ElAnt v1n1 - Thucydides Constructs his Speakers: The Case of Diodotus

Volume 1, Number 1
June 1993


Daniel P. Tompkins,
Department of Classics,
Temple University,
PA 19122,
e-mail: Pericles@astro.ocis.temple.edu

In 428 BC Lesbos seceded from the Delian League. Recapturing the island in the summer of 427, the Athenians deliberated on a suitable punishment of the rebels, deciding first to kill all the adult males, then reconsidering. Here I concentrate on the speech of Diodotus. My main goal is to interpret his speech by attending as closely as possible to its language, treating it with the respect we might accord, say, a chorus in Sophocles or Euripides.

In the course of developing this interpretation, I will comment on some issues that have resonated through the tradition of Thucydidean scholarship.

First, I'll provide some evidence that Thucydidean speakers are stylistically differentiated, that is, that they speak differently from each other.

Second, I'll show that one stylistic 'marker' of this speech is the speaker's use of general or 'gnomic' sentences.

Third, I'll respond to critics who say that general statements, here and elsewhere, are intrusive insertions that upset the balance of the speech. My response will be that the gnomes are essential, not incidental, and that without them the speech would not mean what it does.

Fourth, I'll argue that the sort of 'meaning' this speech creates sheds important light on Thucydides' methodological statement at 1.22.1 about how he wrote speeches. In a nutshell: there is no way to avoid the conclusion that Thucydides himself is responsible for the most important parts of his speakers' speeches, that is, that for all practical purposes he composed them.

1. Stylistic Differentiation

'Impersonality' has long seemed to characterize the speeches of Thucydides, many of which have struck scholars as having a uniformly inelastic prose style. But looking through the speeches we find a number of speakers who are stylistically distinctive:

a) Hermocrates, the Syracusan leader, uses assonance, rhyme, repetition, etymological play, irony, ambiguity, paradox, short staccato antitheses. In all of these respects he differs somewhat from other speakers, and somewhat resembles his countryman, the great sophist Gorgias.

b) Archidamus, in his first speech, is again distinctive, avoiding many of the formations developed in late-fifth century Attic Greek: -sis nouns, neuter abstractions, words with double prefixes, verbal play, redefinition of words, 'Prodican', hair- splitting distinctions between near-synonyms. He is seldom ironic or metaphorical; he seldom separates the article from its noun, seldom indulges in the extravagant antitheses favored by the Corinthians, or the elaborate and complex participial structures favored by Diodotus or the Corinthians. His sentences are far shorter and less complex than those of Nicias, the prolixity champion of this history, and he uses only half as many conditions.(1)

c) Brasidas uses sis nouns, verbal play, and ambiguity.

d) Nicias stands out for his lengthy, subordinated sentences.

e) Alcibiades is notable for a supple, deceptively paratactic style. (2)

It is essential to mention that quantitative stylistics is a treacherous field, filled with problems that include the subjectivity of individual scholars, the elusive nature of statistical 'significance', the problem of finding 'correlations' between bodies of data, and (often) an inverse ratio between detailed precision and depth of interpretation - or, to rephrase it, a direct ratio between exactness and exiguity. I discuss these questions in a footnote. (3)

2. Gnomes

Diodotus' speech is stylistically interesting in a number of ways. For instance, he shares with his opponent Cleon a predilection for double prefixes (3) and for action words - verbs and participles - over adjectives and nouns. He uses participles far more than most speakers. In this paper, I shall focus on gnomic sentences.

Gnomes, in Greek rhetoric, are usually short pithy statements about human behavior. They are always conceptually and syntactically independent of their contexts. Greek is famous for its reliance on this form:

Sweat before virtue Nothing too much Hybris breeds the tyrant No mortal avoids ate

Thucydidean speakers adapt the traditional language of Greek gnomic poetry, speaking of hope, chance, error, passion, and human nature. As in gnomic poetry, the connection with the surrounding context is left implicit. Thucydides differs from other 'gnomic' authors mainly in the complexity of his gnomic sentences. While other authors only occasionally lengthen their gnomes by accumulating alternatives or appositives (e.g. Sophocles, Antigone 175-177), or by adding subordinate or antithetical clauses (e.g. Antigone 613- 619), Thucydides customarily does this. (5)

Turning to the speech of Diodotus, we find not only a number of gnomic sentences - fourteen at least - but also a number that are gnomoid, that is, that reflect generally on human behavior but don't achieve complete syntactic independence.

3. What does it mean?

What do we learn from these general reflections? First of all, that it is more gnomic, more inclined to general statements about human behavior, than all speeches except the great idealizing vision that is Pericles' last speech.

Disproportion incurs disapprobation. Whenever a Greek writer uses gnomes in quantity, modern critics will claim that the gnomes are excessive. We find this tendency in critics of many Greek authors: in Thucydidean criticism, one thinks of Gomme, Andrewes, Hammond and others. (6)

The leading theme of these critics is that general thoughts weaken an utterance by forcing it to change direction or lose focus. Thus Hammond on a later speech by Hermocrates: 'The universal arguments do not tie in with the historical narrative at all. . . . '. Hammond's solution? That the historian added these arguments much later on, and in so doing revealed a new despondency about Athens and the war.

In the face of this, I propose that we view the speech as an internally coherent piece of rhetoric. Since it's chapter 45 that seems most problematic, let's consider the line of thought in that chapter.

45.1 Despite the death penalty for many crimes (hamartemata), men still take risks, excited by hope (elpidi epairomenoi).

45.2 Seceding cities all, in their imaginat7ion (tei dokesei), have adequate resources.

45.3 All men err (hamartanein): this is a natural phenomenon; no law (nomos) will prevent it, as men have learned after setting down penalties in hopes of avoiding injustice (ei pote hesson adikointo).

45.4 So we must find a more terrible source of fear, or else fear has no force: poverty and excess and other circumstances lead men into danger.

45.5 Hope and passion, (he te elpis kai ho eros), invisible tendencies, are more harmful than dangers we can see.

45.6 Chance (tuche) contributes to men's excitement (epairein). Coming unexpectedly (adoketos), it leads people and cities into danger: cities especially, since they fight for what is more important (freedom and rule over others) and each man irrationally exaggerates (epi pleon ti . . . edoxasen) his chances.

45.7 In a word, it is folly (euetheia) to oppose human nature by law or any other threat.

On one level, Diodotus is advancing a magnificent social- psychological statement about the inevitability of revolution: hope and passion drive men and cities in the empire to error. This is as far as most students of the speech get. I think we can move further now by putting 3.45 under a different light and considering its relation to the rest of the speech. There, in chapters 42-44 and 46-48, we find something very interesting: the language Diodotus applied to Mytilene is applied to Athens herself - to the very city that is now judging the Mytileneans.

1) Diodotus had said that despite the death penalty for crimes (hamartemata), men still take risks, excited by hope (elpidi epairomenoi). But he also speaks of the errors of the Athenians (43.5, 47.1), and of Athenian speakers who commit errors (42.4). (7)

2) Then, Diodotus had stated that criminals are guided by impression or appearance, dokesei: the very word he'd invoked earlier when speaking of Athenian gullibility (45.2, 43.1). (8)

3) A little later, Diodotus told us that xuntuchiai, chance circumstances together with temper (orgei), lead men into dangers (45.4): but chance and temper had combined only a page earlier (43.5) to lead Athens to level foolish punishments. (9)

4) Again, criminals like the Mytilenean rebels are misled by hidden powers (aphane 45.5), which the Athenians also suspect (phaneros, aphanos, 43.3); (10)

5) The Athenians passively follow their speakers as the rebels follow their desire (43.5, 45.5); (11)

6) Lesbos does not plan for the future (45, passim), but Athens must: 44.3, 46.4, 48.2. (12)

When we view the central paragraph of this speech in context, the gnomic statements achieve the opposite of what has generally been alleged: they provide that very unity they are said to subvert. Not only Lesbos but Athens as well commits errors; both are guided by specious appearances; Lesbos is urged on by hidden forces, Athens by fantasies about these same forces; chance, rage, and failure to plan for the future affect both sides. In other words, the gnomes in chapter 45 establish a parity between the Athenians and the Mytileneans they hope to punish.

The gnomic tendency of this paragraph, then, far from distracting us, forces us to view the question of punishment in a new light. The psychological description of rebel behavior is so generally applicable that it applies to other parties than the rebels: Mytilene and Athens, judged and judges, citizens and leaders may all err if they allow themselves to be guided by appearances and the 'whim of a moment', their momentary rage, and so commit an injustice.

Intense and polysemous, Diodotus' analysis reminds us of other great moments in Greek literature when a passage's formal independence permits and encourages us to see its multiple applications: when a chorus sings of a lion cub, of hubris breeding the tyrant, of many wonders but none more wonderful than man.

4. Thucydides constructs his speakers

This brings us back to my title. 'Thucydides constructs his speakers'. For this is what Thucydides does if, as most classicists agree he did, he provided speakers with words. (13) Diodotus is not simply someone who opposes mass slaughter: he is a speaker who projects a consistent interpretation of human behavior, and who forces on us a vision, striking and disconcerting, of the psychological similarity between judgets and judged.

Now, how does this claim sit with Thucydides' famous methodological statement in 1.22.1? It should spur reflection. I'd like to make three points:

a) When Thucydides says he strove - and I think that is the correct tone - to attain the entire gnome of what was said, he must mean something more than 'gist' or 'purport'. I say this because there would have been no need to stress the exertion required to recall the that 'Diodotus opposed mass execution'. (14)

b) More importantly, one of the key words here is an. With it - whether it introduces a contrary to fact condition (Rusten) or a potential statement - Thucydides is telling us that he is creating what he speakers said. Too few people, hardly anyone in fact, have acknowledged the enormous imaginative leverage this gives Thucydides. He is not saying:

I wrote what I think they said,


I wrote what I think they would have said,

a sentence that makes no sense if Thucydides is thought to have been on hand, transcribing many or all of the speeches. (15)

c) Assume that Thucydides had 'notes' on what Diodotus said. Even then, it would have been the historian who performed the task of putting these notes into Thucydidean prose, and that act is the crucial one in creating the 'meaning' of a speech. For it is only in the complex and involuted language of Diodotus' speech that the meaning, the notion of the potential equivalence between judges and judged, the subtle reflections of Diodotus on injustice - becomes evident. Compare Cezanne. After looking at one of his views of M Ste. Victoire, if we're asked to decribe it, we don't say, 'It's about a mountain'. We say the painter used color, proportion, brushtrokes and composition to create a vision. That's just what Thucydides does here, in producing a vision of what Diodotus would say, or - better - would have said. Not just the propositions of his speech, but the diction, syntax, sentence structure, and compositional style, the entire intricate verbal network, create the potential identity between Athens and Mytilene and thus create a large part of the speech's 'meaning'.

Now put this together: (1) As Diodotus shows, Thucydidean speeches achieve their goals not just by stating claims in a certain sequence but by using particular words in particular ways. (2) It is Thucydides himself who provides these words. (3) Finally, Thucydides himself is for all practical and meaningful purposes the 'creator' of the speeches.


NOTE TO READERS: This is a short version of a paper (itself part of a larger draft) delivered on 15 May, 1993 to the Association of Ancient Historians. Responses are solicited.

(1) These comments refer primarily to Archidamus' long speech at 1.80-85.2. See Tompkins 1993.

(2) See Tompkins 1972 on Nicias and Alcibiades.

(3) Quantification depends on definition. The definitions of the stylistic features I study in this essay are subject to some dispute: readers differ, for instance, in identifying instances of irony and ambiguity; our assessment of grammatical subordination depends always on our interpretation of the Greek; what appears to some as a participle, e.g. xumpheron, may seem to others a mere adjective shorn of any verbal force. Readers of any stylistic analysis should be aware that the data were derived by a human being, and reflect that reader's judgment. Thus the standard of 'replicatability' so often instanced in the case of scientific experiments, e.g. the recent 'cold fusion' debate, applies in stylistics only in prosaic (and usually uninteresting) cases. A second problem concerns assessment of quantitative data. Archidamus uses fewer articular forms (participle, infinitive, adjective) than anyone else in Thucydides. We can illustrate this with a chart, but when does Archidamus' disinclination to use these forms become significant? Statistical significance is determined by a variety of devices. I have begun to study these, but will not apply them here. Finally, we need a means of correlating stylistic traits, that is, of tracking them all through the forty-odd speeches in Thucydides and finding whether certain traits cluster at all. Again, I shall not here report on explorations in this area. As one surveys recent work on literary statistics, an increasing reliance on computation by computer, even on machine- readable texts, becomes evident. G. R. Ledger's recent book on Platonic 'style' considers only three stylistic indicators: words containing a certain letter; percentage of words ending in certain letters; percentage of words with certain penultimate letters. Thus restricted, the study's author is able to use machine-readable documents and obtain definite and sophisticated measurements of vocabulary distribution, along the way providing clear and useful discussions of statistical methodology. But he hardly begins to discuss 'style' as most of us normally think of it. A second work, by J. F. Burrows, discusses the distribution of the thirty most common words in Jane Austen; a third, Anthony Kenny's introduction to literary statistics, draws its examples only from counts of letters and words. Restricted to these small units, the authors produce data that are accurate, easily replicated, and often without consequence for literary interpretation. Only Burrows makes any attempt at literary interpretation, using Austen's thirty favorite words to discuss, often profitably, the idiolects of her characters. It is an interesting question whether his perceptive comments depend more on his data or his intelligence. Ledger's book on Plato is intended mainly to solve problems of authenticity and chronology; Kenny's is so purely statistical that it might as well concern performance in baseball. Statistical thoroughness won at such cost will win over few literary critics. Finally, Neumann's quantitative study of the Pauline epistles is worth mentioning because of its thorough and careful methodology. Neumann surveys nearly all work done in literary stylistics in Greek (as well as many other languages), then, using a statistical model, produces evidence that the 'disputed' Pauline epistles fit within the realm of Pauline style. Although literary interpretation is not his goal, this work deserves mention because of its exemplary methodology. The main difficulty any stylistic critic faces is to show how meaning is to be found in the sea of data. In what follows, I am concerned with finding interesting things to say about Archidamus and Thucydides: statistical thoroughness will come later.

(4) anupoptotera 43.2, xunaphistatai 47.2, pro[s]kategorein 42.3, xunexamartanein, etc.

(5) Tabulating 'gnomes' is a bit tricky, since every scholar disagrees about what is and what is not gnomic. (Meister finds gnomes in chapter 45 at sentences 1,3,4,7; I would add 2 & 5.)

(6) Cf. Leaf on Il. 9.320; Gomme on 1.120.4-5, 2.11.4, 3.40.6, 3.43.5; Andrewes on 3.37.3ff., Hammond in Stadter on 4.59-64, Welker on Theognis.

(7) 45.1 En oun tais polesi pollon thanatou zemiai prokeintai, kai ouk ison toide, all'elassonon HAMARTEMATON;

43.5b nun de pros orgen hentina tuchete estin hote sphalentes ten tou peisantos mian gnomen zemioute kai ou tas humeteras auton, ei pollai ousai XUNEXEMARTON.

47.1 humeis de skepsasthe hoson an kai touto HMARTANOITE Kleoni peithomenoi.

42.4cd kai pleist'an orthoito adunatous legein echousa tous toioutous ton politon; elachista gar an peistheien HAMARTANEIN.

(8) 45.2 polis te aphistamene tis po hesso tei DOKESEI echousa ten paraskeuen e oikeian e allon xummachiai toutoi epecheiresen?

43.1 Hon humeis t'anantia dromen, kai proseti en tis kai hupopteuetai kerdous men heneka ta beltista de homos legein, phthonesantes tes ou bebaiou DOKESEOS ton kerdon ten phaneran ophelian tes poleos aphairoumetha.

(9) 45.4 . . . hai d'allai XUNTUCHIAI ORGEI ton anthropon hos hekaste tis katechetai hupanekestou tinos kreissonos exagousin es tous kindunous.

43.5b nun de pros ORGEN hentina TUCHETE estin hote sphalentes ten tou peisantos mian gnomen zemioute kai ou tas humeteras auton, ei pollai ousai xunexemarton.

(10) 45.5 . . . onta APHANE kreisso esti ton horomenon deinon

43.3ab . . . ek tou PROPHANOUS. . . . ho gar didous PHANEROS ti agathon anthupopteuetai APHANOS pei pleon hexein.

(11) 43.5a ei gar ho te peisas kai ho EPISPOMENOS homoios eblaptonto, sophonesteron an ekrinete;

45.5 he te elpis kai ho eros epi panti, ho men hegoumenos, ho d'EPHEPOMENE, kai ho men ten epiboulen ekphrontizon, he de ten euporian tes tuches hupotitheisa, pleista blaptousi, . . .

(12) 45.1 kai oudeis po katagnous heautou me periesesthai toi epibouleumati elthen es to deinon.

44.3a nomizo de peri tou mellontos hemas mallon bouleuesthai e tou parontos.

46.4 hoste . . . dei hemas . . . horan hopos es ton epeita chronon metrios kolazontes tais polesin hexomen es chrematon logon ischuousais chresthai . . .

48.2 tade gar es te to mellon agatha kai tois polemiois ede phobera; hostis gar eu bouleuetai pros tous enantious kreisson estin e met ergon ischuos anoiai epion.

(13) Andrewes' 'Appendix 2. Strata of Composition', in Gomme, Andrewes and Dover, Historical Commentary on Thucydides , Volume V (1981), pp. 396-397, claims that Thucydides captured the arguments of the speeches he reports, but not the words.

(14) Here I concur with Andrewes.

(15) Note that Andrewes' discussion, pp. 396-397, never acknowledges that this word occurs.


Andrewes 1962. A. Andrewes, 'The Mytilene Debate. Thucydides III.36-49', Phoenix 16, pp. 64-85

Burrows 1987. J. F. Burrows, Computation into Criticism. A Study of Jane Austen's Novels and an Experiment in Method (Oxford)

Dover 1973. K. J. Dover, Thucydides (Oxford)

Edmunds, Lowell 1975. L. Edmunds, 'Thucydides' Ethics as Reflected in the Description of Stasis (3.82-83)', HSCPh 79, pp. 73-92

Enright 1986. D. J. Enright, The Alluring Problem (Oxford)

Gildersleeve, Basil Lanneau, 'Brief Mention', AJPh 33, pp. 239-240

Selections from the Brief Mention of Basil Lanneau Gildersleeve , ed. Charles William Emil Miller (Baltimore, 1930)

Gomme, Andrewes and Dover 1981. A.W. Gomme, A. Andrewes, and K.J. Dover, A Historical Commentary on Thucydides Volume V, Book VIII (Oxford)

Hammond 1973. N.G.L. Hammond, 'The Particular and the Universal in the Speeches in Thucydides with Special Reference to that of Hermocrates at Gela', in Stadter 1973, pp. 49-59

Holt 1940. Jens Holt, Les Noms d'Action en -SIS (Aarhus)

Kenny 1982. Anthony Kenny, The Computation of Style (Oxford- New York-Toronto-Sydney-Paris-Frankfurt)

Kenny 1986. Anthony Kenny, A Stylometric Study of the New Testament (Oxford-New York-Toronto-Sydney-Paris-Frankfurt)

Ledger 1989. G. R. Ledger, Re-Counting Plato. A Computer Analysis of Plato's Style (Oxford)

Neumann 1990. Kenneth J. Neumann, The Authenticity of the Pauline Epistles in the Light of Stylostatistical Analysis (Atlanta)

Peradotto 1990. John Peradotto, Man in the Middle Voice (Princeton)

Rusten 1989. J. R. Rusten, ed., Thucydides. The Peloponnesian War. Book Two (Cambridge-New York-New Rochelle-Melbourne- Sydney)

Sihler 1881. E. G. Sihler, 'On the Verbal Abstract Nouns in -sis in Thucydides', TAPhA 12, pp. 96-104

Stadter 1973. Philip A. Stadter, ed., The Speeches in Thucydides (Chapel Hill)

Tompkins 1972. Daniel P. Tompkins, 'Stylistic Characterization in Thucydides: Nicias and Alcibiades', YCS 22, pp. 181-214

Tompkins 1993. Daniel P. Tompkins, 'Archidamus and the Question of Characterization in Thucydides', in Joseph Farrell and Ralph Rosen, edd., Nomodeiktes (Ann Arbor)

Wimsatt 1941. William K. Wimsatt, Jr., The Prose Style of Samuel Johnson. (New Haven)

Woolf, 1925. Virginia Woolf, 'On Not Knowing Greek', in The Common Reader (London)

Daniel P. Tompkins
e-mail: Pericles@astro.ocis.temple.edu

Electronic Antiquity Vol. 1 Issue 1 - June 1993
edited by Peter Toohey and Ian Worthington
ISSN 1320-3606