainly implicit meanings. The physical and historical context loads the noun phrase with connotations that evoke quite precise political and philosophical opinions (whose ideological contents go beyond the purposes of this chapter), and yet leave them implicit. Only the individuals or groups who share the history of the dedication, its location, and the acknowledgment of its social/political/ideological value, can interpret the noun phrase as a public act of stance evoking specific evaluations, positions, and alignments.
§51. This brings us to the paramount characteristic of Du Bois’ concept of stance: its social dimension. Stance is produced and understood within sociocultural frameworks. It indexes subjectivity and intersubjectivity, and it invokes sociocultural values. “[N]o stance stands alone” (Du Bois 2007:172).

4.4.2 Positioning, evaluating, and (dis)aligning in Herodotus and Thucydides

§52. In the case of Herodotus and Thucydides, the three vectors of positioning, evaluating and aligning are delineated within multiple sets of sociocultural frameworks. On the one hand the two Histories recreate subjective and intersubjective aspects of the characters’ behavior. On the other the works reflect subjective and intersubjective aspects of the two historiographers’ attitudes as well. [62] A crucial asymmetry, common to all third-person narration and all literary pieces, results. Internal characters are supposed to express only evaluation, positioning and alignment in relation to objects and subjects that are internal to the chronicles. The two authors, however, can represent not only evaluation, positioning, and alignment between characters, but they can also represent the same components in the vector “author – character(s)” and in the further vectors “author – contemporaries” and “author – predecessors” and “author – audience”—all of them deriving from the general vector “Subject 1 – Subject 2” in the original model. These multifarious possibilities invite and enable a great deal of interpretative complexity from readers. [63]
§53. In order to prepare the ground for the analysis of particle use through the lens of stance theory, I will now quote some utterances from Herodotus and Thucydides that I regard as instances of evaluating, positioning, and aligning. In stancetaking, these three sub-acts occur simultaneously; however, I distinguish among the examples to better illustrate Du Bois’ point, that in actual utterances one of the three components tends to be more explicit, while the remaining two are implicitly derived. For the sake of clarity I limit this selection to cases where stance is explicitly encoded. In the third part of the chapter I will offer a wider spectrum of instances, considering implicit vectors alongside further instances of explicit vectors.
§54. The purposes of this sampling are two. First, it establishes connections between the theoretical model of the stance triangle and representative linguistic features by means of which the three vectors are delineated in the two Histories. Second, it singles out features that directly co-occur with particles (which are further features to be taken into account), and are instrumental to the upcoming analysis of particles in stancetaking.
§55. The examples of features that mark stance in Herodotus and Thucydides start with linguistic markers of positioning, which include a variety of forms. In (t6), from Thucydides, the speaker Nicias makes his position known through a combination of first-person markers (pronouns and first-person verbs) and nominative participles.
ὅπερ ἐγὼ φοβούμενος, καὶ εἰδὼς πολλὰ μὲν ἡμᾶς δέον εὖ βουλεύσασθαι, ἔτι δὲ πλείω εὐτυχῆσαι … ὅτι ἐλάχιστα τῇ τύχῃ παραδοὺς ἐμαυτὸν βούλομαι ἐκπλεῖν …
Thucydides 6.23.3
… which fearing, and knowing that the business requires much good advice and more good fortune …, I would so set forth as to commit myself to fortune as little as I may … (tr. Hobbes)
Before the naval campaign to Sicily Nicias had initially expressed reluctance to undertake the expedition. [64] When, as seen in this speech, he finally changes his mind, Nicias clarifies his new position (βούλομαι ἐκπλεῖν) by couching it in terms of some cautious considerations: retaining some fear (φοβούμενος) about the undertaking, having awareness (εἰδώς) of the importance of good advice and fortune, and relying as little as possible on chance (ὅτι ἐλάχιστα … παραδούς).
§56. In Herodotus several utterances on the Nile in book 2 provide especially good examples of authorial positioning. For instance:
Πρόθυμος δὲ ἔα τάδε παρ’ αὐτῶν πυθέσθαι, ὅ τι κατέρχεται μὲν ὁ Νεῖλος πληθύων ἀπὸ τροπέων τῶν θερινέων ἀρξάμενος ἐπ’ ἑκατὸν ἡμέρας, …
Herodotus 2.19.2
Yet I was anxious to learn from them why the Nile comes down with a rising flood for a hundred days from the summer solstice … (tr. Godley)
Predicates are the crucial features (see the predicate nominal πρόθυμος ἔα + πυθέσθαι), as they convey the position of the speaking “I” (that is, eagerness to learn) concerning natural phenomena that look puzzling.
§57. Thucydides’ positioning can be exemplified by this famous authorial statement that occurs towards the beginning of the account of the plague:
ἐγὼ δὲ οἷόν τε ἐγίγνετο λέξω …
Thucydides 2.48.3
For my own part, I will tell what kind of thing happened. (tr. AB)
The author positions himself by characterizing his role as objective: he describes the actual course of events and adheres to actual facts. [65] This positioning is expressed in terms of disalignment with other people’s position in relation to the object of inquiry (the plague), and the types of inquiry they have used. ἐγὼ δέ is a priming act (inferable from the subsequent postpositive τε) whose function is to foreground the first person pronoun. [66] ἐγώ is grammatically unnecessary but pragmatically and argumentatively crucial, as it makes explicit Thucydides’ distance from the doctors and laymen (καὶ ἰατρὸς καὶ ἰδιώτης, 48.3) who have previously spoken about the event. Thucydides’ type of inquiry is the central aspect of his positioning: he wishes to underscore the difference between his historiographical method and those of others, of speaking about the plague according to subjective perceptions and knowledge (λεγέτω μὲν οὖν περὶ αὐτοῦ ὡς ἕκαστος γιγνώσκει, 2.48.3 [67] ) versus speaking about what actually happened (οἷόν τε ἐγίγνετο λέξω, 2.48.3).
§58. The linguistic devices employed to express evaluation span a wider range of features than those used to mark positioning. [68] The following three excerpts offer a glimpse of this range. Let us consider first a stance act encoding evaluation by a character:
Ἴσον ἐκεῖνο, ὦ βασιλεῦ, παρ’ ἐμοὶ κέκριται, φρονέειν τε εὖ καὶ τῷ λέγοντι χρηστὰ ἐθέλειν πείθεσθαι·
Herodotus 7.16.α
O king, I judge it of equal worth whether a man is wise or is willing to obey good advice (tr. Godley)
With these diplomatic words Artabanus begins his response to a proposal that Xerxes had made concerning a disturbing dream of his. Xerxes had been advised in his sleep to invade Greece (7.12-14). Wishing to test the dream’s veracity, he suggested that Artabanus should impersonate Xerxes before the two men go to sleep. If Artabanus in the guise of the king should experience the same terrifying dream as Xerxes had had, then Xerxes and all the Persians would have to trust the dream’s suggestion. Artabanus’ skepticism about both the invasion and the dream experiment makes him opt for starting with a gnomic statement; evoking a consolidated and common belief would probably make his perplexity more acceptable. [69] Indeed the gnṓmē (“being wise and being willing to be persuaded by those who say wise things are comparable qualities”) potentially has multiple implications at this moment of the interaction between king and uncle. [70] The discourse act παρ’ ἐμοὶ κέκριται, literally “it is judged by me,” qualifies the gnṓmē as an act of stance explicitly.
§59. Herodotus also evaluates states of affairs in his own voice:
χρῆν γὰρ Κανδαύλῃ γενέσθαι κακῶς
Herodotus 1.8.2
—of necessity, things for Candaules were evolving badly— (tr. AB)
Right at the beginning of the complicating action in the account of Gyges and Candaules, the author anticipates Candaules’ doom. [71] Three linguistic features overall characterize the utterance as evaluative. γάρ introduces pieces of information external to the account of sequential events, [72] thus suggesting discontinuity in the flow of discourse. [73] χρῆν echoes the historiographical motif of necessity, dear to Herodotus. [74] Finally, the adverb κακῶς is a typical example of an evaluative adverb, as it encodes subjective judgment. [75]
§60. The next instance, from Thucydides, includes evaluative adjectives, and it anticipates the relevance of δή in stancetaking.
κίνησις γὰρ αὕτη μεγίστη δὴ τοῖς Ἕλλησιν ἐγένετο καὶ μέρει τινὶ τῶν βαρβάρων, ὡς δὲ εἰπεῖν καὶ ἐπὶ πλεῖστον ἀνθρώπων.
Thucydides 1.1.2
For this was certainly the greatest commotion that ever happened among the Grecians, reaching also to part of the barbarians and, as a man may say, to most nations. (tr. Hobbes)
γάρ grants that the upcoming thought is associated with the preceding ones (see μέγαν τε … καὶ ἀξιολογώτατον τῶν προγεγενημένω “[a war] big and most memorable if compared to the previous conflicts,” 1.1). At the same time it marks a shift in the discourse, from a list of historiographical moves undertaken by the historian—ξυνέγραψε … ἀρξάμενος … καὶ ἐλπίσας … τεκμαιρόμενος—to an authorial evaluation. Here Thucydides evaluates a general phenomenon that he calls κίνησις (“movement” [76] ). μεγίστη δή is the core of the stancetaking act. μεγίστη encodes judgment in meaning (adjective “great”) and form (superlative); δή, as I shall discuss in more detail later (4.6.2), intensifies μεγίστη in such a way as to mark the evaluation as one in whose certainty the author is personally invested.
§61. The final instances concern aligning and disaligning. Also in this case the relevant linguistic features cover a wide range, from negations to particles and the semantics of adverbs. In the fifth book of Thucydides (85-111), the Athenian envoys have a long exchange with the Melians, in the course of which it becomes necessary for them to express strong disalignment. In the speaking turn that precedes the discourse act in question, the Melians use a cautious litotes to point out that gods favor those who stand pious against unjust men (ὅμως δὲ πιστεύομεν τῇ μὲν τύχῃ ἐκ τοῦ θείου μὴ ἐλασσώσεσθαι, ὅτι ὅσιοι πρὸς οὐ δικαίους ἱστάμεθα, … “Yet, as for fortune, we trust that our righteous stand against injustice will not disadvantage us in divine favour,” 5.104 [tr. Hammond]). The Athenians retort:
Τῆς μὲν τοίνυν πρὸς τὸ θεῖον εὐμενείας οὐδ’ ἡμεῖς οἰόμεθα λελείψεσθαι·
Thucydides 5.105.1
Well, we do not think that we shall be short of divine favour either. (tr. Hammond)
Furley (2006:434n62) interprets τοίνυν here as marking the Athenians’ reservations about the Melians’ belief in gods: “the force of the particle τοίνυν might be cumbrously rendered: ‘that goodwill of the gods you refer to’ indicating that the Athenians do no set much store by it” (italics in the text). According to this reading, τοίνυν has scope over the phrase τῆς μὲν τοίνυν πρὸς τὸ θεῖον εὐμενείας. My examination of occurrences of τοίνυν in Thucydides, Herodotus, and other prose texts, however, has led to a different conclusion. I do not find, first, that the particle has scope over specific phrases; rather, it governs entire clauses and even multi-clause units of discourse. Second, it mostly occurs at the beginning of responses, to convey actual or potential disalignment with the previous speaker or between two parties that are facing each other. [77] Hammond’s “Well, …” his consistent method of rendering τοίνυν in Thucydides, [78] captures my interpretation well. [79]
§62. Now an example of alignment. In the passage below Herodotus aligns with Pindar on the importance of custom:
καὶ ὀρθῶς μοι δοκέει Πίνδαρος ποιῆσαι, νόμον πάντων βασιλέα φήσας εἶναι. [80]
Herodotus 3.38.4
and it is, I think, rightly said in Pindar's poem that custom is lord of all. (tr. Godley)
This utterance seals a famous chapter in which Herodotus judges negatively Cambyses’ disruptive attitude towards Egyptian customs, and asserts: “every human being has come to believe custom matters” (νενομίκασι τὰ περὶ τοὺς νόμους οἱ πάντες ἄνθρωποι, 3.38.2). The linguistic feature indexing alignment with Pindar is the adverb ὀρθώς. The passage shows well that all three vectors of stance are actually represented: μοι δοκέει encodes the act of positioning and evaluating at the same time. [81]
§63. Finally, here is Thucydides disaligning with the Athenians about the fulfillment of an oracle. The excerpt is discussed also in IV.2 §109, (t55), as an instance of καί + the construction μοι δοκεῖ overall introducing a move.
καί μοι δοκεῖ τὸ μαντεῖον τοὐναντίον ξυμβῆναι ἢ προσεδέχοντο· οὐ γὰρ διὰ τὴν παράνομον ἐνοίκησιν αἱ ξυμφοραὶ γενέσθαι τῇ πόλει, ἀλλὰ διὰ τὸν πόλεμον ἡ ἀνάγκη τῆς οἰκήσεως …
Thucydides 2.17.2
My view is that the oracle was fulfilled, but in the reverse of the general expectation. It was not the unlawful occupation which caused the disasters to the city, but the war which forced the occupation. (tr. Hammond)
The oracle in question had prohibited the occupancy of the Pelargikon, a sacred area of Athens. The Athenians held that the oracle was fulfilled in the sense that the people occupying the Pelargikon disregarded the oracle, which caused the disasters of the war. Against the Athenians Thucydides expresses a more rationalistic opinion. Yes, the oracle had been fulfilled, but on a different ground: the occupation of the Pelargikon was not the cause but the result of war. καί μοι δοκεῖ τὸ μαντεῖον τοὐναντίον ξυμβῆναι ἢ προσεδέχοντο includes all the linguistic ingredients of disalignment: the object of disalignment (the interpretation of the fulfillment of the oracle), the content of disalignment (“the contrary,” τοὐναντίον), the disaligning subject (μοι), and the people with whom the speaking “I” disaligns (the unspecified third persons who have certain expectations, προσεδέχοντο).

4.4.3 Epistemic and emotional stance: Avoiding dichotomies

§64. Before moving on to the specifically literary implications of representing stance, we should take a moment to address one of the important ways in which stance has both a social and a cognitive dimension. Acts of stance reflect varying degrees of epistemicity and emotionality. [82] We probably would take “I don’t believe you” as primarily an expression of epistemic stance, “That’s horrible” as primarily an expression of emotional stance, and “I wish she were alive” as a mix of the two (conveying both the speaker’s desire and state of knowledge).
§65. In line with Du Bois, [83] I shall not interest myself in establishing a priori classifications (distinguishing epistemic, emotional, and mixed expressions of stance), but rather proceed by considering how epistemicity and emotionality co-exist in different degrees. In fact, Herodotus and Thucydides are capable of representing stance in terms that are epistemic and emotional at the same time. For example, Ηerodotus conveys the imperialistic calculation (epistemic stance) as well as the hyperbolic fever (emotional stance) of Xerxes when he forges the king’s well-known statement, οὐ γὰρ δὴ χώρην γε οὐδεμίαν κατόψεται ἥλιος ὁμουρέουσαν τῇ ἡμετέρῃ, literally “the sun will not illuminate any land that will border ours,” that is to say, “every land under the sun will be ours,” 7.8.γ.2. [84] Thucydides’ characterization of the Greeks’ attitudes during the civil war, ἑτοῖμοι ἦσαν τὴν αὐτίκα φιλονικίαν ἐκπιμπλάναι (“[they] stood ready to satisfy their present lust for victory,” 3.82.8), represents stance in a manner that has both epistemic and emotional components. The author captures the readiness of the Greeks as well as his own distaste for this attitude through his choice of the word φιλονικία, “lust for victory,” (which in Thucydides always has a pathological connotation [85] ), thus layering the Greek epistemic stance and his own emotional stance.
§66. This brings us to another, more subtle and more important way in which the boundaries in stance are blurred. To define a priori when an utterance reflects someone’s stance and when it does not is intuitively problematic. According to Du Bois, stance is found when a social actor makes use of overt communicative means to perform a public act that refers to positioning, evaluating, and aligning. If just one of the vectors of the triangle is linguistically encoded, the contents of the other two may still be inferred. However, as the example “To the lively mind” suggests (see above §50), the directness of the linguistic encoding can be a problem: a certain linguistic form of an utterance can be perceived as an act of stance even though none of the most direct features occur (such as, for instance, an evaluative adverb, or a first-person verb positioning Subject 1, or an expression of disagreement).
§67. Let us mention a passage that shows the interlacing of narration, the author’s emotional stance, and readers’ emotional involvement. The tragic scene that Thucydides depicts towards the end of book 7, the slaughter of Athenians at the river Asinarus in Sicily (7.84), is dominated by clauses that lack evaluative adjectives and adverbs, typical markers of the author’s stance. Yet the passage is commonly taken as a poignant description that arouses a great deal of emotion in readers. What seems just description or just narration actually conveys author’s emotional stance as well. The audience’s emotional involvement derives from the author’s careful combination of visual details with his representation of selected Athenians’ internal perceptions. All of this is designed to blur the boundaries between what is seen and what is felt, who is seeing things and who is feeling them. This is what makes the passage memorable. [86]
§68. Thus, stancetaking is best understood not as bound to certain types of discourse, such as descriptive or argumentative discourse, but either to linguistic features that in different contexts encode stance components, or to indirect cues to the emotional/epistemic involvement of author and audience. Thompson and Hunston (2000:6) claim that one of the main functions of evaluation (taken as equivalent to stance) is “to construct and maintain relations between the speaker or writer and hearer or reader.”
§69. The next subsections complete the discussion about tracking stance by comparing the key notion to other notions used in literary criticism. The first, focalization, has a bearing on the idea of perception; the second and third (reader response and irony) focus on the relation between author and audience.

4.4.4 Stance vs. focalization

§70. Over the last couple of decades the narratological notion of “internal focalization” [87] has gained currency as an approach to studying the phenomenon of embedded thoughts, especially in Thucydides. [88] R. J. Allan (2013:379-380) equates “implicit embedded focalization” with “vicarious perception,” “represented perception,” and “free indirect perception,” as no explicit verb of perception precedes it. Grethlein (2013:96, with n19) similarly equates “focalization” with “perception,” noting that “focalisation … is not limited to seeing, but embraces all senses as well as intellectual activities and emotional response.”
§71. Rood (1998:11-14, and 294-296) points out the limits of the original definition of “focalization,” if used to analyze texts: the distinction “who speaks” vs. “who perceives” is not always straightforward. Most of all, the focalizing role of the primary narrator has no fixed boundaries; he can choose to access the perspective of his characters to whatever degree serves his purposes. Considering focalization on the cognitive level would help, the scholar suggests (1998:12-13); the level of knowledge that is distinctive of a certain character vs. that of the narrator or the audience can be relevant to the recognition and the extent of focalization.
§72. The picture becomes even more complex if we take into account recent developments in cognitive studies. In the introduction of a monograph on viewpoint in language, the linguists Eve Sweetser and Barbara Dancygier argue for the human capability of multiple viewpoints on the the same scene. [89] While watching the same scene, two subjects can apprehend and simulate the viewpoint of their companion. [90] This suggests not only the sheer possibility to access another viewpoint, but also the capability of keeping into account multiple viewpoints at the same time, without necessarily being involved with only a single viewpoint at a time.
§73. One of the current debates about viewpoint and focalization in historiography focuses on how switches are encoded. How do we perceive, while reading, the shift from someone’s eyes to others’ eyes? And how do the historians’ vantage points influence different focalizations and the language that they use to represent them? [91] I suggest that when contents concern more than one individual—which covers the vast majority of the two Histories—we may assume that the viewpoints represented can be multiple and coexisting. At least the viewpoint of the author always coexists with one more viewpoint that may be assigned to one of the characters. So, for example, the very same imperfect can convey aspects related to more people’s perceptions instead of either X’s or Y’s. This observation leads to more general assessments about why I use stance instead of focalization.
§74. “Stance” shares with “focalization” the idea of revealing individual perception. Both notions also cover a wide range of cognitive activities, from sensory perception to emotional states. However, focalization has no social connotations, whereas the core idea of stance is that it is a social activity. The public character of speeches in historiography makes the concept of stance more suitable than focalization. One could object that stances conveyed in the form of thoughts are not public, but this is true only if the thoughts are viewed as private actions from within the time of the story. The thoughts become public when they enter the historical record, at which point they acquire accountability as factors that influenced political or military strategies. The stances taken by the historians themselves are public in this sense as well: their own explicit or implicit authorial statements presuppose reception by an audience.
§75. The concept of stance has some theoretical advantages over focalization. The analysis of the components of stance is socially oriented, which fits the public context of speeches as well as the strategic (if not directly political) value of people’s thoughts in classical historiography. Also, the three basic components distinguished by Du Bois—evaluating, positioning, and aligning—provide more refined tools of analysis than what focalization offers. Herodotus’ and Thucydides’ language can orient the understanding of the receiver by emphasizing one of the three vectors—for example, by inserting adjectives that mark someone’s act of evaluating something, or by inserting a particle that signals alignment or disalignment between the author and some of the characters. Most of all, focalization does not draw attention to the interplay of interactions between the subjects taking stance, and the multiple interpretive options that this interplay gives rise to. One of them is the relationship between author and audience.

4.4.5 Reader response: Eliciting the audience’s stance

§76. The relationship between author and audience is worth dwelling on, as it is fundamental to interpreting stance in the two Histories. At an earlier point (§52) I refined the “Subject 1 – Subject 2” vector in Du Bois’ triangle by including several non-mutually exclusive variants, such as “author – character/group”; “author – contemporaries”; “author – predecessors”; “author – audience.” The next two sections will bring these refinements to bear as I examine in more detail two aspects of vector: first, authorial strategies for eliciting readerly stance in both Herodotus and Thucydides, and second, the effect of Herodotean irony on interrelations between the author, audience, and character vectors.
§77. Let us first focus on the active role of the audience as subjects potentially involved in stancetaking. [92] Two recent reader response approaches have suggested that Herodotus and Thucydides actively call upon readers to formulate their own stance. In her 2008 monograph, Baragwanath suggests that Herodotus actively invites readers’ individual deliberations and responses on matters. This does not mean that Herodotus does not offer explicit judgments; merely that he is keen to promote evaluative openness. He encourages readers to make inferences about the events he narrates, a narrative technique that tends to yield various or alternative readings. Consider for example the narration of Xerxes’ dream in book 7. On the one hand the council scene preceding the dream yields a hybristic explanation of the attack from a Greek perspective. On the other hand the account of Xerxes’ dream possibly reflects a Persian perspective and potentially yields a Persian explanation, as the dream is imbued with oriental motifs (2008:251). The insertion of the dream and all the details connected to it prompts the reader/listener to contemplate reasons from more than one side; it is a strategy of “enfranchising … readers as capable judges” (Baragwanath 2008:20).
§78. The other reader response approach comes from Grethlein’s perspective on the experiential quality of the narrative in Thucydides (see especially 2010:117-118, and 2013b:29-52). In brief, readers are invited to sense the events and feel the emotions from within the scenarios depicted: “Thucydides’ readers experience the plot from the perspective of characters” (2010:280). Thucydides, Grethlein suggests, adopts strategies that make the past present and align the perception of readers with the perception of internal characters. [93] For example, the author maximizes the unpredictable character of events. [94] He blurs the temporal, spatial, sensorial, emotional distance between narrating subjects and narrated objects. [95] Finally, he indirectly invites readers to take stance about actual situations and alternative developments in events. [96] This last point is what makes Grethlein’s approach similar to Baragwanath’s: both scholars find that the narratives of the historians they respectively study unfold in ways that prompt readers to evaluate and take positions about circumstances and decisions.
§79. Both approaches suggest that Herodotus and Thucydides, albeit with different historiographical purposes, invite readers to observe the complexity of situations, to judge evidence, to reconstruct less immediately evident realities, to draw causal links, to test conjectures, to ponder possible resolutions—in a word, to be inquirers themselves. [97] These views stress that readers have an active role, especially when they respond to the authors’ invitation to take stance.
§80. As we have seen, the two historians use different strategies to elicit audience’s reactions. Herodotus provides direct and indirect authorial signs that call for the audience to make a judgment after the fact. These signs enable them to predict or to think about ultimate reasons and consequences a posteriori. Thucydides, too, involves his audience in a deeper understanding of historical events, but he chooses to make them experience history from the inside, on the spot as it were, and to produce empathic effects that bring the audience closer to characters. [98] In short, from a stance perspective we can say that Thucydides is interested in stressing possible alignments between characters and audience, whereas Herodotus is interested in stressing possible alignments between author and audience. I will retrieve this point while analyzing the discourse functions of δή and ἤδη.

4.4.6 Irony: The “author – audience” vector

§81. This subsection illustrates a pragmatic phenomenon that concerns the alignment between author and audience, and seems to occur primarily in Herodotus. [99] In her monograph on indirect discourse (1993), Fludernik makes an observation that pertains to an exquisitely Herodotean habit. She writes that the use of indirect discourse may yield ironic effects. The narrator may exploit the representation of other voices to stress the distance between his own and the character(s)’ stance. Fludernik (1993:73-74) offers examples from English literature where indirect discourse and free indirect discourse are used as a “distancing device,” especially if the narrator reports them “with obvious disagreement or sarcasm” (74).
§82. The irony conveyed by this distancing device results from the reader’s ability to infer a gap between the author’s knowledge or insight about something, and the characters’. This gap is suggested by the context in which the reported words appear. The clash of perspectives is not made explicit, but is left to the inferential abilities of the reader, who can choose how to react to it—for example by tittering, or by mentally recording that the author is hinting at some misunderstanding. This type of irony, then, is similar to dramatic irony: what characters say on the stage becomes incongruous with what the audience and the playwright know.
§83. Schellenberg 2009 comes to a similar conclusion. The article’s central point is that Herodotus, by reporting a character’s words or commenting on a character’s speech, nudges his readers to notice a knowledge gap. In some situations the character’s circumscribed perspective is shown to diverge from the broader perspective of author and readers. As Schellenberg (2009:131) puts it, Herodotus “insinuates himself among his readers by bringing them into a conspirational alliance against the speeches’ ironic victims.” The ironic effect is achieved when the readers feel amusement upon realizing that different readings of the text are designed for different audiences. [100]
§84. This irony is also noted by classicists who are engaged in defining certain Ηerodotean uses of δή. Cooper (2002:2939, emphasis original) states: “the proper meaning of δή is probably clear as the light of day. … The ironic uses of δή are readily understood from this point of view. A word with the sense clearly, obviously may be encountered in almost any language with an ironic force, if the thought at hand is not actually very manifest and unquestionable.” He also adds (2002:2940), “the ironic force of δή is so close to its proper nature that there is no way to confine its appearance to particular constructions.” [101] Instances of “ironic” δή are discussed in 4.5.5; for now I merely wish to note that Herodotus’ pervasive use of irony functions as a way of establishing a tighter relationship with his audience. In Du Boisian terms, this irony entails an implicit disalignment between author and character, and an implicit alignment between author and audience. [102]

4.5 δή in Herodotus: How it connotes voice and stance

§85. δή is a particle by means of which we can track voice and, in particular, stance. Its contribution will be better understood if we begin with an overview of the basic characteristics of its occurrences. Herodotus is the best source, as δή is one of the most frequent particles in his work (1,395 instances overall, that is, 0.7% of the total words, according to the TLG online). [103]
§86. A general and significant fact regarding the Herodotean use of δή is that in at least 996 (71%) of these occurrences it is found in recurrent combinations, together with either other particles or other lexical items. The most frequent particle combinations are μὲν δή (393x), γὰρ δή (165x), καὶ δὴ καί (81x), δὲ δή (78x), ὦν δή (73x), δὴ ὦν (42x), and καὶ δή (51x). Constructions with other lexical items include ἐνθαῦτα δή (22x), ἔνθα δή (9x), τότε δή (8x), δή τοτε (7x), indefinite pronouns + δή (this being a set phrase known from Herodotus on [104] ), and, most of all, οὗτος forms + δή (οὕτω δή, for example, recurs 109 times, but all the adjectival forms of the demonstrative should be included as well [105] ). In all these constructions I take δή to be signifying together with the lexical item it accompanies. [106] Taking, then, this large proportion of occurrences into account, we see that δή does not appear alone so frequently. This fact helps explain why it is problematic to try to account for an independent and overarching value of δή.
§87. As we propose in all our analyses, approaching the study of a particle by starting from its co-text reveals how the particle works in discourse more clearly than if we make speculations about the particle in itself. A particle’s multifunctionality emerges from its multiple co-texts. [107]
§88. δή appears in Herodotus much more frequently than in Thucydides: the difference is 1,395 vs. 201 occurrences, or 0.7% vs. 0.1% of the total words. This difference in number of occurrences, though noteworthy, is not as important as two macro-phenomena. One phenomenon is that about two thirds of δή occurrences in Herodotus signal narrative progression, [108] or steps through which the author-narrator unfolds his accounts (see next section). Conversely, in Thucydides, δή occurrences that perform this function are significantly fewer (I count just 44 of 201 instances). The second phenomenon is that some Herodotean occurrences are those that scholars regard as an ironic use, a use that does not seem to have a clear parallel in Thucydides. Despite these differences, aspects of δή in Herodotus are essential for understanding δή in Thucydides; for this reason I will begin with Herodotus. [109]

4.5.1 Voicing narrative progression

§89. Just as II.3.3.1 shows about Homer, δή in Herodotus frequently works to move forward narration by marking major steps in an account. In these cases δή does not function temporally, [110] to mark chronology, and pertain to content, but rather aids in the performance of discourse acts that advance narration. This discourse function is mainly fulfilled by the following δή constructions: μὲν δή, ὦν δή/δὴ ὦν, καὶ δὴ καί, καὶ δή, ἐνθαῦτα δή, ἔνθα δή, τότε δή, δή τοτε, οὗτος forms + δή, and δὲ δή. [111] The sense can be translated as follows: “then/so/as I said,” when the construction is used to continue an account; “as I was saying,” when the construction resumes the main thread of narration after embedded material (often after a reported speech); [112] “at that point,” when the construction marks a narrative peak; “and so” when the host act of the construction wraps up an account. [113]
§90. In the combinations listed above the lexical items accompanied by δή would be able to fulfill the narrative functions I described by themselves. What δή adds in the combination is the marking of Herodotus’ voice as narrator. This interpretation is based on two considerations. The first is that while very few δή occur in direct speech, [114] the remaining Herodotean δή occur in third-person narration, but have a strong bearing on the representation of the characters’ voice and stance, as I will show. The second consideration is a comparison with δή in Homeric epic. II.3§54 establishes that across direct speech and narrator text, less than half of the instances mark narrative steps (even less in the Iliad). However, not only the main narrator uses this kind of δή, but also characters. Overall δή occurs more in direct speech than in narrator text. [115] The proportionally higher incidence of δή in direct speech in Homer thus sets a historical precedent for Herodotean usage, where δή in this discourse function is strongly associated with the presence of the speaker’s voice, and in particular Herodotus the narrator’s. This voice conveys no specific evaluation but just moderate involvement, a sort of investment (imbued with some amusement, I am tempted to add) in the process of unfolding his historiographical discourse. [116]
§91. A higher level of involvement may underly the narrator’s use of καὶ δὴ καί and καὶ δή. [117] Both combinations have the effect of arousing interest in the reader, as they mark (thanks to the force of καί) the next piece of information as discursively salient. The narratorial voice participates in selecting the topic that follows (καὶ δὴ καί), or in singling out the visual or temporal detail (if not the narrative peak) that will serve as a focal point.

4.5.2 Perception of evidence

§92. I move on now to looking at δή in co-texts involving speech and perception. In those cases δή occurs mainly alone, and in different act positions. [118] Direct speech is linked with perception because both are instances of immediacy: the subject who speaks or perceives uses δή to convey his or her epistemic experience of evidentiality. [119] The subject can be either a character or the author himself. Here is an example of δή in direct speech, which in Herodotus, I repeat, is not frequent at all: [120]
Ἀλλ’ εἴ τοι φίλον τοῦτο οὕτω γίνεσθαι, χρὴ δὴ τό γε ἐμὸν ὑπηρετέεσθαι ἐπιτηδέως.
Herodotus 1.108.5
So, if you like this to happen in this way, it is definitely my duty to obey scrupulously. (tr. AB)
This is how Harpagus concludes his reply to king Astyages, who has just ordered him to find and kill Cyrus, the newborn royal grandson. “It is definitely my duty to obey”: δή is not only an apodotic particle that stresses the consequent character of the statement after the conditional clause, but it is also a marker of the clear perception that Harpagus expresses at that very moment: he realizes that he has no choice but to obey. [121] γε in τό γε ἐμόν contributes to stressing the individual sense of responsibility, and is a further marker of Harpagus’ voice.
§93. The next example, conversely, comes from an observation by the author:
ἔστι δὲ ἔσωθεν ἄλσος δενδρέων μεγίστων πεφυτευμένον περὶ νηὸν μέγαν, ἐν τῷ δὴ τὤγαλμα ἔνι·
Herodotus 2.138.3
… within is a grove of very tall trees growing around a great shrine where the image of the goddess is (tr. Godley)
Herodotus is describing the temple of the Egyptian goddess Bubastis, whom he takes as the equivalent of Greek Artemis (2.137.5). The moment of the report about the statue of the goddess in the shrine features δή; beyond general encyclopedic knowledge, here the historian anchors the narration and the view of the statue to subjective and embodied perception. [122]

4.5.3 In indirect speech and indirect thought

§94. The point of the present subsection is to show how δή can contribute to signal indirect speech and indirect thought, and therefore can be taken as a marker of voice and stance. Instances encompass not only indirect speech preceded by reporting clauses, but also reported λόγοι marked by infinitive clauses, and, in addition, indirect thoughts, whether preceded by verbs of thinking, believing, and the like, or not.
§95. The lexical items that tend to co-occur with these δή are: γάρ, ἵνα, ὅπως, ὅτι + οblique optative, ὡς + participle, [123] superlatives, [124] and adjectives and adverbs indicating scalarity or extremes (such as πᾶς, πολύς forms, μάλιστα). Further important co-textual signs are those qualifying the communicative environment: “I” markers in a character’s direct speech and in explicit authorial statements (which Herodotus generously employs); verbs of saying which introduce indirect speech (with various degrees of indirectness [125] ); verbs of thinking, considering, meditating, intending; σφ- forms associated with αὐτός forms of pronouns. [126] These features do not always appear as δή’s co-texts, but when they do, it is easier to interpret the particle.
§96. It may seem at first that the function of most of these δή is simply to help encode a character’s act of evaluation. Keep in mind, however, that for Du Bois, evaluation occurs simultaneously with positioning and aligning. In this regard the work of Stephens and of Hartung is relevant: their separate assessments of δή have suggested to us a way in which the particle is involved in connecting evaluation with positioning. According to Stephens, δή is one of the particles that in general indicate “the state of the speaker’s mind”; in particular it may denote “determination of the judgment” and “determination of the will” (1837:57 and 58 respectively). In a slightly earlier publication, Hartung stated that with affirmative and negative particles δή expresses resoluteness and confidence (“Entschiedenheit und zuversichtliche Behauptung,” 1832:285). The force of δή described in these terms by Stephens and Hartung essentially concerns stancetaking. In contexts where δή and surrounding features do not advance stories or accounts of events, δή conveys positioning by expressing certainty, will, commitment, and confidence; at the same time these mental states convey evaluation: “X is certainly Y.” [127]
§97. Some examples of characters’ use of δή will help illustrate this point.
Ἅρπαγος δὲ ἔφη, αὐτὸς γὰρ γράψαι, τὸ πρῆγμα ἑωυτοῦ δὴ δικαίως εἶναι.
Herodotus 1.129.2
“It was I,” said the other, “who wrote the letter; the accomplishment of the work is rightly mine” (tr. Godley)
In spite of Godley’s translation, the Greek reports Harpagus’ words in the form of indirect speech (note ἔφη, and the infinitives γράψαι and εἶναι). At the same time αὐτός and ἑωυτοῦ point to the source of the utterance as the pivot. In this context δή broadly relates to Harpagus’ voice, more specifically to Harpagus’ determined assertion, and even more specifically to his evaluation of the assertion (δικαίως). [128]
§98. The next example is drawn from a point in a Persian λόγος when the Persians express an evaluation of the Greeks. The excerpt is a variant of indirect speech, where the reporting clause (“they say”) refers to an articulated account instead of a single sentence.
τὸ δὲ ἀπὸ τούτου Ἕλληνας δὴ μεγάλως αἰτίους γενέσθαι·
Herodotus 1.4.1
But after this (the Persians say), the Greeks were very much to blame (tr. Godley)
“Αfter this”, as Herodotus will explain a bit later (1.4.3), refers to the fact that the Greeks raised a big army, invaded Asia, and destroyed the empire of Priam on account of a girl from Sparta. In this case the exact scope of δή is ambiguous or difficult to see: it is possible to associate δή either with the Greeks (Ἕλληνας δή), with the evaluating adverb (μεγάλως), or with the act Ἕλληνας δὴ μεγάλως αἰτίους γενέσθαι, whose initial boundary is suggested by δή in peninitial position. [129] Whatever δή actually qualifies, it signals that its associated piece of information bears the seal, as it were, of a specific voice (that of the Persians, in this case), and that the speaker’s sense of certainty is driving its intensifying function. [130]
§99. Finally, an example of free indirect thought:
Οἱ δ’ ἐκ τῆς αὐτῆς κώμης Μῆδοι ὁρῶντες αὐτοῦ τοὺς τρόπους δικαστήν μιν ἑωυτῶν αἱρέοντο· ὁ δὲ δή, οἷα μνώμενος ἀρχήν, ἰθύς τε καὶ δίκαιος ἦν.
Herodotus 1.96.2
Then the Medes of the same town, seeing his [Deioces’] behavior, chose him to be their judge, and he (for he coveted sovereign power) was honest and just. (tr. Godley)
In this passage the verb ὁρῶντες, on the local level, and the preceding content, on a more global level, enable us to interpret the evaluative description of Deioces as an example of free indirect thought. By preceding content we mean, on the one hand, the description of Deioces’ righteousness towards the people (1.96.1), but on the other, Herodotus’ preface to that account, which has a distinctly negative tone: “This Deioces, who had a lust for tyranny …” (ἐρασθεὶς τυραννίδος). The difference between this prefatory assessment and the subsequent characterization of Deioces as “honest and just” is a sign that Herodotus evaluates Deioces differently from the Medes. δή, then, marks the assessment of this man as the Medes’ own opinion, and only their own opinion—one, that is, from which Herodotus wishes to establish his authorial distance. Interpreted this way, δή serves an additional function, to stress the “author – audience” vector: we readers know from the omniscient narrator that Deioces had specific political interests, but the people could have not realized that. [131]
§100. Before turning to δή in authorial statements, let me summarize the question of its position and scope. Τhe examples provided thus far show variability. Unlike δή in constructions that advance the narration of events, which consistently occur in act-peninitial position, and have scope over at least an entire discourse act, the δή that marks stance can occupy different positions and have different scopes. δή intensifying an adjective or adverb (“y is certainly/definitely/absolutely x”), has small scope and is mobile; conversely, if δή has bearing on an entire statement (“I am sure that x,” “certainly x is the case”), δή has act scope, and it occurs in act-peninitial position (which typically happens in the γὰρ δή construction). [132]

4.5.4 In explicit and implicit authorial statements

§101. Authorial statements that explicitly or implicitly convey Herodotus’ stance frequently feature a γάρ move including δή contiguous to γάρ. γάρ by itself has no specific link to voice or stance. However, one of its typical uses is to mark discontinuity by introducing unframed discourse, [133] as for example encyclopedic or cultural information necessary at a certain moment in an account, or explanations and illustrations of a piece of argument. Such stretches of text can be interpreted as authorial interventions in a number of cases. [134] To be sure, γάρ is not to be considered an “authorial” particle. [135] What γάρ does is to start a move that considers an aspect of something mentioned in the previous move, regardless of who performs either move; δή may mark that move as a point where some stancetaking is performed.
§102. In the following example γάρ and δή co-occur with the first person verb “I will say,” the referent of the personal pronoun being Herodotus. In this case the historian is the performer of the preceding moves as well.
Οὐ γὰρ δὴ συμπεσεῖν γε φήσω τά τε ἐν Αἰγύπτῳ ποιεύμενα τῷ θεῷ καὶ τὰ ἐν τοῖσι Ἕλλησι·
Herodotus 2.49.2
… for I will not say that what is done in Egypt in connection with the god [Dionysus] and what is done among the Greeks originated independently (tr. Godley)
In an earlier clause the historian states (ἐγὼ μὲν νυν φημί) that Melampus was the one who introduced Dionysus to the Greeks, having learned about him in Egypt. γάρ and δή sustain this act of authorial positioning by starting an expansion: “this is why: 1. I won’t say that the worship of Dionysus here and there coincides by chance (see (t20)); 2. I won’t say that the Egyptians took these practices from the Greeks” (οὐ μὲν οὐδὲ φήσω, 2.49.3). γε stresses συμπεσεῖν, the coincidental occurrence of the worship of the same god in Egypt and Greece. Ιt also signals, together with δή, the narratorial voice. [136]
§103. Here is an example of δή in an implicit authorial statement:
Τούτου τοῦ ἔπεος Λυδοί τε καὶ οἱ βασιλέες αὐτῶν λόγον οὐδένα ἐποιεῦντο, πρὶν δὴ ἐπετελέσθη.
Herodotus 1.13.2
… an utterance to which the Lydians and their kings paid no regard until it was fulfilled. (tr. Godley)
No “I” marker occurs in this discourse segment, but only the omniscient author knows that the words of the oracle would have come true. De Jong (2013:257) quotes this very passage as an instance of prolepsis. In this case δή’s contribution is to add to the semantic and temporal information “it was accomplished” (ἐπετελέσθη, aorist passive form) the speaker’s (here the author’s) commitment to its veracity. [137]

4.5.5 “Ironic” δή

§104. In the analysis of the Deioces passage earlier (§99, (t19)), I suggested that the certainty with which characters express their stance clashes with the author’s stance. Some dubiousness is conveyed, and only indirectly—through additional information that lies elsewhere. [138] The analysis of (t19) exemplifies a general phenomenon that I had given a foretaste of earlier in the chapter (4.4.6), when I reviewed literature that labels these uses of δή as ironic. I now take up the topic again to develop it.
§105. Recall from §84 that δή, besides marking a character’s determination or certainty, can suggest that the author disaligns with that character. The sense of irony is inferable from the awareness of a gap between, on the one hand, the consciousness of the internal characters as represented by the author, and on the other, the knowledge or perspective shared by the author with his audience, which creates a kind of alliance between them. Herodotus’ audience is invited to react to the characters’ stance by providing their own response (see above, 4.4.5). Herodotus often does not say directly that a character’s perception is or will be proven to be a mistake, but he does use δή to hint at the potential divergence between the character’s sense of certainty and his own scepticism. The presence of δή guides the reader to suspect the possibility of a mistake, first, and then to find the suspicion confirmed as the account proceeds—or, as in the case of Deioces, preceding information enables inference of present irony.
§106. Consider the following γάρ move including δή:
Τὸ μέν νυν ἁρπάζειν γυναῖκας ἀνδρῶν ἀδίκων νομίζειν ἔργον εἶναι, τὸ δὲ ἁρπασθεισέων σπουδὴν ποιήσασθαι τιμωρέειν ἀνοήτων, τὸ δὲ μηδεμίαν ὤρην ἔχειν ἁρπασθεισέων σωφρόνων· δῆλα γὰρ δὴ ὅτι, εἰ μὴ αὐταὶ ἐβούλοντο, οὐκ ἂν ἡρπάζοντο.
Herodotus 1.4.2
Abducting young women, in their [= the Persians’] opinion, is not, indeed, a lawful act; but it is stupid after the event to make a fuss about avenging it. The only sensible thing is to take no notice; for it is obvious that no young woman allows herself to be abducted if she does not wish to be (tr. De Sélincourt and Marincola)
This sequence of discourse acts gives Herodotus’ representation of the Persians’ stance on abducting women. The thought “if women were not willing, they would not get abducted” sounds clear and obvious to them, but it does not mean that Herodotus endorses it. δή, or better to say, δῆλα γὰρ δὴ ὅτι, confines the certainty to the source of the thought (if not to the imagined voice), thus detaching it from authorial thought. The irony springs from the implicit discrepancy between the two views.
§107. Sometimes even a character can share in the irony usually confined to author and audience. Let us take for example a scene from the episode “Rhampsinitus and the thief” (2.121). Herodotus shares the thief’s private intentions in advance. At the climactic point in the story, the thief finds that he must put his fellows out of action, so that he may enact his trick successfully. So, when his companions invite him to join a drinking party, he “lets himself be persuaded,” as De Sélincourt and Marincola put it (2003 ad loc.). Here is Herodotus’ Greek:
τοὺς δὲ αὐτοῦ ὥσπερ εἶχον κατακλιθέντας πίνειν διανοέεσθαι καὶ ἐκεῖνον παραλαμβάνειν καὶ κελεύειν μετ’ ἑωυτῶν μείναντα συμπίνειν. Τὸν δὲ πεισθῆναί τε δὴ καὶ καταμεῖναι·
Herodotus 2.121δ.4
… they lay down there just as they were, disposed to drink, and included him and told him to stay and drink with them; and he consented and stayed. (tr. Godley)
δή, nestled within a series of infinitives that indicate the use of oratio obliqua, signals something subtle. After Herodotus relates how the thief’s companions invited him to stay and drink, τὸν δὲ πεισθῆναί τε δὴ καὶ καταμεῖναι gives the thief’s reaction: “he (said that he) would be persuaded indeed, and would stay.” The medio-passive aorist infinitive form πεισθῆναι, and the δή, show that the words are to be identified with the thief’s voice. This is where stance and irony come into play. The thief’s official positioning (and official alignment with his fellows) is “yes, you convinced me,” but author, audience, and the thief himself know that he had already decided to remain because he wanted his friends to get drunk and fall asleep. In this case three subjects align at the expense of other characters participating in the scene. [139]
§108. There are many other cases where δή suggests discrepancy in stance, with potentially ironic effects. [140] In general, through δή Herodotus may be found stressing the “author – audience” vector in the stance triangle. He implicitly warns the reader about some disalignment between his and the characters’ stance, or, in complex scenes, between his and some of the characters’ stance but not others’. In other words, he uses his character or characters’ certainty as a foil to suggest that the reader should either distrust it outright or at least consider it critically.

4.5.6 Interim conclusion

§109. Ιn Herodotus most instances of δή (which generally appear in combinations with other particles) mark narrative progression, that is, the voice of the author as narrator. δή in accompaniment with various linguistic features marks stance: this occurs when the host discourse act (or move) conveys either sensory perception, or positioning, evaluation, and alignment. More specifically, δή marks a sense of certainty or determination. This stancetaking manifests itself mainly through the modalities of indirect speech, indirect thought, and free indirect thought. The so-called ironic uses of δή are cases where the vector “alignment/disalignment” between author and audience is underlined. In particular, through δή Herodotus implicitly disengages himself from the certainty or determination of at least one of the characters involved in a certain situation. He warns his audience that some aspect(s) of the unfolding account are treacherous or erroneous, and therefore require careful judgment.

4.6 δή in Thucydides: Whose stance?

§110. To understand the use of δή in Thucydides we must first consider it in relation to its most recurrent function in Herodotus, δή combinations that advance narrative. At an earlier point (§§89-90) I interpreted this preponderant function as a manifestation of Herodotus’ voice as narrator. By narrative advance I meant either a scene-change accompanied by the introduction of turning points or facts, or the resumption of a narrative thread following some kind of interruption. In Thucydides the total number of δή occurrences is much smaller than in Herodotus (201 vs. 1,395 instances, i.e. 0.1% vs. 0.7% of total words); accordingly, we find far fewer δή constructions that advance narrative. [141] However, the proportion of οὕτω δή in Thucydides is about the same as in Herodotus (0.6% vs. 0.7% of the respective occurrences of δή). [142] In addition, δή combinations involving temporal markers [143] in Thucydides should include ἐπειδή as well, for the –δή component within the whole word (which functions as a temporal conjunction) corresponds to the pragmatic δή used for narrative progression. There are 236 instances of ἐπειδή in Thucydides (besides the instances of δή), or 0.15% of the word total; compare 46 in Herodotus, or 0.02% of the total. Taking all these numbers together then, we can say that while Thucydides employs δή to advance narrative far less often than Herodotus does, he nonetheless still does employ it to a noticeable degree.
§111. A far more important point of difference is Thucydides’ use of δή in direct speech (which represents 26% of the total occurrences, that is, 54 of 201, against 5%, that is, 32 out of 625, in the Herodotean speeches of books 1, 2, 8, and 9). In general, it is worth exploring how δή marks someone’s voice and stance through different degrees of directness in the representation of speech and thought. Τhe next subsections delve into that.

4.6.1 Characters’ stance in direct speech, indirect speech, and indirect thought

§112. Let us start with an instance of δή in direct speech. At 1.142.3 Pericles affirms that it is difficult (χαλεπόν) for a rival city to build fortresses in time of peace (ἐν εἰρήνῃ), and absolutely so (ἦ που δή) in a territory of enemies (ἐν πολεμίᾳ). At 6.37.2 Athenagoras, leader of the democratic party at Syracuse, doubts that the Athenians will succeed in the expedition, especially if the entire island is hostile to them (ἦ πού γε δὴ ἐν πάσῃ πολεμίᾳ Σικελίᾳ, “definitely, with all Sicily being an enemy territory”). [144] From these examples we can infer that speakers use ἦ που + δή to reinforce their assessment—to express their own absolute certainty about it. [145] Further particles that immediately precede δή in direct speech besides ἦ που are μέν, [146] γε, [147] and γάρ. [148]
§113. Several δή discourse acts that encode someone’s evaluation feature superlatives, comparatives, or adjectives such as “all.” In these cases δή has scope over those words. Here is an example, again from direct speech:
οὐ γὰρ δὴ πεφευγότες αὐτὰ ἐπὶ τὴν πλείστους δὴ βλάψασαν καταφρόνησιν κεχωρήκατε
Thucydides 1.122.4
For certainly you avoid them [stupidity, cowardice, indifference] not by imputing it to that which hath done most men hurt, contempt of the enemy (tr. Hobbes)
At the congress summoned by the Spartans in Sparta, the Corinthians want to solicit a vote for war. In this passage they warn the other Spartan allies that an attitude of disdain does not help to keep major shortcomings away (that is, stupidity, cowardice, and indifference). While expressing this opinion, introduced by γὰρ δή, they embed a second opinion, to which the second δή relates: disdain (καταφρόνησις) has harmed the greatest number of people. δή has scope over πλείστους (note the noun phrase τὴν πλείστους δὴ βλάψασαν καταφρόνησιν that surrounds it) and marks the certainty with which the Corinthians take this secondary stance. [149]
§114. Let us now consider δή in indirect speech. Thucydides is reporting the ruthless trick orchestrated by the oligarchs of Corcyra to eliminate the democrats of their own city. The Athenians are holding these democrats captive on an island.
τῶν ἐν τῇ νήσῳ πείθουσί τινας ὀλίγους, ὑποπέμψαντες φίλους καὶ διδάξαντες ὡς κατ’ εὔνοιαν δὴ λέγειν ὅτι κράτιστον αὐτοῖς εἴη ὡς τάχιστα ἀποδρᾶναι, πλοῖον δέ τι αὐτοὶ ἑτοιμάσειν· μέλλειν γὰρ δὴ τοὺς στρατηγοὺς τῶν Ἀθηναίων παραδώσειν αὐτοὺς τῷ δήμῳ τῶν Κερκυραίων.
Thucydides 4.46.5
They sent to the island friends of the captives, whom they instructed to tell them with seeming good-will that they had better escape as fast as they could, for the fact was that the Athenian generals were about to hand them over to the Corcyraean democracy; they would themselves provide a vessel. (tr. Jowett, adapted)
Through the two δή and the present tense of πείθουσί, λέγειν, and especially μέλλειν, Thucydides makes the voice and positioning of the oligarchs vividly audible. The second δή is suggestive of the tone that the false friends might take in direct speech, a usage reminiscent of Herodotus’ “ironic” δή. This is a trick, and the friends’ apparent benevolence actually serves a ferocious intention; these δή amplify the gap in knowledge existing between the different groups of characters. [150]
§115. δή in the representation of indirect thoughts are easier to detect when δοκέω or another verb of thinking precedes the δή act. [151] The activity of thinking can also be entailed or alluded to through direct references to mental states or stances, as in this passage:
Ἆγις δὲ αὐτοῖς ἑτοῖμος ἦν ἐκείνους μὲν μὴ λύειν δὴ τὰς Ἰσθμιάδας σπονδάς, ἑαυτοῦ δὲ τὸν στόλον ἴδιον ποιήσασθαι.
Thucydides 8.9.1
Agis was willing to let them [the Corinthians] observe the Isthmian truce, and to take on the expedition by himself. (tr. Hammond)
The mental state of the Spartan king Agis in this moment [152] is his willingness to procede with the expedition without the Corinthians. Tucker (1892, ad loc.) observes: “Agis is ambitious; he has been checked already in his fancy for acting alone, and he is quite ready to let the Corinthians stay at home and to take all the control upon himself. … δή is ironical.” The irony that Tucker probably alludes to resides in Agis’ attitude: he pretends to allow the Corinthians to suspend military operations, but he is actually eager to act without them. Independently of whether δή implies that or not, Tucker indirectly confirms that the value of δή here is to be found in connection with someone’s voice and stance (Agis’ “sure!”), which does not emerge from Hammond’s translation (exempli gratia).

4.6.2 Implicit authorial δή, especially with superlatives

§116. Earlier (§113) we have seen that characters may encode evaluations of an object or event by means of δή + superlatives or adjectives indicating extremes. When Thucydides himself evaluates, he uses δή largely with superlatives, especially μέγιστος. [153] Thucydides does not tend to show stance by using explicit authorial markers—he is much less generous than Herodotus in this respect—but his stance is inferrable from content: only the historian can assess the extent and magnitude of certain events in absolute terms.
Καὶ ἡ μὲν μάχη τοιαύτη καὶ ὅτι ἐγγύτατα τούτων ἐγένετο, πλείστου δὴ χρόνου μεγίστη δὴ τῶν Ἑλληνικῶν καὶ ὑπὸ ἀξιολογωτάτων πόλεων ξυνελθοῦσα.
Thucydides 5.74.1
This, or something very close to it, is how this battle evolved. It was the largest battle that had been joined between major Greek cities for a very considerable time. (tr. Hammond)
This is a comment that Thucydides inserts towards the end of the account of the battle of Mantinea (5.64-75.3). There are four superlatives: ἐγγύτατα, πλείστου, μεγίστη, ἀξιολογωτάτων. They represent evaluative acts that reflect his historiographical stance. The second δή qualifies the core of his evaluation: Mantinea was the biggest battle between Greek cities. While the scope of this δή is technically small (μεγίστη δή), its co-occurrence with a superlative suffices to characterize the entire passage as an act of stancetaking. [154]
§117. It goes without saying that Thucydidean authorial statements can include superlatives without δή as well; see, e.g. 8.87.4 (σαφέστατον); 1.138.6 (λαμπροτάτους). The absence of δή indicates that the host discourse act has no focal point in the personal involvement of the speaker while qualifying an object/person/event (which is what δή signals).
§118. Implicit authorial statements can also include δή without superlatives. [155] In the following passage Thucydidean usage aligns with constructions found in Homer and Herodotus.
ἔτι πλέον κατεφρόνησαν καὶ ἠξίουν τοὺς στρατηγούς, οἷον δὴ ὄχλος φιλεῖ θαρσήσας ποιεῖν, ἄγειν σφᾶς ἐπὶ Κατάνην, ἐπειδὴ οὐκ ἐκεῖνοι ἐφ’ ἑαυτοὺς ἔρχονται.
Thucydides 6.63.2
they [the Syracusans] took a yet more dismissive view of the Athenians, and—as the common people tend to do when their spirits are up—they insisted that their generals should let them out to Catana, since the Athenians were not coming against them. (tr. Hammond)
The parenthesis οἷον δὴ ὄχλος φιλεῖ θαρσήσας ποιεῖν, being of a gnomic character, draws attention to the authorial voice. [156] In particular, the construction οἷον δή [157] requires closer examination. To capture its tone correctly we must look to parallels in Homer and Herodotus. In Homer, οἷον δή starts exclamatory discourse acts. [158] In these cases οἷον δή is syntactically superfluous; its pragmatic value rests on conveying emotional involvement and giving an exclamatory tone to the content that follows. [159] Similarly, Herodotus uses οἷα δή three times to convey the same emotionality. One of these instances deserves to be mentioned, as translations do not give it justice. Among the facts that Croesus—on the pyre—recalls about Solon is the time when Solon visited Sardis and made no account of all the prosperity he had seen there. The Greek is καὶ θεησάμενος πάντα τὸν ἑωυτοῦ ὄλβον ἀποφλαυρίσειε οἷα δὴ εἴπας (Herodotus 1.86.5). In Hude’s edition οἷα δὴ εἴπας is put in parentheses; De Sélincourt and Marincola do not translate it, and Godley generically writes, “and [had] spoken as if he despised it.” In light of the Homeric evidence, I read the segment as an independent exclamatory discourse act: “after seeing all the prosperity he would make no account of it—what a speech!” [160]
§119. Returning to Thucydides’ οἷον δὴ ὄχλος φιλεῖ θαρσήσας ποιεῖν: instead of reading οἷον as the neuter form of a relative pronoun, the combination οἷον δή can be read as a construction syntactically non-integrated to the rest of the discourse. The entire utterance can be interpreted as an exclamatory authorial stance (thus reflecting emotional engagement) about mass psychology: “how the throng likes to get to action when the spirit is up!” [161] Αt 8.84.3 we find the only other Thucydidean parallel: τὸ δὲ πλῆθος τῶν στρατιωτῶν ὡς εἶδον, οἷα δὴ ναῦται, ὥρμησαν ἐκραγέντες … ὥστε βάλλειν, which Hammonds translates: “At this the crowd of troops saw red, as sailors will, and surged forward to strike …” I interpret οἷα δή ναῦται as an authorial comment with some exclamatory tone: “—sailors!—,” implying “they were sailors (after all)!” or “as sailors typically do!” [162]

4.6.3 When multiple voices share the same stance

§120. Stancetaking can include the manifestation—linguistic or nonlinguistic—of emotions. In this subsection I take up the emotional effect of stancetaking in situations where it is possible to infer an emotional alignment between Subject 1 and Subject 2. In some contexts Thucydides appears to suggest that he himself shares the same stance and its emotional load with some characters involved in the scene; in other words, the same stance is shared by multiple subjects and multiple voices.
§121. What is interesting in these cases is that such stancetaking acts bring to the reader’s awareness the presence of separate voices speaking, as it were, in chorus. δή signals this potential polyphony. Thucydides uses δή in this polyphonic sense to conceal his authorial persona and to show emotional alignment with his characters.
οἱ δὲ Λακεδαιμόνιοι καὶ οἱ ξύμμαχοι εἵποντο μὲν ὡς ἡγεῖτο διὰ τὸν νόμον, ἐν αἰτίᾳ δ’ εἶχον κατ’ ἀλλήλους πολλῇ τὸν Ἆγιν, νομίζοντες ἐν καλῷ παρατυχὸν σφίσι ξυμβαλεῖν καὶ πανταχόθεν αὐτῶν ἀποκεκλῃμένων καὶ ὑπὸ ἱππέων καὶ πεζῶν οὐδὲν δράσαντες ἄξιον τῆς παρασκευῆς ἀπιέναι. στρατόπεδον γὰρ δὴ τοῦτο κάλλιστον Ἑλληνικὸν τῶν μέχρι τοῦδε ξυνῆλθεν· ὤφθη δὲ μάλιστα ἕως ἔτι ἦν ἁθρόον ἐν Νεμέᾳ, …
Thucydides 5.60.2-3
The Lacedaemonians and their allies followed Agis out of respect for the law, but they blamed him severely among themselves. For they believed that they had lost a glorious opportunity; their enemies had been surrounded on every side both by horse and foot; [3] and yet they were returning home having done nothing worthy of their great effort.—No finer Hellenic army had ever up to that day been collected; its appearance was most striking at Nemea while the host was still one … (tr. Jowett)
At the beginning of this passage Thucydides unambiguously reports the thoughts of Spartans and Argives (see νομίζοντες). But the stancetaking introduced by γὰρ δή has a less clear source. The co-occurrence of the superlative κάλλιστον, with its sense of absolute judgment, suggests a moment of authorial stance; however, the lexical choice of κάλλιστον (which is usually uttered by a character in Thucydides), [163] may suggest the voice and stance of the Spartans and Argives as well. [164] The aorist ξυνῆλθεν could help to disambiguate if we assume that the internal viewpoint of characters is more likely to be expressed by imperfect forms. [165] However, even when imperfects occur in discourse acts that host δή, it is difficult to extract from the context whose stance in particular is communicated—especially if the scene has multiple participants, and if what is in question is just their stance. [166]
§122. My reading complements the literary approaches that have focused on how Thucydides allows readers to enjoy thoughts and actions from within his narrative. Grethlein’s reader response approach in particular suits my findings well. As I pointed out in §78, in his experiential model readers are invited to take stance more from following the hic et nunc of the narrative as it unfolds than from processing the author’s a posteriori judgments. [167] Of course, Thucydides does make sweeping authorial statements several times in his history, but other times he conceals his stance by blending it with the characters’.

4.6.4 Any irony?

§123. A final question is whether there are Thucydidean instances of δή suggesting gaps in knowledge, the apprehension of which would result in readerly perception of irony. Our preceding discussion does not suggest that this would occur, perhaps with the exception of (t26) and (t25). Here I am going to mention a couple of passages that have aroused my interest; I invite the reader to adjudicate on those, and on the question in general.
§124. At 6.80.2 Hermocrates tries to convince the inhabitants of Camarina to join the Syracusan fight against Athens. At one point he says that aligning with the Syracusans would preserve the common interests of Sicily and keep the Athenians from making mistakes. The Greek of the last part is καὶ τοὺς Ἀθηναίους φίλους δὴ ὄντας μὴ ἐᾶσαι ἁμαρτεῖν. Earlier (79.1) Hermocrates had acknowledged Camarina’s alliance with Athens and the help they had provided when the Athenians were in trouble. Here, perhaps, the δή indicates that while he acknowledges the evidence (“the Athenians, being your friends”), he also wishes to distance himself from aligning with the Camarinans’ opinion: “you are sure that they are friends, but I am not so sure.” [168]
§125. At 7.18.1 Τhucydides’ use of a ὅπως δή clause—very much like Herodotean ἵνα δή clauses—suggests that he is distancing himself from the characters. The author begins the chapter by saying that the Spartans were preparing their invasion of Attica ὅπως δὴ ἐσβολῆς γενομένης διακωλυθῇ, which Hammond translates “and imagined that an invasion would stop them.” Without δή the ὅπως clause would simply convey the hope and intention of the Spartans. δή suggests that the Spartans were confident of their success, and the added nuance raises the suspicion of authorial disalignment. That Thucydides expresses skepticism here conforms with the general position he elsewhere expresses, viz. that the Athenians were at this point ready to escalate, rather than end, the conflict. [169]
§126. All things considered, the use of δή to exploit narrative potential for irony seems to me to characterize Herodotean practice more than it does Thucydides’. In any case I wish to underscore that such exploitation of irony does not merely function as a means for the historian to demonstrate his omniscient distance from the events he describes. It has consequences for the reader: the perception of irony triggers an act of evaluation, for the reader must, first, access the stance of the parties involved, and second, assess it against the situation described while getting critically engaged. This is the process that δῆθεν—which I will turn to in the subsequent section—particularly invites.

4.6.5 Interim conclusion

§127. Thucydides’ use of δή resembles Herodotus’ in that Thucydides also uses δή to advance narration and signal others’ voices or thoughts, although he does so far less often than his predecessor does. His usage differs from Herodotus’ in three major ways: he prefers to use δή to express authorial stance implicitly rather than explicitly; he uses δή in direct speech more frequently; and he favors the author’s epistemic and emotional alignment with characters whenever, by means of δή, the source of the stance can ambiguously be attributed to the author as well as the characters.

4.7 Stance and polyphony in the use of δῆθεν

§128. Especially in Herodotus, δή’s ironic potential lies in the author’s ability to play with different characters’ varying degrees of knowledge about things, as well as his deliberate efforts to underscore alignment between author and audience. The etymologically related δῆθεν [170] is a particle that amplifies this potential of δή. In spite of its relative infrequency in ancient Greek texts, [171] δῆθεν deserves special attention, as it has a quite complex communicative role.
§129. Powell (1938:85) says of δῆθεν in Herodotus: “implying falsity of speech or thought.” Denniston uses a litotes to define δῆθεν (1950:264): “the nuance of pretense or unreality, and the ironical colour, which, though often present in δή, do not dominate that particle, are in δῆθεν but rarely absent.” [172] Where does falsity come from in the text, and whose pretense is suggested? Most of all, what does the ironic color rest on?
§130. By analyzing the 13 occurrences of δῆθεν in Herodotus and the 5 in Thucydides, I noticed three consistencies: first, δῆθεν never appears in direct speech; second, it regularly involves someone’s declaration of stance in situations where multiple characters or groups of characters are involved; finally, it regularly implies the author’s disalignment with that stance. Here is an example from Herodotus:
Ταῦτα δὲ οὕτω ἐνετέλλοντο ὡς πιστοτάτου δῆθεν ἐόντος αὐτοῦ ἐν Πέρσῃσι, καὶ πολλάκις τε ἀποδεξαμένου γνώμην ὡς περιείη ὁ Κύρου Σμέρδις καὶ ἐξαρνησαμένου τὸν φόνον αὐτοῦ.
Herodotus 3.74.3
They gave him this charge, because they thought him to be the man most trusted by the Persians, and because he had often asserted that Cyrus’ son Smerdis was alive, and had denied the murder. (tr. Godley)
The subjects of ἐνετέλλοντο are the Magi, who want to make the Persian nobleman Prexaspes their ally in order to enhance their political power against the royal family. According to the passage, they count on Prexaspes’ excellent reputation among the Persians, and on the fact that Prexaspes, king Cambyses’ secret agent, and killer of Cambyses’ brother Smerdis upon Cambyses’ orders, had several times confirmed that Smerdis was still alive. The Magi instruct Prexaspes to publicly declare that the Smerdis on the throne is Cambyses’ brother, thus reassuring the people that he is not an impostor, when in fact he is. The discourse act ὡς πιστοτάτου δῆθεν ἐόντος αὐτοῦ ἐν Πέρσῃσι opens a sequence of three genitive absolutes, which reiterate the Magi’s motivations for this strategic move. Very similar reasons had already been expressed at 3.74.1, including the nobleman’s good reputation (ἐόντα ἐν αἴνῃ μεγίστῃ τὸν Πρηξάσπεα ἐν Πέρσῃσι, “Prexaspes was highly respected among the Persians”). What does ὡς πιστοτάτου δῆθεν ἐόντος αὐτοῦ ἐν Πέρσῃσι add, then?
§131. Herodotus makes it apparent here that he is intruding into the Magi’s thoughts, while at 74.1 he describes their motivations in a more objective way (note ὅτι). The linguistic markers of subjectivity in ὡς πιστοτάτου δῆθεν ἐόντος αὐτοῦ ἐν Πέρσῃσι are the construction ὡς + participle and δῆθεν. [173] δῆθεν’s communicative role is to mark both the certainty with which the Magi evaluate Prexaspes’ reliability, and Herodotus’ disalignment at the same time. That is, Herodotus distances himself from the Magi’s evaluation so that the full sense is, “I disassociate myself from their absolute certainty.” [174] This narrative strategy allows Herodotus to call upon the reader to actively take stock of the certainty’s subjective and epistemically limited nature. This intrusion into the Magi’s subjective thoughts subtly adumbrates the subsequent account of Prexaspes’ spectacular disregard of the Magi’s expectations: he will announce the truth, and then will commit suicide.
§132. This use of δῆθεν is ironic in the sense that it reflects a double understanding, namely that of some of the internal characters, who are unaware of the actual outcome, [175] and that of the omniscient author who is about to prove that the Magi’s understanding of the situation was in fact wrong. From a pragmatic perspective, δῆθεν signals the occurrence of not just a single voice or stance, but two: in the above example, the author’s voice overlays and distances itself from the Magi’s. Duality of voice is used to express duality of stance.
§133. Also in the remaining twelve occurrences of δῆθεν in Herodotus the thoughts and actions marked by δῆθεν are meant to appear clear and logical, but actually turn out to be tricks or other acts of deception—in a word, pretenses. In all cases the reader is able to infer the falsity of these thoughts and actions by comparing them against the author’s omniscient perspective. [176] This consistency of use permits a general conclusion: δῆθεν conveys polyphony; the particle’s presence reveals a certain commitment on the author’s part to giving voice to some characters’ sense of clarity and certainty, while at the same time articulating his own disalignment.
§134. Let us now turn to an example from Thucydides:
ἐν τούτῳ δ’ οἱ Μαντινῆς καὶ οἷς ἔσπειστο πρόφασιν ἐπὶ λαχανισμὸν καὶ φρυγάνων ξυλλογὴν ἐξελθόντες ὑπαπῇσαν κατ’ ὀλίγους, ἅμα ξυλλέγοντες ἐφ’ ἃ ἐξῆλθον δῆθεν· προκεχωρηκότες δὲ ἤδη ἄπωθεν τῆς Ὄλπης θᾶσσον ἀπεχώρουν.
Thucydides 3.111.1
In the meantime, the Mantineans and such as had part in the truce, going out on pretence to gather potherbs and firewood, stole away by small numbers, and as they went, did indeed gather such things as they pretended to go forth for; but when they were gotten far from Olpae, they went faster away. (tr. Hobbes)
Thanks to a private truce between Demosthenes and the Mantinaeans (3.109) after a battle near Olpae (3.107-108), the Mantinaeans, joined by other Peloponnesian commanders, are allowed to secretly escape. They pretend to seek forage and firewood, while actually making their way out of Olpae. As in the Herodotean example, δῆθεν accompanies a piece of information that has already been offered: πρόφασιν … ξυλλογὴν ἐξελθόντες (“having gone out on pretense to gather material”) is subsequently echoed by ξυλλέγοντες ἐφ’ ἃ ἐξῆλθον (“having gathered the things they went out for”). In this case the pretense is made evident from the beginning (see πρόφασιν). What δῆθεν does is to make the entire discourse act ἅμα ξυλλέγοντες ἐφ’ ἃ ἐξῆλθον δῆθεν (“gathering at the same time the things they went out for, sure”) sound polyphonic: on the one hand we hear the voice of the Mantinaeans giving the ostensible version of their intentions; on the other we hear the voice of the author conveying that these intentions are fake.
§135. The account continues from the characters’ perspective: the Mantinaeans go away faster (θᾶσσον ἀπεχώρουν); the Ambraciots, a further group of the Peloponnesian contingent, go out to forage too, and realize that the Mantineans are actually escaping (ὡς ἔγνωσαν ἀπιόντας, 3.111.2) at the point when they catch up with them (ὥρμησαν καὶ αὐτοὶ καὶ ἔθεον δρόμῳ, 3.111.2). Further attempts, assumptions, and uncertainty among different groups (allies as well as antagonists) follow, leading eventually to the killing of about two hundred Ambraciots (by the Acarnanians). The reader is thus invited to experience the different groups’ thoughts and decisions in turn, following their sequential arrival on the scene.
§136. The remaining four occurrences of δῆθεν in Thucydides consistently signal the same kind of polyphony: characters’ intentions and claims are presented as obvious even as the author evaluates them as treacherous. [177] This remarkable consistency further supports the conclusion we drew from Herodotean usage, that δῆθεν’s function is to represent the characters’ internal perspective alongside the author’s external one. It achieves this function at the level of the act (that is, it has act scope). The characters express stance in the form of evaluation; the author, in the form of disalignment. His act of disalignment in turn activates the reader’s ability to detect lies, deceptive maneuvers, and mistaken assumptions.

4.8 ἤδη as stance marker

§137. ἤδη is also etymologically related to δή. [178] ἤδη is more frequent than δῆθεν, especially in Thucydides (373 instances, against 136 in Herodotus, that is, 0.24% vs. 0.07% of total words), and unlike δῆθεν it occurs in direct speech as well. Despite its relation to δή, ἤδη is not considered a particle—for example, Denniston does not include it—presumably because in canonical grammars it is classified as a temporal adverb that modifies the content of the host proposition, like ἔπειτα or πρότερον. However, recent literature shows that temporal adverbs such as πάλιν, ἔτι, and αὖτις relate to the ongoing discourse in a range of ways. [179] ἤδη shows a similar versatility that deserves careful analysis. Powell’s Lexicon to Herodotus, for example, gives the word a puzzling combination of meanings. ἤδη has three main meanings: 1. “Of the past, contrasted w[ith] the pres[ent] …; 2. Of the pres[ent] or fut[ure], contrasted w[ith] the past (…); 3. Logically, introd[ucing] a step in the argument” (Powell 1938:160). What makes ἤδη indicate contrast between past and present? Is this contrasting function totally absent from the “logical” function (number 3)? A discourse approach can resolve these questions. As we shall see, ἤδη’s various meanings, which extend within and beyond the marking of time, in fact possess a common thread.
§138. In what follows, I will first contextualize the discussion by drawing from three bodies of scholarship on the topic: contemporary literature on phasal adverb, contemporary works on Latin iam and on ἤδη in Xenophon, and the work of nineteenth-century scholars on the relation between ἤδη and δή. I will then examine patterns of similarity in the co-texts that tend to occur with ἤδη and δή, and establish how far the two words share the function of revealing an individual’s voice and stance. In the main, the uses of ἤδη generally resemble all the uses of δή I illustrated in sections 4.5 and 4.6. Their co-texts often follow the same pattern (e.g. marking narrative progression in act-initial position; marking the evaluating side of stance near scalar terms). These usages are observable in both Histories. A discourse function that is common to both lexical items and is particularly observable in Thucydides relates to the positioning and aligning vectors of stance. In direct or indirect speech, and in indirect thought, ἤδη gives voice to the characters positioning themselves as witnesses of the situation described. Thucydides may also use ἤδη to position himself in a similar fashion in implicit authorial statements. When ἤδη occurs in situations involving characters, it can blend the positioning of characters and that of the historian on the same situation, and it creates alignment. This may explain why Thucydides employs ἤδη three times as frequently as Herodotus. Finally, two further usages are illustrated with regard to both historiographical works: ἤδη expressing stance about time (typically when it is embedded in participial constructions), and a few cases where ἤδη means “now” propositionally, that is, when “now” modifies the propositional content of the host act.
§139. My inquiry benefits first of all from contemporary works on phasal adverbs, and on the so-called T (temporal) - A (aspectual) – M (modal) markers found in the Romance languages. Adverbs such as “still,” “already,” “not yet,” and “no longer” in English (and their counterparts in other Indoeuropean languages) are called “phasal adverbs.” Phasal adverbs concern an “actual or potential transition between different phases of a state of affairs” (Hansen 2008:85). They have a semantic element that refers to “situation-external time” and are relational in nature; that is, they “relate the SoA [state of affairs] denoted by the host clause to a different SoA, which either precedes or follows it” (Hansen 2008:99). But they also refer to “situation-internal time” in that they “focus on specific internal stages of the SoA”; [180] for instance, they are sometimes called “inchoative markers” and “continuative markers.” [181]
§140. Some scholars take as core elements of phasal adverbs the inferences that the adverbs carry. Conventional inferential elements are what, for example, make us chuckle if we hear someone say: “John is still dead”; the utterance is not acceptable because of the conventional incompatibility of “still” with irreversible states. Other scholars stress the logico-semantic implications of phasal adverbs, which give rise to the notions of negative and positive polarity. [182] “Not yet” indicates positive polarity in that the orientation is towards a positive fulfillment of a phase, whereas “still not” indicates orientation towards a negative description of the situation. Still other scholars study their subjective use by speakers who give to the transitions between phases of events a context-dependent meaning. [183] Hansen in particular (2008:109) shows that in specific contexts the use of phasal adverbs can have a subjective connotation independent from (or even in contradiction with) logico-semantic considerations. [184] This brief theoretical outline introduces ἤδη as a lexical item that indicates phasality and may have a bearing on inchoation. Also, its meaning is not necessarily independent of the context of utterance.
§141. Studies in temporal-aspectual adverbs that in different Romance languages etymologically derive from Latin iam are also varied in their approaches and interpretations. These adverbs can be analyzed in light of their logico-semantic contribution to meaning, as well as in light of what they reveal of the speaker’s attitudes or evaluations (therefore in terms of their pragmatic contribution to meaning). [185] Diachronic developments can widen the range of meanings too. For example, the level of pragmaticization (i.e. the process of shifting from propositional to non-propositional meanings) of these adverbs across time can be so high that in some languages they become interjections—as in the case of northern Italian già, which corresponds to “already,” and can be used as a “confirmative particle” or an “interjectional form” in exclamative expressions such as già!/eh già!, ah già, and di già!. [186] Indirectly these studies warn us about the type of contribution to meaning that ἤδη as temporal-aspectual adverb can signal; a phasal adverb does not necessarily modify the propositional content of the host act.
§142. Kroon and Risselada (1998) interpret Latin iam in light of semantic as well as pragmatic considerations. By analyzing several co-texts of iam, they postulate that iam’s general function is evaluative (“an evaluative particle in all its occurrences”). The three components of its basic meaning are: polarity (affirmative polarity), phasality/scalarity (“iam indicates progress on a temporal or non-temporal scale”), and counter-presuppositional saliency (“iam focuses on the element in its scope as something which runs counter to a particular standard or expectation”). [187] All the various uses depend on a combination of these three components. Kroon and Risselada’s study illuminates one important aspect of this adverb in particular. Phasality, which typically concerns the handling of temporal phases (see above §§138-140), is combined with scalarity, which concerns scales other than the temporal one. This is why the two scholars assert that “iam indicates progress on a temporal or non-temporal scale.” Kroon and Risselada in fact analyze iam together with comparatives, superlatives, and scalar terms such as “all” and “much,” which do not necessarily pertain to the handling of time. [188]
§143. Partly inspired by this work, Wakker in 2002 published an article on ἤδη in Xenophon. ἤδη can also be explained through the notions of polarity, phasal scalarity, and denial of expectation, the author claims (2002:4). [189] In Xenophon ἤδη implies that a high point on a temporal or other scale has been reached (2002:6); therefore, in general it works as an evaluative particle (2002:10). For the evaluation of temporal phases, one must take into account the tense that ἤδη accompanies, be it imperfect, aorist, present, or future. For the evaluation of non-temporal scales, the focus of ἤδη is a point in a scale, which is expressed by adjectives and adverbs such as “all/every,” “much/more,” or comparative and superlative forms (2002:7). By means of ἤδη the speaker evaluates the state of affairs linked to these adjectives or adverbs as occurring contrary to her expectations. The denial of expectation in the case of temporal scales concerns something that happens/has happened before the time expected.
§144. Some of the connections made in recent literature were anticipated by scholars in the nineteenth century. Heller, in a long article to Dunker (Heller 1853), for example, states that ἤδη is comparable with déjà (266), and that across its different uses, ἤδη concerns perceptions and expectations about time: the speaker perceives something that happens/has happened/will happen before he expected. [190] Heller (1853:257-258) therefore disagrees with Hartung’s main conclusion, which sees “instantly,” “in that very place”/“on the spot” as the fundamental meaning shared by ἤδη and δή. [191] Heller sees the origin of Hartung’s reading in a passage from Aristotle’s Physics (222b), which he quotes and criticizes (1853:258-259). Let us reproduce this passage, as it is salient both to Heller’s objection and to the points I will make.
τὸ δ’ ἤδη τὸ ἐγγύς ἐστι τοῦ παρόντος νῦν ἀτόμου μέρος τοῦ μέλλοντος χρόνου (πότε βαδίζεις; ἤδη, ὅτι ἐγγὺς ὁ χρόνος ἐν ᾧ μέλλει), καὶ τοῦ παρεληλυθότος χρόνου τὸ μὴ πόρρω τοῦ νῦν (πότε βαδίζεις; ἤδη βεβάδικα). τὸ δὲ Ἴλιον φάναι ἤδη ἑαλωκέναι οὐ λέγομεν, ὅτι λίαν πόρρω τοῦ νῦν.
Aristotle Physics 222b
‘Presently’ or ‘just’ refers to the part of future time which is near the indivisible present ‘now’ (‘When do you walk?’ ‘Presently,’ because the time in which he is going to do so is near), and to the part of past time which is not far from the ‘now’ (‘When do you walk?’ ‘I have just been walking’). But to say that Troy has just been taken-we do not say that, because it is too far from the ‘now.’ (tr. R. P. Hardie and R. K. Gaye, http://classics.mit.edu/Aristotle/physics.4.iv.html)
Heller’s criticism is that Aristotle’s explanation does not conform with the common use of ἤδη with tenses that are unrelated to the present, such as past and future. We will see, however, that what Aristotle says is not at all incompatible with the use of ἤδη with past and future tenses. [192]

4.8.1 Pragmatic relationship to δή

§145. Although nineteenth and early twentieth century scholarship shows attention for the relationship between ἤδη and δή, in more recent times this relationship is elided. A division of labors is more or less implicitly assumed: δή is a marker of either evidentiality or time, and ἤδη means “already” or “now.” My research, however, aligns me with the approach of the earlier scholarship. As I will demonstrate in this subsection, ἤδη and δή show a significant overlap in scope, position, and function. Later subsections (4.8.2 and 4.8.3) will discuss a particular way in which ἤδη marks stance, especially in Thucydides, while 4.8.4 will single out usages that are not shared by δή (at least not in Herodotus and Thucydides). Let us start with the overlap in scope and position.
§146. ἤδη in act-initial and -peninitial position has scope over the entire act (see e.g. (t34), (t37)). If ἤδη occurs either after or before scalar terms such as comparatives, superlatives, πολύς and πᾶς forms, and semantically evaluative adjectives, the focus of ἤδη coincides with the scalar and the evaluative terms, as it does for δή as well. Another similarity with δή is that when scope is small—for example, over just one contiguous word—the act of stancetaking is considered to be the entire act hosting ἤδη and its focus.
§147. The most conspicuous overlap with δή is through the features that co-occur with ἤδη quite often. For example, we find οὗτος (and τοιοῦτος) forms contiguous with ἤδη at the beginning of acts that move the narration forward. [193] We also find ἤδη ὦν in Herodotus, which is reminiscent of δὴ ὦν and ὦν δή; [194] it not only occurs invariably in act-initial position, but it also constitutes a separate act, like καὶ δή. [195] Sometimes we find ἤδη δέ, [196] which is reminiscent of δὲ δή. [197] Similarly, ἐνταῦθα ἤδη (in Thucydides) and ἐνθεῦτεν ἤδη (in Herodotus) recall ἐνθαῦτα δή. [198] καὶ ἤδη—in the same use as the cluster καὶ δή—and καὶ τότε ἤδη, both introducing salient moments, occur only a few times. [199] In all these cases I take ἤδη to generally relate, on the pragmatic level, to the historian’s voice in narrative progression, rather than to the marking of stance—in line with my reading of δή in parallel co-texts. The contiguity with γάρ (ἤδη γάρ or γὰρ ἤδη) [200] is different: sometimes the two lexical items overall introduce unframed discourse; some other times a γάρ move includes an act of stancetaking signaled by ἤδη.
§148. The features shared by ἤδη and δή that specifically relate to stance are comparatives (including μάλλον [201] ), superlatives, πᾶς and πολύς forms, [202] and a few explicitly visual references. All of them can be found in direct and indirect speech, in indirect thought, and in Herodotean explicit authorial statements. [203] In these linguistic environments ἤδη reflects mainly the evaluating side of stance (even though positioning and alignment can be inferred as well, just as for δή). I draw a connection between these linguistic environments, and Wakker’s definition of ἤδη as “evaluative particle” (see above, §143).
§149. Let us consider an example from direct speech in Thucydides.
Καὶ ἐκ τῶν παρόντων, ὦ Ἀθηναῖοι καὶ ξύμμαχοι, ἐλπίδα χρὴ ἔχειν (ἤδη τινὲς καὶ ἐκ δεινοτέρων ἢ τοιῶνδε ἐσώθησαν) …
Thucydides 7.77.1
Even in our present state, Athenians and allies, we must have hope. Before now men have reached safety from yet worse situations than this … (tr. Hammond)
In the observation that starts with ἤδη Nicias expresses his view that in some unspecified past some people found safety from even worse situations. The fact that the verb is an aorist (ἐσώθησαν) does not necessarily mean that the focus of ἤδη is directly that past; it just reveals that the stance expressed by the speaker in the “now” of the utterance refers to a situation that took place in the past. [204] The scope of ἤδη extends to the entire act. Its position evokes ἦ δή at the beginning of turns of speaking. [205] The point of the act rests on δεινοτέρων, which is a comparative form suggesting scalarity. Evaluations go hand in hand with scalarity, as judgments can very well be made through points in scales denoting higher or lower degrees/extents of things. [206] ἤδη, therefore, expresses assurance while introducing the stancetaking act. [207]
§150. A further commonality between ἤδη and δή that deals with stance concerns visual references (see above §93). These instances exclusively pertain to Herodotus’ work. For these I take our cue from LSJ’s third main meaning of ἤδη: “of place.” Herodotus in particular uses ἤδη to indicate the point from which geographical phenomena start or become visible. [208] I locate these instances within the larger frame of epistemic stance: through ἤδη the speaker assesses spatial locations and relations based on his own sensorial and intellectual activity. This point is the link to the next subsection.

4.8.2 Author’s and characters’ ἤδη to mark firsthand experience

§151. Both δή and ἤδη signal voice and stancetaking. While δή’s particular contribution is to stress somebody’s certainty or determination towards some state of affairs, ἤδη’s particular contribution is to mark somebody’s firsthand experience of something. I see the latter function at work, for example, in this explicit authorial statement by Herodotus:
Ἤδη δὲ ἤκουσα καὶ ὕστατον διαβῆναι βασιλέα πάντων.
Herodotus 7.55.3
[I assert that] I also heard that the king crossed last of all. (tr. AB)
(t34) is about an alternative version of the sequence in which the Persian army and the service train crosses the Hellespont. The aorist form ἤκουσα [209] draws attention to the author’s role as inquirer. By means of ἤδη Herodotus articulates personal confirmation or guarantee. [210] Τhis use of ἤδη reminds us of δή voicing perceptions (see 4.5.2), where δή conveys immediacy through the experience of evidentiality. [211]
§152. Let us now quote a parallel from Thucydides’ implicit authorial statements. The most unequivocal locus is the incipit of book 2:
Ἄρχεται δὲ ὁ πόλεμος ἐνθένδε ἤδη Ἀθηναίων καὶ Πελοποννησίων καὶ τῶν ἑκατέροις ξυμμάχων.
Thucydides 2.1.1
From this point begins the war—here and now, I assert it—between the Athenians and the Peloponnesians and their respective allies. (tr. AB)
In this excerpt the hic et nunc of the war is made to coincide with the hic et nunc of the proposition announcing it. Rather than contributing temporal content (“now”), [212] ἤδη has the pragmatic function of expressing the firsthand experience of the speaker, or the speaker’s pretense of that. Τhe present ἄρχεται contributes to the description of the beginning of the war as happening “now”; however, given the multifunctionality of the present tense, ἄρχεται alone is not strong enough to mark the discourse act as an act of the stance. ἤδη provides that reinforcement. [213] Ι interpret the relation of this use of ἤδη to stance in terms of the positioning vector: Thucydides positions himself with respect to the object ‘spatio-temporal coordinates of events’; he places himself as witness of the beginning of the war. Likewise, any other speaker may use ἤδη to attest to his own immediate sensorial experience of something by positioning himself inside the narrated event as real or pretended participant. [214]
§153. ἤδη, it should be noted, retains the value of marking firsthand experience in all possible combinations of tense—all indicative tenses as well as the imperative present. [215] ἤδη enriches the semantic and contextual implications of the chosen tense—aspect, telicity, distance or nearness to the time of speaking—by anchoring that tense to a subjective act of stancetaking, which I express in terms of bearing witness to something.
§154. This use of ἤδη in implicit authorial statements mirrors, in my view, Thucydides’ use of ἤδη to evoke the voice and to mark the stance of characters in indirect speech. Let us consider an example. At the end of his speech to the Spartans, Alcibiades explains what they can gain and what the Athenians can lose if Sparta fortifies Decelea, a village in northern Attica. Athens would lose revenues and tributes on many sides, including those from the allies, who would pay less attention to the Athenians once they see the Spartans’ fighting power. Then comes this relative clause:
οἳ τὰ παρ’ ὑμῶν νομίσαντες ἤδη κατὰ κράτος πολεμεῖσθαι ὀλιγωρήσουσιν
Thucydides 6.91.7
… their allies, when they see that you are now carrying on the war in earnest, will not mind them. (tr. Jowett)
Alcibiades refers to an indirect thought by the Athenians’ allies (νομίσαντες). [216] The following embedded infinitive construction begins with ἤδη (ἤδη κατὰ κράτος πολεμεῖσθαι). The allies are imagined as taking stance by evaluating the manner in which it is conducted (κατὰ κράτος πολεμεῖσθαι). [217] What ἤδη adds to this imagined evaluation is the allies’ perceptual confirmation, their witnessing to the escalated warfare.
§155. I invite the reader to consider a performative aspect that I see behind these uses of ἤδη. With reference to (t36), ἤδη hints at an inner voice that we could “hear” if the thought would be uttered. It is as if Alcibiades would quote κατὰ κράτος πολεμεῖσθαι: “as soon as the allies think ‘you are indeed fighting in earnest,’ they will not mind.” When ἤδη occurs in indirect speech, the hint at the speaking voice is straightforward. When it occurs in indirect thought (that is, preceded by reporting clauses or phrases such as “he thought” [218] ) or in free indirect thought, the allusion to voice is only potential, but present nonetheless. The historians insert ἤδη to approximate the hic et nunc of the original speech or thought. [219] In this sense ἤδη resembles act-initial ἦ δή, and is a sign of vividness that contrasts with the rules of backshift. Perhaps a similar performative allusion can be seen behind the uses of ἤδη to conclude discourse acts. [220]

4.8.3 Thucydides’ blending of stances

§156. The next example features ἤδη in free indirect thought, [221] and it shows a further way to break rules of backshift. All the four ἤδη appearing in the account of the plague can be read as expressing the stance (whether about time or not) of the ill. [222] Here is one of them:
ὅτι δὲ ἤδη τε ἡδὺ πανταχόθεν τε ἐς αὐτὸ κερδαλέον, τοῦτο καὶ καλὸν καὶ χρήσιμον κατέστη.
Thucydides 2.53.3
So, [223] definitely, what was enjoyable and in every way conducive to it [224] [to what was enjoyable], that became also honorable and useful. (tr. AB)
After introducing the general theme of lawlessness (2.53.1), the author points out people’s indulgence of pleasures, as they felt inescapably near to death. “No one was prepared to persevere in what had once been thought the path of honour, as they could well be dead before that destination was reached” (2.53.2, tr. Hammond). (t37) immediately follows this consideration. The words further exemplify the general atmosphere of degradation and psychological distortion described at this point of the account, by introducing the idea of making a virtue of necessity. On closer inspection, actually, the evaluative adjectives ἡδύ, κερδαλέον, καλόν, χρήσιμον, and the scalar adverb πανταχόθεν can be read as reporting directly the people’s stance, ἤδη being a marker of that. It is also possible that the statement expresses Thucydides’ point of view as well, and that ἤδη reflects the position of the author as witness. This would drive the general sense that in the account of the plague the author is emotionally aligned with the characters. In other words, ἤδη could mark polyphony or blending of perspectives involving the author and the characters. [225]
§157. Earlier in the chapter I suggested that δή, especially in Thucydides, can indicate the blending of the characters’ stance and the author’s stance (see 4.6.3). ἤδη can perform the same function: it can blend the way author and character(s) evaluate states of affairs, [226] or it can blend their positioning within situations. Such blending tends to occur when narration focuses on events (the most common kind of third person narration), not when Thucydides is explicitly commenting on them. The fact that Thucydides is evidently more interested than Herodotus in blending stances may explain why he uses ἤδη more often than his counterpart does (as I already noted, the frequency in Thucydides is 0.24% vs. 0.07% of total words in Herodotus; 373 vs. 136 instances respectively).
§158. Reading ἤδη as a sign of blended stance can change our interpretation of the text considerably. Consider the following passage, which represents the narrative peak of a complex episode involving two divisions of Athenians determined to attack a Peloponnesian garrison in Megara, and some Megarian collaborators of the Athenians. Thucydides explains a Megarian stratagem: there is a wagon that occasionally keeps the city’s gates open at night; in this wagon they plan to hide a boat full of people ready to ambush the city.
καὶ τότε πρὸς ταῖς πύλαις ἤδη ἦν ἡ ἅμαξα, καὶ ἀνοιχθεισῶν κατὰ τὸ εἰωθὸς ὡς τῷ ἀκατίῳ οἱ Ἀθηναῖοι … ἰδόντες ἔθεον δρόμῳ ἐκ τῆς ἐνέδρας, βουλόμενοι φθάσαι πρὶν ξυγκλῃσθῆναι πάλιν τὰς πύλας καὶ ἕως ἔτι ἡ ἅμαξα ἐν αὐταῖς ἦν, κώλυμα οὖσα προσθεῖναι·
Thucydides 4.67.4
At this time that cart was at the gates, which were opened according to custom as for the boat. And the Athenians seeing it …, arose from their ambush and ran with all speed to get in before the gates should be shut again (tr. Hobbes)
Hammond takes ἤδη as meaning “already” (“the wagon was already there”), but Hobbes focuses the attention on τότε. My reading is closer to Hobbes’. Rather than indicating that the wagon was “already” there, ἤδη together with καὶ τότε shifts the beginning of the episode’s climactic moment into the hic et nunc, [227] pointing to the moment of the wagon’s appearance at the gates. As soon as the Athenians see the wagon, they begin the attack. [228] ἤδη signals the convergence of the characters’ and audience’s viewpoints—the moment at which the audience perceives the wagon is one and the same as the characters’. ἤδη encodes experientiality, to recall Grethlein’s terminology from §78.
§159. In §144 I quoted a passage from Aristotle’s Physics commenting on the use of ἤδη. The philosopher finds that speakers of his time do not say (οὐ λέγομεν) Ἴλιον φάναι ἤδη ἑαλωκέναι, because the event is too far from νῦν. Now, in light of the excerpts we have seen, it is possible to reinterpret Aristotle’s meaning to fit with the idea of ἤδη as a marker of firsthand experience, or, in other words, stancetaking by positioning oneself as participating in the “here and now” of the situation. ἤδη can be used when “now” is near, but not when “now” is far. In order to use ἤδη speakers should not perceive (or describe) a great distance between the time of the event and their own time. In a stance-based reading ἤδη can accompany the description of events that take place in a time far from the “now” of the composition of the Histories only to the extent that it positions people in the hic et nunc of that past. [229]

4.8.4 Stance about time, and propositional “now”

§160. The next group of ἤδη appears with participial constructions largely involving past tense participles or present participles of εἰμί. In these cases ἤδη has scope over the whole construction, which works as a separate act (e.g. Ἐσκεδασμένου δὲ ἤδη τοῦ λόγου ἀνὰ τὴν πόλιν, Herodotus 4.14.2; τῶν ἐνθάδε ἤδη εἰρηκότων, Thucydides 2.35.1). In participial constructions ἤδη conveys epistemic stance (that is, stance concerning knowledge states), and exclusively about time. [230] ἤδη marks a subjective assessment about the time in which a state of affairs took place. Consider this passage from Herodotus:
Τελμησσέες μέν νυν ταῦτα ὑπεκρίναντο Κροίσῳ ἤδη ἡλωκότι, οὐδέν κω εἰδότες τῶν ἦν περὶ Σάρδις τε καὶ αὐτὸν Κροῖσον
Herodotus 1.78.3
The Telmessians gave this answer when Croesus was already taken. They did not know yet anything about what happened in Sardis and about Croesus himself. (tr. AB)
The Telmessians were priests of Apollo Telmessus; Croesus had asked them to interpret the sudden appearance of snakes in Sardis. But before they could tell him the answer, Croesus was imprisoned by the Persians. ἤδη’s meaning as “already” here signals the perception of an action that had taken place in the past (ἡλωκότι) from the perspective of the events ‘there and then.’ The Telmessians did not have access at the time to the information about Croesus’ imprisonment (οὐδέν κω εἰδότες τῶν ἦν περὶ Σάρδις τε καὶ αὐτὸν Κροῖσον), but the author chooses to show commitment to the status of information in question from within the situation. “Already” reflects the author’s epistemic stance about time. [231]
§161. Finally, I flag that in a few cases the interpretation of ἤδη may be that of a propositional meaning “now,” which then modifies the semantic content. [232] For example, at 6.29.2 Thucydides reports Alcibiades’ opinion: ἐπεμαρτύρετο μὴ ἀπόντος πέρι αὐτοῦ διαβολὰς ἀποδέχεσθαι, ἀλλ’ ἤδη ἀποκτείνειν, εἰ ἀδικεῖ, … “[Alcibiades] insisted that the Athenians should not credit attacks made on him [Alcibiades] in his absence, but if he really was a criminal, they should proceed to execute him there and then” (tr. Hammond). ἤδη corresponds to “there and then” in Hammond’s translation, which is intended to clarify the time and the place of the hypothetical execution (“here and now” in the unbackshifted speech).
§162. An alternative interpretation could take ἤδη as having scope over an entire move introduced by ἀλλ(ά), and including act ἀποκτείνειν, the act εἰ ἀδικεῖ, and the subsequent act starting with ὅτι. This move would express Alcibiades’ reported stance towards the Athenians’ decisions: in unbackshifted speech he would express: “if I am really a criminal, then kill me!”. [233] Both readings are defendable. The possible use of ἤδη to mark a propositional “now” or “here and now” simply reflects an aspect of multifunctionality that we stress also elsewhere in our work. The functions of ἤδη range from nonpropositional to propositional in a continuum. [234] What does not change is that the interpretation of propositional “now” springs from the understanding of the same kind of positioning by the speaker in the hic et nunc of the situation.

4.8.5 Interim conclusion

§163. The two Histories show that the uses of ἤδη significantly overlap with the uses of δή. Both lexical items are flexible in position and scope. Both occur in direct and indirect speech, indirect thought, and explicit and implicit authorial statements. They both have a bearing on the representation of people’s voice and stance, including that of the authors themselves. The use of both lexical items with comparatives, superlatives, πᾶς and πολύς forms, and evaluative adjectives is a sign of their contribution to the expression of stance, be it perceptive, epistemic, emotional, or a combination thereof. An additional aspect common to δή is the link to visual perception; in some geographical observations by Herodotus ἤδη highlights the vision of space, in other words the hic more than the nunc of events. Finally, both δή and ἤδη are capable of indicating a blending of stances by different individuals.
§164. δή and ἤδη mark someone’s stance by implying similar things, δή a sense of certainty or determination, and ἤδη personal guarantee (ἦ δή) or firsthand experience. ἤδη in particular shows that the speaker or perceiver is positioning himself in the situation described. When the focus of ἤδη is on actions expressed in participial constructions, then the stance shown relates exclusively to time. ἤδη is a marker of relativity from inside the world of events, regardless of whether this takes place in the past, the present, or the future. The blending of author’s and character’s respective stances creates an effect of immediacy, which is consistent with Aristotle’s explanation of its meaning: ἤδη is used when the “now” is near. The fact that Thucydides uses ἤδη three times as often as Herodotus does may be connected to Thucydides’ inclination towards blending perspectives.

4.9 ἄρα between discourse cohesion and the marking of stance

§165. ἄρα does not occur often in Herodotus and Thucydides; there are only 66 and 40 instances respectively (0.03% and 0.02% of the total words). The particle’s functions have been well explicated in landmark accounts that are surveyed elsewhere in this monograph. [235] I find, however, that Herodotus differs sharply from Thucydides in the way he uses ἄρα, and this finding contrasts with general expectations about the functions of ἄρα in prose. Because ἄρα has been labeled a “modal”/ “attitudinal particle,” [236] we expect it to relate almost exclusively to stance. More specifically, ἄρα in post-Homeric Greek is said to convey surprise, [237] mark information as interesting, [238] or express skepticism or disillusionment. [239] However, the picture that emerges from my investigation is quite different. The majority of instances in Herodotus do not relate to stance but rather to discourse cohesion. 41 of 66 instances show that Herodotus employs ἄρα in his own narration either to recall previous parts of the discourse or to recall embedded λόγοι. This pattern of usage is similar to the way ἄρα is used in the Homeric poems. Thucydides, for his part, inserts ἄρα in speeches or reports about others’ speeches and thoughts most of the time (35 out of 40 instances). Thucydides uses the particle to show subjectivity. Rather than conveying surprise or disillusionment, ἄρα in Thucydides shows that the subjects are engaged in inferring possibilities [240] from actual situations or conceivable outcomes.
§166. It is useful to understand ἄρα in terms of three representative functions, which I will now illustrate through excerpts drawn from the two historians. As mentioned above, most of the Herodotean usages show that ἄρα possesses functions that extend beyond stance. Two subtypes may be distinguished. In a minority of Herodotean instances, and in all Thucydidean ἄρα, the particle does play a role in stance; I examine the special relation to stance in εἰ ἄρα + indicative or ἤν ἄρα + subjunctive constructions as the third function. All things considered, the significance of these findings on ἄρα may be viewed as the inverse of what I found on ἤδη: whereas ἤδη was found to relate to stance more than expected, ἄρα relates to it much less.
§167. 13 times in Herodotus ἄρα co-occurs with ὡς δέ in temporal or causal clauses that establish a change to a new setting or a new phase in an episode (ὡς δὲ ἄρα/ ὡς δ’ ἄρα [241] ). In these cases ἄρα has act scope, and its pragmatic function is to stress continuity with or logical consequentiality from preceding pieces of content.
Οἱ δὲ Ἕλληνες, ὡς ἐν Πλαταιῇσι τὴν ληίην διείλοντο, ἔθαπτον τοὺς ἑωυτῶν χωρὶς ἕκαστοι. … Ὡς δ’ ἄρα ἔθαψαν τοὺς νεκροὺς ἐν Πλαταιῇσι οἱ Ἕλληνες, αὐτίκα …
Herodotus 9.85.1, and 9.86.1
The Greeks, after dividing the booty in Plataea, buried their own fallen separately. … As the Greeks buried the corpses in Plataea, immediately … (tr. AB)
In (t40) I have included, along with the ἄρα act, a preceding stretch of text possessing the same content as the ἄρα act in order to show that ἄρα can retrieve that part of the discourse after some intervening details (9.85.2-3). This function directly echoes an exceedingly common Homeric usage, where ἄρα is used in announcing actions that naturally follow from preceding discourse (either directly preceding, or with intervening observations). [242]
§168. In these cases, therefore, ἄρα’s value depends on previous parts of discourse rather than what follows. Sometimes, but not always, the main clause that follows ὡς δὲ ἄρα/ ὡς δ’ ἄρα relates something remarkable, unexpected or simply exciting. [243] It is probably due to such instances that several scholars directly connect ἄρα to the feeling of interest. However, I take ἄρα’s range to be limited to its host act (unlike γάρ, which can mark multi-act discourse units [244] ).
§169. In the second representative function ἄρα does not relate to preceding discourse but to information that is external to the sequence of events. In other words, ἄρα can mark unframed discourse. [245] In these cases the author provides the audience with content that is independent of what characters know or are able to assume, in order to give his audience a comprehensive and cohesive picture of the situation.
ἡ δὲ Λάβδα … εἰδυῖά τε οὐδὲν τῶν εἵνεκα ἐκεῖνοι ἀπικοίατο … φέρουσα ἐνεχείρισε αὐτῶν ἑνί. Τοῖσι δὲ ἄρα ἐβεβούλευτο κατ’ ὁδὸν τὸν πρῶτον αὐτῶν λαβόντα τὸ παιδίον προσουδίσαι. Ἐπείτε ὦν ἔδωκε φέρουσα ἡ Λάβδα, τὸν λαβόντα τῶν ἀνδρῶν θείῃ τύχῃ προσεγέλασε τὸ παιδίον, …
Herodotus 5.92γ.2-3
Labda, … knowing nothing of the purpose of their coming … brought it [the baby] and placed it into the hands of one of them. Now they had planned on their way that the first of them who received the child should dash it to the ground. When, however, Labda brought and handed over the child, by divine chance it smiled at the man who took it. (tr. Godley)
In his long speech to the Spartans—designed to dissuade them from restoring Hippias—the Corinthian statesman Sosicles embeds a few narratives that illustrate the despotic behavior of Cypselus and Periander, former tyrants of Corinth. The passage in question is a climactic point in the story of baby Cypselus. The Bacchiadae, political leaders of Corinth, want to kill the baby to prevent the rise of his power in the future (which would be a threat for them). But when the men in charge of the murder reach the mother (Labda) and get the child, the child happens to smile to them, and they are unable to kill him. ἄρα occurs in a discourse act that inserts a piece of information that interrupts the chronological flow of the narrative, and yet is crucial to the understanding of the climax. In order to appreciate the unexpected outcome of the evil mission, the audience must know—and exactly at that moment—that the plan was to dash the baby to the ground. Tense discontinuity helps the audience process the temporal discontinuity: the verb of the δὲ ἄρα act is a pluperfect (ἐβεβούλευτο), while the preceding and following verbs are aorists (ἐνεχείρισε, ἔδωκε, προσεγέλασε).
§170. This and other ἄρα passages show a consistent pattern in which Herodotus attunes himself to the audience’s need for contextual knowledge, by taking care to include external content necessary for following the narrative. [246] It is the external conditions of reception, then, that influence and motivate this use of ἄρα. [247] In these cases ἄρα stresses the “author – audience” vector not by aligning stance, but by giving attention to discourse cohesion. [248]
§171. The third and final function pertains to ἄρα in εἰ/ἤν discourse acts. Such occurrences make up 87% of the ἄρα found in Thucydides (35 of 40), and is found also in Herodotus, albeit less frequently. Except for four instances in the author’s voice, [249] all are hypotheses voiced or thought by characters in direct speech, indirect speech, or indirect thought. [250] In general, hypotheses that are farther from real perceptions and closer to personal guesses take the form of ἤν ἄρα + subjunctive (against εἰ ἄρα +indicative). For example:
Χρῆν μέν, ὦ ἄνδρες Βοιωτοί, μηδ’ ἐς ἐπίνοιάν τινα ἡμῶν ἐλθεῖν τῶν ἀρχόντων ὡς οὐκ εἰκὸς Ἀθηναίοις, ἢν ἄρα μὴ ἐν τῇ Βοιωτίᾳ ἔτι καταλάβωμεν αὐτούς, διὰ μάχης ἐλθεῖν.
Thucydides 4.92.1
Men of Boeotia, no one among us generals should ever have allowed the thought to enter his mind that we ought not to fight with the Athenians, even although [sic] we may not overtake them on Boeotian soil. (tr. Jowett)
The speaker is Pagondas, the only Boeotian commander to express the view that the Boeotians should join battle with the Athenians. At this point in time he does not know for certain that the Athenians have already left Boeotia, which in fact they had, as Thucydides informs his audience. Pagondas can only surmise that the Athenians are gone. ἄρα here marks Pagondas’ supposition as a logical inference based on his state of knowledge; the particle functions as a signal of epistemic stance. [251]
§172. As a general rule, εἰ/ἤν ἄρα acts in Herodotus and Thucydides emphasize what is logically possible. The protasis does not always have to contain scenarios inferrable from facts known to speakers, as in the above example, but may even hypothesize extreme situations or unfortunate outcomes [252] (this gives rise to the translation of ἄρα as “even”). In other words, these ἄρα are subjectivized, and they are largely employed by characters in acts of epistemic stance. An emotional component may naturally accompany the epistemic one, but the extent and the type of emotion remains unspecified.

4.10 Conclusions

§173. This chapter investigates the way in which the two historians use particles to represent their own and other people’s voices and stances. “Voice” is taken as a superordinate concept indicating the linguistic representation of any people’s words (directly or indirectly conveyed) and any kind of thoughts. “Stance” is subsumed under “voice” as the linguistic representation of people’s evaluating something, positioning oneself, and aligning or disaligning with others. The stance may be fully uttered, reported, or reproduced by the historian.
§174. “Stance” shares with “focalization” the idea of revealing individual perception, whether the character’s or the main narrator’s. However, I privilege “stance” over “focalization” because stancetaking is regarded as a social activity, whereas focalization has no social connotations. Du Bois’ theoretical model of the stance triangle enables us to see the stancetaking subject in relation to objects as well as to other subjects. The identification of three vectors (“positioning,” “evaluating,” and “aligning”) provides important interpretive guidance. They make us see connections between the different components of the triangle and the different linguistic features attached to their expressions. Also, they allow for adding variants, such as positioning oneself with respect to the object ‘spatio-temporal coordinates of events,’ and creating (dis)alignment between author and audience as further subjects involved.
§175. Sometimes particles signal voice regardless of whether the content represents a stancetaking act (see 4.3.1 and 4.3.2 about ἦ μήν and τοι; γε (4.3.3) represents a borderline case). δή and ἤδη that mark narrative progression especially in Herodotus further instantiate the presence of someone’s voice. The bulk of the analysis, however, focuses on particles that signal stancetaking in direct and indirect speech, in indirect thoughts (whether free or not), and in explicit as well as implicit authorial statements. The temporal value that is traditionally assigned to δή and ἤδη is seen here as subsumed under their larger pragmatic value of stance markers.
§176. δή in Herodotus and in Thucydides frequently marks the evaluating side of stance, and lets positioning and aligning be inferred. The evaluating side concerns certainty or determination. A Herodotean characteristic of the use of δή, which in Thucydides is harder but not impossible to detect, involves irony. My reading of “ironic” δή points to the disalignment implicitly suggested between the stance of the author and the stance of some of the characters. A major difference in the usages of δή by the two historians is the following: Herodotus, on the one hand, tends to express authorial stance explicitly, and he usually makes it easy to attribute stance to his characters. Thucydides, on the other hand, prefers to be implicit about his own stance, and tends to blur the distinction between his own views and that of his characters.
§177. δῆθεν, etymological and semantic cognate of δή, occurs infrequently but makes a significant contribution to the issue of stance polyphony. The particle amplifies the ironic potential of δή by marking a clash in knowledge and stance that involves more participants. At once δῆθεν implies certainty, disassociation, pretense, and it blends at least two voices. Furthermore, its use activates the reader’s ability to detect the complexity of some situations on multiple communicative levels.
§178. ἤδη, for its part, contributes especially to the positioning vector of the stance triangle: by underscoring somebody’s firsthand experience of something, the speaker or perceiver positions himself with respect to the object ‘spatio-temporal coordinates of events;’ he places himself inside the narrated event as real or pretended participant. This is where the interpretation of propositional “now” springs from. Perhaps because of the sense of firsthand experience, no ἤδη in either masterpiece has ironic effects.
§179. Finally, the uses of ἄρα in Herodotus and Thucydides are less related to stance than what secondary literature suggests, and, within the marking of stance, the implied meanings have no direct connection to epistemic states such as skepticism, or emotions such as surprise. While ἄρα in Herodotus generally echoes Homeric narrative functions, ἄρα in Thucydides is almost exclusively attached to characters’ logical inferences, that is, to epistemic stance concerning someone’s reasoning and guessing about present or hypothetical situations.
§180. Beyond the analysis of the co-texts and contexts of δή, δῆθεν, ἤδη, and ἄρα in the two Histories, this chapter discusses four general methodological and interpretive points. First, we need to recognize the advantages of viewing particles and their functions in terms of a continuum. Disregarding a priori clear-cut distinctions between particles and adverbs enables readers to appreciate the continuity (and, to some extent, even the overlap) of δή and ἤδη in position and in discourse context. Likewise, distinguishing between “modal” and “connective” particles prevents readers from seeing the important role of ἄρα in connecting upcoming acts to preceding discourse. Finally, considering epistemic and emotional stance along a continuum helps readers see the contribution of both aspects to the expressions of subjectivity and of implicit alignment.
§181. Second, we need to examine indirect thought and free indirect thought more deeply than we have done thus far. Unlike indirect speech, which has received considerable attention in recent years, indirect thought in Herodotus and Thucydides is still understudied. [253] This chapter begins to close this gap by paying close attention to the way in which content that follows verbs like “thinking” “considering,” “pondering,” etc. is formulated. In addition, unlike free indirect speech, which in my view is not practised in the two Histories, free indirect thought does occur. Herodotus and Thucydides use the device not only as a way of accessing the mental states of their characters, but also to create interesting effects of both distance and alignment with those mental states.
§182. The third point is about alignment. For all the lexical items in focus, the communication of stance involves potentially several subjects. Some particle uses appear to reflect the author’s metanarrative alignment with his audience. This has implications for Herodotean irony. Alignment can also occur between readers and characters: accordingly I offer a reading of experientiality in Thucydides. These observations enhance links between the interpretation of literary phenomena and the study of linguistic features. In addition, they pave the way for further interdisciplinary connections between literary criticism in Classics and cognitive studies (for example, on the human mind’s natural ability to access multiple viewpoints in the same scene, as well as on the general phenomenon of conceptual blending). [254]
§183. My final point is related to the third. Several works stress that both historians are able to provide particularly vivid reports of a person’s speech by breaking backshift rules—for example by keeping the tense that direct speech would have featured. Particles have a part to play in creating the effect of vividness: Wakker 1997b and George 2009 have already discussed ἦ μήν in indirect speech, and Scardino 2012 notices that particles reproduced in indirect speech recall the emotional tension of the speaker. This chapter establishes that, in addition to ἦ μήν in indirect speech, the occurrences of γε, δή, ἤδη and ἄρα in indirect speech, indirect thought, and free indirect thought signal the “presence” of a person’s voice, or voice and stance. These usages represent further ways of breaking backshift rules, as third person narration breaks up in multiple strands attributable to multiple subjects. The present analysis reveals a third-person narration that is quite polyphonic. Voices and stances crowd the intricate world of Herodotus and Thucydides, so that, with proper attention to language, we may be witness to a truly “figured stage.”


[ back ] 1. For a recent overview of literature, see Hewings 2012:187-188. Genette (1972:76; 225-267) takes voice as what in general denotes a relationship between verbal action and a person. Bal 2006 overviews the use of “voice” in narratology, and focuses on the ideological presuppositions of “voice” as a conceptual metaphor.
[ back ] 2. See in particular Englebretson 2007, a collection of contributions on stancetaking in natural discourse across several languages.
[ back ] 3. On the auditory imagination that readers enact during silent reading in general, see Chafe 1988b:396-397, and IV.3 §§13-14. Mey 1999:127 posits that the “successful use of a voice in narration, in direct or in (free) discourse, is contingent upon its being used in an interactive situation, in a situation of dialogue” (italics in the text). He continues: “Given the fact that ‘voices’ in texts are born as a result of a dialogic cooperation between author and reader, we must … look into the problem of how the voices, once born, are managed and guided in, and throughout, their textual lives.”
[ back ] 4. Baragwanath 2008 is entirely devoted to the nexus of motivation and action in Herodotus.
[ back ] 5. See below §18 on the “single vs. dual voice” debate concerning indirect speech.
[ back ] 6. The manuscript tradition indicates some uncertainty about breathing in αὐτόν forms, which adds to the complexity (on which see in particular Powell 1933 and 1934). The problem holds for Herodotus only minimally, as the reflexive pronoun he uses is ἑωυτόν; according to Powell (1933:210) the Herodotean manuscript tradition reports variation between reflexive and non-reflexive forms αὑτόν/αὐτόν only in 12 cases (out of 508 total occurrences of ἑωυτόν).
[ back ] 7. See e.g. Bakker 2005:99n13 “[in the Iliad] μέλλειν is frequently used in combination with the ‘subjective’ particle που …, which underlines the necessarily ‘subjective’ nature of the verb: μέλλειν is the verbalization of what is evident to a consciousness.” See also Basset 1979.
[ back ] 8. On verb forms that characterize indirect speech, see especially Rijksbaron 1984:53-58; 101-106; the author discusses also forms occurring after “verbs of perception and emotion” (55). I add the pragmatic nuances of presuppositions that characterize the constructions ὅτι + finite verb vs. accusative and infinitive, as in Huitink 2009. Huitink 2012 is entirely devoted to the interface between grammatical forms and communicative functions of reported discourse in Classical Greek.
[ back ] 9. On the latter point, see below §28.
[ back ] 10. See in particular I.1 §§17 and 18; I.4 §4, and the conclusions of individual chapters.
[ back ] 11. There are numerous variants in terminology and definitions in the scholarship. For example, Chafe (1994:247) distinguishes the following forms of represented speech and thought: “direct – verbatim indirect – indirect – elaborated referred-to – simple referred to.” In landmark works on free indirect discourse (e.g. Banfield 1982 and Fludernik 1993) “discourse” covers both speech and thought. However, “indirect discourse” is not useful for Classical Greek historiography: while instances of free indirect thought may be identified or argued for, indirect speech is never free. The reconstruction of speeches in Herodotus and Thucydides had to meet specific genre expectations; for example, they are mostly public speeches, and in dialogic contexts short turns are avoided. In other words, direct speech in this genre does not encompass every kind of voiced utterance, but only those that are verisimilar enough in the local context, and instrumental to the accounts.
[ back ] 12. Sanford and Emmott specify that backshift is not always used. An example of partially “unbackshifted form” is the second sentence in the following excerpt: “She felt a little trembly in her chin and cheeks. What the hell was this?” (from T. Harris 1989, The Silence of the Lambs, London, p. 373; quoted after Sanford and Emmott 2012:188). “This” lends itself to both “backshifted” and “unbackshifted” interpretations, depending on the extent and the symbolic value of the proximity indicated by the pronoun. Fludernik 1993 discusses features of free indirect discourse extensively.
[ back ] 13. Pascal 1977 and Banfield 1982 are foundational works supporting the “dual voice” and the “single voice” interpretation respectively. The developments of the debate over the past decades exceed the purposes of this chapter. From a cognitive perspective, the notion of voice has been criticized (see especially Vandelanotte 2009:245-251). From the point of view of empirical studies on the “dual/single” ambiguity, Sanford and Emmott consider the current state still inconclusive (2012:189).
[ back ] 14. Also the debate about polyphony developed considerably. For example, Nølke 2013 provides a thorough overview of French studies in linguistic polyphony (Bally, Genette, Ducrot, Anscombre), and proposes ScaPoLine (“la théorie SCAndinave de la POLyphonie LINguistiquE”), a new theory of polyphony that challenges the idea of the uniqueness of the speaking subject.
[ back ] 15. Hohti 1976, for example, divides Herodotus’ speeches into “causative” and “non-causative” functions. Scardino 2012:75-84 analyzes the dramatic functions of indirect speeches in both authors.
[ back ] 16. Grethlein (2013b:99) sees speeches as “unmediated access to the past.”
[ back ] 17. On the issue of authenticity of speeches in Thucydides, see in particular Hornblower 1987:45-72 and Rood 1998:46-48. Various aspects of Thucydidean speeches are discussed e.g. in De Romilly 1956, Hunter 1973, Stadter ed. 1973; Cogan 1981; Morrison 2006a; Scardino 2007:453-701; Pelling 2009. Morrison 1994 illustrates Thucydides’ tendency to treat groups of citizens as analogous to individuals in terms of speech deliveries. The multiple meanings of speeches in Herodotus are discussed e.g. in Solmsen 1943 and 1944; Waters 1966; Hohti 1976; Lang 1984b:18-36; Pelling 2006; De Bakker 2007; Schellenberg 2009.
[ back ] 18. Pavlou (2013:432-433) argues that even preambles to speeches (which she calls “attributive discourse”) shed light on Thucydides’ stance about the speech he is about to report.
[ back ] 19. Scardino 2012:69. Scardino 2012 discusses the linguistic, argumentative, and rhetorical characteristics of indirect speech in the two authors.
[ back ] 20. On the latter point see in particular Cooper 1974:76 on “intrusive infinitives” in Herodotus. Wakker (1997b:217-227) extensively analyses linguistic features that are used in indirect speech in both authors. They include inifinitive constructions, oblique optatives, and deictic shifts. The author makes it clear that in reported speech the wording is responsibility of the reporter, whereas the content or thinking expressed is to be attributed to the reported speaker (1997b:220). She concludes that there is a “sledding scale” between direct and indirect speech, “with at the one end minimal integration of the words reported into the matrix construction (i.e. “direct specch”) and at the other end constructions in which the words reported are entirely nominalized, sometimes merely summarizing the speech reported.”
[ back ] 21. On this point see especially the analysis of Thucydidean passages by Debnar 2013. See also below 4.6.3.
[ back ] 22. Lang 1995 collects a remarkable number of participles fulfilling this role in Thucydides.
[ back ] 23. An example in English is given above, §17.
[ back ] 24. See §9 above, the list of linguistic features that help identifying a shift in the source of voice or thought.
[ back ] 25. De Bakker 2013 analyzes Thucydides’ infrequent judgments about individual characters, and the function that these judgments fulfill.
[ back ] 26. While Herodotus’ text abounds in “I” statements (with the variant of some “we,” on which see in particular Dewald 2002), the explicit authorial statements by Thucydides are countable. See the list in Rood 2004:118, see also Lang et al. 2011:129-138. On Herodotus as “overt narrator,” see in particular De Jong 2004a:102-105.
[ back ] 27. “The relative paucity of Thucydides’ first-person appearances has led to the suspicion that Thucydides manipulates his narratorial stance so as to impose his judgment on narratees without their being aware of it” (Rood 2004:118, with a reference to Loraux 1986:149).
[ back ] 28. On “unframed discourse” see in particular II.4.2, with reference to γάρ.
[ back ] 29. I have already discussed this strategy in the section on direct speech (see above, §22). Classical examples of vicarious voices are Solon for Herodotus, and Pericles for Thucydides; see e.g. Shapiro’s study (1996) on Solon and Herodotus, and Edmunds (1975:7-88 and 209-212) on Pericles and Thucydides. Even though in both cases the vicarious voices should not be straightforwardly mapped onto those of the authors, it has been argued that the ideologies of the characters resonate significantly with the authors’. On Pericles’ voice, for instance, Dewald comments: “Pericles is assigned the only focalized voice within the narrative that in force and scope resembles that of the Thucydides-narrator himself” (1999:243).
[ back ] 30. Immerwahr 1966 was one of the first works highlighting the significance of narrative patterns as revealing of Herodotus’ Weltanschauung. More recently, Rood 1998 examines Thucydides’ use of narrative designs (including pace, displacements, and omissions) to convey specific views about history.
[ back ] 31. See I.1 §16.
[ back ] 32. See also above §23.
[ back ] 33. About backshift in general, see above §16.
[ back ] 34. Thucydides 4.88.1; 4.118.1; 5.38.1; 5.50.1; 6.72.5; 8.33.1; 8.75.2; 8.81.3. The only ἦ μήν in direct speech is in Thucydides 4.86.1. ἦ in Thucydides appears in direct speech in other combinations (5.100.1 ἦ που ἄρα; 1.142.3 ἦ που δή; 6.37.2 ἦ που γε δή), with the remarkable exception of ἦ που δή at 8.27.3, which is part of a reported speech by Phrinichus. Scardino (2012:86) observes that particles in indirect speech “display the emotional tension of the speaker.” In Herodotus we find overall eight instances of ἦ μέν. Four of them appear at the beginning of reported oaths (3.74.2; 3.133.2; 4.154.3; 6.74.1—the latter in Hude’s edition). One more instance is in an oath performed live by the queen of the Messagetae Tomyris at 1.212.3. The only other instance in direct speech concerns the terms of a pledge that Leotychides asks Hegesistratus to undertake (9.91.2). The remaining two instances are found at the start of a reported pledge by Babylonian men guaranteeing their intention to marry girls after buying them (1.196.3), and in the report of a speech by Hippias calling upon the same gods that Socles had invoked, assuring them that ἦ μήν the Corinthians would wish to have the Pisistratidae back (5.93.1). Wakker (1997b:216) quotes the Hippias example together with 1.212.3 (direct speech), and simply observes that ἦ μέν in indirect discourse is unproblematic. ΙΙ.3.2.3 discusses ἦ at beginnings of embedded narratives in Homer; in these cases ἦ shares with the historiographical examples the function of marking the start of a performance.
[ back ] 35. Debnar 2001:189. Besides mentioning the eight instances of ἦ μήν in indirect speech, George (2009:162) remarks that μήν alone in Thucydides occurs six times in indirect speech, and five times passages that he labels “diaphonic” (1.3.3; 6.55.3, 2.97.6; 3.82.6; 7.75.6). About the latter, the scholar comments (2009:162) that μήν contributes to make the historian’s observations “more explicitly personal.”
[ back ] 36. The TLG online records 185 occurrences for Herodotus (0.09% of total words), against 4 in Thucydides. Three Thucydidean instances occur in direct speech (by Pericles at 2.41.4; by Cleon at 3.40.4; by Nicias at 7.77.2), and one in a quotation from the Homeric Hymn to Apollo (3.104.4). On τοι in general, see, e.g. Denniston 1950:537; Ruijgh 1971:197-198; Sicking 1993:64; Wakker 1994:360-362; Wakker 1995:254-255. Wakker (1997b:228) states that τοι “expresses straighaway the speaker’s attitude.”
[ back ] 37. This finding aligns with the observations on the use of τοι in persuasive speeches in tragedy; see III.4.5.1. In Herodotus, at 3.3.3 (τοιγάρ τοι, ὦ μῆτερ, …) the ten-year-old Cambyses assures his mother that he will turn Egypt upside down. At 3.29.2 (ἀτάρ τοι ὑμεῖς γε …) adult Cambyses rebukes the Apis-priests; they cannot make a fool of him. At 3.145.3 (Ἀλλ’ εἴ τοι σύ σφεας καταρρώδηκας) the crazy brother of Maeandrius wants to persuade Maeandrius to give him command of the soldiers if he (Maeandrius) is afraid of the Persians. At 5.39.2 (Εἴ τοι σὺ σεωυτοῦ μὴ προορᾷς) the Ephors strongly invite the former king of Sparta Anaxandrides to marry a woman that can bear him children, even if he does not take care of himself. At 8.57.2 the Athenian Mnesiphilus persuades Themistocles not to withdraw the fleet to the Isthmus but to remain in Salamis (Οὔ τοι ἄρα, ἢν ἀπάρωσι τὰς νέας ἀπὸ Σαλαμῖνος, περὶ οὐδεμιῆς ἔτι πατρίδος ναυμαχήσεις; “No; as a matter of fact, if they will withdraw the ships from Salamis, you will have a sea battle that is not in favour of a unified country”). On ἄρα and stance in Thucydides, see below §§171-172. Finally, at 8.65.5, after hearing Dicaeus’ interpretation of the cloud of dust, Demaratus summons Dicaeus to keep silent, as a word on that by the Persian king would cost his head (Ἢν γάρ τοι ἐς βασιλέα ἀνενειχθῇ τὰ ἔπεα ταῦτα, ἀποβαλέεις τὴν κεφαλήν). Τhe three remaining instances, Herodotus 2.120.3, 3.33.1, and 9.38.1 will be commented on below in §37 (with n44), as they do not occur in direct speech.
[ back ] 38. See Stephens 1837:49-50; Bäumlein 1861:236-239; Denniston 1950:537-542.
[ back ] 39. Lang 1984:21-36 and 100-130 for the identification of dialogues that Herodotus articulates in simpler to complex sets (pairs to decahexads).
[ back ] 40. See e.g. Lang 1984b:134-135 for individuals in Herodotus, and Morrison 1994 for collectives in Thucydides.
[ back ] 41. One of the Thucydidean instances is discussed below (§61 (t12)). The remaining instances are in Thucydides 3.45.4; 5.87.1; 5.89.1; 8.53.3. 6 of the Herodotean tokens occur in second position at the very beginning of the turn, which confirms the relevance of τοίνυν to dialogic arguments.
[ back ] 42. Two more ἤτοι occur in indirect speech (Herodotus 8.108.3 and 1.137.2).
[ back ] 43. E.g. George (2009:161) stresses that in Plato μέντοι quite often relates to dialogic, or at least diaphonic, contexts. See below n49 about “diaphony.”
[ back ] 44. The third τοι not belonging to characters’ direct speech is philologically controversial. At 9.38.1 Legrand reads Ὁ μέν τοι θάνατος ὁ Ἡγησιστράτου ὕστερον ἐγένετο τῶν Πλαταιικῶν (“The death of Hegesistratus, however, came later, after the events of Plataea”); manuscripts A, B, C, P, T, M, p and P and a give μέντοι. Hude, followed by Corcella (in Asheri et al. 2006:62), as well as Rosén (1989), conversely, read μέν νυν in line with manuscript d. In my view μέντοι (instead of μέν τοι, whose only attested occurrence in Herodotus is in direct speech, at 3.155.4) is as defensible as μέν νυν.
[ back ] 45. II.5 §70 comments on a Homeric occurrence of τοι in narrator text (Iliad 10.316) that suggests a contact with the audience. The appeal to the audience and its understanding is what marks the combination καὶ γὰρ τοι in Cooper’s view (1998:1355).
[ back ] 46. μέντοι in Herodotus accompanies explicit authorial statements eleven times out of 115 total occurrences: see 1.139; 1.172; 2.63; 2.130; 4.46; 4.81; 4.155; 6.124; 8.8; 8.87; 8.128. In Thucydides this happens twice (3.113.6 μέντοι οἶδα; 8.87.4 ἐμοὶ μέντοι δοκεῖ).
[ back ] 47. De Jong (2004a:111) includes καίτοι and μέντοι in the list of devices that Herodotus uses to establish or underline a connection with the narratees.
[ back ] 48.First person pronouns often turn out to be syntactically redundant. On unnecessary (and therefore marked) nominatives, see II.2 §68.
[ back ] 49. Kroon (1995:111-115) calls “diaphony” the ability of monological discourse to display “features of a communicative interaction, without having all formal characteristics of a dialogical discourse type” (112). She mentions first and second person pronouns, and metadiscursive expressions such as “I will tell you,” “I think,” “as you know” among the features that convey diaphony. On the authorial voice that conveys authorial stance, and on the relevance of μοι δοκεῖ to the stance triangle, see below 4.5.4; 4.6.2; §§62-63.
[ back ] 50. Self-referential expressions by Herodotus include οὐκ ἔχω εἰπεῖν (2.130.2; 6.124.2; οὐ μέντοι ἔχω γε εἰπεῖν at 8.87.3); ἡμεῖς ἴδμεν (4.46.2); γνώμη μοι ἀποδεδέχθω (8.8.3); ἀπέφαινόν μοι ἐς ὄψιν (4.81.4); ἀναβήσομαι (4.82.); δῆλά μοι ὦν γέγονε (2.146.2); δηλῶσω (4.81.4); ἐγώ δ’ἔχω περὶ αὐτῶν γνώμην τήνδε (2.56.1); and the quite common ἐμοὶ δοκέει(ν), with slight variants: ἐμοὶ δοκέει(ν), 1.131.1; 2.4.1; 2.70.1; 2.77.3; 2.120.5; 2.124.4; 2.137.5; 3.5.1; 3.13.4; 3.45.3; 4.50.2; 4.87.2; 4.167.3; 6.95.2; 7.168.3; 8.30.2; 8.66.1; 9.113.2; μοι δοκέει, 1.152.2; 2.24.1; 2.49.3; 2.103.1; 3.38.5; 3.137.5; ἔμοιγε δοκέει(ν), 1.58.1; 2.53.3; δοκέει(ν) ἐμοί, 1.172.1; 2.56.1; 3.135.3; 5.67.1; 5.69.1; 6.30.1; 7.229.2; 8.22.3; 8.103.1; δοκέει μοι, 2.49.1; 2.98.2; δοκέει(ν) δέ μοι, 2.25.3; 2.42.5; 2.109.3; 2.116.1; 4.29.1; 4.198.1; 7.3.4; 7.173.4; 8.63.1; δοκέει δ’ἔμοιγε, 4.189.3; ἐμοὶ δὲ δοκέει, 6.84.3; ἐγὼ δοκέω, 4.155.1; ὡς ἐγὼ δοκέω, 2.63.3. This list does not include the instances uttered by characters. See above n26 for more on the disparity between authorial “I” statements in Herodotus and in Thucydides.
[ back ] 51. See II.5 §33-35; III.2 §§60-61; III.5 §§46-47; below §44 and n162. Powell (1938:65) records a few instances of καί … γε in Herodotus, which he defines “climactic” (see 2.83.1; 2.111.4; 2.146.2; 2.155.1; 5.118.2; 6.56; 7.176.3); both particles mark a piece of information that is more salient than the preceding ones. For further discussion of γε in the Histories see IV.5 (§§20; 34; 62; 67-68; 73; 110). On γε as making the character on the stage more present as communicator, see III.2.2.5. On ὅ γε in Homer, see II.5.3.2.
[ back ] 52. This happens in 111 of 232 total occurrences of γε and γ’ in Herodotus, and in 89 of 175 in Thucydides where either “I” or “you” markers co-occur with γε within three TLG lines. Several times γε in direct speech seems to contribute to strong emotional involvement, e.g. the expression of one’s anger, as III.5 demonstrates about tragedy and comedy (see III.5.4.1, and §71). A clear example of this is at Herodotus 3.29.2, where Cambyses, furious at Apis’ priests, strikes Apis’ thigh and says: Ἄξιος μέν [γε] Αἰγυπτίων οὗτός γε ὁ θεός· ἀτάρ τοι ὑμεῖς γε οὐ χαίροντες γέλωτα ἐμὲ θήσεσθε. “That is a god worthy of the Egyptians. But for you, you shall suffer for making me your laughing-stock” (tr. Godley). A parallel example in Thucydides is a statement uttered by Gylippus and the Syracusan generals at 7.68.2 (whose first part I already commented at IV.2 §120 (t65) because of καί): ὡς δὲ ἐχθροὶ καὶ ἔχθιστοι, πάντες ἴστε, οἵ γε ἐπὶ τὴν ἡμετέραν ἦλθον δουλωσόμενοι: “It will be clear to all of you that the Athenians are not only enemies but the worst of enemies. They came against our country to enslave it.”
[ back ] 53. See Herodotus 1.49; 1.51.4; 1.60.3; 2.15.2; 2.49.2; 2.122.2; 4.32; 4.59.2; 5.118.2; 6.123.1; 7.139.1; 7.152.3.
[ back ] 54. On γε being the equivalent of prosodic prominence attached to constituents, see e.g. Riemer 1823:366; Thiersch 1826:192-193; Hartung 1832:365; K. W. Krüger 1842:346; Denniston 1950:115; Humbert 1960:394-395; Menge 1914:245. As for this monograph, see below n162, and II.5 §33-35; III.2 §§60-61; III.4 §64; III.5 §§46-47, §§90-91.
[ back ] 55. See Wakker 1995:308 “by using γε the speaker demarcates the applicability of his utterance.” The “limitative” value of γε is pointed out in much earlier scholarship: see e.g. Hoogeveen 1769:219; Hartung 1832:364-365; Kühner 1835:398.
[ back ] 56. See Thucydides 1.3.2; 2.54.3 (t3); 3.89.5; 7.86.5 (the famous evaluation of Nicias’ undeserved misfortune, including a δή + superlative construction as well); 8.41.2; 8.87.4; 8.97.2.
[ back ] 57. See e.g. Cauer 1914:74.
[ back ] 58. Distinctions about these components are blurred especially when the combination γε δή appears (on which see n162).
[ back ] 59. On γε in acts of stancetaking in drama, accompanied by high emotional arousal or not, see III.4 §§63-64 and III.5.4.2.
[ back ] 60. This harmonizes with the findings in III.2 §60 and III.5 §45 about γε in drama, a major difference being that γε in Herodotus and Thucydides is much less frequent than in tragedy and comedy.
[ back ] 61. Thucydides 2.13.1; 4.27.3; 6.72.3; 7.48.4; 8.27.3 (2x). See IV.5 §20 on the passionate tone that Dover assigns to a γε passage in one of Nicias’ speeches in book 6.
[ back ] 62. In the textual analyses I will focus on explicit authorial assessments—that is, assessments expressed by means of “I”-statements where the referent of “I” can only be the author—as well as implicit authorial statements identified as such by secondary literature. Therefore, I will assume that authors and narrators coincide: I am more interested in the expressions of subjectivity and intersubjectivity of the two authors than in establishing the narratological functions of the “I” in “I”-statements. See Rood 1998:10 “Readers of historical texts … tend to identify author and narrator.”
[ back ] 63. 4.4.6 will resume and develop this crucial topic.
[ back ] 64. On Nicias’ speaking style, and on the specific speech from which this excerpt is drawn, see IV.5.2.
[ back ] 65. On this famous wording see in particular Repgen (1982) and Finley (1985:47-48, and 116n4), who comment on the German dictum “wie es eigentlich gewesen” (“how it actually was”), taken as a direct quotation of Thucydides 2.48.3 by the early nineteenth century historian Leopold Ranke. Stroud 1987 disaligns with Repgen and Finley, and takes οἷον as the accusative form of a neuter interrogative pronoun (“what”) referring to αὐτό in the previous sentence, rather than an adverbial accusative of manner (“how”). He also reads τε as anticipating the καί that follows (ἐγὼ δὲ οἷόν τε ἐγίγνετο λέξω, καὶ ἀφ’ ὧν ἄν τις σκοπῶν), and denies that τε has the emphatic function that Repgen sees. As II.4.5.2, III.2.2.3, and IV.2.3 point out, the copulative function of τε and a pragmatic enrichment need not be mutually exclusive.
[ back ] 66. On the notion of priming acts see II.2.5; instances of priming acts in Herodotus and Thucydides are discussed in IV.2 §§38-41 and 87; IV.3 §79, §88, §107, §§110-115; IV.5 §32, §46, §59, §63, §78.
[ back ] 67. μὲν οὖν in this act is commented on in IV.3 §145 (t35).
[ back ] 68. Throughout these sections I will use “evaluation” as a resultative noun related to the act of evaluating in Du Bois’ sense. In literature, conversely, see e.g. notably Hunston and Thompson 2000, “evaluation” is taken as the equivalent of “stance,” thus embracing what in Du Bois is “positioning” and “aligning” as well.
[ back ] 69. Scholarship acknowledges that gnomic statements are a typical trait of Artabanus’ speech style; see, for example, Bischoff 1932:57-58; Shapiro 2000:101-103; Scardino 2007:182-183. In IV.5.5 I offer a discourse-oriented reading of Artabanus’ speeches at 7.49 and 7.51.
[ back ] 70. One of them being that if the king listens to unwise people it means that he is not wise.
[ back ] 71. The Herodotean discourse act formed by the impersonal verb and the infinitive clause is hardly rendered in translation: Godley simply paraphrases “[Candaules], doomed to misfortune”; De Sélincourt and Marincola read “[the king], who was doomed to a bad end.”
[ back ] 72. See above §27, IV.3 §§108-109, and IV.5 §91. In my translation I render γάρ with an “em-dash,” a punctuation mark that generally conveys discontinuity but also associative meaning.
[ back ] 73. The discourse act, in fact, occurs between a participial phrase and the main clause, both focusing on the complicating action: the Herodotean sequence of acts is the following: “Not so much time had passed” (Χρόνου δὲ οὐ πολλοῦ διελθόντος); “—of necessity, for Candaules things were evolving badly—” (χρῆν γὰρ Κανδαύλῃ γενέσθαι κακῶς); “he addressed Gyges with the following words” (ἔλεγε πρὸς τὸν Γύγην τοιάδε).
[ back ] 74. See in particular Munson 2001.
[ back ] 75. See De Jong 2013:255 on this as a metanarrative comment. In fact the discourse act, narratologically, works as a proleptic statement, even though the omniscient stance of Herodotus relies on the use of a finite past tense (χρῆν). A similar act occurs in Herodotus 9.109.2 (note, again, the imperfect): τῇ δὲ κακῶς γὰρ ἔδεε πανοικίῃ γενέσθαι “for she [the daughter of Masistes’ wife] and all her house were doomed to evil.”
[ back ] 76. On this occurrence of the term see in particular Rusten 2015.
[ back ] 77. A particularly illuminating example outside of historiography is Andocides On the Peace 30: Συρακόσιοι δ’ ὅτε ἦλθον ἡμῶν δεόμενοι, φιλότητα μὲν ἀντὶ διαφορᾶς ἐθέλοντες εἰρήνην δ’ ἀντὶ πολέμου ποιεῖσθαι, τήν τε συμμαχίαν ἀποδεικνύντες ὅσῳ κρείττων ἡ σφετέρα εἴη τῆς Ἐγεσταίων καὶ Καταναίων, εἰ βουλοίμεθα πρὸς αὐτοὺς ποιεῖσθαι, ἡμεῖς τοίνυν εἱλόμεθα καὶ τότε πόλεμον μὲν ἀντὶ εἰρήνης, … “Again, an urgent request came to us from Syracuse; she was ready to end our differences by a pact of friendship, to end war by peace; and she pointed out the advantages of an alliance with herself, if only we would consent to it, over those of the existing alliance with Segesta and Catana. But once more we chose war instead of peace, …” (tr. Maidment). Here the speaking “I” uses τοίνυν to express his own disalignment with the Athenians (“we”) about the eventual decision to attack Syracuse. The Athenians in this case are not only one of the parties involved but also the interlocutors. More about τοίνυν can be found above, in §36 and §38.
[ back ] 78. On clause-initial “well” warning the interlocutor about a dispreferred (or thought to be dispreferred) reaction, see e.g. Aijmer 2013:40-41. See also III.4.4.2.
[ back ] 79. In addition to τοίνυν, a series of lexical choices create resonance with the previous words by the Melians, while the Athenians’ opinion is opposed to that of the Melians: πρὸς τὸ θεῖον εὐμενείας (5.105.1) resonates with ἐκ τοῦ θείου and ὅσιοι (5.104); the “we” markers ἡμεῖς οἰόμεθα (5.105.1) resonate with πιστεύομεν and ἱστάμεθα (5.104); finally οὐδ’… λελείψεσθαι, in spite of the different scope of the negation, resonates with μὴ ἐλασσώσεσθαι. III.3 is entirely devoted to phonological, morphological, semantic, syntactic, and pragmatic phenomena of resonance in Classical drama. Resonance can often be employed in contexts of disagreement.
[ back ] 80. Plato in Gorgias 484b quotes this fragment from an otherwise unknown song of Pindar: νόμος ὁ πάντων βασιλεὺς θνατῶν τε καὶ ἀθανάτων. See Immerwahr 1966:319-322; Humphreys 1987:211-220; Evans:142-153.
[ back ] 81. See Rood 1998:246-248: the phrase “seem(s) to me” in general hedges authorial assessments. While in Herodotus it stresses “the subjectivity of hypothetical statements,” or works to qualify superlatives, in Thucydides the phrase is employed more rarely and mostly “when he is re-creating past events.” Rood agrees with Marincola (1989) that at 1.22.2 Thucydides’ intention not to write “as it seemed to me” is not to set a polemical statement against Herodotus, but to assess its appropriateness in relation to the reconstruction of the past as opposed to the narration of contemporary history. In the present chapter the phrase μοι δοκέει/μοι δοκεῖ is considered in connection to the particle that often introduces it, that is, καί (see (t13), (t14)).
[ back ] 82. See e.g. Ochs 1996 and Shoaps 2002 on affective stance; Kärkkäinen 2003 and Heritage and Raymond 2005 on epistemic stance. Conrad and Biber (2000:57) define epistemic stance in terms of “commenting on the certainty (or doubt), reliability, or limitations of a proposition, including comments on the source of information,” whereas attitudinal stance refers to “conveying the speaker’s attitudes, feelings, or value judgements.”
[ back ] 83. Du Bois 2007:144-145.
[ back ] 84. On Xerxes’ “pride in his magnificence,” and on Xerxes compared to Zeus, see Immerwahr 1966:177. On Xerxes depicted by Herodotus as a hybristic monarch, see especially Paduano 1978:78-82, and Nicolai 1998:18. Vannicelli (forthcoming) discusses the elements of Xerxes’ speech at 7.8 that mix imperialistic fever and oriental symbolism for absolute monarchy. Note the occurrence of δή, γε, and the double negation in this discourse act. As for historiographical occurrences of these features, see the later sections 4.5. and 4.6 about δή; see above §§40-44, and IV.5 §20, §34, §62, §67-68, §73 about γε; IV.5 §78 about double negation.
[ back ] 85. See Gomme et al. 1945-1981, vol. IV:407.
[ back ] 86. Hornblower (1991-2008, vol. III:733) “one of the most appallingly memorable chapters.” In epic, for example, attention to visual details can have the narrative function of marking a peak or a complicating action by producing special emotional effects, such as horrifying readers who mentally reconstruct the scene. On the effectiveness of terrifying scenes in Beowulf, for example, see Renoir 1962. On the narrative and emotional functions of inserting visual details in Serbocroatian epic, see Bonifazi and Elmer 2016. On conversational storytelling as a regular locus for displaying affective stance, see Voutilainen et al. 2014. Günthner 2011 shows that speakers are able to index their affective stance in language on several grammatical levels, and by doing so they make the related events emotionally accessible to co-participants.
[ back ] 87. On the original concept of focalization, see Genette (1972:206-217; 206-207 for the quoted terms), who distinguishes between “focalisation zéro” (“zero focalization”; the narrator has unlimited access to the mind of characters), “focalisation interne” (“internal focalization”; the narrator’s access is limited spatially), and “focalisation externe” (“external focalization”; the narrator can only be witness to characters). For Bal, instead, (1985:100-115) internal and external focalizations involve no access to the mind, but coincide with the intra- or extradiegetic locus of the focalizer. In general, the term is meant to clarify the distinction between the person who speaks (narratorial functions) and the person who perceives (focalizing functions). The first classicist to adopt the term “focalization” is Irene De Jong in her 1987 monograph Narrators and Focalizers. The Presentation of the Story in the Iliad.
[ back ] 88. See, for example, Connor 1985 about viewpoint; Hornblower 1994:134-135; Hornblower 1991-2008, vol. II:55-56 about Thucydides focalizing Brasidas through the eyes of various characters instead; Dewald 1999:244-248 (on Herodotean focalization as well); De Jong 2001:76-78; Zimmermann 2005; Scardino 2012:68-71; Allan 2013:378-381; Grethlein 2013b: 96-98; Fragoulaki 2013:129; 159. Focalization in Herodotus is discussed in Baragwanath 2008:41, 54, and 214n17.
[ back ] 89. “… we are not just capable of multiple viewpoints; we are in fact incapable of keeping to one single viewpoint of space, or of cognitive structure, when other humans are present” (Dancygier and Sweetser eds. 2012:2).
[ back ] 90. See, e.g. Jeannerod 2001 and Olsson and Nyberg 2011 on brain simulations of action; Addis et al. 2009 on episodic simulation of the past and the future.
[ back ] 91. For example, scholars have different views about whose perspective is adopted and when in Thucydides 2.84.1-3, which narrates how the Athenian general Phormio succeeds in trapping Peloponnesian ships in the gulf of Naupactus, while the Peloponnesians did not expect to get engaged in a sea battle. Just to mention one of the debated matters, Rijksbaron (2012:363-364) attributes imperfects to the primary narrator’s viewpoint; conversely, Grethlein (2013b:95-96) associates the imperfects with the confusion of the ships, and agrees with Bakker (1997e) in interpreting the imperfect tense as an effort to render the events “as if they are seen on the spot.”
[ back ] 92. On the response by the internal audience of speeches in the two Histories, see, for example, Debnar 2001, and Lang et al. 2011:167-188, where the reactions to indirect speeches is analyzed as well. Particle use, however, is not as relevant to this kind of response as it is to the response by external audience/readership.
[ back ] 93. See above §70 for Grethlein’s understanding of “focalization” in terms of “perception.”
[ back ] 94. On this point Grethlein aligns with Dunn: Thucydides employs a variety of techinques “that have the effect of trapping the reader, who is unable to attain a broader perspective, in the present” (Dunn 2007:132).
[ back ] 95. By evoking Plutarch’s assessment that Thucydides makes his readers spectators (De gloria Atheniensium 347a), Grethlein stresses the experiential nature of using enárgeia, that is, vividness.
[ back ] 96. See Grethlein 2013b:109-111 about Thucydides and the openness of the past, the use of sideshadowing, and the opportunity—left to the reader—to compare speeches with what actually follows.
[ back ] 97. See, for example, Grethlein 2010:279 on the activity of reading as “an exercise in the art of reasoning and conjecturing.”
[ back ] 98. Emmott and Sanford (2012:216-218) refer to the notion of “participatory response,” originally introduced by Allbritton and Gerrig (1991), to describe “reader reactions to situations being depicted in narratives” (216). These reactions include fears, hopes, and wishes for or against the characters.
[ back ] 99. 4.6.4 raises the question of the extent of irony in Thucydides as far as the usages of δή are concerned.
[ back ] 100. This take on irony resonates with Fowler’s definition of the term in his dictionary of English originally published in 1926, cited by Schellenberg as well (2009:133n4): “Irony is a form of utterance that postulates a double audience, consisting of one party that hearing shall hear and shall not understand, and another party that, when more is meant than meets the ear, is aware both of that more and of the outsiders’ incomprehension” (Fowler 1994:295).
[ back ] 101. See also Hartung (1832:282), who states that the sense of determination conveyed by δή sometimes is ironic (“enthält bisweilen Ironie”), and Schraut 1847:19. Denniston (1950:229-236) divides the ironic uses of δή according to seven different syntactic environments in which it occurs; “this use is so widespread that it demands separate treatment” (229). Long before Denniston, ironic meanings of δή were noticed by Stephanus 1572:966 (about Thucydides); Vigerus 1834 [1627]:501; Hoogeveen 1749:298 (about Lucian, and in combination with ὡς); Riemer 1823:425; Kühner 1835:388; Döderlein 1858:362 (about Homer); Thomas 1894; Smyth 1956:647.
[ back ] 102. I posit that a tighter relationship with listeners is easier to achieve and makes more sense than a tighter relationship with readers. Moreover, in the course of aural delivery, nonlinguistic features (such as facial gestures and intonation) could further render the ironic slant of some words.
[ back ] 103. See the δή graph in I.5; Herodotus uses δή more than any other author in our corpus.
[ back ] 104. See Cooper 2002:2959, and Stephens 1837:64, about ὅσος δή.
[ back ] 105. See IV.3.11.3.
[ back ] 106. This does not imply that the different δή constructions work as clusters (on which, see I.1 §19); in δὲ δή for example the two particles work independently. Τhe question of the scope of δή will be addressed in §100.
[ back ] 107. See Bybee 2010 and the framework of CxG (Construction Grammar), which stress a very similar point about learning and storing words in general. I.3.3.4 offers an overview of CxG.
[ back ] 108. E.g. in book 2, 100 of 155 occurrences relate to narrative progression.
[ back ] 109. The upcoming analysis is based on 407 occurrences of δή in Herodotus 1 and 2, and all the 201 occurrences found in Thucydides.
[ back ] 110. Scholars claiming the primarily temporal value of δή include Devarius (1588:63); Hoogeveen (1769:276); Hartung (1828 and 1832:223); Kühner (1835:386); Ellendt (1841:166); Döderlein (1858:363); Wetzell (1879:14-15); and Thomas (1894:85).
[ back ] 111. I say “mainly” because sometimes, instead of the more usual δή constructions, we find δή alone performing this function in act-peninitial position. For example, after deciding to live and kill Candaules instead of being killed, Gyges asks Candaules’ wife how to set on the king. Herodotus introduces Gyges’ words by means of Ἐπειρώτα δὴ λέγων τάδε· “He inquired by saying:” (Herodotus 1.11.4). Wakker (1997b:241) also interprets this δή as a device that draws attention to what happens next, but she identifies a nuance that I do not see: her translation is “and so, you have to know, he asked.”
[ back ] 112. “As I said” and “As I was saying” are the ways in which I render what Wakker (1997b:241) and Van Ophuijsen (1993:141-148) call “anaphoric δή” used to continue a story.
[ back ] 113. IV.3 §§127-143 offers several examples of this kind of δή, including discussion of μὲν δή and various particle combinations at different discourse transitions. Literature refers to the “recapitulating” or “resumptive” function of μὲν δή through e.g. Hartung (1832:262-263), Rieckher (1862:473-474), Denniston (1950:225), and Wakker (1997b:242: δή in combination with μέν is used to “round off a topic and prepare a topic switch”).
[ back ] 114. For books 1 the ratio is 17 out of 248 instances; for book 2, just 2 out of 159 instances. Even though this does not represent the average adequately—the earlier books have little direct speech in general (see Solmsen 1943:194)—the ratio of the last two books does not contradict the ratio of the first two: book 8 includes 5 δή in direct speech, out of 107 total occurrences, and book 9 shows 8 instances out of 111 total occurrences. Helpful tables recording the occurrence of character speeches in Herodotus are to be found in De Bakker 2007:183-237. Scardino estimates that direct speech in Herodotus occupies 18% of the work on average (2007:117), while direct speech in Thucydides occupies 20-25% (2007:453).
[ back ] 115. In II.4 §20 δή is interpreted as a sign of emotional convergence between performer and character, which is very relevant to our upcoming account of δή in stancetaking.
[ back ] 116. I already pointed out that authorial statements tend to coincide with acts of stancetaking. Marking narrative progression is different from making authorial statements; however, sharp distinctions between conveying just the voice and conveying voice + stance should be avoided. Some of the combinations that I mentioned above (§89) may be employed to introduce Herodotus’ explicit stance about something rather than referring to the progress of events. An example is 1.58 πρὸς ὃ δὴ ὦν ἔμοιγε δοκέει οὐδέ τι Πελασγικὸν ἔθνος, ἐὸν βάρβαρον, οὐδαμὰ μεγάλως αὐξηθῆναι “Before that, I think, the Pelasgic stock nowhere increased much in number while it was of foreign speech” (tr. Godley).
[ back ] 117. For further discussion about these combinations, see IV.2 §§96-97 and 100-101, and in III.4 §§51-52.
[ back ] 118. At the end of the next subsection (see §100) I will recapitulate the act positions of δή.
[ back ] 119. See III.2 n181 for references (ranging from 1828 to 2019) indicating the main meaning of δή to be linked to something clear, evident, or obvious within and beyond Herodotus and Thucydides. Van Ophuijsen (1993:141 with n2) states, “the basic value of the particle relates to what is visible to the mind’s eye as well as to the organ of sight,” and gives French voici, voilà as the equivalent. Bakker observes: “[δή] conveys that the consciousness verbalized receives its input from the speaker’s immediate environment, from what is perceptually clear and evident” (1997b:75). As I have already mentioned (§84), Cooper (2002:2939, emphasis original) affirms that “the proper meaning of δή is probably clear as the light of day.” Wakker (1997b:238-241) aligns with Ruijgh 1971:646-647 in saying that δή signals “the importance and interest of the proposition presented” (239). She consistently translates δή as “you must know.” For her the meaning “obviously” is a later development of the orginal “look (how interesting).”
[ back ] 120. For δή in direct speech see also Herodotus 1.115.3.
[ back ] 121. The question of whether he is equally certain and committed to obeying Astyages in his heart is a different matter.
[ back ] 122. Book 2 is rich in parallels: see 2.32.6; 2.60.1; 2.68.1; 2.93.4; 2.96.5; 2.111.1; 2.155.2; 2.156.4. An authorial statement that uses δή in connection with perception occurs when Herodotus observes that the inhabitants of Busiris are beating themselves out of grief: Τύπτονται [μὲν] γὰρ δὴ μετὰ τὴν θυσίην πάντες καὶ πᾶσαι, μυριάδες κάρτα πολλαὶ ἀνθρώπων· τὸν δὲ τύπτονται, οὔ μοι ὅσιόν ἐστι λέγειν. “It is here that everybody – tens of thousands of men and women - when the sacrifice is over, beat their breast: in whose honor, however, I do not feel it is proper for me to say,” 2.61.1 (tr. De Sélincourt and Marincola). Artabanus uses γάρ and δή at 7.16γ.1 to refer to a purely visual experience: φανήτω γὰρ δὴ καὶ ἐμοί, ὡς καὶ σοί, διακελευόμενον “let it [the dream] appear to me as well as to you, with its commands” (tr. De Sélincourt and Marincola).
[ back ] 123. Wakker (1997b:243-247) takes δή in causal and purpose constructions with ἵνα, ὅπως, ὅτι + οblique optative, and ὡς + participle as always expressing the viewpoint of the character involved in the cause or purpose in question. If the causal clause is introduced by ἄτε δή, conversely, then the viewpoint expressed is that of the narrator.
[ back ] 124. On δή accompanying superlatives, see Stephens 1837:63, Krüger 1846:347-348, and Denniston 1950:204-207. See also below 4.6.2, II.5 §68, and III.2 §74.
[ back ] 125. Scardino 2012:70.
[ back ] 126. While the fact that σφ- forms mark indirect reflexivity is generally accepted, the interpretation of αὐτός forms is not straightforward at all. For the editorial instability αὑτός/αὐτός in Thucydides, see in particular Powell 1934). The common reading of αὐτός as “plain” third person pronoun when it is not in attributive position can be challenged. Recent investigations into the pragmatic uses of αὐτός in Homer, for example, are found in Bonifazi 2012:134-155.
[ back ] 127. See below §100 on the scope of δή as “certainly x,” and II.3 §62 on the intensifying function of δή (“I feel/think very strongly that …”) in Homer. I consider these δή as reflecting stance that is somewhere between “epistemic” and “attitudinal” stance according to the definition of Conrad and Biber (see above, n82).
[ back ] 128. A very similar analysis of δὴ γάρ is offered at II.4 §§19-20.
[ back ] 129. δή also appears in indirect speech at 1.94.x; 1.96.2; 1.114.4 (also including γε); 2.13.3; 2.36.6; 2.42.2; 2.45; 2.116.2; 2.121.ε.2; 2.122.2; 2.126.1; 2.132.3; 2.134.1; 2.141.1; 2.141.2; 2.160.4; 2.177.1.
[ back ] 130. Wakker offers a different reading (1997b:240): the Persians “call attention to this important proposition, and especially to its focus, … ‘the Greeks’ by the use of dḗ.” She translates δή “you must know.”
[ back ] 131. For δή in indirect thoughts preceded by reporting clauses or phrases, see 1.114.4; 2.17.3; 2.161.4; 2.172.2; 2.174.1. Further examples of the “author – audience” vector will be discussed later in 4.5.5 on “ironic δή.”
[ back ] 132. The same conclusions are drawn in II.3 §64 with regard to δή in epic, and in III.2 §79 with regard to δή in tragedy and comedy.
[ back ] 133. For the term “unframed discourse,” see II.4.2, and also above §27.
[ back ] 134. Hornblower (1994:134) remarks: “Sometimes γάρ introduces material whose focalizer is really Thucydides himself, the obvious example being that at 2.13.3, from the account of Athenian finances ostensibly taking the form of encouragement by Pericles.” Of Herodotus Lang (1984b:154) remarks: “the ubiquity of gar in the Histories … is evidence of the historian’s readiness to explain.” De Jong (2004a:110) takes γάρ clauses in narrator text as a device that provides “explanations to questions which the narrator assumes the narratees will have.”
[ back ] 135. See below n148 about γὰρ δή uttered by characters.
[ back ] 136. Among other instances of δή in explicit authorial statements, see in particular 1.58; 2.5.1; 2.11.1; 2.11.4; 2.49.1; 2.50.1; 2.120.2; 2.120.3; 2.150.2.
[ back ] 137. See also Herodotus 1.14.3; 1.25.2; 2.16.2 (also including γε); 2.22.1; 2.42.2; 2.46.2; 2.52.2; 2.104.4; 2.94.2 (also including γε and a superlative); 2.156.6; 2.164.2.
[ back ] 138. “Dubiousness” is among the terms that Cooper uses to describe the main aspects of the irony suggested by δή, together with “inaccuracy” and “falsehood”: “Especially in ὡς clauses the pretense is portrayed as bold, and the dubiousness is all the more perceptible.” (2002:2939); “the dubiety … relates to the unworthiness or inadequacy of the fact as a justification for the position taken up on this basis.” (2002:2940); “With participles a suggestion of inaccuracy is the rule. … δή … after ὡς implies falsehood or want of the power to convince in the original thought or assertion restated” (2002:2940).
[ back ] 139. Wakker (1997b:245n62) acknowledges that δή may have irony as one its side effects, but she disagrees on the idea that δή may describe, to use Denniston’s words, “an ingenious stratagem or device” (1950:232); it is not the presence of δή that signals this, but rather the entire context. I point out that δή per se would not signal anything anyway, as it makes sense exclusively in context and with its co-text. Along the same lines, irony through δή is to be seen as an effect springing from an entire situation and from extralinguistic implications, which does not fit Wakker’s prescriptive, rather than subjective, meaning of δή as “you must know.”
[ back ] 140. See, e.g., 1.22.1; 1.80.4; 1.86.2; 2.134.1; 2.147.4; 8.76.2.
[ back ] 141. They include 12 τότε δή, 9 ἐνταῦθα/ἐντεῦθεν δή, 7 μὲν δή, 1 ὕστερον δή, 1 ὅτε δή, and 13 οὕτω δή (plus 5 more οὗτος forms). Overall, they constitute the 36% of δή combinations in Thucydides, against at least 71% in Herodotus (see above, §86). Neither καὶ δὴ καί nor καὶ δή ever occur; however, καί τι καί arguably fulfills a similar function to καὶ δὴ καί (see IV.2 §97). A particular construction used by Thucydides and not by Herodotus is δι’ ὃ δὴ καί (1.128.1; 2.21.1; 2.42.1), meaning “this is why, in particular.”
[ back ] 142. οὕτω δή in Thucydides marks narrative peaks at 1.131.1; 2.6.3; 2.12.4; 2.17.3; 2.19.1. τότε δή does the same at least at 2.84.3 and 8.92.2.
[ back ] 143. Τhese include πρὶν δή (5x in Thucydides; 2x in Herodotus) and πρὶν γε δή (2x in Thucydides; 5x in Herodotus). πρὶν γε δή in Thucydides 1.132.5, 7.71.5 and in Herodotus 6.79.2, 9.22.2 marks narrative peaks.
[ back ] 144. The multi-functions of που (including the propositional reference to unspecified places—“somewhere”) are thoroughly investigated in Koier 2013 within the framework of Construction Grammar (on which see I.3.3.4). Even though κου in Herodotus and που in Thucydides do not pose particular interpretive problems, I record here a couple of interesting data. First, the number of occurrences contradicts Denniston, who remarks (1950:491): “The tone of uncertainty, whether real or assumed, is ill-adapted to the precision of history …. There are a few examples in Thucydides.” The TLG online actually includes 51 κου/που in Herodotus (48 κου and 3 που), against 49 που in Thucydides. Additional observations concern δήπου and δή κου. Two out of the four δήπου in Thucydides appear in a passage (8.87.4-5) where the historian expresses his “considerate opinion” on the Phoenician fleet (see Hornblower 1991-2008, vol. III:1004-1107); therefore the co-text relates to authorial stancetaking. As for Herodotus, conversely, only one δή κου out of five works as a cluster meaning “with no doubt”; in the remainder, κου works separately from δή.
[ back ] 145. In n34 I mentioned that at 8.27.3 ἦ που δή occurs in indirect speech.
[ back ] 146. μὲν δή is exceptionally uttered by characters in Thucydides 3.113.4 and 5.90.1. Τhe passage in book 5 features ἦ μὲν δὴ νομιζομέν γε, with which the Melians react to the bold statement by the Athenians “in practical terms, the ones who dominate do what they can, and the ones who are weak assent” (δυνατὰ δὲ οἱ προύχοντες πράσσουσι καὶ οἱ ἀσθενεῖς ξυγχωροῦσιν, 5.89.1; tr. AB). ἦ marks the reaction as such (see III.2.2.10, where ἦ in drama is described as a marker of emotional involvement while speakers assess something or ask for confirmation about something). μὲν δή projects a move that is going to encompass several discourse acts. γε reinforces “we consider” and anticipates that the object of the consideration is going to be retrieved from something that the Athenians already mentioned (the idea of equal necessity, cf. 5.89.1 ἴσης ἀνάγκης). The thought is going to be: “we do believe that there is an advantage in preserving the idea of common good.”
[ back ] 147. Thucydides 2.62.1; 4.92.4; 8.41.2; 6.37.2. Since the most interesting occurrences of γε δή are in authorial statements, I postpone our remarks on this combination to a later point (see n162).
[ back ] 148. Thucydides 1.81.6; 1.122.4; 2.40.3; 3.57.3; 3.66.2; 4.87.4; 5.111.3; 6.76.2; 6.77.1; 7.62.4. γὰρ δή in Thucydides is more frequently spoken by characters than in Herodotus (10 against 5 times); it consistently introduces associative facts that either look evident, or show the speaker’s determination in stating something. See the upcoming (t24).
[ back ] 149. Further instances of δή in direct speech near superlatives or adjectives and pronouns indicating scalarity are Thucydides 1.33.2 ὀλίγοις δή ἅμάπαντα; 1.74.1 and 3.39.1 μάλιστα δή; 7.62.3 δὴ μάλιστα; 1.122.4 πλείστους δή; 2.64.3, 6.13.1, and 6.17.5 μεγίστ- δή; 2.64.3 πλείστων δή; 3.54.5 ὅτεπερ δὴ μέγιστος φόβος; 4.63.2 τὸ ξύμπαν τε δή; 5.113.1 πλεῖστον δή; 1.83.2 οὐδέν δὴ μᾶλλον; 2.64.1 πρᾶγμα μόνον δὴ τῶν πάντων ἐλπίδος κρεῖσσον γεγενημένον. At 1.84.2 δή near μᾶλλον contributes to signal that the speaker performs an act of stancetaking, especially positioning; the speaker, the king of Sparta Archidamus, conveys that they do not want to be persuaded by vexation: ἤν τις ἄρα ξὺν κατηγορίᾳ παροξύνῃ, οὐδὲν δὴ μᾶλλον ἀχθεσθέντες ἀνεπείσθημεν, “if anyone tries to provoke us with accusations, this is no more successful—we are not goaded into agreement” (tr. Hammond). At an earlier point of the same speech (1.81.6), Archidamus uses δή to convey the decisiveness with which he wishes to position himself and his people on their war expectations.
[ back ] 150. More examples perhaps recalling the “irony” of some Herodotean δή are gathered in 4.6.4. A less complex co-text involving δή in indirect speech occurs at 4.23.1.
[ back ] 151. For example, in 8.87.1 we find the following description of Tissaphernes: βουλόμενος, ὡς ἐδόκει δή, ἀπολύεσθαι πρὸς αὐτοὺς τὰς διαβολάς, literally “ … resolved, as it seemed good to him, to dispel the suspicions against them …” (tr. AB). The imperfect of δοκέω suggests that δή marks Tissaphernes’ perception rather than Thucydides’. More examples of δή in reported thoughts are to be found at 2.24.1 and 4.40.1 (by the Athenians); 4.67.3 (by the Megarian cospirators) and 4.117.2 (by the Spartans, after guessing the thoughts of the Athenians); 6.24.2 (by the Athenians after Nicias’ speech).
[ back ] 152. The Corinthians were supposed to sail together with the Spartans towards Chios, first, and then to Lesbos (8.8.2), but they were unwilling to move, because of the imminent Isthmian festival, which would impose a truce. On the religious scruples of the Corinthians, see Hornblower 1991-2008, vol. III:781-783.
[ back ] 153. δή + superlative is considered a “rather fixed combination” (Wakker 1997b:240n51). Scholars remarking that in earlier times include Hoogeveen 1769:290-295; Kühner 1835:389; Stephens 1837:63; Bäumlein1861:102; Wetzell 1879:17; Kalinka 1890:194; Thomas 1894:101. About 8.1.2 κατάπληξις μεγίστη δή, echoed at 8.96.1 ἔκπληξις μεγίστη δή, Hornblower comments: “both times δή is a regular accompaniment of the superlative and indicates no doubt or irony” (1991-2008, vol. III:751). Besides (t27), see 1.1.2 κίνησις γὰρ αὕτη μεγίστη δή; 1.50.2 ναυμαχία γὰρ αὕτη … μεγίστη δή; 2.31.2 στρατόπεδόν τε μέγιστον δὴ τοῦτο ἁθρόον Ἀθηναίων ἐγένετο; 3.113.6 πάθος γὰρ τοῦτο μιᾷ πόλει Ἑλληνίδι ἐν ἴσαις ἡμέραις μέγιστον δή; 7.75.7 μέγιστον γὰρ δὴ τὸ διάφορον τοῦτο; 8.1.2 and 8.96.1 ἔκπληξις μεγίστη δή; 8.41.2 ὃς αὐτοῖς ἔτυχε μέγιστός γε δὴ ὧν μεμνήμεθα γενόμενος. Further superlatives include γὰρ δή … πρῶτος (1.93.4); βεβαιότατα δή (1.138.3); κράτιστος δή (1.138.3; πλεῖστον δή (2.97.3); πλεῖσται δή (3.10.5); δή πλεῖσται (3.17.4); βέλτιστοι δή (3.98.4); πλεῖστον δή (4.74.4); μάλιστα δή (4.108.3, 5.28.2 and 5.66.2); ἐπὶ πλεῖστον δή (6.54.5); πλεῖστα δή (7.56.4); πλεῖσται (7.70.4); ἥκιστα (7.86.5); παρ’ ἐλάχιστον δή (8.76.4); ἐπικαιροτάτην δή (8.106.2). A few more superlatives are quoted in 4.6.3, as they relate to the blending of multiple voices. κατάπληξις μεγίστη δή (Thucydides 8.1.2) is commented on in IV.5 §38.
[ back ] 154. I read πλείστου δὴ χρόνου as a conflation of a temporal expression and the historian’s stance about the chronology of the battle.
[ back ] 155. See e.g. 7.44.1, where the relative clause coincides with an estimation that expresses scalarity: ἐν δὲ νυκτομαχίᾳ, ἣ μόνη δὴ στρατοπέδων μεγάλων ἔν γε τῷδε τῷ πολέμῳ ἐγένετο, “in a night battle—this was the only one fought between large armies in the whole of the war” (tr. Hammond). At 6.105.2 an authorial superlative is accompanied by ἤδη: καὶ τοῖς Λακεδαιμονίοις ἤδη εὐπροφάσιστον μᾶλλον τὴν αἰτίαν ἐς τοὺς Ἀθηναίους τοῦ ἀμύνεσθαι ἐποίησαν. “Thereby the Athenians at last gave the Lacedaemonians a right to complain of them and completely justified measures of retaliation” (tr. Jowett). On ἤδη and superlatives, see below §148.
[ back ] 156. Hornblower (ad loc.) recalls that a similar gnomic content occurs without δή at 2.65.4 (ὅπερ φιλεῖ ὅμιλος ποιεῖν).
[ back ] 157. οἷον δή is to be distinguished from the idiomatic ὅσος δή or ὅτις δή forms that convey approximation. See n104.
[ back ] 158. See Cooper 2002:2941 on δή “in exclamations.” Already the Homeric scholia established a link between δή and emotions; see I.2 §58.
[ back ] 159. See Iliad 5.601; 13.633; 17.587; 21.57; Odyssey 1.32; 5.183; 11.429; 18.221; 20.393. For example, at Odyssey 5.183 Calypso comments on a speech that Odysseus just delivered: ἦ δὴ ἀλιτρός γ’ ἐσσὶ καὶ οὐκ ἀποφώλια εἰδώς, οἷον δὴ τὸν μῦθον ἐπεφράσθης ἀγορεῦσαι. “You definitely are a rogue, and what you know is not pointless; what a speech you devised to perform!” (tr. AB).
[ back ] 160. εἴπας is the participle aorist in the nominative case, unless we think of a “visionary” aorist, second person “what a speech!” (εἴπας being the epic/Ionic variant of Attic εἶπας). Two further occurences of οἷα δή in Herodotus share a similar function. At 1.122.1 the foster parents of Cyrus rejoice when they see him, as they were sure he had died; the text is μεγάλως ἀσπάζοντο οἷα δὴ ἐπιστάμενοι αὐτίκα τότε τελευτῆσαι, which I paraphrase as “they warmly welcomed him—they assumed that he had died then on the spot!” At 6.26.2 the Chians happen to be ruthlessly killed by Histiaeus: καὶ τῶν λοιπῶν Χίων, οἷα δὴ κεκακωμένων ἐκ τῆς ναυμαχίης, ὁ Ἱστιαῖος ἔχων …, which I render: “and of the rest of the Chians—alas, they were already distressed by the seabattle!” I posit that the “already” component comes from the perfect tense of the participle, rather than from δή. At 3.129.3 οἷα δή is uncertain: both Hude and Legrand leave it in square brackets, as it is omitted in a few manuscripts: Τῇ δὲ δὴ ὀγδόῃ ἡμέρῃ ἔχοντί οἱ φλαύρως [οἷα δὴ] παρακούσας τις πρότερον … ἐσαγγέλλει τῷ Δαρείῳ. “On the eighth day, when he [Darius] was doing poorly, someone who had heard … told Darius of him” (tr. Godley). Indeed, this co-text does not resemble those in the other instances.
[ back ] 161. On the sociological connotations of ὄχλος in Thucydides, see in particular Hunter 1988.
[ back ] 162. I associate the relevance of intonation to occurrences of δή when γε δή also appears. In Thucydides, 7 of 10 instances of γε δή come from the author’s voice; the remaining ones appear in direct speech. For example: ἔθνη γὰρ πλεῖστα δὴ ἐπὶ μίαν πόλιν ταύτην ξυνῆλθε, πλήν γε δὴ τοῦ ξύμπαντος λόγου τοῦ ἐν τῷδε τῷ πολέμῳ πρὸς τὴν Ἀθηναίων τε πόλιν καὶ Λακεδαιμονίων “The greatest number of nations, except the general roll of those which in this war adhered to Athens and Lacedaemon, were together at this one city,” Thucydides 7.56.4 (tr. Hobbes). In Herodotus, 5 of 9 γε δή are uttered by the main narrator. Denniston (1950:246) links the combination γε δή in Herodotus and Thucydides to a fortiori statements. According to him, these statements are conclusions that show even stronger evidence than what was previously adduced. Underlying this argumentative force is a more general pragmatic value: the combination (not a cluster) γε + δή allows the source of the utterance to come to the fore, and it marks an act of stance. Its force is equivalent to the prosodic emphasis that we may attach to a specific piece of information for whatever contextual purpose. Cobet (1876:483) reads irony behind some γε/γε δή in Demosthenes. I do not read irony in the two historians but rather the marking of narrative peaks (see above n143 about πρὶν γε δή). Irony with δή in Herodotus is independent of γε, and concerns selected occurrences, as we have seen.
[ back ] 163. In 10 out of 12 cases the source of the utterance including the καλλιστ- form is a character (three times Brasidas; twice Hermocrates and Pericles; once Archidamus; one instance occurs in Xerxes’ letter to Pausanias, and one in Phormio’s thoughts about how to trap the Peloponnesian ships in the Gulf of Corinth). The eleventh occurrence is in narrator text (“beautiful suburbs,” 2.34.5), and the twelfth is (t29). As for the referents of the adjective or pronoun, in 6 cases they have to do with attacks and military strategies in more or less general terms (1.129.3; 2.11.9; 2.84.3; 5.9.4; 6.33.3; 6.33.4). However, the specific referent “army” occurs only here. The connection to Sappho fr. 16.1-3 (Οἰ μὲν ἰππήων στρότον, οἰ δὲ πέσδων, οἰ δὲ νάων φαῖσ’ ἐπὶ γᾶν μέλαιναν ἔμμεναι κάλλιστον) is inevitable for us modern readers. On κάλλος, καλός, and κάλλιστος as referring to something erotically attractive in ancient Greek literature, see Konstan 2015:49-61.
[ back ] 164. Hornblower (1991-2008, vol. III:157-158) notes that Stroud (1994:291) takes the passive form ὤφθη (at the end of the passage) as a sign of authorial autopsy, but he adds: “The autopsy idea is attractive but not certain, and no amount of emphatic language in Thucydides makes it so. As at 59.3 …, Thucydides makes it hard to say how much is Spartan focalization and how much his own comment and opinion.”
[ back ] 165. See e.g. Bakker 1997e; Grethlein 2013b:33.
[ back ] 166. Let us list here further instances of δή where I see Thucydides blurring his voice and stance, or in other words, where I infer his emotional alignment with characters. At 2.77.2 (πᾶσαν γὰρ δὴ ἰδέαν ἐπενόουν, εἴ πως σφίσιν ἄνευ δαπάνης καὶ πολιορκίας προσαχθείη, “they were now thinking of every possible means of securing control without the expense of a siege,” tr. Hammond), the Peloponnesians’ and Thucydides’ thoughts seem to be blended. At 4.55.2 (ἔς τε τὰ πολεμικά, εἴπερ ποτέ, μάλιστα δὴ ὀκνηρότεροι ἐγένοντο, “in all military matters they [the Spartans] became yet more cautious than ever before,” tr. Hammond) the Spartans’ positioning seems to be joined with the author’s. At 5.16.1 (πολλῷ δὴ μᾶλλον προυθυμοῦντο, “[Pleistoanax, son of Pausanias, and Nicias] became yet more enthusiastic for an end to the war,” tr. Hammond) Thucydides seems to align with the two political leaders. At 5.50.4 (ὥστε πολλῷ δὴ μᾶλλον ἐπεφόβηντο πάντες καὶ ἐδόκει τι νέον ἔσεσθαι, “This greatly intensified the general anxiety, and it was thought that there would be a crisis,” tr. Hammond) Thucydides seems to share the feelings of all the people attending the Olympic games. At 7.55.1 (οἱ μὲν Ἀθηναῖοι ἐν παντὶ δὴ ἀθυμίας ἦσαν, “The Athenians were in complete despair,” tr. Hammond) Thucydides seems to share his stance with that of the Athenians. At 7.71.5 (πρίν γε δὴ οἱ Συρακόσιοι καὶ οἱ ξύμμαχοι ἐπὶ πολὺ ἀντισχούσης τῆς ναυμαχίας ἔτρεψάν τε τοὺς Ἀθηναίους, “until … the Syracusans and their allies routed the Athenians,” tr. Hammond) Thucydides seems to join perspective with the Syracusans at a narrative peak. Then, at the end of book 7 several acts including δή seem to convey the author’s emotional alignment with the losing Athenians, e.g. 7.85.4 (πλεῖστος γὰρ δὴ φόνος οὗτος, “this was a most sizable slaughter,” tr. AB). Finally, there is an occurrence of δή + superlative that I find striking: it is found after the account of the spectacle enjoyed by the Athenians on the harbor at the outset of the Sicilian expedition. The Greek reads as follows: οἱ δὲ ξένοι καὶ ὁ ἄλλος ὄχλος κατὰ θέαν ἧκεν ὡς ἐπ’ ἀξιόχρεων καὶ ἄπιστον διάνοιαν. παρασκευὴ γὰρ αὕτη πρώτη ἐκπλεύσασα μιᾶς πόλεως δυνάμει Ἑλληνικῇ πολυτελεστάτη δὴ καὶ εὐπρεπεστάτη τῶν ἐς ἐκεῖνον τὸν χρόνον ἐγένετο. “The foreigners and the rest of the crowd came there to witness for themselves a remarkable enterprise which would otherwise have seemed incredible. This first expeditionary force was indeed at that time the costliest and most magnificent Greek armadaever to sail from a single city,” 6.31.1 (tr. Hammond). It is possible to interpret the passage as encompassing the joint view (in the literal sense of the word) of both Thucydides and the Athenians.
[ back ] 167. By studying the relation between imperfect tense and the characters’ perspective in Thucydides, Bakker (E. J. Bakker 2007:118) comes to a similar conclusion: “the reader has become a witness who observes the events of the war in situ.”
[ back ] 168. This reading is in line with Thucydides’ introduction to this speech: Hermocrates’ overall communicative intention by giving the speech was “to raise the prejudices of the people of Camarina against the Athenians beforehand” (ὁ Ἑρμοκράτης … τῶν Καμαριναίων βουλόμενος προδιαβάλλειν τοὺς Ἀθηναίους ἔλεγε τοιάδε, 6.75.4).
[ back ] 169. This reading of mine harmonizes with Hornblower’s comment: “δή indicates that this is presumed Syracusan and Corinthian thinking, but also that that thinking is precarious” (1991-2008, vol. III:572), and Wakker’s remark (1997b:245): “By the retainment of the subjunctive, the purpose clause introduced by hópōs is explicitly characterized as presented from the perspective of the Lacedaemonians. But it is the narrator who, with dḗ, asks special attention for this purpose, which is not accomplished, as he will show later in the narrative.”
[ back ] 170. Following Schwyzer and Debrunner (1950:562) and Chantraine (DELG, s.v. δή), Beekes (2010:322) includes δῆθεν in “more or less fixed connections” of δή (δή + adverbial suffix -θεν).
[ back ] 171. Up to literature from the fourth century BCE, the TLG online records only 58 tokens of δῆθεν (13 in Herodotus, 12 in Aesopus; 7 in Euripides; 6 in the Corpus Hippocraticum; 5 in Thucydides; 2 in Aeschylus, Ctesias, Aeneas, and Callimachus; 1 in Sophocles, Xenophon, Plato, Aristotle, Xenocrates, Aratus, and Persaeus.
[ back ] 172. See Suidae Lexicon s.v. δῆθεν (Adler 2004:36): ὡς δή φησι· τοῦτο δὲ προσποίησιν ἀληθείας ἔχει, δύναμιν δὲ ψεύδους. λαμβάνεται δὲ τὸ θεν ὡς παραπληρωματικόν “the δή says: “this has the pretension to truth, but the force of a lie. The θεν is taken as a filler.” I thank José Miguel Jiménez Delgado for drawing my attention to this early definition. In modern times, before Denniston, Stephens retrieves the idea of a discrepancy between utterance and actual truth, and provides ante litteram a pragmatic interpretation of the use of the word: “… the speaker, by the use of δῆθεν, may apprize the audience how the matter really stands” (1837:9). Navarre 1904a:325-326 and Menge (1999 (1914): 246) qualify δῆθεν as often ironic. Navarre 1904a is a rare piece entirely devoted to the particle. The author interprets the semantic evolution of δῆθεν starting from a basic temporal sense (“from that moment on”), and evolving into different “logic meanings” (“significations logiques,” 320), in this order: “therefore,” “evidently,” “consequently,” “of course (ironically/sarcastically),” and “marking the assertion as false.”
[ back ] 173. Note the occurrence of the superlative as well, which resonates with the superlative at 74.1 ἐν αἴνῃ μεγίστῃ.
[ back ] 174. See Navarre 1904a:326: δῆθεν marks that the uttered assertion is contrary to the reality, or at least “qu’on ne la garantit pas, qu’on en laisse la responsabilité à l’intéressé” (“one does not guarantee it, one leaves the responsibility up to the interested person”).
[ back ] 175. Readers may assume that Prexaspes disaligns with the Magi’s evaluation as well; therefore, not all the characters involved are ultimately unaware of the actual outcome. However, Herodotus makes sure that readers do not know how Prexaspes receives the Magi’s orders, and how he will react to the situation, thereby maintaining a high level of suspense. For this reason I do not claim that δῆθεν includes the disalignment of Prexaspes together with Herodotus’: by means of δῆθεν the author establishes dialogue directly with readers, rather than with other co-participants in the story. De Jong (2004a:111) takes Herodotean δῆθεν as a way to draw the narratees’ attention to the ongoing deceit.
[ back ] 176. At 1.59.4 δῆθεν marks a stratagem by Pisistratus; at 1.73.5 it marks the macabre pretense of animal flesh for human flesh; at 3.136.2 a trick by Democedes; at 6.1.1 a lie by Histiaeus; at 6.39.1 feigned ignorance about the murder of Cimon; at 7.211.3 a Spartan trick aimed at deceiving the Persians (De Sélincourt and Marincola ad loc. renders δῆθεν by translating “amongst the feints they employed was to … pretend to be retreating in confusion”); at 8.5.1 Themistocles pretending to pass Eurybiades his own money; at 8.6.2 the Persians’ deceptive plan to wait before attacking the Greeks at Artemisium; at 9.66.3 Artabazus making as if to do battle while the Persians start the retreat after Plataea; at 9.80.3 the helots’ unawareness of selling gold to the Aeginetans as if it was brass; at 9.99.3 a Persians’ trick to move the Milesians away on the pretext that they knew the area of Mycale. At 6.138.4 δῆθεν exceptionally appears in a question that functions as indirect thought: the Pelasgians, worried about the dominance of the Attic mothers’ children, wonder what they will do when they reach adulthood: τί δὴ ἀνδρωθέντες δῆθεν ποιήσουσι. The discrepancy in stance in this case is between the certainty with which the Pelasgians foresee the threat of the Attic boys, and Herodotus’ disalignment with this line of thought, through δῆθεν. From δῆθεν the reader is warned against the characters’ forecast, and is invited to judge the Pelasgians’ decision to kill all those children and also the mothers as drastic, at the very least. Powell (1938:85) comments on δῆθεν in this passage: “perhaps merely emphatic.”
[ back ] 177. At 1.92.1 the Spartans are said to propose a policy “for the common good δῆθεν”; at 127.1 the Spartans want to drive out a curse “to honor the gods δῆθεν”; at 3.68.1 the Spartans “deem worthy to rest from conflicts [with the Plataeans] δῆθεν”; at 4.99.1 the Boeotians do not want “to make truce with the Athenians δῆθεν.” In these instances too I read the scope of δῆθεν as embracing the host discourse act.
[ back ] 178. Ellendt (1835:166-167) sees δή as diachronically derived from ἤδη, but the opposite trajectory (ἤδη derives from δή) is generally accepted.
[ back ] 179. For studies that embrace a discourse perspective, see e.g. Revuelta Puigdollers 2006b on πάλιν; Wakker 2001 on ἔτι, Jiménez Delgado 2014 on αὖ and αὖθις, and Bonifazi 2012:263-273 on αὖτις.
[ back ] 180. “Situation-external” vs. “situation-internal” time is a distinction established by Comrie (1976:5); the former locates states of affairs with respect to other states of affairs or to the time of the utterance (typically expressed by tense), while the latter concerns the way in which states of affairs are temporally constituted per se (typically expressed by aspect).
[ back ] 181. E.g. Van der Auwera 1993 and 1998; Van Baar 1997. König 1991:157-162 discusses the focus particles that work as “phasal quantifiers” in different modern European languages; the author considers non-temporal scales as well (159).
[ back ] 182. E.g. Löbner 1989; Vandeweghe 1992. Polarity in linguistics refers to the relative orientation denoted by certain lexical items.
[ back ] 183. E.g. Michaelis 1996; Hansen 2008:171-172.
[ back ] 184. Let us suppose the following context and example, which I take from Hansen (2008:109): John and Peter agree to meet at some airport at 3pm on a certain day to leave for two different cities. Peter has to fly to Madrid at 4pm. Something causes John to reach the airport only at 4pm. John can utter “Peter is already on his way to Madrid” felicitously at 4pm even though the time of the utterance coincides with Peter’s departure time, as the sense of “early” for John subjectively relates to his own delay, and not to the objective schedule of Peter’s flight.
[ back ] 185. See e.g. Van der Auwera 1998 on propositional vs. context-dependent discursive functions of adverbs in a temporal-aspectual dimension;
Bazzanella, Bosco, Calaresu, Garcea, Guil Povedano, Radulescu 2005
 and 2007, and also Hansen and Strudsholm 2008 on the multiple uses of T (temporal) - A (aspectual) – M (modal) markers across Romance languages; Hansen 2002 and 2008 on French temporal-aspectual adverbs déjà “already” and encore “still”; Squartini 2013 “From TAM to discourse: The role of information status in Northern-Italian già ‘already’,” in Degand, Cornilie, and Pietrandrea eds. 2013:163-190.
[ back ] 186. See Squartini 2013:178: già/eh già are seen as “sheer confirmative particle[s]”; ah già marks “sudden remembrance”; 2013:175 di già is an “interjectional form restricted to a temporal-aspectual interpretation.”
[ back ] 187. All the quotations are in Kroon and Risselada 1998:444.
[ back ] 188. Rosén (2009:360) agrees with Kroon and Risselada only partially: it should be added that iam can highlight also an entire sentence, besides single constituents.
[ back ] 189. Even though polarity is not relevant to our study of ἤδη in Herodotus and Thucydides, for the sake of clarity I note that Wakker (2002:13) reads ἤδη as indicating positive polarity (exactly like “already” in English), and ἔτι as indicating negative polarity (like “still” in English).
[ back ] 190. … non simpliciter ἤδη tempus praesenti coniunctum sive praeteritum sive futurum indicat, sed eius particulae propria vis est ea, ut significet, factum esse aliquid aut fieri aut futurum esse, antequam quis putet vel putaverit “ἤδη does not simply indicate a time connected with the present, or the past, or the future, but its proper force is that of signifying that something happened, or is happening, or will happen before one believes/expects or has believed/expected” (Heller 1853:260) Later in his letter, Heller agrees with Hartung on one aspect by saying: Bene Hartungius vidit, particulae ἤδη locum esse, ubi quae nondum evenerunt, cogitatione praecipiuntur: cuius usus ratio … sponte apparet, quippe in eiusmodi sententiis quod fortasse futurum est, iam praesens et antequam quis expectaverit esse fingatur. “Hartung sees right that the occurrence of ἤδη is where things that did not take place yet are perceived with the mind’s thought: the reason of this use … is clear in itself, for in sentences of this type what will possibly happen is pretended to be already present, and before one had expected” (Heller 1853:275; tr. AB).
[ back ] 191. See δή in ORPS (our online database) for Hartung’s accounts and for the considerable amount of early literature on the relation between δή and ἤδη.
[ back ] 192. I may add that Aristotle’s thought provides insights also into the use of past tenses with ἤδη; in fact, what distinguishes an acceptable from an unacceptable ἤδη utterance is not the specific past tense being used (the examples display a perfect in both cases), but the culturally-bound knowledge of events that are nearer or farther from “now,” whatever moment “now” indicates.
[ back ] 193. E.g. ἀπὸ ταύτης ἤδη ἐς αὐτόν τε τὸν Θερμαῖον κόλπον ἐγίνετο τῷ ναυτικῷ στρατῷ <ὁ> πλόος, Herodotus 7.123.3; καὶ οὕτως ἤδη κατὰ κράτος ἡ Ποτείδαια ἀμφοτέρωθεν ἐπολιορκεῖτο, Thucydides 1.64.3. See also Thucydides 1.118.1; 4.78.6; 5.38.1; 5.76.3; 6.48.1.
[ back ] 194. See above, §§86 and 89. In Herodotus ἤδη ὦν occurs 12 times.
[ back ] 195. See, e.g. Herodotus 4.86.1 Ἤδη ὦν | ἐς μὲν Φᾶσιν ἀπὸ τοῦ στόματος …. The combination occurs also at 1.207.3; 1.209.4; 2.49.1; 2.144.1; 2.172.5; 3.155.4; 4.31.2; 4.100.2; 6.53.2; 7.184.3; 8.100.5. Powell (1938:160) sees especially in ἤδη ὦν, but also in ἤδη γάρ the function of introducing a logical step in an argument. More recently, in his commentary to Herodotus 8, Bowie (2007:191) remarks: “ἤδη ὦν, ‘so then’ combination often used by H[erodotus] to introduce a point that follows logically from what precedes.” Probably Bowie has in mind Powell on this point (see above §137). On καὶ δή as a separate act, see IV.2 §100n153.
[ back ] 196. Herodotus 1.68.6; 2.175.5; 4.18.2; 7.35.1; 7.55.3; 9.14.1; 9.95.1; Thucydides 1.50.5; 1.95.1.
[ back ] 197. See above, §§86 and 89.
[ back ] 198. See Thucydides 4.35.2, 6.44.3, and 7.44.1 for ἐνταῦθα ἤδη; Herodotus 2.25.3; 2.181.4; 7.129.3; 7.225.2; 8.34.1; 9.102.2 for ἐνθεῦτεν ἤδη. At least οn three occasions ἐνθεῦτεν ἤδη introduces a narrative peak in Herodotus: see 2.25.3 καὶ τὸ ἐνθεῦτεν ἤδη ὁμοίως; 7.225.2 ἐνθεῦτεν ἤδη ἑτεροιοῦτο τὸ νεῖκος, and 9.102.1 ἐνθεῦτεν ἤδη ἑτεροιοῦτο τὸ πρῆγμα; in all these cases the discourse act marks a significant change in the course of action.
[ back ] 199. See Herodotus 9.7.1 καὶ ἤδη ἐπάλξις ἐλάμβανε, and Thucydides 6.98.3 καὶ ἤδη ἀντιπαρατασσομένων … τῶν Συρακοσίων στρατηγοὶ … ἀνήγαγον. The remaining 8 instances of καὶ ἤδη in Thucydides are not act-initial, and they do not work as clusters. In Thucydides 7.59.1 we find καὶ τότε ἤδη πᾶσαι ἀμφοτέροις παρῆσαν καὶ οὐκέτι οὐδὲν οὐδετέροις ἐπῆλθεν to describe the two opposing sides at one of the climactic moments of the last battles in the Great Harbor in Syracuse.
[ back ] 200. Ηerodotus uses only ἤδη γάρ (resembling Homeric δὴ γάρ), and only twice (4.45.3 and 6.53.1); Thucydides, conversely, uses ἤδη γάρ at 2.77.4; 4.93.1, 4.130.6; 6.29.1; 7.2.2; 7.6.1; 7.44.3; 8.42.1; and γὰρ ἤδη at 1.51.2; 2.31.1; 2.63.2; 3.52.3; 3.90.2; 3.97.3; 7.31.5.
[ back ] 201. ἤδη and μᾶλλον co-occur in the same discourse act several times; see Thucydides 1.3.2, 2.94.4, 6.105.2, 7.57.9 (ἤδη … μᾶλλον); 1.8.3, 1.8.4, 1.49.7, 2.48.3, 7.8.3, 7.37.1, 7.43.7, 8.71.3 (μᾶλλον ἤδη); 6.49.4, 6.59.2, 7.4.4 (ἤδη μᾶλλον). See also Herodotus 7.223.2 ἤδη πολλῷ μᾶλλον.
[ back ] 202. See LSJ s.v. ἤδη, I.5 with superlatives, and with comparatives. See also Powell 1938:160.
[ back ] 203. The upcoming (t)s will display all of these features.
[ back ] 204. See also, e.g., Herodotus 7.9α.2 Ἐπειρήθην δὲ καὶ αὐτὸς ἤδη ἐπελαύνων ἐπὶ τοὺς ἄνδρας τούτους ὑπὸ πατρὸς τοῦ σοῦ κελευσθείς “I myself have made trial of these men [the Greeks of Asia], when by your father’s command I marched against them” (tr. Godley). ἤδη stresses Mardonius’ own experience of the Greeks of Asia rather than the fact that this happened in the past. I observe that the scope of ἤδη in this act is over καὶ αὐτός.
[ back ] 205. See II.3 §§33-37 on reading act-initial ἤδη in Homer as ἦ δή. Below (§§155-156) I will mention a couple of passages where act-initial ἤδη seems to evoke the performance of ἦ δή viva voce.
[ back ] 206. For ἤδη accompanying the superlative form of an adjective direct speech, see, e.g. Herodotus 1.30.2: at the beginning of the long exchange between Croesus and Solon, Croesus introduces his first question by prefacing it with a preliminary move: “we have heard a lot about you because of your wisdom and of your wanderings, how as one who loves learning you have traveled much of the world for the sake of seeing it.” Then the actual question follows: νῦν ὦν ἐπειρέσθαι σε ἵμερος ἐπῆλθέ μοι εἴ τινα ἤδη πάντων εἶδες ὀλβιώτατον. “… so now I desire to ask you who is the most fortunate man you have seen” (tr. Godley). Neither “already” nor “now” can render what ἤδη does, for the word that immediately follows ἤδη is πάντων, which reflects non-temporal scalarity. Croesus is not interested in knowing at which moment in time Solon saw the happiest man, but, rather, in the greatest extent of Solon’s knowledge and judgment. By means of intensifying πάντων (“the most blessed man out of everyone you have seen”), or ὀλβιώτατον (“THE most blessed man ever”), or πάντων εἶδες ὀλβιώτατον (“the most blessed man you have seen ever”), ἤδη marks Croesus’ evaluating stance towards Solon’s magnitude of wisdom.
[ back ] 207. For further examples of direct speech in Thucydides where ἤδη marks stance with no focus on time, see 1.69.5, with a πολύς form (πολλὰ ἡμᾶς ἤδη τοῖς ἁμαρτήμασιν αὐτῶν μᾶλλον ἢ τῇ ἀφ’ ὑμῶν τιμωρίᾳ περιγεγενημένους); 1.80.1, again with a πολύς form (Καὶ αὐτὸς πολλῶν ἤδη πολέμων ἔμπειρός εἰμι); 1.42.2, with an evaluative adjective (καὶ οὐκ ἄξιον ἐπαρθέντας αὐτῷ φανερὰν ἔχθραν ἤδη καὶ οὐ μέλλουσαν πρὸς Κορινθίους κτήσασθαι); 3.48.2, again with an evaluative adjective (τάδε γὰρ ἔς τε τὸ μέλλον ἀγαθὰ καὶ τοῖς πολεμίοις ἤδη φοβερά); with comparative forms of πολύς (3.11.1 καὶ πρὸς τὸ πλέον ἤδη εἶκον τοῦ ἡμετέρου ἔτι μόνου ἀντισουμένου; 4.62.3 γνοὺς ὅτι πλείους ἤδη καὶ τιμωρίαις μετιόντες τοὺς ἀδικοῦντας); with a superlative form (6.34.9 τὸ δ’ ἤδη τὰς μετὰ φόβου παρασκευὰς ἀσφαλεστάτας νομίσαντας; 6.84.2 καὶ ἐν τούτῳ προσήκετε ἤδη ἡμῖν τὰ μέγιστα; 7.66.2 καὶ ἀρχὴν τὴν ἤδη μεγίστην τῶν τε πρὶν Ἑλλήνων καὶ τῶν νῦν κεκτημένους); finally, see 1.82.3, where ἤδη relates to a verb expressing an opinion (καὶ ἴσως ὁρῶντες ἡμῶν ἤδη τήν τε παρασκευὴν καὶ τοὺς λόγους …), and 3.37.1, where ἤδη relates to a personal assessment whose focus is on ἔγωγε (Πολλάκις μὲν ἤδη ἔγωγε καὶ ἄλλοτε ἔγνων δημοκρατίαν ὅτι ἀδύνατόν ἐστιν ἑτέρων ἄρχειν). Further examples of direct speech in Herodotus where I read ἤδη marking stance with no direct focus on time are 3.53.4, with a πολύς form (Πολλοὶ δὲ ἤδη τὰ μητρώια διζήμενοι τὰ πατρώια ἀπέβαλον); 7.18.2, again with a πολύς form (Ἐγὼ μέν, ὦ βασιλεῦ, οἷα ἄνθρωπος ἰδὼν ἤδη πολλά τε καὶ μεγάλα πεσόντα πρήγματα ὑπὸ ἡσσόνων); 7.157.2, with an evaluative adjective (τοῦτο δὲ ἤδη δεινὸν γίνεται); 7.235.3, again with an evaluative adjective (καταδουλωθείσης δὲ τῆς ἄλλης Ἑλλάδος ἀσθενὲς ἤδη τὸ Λακωνικὸν μοῦνον λείπεται); 4.150.3, with a comparative form (Ἐγὼ μέν, ὦναξ, πρεσβύτερός τε ἤδη εἰμὶ καὶ βαρὺς ἀείρεσθαι); 4.134.2, with scope over the verb δοκέω (Ὡς ὦν οὕτω ἤδη δοκεόντων καὶ αὐτῷ μοι ἔχειν).
[ back ] 208. Examples include: the beginning of Egypt after Lake Serbonis (ἀπὸ ταύτης ἤδη Αἴγυπτος, 3.5.3); utter desert occupying the area northward of the territory of the Androphagi (Ἤδη δὲ κατύπερθε τούτων ἔρημός ἐστι ἐπὶ πολλόν, 4.18.3); the fact that Asia further east of India is inhabited (τὸ δὲ ἀπὸ ταύτης ἔρημος ἤδη τὸ πρὸς τὴν ἠῶ, 4.40.2); Libya adjoining Egypt (ἀπὸ γὰρ Αἰγύπτου Λιβύη ἤδη ἐκδέκεται, 4.41.1); ancient Scythia beginning from the Danube (Ἀπὸ Ἴστρου αὕτη ἤδη <ἡ> ἀκταίη Σκυθική ἐστι, 4.99.2). There is a connection to δή in contexts of visual perception; see e.g. Τὸν ἐγὼ ἤδη εἶδον λόγου μέζω (2.148.1), where the author indicates autopsy: he has seen the Egyptian monumental labyrinth himself.
[ back ] 209. ἤδη accompanies ἤκουσα also at 4.77.1 Καίτοι τινὰ ἤδη ἤκουσα λόγον ἄλλον ὑπὸ Πελοποννησίων λεγόμενον; 7.35.1 Ἤδη δὲ ἤκουσα ὡς …; 9.95.1 Ἤδη δὲ καὶ τόδε ἤκουσα ὡς …. See also 2.29.1 τὸ δ’ ἀπὸ τούτου ἀκοῇ ἤδη ἱστορέων ….
[ back ] 210. Further explicit authorial statements including ἤδη occur at Herodotus 2.9.2 (ἤδη μοι καὶ πρότερον δεδήλωται); 6.53.2. ἤδη ὦν ὀρθῷ λόγῳ χρεωμένῳ μέχρι Περσέος ὀρθῶς εἴρηταί μοι; 1.207.3 Ἤδη ὦν ἐγὼ γνώμην ἔχω …; 2.29.1 αὐτόπτης ἐλθών, τὸ δ’ ἀπὸ τούτου ἀκοῇ ἤδη ἱστορέων. At 8.105.1 the co-occurrence of μέγιστος + πάντων makes the statement an explicit evaluation: Ἐκ τούτων δὴ τῶν Πηδασέων ὁ Ἑρμότιμος ἦν τῷ μεγίστη τίσις ἤδη ἀδικηθέντι ἐγένετο πάντων τῶν ἡμεῖς ἴδμεν “Hermotimus, who came from Pedasa, had achieved a fuller vengeance for wrong done to him than had any man whom we know.” Further instances of ἤδη in unmistakable (if implicit) authorial statements occur, e.g., at 4.45.3 Ἤδη γὰρ Λιβύη μὲν ἐπὶ Λιβύης λέγεται ὑπὸ τῶν πολλῶν Ἑλλήνων ἔχειν τὸ οὔνομα γυναικὸς αὐτόχθονος “For Libya is said by most Greeks to be named after a native woman of that name”; 7.129.3 ἐπεὰν δὲ συμμιχθέωσι τάχιστα, ἐνθεῦτεν ἤδη ὁ Πηνειὸς τῷ οὐνόματι κατακρατέων ἀνωνύμους τοὺς ἄλλους εἶναι ποιέει “As soon as they are united, the name of the Peneus prevails, making the rest nameless.” All translations are by Godley.
[ back ] 211. Besides ἤδη in Herodotean explicit statements, I counted 28 instances occurring in character speech (20%); ἤδη in Thucydides appears 69 times in character speech (18.4%).
[ back ] 212. More below §161 on ἤδη as “now.”
[ back ] 213. At Thucydides 3.52.3 a short parenthetical γάρ move including ἤδη looks authorial: οἱ δέ (ἦσαν γὰρ ἤδη ἐν τῷ ἀσθενεστάτῳ) παρέδοσαν τὴν πόλιν “and they (for they were now at the weakest) delivered up the city” (tr. Hammond). “They” is the Plataeans, who were forced by circumstances to capitulate to the Peloponnesians. Note the presence of the superlative ἀσθενεστάτῳ, which connotes the act as an act of evaluating.
8.20.1 is a less straightforward passage: [καὶ ναύαρχος αὐτοῖς ἐκ Λακεδαίμονος Ἀστύοχος ἐπῆλθεν,] ᾧπερ ἐγίγνετο ἤδη πᾶσα ἡ ναυαρχία “[At which time there came also unto them from Lacedaemon for commander, Astyochus], who was now admiral of the whole navy” (tr. Hobbes). Hammond translates: “who was just now succeeding as overall admiral-in-chief.” Neither translation takes into account that starting with Astyochus ναυαρχία as an institutional profession could have changed. In Smith et al. 1890, Smith s.v. navarchus clarifies: [S]omething more like an admiralty was needed when the new phase of the Peloponnesian War, after the campaigns at Syracuse, extended the sphere of Spartan naval enterprise. The expression in Thuc. 8.20 regarding Astyochus, ᾧπερ ἐγίγνετο ἤδη πᾶσα ἡ ναυαρχία perhaps indicates, by the use of the imperfect tense, that the office grew out of circumstances at that time.” On the historical complexity of the institution of nauarchs, and on the same passage, see, more recently, Hornblower 1991-2008, vol. III: 806-807. Hornblower translates ἤδη “by this time.” My analysis of ἤδη leads to an interpretation that harmonizes with Smith’s suggestion. ᾧπερ ἐγίγνετο ἤδη πᾶσα ἡ ναυαρχία can be read as an implicit authorial discourse act that evaluates the “wholeness” (whatever that means) of the admiralty that started with Astyochus. At 1.22.1 ἤδη does occur within an explicit methodological statement by Thucydides, but the scope of ἤδη is the participial phrase ἐν αὐτῷ ἤδη ὄντες, which I connect to epistemic stance about time (see below, §160). Here the author adopts the hic et nunc of the political and military speakers.
[ back ] 214. This is how I explain the difference between, for example, the same genitive absolute with and without ἤδη, when the verb is in present tense. I read the contribution of ἤδη in τοῦ θέρους ἤδη τελευτῶντος (Thucydides 4.133.4), as opposed to τοῦ δ’ αὐτοῦ θέρους τελευτῶντος (Thucydides 3.86.1) as follows: ἤδη marks someone’s experience of the ending of summer from within the process (which can be linked to the inchoative interpretation “as summer was about to end”); conversely, if ἤδη does not occur, the end of summer is reported without any marker of participants experiencing it.
[ back ] 215. E.g. in Thucydides 7.77.6, the general Nicias, after a parenthetical γάρ move, invites the Athenian troops to consider themselves secure, if they reach some friendly territory of the Sicels; he uses ἤδη and an imperative present: ἤδη νομίζετε ἐν τῷ ἐχυρῷ εἶναι.
[ back ] 216. Further instances of ἤδη preceded by semantic clues about thinking which feature forms of νομίζειν include Thucydides 4.115.3 νομίσαντες ταύτῃ ἑαλωκέναι ἤδη τὸ χωρίον; 4.128.3 νομίζοντες καὶ ἐν μεθορίοις εἶναι αὐτοὺς ἤδη καὶ διαπεφευγέναι; 4.129.2 ἀδύνατος ἤδη ἐνόμιζεν εἶναι διαβὰς τιμωρεῖν; 8.15.1 καὶ νομίσαντες μέγαν ἤδη καὶ σαφῆ τὸν κίνδυνον σφᾶς περιεστάναι. Forms of δοκεῖν occur in Herodotus 6.104.2 δοκέοντά τε εἶναι ἐν σωτηρίῃ ἤδη, …; 3.153.2 and 3.154.1 ἐδόκεε εἶναι ἁλώσιμος ἤδη ἡ Βαβυλών, and Ὡς δέ οἱ ἐδόκεε μόρσιμον εἶναι ἤδη τῇ Βαβυλῶνι ἁλίσκεσθαι. Finally, more specific verbs recalling different mental activities include οἴομαι (Thucydides 8.92.7 οἵ τε γὰρ ἐν τῷ ἄστει ἤδη ᾤοντο τόν τε Πειραιᾶ κατειλῆφθαι καὶ τὸν ξυνειλημμένον τεθνάναι); καταγιγνώσκειν (Thucydides 7.51.1 ὡς καὶ αὐτῶν κατεγνωκότων ἤδη μηκέτι κρεισσόνων εἶναι σφῶν μήτε ταῖς ναυσὶ μήτε τῷ πεζῷ); ἐπιμελέσθαι (Thucydides 7.8.3 ὁ δὲ τὰ κατὰ τὸ στρατόπεδον διὰ φυλακῆς μᾶλλον ἤδη ἔχων ἢ δι’ ἑκουσίων κινδύνων ἐπεμέλετο); and verbs of fear (Thucydides 4.28.2 δεδιὼς ἤδη καὶ οὐκ ἂν οἰόμενός οἱ αὐτὸν τολμῆσαι ὑποχωρῆσαι).
[ back ] 217. The occurrence of κατὰ κράτος significantly occurs also elsewhere near ἤδη; see Τhucydides 1.64.3 καὶ οὕτως ἤδη κατὰ κράτος ἡ Ποτείδαια ἀμφοτέρωθεν ἐπολιορκεῖτο; 3.18.5 καὶ ἡ μὲν Μυτιλήνη κατὰ κράτος ἤδη ἀμφοτέρωθεν καὶ ἐκ γῆς καὶ ἐκ θαλάσσης εἴργετο; 5.116.3 καὶ κατὰ κράτος ἤδη πολιορκούμενοι; 8.1.2 καὶ τοὺς αὐτόθεν πολεμίους τότε δὴ καὶ διπλασίως πάντα παρεσκευασμένους κατὰ κράτος ἤδη καὶ ἐκ γῆς καὶ ἐκ θαλάσσης ἐπικείσεσθαι (on which see IV.5 §41). See also κατ’ἀνάγκην ἤδη, Thucydides 8.2.3.
[ back ] 218. ἤδη can also be found in discourse acts that are not preceded by clear reporting clauses or phrases but which still constitute someone’s thought, calculation, supposition, or expectation. Linguistic features in these instances include verba videndi (“seeing” in the sense of “understanding” or “realizing”), and forms of μέλλειν that characterize the ongoing action as subjectively perceived. See Thucydides 2.90.4 ἰδόντες δὲ οἱ Πελοποννήσιοι κατὰ μίαν ἐπὶ κέρως παραπλέοντας καὶ ἤδη ὄντας ἐντὸς τοῦ κόλπου τε καὶ πρὸς τῇ γῇ, ὅπερ ἐβούλοντο μάλιστα, “The Peloponnesians, when they saw them sail in one long file, galley after galley, and that they were now in the gulf and by the shore (which they most desired) ...” (tr. Hobbes); Τhucydides 7.4.4 προσεῖχέ τε ἤδη μᾶλλον τῷ κατὰ θάλασσαν πολέμῳ, ὁρῶν τὰ ἐκ τῆς γῆς σφίσιν ἤδη, ἐπειδὴ Γύλιππος ἧκεν, ἀνελπιστότερα ὄντα “And he [Nicias] was already beginning to concentrate more on the naval dimension of the war, as he could see that the arrival of Gylippus had now made the Athenians’ prospects by land less encouraging” (tr. Hammond; note the occurrence of μᾶλλον and ἀνελπιστότερα); 7.69.2 καὶ ὁρῶν οἷος ὁ κίνδυνος καὶ ὡς ἐγγὺς ἤδη (note the evaluative adjective ἐγγὺς). See also Thucydides 6.31.1 ὡς ἤδη ἔμελλον μετὰ κινδύνων ἀλλήλους ἀπολιπεῖν lit. “as they were departing with dangers”; 6.96.1 καὶ μέλλοντας ἤδη ἐπὶ σφᾶς ἰέναι; 4.66.3 ῥᾷον δ’ ἤδη ἔμελλον προσχωρήσειν τούτου γεγενημένου.
[ back ] 219. The occurrences of ἤδη in indirect speech are introduced by verbs of saying, or by other verbs that either imply the use of voice, or signal a specific source of the account. Evidence from Herodotus is scarce: see 6.67.3 and 8.108.4 (reported speech depending on ἐκέλευε). Thucydides, on the other hand, offers plenty of examples. See 5.49.3 Ἠλεῖοι δὲ τὴν παρ’ αὐτοῖς ἐκεχειρίαν ἤδη ἔφασαν εἶναι “The Eleans replied that the truce … was already in force throughout Elis”; 7.49.3 [Nicias] ἔφη … ἀλλ’ ὅτι τάχιστα ἤδη ἐξανίστασθαι “they should move elsewhere now, as soon as possible”; 3.2.3 [Τενέδιοι γὰρ ὄντες αὐτοῖς διάφοροι καὶ Μηθυμναῖοι καὶ αὐτῶν Μυτιληναίων … μηνυταὶ γίγνονται τοῖς Ἀθηναίοις ὅτι] … καὶ εἰ μή τις προκαταλήψεται ἤδη, στερήσεσθαι αὐτοὺς Λέσβου “[But people from Tenedos (no friend of Lesbos), and Methymna, and some individuals from Mythilene itself … reported to the Athenians that] … if no pre-emptive action was taken immediately, Lesbos would be lost to Athens” (all tranlsations are by Hammond). Modern punctuation after Stuart and Jones has a comma after προκαταλήψεται; according to my reading ἤδη would make sense also as the beginning of the main clause (εἰ μή τις προκαταλήψεται, ἤδη στερήσεσθαι αὐτοὺς Λέσβου). See also 6.48.1 οὕτως ἤδη Συρακούσαις καὶ Σελινοῦντι ἐπιχειρεῖν (Alcibiades’ reported speech); 6.49.4 τούς τε ἄλλους Σικελιώτας οὕτως ἤδη μᾶλλον καὶ ἐκείνοις οὐ ξυμμαχήσειν καὶ σφίσι προσιέναι καὶ οὐ διαμελλήσειν περισκοποῦντας ὁπότεροι κρατήσουσιν (Lamachus’ reported speech); 8.91.2 and 8.92 6 (Teramene’s reported speech). In Thucydides 8.27.4; 8.46.4; 4.115.5; and Herodotus 8.108.4 the reported speech including ἤδη is introduced by κελεύειν. In Herodotus 7.197.2 ἤδη is part of an account that Xerxes’ guides tell to the king (7.197.1 ἔλεγόν οἱ ἐπιχώριον λόγον).
[ back ] 220. See for example Herodotus 8.49.1 ἡ γὰρ Ἀττικὴ ἀπεῖτο ἤδη, | τῶν δὲ λοιπέων πέρι προετίθεε; Thucydides 2.48.3 καὶ ἔθνῃσκον πολλῷ μᾶλλον ἤδη. | λεγέτω μὲν οὖν περὶ αὐτοῦ …. Unlike particles occurring at the end of priming acts (such as X δέ | …) these ἤδη do not project further acts; they just give them a conclusion. In these cases ἤδη has scope either over the entire act, like ἤδη in initial or peninitial position, or it works as a kind of confirmatory tag-assessment (“it is/was really so”). Thucydides uses ἤδη in this usage more than Herodotus (38 to 9 instances). In our corpus only δῆθεν and ἄρα sometimes fulfill a similar function in a similar position.
[ back ] 221. For ἤδη occuring in discourse acts that report people’s thoughts in Herodotus, see Herodotus 1.137.2 (stance of the Persians); 6.108.1 (stance of the Plataeans); 8.100.2 (embedded stance of the Athenians in Mardonius’ speech).
[ back ] 222. Thucydides 2.51.6; 2.52.4; 2.53.3; 2.53.4.
[ back ] 223. ὅτι δέ is a rare combination that recalls “paratactic” ὥστε. Ιt occurs also in Thucydides 3.104.5; 6.17.3; 6.55.1; 7.15.2. The presence of δέ suggests that a new act starts; no preceding verbs from which ὅτι might depend occur in the previous stretch of discourse. Ιn Stuart and Jones’ edition it consistently appears after a full stop. My translation attempts to capture the value of ὅτι δέ as an act initial discourse marker generally conveying continuation. Herodotus uses this combination 6 times.
[ back ] 224. “Conducive to it” after Rusten’s translation (1989:192).
[ back ] 225. For further ἤδη in discourse acts that seem to convey either the stance of someone in particular, or a blending of the characters’ and the author’s stances, see Thucydides 1.49.7; 4.128.3; 7.23.2; 7.24.3, 7.2.4; 7.43.7 (with μᾶλλον ἤδη); 7.70.8 (including a superlative form); 7.71.1 (including a comparative form); 7.73.1 (including a gnomic act). For an analysis of viewpoint and the theory of Conceptual Blending applied to Homeric passages, see Bonifazi 2018.
[ back ] 226. E.g. in Thucydides 7.84.3 ἤδη marks the blended stances of the Athenians and the author about the Syracusans making the crossing of river Asinarus difficult: καὶ οἱ πολέμιοι ἐπικείμενοι χαλεπὴν ἤδη τὴν διάβασιν ἐποίουν. At 1.57.2 once again the blended stance of the Athenians and the author is in question: οἵ τε γὰρ Κορίνθιοι φανερῶς ἤδη διάφοροι ἦσαν, “for the Corinthians were really patently hostile” (tr. AB). Herodotus uses this strategy as well, albeit not as frequently as Thucydides does. For instance, at Herodotus 9.118.1 ἤδη marks the blended stance of the author and the inhabitants of Sestos in the Chersonese: the latter are under siege, and the evaluation in question is about the desperate conditions of the besieged: Οἱ δὲ ἐν τῷ τείχεϊ ἐς πᾶν ἤδη κακοῦ ἀπιγμένοι ἦσαν, “Inside the town, the besieged were … reduced to the direst extremity” (tr. De Sélincourt and Marincola). The imperfect tense tends to occur in such passages, which may reinforce the sense of internal perception; however, not only imperfects occur: e.g., at 7.55.1 Thucydides’ third person narration reveals, through ἤδη, a blending of his own and the Syracusans’ evaluative stance; note the perfect: Γεγενημένης δὲ τῆς νίκης τοῖς Συρακοσίοις λαμπρᾶς ἤδη καὶ τοῦ ναυτικοῦ, “As the victory for the Syracusans was really brilliant over the fleet as well ….” (tr. AB). I take “brilliant” from Hornblower’s translation (1991-2008, vol. III:648).
[ back ] 227. See IV.2.4.3 on καί at narrative peaks in historiography, and II.3 §58 on δὴ τότε and καὶ τότε to mark narrative peaks in Homer.
[ back ] 228. The fact that I see this phenomenon more often in Thucydides than in Herodotus does not mean that Herodotus does not employ the same discourse strategy. Let us select two particularly vivid moments in the description of past events including ἤδη in Herodotus. At 1.85.4 Croesus’ dumb son, terrified by the imminent killing of his father, hic et nunc starts speaking: μετὰ δὲ τοῦτο ἤδη ἐφώνεε τὸν πάντα χρόνον τῆς ζόης (commonly translated as “from that moment on”). At 7.219.1, in the Thermopylae account, Herodotus conveys the hic et nunc of the moment in which the Greeks realize that the Persians are near: “the third sign was the look-out men running from the hills ἤδη as day was breaking” (τρίτοι δὲ οἱ ἡμεροσκόποι καταδραμόντες ἀπὸ τῶν ἄκρων ἤδη διαφαινούσης ἡμέρης, tr. AB). Τhe peculiar position of ἤδη between two participial phrases suggests a value that comprises both events.
[ back ] 229. Readers may wonder why the narrator/historiographer would not prefer to use νῦν instead of ἤδη. Three reasons can be adduced. First, accented νῦν tends to modify propositional content (think of “I’ll write a message to the administration” vs. “I’ll write a message to the administration now”). Second, ἤδη is richer than νῦν in meaning; it can convey stancetaking about states of affairs independently of time considerations (think of ἤδη κατὰ κράτος). Third, and more important, νῦν is a deictic marker; the moment referred to has to be inferred contextually, as for ἤδη, but this contextual dependency is explicit and straightforward. On the other hand, ἤδη retains the effect of immediacy, because it is not a deictic marker, and it concerns not only time. Most of all, with ἤδη the blending of other voices and stances remains implied without being imposed; ἤδη gives more flexibility in interpretation, and enhances polyphony. I do acknowledge that a very few times ἤδη in Thucydides works as propositional “now” (see §161); however, this is a variation in function that covers a minority of cases, and does not cancel the validity of the general point. ἤδη νῦν co-occur contiguously only in Herodotus 6.50.3 (at the beginning of direct speech).
[ back ] 230. Thucydides features these constructions at least 64 out of 373 times.
[ back ] 231. Among many analogous instances in Thucydides, I mention a passage where the hic et nunc suggested by ἤδη in a participial phrase refers to the epistemic stance of Demosthenes and Eurymedon: ὁ δὲ Δημοσθένης καὶ Εὐρυμέδων, ἑτοίμης ἤδη τῆς στρατιᾶς οὔσης ἔκ τε τῆς Κερκύρας καὶ ἀπὸ τῆς ἠπείρου, ἐπεραιώθησαν ξυμπάσῃ τῇ στρατιᾷ τὸν Ἰόνιον ἐπ’ ἄκραν Ἰαπυγίαν “Meanwhile Demosthenes and Eurymedon, their recruitment from Corcyra and the mainland now complete, took their whole armament across the Ionian gulf to the promontory of Iapygia,” 7.33.3 (tr. Hammond).
[ back ] 232. Compare the discussion on ἤδη vs. νῦν in n229.
[ back ] 233. I read a propositional “now” behind ἤδη twice in direct speech: παυόμενοι λέγομεν ἤδη ὅτι οὐ Θηβαίοις παρέδομεν τὴν πόλιν (Thucydides 3.59.3); and μάθετε ἤδη (Τhucydides 6.90.2 and 6.91.2). At 6.25.1 one of the Athenians asks Nicias to make a declaration about the contingent at vote: ἀλλ’ ἐναντίον ἁπάντων ἤδη λέγειν ἥντινα αὐτῷ παρασκευὴν Ἀθηναῖοι ψηφίσωνται “but he should now declare in front of them all what forces he wanted the Athenians to vote him” (tr. Hammond). It would make sense to relate ἤδη to ἁπάντων, rather than to take it as “now” modifying the verb λέγειν. In Nicias’ reply to this very request, conversely, a further ἤδη possibly refers to the hic et nunc of the situation in a propositional function (note the verb δοκεῖν, however evaluative): ὁ δὲ ἄκων μὲν εἶπεν ὅτι καὶ μετὰ τῶν ξυναρχόντων καθ’ ἡσυχίαν μᾶλλον βουλεύσοιτο, ὅσα μέντοι ἤδη δοκεῖν αὐτῷ [τριήρεσι μὲν…] “Nicias was reluctant to reply, saying that he would prefer to have time to discuss the matter with his fellow commanders: but as far as he could see at present …,” 6.25.2 (tr. Hammond).
[ back ] 234. See I.3 §§11-12 and 54; IV.2 §3 on the continuum between propositional and non-propositional meanings of particles.
[ back ] 235. See II.4.3.2 and II.5.3.3.
[ back ] 236. A representative account in this direction is that of Wakker 1997b, which focuses on Herodotus and Thucydides. In a previous work on conditional clauses, Wakker devotes attention to the constructions εἰ/ἤν ἄρα (1994:343-350).
[ back ] 237. See e.g. Denniston 1950:35; Wakker 1994:345 (“signaling surprise” is taken as a characteristic use of ἄρα); Powell 1938:44 “w. reported sp., and acc. inf., implying surprise.”
[ back ] 238. See, e.g. Hartung 1832:430; once again, Denniston 1950:32-33; Wakker 1997b:231.
[ back ] 239. See Denniston 1950:35 (“ἄρα expressing the surprise attendant upon disillusionment”) and 38.
[ back ] 240. 34 out of 40 instances feature εἰ or ἤν + ἄρα. This use appears in Herodotus 14 times, and it represents the only functional overlap across the two authors.
[ back ] 241. Herodotus 1.24.7; 1.27.1; 1.86.3; 1.111.5; 2.140.1; 3.134.1; 8.59.1; 8.94.2; 9.8.1; 9.19.3; 9.33.1; 9.86.1; 9.100.1. The same function applies to δ’ ἄρα (without ὡς) at 1.111.1 and 2.141.3. The idiomatic construction εἰ μὴ ἄρα (“except”) is considered separately (see Herodotus 4.32.1; 7.10.θ3).
[ back ] 242. See II.3 §66 and II.5 §§51-62 on ἄρα as a marker of frame recall. Further instances in Herodotus occur at 9.19.3 (“once they arrived” at a place, the move being anticipated by the information “they continued their advance”); 9.8.1 (“after hearing that” following reported speech); 9.100.1 (“as they completed the preparations,” after describing the preparations); 8.94.2 (“as they were fleeing” after mentioning a retreat); 9.33.1 (“when Mardonius’ dispositions were complete,” following the description of the dispositions); 8.59.1 (“as they gathered,” after mentioning the decision to summon a conference). In 2.141.3 δ’ ἄρα accompanies “while lamenting,” after the mention of someone lamenting. At 8.135.1 ἄρα alone accompanies the narration of an action involving Mys, someone who had been introduced in a previous chapter (133). In a very few cases of εἰ/ἤν ἄρα clauses in Thucydides, ἄρα performs a similar function—overlapping with the marking of stance. For example, at 2.87.1 (εἴ τις ἄρα δι’ αὐτὴν ὑμῶν φοβεῖται τὴν μέλλουσαν) ἄρα secures the connection to previous content (see φοβούμενοι, 2.86.5, and φοβουμένους, 2.86.6, both referring to the Peloponnesians).
[ back ] 243. E.g. at 8.59.1 people gather, and immediately afterwards Themistocles begins an unexpected and long speech; at 8.94.2 sailors flee, and immediately afterwards meet a mysterious ship; at 9.8.1 the ephors hear the message of the Athenians, and immediately afterwards decide to postpone their plans; at 8.135.1 the action involving Mys is the beginning of a real θῶμα (see 135.1 θῶμα μέγιστον); at 2.141.3 the person who complains in front of a god’s shrine falls asleep all of a sudden, and has a dream.
[ back ] 244. See IV.3 §108; IV.5 §16, §19, §27, §39, §61, §68, §77, §§94-95.
[ back ] 245. On "unframed discourse" see II.4.2, and also above §27.
[ back ] 246. See Herodotus 3.14.8; 3.70.1; 4.45.4; 4.64.3; 4.189.1; 4.205.1; 7.116.1; 7.152.3 (all of the ἄρα acts displaying tense discontinuity with both the preceding and the following act); 2.28.5; 2.58.1; 3.34.4; 8.8.1.
[ back ] 247. I connect Herodotus’ attention to the external conditions of reception to his use of ἄρα near verba dicendi/sentiendi, or expressions retrieving ad hoc λόγοι. In these passages ἄρα seems to indicate adherence to what was said or heard, which relate to occasional accounts external to those of Herodotus. This use well complements the idea that in Homer ἄρα metalinguistically indicates resonance with the discourse memory of epic (II.4.1). The Herodotean passages include 1.141.2 εἰπεῖν ἄρα αὐτὸν πρὸς τοὺς ἰχθῦς (the fable includes the words we report, which are uttered by a flute player addressing some fish; the infinitive depends on a verbum dicendi introducing a fable told to the Ionian messengers by Cyrus: ἔλεξέ σφι λόγον, 1.141.1). 2.28.5 εἰ ἄρα ταῦτα γενόμενα ἔλεγε; 3.34.3 Νῦν ἄρα μέ φασι Πέρσαι; 3.64.4 τὸ δὲ χρηστήριον … ἐν Συρίῃ Ἀγβατάνοισι ἔλεγε ἄρα; 4.134.1 εἶπε ἄρα πρὸς τούς περ ἐώθεε; 7.17.1 εἶπε ἄρα τάδε; 8.111.2 ὡς κατὰ λόγον ἦσαν ἄρα αἱ Ἀθῆναι μεγάλαι τε καὶ εὐδαίμονες; 9.9.1 ἔλεγε ἄρα σφι τάδε; 7.152.3 ἐπεὶ καὶ ταῦτα λέγεται, ὡς ἄρα Ἀργεῖοι ἦσαν; 8.136.2 λεών τε πολλὸν ἄρα ἀκούων εἶναι καὶ ἄλκιμον; 9.8.1 Ὡς δὲ ἄρα ἤκουσαν οἱ ἔφοροι ταῦτα. Furthermore, at 1.111.5 the herdsman Mitridates reports ὡς ἄρα Μανδάνης τε εἴη παῖς (“that the baby was son of Mandanes”) from the account he just heard from a servant (πυνθάνομαι τὸν πάντα λόγον, 111.5). This use of ἄρα finds confirmation in Van den Besselaar’s lexicographical study of ἄρα (1962:257-258): the scholar finds that one of ἄρα’s functions in post-Homeric prose is to introduce others’ words (“em citações,” “in quotations”), sometimes to stress their unacceptability, but sometimes with apparently neutral intentions.
[ back ] 248. In light of that I do not agree with Mulligan (2007), who considers the three instances of ἄρα in clauses that introduce direct speech (Herodotus 1.141.2; 4.134.1; 9.9.2), and connects the use of the particle to the character of the speech immediately following (“all three speeches … are commanding, prompted by vivid evidence, and have an immediate effect on their audience,” 2007:283).
[ back ] 249. Thucydides 3.82.7; 3.84.3; 6.24.4; 6.60.2.
[ back ] 250. For Thucydides the overall division is: 24 ἄρα in direct speech, 3 in indirect speech, 9 in indirect thought (36 out of 40 total instances); for Herodotus, we find 18 ἄρα in direct speech, 2 indirect speech, 5 indirect thought (25 out of 66 total instances).
[ back ] 251. Denniston (1950:37) approaches my reading: “ἄρα in a conditional protasis denotes that the hypothesis is one of which the possibility has only just been realized: ‘if, after all’.”
[ back ] 252. Van den Besselaar (1962:262) refers to εἰ/ἤν ἄρα hypotheses that are considered an “extreme case” (“caso-limite”). εἰ/ἤν ἄρα acts introduce outcomes that look unfortunate in the context of someone’s speech or thought in Herodotus 3.45.4; 5.106.4; 7.9γ; 7.10θ.3; 7.149.1;; Thucydides 1.70.7; 1.84.2; 1.93.7; 1.140.1; 2.5.1; 3.56.5; 6.75.1.
[ back ] 253. A notable exception is Lang 1995, focusing on participles related to these verbs in Thucydides; however, Lang’s aim is to show that historiography uses these participles to draw attention to the importance of characters’ motivations; she does not interest herself in the linguistics of what comes after such participles.
[ back ] 254. Relevant insights come especially from the multi-contribution volumes Schneider and Hartner 2012 (on blending and the study of narrative), and Dancygier and Sweetser eds. 2013 (on viewpoint in language).