The Center for Hellenic Studies

Rhythm without Beat: Prosodically Motivated Grammarisation in Homer

  Use the following persistent identifier: http://nrs.harvard.edu/urn-3:hul.ebook:CHS_BlankenborgR.Rhythm_without_Beat.2014.

2. The Grammatical Clause in Homer

2.0 Introduction

In chapter 1, I have discussed the problems concerning the various ways in which the syntax of Homeric discourse has been analysed by, among others, Chantraine, Bakker, and Clark. Discussing the descriptions of Homeric syntax also included dealing with the issue of enjambment in Homer, and the concept of grammatisation within metrical boundaries. Existing studies of Homeric discourse provide many examples of the emergence of grammar between recurring positions of frequent word end. I have shown, however, that we must allow for a second type of phrasing; a type that elicits the emergence of a higher level of syntactical organisation due to the juxtaposition of metrical phrases. Such juxtaposition is not per se appositional, but regularly shows the mark of well-prepared linking of chunks to achieve more complex syntactical arrangements. The emergence of syntax is not coincidental, but carefully anticipated by a poet who consciously composes his clauses in grammatically complete wholes comprising of several metrical phrases. The grammatical anticipation reduces the importance of the compositional principles of adding style; it rather adds the concept of open-ended clause formation to the description of Homeric verse-making. Homeric style is not as appositional as the subdivision of verses into metrical cola suggests.

Often, the term “clause” is used to refer to units showing syntactical cohesion in Homer. [1] In the examples in chapter 1, recurring metrical boundaries indeed regularly appear to frame phrases that are in themselves grammatical clauses. Still, many instances remain where this does not seem to be the case. In this chapter, I will apply to Homer modern approaches to the grammatical clause. The concept of the grammatical clause as used in functional grammar will turn out to be most suitable for the description of both the grammatically complete metrical colon and verse, and the open-ended type of clause formation.

“Clause” has been used a few times in the preceding chapter—for want of more precise terminology, and in line with common practice. I see no reason to object to the term “clause” beforehand, but I think that the usefulness of the term for the Homeric epic needs to be considered more carefully than has been done so far, before it is accepted. I will first focus on the usefulness of the term “clause” to analyse the units of grammar and, in a way, of sense. In my view, modern methods of linguistic analysis are easily applicable and yield, I think, their best results precisely in Homer.

2.1 Defining the Grammatical Clause

When using the grammatical term “clause”, and applying it to the Homeric epic, we must define what exactly is meant by grammatical clause. Where does such a clause begin? Where does it end? What clues are there in the Iliad and the Odyssey to enable the listening audience to perceive any clausal structure in performance? This final question also deals with clues other than grammatical. Part of the answer will hence be deferred to other chapters. For now, the grammatical clues and characteristics will be the main focus.

In recent years, functional grammar [2] has provided a model for the description of the grammatical clause and the sentence. [3] The central role in this description is that of the predicate, the finite verb, because it is the verb form that dictates the further requirements for a grammatically correct and complete clause. The semantics of the predicate determine the predicate frame: the meaning of the finite verb creates, or requires, one or more valencies for arguments with a specific semantic value in relation to the predicate. In Iliad 1.1, for example, the verb form ἄειδε has a valency of 2: the subject [4] and a direct object:

μῆνιν ἄειδε θεά Πηληιάδεω Ἀχιλῆος

Iliad 1.1

Sing, goddess, of the wrath of Achilles, son of Peleus


In Odyssey 1.1 the verb form ἔννεπε has a valency of 3: subject, direct object, and indirect object:

ἄνδρα μοι ἔννεπε Μοῦσα πολύτροπον ὃς μάλα πολλὰ

Odyssey 1.1

Tell me, Muse, about the man of many devices, who for a very long time


The resulting predicate frame is referred to as the nuclear predication. In Iliad 1.1 the nuclear predication is μῆνιν ἄειδε ‘sing of the wrath’. In Odyssey 1.1 it is ἄνδρα μοι ἔννεπε ‘tell me about the man’. The verbs ἄειδω ‘sing’ and ἐνέπω ‘tell’ are being used as transitive verbs here, but both may be applied as intransitive [5] with similar ease. In functional grammar, this difference in usage transitive-intransitive is seen as a difference in semantics, and described as a difference in predicate frames; a verb can be identified as being used with, or within, various predicate frames, depending on context.

Both in Iliad 1.1 and in Odyssey 1.1, the predication consists of more than the mere predicate frame. In Iliad 1.1, we find, beside the nuclear predication, the word group Πηληιάδεω Ἀχιλῆος ‘of Achilles, son of Peleus’. In Odyssey 1.1 there is the describing adjective πολύτροπον ‘resourceful’. In both Iliad 1.1 and Odyssey 1.1, there is a vocative. These peripheral constituents of the predication, as opposed to the nuclear predication, [6] are labelled satellites.

In general, there are two types of satellites. On the one hand, we find the adjuncts: satellites that further specify the nuclear predication, with the semantic role instrument, beneficiary, purpose, time, place, mode and degree, [7] and the pragmatic function [8] setting. On the other hand, the disjuncts, satellites that render meta-communicative expressions; for example: the point of view or opinion of either the writer or, as understood by the writer, of the audience. The scope of disjuncts is the predication as a whole. Examples of disjuncts are the modal adverbs, opinion-based adverbative elements, style disjuncts, pseudo-final subordinate clauses, parenthetic conditionals, constituents with the pragmatic function theme and tail, [9] and in addition, I argue, the vocative. [10] In the Iliad and the Odyssey, examples for each of the usages mentioned can be found:

Instrument:

μειδήσασα δὲ παιδὸς ἐδέξατο χειρὶ κύπελλον

Iliad 1.596

smiling, she took the cup from her son with her hand

Beneficiary:

μητρὶ φίληι ὲν χερσὶ τίθει καί μιν προσέειπεν

Iliad 1.585

placed it in her hand for his mother and spoke to her

Purpose:

          ἐμοὶ δέ κε ταῦτα μελήσεται ὄφρα τελέσσω
εἰ δ᾿ ἄγε τοι κεφαλῆι κατανεύσομαι ὄφρα πεποίθηις

Iliad 1.523–524

for me these matters will be under my attention to bring about; | well then, I will nod with my head, that you may feel confident [11]

Time:

τήι δεκάτηι δ᾿ ἀγορήνδε καλέσσατο λαὸν Ἀχιλλεύς

Iliad 1.54

on the tenth day, Achilles summoned the army to a gathering


Of course, the expression of “time” can also refer to a specific duration or period.

Place:

     οὐδ᾿ ἔνθα πεφυγμένος ἦεν ἀέθλων

Odyssey 1.18

not even there was he completely safe from misfortunes


Apart from the actual location, “place” encompasses the direction, the origin and the route as well.

Mode:

ἀλλὰ Ποσειδάων γαιήοχος ἀσκελὲς αἰὲν
Κύκλωπος κεχόλωται

Odyssey 1.68–69

but earth-shaking Poseidon is still always, endlessly, | furious because of the Cyclops

Degree:

ἂψ ἴτω ἐς μέγαρον πατρὸς μέγα δυναμένοιο

Odyssey 1.276

she has to leave straightaway for the palace of her father, greatly empowered

Setting as a pragmatic function means that the constituent provides the audience with a situation or an event that serves as the starting point for the nuclear predication. In Greek, setting, like the semantic function purpose, is regularly expressed by means of a verb form. Examples are the participle, or a word group introduced by a conjunction:

τοῦ ὅ γ᾿ ἐπιμνησθεὶς ἔπε᾿ ἀθανάτοισι μετηύδα

Odyssey 1.31

thinking of him he spoke the following words to the immortals

εἰ μὲν δὴ νῦν τοῦτο φίλον μακάρεσσι θεοῖσι
νοστῆσαι Ὀδυσῆα πολύφρονα ὅνδε δόμονδε
Ἑρμείαν μὲν ἔπειτα διάκτορον ἀργειφόντην
νῆσον ἐς Ὠγυγίην ὀτρυνομεν κτλ.

Odyssey 1.82–85

if that (is) indeed dear to the blessed gods, | that Odysseus returns to what is his, to his home, | then let us send Hermes, the guide, the slayer of Argus | to the island Ogygia …


These two examples from the first book of the Odyssey show that the pragmatic function setting, as does the semantic function purpose, creates the possibility to describe embedded predications. The syntax of the Homeric epic, however, is highly paratactic, [12] and allows for a strong progressive and adding tendency (as line 83 in the example Odyssey 1.82–85 clearly shows). I think this severely limits the application of setting, a pragmatic function that seems to me more convincingly identified in written, periodic style.

Modal adverb: in Homer, modal adverbs are always disjuncts. They cannot be used as independent utterances, as in, for example, Plato (πάνυ γε). They strengthen, weaken, or negate meaning:

ἤε κεν νοστήσας ἀποτίσεται ἦε καὶ οὐκί

Odyssey 1.268

whether he will make them pay after his return, or, as is as likely, not


In addition, modal adverbs express the illocutive function of a clause, that is, the pragmatic function of the clause as a whole: what is the aim of the clause as a speech-act? what does the clause want the receiver to do in reaction? An example is the indication of the start of a question:

ἦ φὴις τοῦτο κάκιστον ἐν ἀνθρώποισι τετύχθαι

Odyssey 1.391

let me ask you, do you think that this has turned out to be the worst thing among mortals?


The particles κε (κεν) and ἄν also serve as disjuncts. [13]

Opinion-based adverbative elements (or attitudinal disjuncts):

καὶ λίην κεῖνός γε ἐοικότι κεῖται ὀλέθρωι

Odyssey 1.46

yes, clearly, that man lies low in a destruction that is his due


In many instances, it is difficult to appreciate the opinion-based value of the adverbative element. Three aspects of this difficulty are worth mentioning:

  1. the actual opinion in the clause may be expressed by another word (in the example above ἐοικότι ‘that is his due’), and is merely strengthened, or at least not weakened, by the adverbative element;

  2. often in Greek, this type of disjunct will be expressed by a particle, but the meaning and value of various Greek particles remain in dispute. For an overview, see Denniston 1954, Ruijgh 1971. Clearly, analysis of, for example δή as a near equivalent of ἤδη turns many instances of the latter into opinion-based adverbative elements;

  3. identifying opinions and judgements in ancient Greek texts is the aim of studies into focalising and focalisors: if it is clear through the eyes of whom situations or actions are being seen, presented and judged, it will become clearer whether single words or combinations of words might or should be regarded as attitudinal. Fundamental research in this domain is presented in De Jong 1987.

Style disjuncts:

ὦ γέρον ἤτοι ἐγὼ θεὸς ἄμβροτος εἰλήλουθα
Ἑρμείας κτλ.

Iliad 24.460–461

old man, to be frank, I, an immortal god, have come to your aid: | Hermes … [14]

Pseudo-final subordinate clauses:

ἠ᾿ ἀπόειπ᾿ ἐπεὶ οὔ τοι ἔπι δέος ὄφρ᾿ εὖ εἰδῶ
ὅσσον ἐγὼ μετὰ πᾶσιν ἀτιμοτάτη θεός εἰμι

Iliad 1.515–516

or say no, since for you there is no reason to fear, that I may know for sure | how thoroughly without privilege a goddess I am amidst all


As in the examples given with setting above, the paratactic character of Homeric discourse makes it difficult to distinguish between clauses proper and clauses within a larger scale unit. If ἐπεὶ οὔ τοι ἔπι δέος is not interpreted as a parenthesis, ὄφρ᾿ εὖ εἰδῶ ὅσσον ἐγὼ μετὰ πᾶσιν ἀτιμοτάτη θεός εἰμι is merely pseudo-final.

Parenthetic conditionals:

εἰ δ᾿ ἐθέλεις καὶ ταῦτα δαήμεναι ὄφρ᾿ εὖ εἰδῆις
ἡμετέρην γενεήν πολλοὶ δέ μιν ἄνδρες ἴσασιν
ἔστι πόλις Ἐφύρη μυχῶι Ἄργεος ἱπποβότοιο
ἔνθα δὲ Σίσυφος ἔσκεν ὃ κέρδιστος γένετ᾿ ἀνδρῶν

Iliad 6.150–153

if you want to know that as well, that you may understand fully | my lineage—many men know of it –: | there is a city Ephyre in a far region of horse-breeding Argos; | there Sisyphus resided, who grew to be craftiest of men


Again, as in the preceding example, it proves difficult to label seemingly subordinate clauses in the Homeric epic with semantic or pragmatic functions in relation to what would be called a main clause in written language. [15]

Theme- and tail-constituents:

Λητοῦς καὶ Διὸς υἱός ὃ γὰρ βασιλῆι χολωθεὶς
νοῦσον ἀνὰ στρατὸν ὦρσε κακήν ὀλέκοντο δὲ λαοί
οὕνεκα τὸν Χρύσην ἠτίμασεν ἀρητῆρα
Ἀτρείδης

Iliad 1.9–12

Leto’s and Zeus’ son (theme): for he felt angry towards the king and | sent a foul plague over the army’s camp—the soldiers perished—| since he had dishonoured him, Chryses, the priest, | he, that is, Agamemnon (tail)


The model thus described considers the predication as equal to the sentence as it is understood in written language. [16] With the exception of anacolutha, there is a hierarchical relation between the main finite verb and other verb forms within a complex sentence. The subordinate clause or clauses feature verb forms (and their predicate frame) with semantic or pragmatic functions vis-à-vis the main verb form: they are either argument to the main nuclear predication, or satellite within the complex predication as a whole. Their sub-position to the main predicate is reflected in terminology: they constitute embedded predications. These embedded predications, in turn, serve as the centre of their own nuclear predication and predication frame, possibly adorned with their own satellites and, in case of embedded predications with a finite verb, introduced by a conjunction. Schematic representation of a complex sentence or predication resembles a tree, with new branches representing embedded predications. Iliad 1.57–58 would appear as follows:


Figure 1: the complex predication


Still, the leap from clause to sentence is a giant one; and the denomination “clause” lies well outside the domain of semantic functions and values. Semantics determine the boundaries of the predications. Functional grammar does not consider an (embedded) predication as equal to a clause, for example (as in this case) to a subordinate clause.

“Clause” is merely useful when dealing with pragmatic functions: οἳ δ᾿ ἐπεὶ οὖν ἤγερθεν looks like a complete subordinate clause, but οἳ ought to be labelled extra-clausal. Identification as extra-clausal follows from the identification of pragmatic functions: topic, focus, theme/tail (= left-/right-dislocation) and setting. Pragmatic functions determine whether constituents are part of the clause or not: Dik [17] labels constituents with function theme/tail and setting as extra-clausal. [18] Strictly “clausal” are the constituents with the function topic and focus, in addition to—of course—the predicate. Since theme, tail and setting are extra-clausal, their appearance marks the boundaries of the grammatical clause. Anything in between—between the constituents with pragmatic functions inside and outside (especially following) the clause—is labelled “extra”, but seen as part (“extension”) of the clause. [19] This “extra” is not the main problem for now, as it merely extends the nuclear predication into an “extended” clause. Neither is the setting, itself clause-like shaped. It is especially the theme/tail-constituent that is left in between: between the pragmatically confined clause and the semantically extended predication. The same goes for non-clausal appositions: how to determine where they belong in paratactic syntax? What structuring effect do they render? In what respect are they different from “extra’s”?

2.2 The Grammatical Clause in Homer

In Homer, as in any other text, analysis of the structure and extent of the grammatical clause would start from main verb forms, the finite verbs. All further text elements can be identified in two different ways. First, as single words or word groups, they represent a particular semantic value in relation to a finite verb and its predicate frame. At the same time, as constituents, they serve pragmatic functions within the predication, either within the clause, or as extra-clausals. From these observations, I present the following four working hypotheses for a refined description of Homeric discourse:

  1. Both semantic values and pragmatic functions are considered as being confined in their scope [20] to the predication.

  2. Whereas semantic valuation is derived from the word groups’ informational value for the predicate, the pragmatic function of constituents organises and structures the presentation of information.

  3. The denomination “clause” as grammatical terminology signifies the smallest coherent predicate-centred unit. As such, the clause can be framed by structuring extra-clausal constituents with pragmatic functions as theme, tail or setting. Or, alternatively, by words or word groups with the semantic value connector or coordinator. The clause itself is free from text elements that are used to structure the complex predication.

  4. The clause itself is merely a grammatically coherent whole, without any regard for structuring value or effect. As a constituent, the clause itself can only serve in the pragmatic functions discourse topic and discourse focus. Primarily, usage of the denomination “clause” is solely on the level of illocutionary force, i.e. on the level of pragmatic functions for predicate frames or predications as a whole. Therefore, the clause must be a coherent, and to a large extent, independent and demarcated utterance.


Together, the above four working hypotheses lead to a strictly pragmatic definition of “clause” with a specific illocutionary force. As mentioned above, illocutionary force indicates the function of the type of predication [21] , always considered as an independent utterance. In Greek, there are four identifiable illocutionary types of predications: predications are either declarative (Iliad 1.2 ἣ μυρι᾿ Ἀχαιοῖς ἄλγε᾿ ἔθηκεν ‘that bestowed innumerable pains on the Greeks’), interrogative (Iliad 1.8 τίς τάρ σφωε θεῶν ἔριδι ξυνέηκε μάχεσθαι ‘who then of the gods brought these two together in strive?’), imperative (Iliad 1.1–2 μῆνιν ἄειδε θεά Πηληιάδεω Ἀχιλῆος οὐλομένην ‘sing, goddess, of the destructive wrath of Achilles, son of Peleus’), or exclamative (Odyssey 1.64 ποῖόν σε ἔπος φύγεν ἕρκος ὀδόντων ‘what a word has escaped the barrier of your teeth?’). Regularly, the illocutionary force is closely related to the type of predication: [22] declarative predications are assertive, interrogative predications are inquisitive, imperative predications are directive. The illocutionary force can be expressed or strengthened by the use of specific moods of the verb or modal adverbative elements. [23] It is likely that intonation was a determining factor for illocutionary force.

Defining the clause as the coherent predicate-centred unit raises questions concerning the position of predicative participles. In a style other than periodic, as is Homer’s, labelling a participle word group as “setting” is not necessarily correct or useful. And even if it were, in itself the setting-like participle word group is inevitably predicate-centred, be it in a way different from being centred on a finite verb. Still, the participle is itself a ρῆμα, a more-or-less finite verb form. [24] Its realisation as a minor phrase [25] resembles that of the unit centred on a finite verb. In the appositional syntax of the Homeric epic, the predicative participle may easily be identified as the centre of an individual clause, instead of as a constituent in a clause. [26] Therefore, it is first priority to look at the identification of the grammatical clause within the specific framework of Homeric discourse.

2.2.1 Homeric discourse and the Homeric grammatical clause

In printed editions of Homer’s Iliad and Odyssey, as well as in any other text, the grammatical clause often visibly takes position within a wider, more complex predication as the result of punctuation. The grammatical clause is being fitted into a larger scale whole by means of predication-structuring elements with semantic values like connectors, coordinators, (correlative) adverbs, and particles. Within the written complex predication, the individual grammatical clause appears to be embedded. Writing tends to be much more compelling in terms of semantically suggested hierarchy and demarcation of predications than spoken discourse. Specifically for Homer, the appositional style makes it even more difficult to use terminology such as complex predication: syntactically, the arrangement of grammatical clauses and extra-clausals is through parataxis.

Complex and embedded predication presuppose, on the one hand, an identifiable status of the finite verb, and, on the other, of the co- or subordinating purpose of connectors and coordinators. Identifiable status of verb forms presupposes some sort of inner-predicational hierarchy: the notion that verb forms within the complex predication contribute to the semantics of the “main” finite verb. Alternatively, the verb forms that are not “main” finite verb cannot be properly understood without the proximity to the finite verb they contribute to. They do not give “complete sense”, καταστροφή, thus completing the utterance. [27] When used without a grammatically appropriate relation to a close-by main finite verb, their occurrence is anacolouth.

In Homer, both the hierarchical status of the verb and the purpose of connectors and coordinators are disputable. In case of the finite verb, it can often not be determined what its hierarchical position in relation to other finite verb forms, and predicative participles, is. Chantraine [28] notices that

les termes de la proposition sont définis par des appositions qui se présentent sous des formes diverses et autonomes.


For Chantraine, the compositional principles of verbal autonomy and syntactical coordination and parataxis are clearly visible practically everywhere in Homer. Recent studies [29] show that the Homeric verbal autonomy and syntactical coordination are proof of its identification as spoken language. The written form in which we enjoy the Iliad and the Odyssey must not lure us into the trap of interpreting them as we would interpret written language. Both are material fixations of spoken language, be it of special speech, mainly due to the use of meter and formulas. As summarised in chapter 1, the analysis of Homeric discourse proves that it is largely appositional, and concerned with continuation despite the possibility of larger scale syntactical cohesion and completion.

This may easiest be seen in the unperiodic arrangement of verb forms. The lack of hierarchy in verb forms, and, in general, the absence of identifiable main predications, obfuscates the transition from one predication to the next. The predicate frame may be extended to the nearest metrical boundaries. [30] In that case, the predicate frame is to be put on a par with Bakker’s intonation unit as the chunk of spoken language. Often, however, it is not. Regularly, the predicate frame extends beyond the nearest position(s) of frequent word end due to the progressive tendency that is so characteristic for the additions of Homeric style. From a prosodic point of view, many such additions are “phrases”. [31] In Iliad 1.3–5, verse-initial οἰωνοῖσί τε πᾶσι is a chunk, as is ἡρώων:

πολλὰς δ᾿ ἰφθίμους ψυχὰς Ἄιδι προίαψεν
ἡρώων αὐτοὺς δὲ ἑλώρια τεῦχε κύνεσσιν
οἰωνοῖσί τε πᾶσι

Iliad 1.3–5

many excellent souls of heroes it sent to the house of Hades; | their bodies it turned into loot for the dogs | and all the birds


Both additions, οἰωνοῖσί τε πᾶσι and ἡρώων, are prosodically marked as phrasing or branching. From the syntactical point of view, there is no question that ἡρώων is to be understood as in hyperbaton to ψυχὰς. The syntactical coherence between οἰωνοῖσί τε πᾶσι and preceding τεῦχε κύνεσσιν is equally clear. In these three verses, the predicative frames of προίαψεν and τεῦχε are not limited in size by the boundaries of the metrical cola. If positions of frequent word end, as metrical boundaries, would indeed be decisive, Homer’s compositional technique would result in a monotonous battering in performance, and needlessly disturbing enjambment.

Inevitably making matters more complicated, the status of other than the finite verb forms is not always different from that of the finite verbs themselves. The use of the predicative participle, mentioned above, requires special attention. Its prosodic realisation regularly underlines its syntactical application as apposition. When used as an apposition to the subject of a finite verb, its status as a ῥῆμα equals that of the finite verb in an example like Odyssey 1.156–157:

αὐτὰρ Τηλέμαχος προσέφη γλαυκῶπιν Ἀθήνην
ἄγχι σχὼν κεφαλήν ἴνα μὴ πευθοίαθ᾿ οἱ ἄλλοι

Odyssey 1.156–157

but Telemachus addressed grey-eyed Athena | and held his head close to hers to prevent the others from noticing


Or even in a well-known formula like Odyssey 1.63:

τὴν δ᾿ ἀπαμειβόμενος προσέφη νεφεληγερέτα Ζεύς

Odyssey 1.63

Zeus the cloud-gatherer gave her an answer and spoke


Easily found are the comparable formulaic verses and expressions in which the non-descriptive “setting” ἀπαμειβόμενος is not present. Its presence would neither benefit nor harm the remaining formula.

Both cases of the predicative participle hardly apply as examples of “setting”. Chantraine [32] points out that the predicative participle, especially the perfect participle, is frequently being used as predicate in Homer. The second example, Odyssey 1.63, resembles the usage of two closely related main verbs, itself reminiscent of hendiadys. Numerous are the examples of two conjoining main finite verbs, as in Odyssey 1.231:

ἐπεὶ ἂρ δὴ ταῦτά μ᾿ ἀνείρεαι ἠδὲ μεταλλᾶις

Odyssey 1.231

Since indeed you ask and question me about this


As mentioned in chapter 1, Chantraine assumes that more complex syntactical organisation like subordination is in a stage of development in Homer. An indication of the development of verbal hierarchy is found in the usage of modes, other than indicative. I cannot subscribe to his point of view. Analysis of the various modes being used in the Iliad and the Odyssey does not show even a slight trace of “development” of paratactic clauses into verbal hierarchy within a predication. Chantraine comments: [33]

La nature même du style homérique et son caractère formulaire ont pu entraîner des aèdes à reprendre des formules anciennes dont le syntaxe attestait un emploi authentique d’un mode, dans un autre passage où cet emploi était moins nécessaire. D’une manière générale, le caractère subjectif de l’emploi des modes, qui relève autant de la parole que de la langue, en rend difficile (ou trop aisée) la justification en un passage donné. C’est dans le commentaire de l’emploi des modes que le grammairien est le plus exposé à solliciter les textes en apercevant une intention là où il n’y avait peut-être qu’un emploi mécanique, et, au pire, un emprunt à un autre passage.


As Chantraine observes, the priority of formulaic usage of expressions and verses over the authentic meaning of the clause’s mode, hinders the constitution of verbal hierarchy. Possibly even to such extent that, combined with Homer’s archaism, both the usage of modes other than indicatives, and of conjunctions, merely shows the “remarquable souplesse” [34] of the progressive movement [35] of Homeric discourse:

Enfin, dans l’étude de la subordination, il faudra étudier à la fois les subordonnants et les modes dont certain (subjonctif, optatif) sont devenus, eux aussi, des signes de subordination. D’une manière générale, le développement de la subordination est comparable à celui de la rection: de même que les prépositions en sont venues à gouverner un cas, de même les conjonctions ont gouverné un mode. Mais, chez Homère, ce développement est à son début. [36]


Chantraine further [37] remarks that

il subsiste chez Homère des restes notables d’un usage archaïque, en particulier dans la parataxe, c’est-à-dire l’énoncé où une proposition se trouve mise sur le même plan que la principale si elle exprime une notion accessoire ou complémentaire qui devrait en «dépendre». Cette construction paratactique dont le grec a toujours aimé se servir est fréquente chez Homère …


It is noteworthy to point out that, apart from the lack of verbal hierarchy, the coordinators and connectors do not suggest a strong sense of hypotaxis:

Même lorsque les propositions sont liées par des particules de coordination, il se pose des problèmes et il apparaît en particulier que leur disposition n’est pas toujours logique. Le poète énonce en premier lieu le fait qui lui semble le plus important ou que ses yeux aperçoivent d’abord, sans s’astreindre à une succession chronologique ou logique. [38]


When stressing the frequency of parentheses, Chantraine adds (353):

Ce mouvement caractéristique s’explique par la nature paratactique du style: les propositions se succèdent les unes aux autres sans que leurs rapports soient nettement définis. L’emploi de propositions coordonnées à la suite d’un terme subordonnant est largement attesté et nous en avons rassemblé de nombreux exemples dans l’étude de la proposition relative. Mais il s’agit là d’un phénomène général qui s’observe à propos de toutes les formes de subordination.


The paratactical apposition of verb-centred predications resembles the non-grammatical “flat” apposition of autonomous words. Coordinators and connectors do not “logically” tie predications together. On the contrary: their usage merely underlines the progressive tendency, the movement that characterises Homeric syntax. Next to the “finite” status of the predicative participle, the lack of syntactical hierarchy between individual verb forms is an indication for the independence of the individual grammatical clause as a syntactical unit.

2.2.2 “Isolating” transition

These two arguments (hierarchical status of the main verb remains unclear; connectives and verb modes do not indicate subordination) support the concept of Homeric discourse as a series of autonomous verb forms. The relation of the verb forms to one another is not a hierarchical one. Comparison to the usage, in classical Greek, of certain coordinators may have suggested such verbal hierarchy in the Iliad and the Odyssey. In Homer, the relation between the verb forms remains a paratactic one. With one important restriction, on my part, on the meaning of “paratactic”: the “succession of autonomous finite verbs” is not a description of the internal organisation of the complex predication. It is in fact a negation of any such internal organisation. The progressive movement continues at least [39] up until a shift in discourse type. [40] It is not further, internally, organised towards verbal hierarchy by any predication-structuring devices like the use of connectors in written discourse. Connectors play a role as providers of “continuation” due to prosodic isolation. When analysing the usage of connectors in Homer, the focus should be on a special category of connectors. There is a category of possible “continuations” that is less obvious when judged by the grammatical standards of written discourse. It is, however, understandable as a form of continuation from the point of view of a listening audience.

In written discourse, an important continuing, and structuring, device is the combination of μέν in the first member of a grammatically coordinated antithesis, with either ἀλλά, ἀτάρ, εἶτα, ἔπειτα, αὖτε, ἠδέ, οὐδέ, τε, καί, τοίνυν, or δέ in the following limb or limbs. The common answer to μέν is δέ. When combined, μέν is said to raise expectations for an antithesis to be composed of at least two clauses; δέ introduces the second (and subsequent) limb(s). Still, what seems to be the rule knows many exceptions. [41] In fact, the use of δέ as answer to μέν is often not adversative at all. In many instances δέ is merely continuative. [42] When used as connective, with or without a preparatory particle in the preceding limb, δέ is continuative rather than adversative. In addition, the continuative usage of δέ without a preparatory particle in the preceding limb is much more frequent than the usage with.

In Homer and Herodotus, the use of connective δέ without a preparatory particle in the preceding limb is predominant. Connective δέ may be used to pile up a considerable number of subsequent limbs. In the opening lines of the Iliad, as throughout the Homeric epics, δέ is merely used to mark the introduction of the next step in the narrative:

μῆνιν ἄειδε θεά Πηληιάδεω Ἀχιλῆος
οὐλομένην ἣ μυρί᾿ Ἀχαιοῖς ἄλγε᾿ ἔθηκεν
πολλὰς δ᾿ ἰφθίμους ψυχὰς Ἄιδι προιαψεν
ἡρώων αὐτοὺς δὲ ἑλώρια τεῦχε κύνεσσιν
οἰωνοῖσι τε πᾶσι Διὸς δ᾿ ἐτελείετο βουλή

Iliad 1.1–5

Sing, Goddess, of the wrath of Achilles, son of Peleus | so destructive, it bestowed innumerable pains on the Greeks; | many excellent souls it sent to the house of Hades | of heroes; their bodies it turned into loot for the dogs | and all birds; the will of Zeus gradually became fulfilled


Regardless of its relation to a preceding protasis, connective δέ redirects the listeners’ attention to the next bit of information. In written discourse, there is a limit to the number of δέ-limbs that can be joined to a protasis. The resulting, more balanced alternation of protasis and connective (or adversative) δέ structures written discourse in a way quite unlike the effect of connective δέ in Homer. In written discourse, preparatory μέν and connective (and adversative) δέ contribute to a division of larger scale units into sentences; in Homer, μέν and δέ do not suggest a sentence-like internal division of discourse.

The Homeric usage of δέ requires new terminology. Apollonius Dyscolus labels it δὲ μεταβατικός, modern scholars, following that lead, speak of δέ transitive. In his study on Homeric enjambment, Bakker [43] points out that the transitive usage of δέ may well be compared with the use of “and” when moving to the next step in presenting a narrative in spoken English. Such use of English “and” (as of Greek δέ) is connective, but not so much coordinative as well additive. [44] The particle keeps the narrative continuing, thus playing various roles as connector. As such, the usage of the particle facilitates the poet’s adding style. The position of δέ in the syntax of the Homeric epics corresponds to that in written classical Greek. Occupying the second position in the clause, the localisation of δέ (connective or non-connective) serves a purpose in marking transition. In his 1990 article, Bakker postulates that the position of δέ might turn a preceding word group into a left-dislocated theme constituent. In subsequent publications [45] he dismisses the terminology, since it suggests an aberration from a standard syntactical realisation.

As a continuing, and structuring, device occupying the second position in the clause, the localisation of δέ serves a purpose in marking transition. The prosodic character of δέ as a postpositive [46] concludes the phonetic word, the appositive group, but only after the initial word of the next syntactical unit has already been introduced. The theme of the subsequent unit has already been introduced in one accentual unit together with δέ. The announcement of this theme is syntactically and phonetically isolated due to the combination of the additive usage and the appositive character of δέ. The location of δέ enables the phonetic word (of which δέ is the closure; for example αὐτοὺς δὲ in Iliad 1.4) to fill the metrical colon between two positions of frequent word end. Similar conclusions may be drawn from the usage of γάρ and ἄρα. [47] These two particles are also postpositive; like δέ, γάρ is not only postpositive, but to an extent enclitic as well. The particles δέ, γάρ and ἄρα all isolate an added theme in a metrical colon: as postpositives they conclude the phonetic word introducing the new theme before expanding this theme into a new clause. Explicit mention ought to be made here of the use of the pronoun ὁ / ὅς with δέ to indicate a change of subject or a topic shift. The combination of the pronoun with the particle γάρ or δέ can signal either a topic shift, or a semi-relative continuation. An example of topic shift may be found in Iliad 1.56–57, [48] of semi-relative continuation in Iliad 1.46–47:

κήδετο γὰρ Δαναῶν ὅτι ῥα θνήισκοντας ὁρᾶτο
οἱ δ᾿ ἐπεὶ οὖν ἤγερθεν ὁμηγερέες τε γένοντο

Iliad 1.56–57

For she felt for the Greeks as she saw them perish, | as for them, when they had thus gathered and come together


In line 56 the subject is Hera who pities the Greeks, expressed in accusative case. The combination of pronoun and particle at the start of line 57 indicates the shift: from the start of line 57 the Greeks are the grammatical subject. In Iliad 1.47 below the shift is not as radical as in Iliad 1.57. The person referred to by means of the pronoun (in combination with the particle) was already the “subject” in the preceding participle clause:

ἔκλαγξαν δ᾿ ἄρ᾿ ὀϊστοὶ ἐπ᾿ ὤμων χωομένοιο
αὐτοῦ κινηθέντος ὁ δ᾿ ἤϊε νυκτὶ ἐοικώς

Iliad 1.46–47

The arrows rattled on the shoulder of the enraged | himself as he moved himself into position: he approached like night


In combination with pronouns the particles δέ, γάρ and ἄρα do not suggest any syntactical hierarchy. The usage of γάρ shows that interpretation of units as subordinate is more often than not highly conjectural. The localisation of γάρ resembles epic τε and the Aeolic particle κε. It also resembles their effectiveness as modifiers. The exact modification resulting from γάρ is immediately clear. What is not clear is whether the usage of γάρ results in parataxis or subordination. [49] The formal syncretism of the relative and personal / demonstrative pronoun makes the issue hardly relevant. [50] Still, there are noticeable differences when it comes to the exact localisation of the various connectors, and of the phonetic word groups they are part of (these differences will be discussed in chapter 5). The phonetic word that fills the metrical colon may well be a chunk itself. [51] The same holds true for the localisation of extended “subordinate conjunctions” (like δ᾿ ἐπεὶ οὖν in Iliad 1.57) in Homer. The continuation, the movement of Homeric discourse, is thus not only visible in the occurrence of discourse markers; it is audible in the isolating characterisation of discourse markers.

2.2.3 Extra-clausal, intra-clausal, and inter-clausal transition

The progressive movement that defines Homeric syntax, or rather replaces a true Homeric “syntax”, presents the listening audience with bits and pieces of information. The audience receives grammatical clauses, which do not necessarily have to be interpreted in a periodic relation to what preceded or to what will follow. New, autonomous ideas are being added as grammatical clauses onto what has been told so far. Adding sometimes implies the usage of “transitional” [52] connectors, coordinators, adverbs or particles, but most often, it does not. Continuation allows the pragmatic positioning of vocatives, and ”causes” that of theme and tail constituents in a way reminiscent of Clark’s semantic “hooks”. The presence, however, of such constituents is not a requirement. For Homer’s audience, having missed one bit of information need not impede the understanding of subsequent bits. Nor will absorption of every bit of information so far necessarily heighten the level of the audience’s understanding with regard to what is about to be told.

On the basis of what I said so far, we can divide the verses of the Iliad and the Odyssey into grammatical clauses, and extra-clausal constituents. I will present the first 16 lines of the Iliad as an analysis in accordance with this division. It is a provisional analysis, though, as it has to be modified based on observation that will be dealt with later in this section. At the start of the Iliad, the perceptible, translated presentation of bits of information may thus be rewritten in the following way, using subscript for punctuation and all non-clausal constituents in the translation:

1       μῆνιν ἄειδε θεά Πηληϊάδεω Ἀχιλῆος
          οὐλομένην ἣ μυρί᾿ Ἀχαιοῖς ἄλγε᾿ ἔθηκεν
          πολλὰς δ᾿ ἰφθίμους ψυχὰς Ἄϊδι προΐαψεν
          ήρώων αὐτοὺς δὲ ἑλώρια τεῦχε κύνεσσιν
5       οἰωνοῖσί τε πᾶσι Διὸς δ᾿ ἐτελείετο βουλή
          ἐξ οὗ δὴ τὰ πρῶτα διαστήτην ἐρίσαντε
          Ἀτρείδης τε ἄναξ ἀνδρῶν καὶ δῖος Ἀχιλλεύς
          τίς τάρ σφωε θεῶν ἔριδι ξυνέηκε μάχεσθαι
          Λητοῦς καὶ Διὸς υἱός ὁ γὰρ βασιλῆϊ χολωθεὶς
10      νοῦσον ἀνὰ στρατὸν ὦρσε κακὴν ὀλέκοντο δὲ λαοί
          οὔνεκα τὸν Χρύσην ἠτίμασεν ἀρητῆρα
          Ἀτρείδης ὁ γὰρ ἦλθε θοὰς ἐπὶ νῆας Ἀχαιῶν
          λυσόμενός τε θύγατρα φέρων τ᾿ ἀπερείσι᾿ ἄποινα
          στέμματ᾿ ἔχων ἐν χερσὶν ἑκηβόλου Ἀπόλλωνος
15      χρυσέωι ἀνὰ σκήπτρωι καὶ λίσσετο πάντας Ἀχαιούς
          Ἀτρείδα δὲ μάλιστα δύω κοσμήτορε λαῶν

Iliad 1.1–16

Sing of the wrath, goddess, (vocative) of Achilles son of Peleus, (?) so destructive: (personal pronoun seemingly used as anaphoric) it bestowed countless pains on the Greeks, yes many: (δέ) excellent souls it sent to the house of Hades of the heroes, as for them selves: (δέ) it made them into loot for the dogs and (τε) all the birds, as for Zeus’s: (δέ) his will gradually became fulfilled, from the moment (ἐξ οὗ δή) the two of them stood up against one another in anger for the first time, (verse-end) son of Atreus lord of man and (καί) godlike Achilles (shift to another illocutionary force), who then of the gods made the two of them come together to fight in strife: (shift to another illocutionary force) Zeus’s and (καί) Leto’s son (?) for it was him: (γάρ) out of anger for the king he sent a foul plague over the army’s camp, (δέ) they fell, the soldiers, because (οὕνεκα) he had dishonoured the priest Chryses, he, the son of Atreus (verse-end), for it was him: (γάρ) he came to the fast ships of the Greeks : (τε) to free his daughter and (τε) bringing countless gifts and? (τε) holding in his hand the ribbon(s?) of far-shooting Apollo around the golden priest’s staff [53] and (καί) he begged all the Greeks, and as for the two sons of Atreus (δέ), especially those two, the two arrangers of the army: (discourse shift)


A stop at the end of Iliad 1.16 is not coincidental: it coincides with a shift in discourse type.

In this representation, the role of δέ, τε and γάρ to mark the boundaries of the clause is evident. At the same time, it is equally clear that they do not function as organisers in complex predications. The postpositive (and enclitic) character of δέ, τε and γάρ shows their respective roles in the development of the flow of ideas. This characterisation of the particles brings out their usage as “audible punctuation” far better than their semantic contribution to predication-structuring. In verses 3, 4, 5, 10 and 16, δέ is merely used, as throughout the Homeric epics, to mark the introduction of the next step in the narrative. The use of the particle redirects the listeners’ attention to the next bit of information. The particle is not meant in any way to structure the predication or restructure the complex predication. Its use does not even suggest a predication limited in size as the result of verbal hierarchy, with every subsequent hierarchical tree as the domain of a new predication. As a means of transition from one clause to the next, δέ itself stands in-between the two clauses but is not part of either. Bakker [54] points out that the transitive usage of δέ may well be compared with the use of and [55] when moving to the next step in presenting a narrative in English. The particle links the various bits of information. As such, the usage of the particle facilitates the poet’s adding style. Still, the position of δέ as “second constituent” puts both the particle itself (as an appositive) and the preceding constituent in a special light. This special light, however, can only be fully appreciated if the extra-clausal status of the connector retains its pragmatic function as theme-constituent: in other words, as long as its prosodic “isolation” separates the connector from the grammatical clause. As a connector, δέ is part of the developing clause if the particle is elided. This observation must result in modification of the provisional analysis of Iliad 1.1–16: semantically dislocated constituents ending in an elided particle, cannot retain their pragmatic function as an extra-clausal. Only if the connector is not elided, is the preceding word to which δέ is postpositive not a part of the clause that is about to start. [56] Nor is it of the preceding clause, despite the clause’s apparent ability to extend by means of “extra’s”: the constituent and its subsequent postpositive are truly “on their own”. Together, they constitute the completion of a chunk that is audibly separated from what follows appositive δέ. Depending on speech rates, the position and location of δέ audibly cuts the narrative into pieces that include ones that are not predicate-centred in themselves. Often a verb form appears to be singled out, as in formulaic λῦσε δὲ γυῖα ‘and loosed his limbs’. Even more often (as is the case in, for example, Iliad 1.5), the particle is elided. It is then no longer capable of turning the phonetic word into an extra-clausal. The elided phonetic word is drawn into the clause:

νίκη δ᾿ ἐπαμείβεται ἄνδρας

Iliad 6.339

Victory randomly finds its way to men


The closure of the appositive group by means of an enclitic particle turns the “transitional” constituent into a separate phonetic word. When elided, the particle and the constituent preceding it are the initial part of a chunk. As appositive group, however, they remain “sentential-prepositive”: the appositive group is itself syntactically proclitic to the subsequent grammatical clause that is its scope (indicated as [ …]) [57] :

[Ἀργεῖοι δὲ → μέγα ἴαχον] [ἐρύσαντο δὲ → νεκρούς]

Iliad 17.317

The Greeks cheered loudly, and started dragging the corpses away


It depends therefore on the realisation of the appositive group as an independent phonetic word whether it functions as extra-clausal transitional constituent, or, in case of elision of the particle, as constituent within a grammatical clause (intra-clausal). The provisional analysis of Iliad 1.1–16 should hence be modified with respect to line 5: the dislocated constituent in the line is in fact intra-clausal.

Like δέ, the connector γάρ is postpositive and tends to occupy a position as sentential “second constituent”. Like δέ, and other appositive particles, [58] the occurrence of γάρ actually defines the transitional constituent, and hence the phonetic word. As with δέ, γάρ can turn the preceding constituent, especially when this constituent is a pronoun, into a theme constituent. Notwithstanding that, γάρ has another semantic value than δέ. Where δέ functions as a transitional particle merely used to further the continuation of discourse, γάρ suggests a connection on the level of causality. [59] Scholars have noticed the often lacking logic in the presupposed connexion of thoughts:

Compression of thought is often the source of difficulty, and formal exactitude can then be achieved by supposing an ellipse … this, though a convenient method of exposition, is psychologically somewhat misleading. [60]


It is indeed hard to see what causal connexions underlie Iliad 24.66–70 (or, alternatively, Herodotus III, 80–82 [61] ):

οὐ μὲν γὰρ τιμή γε μί᾿ ἔσσεται ἀλλὰ καὶ Ἕκτωρ
φίλτατος ἔσκε θεοῖσι βροτῶν οἳ ἐν Ἰλίωνι εἰσίν
ὣς γὰρ ἐμοί γ᾿ ἐπεὶ οὔ τι φίλων ἡμάρτανε δώρων
οὐ γάρ μοί ποτε βωμὸς ἐδεύετο δαιτὸς ἐίσης
λοιβῆς τε κνίσης τε τὸ γὰρ λάχομεν γέρας ἡμεῖς

Iliad 24.66–70

For the honor will not be the same, but Hector too | was dearest to the gods of the mortals who are in Troy, | for so he was to me as he did not fail in any way in gifts dear to me, | for never at any time was my altar lacking in the equal banquet, | the drink offering and the savor of burnt offering, for that we have received as our privilege


The connective use of the combination γε and ἄρ is far more prominent than the causal value. [62] Rather, the combined use of the postpositive deictic particle γε and resumptive ἄρ highlighting the value of the preceding constituent, together explain the seemingly explanatory character of the subsequent clause. I argued that the particle is better described as “transitional”. Hence the ability of γάρ to turn the preceding word or word group into a theme constituent: it resembles a similar ability of δέ. The main difference between the two connectors, as the passages quoted from the Iliad show, is that γάρ implies topic continuation, whereas δέ results in topic shift. The topic continuing effect of γάρ stems from two aspects: the extra stress on the preceding constituent rendered by γε, but essentially the resumptive value of ἄρ, turning the preceding constituent into a continuing topic. On the other hand, δέ signals a change of camera-position.

The postpositive character of both γάρ and δέ is functional in that it creates an audible separation of the phonetic word from what follows, thus emphasising not the constituent preceding γάρ and δέ, but rather the pragmatic value of γάρ and δέ itself. As “rightbranchers”, both conclude the phonetic word before continuing into the informational unit that forms the subsequent clause. The use of the connector τε is not in all instances comparable to that of the connectors γάρ and δέ. The latter two turn the preceding word or word group into an extra-clausal constituent preparing for the subsequent clause. The former is used to play down the disturbance of transition to another addition within the grammatical clause. [63] The connector τε is, in other words, an inter-clausal transition.

2.2.4 “Subordinating” transition?

Apart from καί, δέ and τε the first 16 lines of the Iliad contain a few conjunctions that, in classical Greek, are undoubtedly subordinating. In Iliad 1.6, the combination ἐξ οὗ is taken as the introduction to a subordinate clause. [64] In Iliad 1.11, οὕνεκα introduces a clause that seems to function as the explanation of the start of the plague and the subsequent suffering of the army as mentioned in Iliad 1.10. Both conjunctions are built on the pronoun ὁ/ὅς that appears as what can be called in classical Greek, both demonstrative and relative. [65] The problematic identification of the type of pronoun has been dealt with in the description of syntax, especially hypotaxis, by Chantraine in chapter 1. In Homer, the difference between the two denominations of the pronoun is often irrelevant or even non-existing. [66] Nonetheless, the form of the pronoun (οὗ/οὑ(-) versus demonstrative τοῦ/του(-)) suggests usage as relative. Still, even if the conjunctions are analysed as subordinating, the exact correspondence with a preceding piece of information can remain unclear, as in Iliad 1.6. When not elided, the subordinating conjunctions resemble καί, δέ and τε in that the conjunctions are isolating: they cannot span the subsequent clause. Conjunctions introduce the subsequent clause and, as such, represent the transition from one clause to the next. Their occurrence signals the beginning of a clause. As a consequence, their presence can result in the apparent conclusion of the preceding clause, but they are never being awaited nor expected. [67] In other words, the grammatical clause itself invariably remains open-ended when it comes to its potential length. In Iliad 1.6, ἐξ οὗ is a fine example. There is no question about which grammatical clause ἐξ οὗ introduces. But the ongoing and persistent call for a correspondence with μῆνιν ἄειδε θεά in Iliad 1.1 sufficiently shows to what extent the preceding informational unit or “clause” can be expanded in the opinion of various scholars. [68]

Then again, when used in what looks like a periodic utterance, the conjunction does create expectations, at least for a main clause to follow. But syntactical difficulties abound. In an example like Iliad 1.37–41, the first periodic subordinate clause (line 39) may easily be understood as dependent, if dependent at all, [69] on the imperative κλῦθι in line 37:

κλῦθί μευ ἀργυρότοξ᾿ ὃς Χρύσην ἀμφιβέβηκας
Κίλλαν τε ζαθέην Τενέδοιό τε ἴφι ἀνάσσεις
Σμινθεῦ εἴ ποτέ τοι χαρίεντ᾿ ἐπὶ νηὸν ἔρεψα
ἢ εἰ δή ποτέ τοι κατὰ πίονα μηρί᾿ ἔκηα
ταύρων ἠδ᾿ αἰγῶν τόδε μοι κρήηνον ἐέλδωρ

Iliad 1.37–41

hear me please, Silverbow, who stand protective over Chryse | and holy Killa and over Tenedus rule with iron fist, | Smitheus, if ever I covered a temple with a roof pleasing to you, | or if ever I gave you full ration when burning the fat shanks | of bulls and goats: fulfil this one hope for me


If it was not dependent on the imperative, alternative expectations are at first frustrated by the addition of another subordinate clause in Iliad 1.40–41a (ἢ εἰ δή ποτέ τοι κατὰ πίονα μηρί᾿ ἔκηα | ταύρων ἠδ᾿ αἰγῶν). Finally, there is no preparatory effort to create a correlation between the seemingly periodic subordinate clauses in lines 39–41a and the imperative κρήηνον in 41b, unless it were the use of εἰ itself as in, for example, εἰ δ᾿ ἄγετε. Then again, in that case the remaining clauses in lines 39 and 40 should be analysed as paratactic main clauses, as εἰ merely prepares for the imperative and not for a subordinate clause. There is not such a huge difference between this passage and passages with an independent [70] protasis. As a result it seems reasonable to consider the conjunction (in combination with an appositive particle) an extra-clausal constituent with an identifiable pragmatic function. [71]

Following Chantraine, [72] I think it is safe to assume that in Homer the development of particles into subordinating conjunctions is still in its infancy. The Homeric epics present the audience mainly with instances of the original usage of the particles in parataxis.

2.3 The Grammatical Clause and Enjambment

Having discussed the various types of enjambment (in chapter 1) and their value as acknowledgement of metrical-syntactical disparity, and having presented the model of the grammatical clause as a description of Homeric syntax, there remains the issue of “clausal enjambment”, the type of enjambment that stresses the metrical-syntactical synchrony. If this type has nothing to do with disparity or with emphasis within the grammatical clause, is there such a phenomenon as “clausal enjambment” at all in Homer? Maybe there is, but my use of the terminology will differ from that used by Higbie.

As shown, the whole-line formula gradually became to be understood as “whole-clause formula”. The whole-clause formula was seen as the basic unit of Homeric composition. Homer would then have added one line after another, at the same time adding one thought after another to form syntactically complete “sentences”. The mere use of the word “adding” to describe this process of composition suggests that hardly any expectations, or none, are raised by the verse-end straddling clauses that make up the larger part of these “sentences”. Certain exceptions were allowed: reasonable expectation that a sentence will continue

can either be due to the order of the clauses […] or because some element in the first clause, a particle or word, sets up an expectation for the second clause. [73]


These instances call for what Higbie calls “clausal enjambment”. She then sub-divides it into “clausal-internal enjambment” and “clausal-external enjambment”. Clausal-internal enjambment separates the constituents of a grammatical clause, though they are closely connected by means of correlative adverbs like οὔτε and ἤ / ἠέ. The verse-end straddling clause may be grammatically complete in itself:

οὔτέ ποτ᾿ ἐς πόλεμον ἅμα λαῶι θωρηχθῆναι
οὔτε λόχονδ᾿ ἰέναι σὺν ἀριστήεσσιν Ἀχαιῶν
τέτληκας θυμῶι

Iliad 1.226–228

Never to don armour together with the soldiers in preparation for battle, | not even to join in an attempt to ambush with the best of the Greeks | you were courageous enough in your heart


Clausal-internal enjambment is rare in Homer. Clausal-external enjambment, on the other hand, with a dependent clause preceding a main clause, is common:

Σμινθεῦ εἴ ποτέ τοι χαρίεντ᾿ ἐπὶ νηὸν ἔρεψα

Iliad 1.39

Smintheus, if ever I covered a temple with a roof pleasing to you


One would expect a main clause to follow. Probably intonation supported the audience’s sensation that the end of Iliad 1.39 was by no means the end of the “sentence”. The same sensation was felt by the audience when they listened to Iliad 8.66:

ὄφρα μὲν ἠὼς ἦν καὶ ἀέζετο ἱερὸν ἧμαρ

Iliad 8.66

As long as it was morning and the sacred day still increased


For the audience, verse-initial ὄφρα (followed by μὲν in second position) and intonation of this line were probably indicators for a main clause to follow. [74] Such a sentence structure even seems to fit the definition “formulaic”. [75] Of course, subordinate clauses that end in clausal enjambment, that is, requiring a main clause to follow in a subsequent verse, are not always followed by a main clause to complete a meaningful sentence. Several lines in the Homeric poems apparently feature a “subordinate” clause without continuing into a main clause after enjambment:

ξεῖν᾿ ἐπεὶ οὔτε κακῶι οὔτ᾿ ἄφρονι φωτὶ ἔοικας
Ζεὺς δ᾿ αὐτὸς νέμει ὄλβον Ὀλύμπιος ἀνθρώποισιν

Odyssey 6.187–188

Stranger, as you do not resemble a coward or an unwise man,
Zeus himself, the Olympian, grants a share of fortune to humans


The speaking character, Nausicaa, proceeds with a general statement (Ζεὺς δ᾿ αὐτὸς νέμει ὄλβον Ὀλύμπιος ἀνθρώποισιν) that meets the syntactical requirement of the expected main clause but should, in the semantic and pragmatic approach, be explained as a parenthesis. In line 190 (καί που σοὶ τάδ᾿ ἔδωκε σὲ δὲ χρὴ τετλάμεν ἔμπης ‘I think he gave you this as well, and you must endure it till the end’) all hope of an appropriate main clause is lost since a new main clause starts. Nausicaa will actually start her “ἐπεὶ-sentence” once more, in line 191 (νῦν δ᾿ ἐπεὶ ἡμετέρην τε πόλιν καὶ γαῖαν ἱκάνεις ‘now since you have arrived at our citadel and country’), this time with a more appropriate continuation. One may assume that the force of the verse end as a clausal divider, preparing for a main clause to follow, was stronger than the compositional need for a logical and appropriate sentence here—both in the good MSS and in the testimonia: Plutarch (De profect. in virt. 82 e) cites, between 187 and 188, as 187a:

οὖλέ τε καὶ μέγα χαῖρε θεοὶ δέ τοι ὄλβια δοῖεν

greetings and be most welcome; may the gods grant you happiness


Plutarch almost quotes an existing line from the Odyssey, [76] to end in an apodosis.

There are other examples of verse-end straddling “subordinate” clauses without an apodosis, like Odyssey 3.103, Odyssey 4.204, Odyssey 8.236, Odyssey 14.149, Odyssey 17.185 etc. The structure of these syntactically rambling “sentences” varies: the subordinate clause is followed by a relative clause, another subordinate clause, a participle, a main clause in parenthesis or a combination of any of the above. In all cases, however, the subordinate clause is followed by at least one other verse and the end of the “incomplete sentence” always coincides with the end of the dactylic hexameter. This suggests that the force of the verse end as a clausal divider has much in common with the force of the verse end as an intonation boundary. So much, that the former can be mistaken for the latter and thus create a syntactically open-ended grammatical clause. The examples cited above show that several such incomplete lines were maintained throughout the manuscript tradition. Consideration for the “mistake” makes Plutarch’s continuation in Odyssey 6.187a unnecessary and superfluous.

The examples of subordinate clauses ending in enjambment at verse end, but without apodosis, are not the only examples in Homer showing a preference for combining clausal division with verse end. Odyssey 23.225–230 provide an extreme version of such a subordinate clause followed by a relative clause, an embedded relative clause, a parenthetic main clause and a proper main clause. The proper main clause starts at the beginning of line 230, so the true clausal enjambment coincides with the verse end of line 229:

νῦν δ᾿ ἐπεὶ ἤδη σήματ᾿ ἀριφραδέα κατέλεξας
εὐνῆς ἡμετέρης ἣν οὐ βροτὸς ἄλλος ὀπώπει
ἀλλ᾿ οἷοι σύ τ᾿ ἐγώ τε καὶ ἀμφίπολος μία μούνη
Ἀκτορίς ἥν μοι δῶκε πατὴρ ἔτι δεῦρο κιούσηι
ἣ νῶιν εἴρυτο θύρας πυκινοῦ θαλάμοιο
πείθεις δή μευ θυμὸν ἀπηνέα περ μάλ᾿ ἐόντα

Odyssey 23.225–230

But now, since you have listed the unmistakable signs | of our bed, which no other mortal has seen, | but just you and me and one single maid-servant, | Aktoris, whom my father gave me long ago when I moved here—| she used to guard the doors of the well-built bed room for the two of us—| you finally convince my heart though it is very suspicious


This extraordinary example goes to show, I argue, what is evident throughout the Iliad and the Odyssey: when the subordinate clause precedes the main clause, the boundary between the subordinate clause and the main clause usually coincides with the verse end. Odyssey 23.225–230 further strengthens the idea that in such cases the poet will postpone the start of the main clause until the start of a verse. If that is indeed the case, the significance of the “clausal-external enjambment” is the same as that of the punctuation used in prose to mark the boundary between the subordinate and the main clause, and that of the audible pause in prose when read aloud and, in spoken language, the audible rise in tone at the beginning of the main clause after the audible fall of tone at the end of the subordinate clause. “Clausal external enjambment” serves the same structuring purpose as these phenomena in prose. [77] The intonation boundary is as audibly non-disruptive and—despite the verse end—as insignificant as these prose phenomena.

2.4 Conclusion: The Grammatical Clause is a Suitable Tool for the Analysis of Homeric Discourse

In this chapter I have examined the clause in Homer as a grammatical clause, an approach that covers both the metrical-unit clause and the clause that develops from the alignment of chunk-like metrical cola. As such, the clause is not similar to a sentence or to parts of the sentence. It is rather the smallest grammatically coherent unit centred on a predicate. The grammatical clause has a pragmatic function and illocutionary force on its own, regardless of any structuring function within a larger whole. In Homer, the grammatical clause is open-ended, and framed by extra-clausal constituents. Its independence and lack of structuring function are reflected in Homer’s appositional style.

I applied the model of the grammatical clause to both the metrical-unit clause and the clause that develops from alignment of units to show that Homeric syntax is best described as a movement featuring grammatical clauses and extra-clausal transitional constituents. From the grouping together of clauses and extra-clausal constituents—some syntactically transitional, others preparatory or additive to a verb-centred unit—larger scale grammatisation emerges, sometimes seemingly developing into subordination and complex predications. Throughout the description of Homeric syntax as a movement featuring grammatical clauses, enjambment was considered as the acknowledgment of verse-end crossing grammatisation.

In chapter 3, I will investigate the metrical and phonological shape of Homer’s grammatical clause. If there is a second type of phrasing that features rather free clause-formation, there may be metrical and phonological evidence for the different ways in which clauses terminate.

Footnotes

[ back ] 1. Cf. e.g. Clark 1997:24-30.

[ back ] 2. The model was introduced into linguistics by Tesnière 1959, and adopted by Chafe 1970, Helbig 1971, Korhonen 1977, Lyons 1977:147–154, 434–438, Matthews 1981, Allerton 1982. The terminology used here is that of S.C. Dik 1997 and Pinkster 1990. For an overview of functional grammar in relation to the Homeric epics, see Edwards 2002:9–13. The most recent attempt to define the clause is H. Dik 2007:22–28.

[ back ] 3. Sentence includes complex sentence-structures comprising of at least one main clause and one subordinate clause. Functional grammar uses the technical term predication to indicate the sentence; in Homer, as I will explain below, predication is preferably understood as equivalent to clause.

[ back ] 4. Though the subject of an imperative is often included in the verbal form itself, as it is here. The vocative θεά is an apposition to the included subject.

[ back ] 5. For the intransitive use of ἄειδω, cf. Iliad 1.604, 2.598, 4.125, Odyssey 1.154, 1.155, 1.325, 19.519, 21.411; for ἐνέπω, cf. Iliad 11.643, Odyssey 23.301.

[ back ] 6. The terminology has been applied to the grammar within Homeric metrical units in Bakker & Fabricotti 1991.

[ back ] 7. Bartsch 1972; Verkuyl 1972; S.C. Dik 1997(I):191–209.

[ back ] 8. S.C. Dik 1997(I):313–338, (II):401–405.

[ back ] 9. Greenbaum 1969; Meier-Fohrbeck 1978; Quirk 1972; S.C. Dik 1997(I):132–140.

[ back ] 10. The vocative is not regularly treated as a disjunct, but I would argue that comparison with e.g. the description of parenthetic verbal forms to identify illocutive functions and discourse types, justifies the identification of vocatives as disjunct satellites; cf. Lyons 1977:738.

[ back ] 11. Contra Chantraine 1953(II):297. I do not believe that there is a fundamental difference between the two final clauses that should lead to identification of ὄφρα τελέσσω as an argument. The correspondence in the prosodic patterns of lines Iliad 1.523–524 is an extra ground to declare impersonal use of μελέομαι unlikely here.

[ back ] 12. Chantraine 1953(II):351–364; Bakker 1997b.

[ back ] 13. Cf. Chantraine 1953(II):345–350.

[ back ] 14. Cf. the use of e.g. τοι in Odyssey 11.252.

[ back ] 15. Cf. the discussion on sentence-structure in these lines in Kirk 1990:176–178.

[ back ] 16. Slings 1992.

[ back ] 17. S.C. Dik 1997(II):381: ‘[these constituents] are typically set off from the clause proper by breaks or pause-like inflections in the prosodic contour’.

[ back ] 18. To which I would like to add all non-clausal appositions, including the vocative.

[ back ] 19. Since the “extra’s” are normally found only following the pragmatically labelled constituents within the clause, their identification makes the clause highly “open-ended”. For the word order of the constituents within the clause, see H. Dik 1995; 2007.

[ back ] 20. Devine & Stephens 2000:72–73.

[ back ] 21. Franck 1980; Lyons 1977; Searle 1969; 1976.

[ back ] 22. An example of the opposite is the so-called rhetorical question: an interrogative predication with assertive illocutionary force.

[ back ] 23. Both the usage of specific modes and modal adverbs create the possibility to identify optative and concessive illocutionary force as well.

[ back ] 24. Cf. Ruijgh 1987:342n62.

[ back ] 25. Cf. Devine & Stephens 1994:387–388.

[ back ] 26. Chantraine 1953(II):329.

[ back ] 27. Aristotle Rhetoric 1409b8, 1409a29–34.

[ back ] 28. Chantraine 1953(II):12.

[ back ] 29. Slings 1992; Oesterreicher 1997; Bakker 1997b; 2005.

[ back ] 30. Bakker 2005:11.

[ back ] 31. Cf. Bakker 1997b.

[ back ] 32. Chantraine 1953(II):321, with examples.

[ back ] 33. Chantraine 1953(II):205.

[ back ] 34. Chantraine 1953(II):234.

[ back ] 35. Bakker 2005:69. Cf. chapter 1.

[ back ] 36. Chantraine 1953(II):235.

[ back ] 37. Chantraine 1953(II):232.

[ back ] 38. Chantraine 1953(II):351–352.

[ back ] 39. Cf. Aristotle’s remark in Rhetoric 1409a29–31.

[ back ] 40. Possibly even further, if direct speech is considered as embedded predication.

[ back ] 41. Denniston 1954:369.

[ back ] 42. Denniston 1954:177.

[ back ] 43. Bakker 1990; cf. 1997a:54–85.

[ back ] 44. Terminology coined by Halliday & Hassan 1976:238.

[ back ] 45. Cf. Bakker 1997a.

[ back ] 46. Koster 1953:51–52; Devine & Stephens 1978. Devine & Stephens 1994:354–355 discuss the possibility of clisis for the particle. Clisis seems to be well possible at higher rates of speech. At lower rates, as evidenced in the musical settings, the grave accent of the non-lexical appositive does not seem to be part of the rising trajectory, as opposed to other word-final grave accents.

[ back ] 47. Ruijgh 1990; Devine & Stephens 1994:422–423; Bakker 1993:15–25; 1997b; 2005:92–113.

[ back ] 48. The word group δ᾿ ἐπεὶ οὖν may be considered as an expansion of nuclear δέ, see Bakker 1993:15–25.

[ back ] 49. Cf. the note on Iliad 7.73 in Denniston 1954:72–73.

[ back ] 50. That is, for the Homeric epic. In later Greek, the use of a relative pronoun instead of a personal pronoun to start a main clause leads to the identification of the relative pronoun as (part of) a connecting device.

[ back ] 51. Cf. the remarks on extra-clausal constituents in Hellenistic hexameters in Van Raalte 1986:69–70.

[ back ] 52. I would argue that connectors and coordinators regularly are best described as adverbial particles.

[ back ] 53. Or rather: ‘he wanted to free his daughter … he brought countless gifts … he held in his hand the ribbon(s) of far-shooting Apollo around the golden priest’s staff’.

[ back ] 54. Bakker 1990.

[ back ] 55. In my view English progressive ‘and’ may better be compared to Homeric καί, when preceded by an audible pause. In Bakker 1997b, Bakker describes καί as additive. See further discussion below.

[ back ] 56. Devine & Stephens 1994:303: ‘they are moved into that slot out of the phrasal constituent in which they would have remained had they been nonclitic words.’

[ back ] 57. Devine & Stephens 1994:303.

[ back ] 58. In accordance with Wackernagel’s Law, cf. Ruijgh 1990.

[ back ] 59. Denniston 1954:56–74.

[ back ] 60. Cf. Denniston 1954:61.

[ back ] 61. Denniston 1954:63 notes that γάρ ‘refers, not to the immediately preceding sentence, but to something further back. This looseness of structure is characteristic of Homer and Herodotus: the Attic examples are few, and not remarkable.’

[ back ] 62. Cf. τάρ < τε + ἄρ.

[ back ] 63. For an overview of the usage of τε, with the exception of so-called epic τε, see Denniston 1954:495 ff.

[ back ] 64. Though the dispute continues on whether the subordinate clause is depending on the imperative in line 1 or ‘Zeus’s plan’ (according to Aristarchus) in line 5. For recent discussion see Bakker 1997b and Lacatz 2000 ad loc.

[ back ] 65. In the Iliad and the Odyssey, the relative use of οὕνεκα seems most easily defendable when applied as correlative as in e.g. Iliad 1.111, 3.403, 13.727, 14.192 and Odyssey 13.332.

[ back ] 66. Chantraine 1953(II):166 treats the pronoun ὁ as a rudimentary Homeric article, the result of further development of the Mycenaean demonstrative pronoun; as such, he recognises the ability of the article to retain either demonstrative or relative value: ‘L’usage homérique appelle diverses observations qui s’expliquent par le caractère originellement démonstratif du pronom. Au nominatif féminin singulier (ἥ) et au nominatif pluriel masculin et féminin (à l’exception de τοί, ταί), il n’est pas possible de distinguer entre le thème de l’article et celui du relatif proprement dit : l’identité de ces formes a pu aider à l’extension du thème d’article à l’emploi de relatif. Les formes atones de l’article (ὁ, ἡ, οἱ, αἱ) sont accentuées lorsqu’elles équivalent au relatif (cf. A 388, etc…). Dans plus d’un exemple, il est malaisé de déterminer si l’article est proprement l’équivalent du relatif ou s’il est démonstratif. A la vérité, la question ne doit pas être tranchée, mais les exemples montrent l’origine de l’emploi « relatif » de l’article. Ces cas ambigus se trouvent surtout dans l’Iliade. […] Ajoutons que la tradition hésite entre ὅ et ὅς.’

[ back ] 67. All correlative adverbs can be used without correspondence.

[ back ] 68. I agree with Aristarchus that line 6 gives the starting point of Zeus’s plan, on the basis that ἐξ οὗ not only introduces the subsequent clause but concludes a separate grammatical clause ἐτελείετο βουλή as well; in parataxis with a high level of nonconfigurationality (Devine & Stephens 2000:142–153, especially 145) it is too farfetched, I think, to suppose a correspondence between grammatical clauses that are not linked by one and the same pragmatic constituent.

[ back ] 69. Cf. the use of εἰ in combinations like εἰ δ᾿ ἄγε, εἰ δ᾿ ἄγετε, εἰ δέ, reducing εἰ to an introduction of an imperative.

[ back ] 70. Cf. Chantraine 1953(II):351–352.

[ back ] 71. Functional grammar does not yet have terminology to cover such a pragmatic function. One might think of “transition-constituent”, reorganising the audience’s expectations and hence facilitating the addition of new informational units.

[ back ] 72. Chantraine 1953(II):232.

[ back ] 73. Higbie 1990:41.

[ back ] 74. In a line like Iliad 15.547 ὄφρα is in third position. The particle δέ is in second position, equaling ὄφρα to the adverb τόφρα (‘meanwhile’ cf. Kühner &Gerth 1963(II):228). The same combination of intonation cum correlative adverb as the first word of the “sentence” to signal a following main clause, occurs in examples like Iliad 1.53: ἐννῆμαρ μὲν ἀνὰ στρατὸν ὤιχετο κῆλα θεοῖο (‘nine days the god’s arrows swept through the army’), whereas ἐννῆμαρ occurs as a temporal adverb, but without a corresponding adverbial expression following it in the next line, in Iliad 12.25.

[ back ] 75. Clark 1997:22 speaks of “correlative couplet formulas” and cites as examples Iliad 8.66–67 and Iliad 11.84–85.

[ back ] 76. Odyssey 24.402 reads μάλα in stead of μέγα (Vulgate). The line cited by Plutarch is identical to Hymn to Apollo 466.

[ back ] 77. According to this approach, “clausal external enjambment” occurs two times in the sample of the Iliad (see Appendix), and three times in the Odyssey (if phrasal metarrhythmisis is accepted in Odyssey 1.83).