IV.5 Analysis of four excerpts

§1. This chapter combines and exemplifies the main points made in IV.2, IV.3, and IV.4. Particles are observed in situ over relatively long passages, and their use is seen in relation to discourse segmentation, and-coordination strategies, and traces of peoples’ voice and stance, all at once. The aim is to demonstrate how substantially the process of reading and comprehending the text changes if the aspects mentioned above are added to a more traditional analysis of sub- and main clauses.

§2. The four selected excerpts are representative of the major discourse settings of the two Histories, namely narrative sections and speeches. For the sake of comparison, they share some basic thematic affinities (two of them are accounts of reactions to major military enterprises, and two are “wise-advisor” speeches). Each of the four excerpts is thirty OCT lines long.

§3. The texts display modern punctuation as given in Stuart Jones’ edition (OCT and TLG online) for Thucydides, and Hude’s edition (OCT) for Herodotus (including extra spaces and indentation in square brackets), as well as the medieval punctuation from two important manuscripts (in italics within curved brackets). [1] Notes on philological aspects such as variants occur only when relevant to punctuation or to particles. Analogously, paleographical matters are not discussed; the sole purpose of including a sample of medieval punctuation is to remind us of one way in which pre-print texts were used over many centuries.

§4. The segmentation that I propose is in acts, small communicative steps of the discourse, and moves, multi-act units sharing an overarching communicative goal. [2] In the layout each line corresponds to an act, and each line space corresponds to a move boundary.

§5. My readings are not offered as objective descriptions, let alone as reconstructions of the original segmentation that Herodotus and Thucydides had in mind. Rather, they are interpretations of how discourse might be segmented if, on the one hand, we pay attention to several linguistic features of the language besides hypotactic segmentation, and, on the other hand, we take into account the ancient notions of kôla and kómmata, as well as the contemporary notion of the intonation unit.

§6. The upcoming analyses ultimately illustrate the holistic approach to discourse that we present throughout the work. [3] Boundaries are inferred on the basis of postpositive particles, participial constructions, parallel constructions, hyperbata, and negations, besides independent clauses and subclauses. As for particles, keys to segmentation are not only postpositive δέ and τε, but also καί, appearing in different constructions. Moreover, the occurrence of particles possibly evoking someone’s voice and stancetaking (such as γε, δή, and ἤδη) help to single out parts of third-person narration that incorporate reported thoughts besides indirect speech. A further important key to segmentation is the lexical repetition of terms, which creates coherence within and across moves. Finally, verbs seem to frequently occupy act-final or move-final positions; even though this last remark calls for future research, the result per se is relevant to the purposes of the chapter.

§7. Comments in the next sections are made move by move (including a literal translation of my own), but the continuous texts are reported in the Appendix. General results and comparative data are all gathered in the conclusions.

§8. The first excerpt comes from the second speech of the Athenian general Nicias (6.20-23). [4] Nicias delivers this speech in Athens during the assembly debate over the proposal to invade Sicily (6.1-26). As the scholarly literature underscores, [5] Nicias embodies an attitude of stark caution and doubt towards the undertaking. He claims that the Athenians must possess a great contingent as well as military superiority in order to handle the high risks. Prior to the selected excerpt Nicias has explicitly suggested that the Athenians let the Sicilians live and settle their own affairs (6.13.1). He has also revealed his thoughts on the might of the Sicilian cities, their autonomy in food provision, and the considerable size of their fleets and cavalry (6.20). In light of these factors—Nicias warns his fellow citizens—careful plans should include a large force of infantry, and especially adequate equipment and supplies (6.21). In the present excerpt he substantiates and details these points. The final statements summarize his own view: Athens needs good planning and good luck, and therefore the best security is to maximize armament and minimize risks.

§9. In an attempt to define stylistic characterization in Thucydides, Tompkins (1972) analyzes Nicias’ and Alcibiades’ speeches as opposites both in content and style. While Alcibiades uses parataxis and presents his thoughts straightforwardly (especially by means of sentence-initial καί, 205), Nicias expresses possible complications in action, Tompkins believes, by means of syntactic complication, which results in a heavy use of hypotaxis and a preference for concessions. In a section discussing Nicias’ “sentence complication” (184-204), Tompkins singles out semantic features that point to Nicias’ conservative stance against adventurous enterprises: “concessions and reversals” (185) marked by the use of “still, yet, however, but”; several levels (or degrees) of subordination (185-188); impersonal verbs and abstract nouns (189-193). These features characterize, in Tompkins’ view, Nicias’ tendency to express concerns and reluctance.

§10. The analysis of particles, their immediate co-text, semantic input, syntactic contructions, and lexical repetition makes me divide the speech excerpt into 10 distinct moves. The communicative goals underlying them are the following: Nicias remarks that infantries are needed (acts 1-12); ships are needed as well, to carry supplies (13-20). He reminds his fellow citizens that cities cannot be in charge of (the supplies for) the army (21-22); total independence is needed as far as supplies are concerned (23-24). He warns about the necessity of money (25-28). He argues that military excellence in all departments is needed (29-35). Then he points out that founding new colonies requires exerting power from the very beginning (36-43). Even with fear and uncertainties, Nicias declares that he wants to set sail with an appropriate armament (44-52). He adds that only so there is success for the city and the army (53-55). Finally, he says he is willing to give up commanding in case someone else disagrees on these points (56-57).

§11. In the first discourse move (acts 1-12) Nicias details the requirements mentioned in the previous part of the speech (6.21). He starts with the infantries necessary for facing the enemies. οὖν marks the beginning of the move. [6] Together with οὖν, two more elements signal the structure of the unfolding argument, ὁπλίτας … πολλούς, and τε. They have separate functions. ὁπλίτας … πολλούς conveys the subject of the upcoming discourse acts. The adjective πολλούς adumbrates Nicias’ rhetorical insistence on the large amount of resources the Athenians must provide. τε attached to the noun phrase projects a second conjunct introduced by καί—see act 10, καὶ τοξότας πολλούς, a symmetrical noun phrase including another accusative plural and the repetition of πολλούς. [7] In line with what we suggest about different genres, ὁπλίτας τε οὖν πολλούς is a priming act: it orients the attention of the audience to a discourse topic that projects at least beyond the following act. [8]

§12. μοι δοκεῖ in 2 is a crucial construction. I take it as an independent act surrounded by elements of a subclause depending on it. Most of all, μοι δοκεῖ projects not only the stancetaking character of the considerations that follow, [9] up to the mention of an alternative stancetaking (εἰ δέ τῳ ἄλλως δοκεῖ, act 56), but also a long series of infinitive constructions [10] that enhance cohesion among multiple moves. In fact, δοκεῖ’s status as the only finite verb in an independent clause before νομίσατε, 28, is what persuaded editors to put the next full stop only before the sentence including νομίσατε.

§13. Apart from the central role of μοι δοκεῖ, the main component that shapes this specific move is the naming of the individual contingents (ὁπλίτας τε, 1; καὶ τοξότας, 10; καὶ σφενδονήτας, 11). Further articulation within this list is marked by parallel constructions introduced by καί (“from our own and from the allies,” 4-5), by τε followed by καί (“[and within the allies] from our subject allies and from whomever…, 6-7 [11] ), and, finally, by ἤ … ἤ (“…we may either convince or induce by payment,” 8-9). A further component wraps the move up, the antagonistic connection established between “us” (ἡμᾶς, 3; ἡμῶν αὐτῶν, 4; δυνώμεθα, 7) and “those” (ἐκείνων, 12). Act 12 closes the first part of the argument by articulating the expected objective, that is, to oppose the enemy cavalry.

§14. The next move (13 to 20) conveys Nicias’ speculation that ships will make it easier to transport supplies. Acts 13 and 14 connect infantry to naval army, and soldiers to supplies. The particle τε in ναυσί τε, 13, echoes ὁπλίτας τε, 1, and starts a new general entry (“we need ships as well”). [12] This is an example of a mismatch between modern section division and move division: while the OCT text goes on with section 1, and it does not punctuate with any full stop, ναυσί τε, 13, establishes the new discourse topic. Indeed, τε in Herodotus and in Thucydides often signals the beginning of a new move. [13]

§15. Some grammatical elements of this second move resonate with elements of the first. For example, the subordinating conjunction ἵνα, 14, echoes the necessary anticipation of plans in 12 (with ὅπως and the subjunctive), and the first person plural verb ἐσκομιζώμεθα, 14, resonates with δυνώμεθα, 7. In 14 the theme “supplies” is introduced (τὰ ἐπιτήδεια), which forms the subject matter of the following acts (15 to 20). Later, ἵνα and a first person plural subjunctive are repeated (see 18 and 19 [14] ); furthermore, the close of the move is sealed by the repetition of ἐπιτήδεια, 20. All these lexical repetitions help stressing the thematic importance of the elements referred to, and at the same time have the organizational function of framing the piece of discourse devoted to them.

§16. The next move consists of just two acts. It is a γάρ move that illustrates the necessity for autonomy on the part of the military forces. Syntactically, this is an independent period, and it forms a complete thought. [15] Because it is followed by a further infinitive construction dependent on μοι δοκεῖ (i.e. ἑτοιμάσασθαι, 23), the unit works as a discrete whole embedded within the larger discourse. Its syntactic autonomy is what presumably led editors of printed editions to encapsulate the thought in a parenthesis. The eleventh-century-manuscript that I checked (British Museum 11727) has no parenthesis, but only a high dot and a middle dot framing the unit. The dots do not imply the interpretation of the move as secondary to the main argumentative line—an effect that parentheses do have. In a discourse perspective, the particle γάρ together with the shift of tense in ἔσται, 22 (the previous finite verb is ἔχῃ, 20), are more precise than a dot, and less loaded than a parenthesis: they indicate that the upcoming piece has to be processed as a thought that is associated but also discontinuous with respect to the preceding discourse. [16] The discontinuity in tense expressed by ἔσται, 22, foreshadows οἷοί τε ἐσόμεθα, 33, and Nicias’ further remarks about the future decisions to take.

§17. The next move (23-24) concludes the topic of the supplies by reinforcing the point about autonomy. The subject matter here is “the remaining supplies” to be prepared in addition to those already mentioned. τε in τά τε ἄλλα, 23, resumes the discourse about τὰ ἐπιτήδεια, 20, after the γάρ move. Its pragmatic function here resembles that of τε in ναυσί τε, 13; in both cases it marks a link between parallel constituents that are separated by different linguistic material (see IV.2 §83). καί in 24 does not simply answer τε; rather, while marking an addition it corrects the immediately preceding utterance by putting it in more precise terms: καὶ μὴ ἐπὶ ἑτέροις γίγνεσθαι means, “that is to say, (it seems to me that supplies should) not be provided thanks to others.” [17]

§18. The subject matter of the next move is χρήματα, “money.” The δέ act in 25 qualifies it as a new, separate act. A lexical item occurs twice, at the beginning and at the end of the move, μάλιστα (25 and 28), albeit in different act positions and with different scope. [18] This move is divided into two parts. The first (25), introduced by δέ, makes the point about financial autonomy. The second part also starts with δέ, and it expands on the same topic (see τά, 26, retrieving χρήματα). Overall it works as a warning, introduced as if almost incidentally, about the alleged reality of the Egestan contribution (later revealed in chapter 46 of book 6). In light of this reading, the full stop posited by editors after ἔχειν, 25, is questionable. On the local level it would make sense, if we take it to mark the boundary that δέ indicates. On the global level, however, it happens to be an exceptional full stop after a considerable amount of discourse, and therefore it represents a major pause—the immediately preceding full stop in the OCT edition is to be found at the end of 21.2, that is, before act 1. This segmentation based on hypotaxis obscures the coeherence and the thematic connection between the two thoughts about money. [19]

§19. The next move starts with γάρ; the overarching communicative goal is to illustrate and expand on the crucial point of the capabilities of Athenian resources alone. The move coincides with the next períodos, and it corresponds to chapter section 23.1. However, the individual discourse acts (29 to 35) do not coincide with syntactic clauses, and their ranking is not to be considered in terms of “main” vs. “subordinate.” Here I have identified acts based on the occurrence of particles, negations, and a οὗτος form. [21]

§20. The result is a series of segments whose individual syntactic shapes vary significantly. The first act establishes the subject matter: “if we depart from here with our own force”; the subordinate clause ἢν … ἔλθωμεν forming it, strategically sets up the hypothetical scenario Nicias intends to comment on. Act 32 answers act 30 in construction and syntactic parallelism (μὴ … παρασκευασάμενοι, 30; ἀλλὰ … ὑπερβάλλοντες, 32). Both are participial phrases; they frame act 31, where the particle γε gives emphasis to the encapsulated act. Let us focus on the discourse function that the encapsulated act performs. πλήν γε πρὸς τὸ μάχιμον αὐτῶν, τὸ ὁπλιτικόν led Dover (1954-1955:8) to imagine “a tone of warning and an admonitory gesture” accompanying πλήν γε, and “vehement emphasis and a change from admonition to a more menacing tone” accompanying τὸ μάχιμον. He paraphrases the content of segments 29-32 as follows: “if we go to Sicily after ourselves raising at Athens a force not simply (as has been proposed) a match for the Siceliots (as it is called by the proposers)—only, mind you, it isn’t a match for that element in their forces which really does the fighting, their hoplites—but (as I would propose) superior in all arms, …” From this reading I infer that the speaker, while uttering πλήν, τὸ μάχιμον, and τὸ ὁπλιτικόν implies meanings that are hinted at, and left unsaid. γε is a most apt particle to do that. [22] In this case γε has scope over the entire noun phrase (πλήν γε πρὸς τὸ μάχιμον αὐτῶν, τὸ ὁπλιτικόν, 32). [23] Perhaps Nicias intends to implicitly contrast his opinion about the necessity of a superior force in all respects with Alcibiades’ confidence that the naval superiority of Athens is enough to grant success. [24] The effect on the pragmatic level is the stress on the act including γε. Dover concludes his analysis by saying: “Just as we have not really read a Greek orator’s speech until we have heard in our imagination the sound of the speaker’s voice and seen his gesticulation, so a Thucydidean speech requires us to recreate in our own minds as much as possible of the visual and auditory images that were in the author’s mind when he wrote it” (1954-1955:8). It is significant that Dover associates this comment with the occurrence of πλήν γε; our joint analyses in this monograph show, in concurrence with Dover’s observation, that γε is frequently connected to the speaker’s voice and a sense of liveliness. [25]

§21. The discourse act in 33 is characterized by the “wrapping up” use of the demonstrative adverb οὕτως, [26] by the negatively-oriented adverb μόλις, and by the future tense of ἐσόμεθα: “We will scarcely in this way be able to …”. The condition expressed at 29-32 is hard to fulfill; but even if it is fulfilled, it will result (that is Nicias’ prediction behind the future tense) in goals difficult to achieve: to rule what Athens wishes to conquer, and to save what Athens wishes to save. In the manuscript a low dot or comma (it is difficult to tell which sign it is precisely) occurs after both τῶν μέν, 34 and τὰ δέ, 35. These are paralinguistic signs that suggest a performative pause. Such a pause would enhance the sense of the twofold difficulty that Athens faces, which Nicias goes on to elaborate in the two acts formed by the two infinitive constructions.

§22. A future tense also closes the next move, made up of acts 36 to 43 (ἕξουσιν, 43), which again coincide with a períodos and a chapter section. Rhetorically this move represents the climax and final point of the argument; τε is the particle that introduces it (analogous τε occur in 13 and 23). Lexically and syntactically the move illustrate Nicias’ stylistic preferences: hypotactic “complications” (see the relative-conditional clause ᾗ ἂν κατάσχωσιν, 39, and the hypothetical clause ἢν σφάλλωνται, 42); and references to cautious pondering (νομίσαι, 36; εἰδέναι, 41). On the one hand he refrains from being direct (see the impersonal verbs χρή, 36, and πρέπει, 38, and references to abstract individuals—“they” instead of “those of us”). On the other hand, he depicts potentialities as certainties, dangers as actual facts (“founding a city among strangers, even enemies,” 37; “if they fail, they will gain an environment utterly hostile,” 42-43). Overall, the move conjures up the risky situation of conquering a potentially hostile territory.

§23. The next move focuses on Nicias’ own deliberation, which opens the very last part of his speech (6.23.3). In ring-composition fashion, the move is framed by first person markers (here we see ἐγώ, 44, and ἐμαυτὸν βούλομαι, 51), recalling μοι δοκεῖ, 2. [27] As we saw with previous stretches of discourse, the force of each act within a move is obscured if we aim at finding periodic perfection—that is, if we read impatiently, so to speak, through the text from the beginning of 23.1 by assuming that the information given is secondary or subsidiary, until we reach the main clause (ἐμαυτὸν βούλομαι ἐκπλεῖν, 51). To read in this manner would be to miss the strategic sequence of the acts that precede the main clause, as well as the pragmatic force conveyed.

§24. The concessive flavor of ὅπερ (44), combined with the possibly concessive nuance of the participles φοβούμενος, 44, and εἰδώς, 45, reflects Nicias’ cautious internal state: he fears, he is aware, and yet he decides to sail. A further element of hesitation concerns his confidence in τύχη: he knows that more than good counsel, they need good luck (ἔτι δὲ πλείω εὐτυχῆσαι, 47). At the same time, he does not want to be exposed to τύχη (ὅτι ἐλάχιστα τῇ τύχῃ παραδοὺς, 50). [28] What makes the two utterances logically harmonious instead of contradictory is the idea expressed through acts 48 and 49 (χαλεπὸν δὲ | ἀνθρώπους ὄντας), which I am going to focus on next.

§25. Modern editors put χαλεπὸν δὲ ἀνθρώπους ὄντας in parentheses, which leave the impression that the thought is incidental or superfluous. Hornblower (1991-2008, vol.III:359) on the contrary rightly argues that the content of this parenthesis is actually an important argumentative step: “The ‘parenthesis’ is nothing of the sort, but an organic element in a characteristically paratactic and jerky sequence: ‘we shall need the help of fortune; but that is something hard for mortals to be sure of; so I would rather, after all, trust as little as possible to fortune’. That is to say, there is an understood accusative absolute ὄν, thus: χαλεπὸν δὲ ὂν ἀνθρώπους ὄντας.” The first of the two acts is a syntactically elliptical statement commenting on 47: “we need good luck even more—that’s difficult, though.” The second act is also a grammatically elliptical segment: “as we are human beings.” Parentheses are paralinguistic additions that, while clarifying syntactic embeddedness, constitute an interpretive filter as well, in that they imply the communication of secondary information. In this case it is essential to follow the individual discourse steps as Thucydides conjoins them, as Hornblower suggests. In the Greek, the particle δέ at 48 suffices to signal the beginning of a discrete discourse act; so, no parentheses are necessary. [29]

§26. ἐκπλεῦσαι at the end of 52 has puzzled Poppo and Stahl (1875-1889, Vol. III, Sect. 1:61), who hypothesize that it is a spurious addition (mainly because of the lexical repetition ἐκπλεῖν, 51, and ἐκπλεῦσαι, 52), and put ἐκπλεῦσαι in square brackets. The odd variation in tense between the two verbs also puzzled Dover (1965:34): “The semantics of verbal aspect in Greek prose literature have not yet been investigated with sufficient care to provide an objective explanation of the change of aspect here.” I read the act containing ἐκπλεῦσαι as different in force from the preceding act, and I take the variation of tense in ἐκπλεῦσαι as one of the signs. Even though in syntactic terms the two infinitives are equally dependent on βούλομαι, the first infinitive clause (with ἐκπλεῖν) may be interpreted as a declaration (“I decide that I set sail”), while the second clause (with ἐκπλεῦσαι), act 52, may be interpreted as an exhortation or strong wish (“I wish that we set sail with an appropriate armament”). Several elements are in favor of this reading. The juxtaposition ἐμαυτόν - παρασκευῇ marks the distinction between the personal participation in setting sail, and the decisive element on which a successful participation depends, namely the armament. [30] The particle δέ suggests a different act altogether, rather than a link between two equal elements (given by the content of the two infinitives). The lexical repetition of the same verb ultimately emphasizes the difference in aspect, which contributes to the idea of acts with different illocutionary forces. Finally, a very similar wording right after the end of Nicias’ speech matches the interpretation of the general’s intention, however resolute, to set sail safely, rather than an effective decision taken on behalf of all:  ὁ μὲν Νικίας τοσαῦτα εἶπε νομίζων τοὺς Ἀθηναίους τῷ πλήθει τῶν πραγμάτων ἢ ἀποτρέψειν ἤ, εἰ ἀναγκάζοιτο στρατεύεσθαι, μάλιστ’ <ἂν> οὕτως ἀσφαλῶς ἐκπλεῦσαι (“Nicias said such things thinking that either he would dissuade the Athenians with the argument of the magnitude of matters, or, if he was forced to do the expedition, they would set sail in this absolutely safe way,” 6.24.1 [tr. AB]). [31]

§27. For the fifth time in this excerpt (after 13, 21, 23, and 25), particles and other co-occurring features make me see a new move start that does not coincide with modern section division: Acts 53-55 represent a recap move signalled by an “epimythic” οὗτος [32] form and the particle γάρ. The referent of ταῦτα must be all the suggestions that Nicias expressed during the speech he is about to close. The function of γάρ is to insert the final general assessment (see ἡγοῦμαι) by the speaker. [33] τε in 54 and καί in 55 mark the two parallel components of the assessment. The act in 54 specifies the benefits for the city; the act in 55 the benefits for the soldiers.

§28. The final acts mention the possibility of another opinion and of another command. In this case the extra space preceding εἰ δέ in the OCT edition matches my pragmatic reading of 56 and 57 as a separate move. τῳ … δοκεῖ is the key feature, as it thematically juxtaposes the opinion of someone else to Nicias’—see μοι δοκεῖ at 2. The move boundary is even greater if we consider that 56 and 57 constitute the counterpart of a multi-move unit starting with μοι δοκεῖ. Ring-composition by means of lexical repetition is also at work here.

§29. All comparative results are gathered in 5.6, at the end of the four analyses. Here I just conclude by relating this analysis on the whole to Tompkins’ idea that Nicias is a master of “sentence complication.” The suggested articulation of the text in moves and acts shows that Nicias as a speaker is capable of expressing his thoughts by using a variety of means in addition to syntactic complexity. Each small step seems shaped in such a way to facilitate comprehension rather than to complicate it.

§30. A priori Hellenists may be inclined to think that, however structurally complex Thucydidean speeches are in general, speakers nonetheless tend to present information in rather small segments. Therefore an act division such as the one presented in the previous section may not surprise, after all – it would simply reflect the character of spoken discourse. Now, let us see if the size of acts and moves in 8.1, a narrative section, is different with respect to 6.22-23. Book 8 starts with the reactions of the Athenians to the news of the devastating defeat in Sicily. Just as many other chapters, chapter 1 weaves together multifarious facts, feelings, decisions, and courses of action with the kind of exemplary elegance that has caused modern readers to observe in Thucydidean style the sharpness of the author’s mind. [34]

§31. I divide the chapter into 12 moves. The overarching communicative goals I infer, in brief, are the following: the historian stresses the disbelief of the Athenians, as the news spreads (acts 1-6). Then he focuses on the Athenians grudging against those who voted for the expedition (7-9). He enacts a close-up on the Athenians’ anger towards seers (10-13). Then he comes to the highest point of the description of feelings: distress and consternation are pervasive (14-16). Afterwards he expands on the deep sense of deprival (17-22), and on the lack of hope for survival (23-26). The historian moves on to detailing the fears of threats and attacks (27-35). At that point his access to the mind of the Athenians shifts to the “nevertheless” components of their thoughts: it is necessary to act (36-44); thrift and measures have to be managed wisely (45-49). Thucydides concludes that the Athenians are ready to act (50-52), and immediately (53-55). At that point he wraps up the account of the entire summer (56).

§32. By means of the first move, Thucydides reports the announcement of the defeat, and he depicts the Athenians’ disbelief of the surviving soldiers even before mentioning their disbelief of the messengers. In order to illustrate the importance of act 1, we need to consider the closure of book 7. Hornblower (1991-2008, vol. III:749) remarks that what straddles books 7 and 8 is a μέν … δέ construction. [35] The text is: ταῦτα μὲν τὰ περὶ Σικελίαν γενόμενα. [OCT 8.1] ἐς δὲ τὰς Ἀθήνας … “These were the accounts about Sicily. In Athens…” Τhe combination of μέν and a οὗτος form within such co-texts indicates that a section is rounded off but more is to come—which in several modern languages can be rendered by means of a nonfinal intonation contour. The δέ act that follows suggests a major shift in subject and setting, from the expedition in Sicily to the reactions and consequences in Athens. It is a priming act: it is short and lacks predicates, but it nevertheless performs a crucial cognitive and pragmatic re-orientation. Now the historian moves to Athens, a post-Sicily Athens. The act primes the geographical/physical setting of the upcoming account.

§33. The move goes on with a short and simple subclause (ἐπειδὴ ἠγγέλθη, 2), which establishes the temporal setting complementing the spatial one. The ἀγγελία, the public and formal announcement of the defeat, works as the (only) relevant chronological reference. In act 3, μέν (following the entire phrase ἐπὶ πολύ) marks the beginning of the account of reactions, and therefore projects several components. [36] The co-occurrence of ἠπίστουν with μέν—a verb form whose subjects (the Athenians in general) are so much in focus that they are not even mentioned [37] —triggers the expectation that some other action (or, better, reaction) by the same subjects will follow.

§34. 4 and 5 are separate acts marked by syntactic parallelism (καί + participle in dative); despite its apparent length, 4 forms just one act because of the bundling effect of the hyperbaton τοῖς … διαπεφευγόσι. γε in act 6 works to retrieve implicit information [38] and signal emotional involvement at the same time. The retrieval, in this case complemented by the anaphoric marker οὕτω, concerns the content of the announcement, which ἐπειδὴ ἠγγέλθη, 2, leaves implicit. The emotional involvement partly resides in the stress given to οὕτω: capital letters for “so” in my translation render the probable prosodic prominence of οὕτω γε (compare the way “SO” in English would signal the speaker’s emotional involvement as well). The presence of γε tells us also that a voice is represented here; the emotional involvement may ambiguously be that of Athenians blending with that of the historian. [39]

§35. The next move focuses on a different emotional state: the initial distrustful attitude gives way to shocked acknowledgment, as the truth comes to be accepted. The OCT extra space before ἐπειδή at 7 does not accompany a section number, and may have been decided on the basis of the editors’ perception of a major boundary in the discourse—or, otherwise, the acknowledgement that medieval scribes put such an extra space in the manuscript. My reading is that a move boundary occurs, indeed, and lexical repetition, once again, comes into play. Just as ἐπειδή in 2 provides the temporal setting for the first move, ἐπειδή in 7 provides the temporal setting for the second move. All the manuscripts, except for B, give ἐπειδή τε instead of ἐπειδὴ δέ, which is interesting. Poppo explicitly says that τε does not make sense here, because it does not answer μέν. [40] However, τε may be fully justifiable, if we think of the Thucydidean usage of τε at discourse transitions. [41] In addition, the sense of contrast that can be inferred from the act (between the moment of the announcement, at 2, and the moment of the actual acknowledgment of the announcement, at 7) does not depend on δέ in itself, but it depends on the semantic co-text (from act 2 on). [42] As for μέν in χαλεποὶ μὲν ἦσαν, 8, it marks the first element of several, in keeping with the function of μέν in 3 and (later) μέν in 17.

§36. The escalation in Thucydides’ report about the emotional arousal goes on with a “close-up” involving the same unspecified Athenians, this time directly angry at vaticinators, seers, and those practicing divination. Modern editors presumably put a comma (instead of a full stop) after ψηφισάμενοι, 9 (see the end of the previous move) because they assume that χαλεποὶ μὲν ἦσαν, 8 and ὠργίζοντο δέ, 10, reflect a thematically related μέν … δέ construction. However, the syntactic symmetry provided by the two verb forms in the indicative, and the semantic affinity (negative feelings towards somebody) is not as important as the asymmetry Thucydides conveys. ὠργίζοντο δέ expresses a climax, and δέ, far from hinting at a contrast, simply introduces a new act and a new move. I note that τε καί in 11 stresses the “common knowledge” connection that links the two professional categories of vaticinators and seers. [43]

§37. Starting from act 6 the historian not only pretends, as he usually does, to have access to the internal states of characters, but he also directly presents in third-person the thoughts of the Athenians. In other words, the account of reactions shifts to an indirect thought modality. The use of αὐτός is revealing (see αὐτοί, 9; αὐτούς, 12; and, later, αὐτούς, 15) [44] : the subjects around whom facts and feelings occur are the Athenians (in direct speech it would be “we ourselves”), but the narration keeps being in the third person. The overarching effect is a blending of voices: the narrator’s and the characters’. [45]

§38. The climactic moment of the description is reached when the historian changes the grammatical subject (from individuals to concepts), and employs comprehensive terms to generalize and deepen the pervasiveness of the Athenians’ internal states. The feelings that predominate among the citizens are distress (see ἐλύπει, 15), fear and consternation (φόβος τε καὶ κατάπληξις, 16). The act in 16 is quite emphatically different and separate from the preceding. It contains μεγίστη δή, a feature that is explored more closely in IV.4 §116 (concerning the two Histories). In a pragmatic perspective, the construction δή + superlative relates to voice and stance. δή signals not only the speaker’s voice but also her communicative actions of stancetaking. Τhe speaker positions herself with respect to that reality (“I feel strongly abοut X”), and at the same time evaluates that reality (“X is really Y”). A fascinating aspect of δή marking someone’s voice and stance in Thucydides is the potential polyphony, namely the potential blending of more voices through the same utterance—which I already advanced for acts 9-12, independently of the use of δή. In act 16 the voice and stance could be the despairing Athenians’ or the historian’s. I advance that Thucydides may deliberately leave the reference ambiguous, possibly to stress—in an indirect form—some sense of sharedness or empathy.

§39. A quite elaborate γάρ move follows, which expands on the emotional atmosphere established in 14-16. γάρ in fact introduces two moves, starting with ἅμα μέν, 17, and ἅμα δέ, 23 respectively (see the space in between, and the indentation in my translation). In Thucydides ἅμα not infrequently forms a construction with μέν or δέ; the construction works as a discourse marker introducing facts or states that co-occur with each other. [46] In this passage strong lexical and semantic parallelisms across the ἅμα μέν and the ἅμα δέ units reinforce this idea of co-occurrence: participles referring to the generic “they,” the Athenians (στερόμενοι, 17; ὁρῶντες, 23), and negative or privative concepts (στερόμενοι, 17; οὐχ, 21; οὐχ, 23; οὐδέ, 24; οὐδ’(έ), 25; ἀνέλπιστοι, 26). In view of this, μέν and δέ serve, together with ἅμα, to index the desolation of a people towards a parallel sense of lack of means, namely at the present and for the future. In these two parallel units the only particles occurring except for μέν and δέ are καί and τε. The repetition of καί gives rise to an accumulation effect [47] : καὶ ἕκαστος, 18; καὶ ἡ πόλις, 18; καὶ ἡλικίας, 20; see also its negative counterpart οὐδὲ (οὐδὲ χρήματα, 24; οὐδ’ ὑπηρεσίας, 25). τε signals a construction of two conjuncts (ὁπλιτῶν τε πολλῶν καὶ ἱππέων, 19), which along with ἡλικίας, 20, grammatically depend on στερόμενοι, 17, and semantically instantiate what “the city” means to the Athenians. Something particularly striking in these moves is the strategic placement and the length of verbal forms. ἐβαρύνοντο, 22 is an act in itself; the relatively high number of syllables forming it iconically mirrors the heaviness of the semantic meaning (“they had a heavy heart”), and possibly of its utterance. Symmetrically, at the end of the ἅμα δέ move, ἀνέλπιστοι ἦσαν, 26, concludes the picture of the hopeless atmosphere in Athens. Act 26 is masterfully filled with verb forms: ἀνέλπιστοι ἦσαν ἐν τῷ παρόντι σωθήσεσθαι. [48]

§40. The modern comma after σωθήσεσθαι does not sufficiently convey the strength of what I see as a move boundary: act 27 (below) begins a new phase of the account of reactions and thoughts about threats and further attacks from enemies, which constitutes a separate discourse unit. The hyperbaton that frames act 27 (τούς … πολεμίους), and the occurrence of τε—which at this point is unsurprising at the beginning of moves—introduce a very different subject, that is, the threatening thought of the enemies, of their possible move directly against the Piraeus. ἄλλως τε καί in 30 signals the climax of this thought: “especially after such a victory.” [49] καί in 31 (καὶ τοὺς αὐτόθεν πολεμίους) answers τε in 27 (τούς τε ἀπὸ τῆς Σικελίας πολεμίους). The two conjuncts, then, span more than one act (27-30, and 31-34); the content refers to the Sicilian enemies and the “local” enemies. [50] In act 35 καί introduces the final “entry” of potentially threatening people, namely the newfound allies of the Greek enemies.

§41. τότε δή in 32 links the situation described (the possession of twice as many resources as those at the enemies’ disposal) to the very moment of the thought of the Athenians about it. [51] I draw a pragmatic connection between δή in 32 and ἤδη in 33 (κατὰ κράτος ἤδη). The two lexical items introduce two acts that express a subjective perception of evidence as well as certainty by the Athenians (see “clearly” and “certainly” in my translation). [52] Once again, Thucydides uses a third-person description of the thoughts of the Athenians (see ἐνόμιζον, 28), but he then manages to blur the distance between his own and the Athenian’s cognition in representing the content of the thoughts. As for κατὰ κράτος ἤδη, 33, I see it as an independent act because its content can be referred to both 32 and 34: by force the enemies would gain double resources; by force they would attack by land and sea.

§42. By means of the next move Thucydides illustrates the resilient attitude of the Athenians. [53] In this act division ὅμως δέ, 36, is separate from ὡς ἐκ …, 37, because the latter belongs to the infinitive subclause. ὅμως δέ is in fact a powerful act in itself: the construction suggests that the argument will take a totally different direction. Because it activates a new frame, this very short act works as a priming act. The new frame is not spatial or temporal, but rather argumentative: it includes the series of decisions that the Athenians take in spite of all their misgivings and feelings of despair that Thucydides has carefully depicted up to now. [54] In content, ὅμως δέ starts a unit that embraces acts 36 to 49, which coincides with a long períodos identified by modern punctuation. However, I sub-divide the move into two: 36-44, and 45-49.

§43. ἐδόκει in 37 projects multiple entries illustrating what seemed to the Athenians to be the right course of action. μὴ ἐνδιδόναι and ἀλλὰ παρασκευάζεσθαι, 38-39, represent a further “abstract,” that is, an anticipated summary of the Athenians’ opinion. The two acts are syntactically parallel and semantically opposite: not to give in, but to get ready. What follows is a list of the three kinds of preparations chosen, plus a specification regarding the third type of preparation, stabilizing relationships with allies. καί links these preparations (καὶ ναυτικόν, 40; καὶ χρήματα, 42; καὶ τὰ τῶν ξυμμάχων ἐς ἀσφάλειαν ποιεῖσθαι, 43); at the same time, the parallel repetition of καί makes them discrete. The last καί, however (44), introduces a different kind of information, namely a detail concerning one of the allies. [55]

§44. Starting at 45 below, the historian narrates about a different set of decisions by the Athenians; they want to take some political measures regarding the city itself. In 50 τε signals a move boundary, in line with several preceding cases (see 1, 13, 23, and 36 in excerpt 1; 27, here, and 50 in this excerpt). The last verb form, προβουλεύσουσιν, coincides with the last act of the move, in line with ἐβαρύνοντο, 22. [57]

§45. The next move (50-52) also starts with τε: This time Thucydides adds a closing and comprehensive (see πάντα, 50) note about the Athenians. He seems to start zooming out from the details of their thoughts. Act 51 is syntactically a subclause, but in content it is the key utterance of the move: “which the δῆμος likes to do.” Poetic flourishes surrounding the clause both help set the statement off rhetorically and underline its solemn character: note, first, the strong alliteration in the two framing utterances (in 50 through the repetition of initial π, in 52 through the ε/η sound); and the assonance across 51 and 52 (of the diphthong ει in φιλεῖ and ποιεῖν in 51, and εὐτακτεῖν in 52).

§46. Acts 53-55 exemplify the statement made in the previous move: καί in 53 signals the start of a new move, and the extra space in OCT seems to acknowledge that. It stands alone before the subclause of 54 because it projects multiple acts, and therefore works as a priming act. The second καί in 55 is an “apodotic” καί that has clausal scope and marks an immediate correspondence between thoughts and actions. [58] The zooming out from thoughts results in the shift to action.

§47. The last move (and act) of this excerpt, which opens with a further καί, is a coda that completes Thucydides’ zooming out. The conjunct that καί links to τὸ θέρος ἐτελεύτα is the entire Thucydidean account of that particular summer (the nineteenth since the start of the war; it occupies chapters 19 to 87 of book 7, and the current chapter of book 8). The statement functions as a formulaic closure of a substantial stretch of text, and as such presents itself as a unit with more independence than modern punctuation gives it credit for (see the modern comma after ταῦτα, 55). Conversely, in his English translation Hammond (2009:415) indents, which harmonizes with what a discourse analysis suggests. [59]

§48. All in all, the number of acts and moves across these two first excerpts is similar: the speech includes 57 acts and 10 moves; the narrator text 56 acts and 12 moves. Both stretches of text allow the audience to chunk discourse in manageable units, regardless of hypotaxis and the complexity of the content.

§49. The next excerpt is thematically similar to the passage just analyzed. Just as Thucydides 8.1 narrates the reactions of the Athenians after the devastating news of the defeat in Sicily, Herodotus 8.108.1 narrates the reaction of the Greeks, this time to the successful battle at Salamis against the Persians. The Herodotean account, like the Thucydidean one, displays a narrative style typical of the historian, masterfully combining thoughts and actions, events and words into a harmonious whole. [60]

§50. I have chosen to juxtapose similar sections from the two authors so that we may be able to compare their styles at the level of discourse segmentation. Conventional wisdom might lead us to expect Herodotean narrative to be divisible into more (i.e. smaller) segments than Thucydides’, for example, or to give evidence of more extensive use of parataxis. As the analysis will show, neither of these two expectations is met. [61]

§51. The excerpt is divided into 18 moves, by means of which Herodotus points out the following: the Greeks expect the Persian ships to be at Phalerum (1-6); they get ready to chase them (7-8); as they do not see them, they move towards Andros (9-11). They have an assembly in Andros (12-13). The historian zooms in on Themistocles, who expresses his opinion (14-18). Then Eurybiades takes the floor (19-21). Herodotus presents Eurybiades’ idea in brief (22-23). The first point is that it is problematic if Xerxes is forced to stay in Greece (24-27); either if he keeps things quiet (28-33); or if he takes initiatives (34-39). In the latter case, Eurybiades remarks, the Persians would steal the nourishment of the Greeks (40-41). Eurybiades thinks that Xerxes actually will not stay (42-45). It is better to let him go (46-47); it is better that the battle goes on in Persia (48-49). The other generals agree with Eurybiades (50). Themistocles realizes his low chance of convincing the Greeks to sail (51-54). But he also realizes that he could count on the Athenians (55-58). So, he starts a speech (59).

§52. Here is the text of the first move: From a discourse perspective temporal or causal clauses accompanied by δέ often provide new starting points for subsequent action. [62] Accordingly, this δέ marks the beginning of a new act that is also the beginning of a move that extends beyond the temporal or causal clause and embraces segments 1 to 6. At the same time, act 1 provides the setting for the entire account of the day after the battle, chapters 108 to 112. It also resumes the narrative thread: after the account of the battle, which ends at chapter 96, Herodotus shifts to narrating events on the Persian side, recounting Xerxes’ decision to retreat and leave Mardonius in charge of a new campaign (chapters 97 to 107). Acts 1 and 2 (ὡς δὲ ἡμέρη ἐγίνετο | ὁρῶντες οἱ Ἕλληνες κατὰ χώρην μένοντα τὸν στρατὸν τὸν πεζόν) play the cognitive role of reorienting the audience’s attention to the Greek side of the battlefield, not differently from act 1 in the Thucydidean excerpt above (ἐς δὲ τὰς Ἀθήνας, 8.1.1).

§53. Act 2 (ὁρῶντες οἱ Ἕλληνες κατὰ χώρην μένοντα τὸν στρατὸν τὸν πεζὸν) actually narrows the field of vision in the audience’s mind’s eye by orienting it through the vision of the Greeks: the Greeks see that the Persian land army is still there, at Phalerum. The participle ὁρῶντες instantiates a strategy very common in historiography, that of providing “participial preludes” (the use of participles to encode introductions to actions). [63]

§54. I read what follows as a subtle articulation of the Greeks’ reaction, step by step, and, from the perspective of Herodotus’ narration, act by act. Acts 3 to 5 all start with an imperfect verb (ἤλπιζον, 3, ἐδόκεον, 4, and παραρτέοντο, 5). However, the symmetry is interwoven with a different type of design, embracing the verbs starting 2 to 5: “observing - expecting,” 2-3, on the one hand, and “expressing a view – preparing,” 4-5, on the other. Particles suggest a further link, which distinguish acts 1-3 from acts 4-6. ἐδόκεόν τε ναυμαχήσειν σφέας, 4, and παραρτέοντό τε, 5, display a bisyndetic construction of τε with clausal scope. This construction separates 4 from 3, it stresses a particularly tight link between the two conjuncts, [64] and makes acts 4-6 a whole. A further feature that makes 4-6 cohere is the double use of future tense (ναυμαχήσειν, 4, and ἀλεξησόμενοι, 6). A full stop after Φάληρον, 3, would reflect the relevance of the bisyndetic construction to discourse segmentation. The comma of modern editions, however, invites a continuous reading that obscures the distinction between observation and its practical consequences. The final construction ὡς + future participle confirms Herodotus’ access to the internal stance of the Greeks in line with the preceding acts.

§55. An extra space in the OCT not accompanying a section or chapter number matches my reading of a move boundary between 6 and 7. Act 7 contains a construction parallel to act 1, a temporal clause with δέ. This is the clue that a new move is being established: just as the previous one did, the same construction here provides a new contextual frame for a new action. [65] Now, the focus is on the decision to chase the Persian ships, which constitutes the next step in reasoning after the initial defensive thought (acts 1-6). This decision, in turn, establishes the ground for a series of subsequent narrative steps, made up of an account of the chasing trip and events that occur along the way (acts 7-11), and a debate taking place in Andros (intermediate stop on the way to the Hellespont, acts 12-13).

§56. The next move boundary in my reading once more coincides with the paralinguistic suggestion given by an extra space between words in the OCT edition. It it tempting to think that the extra space is connected to the occurrence of μέν νυν, 9. As IV.3.11.4 has shown, this combination often occurs either at the beginning or at the end of episodes. In this case μέν νυν may indicate the end of the Greeks’ speculations, or alternatively, mark the beginning of the Greeks’ realization that they had made an incorrect assumption. This beginning puts upfront the fundamental object of the Greeks’ considerations thus far, the Persians’ ships, to which the denial of their expectation follows. I read τὸν … ναυτικὸν τὸν Ξέρξεω στρατόν as a separate act that isolates a grammatical object preceding the verb it depends on (ἐπεῖδον), and οὐκ ἐπεῖδον, 10, as a further discourse act stressing the negative outcome. Acts 10 and 11 in fact close the scenario opened by ὁρῶντες, 2, τὸν στρατόν, 2 (even though at 2 it refers to the land army), and τὰς νέας, 3. The particles and the semantic design of these acts make me see a separate move, rather than the first part of a μέν … δέ construction (see δέ in 12 here below) with νυν working as a distinct discourse marker. [66] Accordingly, the modern comma at 11 could be replaced by a full stop.

§57. In 11 and 12 “Andros” is repeated, which makes more sense if 11 and 12 are taken as belonging to two different moves. The length of the verb form and the imperfect of ἐβουλεύοντο suggest an iconic rendering of what could have been a long consultation. [67]

§58. The next move indicates a shift from collective ideas and actions to the individual proposal of Themistocles. Once again, μέν νυν performs the segmenting function. “Themistocles” semantically provides a new entry—an individual emerging from an unspecified group of Greek soldiers. μέν νυν tells us that something distinct from the preceding discourse is going to start. The semantic juxtaposition that the historian inserts later by mentioning a second individual and “the opposite opinion” (Εὐρυβιάδης δὲ τὴν ἐναντίην ταύτῃ γνώμην ἐτίθετο, 19-20) makes the two characters the salient figures and dialectical voices of the moment. [68] Tο be sure, μέν (even within the combination μέν νυν) and δέ do not imply the semantic juxtaposition. They just tell us to process the content in terms of two aspects of a larger whole; it is the surrounding components that make the two aspects semantically symmetrical and opposing. [69]

§59. The next move, actually, anticipates something asymmetrical in the juxtaposition, on the pragmatic and cognitive level: The medieval comma after Εὐρυβιάδης δέ suggests a possible performative pause or intonational break in line with the prosodic treatment of Εὐρυβιάδης δέ as a “first position”-constituent. [70] I read the phrase as a priming act, even though syntactically it is fully integrated into the main clause of which it is the grammatical subject (see 20). The point is that the historian re-orients the audience’s attention to Eurybiades, to expand on the latter’s opinion. The key to that is λέγων ὡς, 21, a metanarrative act introducing Eurybiades’ relatively long speech (indirectly reported). The occurrence of εἰ immediately afterwards, and the function of ὡς following “saying”—i.e. to signpost the upcoming reported speech—suggests an act boundary between ὡς and εἰ. On the whole this move anticipates Herodotus’ asymmetrical way of reporting Eurybiades’ opinion. If we consider the account up to now, 19-21 represent a further step in a progressive zooming in on a specific moment of the reactions after Salamis (collective thoughts and actions – assembly – general opinion of Themistocles – reported words by Eurybiades).

§60. Acts 22-48 constitute Herodotus’ summary of Eurybiades’ words. I subdivide the unit into five moves. The first represents a summary of Eurybiades’ ultimate point, which the subsequent γάρ move illustrates in detail. [71] Before I introduce the γάρ move that follows, I highlight a mismatch between the act and move-based segmentation that is under consideration here, and the modern full stops occurring at the end of 13 and, here, at the end of 23. These two full stops make 14-23 a whole. The resulting periodic unit (Θεμιστοκλέης μέν νυν …, 14; the semicolon at 18; Εὐρυβιάδης δέ …, 19; the full stop at 23) embraces Themistocles’ idea (14-18) and the very beginning of Eurybiades’ idea (19-23). I believe this segmentation is largely inspired not only by the sequence of sub- and main clauses, but also by the semantic juxtaposition “Themistocles – Eurybiades.” The act and move-based segmentation suggests something quite different. It follows the flow of discourse beyond hypotaxis, and beyond the presupposition that μέν and δέ must always signal semantic symmetry and opposition. The flow of discourse indicates two separate moves that end before 23: 14-18 presents Themistocles’ idea, and 19-21 projects the account of Eurybiades’ idea. Then, a major discourse boundary is conveyed by λέγων ὡς, 21, in spite of the suspended hypotaxis and the modern comma. Act 21 introduces the whole of Eurybiades’ reported words, which spans acts 22-48.

§61. The next move is quite elaborate; it illustrates what is likely to happen if the Greeks break the bridge and Xerxes remains in Europe. γάρ is the pragmatic signal that the illustration is going to take place. Further moves within the γάρ move can be identified as well. I present here the entire illustration of Eurybiades’ argument (24-41) with a space indicating what I read as further move boundaries in between (see the corresponding indentations in my translation). The acts in this passage are easily identified through syntactic and pragmatic boundaries, mainly due to parallelism and repetition. [72] ὡς in 28 is isolated as it is followed by a μέν act, and thus projects multiple segments that articulate the causal link it expresses. [73] The sequence οὔτε, 30, οὔτε, 32, and τε, 33, lists the consequences that Xerxes will experience if he remains inactive (ἄγοντι μέν οἱ ἡσυχίην, 29). Conversely, the τε in 31 and 36 belong to the idiomatic phrase οἶός τε εἰμί.

§62. Acts 35-39 unfold what 34 projects, that is, what could happen in case Xerxes becomes active (ἐπιχειρέοντι δὲ αὐτῷ καὶ ἔργου ἐχομένῳ, which parallels ἄγοντι μέν οἱ ἡσυχίην, 29). Ιt is noteworthy that also in this case the sense of juxtaposition rests on the semantic opposition given by the co-texts of μέν and δέ, rather than by the bare μέν and δέ. τε καί in 37 signals a shared knowledge-based association of items (κατὰ πόλις τε καὶ κατὰ ἔθνεα). γε at the end of 38 may be seen as corresponding to an intonational mark distinguishing the first from the second conjunct. [74] It also evokes Eurybiades’ voice, as if the historian’s report is about to break the rules of indirect discourse and let the actually uttered words emerge, or pretend to do that. [75] Finally, the τε in act 40 (τροφήν τε ἕξειν …) has scope over 40 and 41: the content and the size of the last sub-move makes me assimilate this τε to those serving to close an argument (IV.2.3.4).

§63. The next acts (42 to 45) comprise a move starting with ἀλλά followed by γάρ: γάρ marks the unfolding of the ἀλλά move; at the same time, its peninitial position isolates ἀλλά, and makes ἀλλά a priming act in itself (like several καί in Thucydides). ἀλλά decidedly re-orients the discourse by signaling a substantial replacement of all the hypotheses outlined in the previous segments. In fact, acts 43 to 47 articulate the precisely opposite scenario: οὐ μενέειν ἐν τῇ Εὐρώπῃ τὸν Πέρσην, 45 vs. ὁ Πέρσης μένειν ἐν τῇ Εὐρώπῃ, 26.

§64. A ὦν move follows: ὦν tells us that the upcoming thought follows from the previous one: Xerxes won’t stay quietly; therefore, let him go. [76]

§65. With the final move Herodotus begins to step out of the reported speech: The historian in 48 still allows the audience to hear Eurybiades’ voice through use of expressions such as the temporal marker τὸ ἐνθεῦτεν—which anchors the report to the moment of the speech—the distancing/dismissive tone of ἐκεῖνος for Xerxes, and the assenting force of ἤδη in his stancetaking. [77] Then (49) he inserts ἐκέλευε, which is the pivotal word of the stepping out, as it occurs after a series of elliptical infinitives. Therefore, I take ἐκέλευε as a separate act. [78]

§66. The next move takes Herodotus definitively out of the reported speech, as the opening οὗτος form indicates. [79] Note the extra space in the OCT edition. As in other occasions, [80] there is a match between this paralinguistic segmentation and my reading of a move boundary. The next move boundary is suggested by even more linguistic features: A new moment of the scene is established by a temporal clause with δέ, ὡς δὲ ἔμαθε, 51. The construction resonates with similar frame-setting clauses used earlier (see ὡς δὲ ἡμέρη ἐγίνετο, 1, and ἐπεὶ δὲ ἐπύθοντο, 7). In addition, the focus turns away from the reaction of a group of listeners, the Peloponnesian generals in segment 51, to the reaction of an individual, who responds by taking action. The role of ὅτι at the end of act 51 is comparable to the role of ὡς at the end of 21: ὅτι signposts the upcoming content of Themistocles’ understanding. I take ὁ Θεμιστοκλέης, 53, as a separate act because of its position within the part of the account that turns the attention back to Themistocles: he is the subject of all the clauses in 51, 52, 54, and, below 59.

§67. γε in 54 puts prosodic emphasis on τοὺς πολλούς in such a manner as to suggest that Themistocles, though realizing that he cannot persuade the majority, still hopes to persuade some of them. The capital letters in my translation are an attempt to render that emphasis. Because γε’s scope in this case is over a πολύ form, the particle works as a scalar quantifier, generating what we would call a scalar implicature: [82] “the majority” evokes a quantitative alternative either “upwards” in the direction of everyone or “downwards” (as here) in the direction of a minority. [83] Combined with the future tense (πείσει, 52), the γε here seems to signpost not only Themistocles’ mental state as he prepares to deliver his speech of response, but also his voice. The whole act 52 stands out as a reproduction of the general’s immediate words, as if Herodotus could present his thought re-uttered. [84]

§68. I divide the remaining text into two moves, which here below are combined for the sake of convenience: γάρ in 55 indicates that Herodotus is about to expand on why Themistocles turns to the Athenians. Editors put this γάρ move in a parenthesis. In a hypotactically-based segmentation it is fundamental to signal the occurrence of syntactic non-integratedness; here, the parenthesis signals that the participle μεταβαλών in 54 is temporarily separated from “its” main clause, ἔλεγέ σφι τάδε, 59. However, as I have noted before, [86] the paralinguistic sign of parenthesis suggests that its enclosed contents have minor or peripheral status in the discourse. Not differently from earlier examples, segments 55-58 prove the contrary. οὗτοι, 55, is a non-neutral retrieval of referents. On the one hand it resembles the Herodotean use of οὗτος forms for individuals at the beginning of moves. [87] On the other, it retrospectively clarifies Themistocles’ mental calculation when he assessed τούς γε πολλούς in 50: knowing that the Athenians were already angry and determined to set sail, Themistocles thinks he can count on them. That is, the prosodic emphasis in 52 makes the suggestion – later confirmed by οὗτοι in 55 – that Themistocles hopes to convince, if not the entire majority, at least the Athenians, who make up part of the majority constituted by Athenians and Peloponnesians. [88] Thus γε, especially as a negation precedes it, signposts a contrastive thought that is left unsaid: “not the majority γε” implies “but perhaps some of them.” Taken all together, the thought expressed in 55-58 is not minor or peripheral but crucial, and it is crucial at the place where it occurs—that is, after Themistocles turns his gaze, so to speak, to the Athenians, but before he actually begins to speak, at which point he specifically addresses his comments to them.

§69. The end of the excerpt coincides with ἔλεγέ σφι τάδε, 59. This act has the fundamental segmenting role of announcing the direct speech that follows (109.2-4).

§70. Xerxes’ uncle Artabanus is commonly regarded as one of the “tragic warners” in Herodotus along with Solon, Croesus, and Amasis. The role they all play is “to foreshadow what is going to happen and help the reader notice the blindness of the rulers who did not pay attention to them.” [89] Artabanus plays a major role in Xerxes’ deliberations about whether or not to attack Greece. He contributes thoughts on the gods’ jealousy (7.10), initially objects to Xerxes’ assumptions about dreams (7.16), but changes his mind after having the same dream that Xerxes had (18.2-3). Later, he meditates on the uncertainty of human success (7.46), and at 7.47.2 divulges to Xerxes his dread about two factors that are greatly hostile to the king. His final speeches take place in chapters 49, where he expands on these two factors, and in 51, with his final warning on the Ionians. Xerxes’s final reaction is harsh: “you are most mistaken” (σφάλλεαι … δὴ μάλιστα, 52.1). The king repudiates his uncle and sends him back to Susa (53.1).

§71. Artabanus’ warning speeches have many parallels with Nicias’ speech in Thucydides, and their similar motifs have been discussed by Scardino (2007:720-722). The main difference, Scardino argues (2007:721), is that while Nicias focuses on concrete problems and practical details, Artabanus abstracts philosophy and wisdom from historical events. [90] Not by chance are chapters 49 and 51 rich in gnômai. [91]

§72. The verbal exchange between the king and his uncle occupies chapters 46 to 51, and is quite elaborate. Lang (1984a:34) considers the ten extended turns that occur in these chapters one of the “most impressive and most patterned dialogues in the whole work.” Overall, Xerxes and Artabanus share the same number of turns (five to five). 7.49 and 51 (30 OCT lines on the whole) represent the final turns by Artabanus within this highly dialogic section. Even though the interruption of chapter 50 (the king’s final turn) makes the excerpt(s) less ideal for the comparative purposes of IV.5, Artabanus’ moves are actually quite consistent: while reinterpreting the king’s points in his own terms (see later, §88), in chapter 51 Artabanus fundamentally prolongs his warnings.

§73. I divide Herodotus 7.49 in 11 moves conveying these major points: in general, no blame is to be addressed about the Persian army and ships (1-5); still, land and sea are worrisome (6-11); sea is hostile (12-18); several harbors would be needed (19-22); but that is not the case (23-24); circumstances rule over human beings (25-27); land is the second element (28-30); land is hostile too (32-37); people think that no success is enough (38-39); more territory would mean more trouble (40-43); a man that is ἄριστος has to show sense of fear and boldness (44-49). The first move (Artabanus asserting that no blame is to be addressed about Persian army and ships, in general) corresponds to acts 1-5. After the initial act, the formal address to the king, [92] the first point of the argument is constructed by means of litotes (see οὔτε, 2 and οὔτε, 5). Instead of expressing direct approval, Artabanus stresses that the Persian army and navy can draw no complaint. Act 3 (ὅστις γε σύνεσιν ἔχει) closely resembles ἥτις γε σώφρων (Euripides, Medea 1369, discussed in III.3 §79). In both passages γε marks an element of the discourse that stands out because of some intended contrastive implications. In Artabanus’ reasoning, “whoever has sound judgment would not blame …,” suggests that there could exist one who is not so intelligent. [93] γε works like prosodic stress in English and other modern languages (see the capital letters in my translation); I see it as a reflection of the speaker’s voice. [94]

§74. The next move enacts a shift of attention to two crucial elements of Artabanus’ argument that he mentioned at an earlier point (7.47.2; see below): land and sea. Τhe move is divided into two parts: 6-9 and 10-11. Both parts center on the notion “two things” (τὰ δύο, 7; τὰ δέ δύο, 10). As for the move start, both Hude in the OCT and Legrand in the TLG online have τε, even though the “a” manuscripts (including Laurentianus 70,3) and “P” (Codex Parisinus 1633) have δέ. The and-coordination structure of the passage (οὔτε, 2, οὔτε, 5, τε, 6, and, later, οὔτε, 12) and the regular occurrence of τε at move starts [95] fully support Hude and Legrand’s choice. [96] δέ in act 10 opens the second part of the move by simply marking a new step. The modern full stop after γίνεται, 9, matches the discourse perspective: the speaker stops before making a new and short statement by means of which he names the two elements, “land and sea.” However, the extra space between words notwithstanding, I do not see 10-11 as an independent move, because of the strong thematic coherence provided by the lexical repetition δύο - δύο.

§75. Back to act 7, τοι deserves special consideration, especially since Artabanus uses it seven times in chapter 49 (see also acts 24, 29—again with δύο—, 33, 34, 35, 40, and also 19 by means of καίτοι). As I say in IV.4.3.2, in Herodotus τοι oscillates between the reference to “you” and the interactional usage. Ultimately its use has a crucial interpersonal and argumentative function: it marks the exclusivity of the face-to-face interaction (see the translation “let me tell you,” [97] ), and it generally suggests a persuasive intention by the speaker. We may say that τοι characterizes the move pragmatically, just as “the two things” characterizes the move semantically.

§76. In fact τὰ δύο τοι, 7, constitutes a crucial independent act. It retrieves the topic of the two “most hostile” elements anticipated in a previous speech and left unexplained. At 7.47.2 Artabanus’ words are ὁρέων τοι δύο τὰ μέγιστα πάντων ἐόντα πολεμιώτατα (“as I see it, let me tell you, two greatest things are the greatest enemies of all”). Those words are retrieved in chapter 49 by means of τοι, 7, δύο, 7 and 10—and later in 29—and πολεμιώτερα, 9). Essentially τὰ δύο τοι establishes the subject matter of what follows. The verb λέγω in 8 semantically and pragmatically stresses the ongoing discourse being performed by Artabanus.

§77. The next move, a γάρ move, explains why Artabanus thinks of the sea as a great enemy. οὔτε γάρ in 12 has a double significance. First, οὔτε resonates with οὔτε, 2, and οὔτε, 5, and it progresses the series of thematic additions by means of τε (a big army, many ships, gathering more, and now a large harbor). Second, however different the scope and the syntactic functions of the two particles may be, τε near γάρ in Herodotus can accompany the presentation of knowledge shared beyond the context of the ongoing narrative or speech. [99] In cognitive terms οὔτε γάρ in 12 introduces “unframed discourse” [100] about harbors that can (cannot, in this case) receive a fleet adequately.

§78. Even though οὔτε γὰρ τῆς θαλάσσης ἔστι λιμὴν τοσοῦτος οὐδαμόθι is syntactically unified, my pragmatic and cognitive reading divides it into two distinct acts. Beyond syntax, οὔτε γὰρ τῆς θαλάσσης, 12, is a priming act establishing “the sea” as the subject to which Artabanus wants to draw Xerxes’ attention; [101] the subject is going to be articulated in several acts. I take the double negation (οὔτε, 12, οὐδαμόθι, 13) and the existential use of ἔστι as further suggestions about an act boundary between 12 and 13. [102]

§79. Acts 19-22 constitute a καίτοι move, whose purpose is to adjust the previous statement by adding further details. [103] The extra space not accompanying any new section number in this case matches a reading that gives pragmatic importance to καίτοι. Because of its overarching function, and because of its position before the negative marker οὐκί, καίτοι represents an act in itself. ἀλλά in 21 “answers” the negation οὐκί in 19 by introducing a replacement on the semantic level (“not X but Y”). [105] δή in 22 occurs in act-peninitial position after the constituent παρ’ ἥν. In the light of its co-text and context, the particle marks a piece of information that is interactionally evident: from the preceding discourse it is obvious to speaker and addressee that the need of multiple harbors refers to a piecemeal coastal navigation. [106]

§80. The next move is an ὦν δή move: in this case δή has a different role, namely in combination with ὦν it marks the next major step in discourse. ὦν’s contribution is to project a logically consequent step. [107] Translations tend to mingle the two acts in one coherent clause (e.g. De Sélincourt and Marincola write, “But there is not a single one,” 2003:434). However, the pace of speech according to the two acts is different: “That is not the case—mind you, that there are capacious harbors.” The act boundary is given, in my view, by the genitive absolute construction, which makes 24 a separate act. τοι turns out to be in act-peninitial position. [108]

§81. Under the influence of hypotactic principles, translations tend to link the genitive absolute above, “that there are capacious harbors” (οὐκ ὦν δὴ ἐόντων τοι λιμένων ὑποδεξίων, 24) to the immediately following finite verb μάθε, 25 (see below). The rationale is that a secondary clause such as ἐόντων … ὑποδεξίων cannot stand alone, but needs to be attached to an independent finite verb—hence the comma after ὑποδεξίων. If we conversely segment the text on the pragmatic basis of acts and moves, the result is that 25-27 can be seen as a distinct move: The goal of Artabanus’ move is to let the king learn from proverbial wisdom. While 23 and 24 negate the existence of capacious harbors, 25-27 generalize Artabanus’ point: actual circumstances can be more decisive than human determination. As argued in II.3 (§§50, 68, 70, 74), Pindaric beginnings that lack syndetic items tend to encode major discourse transitions. Here the imperative form μάθε is discontinuous enough, especially if we consider that it conveys an authoritative instruction given by a king’s uncle to the king himself. [109] Furthermore, as III.4.5.3 shows, dialogic turns in drama that start with no connective items do usually encode a connection with the context or co-text in other ways (for example by means of an anaphoric οὗτος form, or an appeal to the “you” interlocutor). This seems to be what happens here as well, even if we deal with a move within the same speech rather than a new turn in drama (which does not necessarily coincide with a new move).

§82. In a discourse perspective, a salient element in segments 26 and 27 is οὐκί, 27, as it introduces what in pragmatics is called a metalinguistic negation. A metalinguistic negation is when an entire utterance is negated, and not just the individual element(s) in it. [110] Or as Miestamo (2009:221) puts it, “metalinguistic negations are objections to utterances.” With his utterance Artabanus attempts a radical alteration of thought; he instructs Xerxes that the saying “circumstances control human beings” means (see καί, “that is to say”) that the reverse “human beings control circumstances” does not hold true. The translation offered here takes the role of ὅτι to be the equivalent of a colon, the paralinguistic sign that we use to introduce a quotation or direct speech. [111]

§83. The next move starts with an extra space in OCT that seems to acknowledge the force of καὶ δή at move starts. καὶ δή works as a cluster, and constitutes an act in itself. [112] Its function is to focus the hearer or interlocutor’s attention on something in particular, and at that particular moment of the discourse—hence the translation “now look.” I read the position of τοι in 29 to be peninitial after the constituent τῶν δύο. The assonance and alliteration in segments 30-31 (ἑτέρου εἰρημένου + ἕτερον ἔρχομαι ἐρέων), the lexical insistence on forms of the same verbum dicendi and on ἕτερος forms, and the explicit performative verbs ἔρχομαι ἐρέων add prominence to the move that Artabanus is about to perform.

§84. Overall, acts 28-31 represent an elaborate introduction to the topic “land,” second cardinal entry of the speech, and subject of the next move (32-37). Modern editions make the beginning of section 4 of chapter 49 coincide with γῆ δὴ πολεμίη. My discourse analysis backs this reading up: after the elaborate announcement about the second hostile element that Xerxes should take into account, Artabanus opens a new move by uttering γῆ δὴ πολεμίη, 32, a communicative step that is relevant for three reasons. First, the speaker includes the explicit mention of the element (γῆ); second, he reconnects the main point to previous discourse (πολεμίη recalls πολεμιώτερα at 9 and πολεμιώτατα, 47.2); third, he uses δή to openly mark γῆ … πολεμίη as his own stance: he thinks that yes, definitely, land is hostile. [114] In the remainder, lexical repetition reinforces the idea (see πολεμιωτέρη, 35) and the pragmatic intention to persuade the king (see τοι repeated in 33, 34, and 35).

§85. In line with the gnomic conclusion of the argument about sea, Artabanus concludes the argument about land with another gnṓmē: The word order of this statement suggests a possible segmentation in two acts. Let us recall οὔτε γὰρ τῆς θαλάσσης | ἔστι λιμὴν τοσοῦτος οὐδαμόθι, 12-13. εὐπρηξίης δέ gives cognitive prominence to the topic of the statement, that is, success (as in 12 “the sea” does). Of the 13 remaining instances of οὐκ ἔστι in Herodotus, nine can be read as act-initial. [116] The reinforcement of the negation through οὐδεμία within the same act occurs also at 8.118.3, also in direct speech: Δέσποτα, οὐκ ἔστι οὐδεμία [sc. σωτηρίη] “Sire, there is none [no rescue is possible].”

§86. After closing the parallel arguments on sea and land, Artabanus tries to persuade his interlocutor by making a point on a major risk involving territorial expansion in general. In this new move we find a further καὶ δή (cf. 29), a further τοι (cf. 7, 24, 29, 33, 34, 35), and a further λέγω (cf. 8; cf. also ἐγὼ εἰκάζω, 14, and ἔρχομαι ἐρέων, 31). These features strengthen the direct, attentive, and authoritative tone adopted by Artabanus. In act 43 word order and alliterative as well as assonantal patterns add rhetorical force to the message (τὴν χώρην πλεῦνα ἐν πλέονι χρόνῳ).

§87. Artabanus’ last move consists in revealing two fundamental qualities that a man excelling over the others (ἀνὴρ … ἄριστος, with indirect reference to Xerxes) should demonstrate. This last move mixes features of gnômai with wishes: general terms (ἀνήρ, 44), and general states (πᾶν … χρῆμα, 48; τῷ ἔργῳ, 49) are combined with desirable states (ἄριστος, 45; θρασύς, 49), and a conditional period framing the whole. Particles, subordinating conjunctions, and adverbial pronouns signpost almost each argumentative act. οὕτω projects a longer exposition. [117] εἰ appears alone as it starts the expression of a multiple condition, articulated by μέν, 47, and δέ, 49. The hyperbaton πᾶν … χρῆμα makes for 48 as an individual act.

§88. In his reply to this speech (chapter 50), Xerxes points out that taking risks and running into trouble is better than fearing everything. Great gains require great risks (μεγάλα γὰρ πρήγματα μεγάλοισι κινδύνοισι ἐθέλει καταιρέεσθαι, 50.3). So, he continues, they will conquer all Europe and will come back home safely (καταστρεψάμενοι πᾶσαν τὴν Εὐρώπην νοστήσομεν ὀπίσω, 50.3). Artabanus’ tactic in responding to Xerxes is to summarize the king’s views by offering an account in his own terms (see, for example, ἀρρωδέειν, below in 2, which reminds us of ἀρρωδέοι, act 47 of chapter 49).

§89. Chapter 51 can be divided into 8 moves, with the following communicative goals: to invite Xerxes to follow his counsel (1-3); to motivate more warnings as more facts deserve attention (4); to recall Cyrus’ conquest of Ionia except for Athens, the mother state (5-6); to suggest that it is not a good idea to put the Ionians against Athens (7-9); to reassure that the Persians can be successful without the Ionians (10-11); to state that the Ionians rightly would turn out to act disadvantageously for the Persians (12-15); to further illustrate the disadvantages (16-19); finally, to recall that end points are not necessarily as clear as beginning points (20-22).

§90. By means of the opening move, Artabanus insists that the king should accept his advice. The boundaries in these three acts fully coincide with medieval interpunction and modern punctuation. It is not by chance that after the vocative particle ὦ, 1, a subordinating conjunction (ἐπείτε, 2) and a postpositive particle (δέ, 3) occur at each beginning; they signpost the start of discourse acts, however syntactically diverse. δέ in 3 is an instance of “apodotic” δέ. [118] Even though it has no syntactic function, on the pragmatic level the particle marks a discourse discontinuity between the act motivating Artabanus’ further advice (2) and the authoritative act of recommending Xerxes to accept his counsel.

§91. The following move strategically motivates the need for further counsel, by means of a call for fair proportion: a longer series of facts requires a longer series of statements. γάρ and ἔχει in 4 encode the marking of unframed discourse through the associative force of γάρ, and the comprehensiveness of the present tense. The statement sounds gnomic, in line with some of Artabanus’ previous moves (acts 26-27, 38-39, and 44-49 in chapter 49). Note in addition the alliteration of “p” through περὶ πολλῶν πρηγμάτων πλεῦνα. It is as if Artabanus exploits general wisdom to back up his will to extend the warnings. The proposed act-based segmentation enhances what could be a deliberate pun by Herodotus: exactly when Artabanus points out the necessity of more words, he delivers a particularly long segment, ending with ἐκτεῖναι, “stretching out.”

§92. The start of the next move is asyndetic: The formal method of referring to Xerxes’ grandfather Cyrus, by giving his proper name together with the patronymic, enables Artabanus to strategically place his whole statement in a broader, countrywide perspective, while at the same time calling on the admonitory power of historical precedent. [119] The extra space in the OCT (without section number) matches my reading of a move boundary. By means of this asyndetic statement Artabanus starts a series of considerations centered on the Ionians, the subject of his last elaborate argument. Therefore, retrospectively I interpret acts 5-6 as conveying a major transition in the discourse. [120]

§93. Artabanus’ considerations about the Ionians start with an ὦν move: it is not good to ask the Ionians to assist in the conquest of Athens, their mother city: ὦν in 7 suggests to the audience that they process the content of the move as a logical consequence to the previous utterance. τούτους … τοὺς ἄνδρας is an anaphoric expression; the referents (the Ionians) are to be mentally retrieved on the basis of the preceding phrase Ἰωνίην πᾶσαν, 5. [121] τοι at 8 conveys the speaker’s persuasive intention, and is a further marker of the move as a whole (see the performative verb συμβουλεύω “I advise,” which recalls previous analogous verbs such as λέγω, and 8, ἐγὼ εἰκάζω, 14, and ἔρχομαι ἐρέων, 31).

§94. The following καὶ γάρ does not work as a cluster. γάρ signposts the new move as supporting the advice provided, whereas καί simply intensifies ἄνευ τούτων (“even without them”). [122] Although τε belongs to the idiomatic phrase οἶός τε εἰμί, it serves as a postpositive word that indicates an intonational and discoursive boundary before its preceding word (οἷοί in this case). This makes for 11 as a separate act.

§95. The following is an extended γάρ move. γάρ in 12 introduces an expansion of the point, “it is not a good idea to pit the Ionians against their fathers (i.e. the Athenians).” σφεας, near γάρ, grants referent and topic continuity. The move in turn consists of two parallel expansion-moves. The first illustrates equally unfortunate scenarios: “If the Ionians were to enslave their mother state on behalf of us, this would be unrighteous for them and for Athens; if they would liberate their mother state on behalf of Athens, this would be inconvenient for us” (13-15). The second expansion spells out the negative consequences: “neither behavior would be advantageous for the Persians: we would have either no profit, or—directly—damage” (16-19). Alongside γάρ, 12, and νυν, 16, the semantically antithetical terms “most unrighteous” vs. “most righteous” (ἀδικότατος vs. δικότατος) determine the structure of these expansions, where each term is used twice (ἤ … ἀδικωτάτους, 14, ἢ δικαιοτάτους, 15; [124] ἀδικώτατοι μέν, 16; δικαιότατοι δέ, 18). In the light of this semantic structure, Ι do not see νυν in 16 as part of a μέν νυν cluster, but, rather, as a particle introducing the second expansion. [125]

§96. Overall, there is a discrepancy between this discourse analysis and the modern indication of discourse boundaries. Section numbers 2 and 3 are located near τούτους ὦν τοὺς ἄνδρας and ἀδικώτατοι μέν νυν respectively; they suggest that major transitions occur first between “Cyrus did not conquer Athens” and “no Ionians against the Athenians”, and, second, between “righteousness and unrighteousness of the Ionians” and “bad effects of both the righteousness and the unrighteousness of the Ionians.” Attention to particles and to other co-textual elements indicates a different articulation. The major boundary occurs at 5-6 with the asyndetic start “Cyrus …”. Then the consequent thought (“ὦν no Ionians against the Athenians,” 7-9) is illustrated through two γάρ moves (“we can win wthout them,” 10-11, and “both their righteousness and unrighteousness would be detrimental for us,” 12-19; the latter being split into two parts).

§97. In line with previous argumentative closures, Artabanus inserts at this point a final gnomic thought. For the third time a stern instruction to listen to the old uncle occurs (see the imperative βάλεο, 20, which reminds us of μάθε (act 25 in chapter 49), and ἔνδεξαι, 3 earlier in this chapter). ὦν marks the move as a thought that naturally follows from the preceding discourse. Actually this conclusive thought rounds off all of Artabanus’ conversational turns in chapters 46 to 51. [126] What makes it sound like an overarching consideration by the wise warner is the double meaning of the saying. Act 22 literally means: “the end is not made all clear together with the beginning” (τὸ μὴ ἅμα ἀρχῇ πᾶν τέλος καταφαίνεσθαι). However, the word sequence highlights ἀρχῇ πᾶν τέλος as the core part of the message, which seems to allude to the end of Xerxes’ reign. [127] The strategically inserted subclause ὡς εὖ εἴρηται (21) stresses the clever formulation. [128]

§98. Let us first compare some data across the four excerpts. Table 1 assumes that sections are identified via numbers, períodoi are identified by full stops, and clauses include main clauses, subclauses with finite verbs, and participial clauses. A qualitative difference should be stressed concerning clauses vs. acts in general: while infinitives often fall in the same act together with the verb they depend on, several acts include no verb form at all.

§99. In terms of large-scale segmentation (sections, períodoi and moves), the Herodotean speech, excerpt 1, consistently shows higher numbers, roughly doubling those of Thucydides’ speech, excerpt 4 (8 vs. 4 sections; 15 vs. 7 períodoi; 19 vs. 10 moves). There is a convergence in the perception of a more segmented discourse in excerpt 4 as opposed to the less segmented discourse in excerpt 1. However, while the Herodotean excerpts (3 and 4) significantly differ in number of períodoi (8 vs. 15), the number of moves is comparable (18 vs. 19). This suggests that from a pragmatic perspective major discourse boundaries are found equally often in narrator text and in speech, at least in the two excerpts under examination. This invites us to consider that a holistic discourse approach can capture discontinuities that modern punctuation and its focus on hypotaxis ignores.

§100. Admittedly, most full stops coincide with move ends. Also, all section numbers coincide with move starts. However, the reading in moves only partially coincides with the OCT segmentation provided by section numbers and by full stops wrapping up períodoi. For example, in excerpt 1 the (modern) very long period almost overlapping with Thucydides 6.22—which, I repeat, in Stuart and Jones’ edition has no internal sections—is here divided into five distinct moves. Moreover, in spite of the modern full stop closing the supposedly long period at the end of 25, I see thematic continuation between 25 and 28. Further evidence of mismatch between my discourse segmentation and section divisions concerns γάρ in all the excerpts. γάρ is a move starter, just as in the other ‘monological’ texts of our corpus (Homer and Pindar). However, only in Thucydides 23.1 (excerpt 1), Herodotus 108.3 (excerpt 3), and Herodotus 7.49.2 (excerpt 4) does a new section number co-occur with the act hosting γάρ (against 10 γάρ moves).

§101. In terms of small-scale segmentation, the similar number of acts across the first three excerpts is representative of a finding that I believe to be general and crucial. In both authors the flow of discourse proceeds by steps that are relatively short. Not only do syntactic boundaries, particles, and particle combinations contribute to that, but also repetitions and various kinds of parallelism.

§102. The division of the Herodotean speech in 71 acts reaffirms the possibility of chunking discourse in small steps. The remarkable difference between acts and clauses in excerpts 3 and 4 (71 acts and 52 clauses in the Herodotean speech, vs. 59 acts and 58 clauses in the Herodotean narrator text) consolidates the idea that several act boundaries can be found on the sub-clausal level.

§103. The segmentation in acts produces the effect of evening out syntactic hierarchies; therefore, it enhances the importance of single communicative steps whose syntactic form may suggest a subsidiary status. Priming acts constitute a particularly good case in point: they are short and syntactically dependent on what follows; still, they have an important pragmatic and cognitive function (e.g. ὁπλίτας τε οὖν πολλούς, 1 in excerpt 1; ἐς δὲ τὰς Ἀθήνας, 1, and ὅμως δέ, 36, in excerpt 2; Εὐρυβιάδης δέ, 19, in excerpt 3; οὔτε γὰρ τῆς θαλάσσης, 12, in excerpt 4, chapter 49).

§104. Independently of whether they have a priming function or not, some of the acts that I identified are very short (e.g. in excerpt 3: 13, 25, 42, 49—one word; 5, 6, 10, 19, 21, 43, 53—two words; 15, 24, 30, 38, 44, 59—three words). I draw a potential link between them, the kómmata mentioned by ancient writers (see IV.3 §§43-45, 46, 48-49), and Fraenkel’s Kurzkola (II.2 §14). The medieval punctuation sometimes indicates boundaries between very short stretches of text as well—invariably, phrases. [130] However partial this result is, it suggests that our canonical segmentation in subclauses and main clauses is not the only way to see how discourse progresses. A holistic approach lets the role of particles and other features emerge together with the canonical segmentation.

§105. As for matches and mismatches with modern commas, full stops, and with medieval punctuation, two simple results emerge. First, more than once medieval punctuation in the manuscript appears at the end of what I read as an act, whereas no modern punctuation is inserted (see 3, 6, 20, 29, 33, 39, 45, 54 in excerpt 1; 3, 11, 12, 15, 17, 21, 23, 24, 25, 27, 32 in excerpt 2; 2, 4, 14, 16, 19, 29, 31, 34, 44, 54, 58 in excerpt 3; 8, 30 in chapter 49, and 16 and 18 in chapter 51 in excerpt 4). In far fewer cases modern punctuation appears where there is no medieval sign (see 9, 15, 18, 26, 41, 44, and 49 in excerpt 1; 2, 7, 29, 40, 41, 42 in excerpt 2; 11, 46, and 57 in excerpt 3). This suggests that medieval punctuation matches the division in acts more than modern punctuation does, at least in three excerpts—excerpt 4 (Artabanus) is different: at 2, 3, 13, 36, 37, 40, 41 in chapter 49, and 12, 13 in chapter 51, modern punctuation appears where there is no medieval sign.

§106. Second, there is no clear correspondence between modern full stops and medieval high dots, and between modern commas and medieval mid dots: high dots are equally present when full stops and commas occur; and modern commas may equally correspond to mid as well as to high dots. This means that the modern indication of weaker and stronger boundaries is not directly related to the occurrence of specific medieval signs.

§107. Let me now comment on the use of δέ, τε, and καί in relation to discourse segmentation. In excerpt 1 the eight occurrences of δέ (see the beginning of acts 15, 25, 26, 35, 47, 48, 52, 56) and their surrounding text indicate a variety of ways in which the δέ act is discrete—only at 56 does the act mark a semantic opposition. Conversely, in the two narrator texts (excerpts 2 and 3) all δέ οccur at move starts.

§108. As for τε, the most important result is that it can introduce entire moves, especially in Thucydides (see 1, 13, 23, 36 in excerpt 1; 27, 45, 50, in excerpt 2; see also 40 in excerpt 3, and 6 in excerpt 4, chapter 49). In these instances τε follows a term that represents the topic of the next few acts. Almost all the occurrences of τε καί are between noun phrases that are combined on the basis of broadly known semantic frames (τοῖς χρησμολόγοις τε καὶ μάντεσι, 11; φόβος τε καὶ κατάπληξις, 16 in excerpt 2; κατὰ πόλις τε καὶ κατὰ ἔθνεα, 37 in excerpt 3; γῆ τε καὶ θάλασσα, 11 in excerpt 4, chapter 49).

§109. In line with the general frequencies of καί in the two Histories (see I.5) the Thucydidean excerpts include 42 instances, while the Herodotean excerpts include only 10. Several discourse acts start with καί linking conjuncts (with or without particular enrichments); in excerpt 2 this happens 17 times out of 25 total occurrences. Conversely, καί intensifying one constituent has a mobile position within the act (see, e.g. 13, 14, 15, 28, 32, 35, 37 in excerpt 1). It is noteworthy that καί in Thucydides’ narrator text is more frequent than in Thucydides’ speech (25 against 17 tokens).

§110. Some particles are not cues to segmentation, but, rather, markers of someone’s voice or voice and stance in specific acts. γε (πλήν γε πρὸς τὸ μάχιμον αὐτῶν, τὸ ὁπλιτικόν, 31, in excerpt 1; μὴ οὕτω γε ἄγαν πανσυδὶ διεφθάρθαι, 6, in excerpt 2; ἤτοι ἁλισκομένων γε, 38, and οὐ πείσει τούς γε πολλοὺς πλέειν ἐς τὸν Ἑλλήσποντον, 52, in excerpt 3; ὅστις γε σύνεσιν ἔχει, 3 in excerpt 4, chapter 49) signposts vocal liveliness and emphasis on meanings left unsaid. δή (φόβος τε καὶ κατάπληξις μεγίστη δή, 16, in excerpt 2; γῆ δὴ πολεμίη, 32, and παρ’ ἣν δὴ κομίζεαι, 22, in excerpt 4, chapter 49) conveys someone’s certainty or the evident character of something. Finally, ἤδη (κατὰ κράτος ἤδη, 33, in excerpt 2; τὸ ἐνθεῦτεν δὲ περὶ τῆς ἐκείνου ποιέεσθαι ἤδη τὸν ἀγῶνα, 46, in excerpt 3) reflects a sense of confidence as if characters express their stance viva voce.

§111. While particles and particle combinations often relate to move boundaries, an important element contributing to shaping moves is lexical repetition. I have pointed out several repetitions of terms that contribute to thematic and pragmatic coherence. With respect to the broader phenomenon of resonance, which includes multifarious reuses of verbal features across different turns of speaking (see III.3), here we deal with a monologic reuse of words. Further research into the relation between lexical repetition and morphological similarity, on the one hand, and our way of processing the sequence/order of linguistic elements, on the other, could lead to significant results. [131] For the moment I flag three aspects of lexical repetition that make it relevant to segmentation. First, it exemplifies a way in which discourse segmentation goes beyond hypotactic divisions. Second, in spite of Thucydides’ fondness for linguistic variation, the historian offers evidence of lexical repetition at strategic moments of the articulation of the discourse (see e.g. πολλούς, 1 and 10, τὰ ἐπιτήδεια, 14 and 20, ἐκπλεῖν and ἐκπλεῦσαι, 51-52, in excerpt 1; ἐπειδή, 2 and 7, in excerpt 2). Third, the power of lexical repetition does not emerge as much in our silent reading as it does if the text is read aloud. This harmonizes with the idea of “audible” strategies of segmentation, namely the possible link between particles and particular intonational contours, and, more importantly, the potential match between discourse acts and intonation units.

§112. The final point concerns the position of verbs. In his monograph on Thucydides’ narration, Lamb observes (1914:250-254) that verb forms frequently occur at the end of sentences in the form of clausulae (often they are four-syllable verbs). According to the discourse segmentation that I propose, the narrator-text excerpts (2 and 3, thus not only in Thucydides but also in Herodotus) do display verb forms that seem to fit a performative closure: ἐβαρύνοντο, 22, and προβουλεύσουσιν, 49, in excerpt 2; ἐβουλεύοντο, 13, and ἐκέλευε, 49, in excerpt 3. In Thucydides’ excerpts moves mostly end with verbs (9 out of 12 times, and 7 out of 10, respectively). I would take this point further: act division reveals this tendency on an even smaller scale. Verb forms are predominant at the end of acts, at least in narrator text. Besides individual occurrences, let me point out six times in succession (ἠγγέλθη, 2; ἠπίστουν, 3; διαπεφευγόσι, 4; ἀγγέλλουσι, 5; διεφθάρθαι, 6; ἔγνωσαν, 7) and later five times in succession in excerpt 2 (σωφρονίσαι, 45; ἑλέσθαι, 46; παρόντων, 47; ᾖ, 48; προβουλεύσουσιν, 49); four times in succession (περιημέκτεον, 55; πλέειν, 56; βαλόμενοι, 57; βουλοίατο, 58) in Herodotus’ narrator text (excerpt 3). This speech excerpt shows that too: three times in succession, for example, in chapter 49 (λέγω, 8; γίνεται, 9; ἐστί, 10) and four in chapter 51 (γινόμενοι, 16; προσβάλλουσι, 17; γινόμενοι, 18; γίνονται, 19). These data are far too little to make general claims. However, especially for Thucydides, my limited results not only corroborate general findings about SOV clauses in ancient Greek, [132] but also suggest that verb position might be one of the features shaping discourse acts.

§113. To sum up: this analysis shows that the discourse seen act by act and move by move is segmented much more than, and often differently from, what a modern hypotactic analysis yields. The main advantage of this kind of segmentation is the slower pace of reading it institutes, which allows a number of phenomena to be appreciated below and above the level of the clause. It also lets the occurrence and the role of particles and particle combinations emerge. Together with different co-occurring words, particles mark moves with overarching communicative goals, individual acts, the speaker’s and narrator’s voice and stance, and various coordination strategies.