Chapter 2. Formula, Context, and Synonymy

Milman Parry’s 1928 dissertation, [1] insofar as it drew upon and continued the work of earlier scholars, provided a functional dimension to the findings of such earlier scholars as Düntzer, Ellendt, Witte, and Meister. Parry pointed out that the coexistence of semantically equivalent but metrically different forms in the Homeric diction was not only the creation of epic verse (cf. the notion of the epic Kunstsprache as a “hexametric” language), [2] but was also due to the need of the epic poets to have metrical alternatives for the sake of easy versification. In other words, many peculiarities of the Homeric language and diction could be shown to have not only a cause, but also a motivation.

As is well known, Parry’s main concern was to show that by adding epithets to names or substantives, the epic poet had various, metrically different forms at his disposal for one and the same “essential idea.” [3] The regularity in the use of epithets, on the basis of which they form a system, could be shown to lie in their economy: for each relevant metrical form, there was one and only one noun-epithet combination, which suggested that the point of epithets was metrical differentiation in the first place, rather than to contribute to the content of the sentence.

The idea that epithets are used in systematic ways for the sake of metrical diversity could be added to the observations made by Parry’s predecessors, concerning the occurrence of morphologically and phonologically different alternatives (e.g. dialectal and even artificial forms) for a given word. To these, the same principle of economy applied. A third means to achieve metrical diversity was what can be called “functional synonymy,” that is, the use of synonyms, not so much for the sake of stylistic variation as, again, for the sake of metrical diversity. Parry was aware of the importance of synonymy and introduced the concept in his discussion of noun-epithet formulae for the Achaeans [4] and for the human race. [5] In both cases he followed Düntzer. [6]

The first systematic survey of synonymy in the Parryan context is offered by H. Paraskevaides. [7] Paraskevaides gives the term “synonymy” a very broad sense, applying it whenever a group of formulaic expressions conveys the same or a similar idea. Paraskevaides’ treatment of the issue is strictly Parryan, in that synonymy is consistently viewed as functional synonymy. This means that whenever in any given group of synonymous expressions two expressions occur with the same metrical form and prosodic characteristics they are considered to be a violation of the law of economy.

Paraskevaides has demonstrated the importance and scope of synonymy as a factor in Homeric formulaic diction, but in some cases more can be said about a given pair of “synonyms” than that they are either metrically different (and thus functional synonyms) or metrically equivalent (and thus a violation of economy). In other words, I will suggest that some of Paraskevaides’ alleged synonyms are, on closer inspection, not synonyms. The consequence of this is that in the case of metrically equivalent pairs the case for functional (i.e. formulaic) synonymy is strengthened: when two metrically equivalent expressions can be shown to have a different meaning, they cease to be a violation of the “law” of economy, which can only be favorable to the notion of functional synonymy. In the present chapter I present a representative instance of this phenomenon.

The case to be discussed is the subject of the previous chapter, the set of dative expressions for “spear.” This formulaic set, which in itself is based on functional synonymy, displays a number of metrically identical pairs. My argument to the effect that this duplication is motivated by various differences in meaning primarily applies, of course, to the expressions themselves, but in its general orientation it is applicable, it is to be hoped, to other aspects of Homeric diction.

The system of singular dative epithet-expressions for “spear” can be presented as follows, after Paraskevaides (1984:26):

The metrical diversity displayed by this set is effected by the use of epithets as well as by the functional synonymy of the two main words, ἔγχος and δόρυ, which are metrically unequivalent. In spite of their different archeological denotation, [8] ἔγχος and δόρυ are synonyms in the Homeric diction. [9] This appears, among other things, from the fact that the two terms can be used coreferentially. [10] A clear case of this is the following example that is also cited on p. # above:

In this passage, one and the same spear is referred to three times in succession, twice with δουρὶ, once with ἔγχος. This shows that the two words are synonymous in the sense that they may be used indiscriminately from a referential point of view. The interchangeability of the two lexemes lies at the basis of the datives being part of a formulaic system that in the previous Chapter is discussed as a case of “peripheral semantics.” It is on account of their function within this system that δουρί and ἔγχεϊ have their specific function in Homeric verse.

The metrical diversity displayed by the list cited above may be motivated by the formulaic purpose of the system of which the expressions are a part. But the list contains a number of metrically and prosodically identical pairs. The two main words form such a pair by the addition of metrically different epithets (ὀξέϊ δουρὶ ‘with the sharp spear’ vs. ἔγχεϊ μακρῷ ‘with the long lance’); furthermore, every instance of χαλκῷ ‘with bronze’ with epithet is consistently matched by an equivalent expression containing δουρὶ or ἔγχεϊ. Thus beside ὀξέϊ χαλκῷ ‘with the sharp bronze’ we have the two expressions just mentioned (ὀξέϊ δουρὶ and ἔγχεϊ μακρῷ (all three begin with a vowel); beside νηλέϊ χαλκῷ ‘with merciless bronze’ we have δουρὶ φαεινῷ ‘with the shining spear’ (both begin with consonant); and, finally, beside ταναήκεϊ χαλκῷ ‘with the sharpened bronze’ we have χαλκήρεϊ δουρὶ ‘with the bronze spear’. When we view the list as a formulaic system, that is, as a group of synonymous expressions conveying the idea “with the spear,” these pairs are mere duplicates and, as such, violations of the law of economy. This is the way they are treated by Paraskevaides (1984:23).

But Paraskevaides’ analysis is not immune to criticism, and in what follows, I challenge his view, offering some considerations on synonymy and metrical equivalence in Homer. These fall into three parts, covering three aspects of synonymy in Homer. The diachronic aspect (section 1) applies to the pair ὀξέϊ δουρὶ vs. ἔγχεϊ μακρῷ. It consists in the insight that two alleged synonyms may still be different in the diachronic dimension. The synchronic usage in Homer may reflect this former difference, in spite of the interchangeability of the expressions on other grounds.

Then follows the discussion of the semantic aspect (section 2), which applies to cases in which two allegedly synonymous expressions turn out, under closer investigation, not to be synonymous at all. The case here is ὀξέϊ δουρὶ vs. ὀξέϊ χαλκῷ. On the basis of a linguistic examination of the contexts in which they occur it can be shown that the two expressions conform to entirely different distributional patterns.

The poetic aspect (section 3), finally, lies in the significant addition of an epithet to a nuclear noun. The specific value of the epithet may preclude the equivalence in meaning of two allegedly synonymous expressions. This section will be concerned with the difference between νηλέϊ χαλκῷ and δουρὶ φαεινῷ.

It should be stressed that these three aspects of synonymy are not necessarily distinct, in that the presence of one of them excludes the other two. For example, the preservation of an older, metrically equivalent form may be motivated by semantic or poetic factors, and a semantic difference between two metrically equivalent expressions may be put to poetic use. The three aspects merely represent three vantage points from which to describe synonymy in Homer beyond the confines of formal analysis.

We saw that δόρυ and ἔγχος are functional synonyms in the Homeric diction. But on this basis, the co-occurrence of ὀξέϊ δουρὶ and ἔγχεϊ μακρῷ would be entirely dysfunctional and would disrupt the economy of the system: both expressions have exactly the same metrical form and localization (at the end of the verse, after the bucolic diaeresis).

Ὀξέϊ δουρὶ and ἔγχεϊ μακρῷ seem to be wholly interchangeable in the following pair, where they are used as periphery to one and the same verb of wounding:

There does not seem to be a specific reason why ὀξέϊ δουρὶ should have been used specifically in the one case and ἔγχεϊ μακρῷ in the other. [11] Consequently, examples like this one seem to suggest that the phrases ὀξέϊ δουρὶ and ἔγχεϊ μακρῷ are just as interchangeable as the lexemes δόρυ and ἔγχος in general (see Iliad XIII 516–20 above, p. #). In the following pair, however, there is a difference:

The wider context of V 238 makes it clear that the spear mentioned will be used by its owner (Pandarus) for throwing (V 280 προΐει ‘hurled’; 281 βάλε ‘hit’; 282 πταμένη ‘flying’), [13] whereas at XV 745 Aias is keeping the Trojans away from the ships (XV 730–731 ἔγχεϊ δ᾿ αἰεὶ | Τρῶασ ἄμυνε νεῶν ‘each time he warded off the Trojans with his lance'), thus using his spear as a defensive weapon rather than as a missile.

Cases like XV 745 suggest what was already stated by Trümpy and Lorimer: [14] originally, δόρυ and ἔγχος did not designate the same type of weapon. Ἔγχος, the older word, [15] is the term for the long Bronze Age thrusting-spear, whereas δόρυ, the normal term for “spear” in post-Homeric Greek, refers to the lighter throwing-spear that came into use in the late Mycenaean era. The terms are (functional) synonyms in the Homeric diction, for reasons specified above, but when it comes to describing what one actually does with a spear, [16] or to singling out a truly heroic weapon, handled only by a special class of heroes (viz., Achilles and Aias), [17] ἔγχος can still have its proper, original meaning, which fully justifies the presence of the epithet μακρῷ ‘long’. This would seem to be the case in the following example:

Here a typical ἔγχος-bearer does what one typically does with an ἔγχος in its historically correct denotation: he thrusts (νύξ᾿) the weapon into his opponent’s body. [18]

The original difference between ἔγχος and δόρυ can be said to have two linguistic reflexes. First, δόρυ, and not ἔγχος, has an affinity with verbs denoting throwing (e.g. ἀκοντίζω ‘throw ’). [19] Thus it would seem that the use of ὀξέϊ δουρὶ in the following example is no more accidental than that of ἔγχεϊ μακρῷ in the previous one: [20]

Second, as Trümpy has noted, [21] δόρυ is regularly used in the dual form (δύο δοῦρε ‘two spears’, ἄλκιμα δοῦρε ‘two strong spears’), explicitly designating a pair of light throwing-spears. Also, warriors regularly have a second δόρυ at their disposal when they have used their first one. None of these facts applies to the ἔγχος: ἔγχος never appears in the dual form, and it typically refers to a single weapon. The distinction is clearly brought out in the arming scene of Patroklos:

The unusually clear distinction between δόρυ and ἔγχος in this passage is due, of course, to the particular significance of Achilles’ ash spear. [22]

We may conclude, then, that even though ὀξέϊ δουρὶ and ἔγχεϊ μακρῷ may be used as genuine metrical duplicates, owing to the extensive functional synonymy of δόρυ and ἔγχος, they are not interchangeable without further ado: sometimes the original technical distinction between two kinds of weapons plays a role. The need to use ἔγχεϊ μακρῷ, in full accordance with the meaning of its epithet, to designate a specific kind of spear, still motivates its occurrence in the diction, as the metrical equivalent of ὀξέϊ δουρὶ. A detail of past heroic warfare belongs to the present of the epic tradition. In later chapters we will see that such pointing to the past is not limited to epic realia, but applies no less to epic language itself. [23]

We now turn to the discussion of the meaning of ὀξέϊ (ταναήκεϊ) χαλκῷ ‘with the sharp(ened) bronze’ with respect to that of ὀξέϊ (χαλκήρεϊ) δουρὶ ‘with the sharp (bronze) spear’ and ἔγχεϊ μακρῷ ‘with the long lance’. Of course, χαλκός ‘bronze’ may come to mean “spear” only by a possible metonymical extension of its meaning; there are cases where the use of χαλκός is irrelevant for our purposes. At Iliad XXIII 118, for example, it is used to refer to an axe, and it may also just denote the bronze of which the spear point is made (Iliad X 135; XIV 12; XV 482). In these cases, obviously, even the alleged synonymy of χαλκός and the corresponding δουρὶ- or ἔγχεϊ-expression is ruled out.

But in the cases where the reference to a spear is not ruled out, an investigation of the contexts suggests that ὀξέϊ (ταναήκεϊ) χαλκῷ and ὀξέϊ (χαλκήρεϊ) δουρὶ are used under entirely different circumstances. ὀξέϊ χαλκῷ typically combines with predicates that denote states, such as (medio-) passive participles, for instance:

The verbs of the δουρὶ-expressions, by contrast, denote less “time-stable” phenomena, such as actions (consisting of killing, thrusting, hitting, or throwing), rather than states, for example:

Χαλκῷ-expressions are not impossible with this kind of verb, but when they occur with it, the expression as a whole is still different from the examples just quoted, in which an action in combat is reported: instead of being factual and declarative, describing one specific killing or wounding, the context in which a χαλκῷ-expression occurs with an action-denoting verb is either nonfactual or generic in the sense that it does not describe one particular event. Typical environments for a χαλκῷ-expression used in this way are conditional clauses, which are by definition nonfactual, in describing what may happen or what has not yet happened: [24]

Another typical environment for ὀξέϊ (ταναήκεϊ) χαλκῷ is the following generic (“iterative”) sentence, which does not refer to a particular killing:

Thus we see that ὀξέϊ δουρὶ is at home in contexts that describe one particular, factual killing, while ὀξέϊ χαλκῷ typically occurs in contexts that are characterized either by a stative verb, denoting a time-stable phenomenon, or by a nonfactual or generic modality. [25] Now what these contexts have in common in opposition to the sentences describing one particular killing or hitting is that they are less transitive.

Following an influential article in modern discourse linguistics, [26] I define “transitivity” as “the degree to which an action is ‘carried over’ from one person or entity to another.” The notions “transitive” and “intransitive” are thus not the members of a binary pair, [27] but the two extremes of a continuum of increasing transitivity. Transitivity is what may be called a “cluster-concept” in that it consists of a number of independent parameters. One of these is individuation of the object. By this criterion “Odysseus killed Pidytes” is higher in transitivity than “Odysseus killed fifty Trojans.” A second parameter is reality (or factivity): real events are more transitive than events that have as yet not occurred; by this criterion “Odysseus killed Pidytes” is higher in transitivity than “Odysseus will kill Pidytes.” A third parameter is volitionality: “Odysseus jumped” is more transitive than “Odysseus fell,” even though neither is transitive in the traditional definition. [28]

The degree of transitivity may be reflected by properties of the sentence, and this insight is, I suggest, the appropriate basis to differentiate ὀξέϊ χαλκῷ from ὀξέϊ δουρὶ. The contexts in which ὀξέϊ χαλκῷ typically occurs consistently display a lower degree of transitivity by one or more of Hopper’s and Thompson’s parameters than the contexts in which ὀξέϊ δουρὶ occurs. The examples with medio-passive participles cited above represent states with only one (nonvolitional) participant, a typical case of low transitivity. The conditional clauses are less transitive by the factivity criterion: no transfer of energy has yet taken place. The generic (“distributive-iterative”) sentence, finally, is less transitive, too. This is obvious, because a repeated act performed on new victims all the time, belonging to the same class (“every time he was killing Greeks”), has a considerably less individuated object than a specific act that occurs only once. [29]

The preference for ὀξέϊ χαλκῷ in these contexts is now easy to explain in the light of the argument above. Being less specific by its lexical value (which denotes only the material), and as such not necessarily referring to a particular weapon, this dative for “spear” has a natural affinity with less transitive contexts that favor noun phrases with a less specific denotation. [30] Consequently, ὀξέϊ δουρί, being more specific, is blocked from these contexts, occurring instead in the highly transitive sentences that report an actual killing. [31]

Returning, after this linguistic investigation, to the system of dative expressions for “spear” (see the list on p. #), we have to conclude that on account of their different distributional patterns the χαλκῷ-expressions in fact do not belong to this system at all; they form a mini-system of their own. To realize that the χαλκῷ-expressions are not synonymous with the δουρὶ-expressions, but have a different function in the language, enhances the economy of the system in Parry’s sense, since a number of violations have disappeared. In the new χαλκῷ-subsystem, metrical diversity is realized by the choice of different epithets (ὀξέϊ, ταναήκεϊ, νηλέϊ). Metrical-prosodic considerations, however, are not always sufficient to account for the presence of these epithets, as appears from the use of νηλέϊ to which we now turn.

It might seem tempting to assume that the relation between δουρὶ φαεινῷ and νηλέϊ χαλκῷ is analogous to that between ὀξέϊ δουρὶ and ὀξέϊ χαλκῷ, so that νηλέϊ χαλκῷ would be the prosodic counterpart of ὀξέϊ χαλκῷ (having an initial consonant instead of a vowel). In a system of functional synonyms in which each item has a reason for being precisely because of its metrical and prosodic properties this is to be expected and it is the reason which Paraskevaides gives for the coexistence of the two expressions. [32] It is indeed true that νηλέϊ χαλκῷ may function as what seems to be the prosodic duplicate of ὀξέϊ χαλκῷ in less transitive contexts where an initial consonant is needed, as in:

The first example (Iliad XII 427) is less transitive because the agent of the act is not expressed and also because the subject is unspecified; in the second example, we have again a conditional environment. In these cases, then, the reasons for using νηλέϊ χαλκῷ instead of its metrical and prosodic doublet δουρὶ φαεινῷ seem to be the same as those for preferring ὀξέϊ χαλκῷ to ὀξέϊ δουρὶ in the less transitive contexts cited earlier.

However, whereas ὀξέϊ χαλκῷ and ὀξέϊ δουρὶ differ only with respect to the main word, νηλέϊ χαλκῷ and δουρὶ φαεινῷ have different epithets as well. And it is due to the force of its epithet that νηλέϊ χαλκῷ is not equivalent to δουρὶ φαεινῷ. In the context of functional synonymy and Parryan economy, when an alleged violation of the law of economy involves epithets, extra caution is in order: the epithets are more likely to have a specific “poetic” meaning precisely because they do not serve to differentiate two expressions metrically from each other. [33] In a number of instances, in any case, the metrical duplication seems to be motivated by the lexical value and expressive potential of the epithets.

In a case not unlike the present one, for example, Eide successfully describes the (poetic) difference between the metrical duplicates χειρὶ παχείῃ ‘with his/her thick hand’ and χειρὶ βαρείῃ ‘with his heavy hand’: [34] the former is used in more or less neutral contexts where the instrument expression can be said to have, in terms of the previous chapter, a peripheral function, while χειρὶ βαρείῃ is used when the hand of someone is referred to as a frightening, threatening instrument. Likewise, within the set of nominative-accusative expressions for “spear,” the metrically equivalent pair μείλινον ἔγχος ‘ash spear’ vs. χάλκεον ἔγχος ‘bronze spear’ seems to be a violation of economy,” [35] but Évelyne Cosset plausibly argues against the treatment of the pair as merely equivalent: [36] χάλκεον ἔγχος is used to refer to a spear as a dangerous and heroic weapon with high killing power, whereas μείλινον ἔγχος is used when a spear misses its goal or when a given action is not forceful or heroic. [37]

In a similar way, νηλέϊ χαλκῷ may add an overtone of menace and imminent terror to the context that would be absent if the more neutral δουρὶ φαεινῷ were used, for example:

One might argue here that νηλέϊ χαλκῷ is used, not because of the epithet νηλέϊ, but because of χαλκῷ, in view of the nonfactual context (in the case of XIII 501 and XVI 761 the spear-expression occurs within the complement of ἵεντ᾿ ‘they wished to…’, and in the case of XIII 551–553 the context is negated, as complement of οὐδὲ δύναντο ‘they could not …’). It is true that these contexts are different from factual contexts like “He killed/hit him with the spear,” but that does not mean that νηλέϊ χαλκῷ is a mere prosodic alternative to ὀξέϊ χαλκῷ. Again, the difference between δουρὶ φαεινῷ and νηλέϊ χαλκῷ lies both in the opposition between “pitiless” and “shining” no less than between “spear” and “bronze.” Besides, νηλέϊ χαλκῷ is sometimes used in factual contexts that seem to be the domain of the δουρὶ-expressions:

The force of the epithet νηλέϊ as opposed to that of φαεινῷ, is particularly suited to the gruesome detail in which the hit is described, which is another way of saying that νηλέϊ χαλκῷ is not just a prosodic alternative to ὀξέϊ χαλκῷ, or a doublet of δουρὶ φαεινῷ.

Insofar as the preceding discussion of the “doublets” in the system for “spear” in the dative has implications beyond that system, it shows that the issue of synonymy in Homeric diction involves more factors than metrical-prosodic ones only. The Parryan approach to synonymy is functional insofar as it is concerned with the similarity in function of two or more expressions with different metrical forms. But to the extent that this similarity is subservient to a system, the Parryan approach had better be called structuralist, in that the emphasis is on the function of expressions within the system to which they belong rather than to their actual contexts.

To recognize the existence and importance of formulaic “systems” in Homeric diction is of course wholly legitimate and indeed one of the ways to make progress in understanding Homeric discourse. [38] Formulaic systems are indispensable for oral verse-making, but essentially they are no more than the adaptation of language to the verse, not language itself. The function a given element has in a given system does not tell us anything about the function of that element in actual contexts, for the simple reason that “context” is much more than “metrical-prosodic environment.” The discussion of the set of dative expressions for “spear” has shown that the structural and “systematic” discussion of a set of synonyms can never replace the study of its constitutive members in actual contexts. Not only may a closer study of “context” be necessary for determining, on a linguistic basis, the extension of the set (as in the case of δουρὶ vs. χαλκῷ), it can also reveal that “context” may acquire the value of “poetic significance.” Both aspects, the meaning of language and the meaning of poetry, cannot be captured by the rigor of any system, but it cannot be anyone’s intention to rule them out.