Chapter 1. Peripheral and Nuclear Semantics

In recent years a number of studies on Homeric versification have appeared which aim to show a way out of the deadlock at which the study of the Homeric formula ended in the ’60s and ’70s. [1] Milman Parry’s analysis of the systematic precision in the use of noun-epithet formulas, as presented in his 1928 dissertation, has been found convincing by most scholars in the English-speaking world, yet the ways which Parry showed for the application of his formula concept to the whole of the Homeric poems has not met with general consent. [2] There were problems with the definition of “formula,” and questions as to the production of Homeric verse in actual performance. It was one thing to speak of a “grammar” of poetry, [3] a grammar necessitated by the verse, but quite another thing to show how it worked. Somehow the impression was created that an oral poet simply had to “know” a great many formulas in order to be able to compose epic verse. And the tacit equation of the “grammar” of formulaic poetry with the rules and regulations of the epic verse begged the question as to the differentiation of oral versification from written versification—this was one of the first and most important objections to Parry’s theory in the ’60s, which need not be restated here. [4]

The recent publications converge in one crucial respect. They abandon the conception of the formula as a ready-made phrase tout court in favor of a view in which the formulaic in Homeric diction is subservient to the singer’s communicative intention: Homeric diction is formulaic in that it provides the singer standardized, indeed, grammatical, ways of adapting what he wants to say to the conditions in the verse. This conception implies a distinction in the diction between communicatively and poetically important linguistic material on the one hand and on the other hand the adaptation of this material to the metrical context. In this chapter I will speak of material that is “peripheral” to a “nucleus.” [5]

This distinction amounts in fact to a reformulation of Parry’s own conception of a formulaic system, in particular with respect to the twin notions of “length” and “thrift” (or “economy”). The former referred to the number of formulas for a single “essential idea” and the latter to the metrical uniqueness of each component member of such a formulaic system: [6] economy, and hence formulaicness, in this systematic sense would be restricted to the set of peripheral elements clustering around certain semantic nuclei.

A further important point of this approach is that it makes the specific nature of oral poetry with respect to written hexameter poetry very clear: while oral, spontaneous versification does not differ from written, planned versification in the metrical localization of phrases that convey poetic or narrative meaning—both oral and literate hexameter poets have to observe the same positive and negative metrical factors [7] —it does visibly differ from written versification in the degree to which it makes systematic use of flexible, metrically adaptable material.

There is much in Homeric formulaic diction that is “peripheral.” The principal example is the way in which an epithet can be described as peripheral to a nucleus (the noun or name). Parry himself, in his French thesis, writes about the relation between epithets and their nouns or names in this way: one out of a number of epithets can be added to a given name, depending on the metrical circumstances. [8] But later he comes to describe the epithet as being indissolubly linked with its name, meaning that there are as many formulas for “Odysseus” as there are noun-epithet-combinations for this hero. [9] It is the neglect of the distinction between “fixed” and “flexible” for the name and the epithet, respectively, that caused the problems mentioned at the beginning of this chapter: duplicating the formulas for any single concept raises the question as to how many formulas a poet must know in order to tell the epic story.

Parry’s earlier distinction between “fixed” and flexible,” rephrased as “nucleus” and “periphery,” is where the new publications find common ground. Thomas Jahn analyzes the very frequent expressions for “in his heart,” consisting of prepositional phrases with, or inflected forms of, such lexemes as θυμός, φρήν/φρένες, ἦτορ, κραδίη, πραπίδες, κῆρ, and στῆθος. He describes them as peripheral elements to a nucleus that consists of verbs of feeling or thinking. [10] I myself have discussed the peripheral extensions of the well-known concessive participial phrases modified by the particle περ (e.g. ἀχνύμενός περ ‘grieved though he was’) and have shown that the distribution of the satellite particles καί, μάλα, and ἔμπης is entirely in service of the automatic adaptation of the participial phrase to the metrical circumstances. [11]

Finally, Edzard Visser argues that the distinction between nucleus and periphery may be applied as well to the verse as a whole. The main tenet of Visser’s illuminating work is that the typical Homeric verse does not consist of the formulaic building blocks we are familiar with in Parry’s model. Rather, he claims that a Homeric verse is a combination of “determinant” material, whose metrical form is an active factor in the localization, and “reacting” material, which is dependent in its metrical form and localization on the determinant material. Visser shows that in verses containing the statement “A killed B” normally the names of the victor and the victim are the metrical determinants: as such they have “priority” in the localization. The verb “(he) killed,” on the other hand, is a flexible and “reacting” element: its form, lexical articulation, and localization depend on the form and localization of the two other elements. [12]

Together these studies point to “peripherality” as an essential and structural property of Greek epic diction. And this basic insight suggests an obvious line of research: to investigate which further part of Homer’s diction can be characterized as peripheral material with respect to which nucleus. In this chapter, I address, by way of case study, the dative expressions for “spear” in this light, seeing whether they can be described as peripheral, extending material with respect to verbs denoting killing or wounding. Also, I will be concerned with the conditions under which a given element may be called “peripheral” in Homer and the conditions under which it may not.

The following passage may serve as our starting-point: [13]

This passage consists of seven factual statements of the type “A killed B.” Verses in which this kind of simple assertion is made form the main subject of Visser’s (1987) study of Homeric versification. In his discussion of line 32, Visser (1987:80–82) states that the verse-final expression δουρὶ φαεινῷ ‘with the shining spear’ is the weakest element in the verse, being a mere verse-filler which bridges the open metrical space left after the verb, the space between the bucolic diaeresis and the end of the line. [14]

In this chapter I will elaborate on this point, analyzing δουρὶ φαεινῷ and expressions for “with the (his) spear” in general from the point of view of their verse-technical function. I will argue that very often these expressions are not uttered to convey what they mean by their lexical value. Rather, they are uttered to adapt the verb of killing or wounding to its metrical context, by giving it the appropriate length. The fact that the nucleus is a verb of killing or wounding puts contextual constraints on the functioning of this formulaic “system”: it is confined to contexts that are concerned with androktasiai ‘killings in battle’. I will return to this point below.

There are two ways for dative expressions for “spear” to have a verse-technical function. First, they may be a peripheral element to a nucleus, which is constituted by a verb denoting killing or wounding. This means that expressions for “with the spear” are related to the verb in the same way as epithets to their name or noun, or as Jahn’s expressions for “in his heart” to a verb of thinking or feeling. This is the appropriate characterization, I suggest, of δουρὶ φαεινῷ at Iliad VI 32 in the extract above: the phrase is a peripheral element to ἐνήρατο ‘he killed’, giving this verb the length needed to fill the verse. Second, “with the spear” may have a versifying function without being immediately added to a nuclear verb. This typically happens when the verb is located in another verse. This can be observed in line 31 in the extract, where ἔγχεϊ χαλκείῳ ‘with the bronze spear’ fills the first half of the verse in a situation where the second half is to be filled by the metrical determinants, the names of the victor and his victim. [15]

The peripheral status of an expression in Homeric diction entails two important properties. [16] Peripheral elements have to be (i) neutral with respect to their context, and (ii) metrically variable. I will now go on to review these properties (sections 1 and 2) as a prelude to a discussion of the meaning of peripheral elements and of ornamental adjectives in particular (section 3).

Peripheral elements are semantically neutral in that they may just be present or absent without any difference for the intended meaning of the combination nucleus-periphery. This is the logical consequence of the notion of peripherality: a peripheral element is peripheral precisely because it may be absent without leaving a semantic gap. And when it is present, it serves primarily a verse-technical, rather than a semantic role.

But this neutrality with respect to context does not mean that peripheral elements are meaningless. To deny that a peripheral element has any meaning of its own, as Parry did in his later publications, is to take too strong a position. Parry’s treatment of the semantics of what I am calling peripheral elements has invoked, understandably, considerable reaction from scholars who claimed the contrary. [17] Epithets and other peripheral elements do indeed have meaning; they have a sense that is very often very appropriately (poetically) in accordance with the meaning of the nucleus to which they are attached (see below). But it is still a meaning that is subservient to the ultimate goal for which they are used, which is the metrical extension of their nucleus. This is why the meaning of any peripheral element is intrinsically “innocuous”: if its presence or absence would matter in any way, the element in question would cease to be a useful peripheral element.

Neutrality with respect to context can be observed in the extract cited above: ἔγχεϊ χαλκείῳ ‘with the bronze spear’ and δουρὶ φαεινῷ ‘with the shining spear’ are used in the expressions reporting the killings of Pidytes by Odysseus and of Ablerus by Antilochus, respectively. They are not meant as descriptive details distinguishing these particular killings from the other killings in the list. To be killed in the Iliad often means to be killed by the thrust or the throw of a spear. [18] This means that spears may be present even when they are not mentioned, and that when they are mentioned, they need not carry the bulk of the descriptive load of the scene. Their occurrence in the description of a killing has the typical innocuous quality of peripheral elements.

The function of a peripheral element not only hinges on its neutrality with respect to context, but also on its variable metrical form, which is what Parry called “length” (see above): if the verse-filling and extending function of a peripheral element is to be fully productive in the diction, the peripheral element has to be able to fill any incident metrical slot in an automatic way. Consequently, diversity of metrical form can be seen as an index of peripherality. The diversity can be achieved by a number of means, each of which is in its own right a salient feature of the Homeric diction. We may mention (i) morphological and/or dialectal diversity, (ii) the addition of optional (“peripheral”) elements, and (iii) synonymy.

The first of these is the basic ingredient of Witte’s and Meister’s notion of Kunstsprache, [19] of which Parry discovered the functional motivation. Morphological and/or dialectal diversity (e.g., in the dative plural expressions for “ships”: νηυσί beside νήεσσι and the artificial form νέεσσι) is at least partly motivated by the need for metrically diverse and semantically interchangeable forms. [20] In fact, the dative expressions for “ships” can be said to function, as part of prepositional or instrumental phrases, as peripheral system to verbs denoting location or displacement of the Greek besiegers of Troy. [21]

The second means to effect metrical diversity implies that a peripheral element may consist of a nucleus and a periphery itself (as in the case of δουρὶ φαεινῷ, where the epithet φαεινῷ ‘shining’ functions as periphery to the nucleus δουρί, or in the case of the ships just mentioned, where the dative phrase for “ships” can be modified itself by a number of epithets. Peripherality is thus a recursive affair: it applies within expressions that are as a whole peripheral to something else. The addition of epithets will be further discussed in section 3 below.

The third factor, synonymy, means that the very frequent phenomenon in Homer of the existence of various lexemes with the same meaning is not just a matter of poetic style; synonymy in Homeric diction is motivated by the need of different metrical forms for one single semantic concept. A good example of metrically motivated synonymy in Homer is the large number of verbs meaning “to kill” (Visser 1987:67–79), which reflects the non-determinant (reacting) status of the verb in verses reporting a killing.

In the case of dative expressions for “with the spear,” synonymy and the concomitant metrical diversity lies in the co-occurrence of the lexemes δόρυ (δουρὶ) and ἔγχος (ἔγχεϊ or ἔγχει). [22] Each of these can be combined with its own epithets. The functional synonymy of δόρυ and ἔγχος appears from the fact that both lexemes may be used “co-referentially” (referring to one and the same object in a single description), of which the following extract is one of the numerous examples: [23]

Together with their epithets, δουρὶ and ἔγχεϊ/ἔγχει yield the following list, which we may call the “system”: [24]

ἔγχει	V
δουρί	C ⎯ ∪
ἔγχεϊ	V ⎯ ∪ ∪
ὀξέϊ δουρί	V ⎯ ∪ ∪ ⎯ ∪
ἔγχεϊ μακρῷ	V ⎯ ∪ ∪ ⎯ ⎯
δουρὶ φαεινῷ	C ⎯ ∪ ∪ ⎯ ⎯
ἔγχεϊ χαλκείῳ	⎯ ∪ ∪ ⎯ ⎯ ⎯
ἔγχεϊ ὀξυόεντι	⎯ ∪ ∪ ⎯ ∪ ∪
	⎯ ∪
χαλκήρεϊ δουρί	⎯ ⎯ ∪ ∪ ⎯ ∪

Metrical and/or prosodic diversity as discussed in the previous section is greatly augmented when more than one peripheral element may be added to the nucleus. Thus in the system of the peripheral element “with the spear” just listed, we have δουρὶ φαεινῷ and ὀξέϊ δουρί beside χαλκήρεϊ δουρί, and ἔγχεϊ μακρῷ beside ἔγχεϊ χαλκείῳ and ἔγχεϊ ὀξυόεντι. The co-existence of various epithets to one nucleus leads us again to the question of the meaning of these elements. We already saw that peripheral elements are neutral with respect to their context (subsection 1 above); we may now say that they are mutually interchangeable as well. Like neutrality with respect to context, interchangeability is a crucial feature of peripheral elements: for δουρὶ φαεινῷ and ὀξέϊ δουρὶ, for example, to function as a metrically identical pair that allows for adaptation to the prosodic circumstances, and so to adequately perform their function as peripheral elements, there must be, in this particular function, no semantic (lexical) barriers between the epithets “shining” and “sharp” which would restrict the choice of either of them.

But again it should be stressed that interchangeability does not imply loss or superfluity of meaning: the fact that δουρὶ φαεινῷ is in principle interchangeable with ὀξέϊ δουρί does not mean that the lexical difference between the two epithets is irrelevant. φαεινός ‘shining’ is the epithet for heroic weaponry in general as well as for “eyes,” those of Zeus in particular, the eyes that not only saw the battle but also oversaw its crucial developments. [25] The spear’s visibility as it shines in the sun when brandished, thrown, or missing its target is consistently brought out by the use of the φαεινός-epithet as periphery to verbs denoting these actions: ^Tἀκόντισε δουρὶ φαεινῷ || ‘made a throw with the shining spear’, ^Tτιτύσκετο δουρὶ φαεινῷ || ‘brandished the shining spear’, ^Tἀπήμβροτε δουρὶ φαεινῷ || “missed with the shining spear.” [26] In two of these cases, the other phrase ὀξέϊ δουρί could have been used on account of so-called nu movable in the Ionic dialect (ἀκόντισεν ὀξέϊ δουρὶ, ἀπήμβροτεν ὀξέϊ δουρί), but this happens only at Iliad IV 490. By contrast, ὀξέϊ δουρί is consistently used when the spear actually hits an opponent’s body or armor and so is true to its sharpness. Typical verbs in this connection are βεβλήκει ‘hit’ (Iliad V 73), οὔτασε ‘wounded’ (e.g., Iliad V 336, XVI 317), νύξ(ε) ‘thrust’ (XX 488–489), etc. Still, the poetic or factual difference between the two epithets can be easily overruled. At Iliad VI 32 (ἐνήρατο δουρὶ φαεινῷ), cited above on p. #, for example, δουρὶ φαεινῷ is used simply because the nuclear verb ἐνήρατο ends on a vowel.

Meaning, then, is real but not unconditional, and some writers have gone too far in applying semantic distinctions that are in themselves justified. An example is Odisseas Tsagarakis’s discussion of the semantic distinction between the two ornamental adjectives χαλκείῳ ‘bronze’ and ὀξυόεντι ‘sharp’ to ἔγχεϊ. By its very meaning, || ἔγχεϊ ὀξυόεντι is more suited to be applied to a killing than the more neutral || ἔγχεϊ χαλκείω^P. [27] Accordingly, Tsagarakis claims that the only time that || ἔγχεϊ ὀξυόεντι^T is used in the first half of the verse (Iliad XVI 309—in six other cases the phrase is used at the end of the line) it must be sensitive to the context, which is explicitly concerned with killing. [28] But the extra syllable is important and is probably the reason why ὀξυόεντι is used at Iliad XVI 309. Compare Iliad VI 30–31 (cited above on p. #), where χαλκείῳ is used simply because a penthemimeral caesura was needed, in spite of the fact that the context, on Tsagarakis’s account, would favor ὀξυόεντι. In sum, then, the semantics of epithets and ornamental adjectives in Homeric diction is more complex than either the notion of “ornamental epithet” or unconditional full significance would lead us to believe. [29]

We will now have a closer look at the meaning of δουρὶ and ἔγχεϊ in their context, in particular with respect to the verb to which they are joined. As will appear, the question as to whether or not a spear-expression can be seen as peripheral with respect to a nucleus depends to a great extent on the context in which the verb occurs and on the function it has in that context. The following tri-partition in the material appears to be meaningful: (i) the spear-expression is truly peripheral; this occurs when the nuclear verb denotes a killing or wounding in an ongoing narrative that is concerned with androktasiai with catalogue-like features; [30] (ii) the spear expression occurs in the context meant under (i), but it loses (some of) its peripheral status on account of some contextual feature; (iii) the spear-expression occurs outside ongoing battle-narrative and thus cannot be assigned a peripheral status. The three subdivisions will be dealt with in the following three subsections.

What is omnipresent in consciousness may be taken for granted to such a degree that its presence speaks for itself. And when it is mentioned, there is either a specific reason for doing so, or the mentioning serves the purpose of peripherality which we are studying here. The idea “with the spear” appears to be so prominent when battle is described that it is subsumed in the semantics of verbs of spear-handling (killing, wounding, aiming etc.): the modifier “with the spear” may be omitted with any of these verbs in contexts in which the use of spears goes as a matter of course. The consequence of this is that when a spear is explicitly mentioned in this context, the purpose is versification: a spear-handling verb is extended backwards or forwards to the nearest metrical boundary (verse-beginning or –end, caesura). In other words the dative modifier becomes a peripheral element that is suited to this particular context and this particular nucleus.

Let us start again from instances like Iliad VI 32 cited on p. #. Here a spear-expression of the form ⎯ ∪ ∪ ⎯ ⎯ || extends a verb of the form ^T∪ ⎯ ∪ ∪^B (falling between the trochaic caesura and the bucolic diaeresis) to the end of the line. δουρὶ φαεινῷ as extension of ἐνήρατο occurs only once, as already indicated, but as extension of ἀκόντισε, a verb of the same metrical form, it is very frequent (14 times in the Iliad), for example:

These verses are as to their internal structure similar to Iliad VI 32: the names of the agent and his target or victim, being semantically the most important, have “priority” in the production of the verse; together they lay down the verse-structure, the object occupying the important position just before the trochaic caesura and the remaining metrical space (the second half of the verse) being filled by the verb and its extension.

That δουρὶ φαεινῷ is no more than an optional extension of the verb appears from the fact that it can be easily dropped when the names of the aggressor and his intended victim cannot, for some reason, be placed in the first half of the line. Consider:

The name of Diomedes (Διομήδης, ∪ ∪ ⎯ ⎯) can be placed only at the end of the line; [31] only in the form of Τυδέος υἱός ‘son of Tydeus’ (⎯ ∪ ∪ ⎯ ∪) can it be placed in other locations in the verse, in particular before the trochaic caesura. The following hypothetical verse for “And to him the son of Tydeus aimed his shining spear” would therefore seem to be acceptable: [32]

However, this verse was never created. Perhaps the desire to add the descriptive detail (ἰθὺς μεμαῶτος ‘charging directly’), applying to the warrior aimed at (Hektor), played a role. But there is also a more important reason: Τυδέος υἱός before the trochaic caesura would be too prominent from an informational point of view. The hypothetical verse would be appropriate in a context in which the one who aimed at Hektor (who appears in pronominal form: τοῦ ‘him’) would still have to be identified as Diomedes, so that the name would convey new information in the context. But in the context of the actually attested line Diomedes is already present, and so the mentioning of his name is not new information; it merely serves to disambiguate the subject of ἀκόντισε. This is why Τυδέος υἱός is placed at the end of the line, after the verb, where it can be interpreted as a clarifying apposition to ἀκόντισε: “and at him (. . .), he aimed his spear, the son of Tydeus”: the patronymic expression has itself become a peripheral element. [33] The “dislocation” of Τυδέος υἱός goes at the cost of δουρὶ φαεινῷ, but this merely proves that this expression is a truly peripheral element, which can be dropped whenever the context gives reason to do so.

Consider also:

This verse is as to its propositional content identical to Iliad XVII 304 cited earlier: both state the fact that Hektor aimed his spear at Aias, and one could ask why the two verses are different. Again, the difference appears to be motivated by contextual considerations, which take precedence over the question as to whether or not to use δουρὶ φαεινῷ. The crucial difference between Iliad XVII 304 and Iliad XIV 402 is that the latter is not a neutral statement of the type “A (Hektor) aimed at B (Aias).” The line follows on a description of the lining up of the two armies in battle order, the Greeks by Poseidon and the Trojans by Hektor. Similes underline the shouting of the two armies and the tension rises: what will happen now? The idea of “the first” to do something comes to the fore, and this is in fact what determines the first verse after the description: πρῶτος ‘the first’ is placed at the important pre-caesural position and the usual constituents are rearranged accordingly. The name of the target, Αἴαντος, is placed first in the verse, and the name of Hektor, deducible from the previous context, is placed at the end of the line, as the kind of disambiguating apposition we saw earlier. The name is extended by the epithet φαίδιμος, so that it occupies the same metrical space as δουρὶ φαεινῷ.

The following example shows that apart from πρῶτος there can be more factors at work:

Here it is the metrical form of Ἰδομενῆος that causes the divergence from the basic pattern of Iliad XVII 304 and 525 cited earlier: if this form (⎯ ∪ ∪ ⎯ ∪) is placed at the normal position for new information before the trochaic caesura, there is no more room left for the subject Aineías, and as this form cannot be placed after the bucolic diaeresis, the object has to move to the end of the line.

These examples show that Homeric diction is capable of expressing subtle contextual nuances even in such stereotyped narrative situations in which the one warrior aims at the other. The central verb ἀκόντισε is fixed and immobile in its metrical position, and the flexibility in reacting to contextual factors is facilitated by δουρὶ φαεινῷ as it either fills the space between the verb and the end of the line in a heroically correct way, or leaves that space for contextually more significant material in a given case. For the meaning there is no difference, since the idea “spear” is already inherent in the lexical value of ἀκόντισε. [34]

Another example of the peripherality of spear-expressions is constituted by cases where δουρὶ or ἔγχεϊ are used in a context where the spear was mentioned just before:

The narrative situations underlying these examples are very similar. In both cases, a warrior aims his spear at a particular enemy, but misses; instead, he hits, by accident, another man who is present at the scene. [35] Both times it is stated that this accidental hit was done “with the spear,” a redundant detail, since the same throw is described just before as ἀκόντισε δουρὶ φαεινῷ. The conclusion seems inescapable, then, that in these two cases βάλε δουρὶ ‘hit with the spear’ is simply an extended form of βάλε ‘hit’, and that δουρί is a truly peripheral element. [36]

But what about the first mention of the spear in these two examples? Would not the fact that the spear persists in the narrative be a factor against its peripheral function? Second mention of the spear also occurs in the following extracts: [37]

On account of the second mention, by a full noun phrase (αἰχμὴ χαλκείη, 185), it might seem that the first mention, δουρὶ in 183, is fully integrated in the structure of the discourse, contrary to what we might expect on the basis of other examples. However, there are good reasons for analyzing δουρὶ in this extract, just as δουρὶ φαεινῷ earlier, as peripheral elements, in view of instances where a spear is referred to that is not mentioned earlier, e.g: [38]

When the wording is similar between two formulaic passages, the differences become revealing. Polupoites’ killing of Damasos and Akhilleus’ killing of Demoleon are identical battle events. What seems at first sight in Polupoites’ killing to be a genuine, referential mentioning of a spear (δουρὶ βάλεν Δάμασον ‘hit Damasos with the spear’) that makes possible the use of αἰχμὴ χαλκείῃ ‘bronze spear point’ two verses later, appears in light of Akhilleus’ killing in Book XX to be no more than the optional backward extension of βάλεν to the beginning of the line. For at Iliad XX 399 αἰχμή is used in an identical way, both in battle and in the verse, without any overt preparation. Of course, verbs such as νύσσειν ‘thrust’ and ἀκοντίζειν ‘throw (a spear)’ virtually imply the use of a spear, but the same principle equally applies to more neutral verbs such as βάλλειν ‘hit’ and οὐτάζειν ‘wound’. This proves that the omnipresence of spears in the depicted battle may correlate with absence of spears in the text. And the fact that this absence does not make the text illogical or incoherent is the basis for the use of δουρὶ/ἔγχεϊ as a context-neutral peripheral element.

Peripherality is a function, not a category. [39] In other words, a given expression can never belong to the category of peripheral expressions, viz. be peripheral by its very nature in the way in which a word can be a noun or a verb. The peripheral status of an expression, however self-evident that status might seem to be in some cases, always depends in the last resort on the use that is made of it as a peripheral element in a specific context. The neutral and hence “innocuous” meaning of certain elements in certain contexts is exploited for the sake of easy and smooth versification. But nothing prevents the element from being used in its full proper meaning. Conversely, even the names of the protagonists of the epic, whether or not with their epithets, may become peripheral additions to verbs, as part of a stylized epic form of spoken language. [40]

Spear-expressions, too, may be used as a significant, context-sensitive element in the same contexts and in the same metrical positions as the examples discussed earlier. Below follow two examples of this phenomenon; they do not have a special poetic effect, but they show that the peripheral status of an element can always be overruled whenever the context motivates this. Consider first Diomedes’ encounter with an immortal opponent in the Iliadic battle:

As to their form and metrical position, both instances of ἔγχεϊ χαλκείῳ in this passage, are identical to ἔγχεϊ χαλκείῳ at Iliad VI 31 (cited on page #). This time, however, the phrase is not used merely because the part of the verse before the penthemimeral caesura has to be filled. But then this passage is not the standard account of a killing. The confrontation of Diomedes and Ares, and the repeated intervention of Athena in this fight, constitute a highly specific narrative situation, and this is immediately reflected in the function of ἔγχεϊ χαλκείῳ. The spears that Ares and Diomedes throw at each other are objects in their own right, whose course we are allowed to follow in the narrative. Each time the subsequent discourse, in which the intervention of Athena is described, is concerned with the spear, as Athena’s intervention consists in operations upon the weapon. In terms of text linguistics, ἔγχεϊ χαλκείῳ twice functions as the first mention of a topic which “persists” into the following clause. The syntactic reflex of this persistence in line 853 is the anaphoric pronoun τό ‘it’; in line 856 there is what may be called “zero-anaphora”: the topic is so continuous that it can be omitted as the syntactic object of the following verb. [41]

The full integration of the two instances of ἔγχεϊ χαλκείῳ in the chain of ideas in the narrative appears from the fact that each time the deletion of the phrase would disrupt the coherence of the discourse: τό γε in line 853 would be left hanging in the air, and there would be uncertainty as to the object of ἐπέρεισε ‘she pushed’ in line 855. The difference with Iliad XII 182–186 on p. # above is clear. There the second reference to the spear is not pronominal or zero, but fully nominal. This alone makes the preceding dative redundant: it can easily be missed, as is shown by Iliad XX 395–400 on p. #. Furthermore, and more importantly, δουρὶ at Iliad XII 183 (p. #) cannot be called the first mention of a persistent topic: the subsequent discourse in those examples is not concerned with what happened to the spear, as in the encounter of Ares and Diomedes, but with what happened next, or what happened to the victim.

In the following example, the dative expression is significant for another reason:

Here we have two contrasting sentences: two different warriors, functioning as contrasting topics (τὸν μὲν … τὸν δ᾿ ἕτερον ‘the one … the other’) are hit by two different weapons at two different parts of their body. Consequently, the dative χαλκήρεϊ δουρὶ is a means to differentiate two different killings from one another and cannot be a peripheral element.

The notion of peripherality is entirely bound up with the appropriateness of a given element with respect to a given nucleus. This means that outside those contexts the bond is gone and that the element conveys independently its proper meaning. For example, it makes a real difference whether or not the particle καί is followed by the combination “participle + περ”: before the participle, καί is a peripheral element whose function it is to adapt the participle, by backwards extension, to the metrical circumstances. [42] The particle can have that function because it has a meaning that is neutral with respect to the concessive context constituted by the participle (cf. though and even though in English). But without the participle, καί is used for its own sake. In other words, the particle’s meaning is now not exploited, but used. [43] The same applies to our spear datives.

Outside battle narrative in the sense delimited above, spears may be referred to together with other weapons. The dative for “spear” is then coordinated with other expressions:

“With his spear” as a coordinated phrase yields the recurrent phrase δουρί τε μακρῷ ‘and with the long spear’, which cannot function as a peripheral element, since in containing the connective particle τε it can hardly be called context-neutral (the context has to be concerned with coordination):

The datives ἔγχεϊ or δουρί cannot be a peripheral element when they refer to a spear that is not used as a weapon, for instance when the wounded Diomedes and Odysseus come to the Assembly “leaning on their spear” (|| ἔγχει ἐρειδομένω, Iliad XIX 49). Furthermore, “with the spear” can be used metonymically, so that the spear stands for warfare. [44] In this use ἔγχεϊ is the complement of predicates denoting excellence, e.g.: [45]

When characters mention their spear, even when referring to the type of warfare that generated the peripheral system for “with the spear,” the mention is never peripheral in the sense of this chapter. See for example Akhilleus’ threat to Hektor: [46]

It may be that a Homeric warrior cannot rage but with his spear, but this has not created a need for a peripheral system around the nucleus “rage.”

To conclude, we can say that qualifications like “meaningful” or “significant” in the study of Homeric discourse have to be used with circumspection. Nothing in Homer is meaningless, but much is not intentionally meaningful either. To be keen on a poetically effective use of a given expression may be understandable as a reaction to what some Parryists have done to Homer the creative poet, and it is justified by the poetic effects that emerge at the most unexpected moments. But meaning is often qualified in Homeric diction, depending as it does on context-type, subject matter, and most of all on what I have been calling peripheral systems, as a response to important and recurring tasks. Such a response is in the last resort nothing but the emergence of a grammar. [47]

The notion of grammar, in a direct linguistic sense, may be used to get the relation between “formula” and “original meaning” in sharper focus. Grammar is not a constraint but a set of emergent rules that make purposeful expression and communication possible. True, grammar may create standardized, indeed formulaic phrases, that have to be used, and remembered, as such in their idiomatic fixity. But grammar may also involve the kind of regularity by which speakers behave according to flexible rules that they help shape in using them. Here the distinction between “traditional” and “original” loses its meaning, since speakers are both. In this way Homeric formulas can certainly be the ready-made building blocks that appear in many discussions. But no poetry or narrative would be possible without more flexible strategies, such as the regularized adaptations of nuclear elements to the metrical circumstances that we have studied in this chapter. The notion of peripheral semantics thus explicitly leaves room for “free will” without precluding the existence of a style and a method of versification which significantly differ from unequivocally literate poetry.