Cornish’s description points to the importance of the cognitive processes involved in the production and processing of discourse. The textual antecedent problem is one of the factors that have led to a cognitive approach to reference.

It is important to be aware that the labels of “given” and “new” in this framework do not relate to the status of the referent in the speaker or hearer’s knowledge. In fact, (t3) reads most naturally when we assume that both speaker and hearer know who “Sam” refers to. The newness or givenness is rather a status relative to the ongoing discourse, which has led people to prefer the terms discourse-old and discourse-new. However, even this terminology is insufficient for explaining the forms of referential expressions used in naturally occurring discourse. Most obviously problematic are referents that are introduced into the discourse, then not mentioned for a certain span of text, and then retrieved. Although they are discourse-old (i.e. “given”), they cannot generally be retrieved by an unaccented pronoun.

This description of the process of reference has been widely taken up, as witnessed by the following relatively recent quotes:

Gundel, Hedberg, and Zacharsky focus on the presuppositions inherent in referential expressions:

What all authors agree on [20] is that anaphora is not to be understood as a relation between a referential expression and its antecedent in the co-text, but as an instruction to the hearer to focus on a certain referent in his discourse model. It is not primarily a process of verbal memory, but of cognitive focus within the mental representation of discourse. [21] As a result, the accessibility of a referent is its status in the current representation of the discourse: regardless of whether it has been mentioned before and how long ago, the referent’s current status as more or less in focus determines the anaphoric expression used.

Since we have two female referents in this narrative the “she” in the second sentence is underdetermined at first sight. However, we have no problem interpreting the line since we visualize the scene and imagine the victim falling, rather than the aggressor. From Homer, we might compare the moment of Patroclus’ death by the hands of Hector:

In English we find the unaccented personal pronoun “he,” where in Greek we find the equivalent least marked reference, null anaphor, within the third-person singular verb (δούπησεν). Despite the change of grammatical subject (from Hector to Patroclus), the lack of an expressed new subject does not lead to confusion, for three reasons. First, the semantics of πεσών makes Patroclus the logical subject, since he has just been wounded. Second, δούπησεν δὲ πεσών is a formula that always refers to the stricken hero, so the audience’s knowledge of epic will prevent any potential ambiguity of reference. [23] Third, the form of the formula in itself suggests that not just “the referent Patroclus” is in focus, but the entire image of his fall. The finite verb does not strictly refer to the noise Patroclus makes, but to that of his armor and weapons. [24] The Homeric performer can present this complete, synaesthetic image, since from the moment that he is stabbed and wounded (line 821), Patroclus is in focus in the mind’s eye. It is the idea of “focus” that explains the two examples above: in both cases the mind’s eye inevitably moves to the recipient of the attack, putting them in mental focus, and thus making them accessible. [25] In earlier work, Chafe speaks of the “subject of consciousness” in terms of attention, “the mechanism by which the spotlight of consciousness is directed at one or another area of the material accessible to the mind.” [26] A higher accessibility leads to a lower “activation cost,” in Chafe’s terms, which in turn translates to a less specific referring expression. [27] The focus of the mind’s eye is an important factor in the process of anaphoric reference. It is especially relevant in passive constructions, where there is a clear distinction between subject/object and patient/agent. In Homer and Pindar, moreover, the grammatical subject can be a part of the body or a hero’s weapon or armor, as it might be here, while the subject of consciousness always remains the character (see also t13 below).

An alternative reading of the final line is: “…by far the best. He knew..,” if we interpret the pronoun as demonstrative rather than relative. [37] In (t10) it is perhaps unnecessary to choose a demonstrative reading, but consider this parallel:

Here, Murray prefers to read the pronoun as demonstrative (rather than relative), and translates accordingly. In the language there is no formal distinction between the two possible readings of the pronoun, which suggests that Murray was guided by the context.

In this passage the narrator describes the fight between Paris (Alexander) and Menelaus. For this narrative stretch, those two characters are the only relevant referents, and the focus of the mind’s eye shifts steadily from one to the other. In lines 346 and 347, agency remains with Paris and as a result the new act in 347 is introduced without a pronoun (∅ βάλεν). [45] A few lines later we have followed the thrown spear to its mark, Menelaus, who has now become focused in the mind’s eye. As Tomlin explains, the nominative case is a grammatical reflex of attention management: in English the nominative typically refers to the referent that has become the focus of attention in the directly preceding discourse. [46] However, in instances like (t12) the new grammatical subject (Menelaus) is in focus in the mind’s eye, but has not yet been primed in the language when he is activated with ὁ δέ in 349. This is the reason that we find the pronoun ὁ, and not just the verb. [47] Contrary to our intuition, then, ὁ is a marked rather than an unmarked anaphoric expression and should be taken as such. [48]

The audience cannot but picture the scene of Achilles tearing through the enemy ranks like a forest fire spurred on by the wind, trampling their bodies like grain on a threshing floor. In the final lines the narrator sketches an image of Achilles triumphant on a chariot spattered with blood, riding over the bodies of his adversaries. Although he has not been the grammatical subject over the last eight lines, and has not been named in four lines, ὁ δέ suffices to restore Achilles as grammatical subject in 502. The images evoked are strong, but unlike some other similes they apply readily to the current situation on the battlefield. In 498 the horses are the subject (ἵπποι) and the axle (ἄξων) in 499, but Achilles is constantly at the forefront of our mind, literally in the center of the image, in focus. As in the case of Jason below (t32), ὁ in 502 does not just retrieve “Achilles” but it retrieves the raging and bloody Achilles that has just been created in the mental representation of the discourse.

After the first plural form περισχόμεθα, the singular verb form ᾤκει suffices to avoid ambiguity. It might seem all the more surprising, then, that the pronoun is used in the following act, even though there is total continuity of grammatical subject.

In this passage ὁ δέ is problematic, since it would most naturally establish Achilles as the grammatical subject of the new act, instead of continuing to refer to Poseidon as it actually does. This apparent mismatch may have been Aristarchus’ reason for athetizing lines 322-324, since taking out these lines creates an attractive symmetry between τῷ μέν (sc. Achilles, 321) and Αἰνείαν δέ (325). [63] As we follow the reading of the manuscripts and editions, however, this requires an explanation. In II.2.5 we devote some attention to small fronted acts that serve to guide the mind’s eye of the audience. These priming acts take the form of a referential expression + a particle (often δέ), and are followed by some performative discontinuity. Even though nothing linguistically suggests that there is a boundary after δέ (the accusative that follows seems in no way to be independent), the manuscripts suggest that some kind of discontinuity was assumed after ὁ δέ. [64] Here, the reason for such a discourse act is not obvious, but the motivation may be visual. After his arrival Poseidon sheds mist over Achilles’ eyes, which leads the mind’s eye to focus on Achilles. The expectation of the audience might have been that focus and agency stayed with Achilles, but in fact the scene moves back to Poseidon with ὁ δέ (322). The god then first enacts a ritual return of the spear to the hero, after which he performs a truly awesome deed: he throws Aeneas over the entire army to the other side of the battlefield. The scene is climactic and highly vivid, [65] and the use of the pronoun rather than a null anaphor may serve to prepare for the upcoming image that has Poseidon as its center.

Here and in the parallels, the sense of the act introduced by ὁ δέ is gnomic. [73] In all cases I read the first part of the thought as a copulative construction with ἐστί left out, which means that ὁ must be read as a demonstrative pronoun rather than as an article. Because in this context ὁ δέ introduces a gnṓmē – unframed, generalizing discourse – there is always a change of grammatical subject. However, in gnômai the referent of ὁ can be ambiguous; either it can be the indefinite “he”, or it can be the main referent of the preceding discourse, often the victor.

The text illustrates neatly how problematic generalizations about ὁ δέ and ὅ γε in Homer are: ὅ γε in 190 marks change of grammatical subject from ἦτορ, “heart,” to Achilles, whereas ὁ δέ in 191 accompanies subject continuity. [79] The choice of the pronoun over null anaphor is expected in 190, since null anaphor would have led to the jarring image of the heart drawing a sword. In 191 the anaphoric pronoun serves to juxtapose the image of the assembly dispersing with the image of Achilles and Agamemnon staying and fighting. This explains the use of the pronoun in both instances, but the question remains what the function of γε is in ὅ γε.

The passage seems straightforward, but on closer inspection it involves an interesting shift. Whereas in the majority of instances of ἀλλ᾽ ὅ γε there is a single referent who functions as the agent in both parts of the construction, here the tracking of referents is more complicated. [86] The initial assertion is that “no-one sails by before hearing the Sirens,” so what does ὅ γε refer to? To no-one? The participle τερψάμενος makes it clear that who is meant by ὅ γε is the sailor who did indeed stay and listen, and was pleased as a result. The automatic inference triggered by “no-one sails by” creates a pool of available referents who did not sail by, one of whom can then be referred to with ὅ γε. [87] A similar example occurs in the following gnomic thought that Odysseus imparts to the suitors:

Again, an unproblematic reading belies the underlying referential complexity. As in (t18) ὅ γε directs attention to the indefinite referent that we infer from the preceding act. [88] Alternatively, one might say that ὅ γε triggers the creation of a generic referent, about which the only inferrable information at the point of utterance is that he has not “been a lawless man.” It is this cognitive action, I propose, that justifies the use of ὅ γε over null anaphor or the pronoun only. [89]

Before we look at the two instances of ὅ γε, consider the use of μιν in line 181. μιν is an unstressed (enclitic) third-person pronoun in the accusative that serves to retrieve accessible referents. Pandarus’ speech starts in line 180, and he has not mentioned the referent of μιν (“that man”) up to this point. However, in the preceding turn Aeneas has already pointed the man out (τῷδ᾽…ἀνδρί 174). The referent of the strong demonstrative ὅδε is the man they both see, and Pandarus retrieves him with the much less emphatic μιν, since by now he is well established in both their mental discourse models (as well as in that of performer and audience, of course).

The situation here is different from most preceding examples, since the two available referents have both been mentioned only in an oblique case. Neither a null anaphor nor a pronoun suffices to retrieve Irus or the stranger. In the earlier narrative, however, it was told that the “stranger” was indeed the stronger, and in fact the whole co-text suggests that the suitors backed Irus. As a result, the referential ambiguity is resolved by the time that φέρτερος was uttered. [93] There is yet another layer to this reference. In the mind of Telemachus, who is speaking, and in the minds of performer and audience, ὅ γε refers not to a stranger but to Odysseus. The reference means something different to the internal audience (Penelope and the suitors) than to the performer and his audience, and this layering has its effect on the content. Bonifazi has demonstrated convincingly that references to the disguised Odysseus are particularly sophisticated in this part of the Odyssey, and there is no doubt in my mind that this instance is another example of this complexity. [94] The performer uses ὅ γε because the referent – and the sense of Telemachus’ utterance – can only be grasped fully if the listener makes the necessary inferences. It may be surprising to Penelope and the suitors that the stranger beat Irus, but this is decidedly not the case for Telemachus or the audience: obviously Odysseus is stronger in might than the resident beggar of his own palace.

The addition of γε in this instance may not appear to be needed, since Telemachus is surely constantly in focus as he is being washed and clothed. His continued prominence in the mental representation of discourse is confirmed by the fact that he is retrieved in line 468 as the unexpressed subject of βῆ. Since Telemachus remains the subject, ὅ γε looks marked in 469. I propose that we read the overdetermination here as an instruction to visualize not just Telemachus, but Telemachus as he is after the ablutions described in the preceding lines. The instruction is perhaps anticipated by the description δέμας ἀθανάτοισιν ὁμοῖος (469). It is not just Telemachus who joins Nestor, but it is Telemachus looking like a god. The same expression is used of Odysseus when he arrives at the palace of the Phaeacians and after he has been washed by Eurynome in book twenty-three. [95] Even more salient, Telemachus himself utters similar words when he mistakes his father for a god during the recognition scene in book sixteen. [96] The scene in book three thus reveals its importance on a macro-discursive level. The first books of the Odyssey are about how Telemachus finds himself in the role of man of the house. Perhaps the moment when he joins Nestor is the moment when he finally becomes a worthy heir to his father, and thus like his father looks “like a god.” If so, it really is a new Telemachus that sits down next to Nestor. [97]

ὅ γε creates the contrast: “Don’t listen to him, listen to me.” The referent of ὁ is highly accessible, since it refers to the last speaker (Telemachus), to whom both current interlocutors have presumably been listening. [100] Thus he is part of the speech situation, visually accessible, and in focus in the discourse model. [101]

In the little narrative about Tlepolemus we hear how he kills his uncle, and then starts building ships: he has to flee, because the other children of Heracles are on his tail. The contrast is made explicit by the adjective ἄλλοι, which justifies the use of ὅ γε in 664. As (t24) demonstrates, contrastiveness is a factor that functions separately from activation cost: at that point Tlepolemus is clearly accessible. [103]

The passage is far from straightforward, but in line 409 ποιήσεται retrieves Paris, who is clearly very accessible at that point (κεῖνον, ἑ 408). Since the construction changes from imperative to indicative in 409, the third person verb form suffices to disambiguate among the three available referents: I (Helen), you (Aphrodite), and he (Paris). Then follows ὅ γε in the second part of the disjunction. The occurrence of ὅ γε in either part of a disjunctive construction is well documented, but little discussed. [104] Contrary to expectation, ὅ γε in disjunctions is not used to juxtapose two referents, but rather two possible events involving the same referent. [105] Here the two items in disjunction, ἄλοχον and δούλην, are syntactically symmetrical and semantic opposites. However, the addition of ὅ γε to the second suggests that the expression ἢ ὅ γε δούλην must be read with emphasis placed on ὁ (“or he makes you a slave”). The opposite of this statement is not “until he makes you a concubine” [106] (this is the opposite of “until he makes you a slave”) but rather of “you make him a slave.” It is exactly this presupposition that is triggered through the use of ὅ γε. [107] The expected situation in Greek culture is that the goddess of love makes a human her slave, not the other way around. [108] The markedness of the (hypothetical) situation is brought out by the combination of the pronoun and the particle. [109]

After the story of the scepter, the current frame is recalled with τῷ ὅ γε, the first pronoun retrieving the scepter, and the second guiding attention toward Agamemnon. [110] The latter had been out of focus for several verses, and even though his name is mentioned in the oblique in line 107, this is not actually the referent of ὅ γε. Rather, the act containing ὅ γε re-establishes the frame of Agamemnon in the council, and that is why the pronoun is used rather than a simple verb form. There is, however, no need for further emphasis: γε is there only to accompany the pronoun in its non-initial position. Compare the following example from the Odyssey:

The co-text is essentially the same as in (t27): a speaker is introduced, then follows a brief piece of unframed discourse (γάρ), [111] and finally a pronoun referring to the unframed discourse (τοῦ) and a second pronoun with γε, in the act that re-establishes the narrative frame.

Since this verse follows direct speech, subject continuity is implicit. It is not hard to see that in this construction ὅ γε serves to set up a light contrast between this speaker and the next.

The reference to Alcinous here serves at once as the climax of the imagined entrance into the palace, and as the beginning of the long episode in which Alcinous and Odysseus are the main characters. At this crucial moment Alcinous is granted agency with some emphasis, which prepares the audience for his importance in the upcoming narrative.

There is clear subject continuity from line 393 onward, yet after two verbs with null subjects (ἐνόησεν, λήθετο) we find ἀλλ᾽ ὅ γε. There is a significant difference between the sense of the verbs before and after the pronoun and particle combination, marking the contrast between inaction and action.

After a gnomic passage, a demonstrative pronoun and a full noun phrase serve to retrieve Agamemnon (αὐτὸς ἥρως Ἀτρεΐδας 31). [131] Although he had been mentioned in the oblique before (φονευομένου πατρός 17, σὺν Ἀγαμεμνονία 20), he had not yet been fully attended to, only mentioned as part of an event relevant to Orestes. Then Pindar narrates a selection of Agamemnon’s adventures in reverse: he elliptically retrieves the story of Agamemnon’s taking of Priam’s daughter Cassandra, and then bringing her to Mycenae where she too is killed by Clytaemnestra (in Pindar’s version, lines 19-21) with the four words μάντιν τ’ ὄλεσσε κόραν. The particle τε here, I submit, serves to mark the sharedness of this episode, and allows Pindar to waste no more words in telling it. [132] The mention of Agamemnon’s death activates the event of line 16 again in the hearer’s mind. This makes Orestes, who has not been named or even mentioned for seventeen lines, accessible enough to be referred to with ὁ. In the preceding lines Agamemnon has been the grammatical subject, so the pronoun suggests a change of subject; the only other available masculine singular referent is Orestes. [133] The possible ambiguity in the anaphoric expression is perhaps the reason for the apposition νέα κεφαλά in line 35, [134] since this can only refer to Orestes. This passage demonstrates that ἄρα functions on the level of the larger discourse or interaction rather than as a link between two contiguous clauses. In narrative terms, ἄρα recalls the frame of the main storyline: lines 32-34 function as a little flashback, told regressively, and δ᾽ ἄρα serves to re-activate the main narrative frame of Orestes’ story. At the same time ἄρα marks its host act, as well as the upcoming narrative, as a part of the tradition shared between Pindar and his audience.

I read ἆρα rather than the ἦρα proposed by Schroeder and printed in most editions. [136] Braswell regards ἆρα as a prosodically enriched form of ἄρα, and I would add that it may be regarded as pragmatically enriched as well. Again, ὁ δ᾽ ἆρα marks the recall of the main narrative frame, after the explanation of the prophecy. The “one-sandaled man” of the prophecy is Jason, as Pindar’s audience would have known. What makes this instance different from (t31) is that here there is apparent continuity of grammatical subject, so the use of the nominative pronoun must be explained otherwise.

I discuss the use of ὅ ῥα to introduce unframed discourse, as in line 380, in §§58-59 below. For the present, I focus on the instances of ὁ δ᾽ ἄρα in lines 383 and 385. In 383 ὁ δ᾽ ἄρα retrieves Ajax, last named in 378, who picks up a stone no normal man could have lifted. The referential expression comes after the description of the rock, and ὁ δ’ ἄρα serves to retrieve the narrative frame. [141] After the image of Ajax throwing the rock at someone, his victim is made an agent by ὁ in 385, while what happens to him (he tumbles off the wall head first) is marked by ἄρα as the expected outcome. In addition, the next words (ἀρνευτῆρι ἐοικώς) suggest that the hearer is invited to imagine the scene by superimposing upon it the (shared) image of a diver. The vivid nature of the acts introduced by ὁ δ᾽ ἄρα is paralleled elsewhere in Homer, and it may be linked to Bakker’s reading of ἄρα as an “evidential” particle. [142]

ὅ(ς) ῥα introducing unframed discourse serves to recall little bits of information about the referent which the speaker expects to already be part of (or inferrable from) the discourse memory. [151] In line 643 the next assailant of Menelaus enters the scene, Harpalion son of Pylaemenes. After the name is stated in line 644 we find ὅ ῥα and a verb in the imperfect: the pronoun and particle introduce a piece of unframed discourse that consists of a flashback and a flash-forward. In this instance, one can just as well translate “Harpalion. He followed”; either way the act introduced by ὅ ῥα is unframed. Both the flashback and the flash-forward (οὐδ᾽…γαῖαν 645) are unframed discourse in the sense that the performer reveals his omniscience and informs the audience outside the frame. The use of ἄρα in unframed discourse marks a piece of knowledge from the discourse memory that is retrieved to become part of the current discourse model: it may be regarded as activating a piece of information in the long-term memory to become part of the working memory.

My translation and punctuation reflect a reading of line 254 as unframed, despite the presence of the aorist. Two things support this reading: (1) ἀποναίω is not attested in the imperfect, and (2) there is a parallel passage in the Iliad. Compare this passage from the Catalogue of Ships, from the entry about the people from Doulichion:

It is clear that the resemblance between these two passages reaches beyond the repetition of the verb. [154] For the current purpose, it suffices to note that the information about Phyleus moving from Elis to Doulichion is necessary, yet clearly out of the current frame, as the use of ποτε confirms. I believe this justifies a similar reading of the passage from the Odyssey (t35): ὅς ῥα in line 254 introduces a piece of unframed discourse containing shared knowledge. [155] The passage preceding (t35) is a genealogy leading up to Polypheides and eventually his son, Theoclymenus. At this point in the genealogy, the family is in Argos, but it appears that the performer knows that Polypheides is connected to Hyperesia rather than to Argos. That would explain why he adds the line, with ὅς ῥα, to avoid a discrepancy between his performance and shared tradition.

The Trojans are at the wall, and Ajax does all he can to get the Greeks to fight back and keep them from the ships. His rhetorical question has a note of incredulity, and in the next utterance he vehemently adds some information that his audience apparently does not yet know. Although I do not see δή as directly connected to what is evident and visible, I do believe that some of the utterance’s tone of insistence comes from the implication that the speaker directly perceives what he is claiming. [162]

This instance, like the majority of ὅς/ὁ δή instances, occurs in direct speech. Here Odysseus introduces a new character, a Phoenician (288), about whom he introduces several descriptions that his audience cannot yet know. One aspect of the “newness” inherent in relative clauses introduced by pronoun + δή is that it generally contains an expression of stance, a personal judgment or feeling of the speaker. [165]

Aeneas’ intensified superlatives serve to strengthen his boast about his forefathers. The second of the two examples shows that the link between relative clauses introduced by δή and “new information” is relative. There can be no doubt that both the internal audience (Achilles) and the audience at the performance was expected to know that Ganymedes was the most beautiful man of all. However, the presence of δή here, rather than ἄρα or τε, presents the statement not so much as a foregone conclusion, but as an expression of a personal opinion, with a clear rhetorical goal.

Here, as in all the parallels, [166] the intensified expression occurs in direct speech. In all instances, the scope of δή is ambiguous, but here at least there seems to be a reference to a specific earlier scene (335-341). Eumaeus and the herald sent by Telemachus arrive at the palace at the same time, and the herald gives his news first (337). It is for this reason, I believe, that δή intensifies πρῶτος here: to render precisely what happened. [167]

This is one of only two instances of δή τοι in narrator text, and one of the relatively few instances in Homer where τοι is clearly the particle rather than the dative second person pronoun. Just as in (t39), this piece of discourse introduces a character. [168] The difference is that the introduction here first mentions the father of the character in focus. The performer brings the attention back to Dolon with the priming act ὅς δή τοι. [169] The priming act also projects Dolon’s importance in the upcoming long episode (316-457). [170] The presence of the particle τοι especially suggests that there may be one more factor at work in this passage. It is as if the performer turns to the audience and speaks to them directly, inviting them to imagine this Dolon, and to share the performer’s opinion of him. Since it occurs in the priming act, I believe that δή may be regarded as having scope over the entire line, the entire description of this antagonist.

The use of γάρ after pronouns in both Homer and Pindar often introduces unframed discourse, as here. [171]  γάρ introduces a fact about Lynkeus’ sight, which is triggered by association with the preceding narrative. The difference between unframed discourse introduced by δή and γάρ is that δή typically introduces a personal judgment. This does not mean that γάρ cannot introduce a personal judgment, as δὴ γάρ in Homer illustrates (see II.4 §20). Unframed discourse introduced by ἄρα, conversely, is expressly presented as shared between performer and audience or between speaker and internal audience. The difference between ὅς ῥα on the one hand and ὁ/ὃς δή and ὁ/ὃς γάρ on the other may be represented quite well with the paraphrase “who of course” for ὅς ῥα and “who actually” for ὁ/ὃς δή and ὁ/ὃς γάρ.

5.5 Conclusions

Footnotes