Black Doves Speak: Herodotus and the Languages of Barbarians

  Munson, Rosaria Vignolo. 2005. Black Doves Speak: Herodotus and the Languages of Barbarians. Hellenic Studies Series 9. Washington, DC: Center for Hellenic Studies.

4. The Meaning of Language Difference

On misunderstanding language difference

Confronted with these two traditions, Herodotus offers his own interpretation (γνώμη), which is far more than an attempt to reconcile them and establish what really happened. In speculating on how the more fantastic Dodonian legend may have originated, the narrator reflects on the process of naming and the transmission of names. Since the Egyptian version seems at least plausible from a rationalist viewpoint, Herodotus starts by assuming as true (εἰ ἀληθέως) the story of the Phoenician abduction that caused the arrival of two priestesses (not doves), one to Libya and the other to Greece, or—as it was then called—Pelasgia. [2] The woman who came to Dodona, he says, must have built a shrine of Zeus under the oak tree in remembrance of the sanctuary she used to tend at home. After learning Greek (συνέλαβε τὴν Ἑλλάδα γλῶσσαν), she founded an oracle and told the local people about how her sister was sold in Libya by the same men who sold her in Thesprotia. Where, then, did the Greek tradition about doves come from? This is how Herodotus explains it:

It seems to me that the reason why these women were called “doves” by the Dodonians is this, that they were barbaroi, and seemed to them to speak like birds. After a while they say she started to speak with human voice (ἀνθρωπείῃ φωνῇ), when the woman uttered sounds that were intelligible to them (συνετά σφι ηὔδα). So long as the woman was speaking barbarian (ἐβαρβάριζε), she seemed to them to be speaking with the voice of birds—for could a dove really speak with human voice? And when they say that the dove was black, they indicate (σημαίνουσι) that the woman was Egyptian.


Black, foreign, and a slave, the woman has been confined to the animal world, yet at the same time regarded as the bearer of a divine message. To the ancient Pelasgians of Dodona, “human voice” is the Greek language and barbarian speech is abnormal in either direction. [
3] But how could anyone fathom a speaking dove? To the Pelasgians, no doubt, the Egyptian priestess is a “dove” in somewhat the same way as to the Scythians snow is “feathers” or to the narrator himself certain Ethiopians sound like bats. The process of transmission, however, has turned the woman into a literal dove just as—less innocently—Cyrus’ nurse Spako became a dog in the discourse of political propaganda. In the case of the Dog-woman the motivation of the initial name-givers is unimportant and unclear, but for the Dove-woman it is crucial and solves the riddle. When the present-day Dodonians say “black dove,” they unwittingly give a sign (sēmainousi) of the truth to Herodotus, who speaks their same “language,” but is also aware of many different languages and knows that common to all of them is the mythopoetic power of names. [4]

Unlike the logos of the doves, this one is sparingly narrated, with no metanarrative interventions. Its mildly anti-Greek bias derives from its being focalized through the Paeonians and from the outcome of the action, which is comically detrimental to the Perinthians. The Paeonians had received an oracle from “the god,” encouraging them to make war on the Perinthians. When the two armies would be ranked opposite to each other, the oracle had said, if the Paeonians should hear the Perinthians call their name, they should immediately attack and victory would be theirs. Subsequently, Perinthians and Paeonians got together for a series of contests. The Perinthians won and began to sing the paean. The Paeonians thought that the Perinthians were calling their name and that their moment had come, just as the oracle had said. They promptly attacked the Perinthians and inflicted on them a severe defeat.

Language makes no difference

The topic of speech forms a counterpoint to the main plot of an ethnographic-historical narrative—another ainos—designed to undermine conventional notions of alterity: [22] the story of the gradual coming together and ultimate fusion of a band of Scythian young men and a parallel contingent of stranded Amazons or, as the Scythians call them, “Mankillers.” [23] The two groups are in principle as mutually incompatible as Amazons and Greeks—or as Greeks and barbarians, the antithesis they symbolize. In practice, however they turn out to be equal and remarkably similar, except in gender and language. As one of the Amazons and the young Scythian who approaches her start dealing with one another in their primitive setting, the narrative again specifies that the two were not able to communicate through speech, but that it was not a problem:

[the Amazon] was not able to speak with her voice (φωνῆσαι) since they did not understand one another, but she told him with her hand (χειρὶ ἔφραζε) to come the next day to the same place and bring another man with him, making a sign (σημαίνουσα) that there should be two, and that she would bring another woman with her.


Just as sex replaces war, here the universal language of gestures overcomes the language barrier. [
24] This first encounter leads to others, and at the end of the story they all get married, Scythians and Amazons, and the Sauromatian nation is born. The text brings up again the issue of language as part of the well-balanced distribution of powers and privileges between the two equal groups in the new society. The Amazons adopt the language of their Scythian husbands. This corresponds to their subordinate position in the sexual sphere but it is also, in Herodotus’ formulation, an index of superior intelligence: “the men were not able to learn (οὐκ ἐδυνέατο μαθεῖν) the language of the women, but the women picked up the language of the men” (4.114.1). Nevertheless, “the Amazons did not learn it perfectly”; as a result the Sauromatae basically speak Scythian, but they distort it (σολοικίζοντες, 4.117). While Herodotus employs the verb soloikizein ‘distort’ in this remark on a dialect which is not, alas, pure Scythian, in other texts the word is applied to bad Greek as a virtual synonym of barbarizein. [25] But by cheerfully assuming the Scythian viewpoint, the ethnographer once again scores a point for relativity on the language front.

The last two sets of interpreters we must consider more deeply relate to the narrator’s overall ideology in the sphere of language difference because they help to stage verbal exchanges where other discrepancies come into play that are more problematic than linguistic difference. [33] As the pyre is already blazing, Croesus remembers his past conversation with Solon and invokes Solon’s name three times:

Cyrus heard him and bid the interpreters to ask Croesus who was this man whose name he had called. They approached him and asked. Croesus was silent for a while, then, as he was being compelled, he answered, “One with whom I think it would be worth a lot of money for all tyrants to have a talk.” Since what he said was unintelligible (ἄσημα) to them, they asked him again. At their insistence … Croesus explained how the Athenian Solon had come to him and, looking at his fortune, had made little of it. He repeated what Solon had told him and remarked how things had gone for him just as Solon had said at the time, speaking not so much for him (Croesus), but for the whole human race (ἐς ἅπαν τὸ ἀνθρώπινον), and especially for those who regard themselves fortunate … And Cyrus, after hearing from the interpreters what Croesus had said, changed his mind, reflecting that he himself was a human being (ἄνθρωπος) and was giving to the flames another human being (ἄλλον ἄνθρωπον); he therefore ordered that the pyre be extinguished as quickly as possible …


The meaning of anthrōpinos as ‘cross-cultural’ in Herodotus is confirmed by the narrative of a verbal exchange inserted in the important series of generalizations at 3.38. This dialogue includes three sets of participants, each using three different types of cultural codes: a linguistic code, a code of communication (how things are said, including expressions and gestures), and a code of customs, the last representing the substance of the discourse. I quote the narrative partially in the context of the interpretive passage where it appears: [37]

1. If one should place all the nomoi in front of all men (πᾶσι ἀνθρώποισι) bidding them to choose the most beautiful, after examining them carefully each would choose his own. Indeed, all men (πάντες ἄνθρωποι) believe that their own nomoi are by far the best … 2. That all men believe this one can see from many pieces of evidence (τεκμηρίοισι) among which the following: 3. Once during his reign Darius called some Greeks who were in the area and asked them how much money it would take for them to agree to eat their dead parents. The Greeks answered that they would not do that for anything. 4. Then Darius called some Callatian Indians, who are wont to eat their parents, and in the presence of the Greeks, who understood through interpreters what was being said, asked them how much money it would take for them to agree to burn their dead parents. They shouted out and bid him to refrain from blasphemy. Thus such things are set by custom νενόμισται and it seems to me that Pindar correctly said in poetry (ὀρθῶς … ποιῆσαι) that Nomos is king of all.


Language difference as paradigm

Transcending culture

In relation to Pindar and other sources on the colonization of Cyrene—as well as, we might add, Greek colonization in general—Herodotus’ narrative presents a striking peculiarity. On the one hand, it reports no fewer than four Delphic oracles that order the Greeks to colonize Libya. On the other hand it also strives to legitimize the foundation not only from a Greek but also from a barbarian viewpoint. [59] It does so, moreover, in a context that exploits the Pindaric themes of Battus’ disability, the voice of the oracle, and the issue of Greek versus native speech. In Herodotus’ scene of Battus’ consultation of the oracle, Apollo, Battus, and Herodotus’ Greek sources engage in a complicated dialogue that is, at one level, about language. Louder than everyone else’s is the voice of the narrator himself, who interferes at every sentence as the interpreter of the Delphic god:

After some time a child was born whose speech was defective and halting, to whom they gave the name Battus [‘Stammerer’], as the Thereans and the Cyreneans say, but in my opinion some other name: he changed his name to Battus after he came to Libya, taking this name as a result of the oracle he received at Delphi and of the honor he was granted there. For battos in Libyan means ‘king’ ( basileus ), and for this reason I think the Pythia addressed to him in the Libyan tongue, knowing that he would be king in Libya. For when he became a man, he went to Delphi about his voice, and the Pythia prophesied to him as follows:

“O Battus, you came for your voice (phōnē), but Phoebus Apollo the Lord sends you as oikist in sheep-nurturing Libya,” just as she would have said in Greek “O king, you came about your voice.”


Herodotus’ translation from the Libyan refashions the Therean and Cyrenaic tradition he himself reports. [
60] It also reverses the pattern, testified in other sources, of “bilingual” oracles appropriating foreign phenomena to the Greek language. In one of the foundation stories, for example, Delphi prescribes that the Sicilian city be named “Gela” (a local name) after the “laughter” (gelas) of the future oikist. [61] With “Battus,” on the contrary, the histōr insists in his own voice that the name by which the Greek founder of Cyrene is most commonly known is a Libyan word because the god who elected him king in Libya “named” him “King” in the local tongue. [62]


[ back ] 1. Cf. 2.42.5: “The Egyptians call Zeus ‘Ammon’.” For the tradition of the common origin of Siwa and Dodona and the controversy about the origins of the two stories reported by Herodotus, see Lloyd 1976.253–254; Fehling 1989.65–70; Nesselrath 1999.

[ back ] 2. On Herodotus’ rationalization on the basis of the likely, see Lloyd 1976.252, Campos Daroca 1992.57.

[ back ] 3. The Greek assimilation of barbarian language to the voice of birds (see above, p. 3 and note 15) coexists with the older tradition according to which the voice of birds is also a metaphor for the voice of the poet (see e.g. Alcman PMG 39, 40) who produces speech that is abnormal but close to that of the gods. See Nagy 1990.88. Here skin color is predictably a secondary index of otherness. Snowden 1983.

[ back ] 4. Spako: 1.110.1. Scythian “feathers”: 4.31. Ethiopians speaking like bats: 4.183.4. Cf. Mora 1985.191. See above, pp. 40–41 and 25.

[ back ] 5. In Greek ainos may refer to any allusive utterance; in Nagy’s definition, it represents “a code that carries the right message for those who are qualified and the wrong message or messages for those who are unqualified” (Nagy 1990.148). Herodotus never uses the term, but the epea or logoi uttered by some of his characters bear a secondary meaning that makes them substantially equivalent to ainoi: see, for example, the story about the flute player and the fish that Cyrus tells the Ionians (1.141.1), and the other cases discussed by Hollmann 1998.116–125. Moreover, as Nagy has argued, the Herodotus narrator himself, though he employs the medium of historiē, also participates in the mode of the ainos vis-à-vis his audiences. Nagy 1990.215–338. Also Payen 1997.66–74; Munson 2001.5–8.

[ back ] 6. Malice of Herodotus 40 = Moralia 871D: “Thus Herodotus no longer picks Scythians or Persians or Egyptians to utter the sayings which he invents himself—as Aesop picks crows and apes …” (trans. L. Pearson). Nagy 1990.322 quotes this passage in his discussion of the generic connections between Herodotus and Aesop.

[ back ] 7. See above, p. 12 and note 28; pp. 35, 44, 48–49.

[ back ] 8. See e.g. Homeric Hymn III to Apollo 500, 517–519.

[ back ] 9. Persians 393, 406. Hall 1996.139. See also 605, where a κέλαδος οὐ παιώνιος (i.e., a cry antithetical to a paion, because barbarian) rings in the queen’s ears. Discussed by Broadhead 1960.159–160.

[ back ] 10. Timotheus Persians 158–159: Ἑλλάδ’ ἐμπλέκων Ἀσιάδι φωνάν. I am quoting the text as edited by Wilamowitz-Moellendorf, reprinted by Janssen 1984.

[ back ] 11. A barbarian scream in a battle context appears in the oracle reported at 9.43.3. For the ambivalence of Herodotus’ representation of the battle of Salamis, see Munson 1988; 2001.223–224.

[ back ] 12. For Herodotus’ attitude toward wars of aggression, see Munson 2001.211–217.

[ back ] 13. Cf. e.g. Polybius 1.67.9 on the difficulties of communication in the Carthaginian army (cited by Rotolo 1972.405).

[ back ] 14. 9.98.2. No such attempt appears in the parallel message of Themistocles (8.22.2). Cases of strategic exploitation of language difference in other sources include Thucydides 3.112, Aeschylus Libation Bearers 563–564, Liv. 1.27.9; cited by Rotolo 1972.402 and note 35.

[ back ] 15. 6.29.2. Foreign-speaking Greeks are remarkably rare also in other text, but see, e.g. Themistocles, who devoted a year studying Persian before presenting himself to the Great King (Thucydides 1.138.1, and see Gehman 1914.35 for other sources on this story).

[ back ] 16. For the convention, see Lejeune 1940–1948.51. E. Hall 1989.117–121.

[ back ] 17. 9.16.2. The specification agrees with the extreme precision of the source citation in this case.

[ back ] 18. See the words by which Croesus exonerates Adrastus (1.45.2) or Idanthyrsus challenges Darius (4.127.1–4). The introduction to the Constitutional Debate (3.80.1) communicates the narrator’s deliberate program to demonstrate to his audience that intellectual insights, moral principles, and values which contemporary ideology regarded as typically Greek are actually not the exclusive prerogative of Greeks. In Aeschylus’ Persians a barbarian has to be a ghost and, as Georges put it, “laundered” and universalized by death before he can speak with the moral consciousness of a Greek (Georges 1994.82, cf. 110.).

[ back ] 19. See e.g. the Persian difficulties in translating the non-verbal Scythian message of Darius. For oracles, cf. Heraclitus DK 93: “the god neither discloses nor hides his thought but indicates through signs (σημαίνει).” Hence the difficulty normally consists in recognizing a metaphor as such and decoding its correct referent (e.g. 1.55). Kirchberg 1965; Bernabò 1977. For this and other meanings of the verb σημαίνειν in Herodotus, see Hollmann 1998.10–16.

[ back ] 20. 1.29–34.1. The ambiguity of the concept of olbos etc. is discussed by Nagy 1990.274–282. For Croesus as the symbol of man bombarded with verbal communications he is unable to interpret, see Sebeok and Brady 1979.

[ back ] 21. For the role of negation in the ethnographies, see Munson 2001.148.

[ back ] 22. For ainos, see above, p. 69 and note 5.

[ back ] 23. 4.110–117. The gloss with the Scythian name Oiorpata has been quoted above, p. 38. On this episode, see, especially Dewald 1981, Hartog 1988.217–224, Flory 1987.108–113, Munson 2001a.123–132. Also Campos Daroca 1992.45, 47.

[ back ] 24. For gestures that cross cultural boundaries, see above, p. 22.

[ back ] 25. The word occurs only here in Herodotus and is not attested earlier. But L.S. s.v. cite the definition τῇ λέξει βαρβαρίζειν in Aristotle SE 165b20.

[ back ] 26. See Geham 1914.37–40, 43–46; Rotolo 1972.403–404.

[ back ] 27. Campos Daroca 1992.65–66.

[ back ] 28. These include royal scribes or secretaries (grammatistai), especially the ones who read aloud the letters bearing the orders of the king at 3.128.5. Hollmann, who draws attention to this passage, observes that “just as grammatistai are machines converting signs from one medium to another, translators (ἑρμηνεῖς) in Herodotus transfer signs from one language to another” (the quotation is from Hollmann, forthcoming; but see also Hollmann 1998.207–209 and 223–224).

[ back ] 29. See above, pp. 28–29.

[ back ] 30. 3.20–25, esp. 22.4. Longo 1987.19–20. Fehling 1987.100. Rosellini and Saïd 1978.

[ back ] 31. Van der Veen 1996.63, who also stresses the ambivalence of these consequences. Braund 1998.161–162.

[ back ] 32. 8.85. See above, p. 57. For interpreters in this episode as evidence of the presence of Greeks at the Persian court, see Lewis 1985, esp. 105.

[ back ] 33. Campos Daroca 1992.65.

[ back ] 34. The scene between Cambyses and the defeated Psammenitus (3.14) is structurally analogous to this one but lacks the dimension of understanding or not understanding the hidden message of an utterance. Accordingly, no attention is drawn to language, and the role of intermediary, which in the scene with Croesus is fulfilled by interpreters, is assigned to an angelos (‘messenger’), with no suggestion of linguistic translation (3.14.8).

[ back ] 35. Hollmann 1998.221–222. Cf. also the Telmessians’ interpretation of the snake-horse portent, that a host “of alien speech” (ἀλλόθροον) would come and overcome Croesus (1.78.3).

[ back ] 36. 1. 86.5, cf. 86.6. See Long 1987.111.

[ back ] 37. For discussion of this chapter, see Gigante 1956.109–113; Humphreys 1987, Burkert 1988.21–22; Thomas 2000.125–127; Munson 1991.57–63 and 2001.168–172.

[ back ] 38. The expressions in both cases reflect the quantitative mentality that Konstan 1987 describes as a characteristic of Persian kings.

[ back ] 39. Cf. Braund 1996.173. The fact that a Callatian Indian-to-Greek interpreter must have represented “surely a rare commodity” (Harrison 1998, ch. 2 “Herodotus’ presentation of foreign languages”) corroborates the argument that their presence is here symbolic.

[ back ] 40. For the meaning of tekmērion in this and other Herodotean passages, see Hollmann 1998.4–6.

[ back ] 41. Humphreys 1987.214. On onomastic correctness in Herodotus, see above, pp. 41–51. For a survey of the various meaning of nomos in different texts to the end of the fifth century, see Ostwald 1969.20–54. The distinction made here between Nomos and nomoi is that between men’s universal (one would almost say “natural”) impulse to regulate themselves in certain spheres of behavior (e.g. disposal of the dead) on the one hand and, on the other, the different forms of regulation (e.g. burial, cremation, etc.) that are produced by such an impulse in a given area of behavior in different societies. The gloss of corroboration of Pindar’s phrase with gloss of opinion (ὀρθῶς μοι δοκέει) compensates for ποιῆσαι, a term for the fictions of poets which Herodotus tends to use critically. See Svenbro 1977.210. The interpretive resonance of the reference would be richer to us if we knew more about the meaning of the phrase in its original Pindaric context (fr. 169 SM); see Schroeder 1917, Stier 1928, Gigante 1956, Ostwald 1965, Humphreys 1987.

[ back ] 42. See e.g. 1.199, 2.64, 4.62, 4.93 for a discourse marked variously by the narrator’s ambivalent reaction to cultural peculiarities in the areas we have mentioned.

[ back ] 43. See above, pp. 11–12.

[ back ] 44. E.g. 1.202.2 and 4.26.

[ back ] 45. See above, p. 77 and note 41.

[ back ] 46. On Herodotus’ use of the words ὁ θεός and τὸ θεῖον, see Linforth 1928.

[ back ] 47. For the oracles’ linguistic mediation in connection with their international status, see Campos Daroca 1992.70–73, who also recalls the role of Dodona in the transmission of the names of the gods (2.52). See above p.13. But, aside from the two examples in Herodotus, we know of no other Greek oracle that expressed itself in a non-Greek language. See Parke and Wormell 1956; Fontenrose 1978.

[ back ] 48. Iliad 2.867. See above, p. 2. There is no trace in Herodotus of Plutarch’s interpretation of the prodigy as signifying that the oracle wished to communicate that the Greek language was not at the service of barbarians (De defectu oraculorum 412a; cf. Aristides 19). Lejeune 1940.48–57.

[ back ] 49. See Moggi 1991.35, 40n29, where the examples of exceptional recognition of the role of foreign populations are all from Herodotus.

[ back ] 50. Libya is the land of the Atarantes, who have no personal names 4.184.2–3; of the troglodytes who speak like bats (4.183.4); of the encounter between Nasamones and pygmies (2.32.6); and of the silent transactions between Carthaginians and Libyans. Campos Daroca 1992.74, 128.

[ back ] 51. Pindar Pythian Odes 4.62: δυσθρόου φωνᾶς. Herodotus 4.155.1: ἰσχνόφωνος καὶ τραυλός. For Battus’ disability in the context of the morphological category of the hero, see Giangiulio 1981; Cosi 1987; Calame 1993.143–144. For a survey of the traditions about Cyrene, see Gentili 1990; Calame 1996.

[ back ] 52. Pythian Odes 4.61–62: αὐτομάτῳ κελάδῳ; cf. Pythian Odes 4.5–6. Herodotus 4.155.

[ back ] 53. Dougherty 1993.107–117; Calame 1996.116–128. Pythian Odes 5 and 4 both celebrate the victory of Arcesilaus’ brother in law in 462. The colonization themes in these odes refer back to Pindar’s earlier Cyrenaic ode, Pythian Odes 9 (474 BC), especially the section on the marriage of Apollo with the nymph Cyrene (5–69). Dougherty 1993.147.

[ back ] 54. For the anti-Battiad thrust of Herodotus’ narrative, see Chamoux 1953.207; Laronde 1990.35–37.

[ back ] 55. Pythian Odes 5.55–59; cf. the rationalized version of Paus. 10.5.7. Dougherty 1993.107. Pindar’s episode about Battus overcoming the lions with his voice recontextualizes the mythical tradition of the nymph Cyrene wrestling a lion barehanded (Pythian Odes 9.26–28). Cosi 1987.132–133; Dougherty 1992.147.

[ back ] 56. At 8.30 a Delphic oracle to Arcesilaus reproaches the kings of Cyrene for causing the anger of the gods because they did not rule in the manner of Battus.

[ back ] 57. The model of the unambitious leader is Leonidas, who did not expect to become king (7.205.1). For the deformity of Battus in the context of traditional stories of colonization, see Giangiulio 1981; Vernant 1981; Cosi 1987.

[ back ] 58. 4.159.1. For evidence, outside of Herodotus, of Battus’ wars with the natives and buildings of other settlements, see Malkin 1994.174 and note 13.

[ back ] 59. Calame 1993.141. Scholars currently tend to think that the native populations played a more important role in Cyrenaica than in other areas colonized by the Greeks. Laronde 1990.48. Giannini 1990.70–71.

[ back ] 60. Dougherty 1993 sees an allusion to the Libyan meaning in Pindar’s “automatic” oracle, which is similar to the oracle reported by Herodotus (Pythian Odes 4.59–63). However, Herodotus’ explicit translation is unique in early sources (it occurs later in a scholium on Pindar, Drachmann 1984.2:93, and Hesychius s.v. Βάττος), and he presents it as not generally known.

[ back ] 61. Stephanus of Byzantium s.v. Γέλα; Etymologicum Magnum 225. Parke and Wormell 1956.2.166. On bilingual oracles, see Dougherty 1992.45–48, with this and other examples.

[ back ] 62. Herodotus does not mention Battus’ alternative name, Aristoteles, known from Pindar (Pythian Odes 5.87) and other sources.

[ back ] 63. Campos Daroca 1992.128. Calame 1996.136.

[ back ] 64. Cosi 1987.121,130.

[ back ] 65. In this case, from our point of view, they were apparently right: see Masson 1976, who argues that the name Battus appears in different parts of Greece, is really Greek and derives from βατταρίζω.