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.

5. Breaking the Measure: Grammatisation through Prosody

φανήσεταί σοι λόγος εἷς εἰρόμενος

Dionysius of Halicarnassus On Literary Composition 26/141.3 U-R.

You will think it all one continuous piece of prose

5.0 Introduction

Many verses and clusters of verses in Homer show a level of syntactical coherence over, or beyond, the boundaries of the metrical phrase, sometimes even over the boundaries of several metrical phrases. At the same time, as I have shown, there is syntactical coherence involving several minor phonological phrases; the identification of the minor phonological phrase is only one step in the analysis of Homeric larger scale composition. Focusing on the metrical phrase, both the verse and the colon, as a unit of composition and performance, reduces listening to the Homeric epic to an intolerable experience: the listening audience receives staccato-like audible chunks that offer bits and pieces of larger scale semantically coherent wholes. In chapter 4, I have provided the theoretical basis to lessen the value of metrical cola as an expressive medium. Starting from the analysis of the possibilities meter offers for phonetic disruption caused by word- and phrase-final lengthening, I have identified primary and secondary pauses. These pauses are audible and often do not necessarily correspond with metrical boundaries that demarcate cola, not even the verse end. A pause disturbs the metrical repetition. A primary pause is so disturbing that it can only be localised on a metrical element that allows for additional phrase-final lengthening. [1] The primary pause is found where word end is in a heavy syllable, not just a segment, on the metrical thesis. To the ear, the primary pause causes a change of direction in rhythm: the rising word end causes metarrhythmisis to anapaests despite the flow of dactyls. The secondary pause is found on the arsis at the end of the phonetic word, or on the right branch of minor phonological phrases that are internally tied together through sandhi, resyllabification, elision, and shortening. Just like the primary pause, the secondary pause means phonetic word end in a syllable. A secondary pause is only a very mild disturbance: the realisation of the secondary pause depends on rates of speech, as well as the identification as non-breathing pause. At higher rates of speech, secondary pauses become practically imperceptible. Enclitics, as word-final syllables leaning on the preceding lexical word and “clitic” in both prosodic and syntactic respect, signal the secondary pauses that are the first to be perceived at slower rates of speech. My working hypothesis was that as primary and secondary pause make phrases audible in the continuous flow of dactyls of the Iliad and the Odyssey, they will suggest patterns that remind us of the hierarchical constituency of spoken language: the primary pauses recall the boundaries of the major phonological phrase, the secondary pauses that of the minor phonological phrase. As the primary pause offers the opportunity for breathing, it concludes the major phonological phrase. When judging the phonetic disturbance word end on the arsis causes within the major phonological phrase, it would be theoretically sound to not presuppose any serious disturbance at all, but allow for mild disturbances. That includes phonological phrases that straddle the verse end.

The progressive movement of Homeric syntax suggests that the development of larger scale syntactical coherence is elicited, or at least furthered, by prosody. It is not the metrical phrase, verse or colon, that dictates the boundaries of grammatisation, though its boundaries frame grammatical organisation to some extent. Larger-scale organisation is framed by positions of frequent word end; not by every such position, but by some, whereas others are being straddled. I have shown that in this prosodic organisation, the boundaries demarcating coherence stem from audible pause. As audible pauses come in two qualities, a stronger and a milder pause, the audible punctuation of the Iliad and the Odyssey results in a patchwork of rather freely formed grammatical clauses and extra-clausal constituents. In this chapter, I will show that the patterning that represents the greater variety in phrases causes the addition and continuation that characterise the syntactical movement. When that happens, progressive movement of syntax is the result of the primary/secondary-pause alternation. There is reason to think then of (part of the) Homeric syntax as emerging from the reorganisation of the grammatically appositional minor-phrase metrical cola into a larger scale syntactic hierarchy of variable major phonological phrases.

The two separate aspects of verse-making, syntax and prosody, affect one another in their realisation. The coherence of one may, or may not, coincide with the coherent phrasing of the other. If there is disparity between the two, the boundaries of one disrupt the coherence of the other. The “audible-pause” test will show that the two coincide on minor phrase scale: the (extra-)clausal and the minor-phrase phonological phrasing share the same pattern. I will also present examples that clarify the prosodic determination of major phrase syntactical reorganisation of minor phonological phrases.

5.1 Audibly Continued Grammatisation

The description of Homeric syntax as a progressive tendency is in accordance with the observation that Homeric syntax has few ways to reach completion, but many ways to add and to continue. The text of the Iliad and the Odyssey as we have it shows addition and continuation in various forms. Among these forms, Chantraine 1953 and Bakker 1997 list connectors, subordinating conjunctions, hyperbaton, appositions and attributive elements. My aim will be to provide examples, that show to what extent these discourse-relations are prosodically supported, or supportive, in their role as “continuation-providers”. Contrary to what we find in written literature, these means of syntactical movement are not meant to provide the internal structure for a smaller or larger scale unit within the wider frame of the discourse: there is no “periodic” unit that consists of a certain maximum of hierarchically coordinated intonation units – only to be followed by the next, more or less similar, periodic unit. Characteristic of a periodic smaller or larger scale unit is the progressive determination due to syntax: subsequent, hierarchically organised limbs often determine the exact syntactical function and value of the preceding limbs, for example in terms of subordination. In Homer discourse, addition and continuation hardly ever affect what came before. Nor does its syntax force the listener to interpret additions and continuations as explicable only depending on what was said just before. Still, there is a form of “determination” in Homeric syntax: the divergence of, on the one hand, the metrical location of additions and continuations, and, on the other, the distribution of audible pauses, makes syntax emerge from the phonological reorganisation of metrical cola. This effect of the compositional process, an audible effect, stands at a great distance from the notion of “sentence”. It must, however, be remembered that this audible effect has fallen victim to increasing prosodic rigidity rather soon after Homer: or perhaps in Homer, as the differences between the Iliad and the Odyssey in the applications of Wernicke’s law (chapter 4) suggest.

Audible pauses suggest another type of “structuring”: a type that involves prosodic units whose alignment stems from different types of glue, differently realised phonetic disturbances. The disturbances in the examples of this chapter seem to reflect different syntactical transitions: transitions between constituents that appear to need each other to make sense on the one hand, and, on the other hand, transitions between sense-units. The alternation of secondary and primary pauses brings out a pattern of major phonological phrases divided into minor phonological phrases, but the reorganised minor phrases contribute to the resulting major phrase in different ways, so that it is difficult at times to recognise the minor phrases. The major phonological phrase is not a “sentence”. In the examples of means of continuation below, the major phrase encompasses various minor-phrase constituents, both belonging to a single syntactical unit and signalling continuation and more to come.

5.1.1 Audible connection

Audible pauses tend to demarcate developing phrasing only after the introduction of the connector. The prosodic character of, for example, δέ as a postpositive [3] may facilitate a secondary pause, if δέ is realised as syllable. This pause, however, only appears after the initial word of the next syntactical unit has already been introduced. If the completion of a preceding syntactical unit is not audibly realised as (strong) phonetic disruption, δέ as word-final syllable of the phonetic word facilitates a secondary pause that more or less diminishes the listener’s perception of the preceding syntactical break. The listener may find himself in the start of a new syntactical unit before having heard the completion of the unit that closely precedes it. In such instances, the audible termination of a minor-phrase theme constituent not only signals continuation by means of the start of a new syntactic unit, but the completion of the preceding unit as well. Examples can be found in many Homeric verses, like:

ἀλλὰ πίθεσθ᾿ ἄμφω δὲ νεωτέρω ἐστὸν ἐμεῖο

Iliad 1.259

But please listen to me: both of you are younger than I am

ὧς φάτο Τυδείδης δὲ διάνδιχα μερμήριξεν

Iliad 8.167

Thus he spoke and the son of Tydeus considered the two alternatives

The last example, Iliad 8.167, seems to feature sense-pause [
4] at position 2, the other at position 3. In both examples the sense-pause, the completion of a syntactic unit due to the appearance of a theme constituent, is muted. At position 5½ the theme of the subsequent syntactical unit has been introduced in one phonetic word together with enclitic δέ. The phonetic disruption before the announcement of the theme, is muted at position 3 in Iliad 1.259 through elision. At position 2 in Iliad 8.167, it is milder than any disruption at position 5½: at position 5½ there is clisis in addition to word end in a syllable. The “isolated” transitional constituent is remarkably strongly tied to the preceding syntactic unit.

The combination of the additive usage and the prosodic character of δέ shows that Homeric δέ is not strongly demarcative. The appositional location of the particle and the preceding word may be used to cover up sense pauses at metrical boundaries. [5] In a way, the minor-phrase phonetic word consisting of the clause’s [first constituent + (appositive) connector] straddles positions of frequent word end: it ties constituents on both sides of word junctures together too tightly to allow for a strong audible pause reflecting the syntactical break. In the example above, however, δέ brings out the divergence of syntactical and phonological patterns. The prosodic quality of δέ as a postpositive gives special relief to its “second position”. [6] The “second position” is not the important observation here, though it may reveal the presence of relatively unmarked syntactical transitions. [7] The location of δέ, and its realisation as a syllable, allow for a mild audible pause without seriously interrupting a developing major phonological phrase. The location of δέ enables the phonetic word (of which δέ is the closure) to bridge the metrical gap between two positions of frequent word end, for example the gap between positions 5 and 7, and provide continuation over at least one of the two positions: [8]

ὧς φάτ᾿ Ἀθηναίη :5 τῶι δὲ (:) 7 φρένας ἄφρονι πεῖθεν

Iliad 4.104

Thus spoke Athene, and as for him, for the fool she convinced the mind

This way, the phonetic-word metrical colon between positions 5 and 7 is not audibly realised as an independent phrase, but as a subphrase within a developing major phrase. Similar conclusions may be drawn from the usage of γάρ and ἄρα [
9] when they form the closure of minor-phrase phonetic words. The phonetic word bridging the metrical gap may well be a chunk itself. [10] If it is considered too small to be an individual chunk, its prosodic character may still render the phonetic word more or less “prepausal”. This pause at the closure of the phonetic word is only slightly stronger, due to the closure through an (enclitic) appositive, than that of word end in a secondary pause earlier in the same line:

ὧς φάτο :2 Τυδείδης δὲ : διάνδιχα μερμήριξεν

Iliad 8.167

Thus he spoke, and Diomedes considered his alternatives

The syntactical break at position 2 of Iliad 8.167 is less audible than the “pause” at position 5½. Any additional word-final lengthening in the third foot turns the preceding phonetic word into a pragmatic constituent that “stages” the next theme of the checklist without interrupting the major phonological phrase.

Word end in (the syllable) δέ features a secondary pause, with the possible exception of the verse end. [11] Regularly, when (the segment) δέ precedes a position of word end that otherwise may feature a primary pause, for example position 5 or 7, the phonetic word group of which δέ is a part, is preceded by a primary pause, as in the following two examples:

μή τι πάθοι :3 μέγα δέ 5 σφας ἀποσφήλειε πόνοιο

Iliad 5.567

Lest he may get hurt, and completely frustrate them of their toil

αἰχμὴ χαλκείη :5 τὸν δὲ 7 σκότος ὄσσε κάλυψεν

Iliad 4.461

The bronze tip; as for him, darkness covered both his eyes

The phonetic word that serves as a connective device is thus realised in two distinctive ways. Both realisations end in a secondary pause at best, but they may start from either a primary or a secondary pause. Both phonetic words prosodically determine continuation from their completion, but not necessarily at their beginning. When preceded by a secondary pause, the phonetic word provides phonological continuation until the first subsequent primary pause. Following a primary pause, the phonetic word may, to certain extent, be singled out as the start of a major clause. If the phonetic word does not equal a minor phonological phrase, it will give additional relief to the start of a unit. [
12] If elided, the postpositive that forms the closure robs the phonetic word of this capacity. If extended with a periphery (for example [periphery underlined] δ᾿ ἄρα, δ᾿ ἐπεὶ οὖν, δ᾿ ἔπειτα), the capacity is maintained to an extent: I would argue that the extension is motivated by the prosodic realisation of the phonetic word as a subphrase (to be reorganised within a major phrase), and the audible consequences. The most important consequence is variety in the word-end shapes, and hence in phonetic disruption within the line. When I indicate major phrase termination with °, phonetic word end with ˺, and phonetic word end in an (enclitic) appositive (being more likely than mere phonetic word end on the arsis to become audible at slower rates of speech) with ˙˙, such variety can be made visible around the transitional constituents in, for example, Iliad 6.318–320:

ἔνθ᾿ Ἕκτωρ εἰσῆλθε διίφιλος ˺ἐν δ᾿ ἄρα ˙˙χειρί
ἔγχος ἔχ᾿ ἑνδεκάπηχυ ˺πάροιθε δὲ ˙˙λάμπετο δουρός
αἰχμὴ χαλκείη °περὶ δὲ ˙˙χρύσεος θέε πόρκης

Iliad 6.318–320

Hector, loved by Zeus, stepped into it, and in his hand | he held the eleven-cubit spear: at its front gleamed the spear’s | bronze tip, around it ran a golden ring

The phonetic words ending in a postpositive, πάροιθε δὲ and περὶ δὲ, and the phonetic word ending in the extended postpositive, ἐν δ᾿ ἄρα, bridge the metrical gap between two positions of word end. πάροιθε δὲ bridges the gap between positions 5½ and 8, περὶ δὲ the gap between positions 5 and 7, and ἐν δ᾿ ἄρα between positions 8 and 10. The two metrical positions on both sides of each gap are identical with respect to the possibilities they offer for primary or secondary pause: both positions allow for a primary pause, or both allow for a secondary pause. Had the phonetic word end in these three instances been anything other than an enclitic, the two metrical positions would have been realised identically in terms of phonetics. If both positions of word end, on either side of the gap, were realised as audible pauses, their rhythmical realisation would have been identical. The occurrence of the (extended) postpositive frustrates such repetitive realisation.

Other connectors, like τε, have a role similar to δέ in providing continuation, but their prosodic characterisation is different. This already became clear in 2.2.3, where was analysed as an inter-clausal connector In Iliad 1.5, the main function of τε is to keep the clause going despite the recent verse end at the completion of Iliad 1.4:

οἰωνοῖσί τε πᾶσι Διὸς δ᾿ ἐτελείετο βουλή

Iliad 1.5

And for all the birds; the will of Zeus gradually became fulfilled

In Iliad 1.13–14 [
13] the use of τε illustrates the problem of the verbal status of the predicative participle:

λυσόμενός τε θύγατρα φέρων τ᾿ ἀπερείσι᾿ ἄποινα
στέμμα τ᾿ ἔχων ἐν χερσὶν ἑκηβόλου Ἀπόλλωνος

Iliad 1.13–14

To free his daughter bringing countless gifts and | holding the ribbon in his hands of far shooting Apollo

In 2.2 and 2.2.1 I argued that, under specific circumstances, the predicative participle can be understood as a predicate-centred grammatical clause in itself. Often, however, it tends to align with a finite verb within a clause. Homer’s use of τε in Iliad 1.13–14 subtly reminds the audience of both the possibilities and the restrictions of the clausal use of predicative participles. The connector τε serves the same function as καί in concealing the disturbing effect that additions may have on the phonology within the clause. If we compare, for example, Iliad 1.7 and Iliad 1.9, we see that both τε and καί form word groups that straddle position 3:

Ἀτρείδης τε ἄναξ ἀνδρῶν καὶ δῖος Ἀχιλλεύς

Iliad 1.7

Atreus’s son, lord of men, and godlike Achilles

Λητοῦς καὶ Διὸς υἱός

Iliad 1.9

Leto and Zeus’s son

They differ in as much [
14] that an addition introduced by καί, either within the clause (Iliad 1.9) or within an extra-clausal constituent (Iliad 1.7), does not emphasise the constituent that is being added. The mentioning of two heroes in Iliad 1.7 has already been prepared for; there is no need to remind the audience of the appearance of a second character. The application of τε throws Ἀτρείδης into relief before the utterance is interrupted by, and continued by means of an addition, after an audible pause at position 7. [15] In Iliad 1.9, there would have been no audible pause following the first foot, whereas the audible pause coinciding with the verse end of Iliad 1.8 is a strong one. [16] Again, it is the patchwork of pauses that causes the emergence of larger-scale grammatisation.

When constituents that the performer means to utter together can easily or automatically be uttered in one and the same phonological phrase, καί is much more convenient than τε when introducing an addition within the clause. The appearance of τε serves as a strong indication of the realisation of a mild pause while the preceding phonetic-word end within the verse or at verse end is muted. It is possible, however, that both particles are used to facilitate continuation. In Iliad 1.22–23 τε and καί are both used in that way within the clause:

ἔνθ᾿ ἄλλοι μὲν πάντες ἐπευφήμησαν Ἀχαιοὶ
αἰδεῖσθαι θ᾿ ἱερῆα καὶ ἀγλαὰ δέχθαι ἄποινα

Iliad 1.22–23

at that moment all the other Greeks agreed | to pay proper respect to the priest and accept the precious gifts

The use of τε aims at bridging: not only the phonetic-word end at verse end of Iliad 1.22, but position 3 of Iliad 1.23 as well. The elision makes the particle bidirectional. [
17] Addition is the aim of καί: it helps the discourse straddle the third foot caesura. Connective καί could have done without τε in the first half of line 23, but not the other way around. Elided τε may seem useful for prosody though (metri causa). If the particle had not been there, position 3 would suffer from either shortening or hiatus. Shortening does not have to be avoided in the thesis, but hiatus does. But the usage of τε metri causa is not without syntactical consequences: it prepares for continuation, preferably without an intervening pause. Compare, by contrast, the use of καί in Iliad 1.15:

χρυσέωι ἀνὰ σκήπτρωι καὶ λίσσετο πάντας Ἀχαιους

Iliad 1.15

around the golden priest’s staff, and he begged all the Greeks

Here, resumption of the narrative by καί follows the pause in the third foot. The particle is a syntactical reset. It is in such examples that it becomes clear that it is not so much δέ, but rather καί that is best comparable with additive ‘and’ in English. [

The audible punctuation that results from the alternation of phonetic disruptions creates two types of phonologically characterised “units”: first there is the unit that is framed by primary pauses; then, within the unit framed by primary pauses, there can be several smaller units that have secondary pauses as their boundaries. An example like Iliad 9.165–166 shows that the phonological reorganisation of minor-phrase metrical cola around the verse end facilitates the composition of verb-centred (parts of) major phonological phrases:

ἀλλ᾿ ἄγετε κλητοὺς ὀτρύνομεν8 οἵ κε τάχιστα
ἔλθωσ᾿ ἐς κλισίην 5 Πηληιάδεω Ἀχιλῆος

Iliad 9.165–166

But come, let us encourage selected men, who may, as soon as possible, | go to the tent of Peleus’s son Achilles

In the example Iliad 6.318–320, there are four verb-centred (parts of) major phonological phrases (a-d):

ἔνθ᾿ Ἕκτωρ εἰσῆλθε διίφιλος ἐν δ᾿ ἄρα χειρὶ
ἔγχος ἔχ᾿ ἑνδεκάπηχυ πάροιθε δὲ λάμπετο δουρὸς
αἰχμὴ χαλκείη περὶ δὲ χρύσεος θέε πόρκης

Iliad 6.318–320

a     ἔνθ᾿ Ἕκτωρ εἰσῆλθε διίφιλος
b     χειρὶ ἔγχος ἔχ᾿ ἑνδεκάπηχυ
c     λάμπετο δουρὸς αἰχμὴ χαλκείη
d     χρύσεος θέε πόρκης

Iliad 6.318–320

These four verb-centred units are linked by phonetic words that serve as connectors: ἐν δ᾿ ἄρα, πάροιθε δὲ, περὶ δὲ. As discussed above, the phonetic disruption at the end of these phonetic words is likely to be stronger than that of other phonetic words ending in a light segment or syllable: especially at slower rates of speech, the closing (enclitic) postpositive is a relatively strong demarcation of the accentual unit that is the phonetic word. If the passage is divided into phonological phrases along these criteria, the result is:

a     [ἔνθ᾿ Ἕκτωρ εἰσῆλθε διίφιλος (:) ἐν δ᾿ ἄρα (] [) χειρὶ ἔγχος ἔχ᾿
       ἑνδεκάπηχυ (:) πάροιθε δὲ (] [) λάμπετο δουρὸς αἰχμὴ χαλκείη]
b     [περὶ δὲ (][) χρύσεος θέε πόρκης]

Iliad 6.318–320

The ‘(:)’ indicates the syntactical boundary at a metrical boundary that will go practically unnoticed in performance, as ‘(][)’ indicates the metrical boundary at a syntactical boundary that cannot be made audible. The ‘(] [)’ signals a syntactical boundary that may be audible as a secondary pause depending on rate of speech. A and b are (parts of) different major phonological phrases. Both contain connective phonetic words that seems to have possible audible pauses both before and after. At normal rates of speech, the connective phonetic word is itself only slightly phonetically marked, within the larger phonological phrase. It is phonologically and phonetically concluding, but most of the time only by means of a secondary pause: it is a unit, within a unit. As a performer chooses to realise nonbreathing pauses within the rhythmical phrase, the concluding character of the postpositives will make a mild pause following it (indicated as ‘(] [)’) preferable to a pause preceding it (at a position of frequent word end, indicated as ‘(:)’). The result is that at normal rates of speech the connective is the closure of the minor phonological phrase when preceded by a secondary pause. In other words, the connective prepares for continuations before the performer can take a breath.

Often, the connective phonetic word starts from a primary pause, as in περὶ δὲ χρύσεος θέε πόρκης. As such, it resembles “additive” καί connecting verbal forms and filling the unmarked, second element of the foot:

φράζεο Τυδείδη5 καὶ 6 χάζεο μηδὲ θεοῖσιν
ἶσ᾿ ἔθελε φρονέειν

Iliad 5.440–441

Think again, son of Tydeus, and step back; do not | wish to be equal in thoughts to the gods

Of course, the continuation is determined in a different way, from the prosodic point of view: καί is prepositive, περὶ δὲ is a separate phonetic word, allowing for a slight pause after δὲ. Serious phonetic disturbance, however, is avoided [
24] after δὲ. Prepositive καί is truly “additive” as the continuation is not audibly prepared for: the previous phonological phrase has been completed without any clue concerning the type of continuation. Phonetic word περὶ δὲ prepares for continuation, but the preceding primary pause creates the opportunity to single out the connective. When starting from a secondary pause, the connective phonetic word is hardly singled out. Both syntactically and phonologically, continuation is being guaranteed. An example is πάροιθε δὲ in ἐν δ᾿ ἄρα χειρὶ ἔγχος ἔχ᾿ ἑνδεκάπηχυ πάροιθε δὲ λάμπετο δουρὸς αἰχμὴ χαλκείη cited above. This usage of the connective phonetic word may be compared to καί as the realisation of the second light segment in the foot’s double short:

ὦ φίλοι ἀνέρες ἔστε καὶ 6 ἄλκιμον ἦτορ ἕλεσθε

Iliad 5.529

Comrades, be men, and please take heart to defend yourselves

5.1.2 Audible “subordination”

Another means of continuation as identified by, among others, Chantraine and Bakker, is subordination. In my description of the grammatical clause in chapter 2, I have pointed out that what seem to be subordinated clauses in Homer can not automatically be seen as such. In Homeric discourse, the verb-centred grammatical clauses are loosely tied together by transitional constituents that do not always imply syntactical hierarchy. As mentioned in chapter 2, the Iliad and Odyssey contain conjunctions shaped as transitional minor phrases that, in classical Greek, are subordinating. The problems around conjunctions built on the pronoun ὁ/ὅς have been dealt with in the description of syntax, especially hypotaxis, by Chantraine. In chapter 2, I discussed the syntactical isolation of the subordinating conjunction as an extra-clausal. Following Chantraine, I thought it safe to assume that in Homer the development of particles into subordinating conjunctions is still in its infancy. However, to fall in with a respected authority on this matter is far from being conclusive. In order to consider seemingly subordinating conjunctions as pragmatically equivalent to coordinators like καί, δέ and τε, proof had to be presented that these conjunctions have similar discourse-organising abilities. These conjunctions resemble καί, δέ and τε in that they facilitate the progressive movement that replaces true syntax in Homer. In addition to the observations in chapter 1 and 2, I will provide an additional argument based on prosody concerning their resemblance to coordinators within the framework of the clause.

(ii) In Iliad 1.56, the feelings of the goddess Hera are described, now that she chooses to urge Achilles to summon the assembly of the Greek army:

κήδετο γὰρ Δαναῶν ὅτι ῥα θνήισκοντας ὁρᾶτο

Iliad 1.56

For she felt for the Greeks, as she saw them perish

(iii) In the subsequent verse, the Greek army gathers in what turns out to be a formulaic expression:

οἱ δ᾿ ἐπεὶ οὖν ἤγερθεν ὁμηγερέες τε γένοντο

Iliad 1.57

as for them, when they had thus gathered and come together

That is, only ἤγερθεν ὁμηγερέες τε γένοντο is demonstrably a formulaic expression. [
31] With that knowledge, this seemingly ordinary line contains a few interesting features. First, the introduction by οἱ δ᾿. This introduction looks like an elided theme constituent: if the verse had been introduced by non-elided οἱ δέ, the introductory phonetic word would have served as what Bakker calls a “staging” constituent. Its phonetic shape would have thrown it into extra relief. Next, there is the combination of a conjunction and a particle, resembling the “extended” particle: ἐπεὶ οὖν. The actual formulaic expression ἤγερθεν ὁμηγερέες τε γένοντο occupies the hexameter from position 3 onwards. The first phonetic word [32] of the verse, οἱ δ᾿ ἐπεὶ οὖν, up until position 3, shows a hierarchy in the usage of clause-introducing constituents: both the combination pronoun + postpositive particle, and the combination conjunction + particle are used as introductory or transitional constituents. The combination of the two different constituents reveals the adaptable nature of the theme constituent in reorganised phrasal domains. We have an abundance of examples that feature the non-elided theme constituent. Compared to those non-elided introductory constituents, the elided postpositive particle (δ᾿) in Iliad 1.57 prevents hiatus and, as a syllabic onset, [33] also does not end a separate minor phrase but allows the two introductory constituents to be combined into one minor phrase. The combination of two introductory constituents still results in an appositional introductory minor phrase, so the level of subordination here is not different from (i) and (ii).

(iv) In Iliad 1.60–61, there is an interesting difference between two instances of εἰ:

Ἀτρείδη νῦν ἄμμε πάλιν πλαγχθέντας ὀίω
ἂψ ἀπονοστήσειν εἴ κεν θάνατόν γε φύγοιμεν
εἰ δὴ ὁμοῦ πόλεμός τε δαμαῖ καὶ λοιμὸς Ἀχαιούς

Iliad 1.59–61

Son of Atreus, now we shall be driven back home, I belief, and return having achieved nothing, if we can at least escape death | if finally war and plague together will destroy the Greeks

The first εἰ, in line 60, is followed by κεν and an optative φύγοιμεν. The second, in line 61, is followed by future indicative δαμαῖ. Commentators [
34] explain the optative in 60 as ‘a mere supposition, which is expressed as unlikely, remoter and less emphatic’. The speaker here, Achilles, presents the condition in line 61 as much more likely, maybe even as a fact (‘since war and plague together will destroy the Greeks’) than the condition in line 60. The condition in line 60 may as well be a parenthesis. Either way, it is remarkable that the combination εἴ κεν facilitates phonetic disruption due to the movable nu of κεν at position 7, as the end of the minor phrase. Though not-isolated due to the proclitic character of κεν, the transitional constituent provides for an emphatic second position [35] in the subsequent minor phrase for γε stressing θάνατον. The transitional constituent itself is part of a reorganised (major-)phrasal domain despite the appositional alignment of minor phrases. In line 61, εἰ is followed by δή, but combination of δή with the conjunction does not result in a phonetic disruption isolating the conjunction. On the contrary: the occurrence of ὁμοῦ leads to shortening of δή. There is word end at position 3: the minor phrase εἰ δὴ ὁμοῦ is prosodically isolated as an extra-clausal transitional constituent. The identifiable minor-phrase ending in an audible pause calls for even more caution when identifying a subordinate clause in Iliad 1.61. We may even have to come to the conclusion that, again, there is no subordination—not in line 60, and not in line 61.

(v) Quite the opposite situation is found in line Iliad 1.83, in the word group εἴ με σαώσεις ‘if you will rescue me’. The words are joined in one phonological phrase due to clisis: the resulting phrase is a complete subordinate clause, including the conjunction. It may still be true that the clause is not strictly subordinate, as εἰ can be used to introduce paratactic parentheses. What is remarkable is that one entire clause is contained in a single minor phonological phrase, and that the introducing conjunction is included. This minor phrase is quite different from the phrases that contain either (parts of) the grammatical clause or transitional constituents. A parallel for minor phrases containing the transitional constituent can be found in the short comparison of the type Odyssey 5.433 ὡς νυκτερίς ‘like a bat’, and, derived from it, Iliad 8.35 ὡς σὺ κελεύεις ‘as you command’, and, encompassing another conjunction, Iliad 8.142 αἴ κ᾿ ἑθέληισι ‘if he will want to’. In these instances, higher-level grammatisation is not the result of reorganisation of minor phrases into major phrases, but innate to the individual minor phrase. I suggest that the occurrence of such conjunction/particle-introduced minor-phrase clauses may be the origin of the clause that develops from the reorganised phrasal domain. At the same time, I admit that the paratactic character of both the minor-phrase clause and the “reorganised” clause need not be reviewed based on above observations: any trace of higher-level grammatisation as a result of reorganisation of minor phrases does not necessarily, and surely not immediately, lead to hierarchical complex predications in Homer. The paradigm shift from sentential particle to subordinating conjunction may find a necessary, preliminary step, not its beginning, in the prosodically motivated emerge of higher level grammatisation.

(vi) Special attention should be given to conjunctions followed by subjunctives, such as Iliad 1.82 ὄφρα τελέσσηι ‘so that he will bring it about’, and Iliad 1.90 οὐδ᾿ ἢν Ἀγαμέμνονα εἴπηις ‘not even if you were to mention Agamemnon’. The absence of prepausal postpositive particles strengthens the suggestion of genuine subordination here.

These seven examples from the Iliad’s first book illustrate the usefulness of the “audible-pause” test to identify the emergence of larger scale grammatisation. They do not, however, answer the question concerning the possibility of subordination in Homer. In short, it is—again—overreaching to come to the conclusion that there is subordination in Homer when analysing the exact size and juxtaposition of grammatical clauses. We are literally centuries away from anything like a hierarchical complex predication. There is no “sentence-structure” in the Iliad and the Odyssey as if the poems could be compared to the composition of classical Greek prose writings. Certain grammatical and syntactical features that constitute the backbone and the essence of literary works composed with the aid of (or as a sign of craftsmanship in) writing are absent from the Homeric epics. With others, like the use of conjunctions and subordination next to appositional parataxis, the Iliad and the Odyssey seem to experiment. [38] The experiments can be identified (as I have for the Iliad’s first 100 lines) as the result of reorganisation of independent minor phrases into major phrases, following the example of single-phrase subordinate clauses. The occurrence, however, of experiments side by side with the original and, incidentally, opposite usage (the latter being much more frequent) within the same body of texts makes it troublesome to accept them straightforwardly as successful first steps towards a more periodic and internally hierarchical sentence-structure. [39]

5.1.3 Audible attribution

Attributive elements in genitive case are mentioned by Bakker among the characteristic of adding style. It would be interesting to apply the “audible pause” test to find out if the mapping of attributive elements prepares for continuation in a way similar to the mapping of particles and conjunctions. The first example of an attributive element in genitive case in the Iliad is the word group Πηληιάδεω Ἀχιλῆος in Iliad 1.1:

μῆνιν ἄειδε θεά Πηληιάδεω Ἀχιλῆος

Iliad 1.1

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

Whether or not the word group Πηληιάδεω Ἀχιλῆος constitutes a single spurt of vocalisation (or is it too short?) is hardly the issue. The listening audience will have understood the genitive case Πηληιάδεω Ἀχιλῆος as a coherent whole. The genitive case word group specifies the wrath that is going to be the central theme, if not of the whole epic, then at least of its opening scene. [
40] As long as more information is being added concerning the wrath of Achilles, the theme of the wrath is still being further and further elaborated. [41] The continuing effect of this genitive case attributive element on syntax can be compared to the genitive case attributive elements in Iliad 1.7, Iliad 1.14, and Iliad 1.34:

Ἀτρείδης τε ἄναξ ἀνδρῶν καὶ δῖος Ἀχιλλεύς

Iliad 1.7

The son of Atreus, lord of men, and godlike Achilles

στέμματ᾿ ἔχων ἐν χερσὶν ἑκηβόλου Ἀπόλλωνος

Iliad 1.14

Holding the ribbons in his hands of far shooting Apollo

βῆ δ᾿ ἀκέων παρὰ θῖνα πολυφλοίσβοιο θαλάσσης

Iliad 1.34

And he strode in silence along the shore of the roaring sea

In Iliad 1.1, 1.14, and 1.34, the attributive element occupies the second half of the verse, from the third foot caesura to the verse end. The continuation that the attributive element provides seems to be consistent with the mildness of the trochaic caesura as an audible pause in Iliad 1.14 and 1.34. The second half-verses of Iliad 1.14 and 1.34 seem to qualify as metrical fillers: they extend the verb forwards to the nearest metrical boundary. Of course, ἐν χερσὶν ‘in his hands’ and παρὰ θῖνα ‘along the shore’ qualify as fillers as well: both occupy the metrical colon between positions 3 and 5½. Word end at position 3 in Iliad 1.34 allows for an audible pause (ἀκέων : παρὰ). The trochaic caesura hardly allows for phonetic disruption: syntactical continuation by means of an attributive element does not only extend the major phonological phrase but the filler as well. The result is not an extended metrical filler (ἐν χερσὶν ἑκηβόλου Ἀπόλλωνος / παρὰ θῖνα πολυφλοίσβοιο θαλάσσης) but a sizable phonological-phrase filler that consists of more than one metrical filler.

Especially interesting from the audible-pause point of view is the attributive element ἀνδρῶν in Iliad 1.7:

Ἀτρείδης τε ἄναξ ἀνδρῶν καὶ δῖος Ἀχιλλεύς

Iliad 1.7

The son of Atreus, lord of men, and godlike Achilles

A demarcating word end at position 5 in this line would result in division of the word group ἄναξ ἀνδρῶν ‘lord of men’ into ἄναξ and ἀνδρῶν. Such separation may be seen as verse-internal enjambment, though, as mentioned before, the occurrence of enjambment at the caesura or verse-internal dieresis is not commonly accepted. If verse-internal enjambment were accepted, its poetic purpose in Iliad 1.7 is not immediately clear. [
43] It is tempting instead to grant the attributive element ἀνδρῶν the effect of bridging the word juncture at position 5. As filler, ἀνδρῶν extends the minor phrase to the nearest pause, in this case the primary pause at word end at position 7. The genitive ἀνδρῶν is itself context-neutral, and innocuous with respect to the line’s content. [44] It qualifies as metrical filler. In the case of the word group ἄναξ ἀνδρῶν, word-final -ναξ leaves room for phonetic continuation through phonological sandhi. Word-final -δρῶν, as a superheavy syllable, leaves none. If rhythmical variety overrules metrical repetition, Iliad 1.7 features a primary pause at position 7. In Iliad 1.7, the attributive element phonologically bridges the third foot word end. The continuation that the attributive element provides is grammatical as well as phonological.

Genitive case attributive elements can be used to similar effect to bridge the verse end. When considering the sense pauses in the Iliad and the Odyssey, rhythmical muting of the verse end would often a priori seem appropriate in order to extend the phonological phrase forwards to the nearest audible pause. The first examples encountered in the Iliad’s opening book are Iliad 1.3–4 and Iliad 1.40–41:

                    ψυχὰς Ἄιδι προίαψεν
ἡρώων : 3

Iliad 1.3–4

The souls it sent to the house of Hades | of the heroes

                    κατὰ πίονα μηρί᾿ ἔκηα
ταύρων ἠδ᾿ αἰγῶν : 5

Iliad 1.40–41

I have burnt fat tights | of bulls and goats

But appropriate does not automatically mean that the verse end is ignored as an audible boundary: syntax does not count as proof for prosodic realisation. As I have shown in chapter 4 and in the previous sections of this chapter, it is the other way around: prosody may confirm what we find on the level of syntax.

Genitive case attributive elements can also serve to extend the verb backwards. [45] This preparatory task can be phonologically characterised as well. Some examples, in addition to Iliad 1.1–2 above, may serve to illustrate this. In Iliad 1.5 Διὸς and elided δ᾿ together bridge word end at position 7:

οἰωνοῖσί τε πᾶσι Διὸς δ᾿ ἐτελείετο βουλή

Iliad 1.5

And all the birds; the will of Zeus gradually became fulfilled

At the same time, the elision of the phonetic word Διὸς δ᾿ avoids a repetitive (trochaic) rhythmical pattern. In Iliad 1.9 [
46] genitive Λητοῦς καὶ Διὸς extends the missing verb backwards to the start of the line:

Λητοῦς καὶ Διὸς υἱός ὃ γὰρ βασιλῆι χολωθείς

Iliad 1.9

The son of Leto and Zeus (it was); for he, enraged with the king

In Iliad 1.21, perhaps the model for Iliad 1.9, the same localisation (though in another case) of Διὸς υἱὸν (this time in a new phonological phrase, after a primary pause) extends the verbal form ἁζόμενοι forwards:

ἁζόμενοι Διὸς υἱὸν ἑκηβόλον Ἀπόλλωνα

Iliad 1.21

With proper respect for the son of Zeus, far shooting Apollo

The nearest metrical boundary is position 5½; the first possible stronger audible pause, the verse end. In Iliad 1.19, genitive case attributive element Πριάμοιο is preparatory in bridging word end in the third foot in order to extend the verb ἐκπέρσαι all the way to position 7:

ἐκπέρσαι Πριάμοιο πόλιν7 ἐὺ δ᾿ οἴκαδ᾿ ἱκέσθαι

Iliad 1.19

To destroy Priam’s citadel, and to return home safely

Many genitive case attributive elements appear as the final word of a metrical colon or the verse. Their localisation, especially with word end on the arsis (type: Iliad 1.19 Πριάμοιο), contributes to the bridging of positions of frequent word end. [
47] It is tempting to assume a compositional principle underlying the audible signalling of Πηληιάδεω Ἀχιλῆος in Iliad 1.1. Is it an audible preparation for syntactical continuation, in this case by means of hyperbaton (μῆνιν … οὐλομένην)?

5.1.4 Audible predicate extension

The position of οὐλομένην in Iliad 1.2 shows, that the addition [48] of the participle is in fact a means of syntactical continuation from Iliad 1.1 into Iliad 1.2:

μῆνιν ἄειδε θεά Πηληιάδεω Ἀχιλῆος
οὐλομένην ἣ μυρί᾿ Ἀχαιοῖς ἄλγε᾿ ἔθηκεν

Iliad 1.1–2

Sing, goddess, of the wrath of Achilles, the son of Peleus, | so destructive, that bestowed countless pains on the Greeks

The adjective seems to occupy a prominent place, using the “emphatic” first position in the line to extra effect. [
49] It is a phonological continuation as well, as the verse end is muted compared to the sense-pauses at position 5 in Iliad 1.1, and position 3 in Iliad 1.2. The localisation of the attributive participle (choriambic ῥυθμός, descending-blunt verbal rhythm, progressing from a mild to a stronger audible pause) suggests extension of the verbal form [50] backwards: [Πηληιάδεω Ἀχιλῆος | οὐλομένην]. The result is a major phonological phrase that crosses the verse end: [51] Πηληιάδεω Ἀχιλῆος οὐλομένην. Syntactically, the adjective has a continuing effect similar to, for example, a relative clause. Phonologically, the participle οὐλομένην is more closely tied to the preceding verse-final words, than the relative pronoun ἣ is to preceding οὐλομένην. The recurring phenomenon of the “run-over participle + relative pronoun” [52] (rhythmical result: muted verse end + audible pause at position 3 [53] ) is an effective way to create rhythmical variety. At the same time, however, it is unlikely that this adjective in first position in the verse can be identified as an example of poetically effective enjambment. Of course, the adjective follows enjambment when run-over over the verse end is considered as definition of enjambment. Technically, οὐλομένην stands in hyperbaton, [54] but enjambment does not contribute: this hyperbaton is not “strengthened” by a pause at verse end. If in this case hyperbaton is strengthened by a pause at all, [55] it must be by the audible pause at position 5 in the first line. In every respect, the positioning of the adjective is more effective in continuing the phonological phrase at hand than is the verse end in creating a phonetic disturbance or poetically meaningful enjambment.

The predicative participle, however, syntactically and phonologically extends forwards in Iliad 1.13:

λυσόμενός τε θύγατρα φέρων τ᾿ ἀπερείσι᾿ ἄποινα

Iliad 1.13

To free his daughter, bringing countless gifts

5.1.5 Audible apposition

L’autonomie de chaque terme a pour conséquence que l’aède peut, à l’occasion, perdre de vue le mot auquel il se réfère, d’où des libertés dans les règles d’accord, d’ où aussi l’intervention de groupes de mots qui ne se rattachent pas strictement à ce qui précède ou à ce qui suit.

In the words of Chantraine: [
57] this appositional character of Homeric syntax is evident from the paratactic organisation of the discourse. Devine and Stephens consider the ‘rather flat, as opposed to hierarchical, phrase structure’ [58] of the Homeric epic as illustrative for the “nonconfigurationality” of ancient Greek. At the same time, of course, Homeric discourse features manifold true, grammatical, appositions. These appositions have been viewed from different angles in the past century of scholarship. Many of them, especially the epithets, have been described as compositional fillers, and thus as the building blocks of the oral compositional technique. Analysis as metrical filler might be functional: when the epithet is used to fill the remainder of a verse in order to facilitate the usage of an audible pause, for example verse end, for discourse shift or sense-pause. The first examples the Iliad offers are Iliad 1.16, Iliad 1.21, and, even more evidently, Iliad 1.35–36:

Ἀτρείδα δὲ μάλιστα δύω κοσμήτορε λαῶν

Iliad 1.16

And especially the two sons of Atreus, the two arrangers of the army

ἁζόμενοι Διὸς υἱὸν ἑκηβόλον Ἀπόλλωνα

Iliad 1.21

With proper respect for the son of Zeus, far shooting Apollo

πολλὰ δ᾿ ἔπειτ᾿ ἀπάνευθε κιὼν ἠρᾶθ᾿ ὁ γεραιός
Ἀπόλλωνι ἄνακτι τὸν ἠύκομος τέκε Λητώ

Iliad 1.35–36

And then, having arrived at a great distance, the old man spoke in prayer | to lord Apollo, whom fair Leto had given birth to

In these examples, the grammatical appositions conclude a larger whole. Their appearance functions as an afterthought: the information in the apposition may well be meaningful in its own right, but the apposition itself does not prepare the audience for what is to follow. [

In many similar instances, however, the apposition seems to be a break in a larger whole, in something that still needs to be picked up after the apposition. Iliad 1.19–20 is an example:

ὑμῖν μὲν θεοὶ δοῖεν Ὀλύμπια δώματ᾿ ἔχοντες
ἐκπέρσαι Πριάμοιο πόλιν ἐὺ δ᾿ οἴκαδ᾿ ἱκέσθαι

Iliad 1.19–20

May the gods grant you, the gods who have their houses on mount Olympus, | to destroy Priam’s citadel, and to return home safely

The appositional epithet Ὀλύμπια δώματ᾿ ἔχοντες of θεοὶ suits the metrical requirements of the verse up to verse end. At the same time, the epithet disrupts the developing wish that has been initiated, but not been brought to completion, in the first half of Iliad 1.19. Phonetically, the disturbance in Iliad 1.19 is minimal. Iliad 1.19 does not contain a strong audible pause; the first primary pause is the word end at position 3 in Iliad 1.20. Both the Iliad and the Odyssey abound in examples that are rhythmical parallels to Iliad 1.19–20.

I will explain why, within the variety of constituents that may be labelled “appositions”, special attention should be given to those in the nominative case. In an example like Iliad 1.19–20, the apposition appears to be a metrical filler. Its location and phonetic non-disruptiveness enable the phonological phrase to straddle the verse end boundary rather easily. Other nominative case appositions, on the other hand, seem to contribute to the syntactical movement in quite a different manner. They appear to be grammatical requirements of a clause, rather than true appositions. In the Iliad’s first lines, an example can be found in Iliad 1.11–12:

οὕνεκα τὸν Χρύσην ἠτίμασεν ἀρητῆρα
Ἀτρείδης ὁ γὰρ ἦλθε θοὰς ἐπὶ νῆας Ἀχαιῶν

Iliad 1.11–12

As he had dishonoured him, Chryses the priest, | the son of Atreus; for he had come to the fast ships of the Greeks

The patronymic Ἀτρείδης seems to function as the subject of Iliad 1.11. As such it extends the verbal form ἠτίμασεν ‘had dishonoured’ forwards to position 3 of Iliad 1.12. In his analysis of Homeric syntax, Bakker [
60] labels such nominative case appositions “additions”. In his analysis, the verse end boundary renders such appositions tail-constituents to the preceding intonation unit. [61] The main purpose of the nominative case apposition lies in the staging or restaging of persons or objects. The syntactical checklist-structure is characterised by a certain degree of looseness between consecutive intonation units. Such looseness requires nominative case appositions for the purpose of “disambiguation” every once in a while. [62] Ἀτρείδης in Iliad 1.12 is identified as an apposition to the subject implicit in the verbal form ἠτίμασεν . [63] In translation, Iliad 1.11–12 may be something like:

since he had dishonoured him, Chryses the priest, | he, that is, the son of Atreus; for he came to the fast ships of the Greeks

The apposition itself may thus be seen as a separate intonation unit, [
64] according to Bakker.

I think, however, that similar to what has been said about the attributive elements, participles, and “disruptive” appositions above, the location of this nominative case “apposition” suggests something else. Ἀτρείδης in Iliad 1.12 may well have a pause preceding it, but the force of that pause is being muted compared to that of the word end at position 3. As phonetic disturbance, the metrical boundary of the verse end may still have some significance, comparable to the trochaic caesura in Iliad 1.20. The phonological tie between the final words of Iliad 1.19 and the run-over word Ἀτρείδης is stronger than that between Ἀτρείδης and the phonological phrase that follows it. Depending on the prosodic realisation of the verse end, Ἀτρείδης may well be an apposition. The rhythmical structure of the verse-end-straddling phonological phrase, however, suggests that the grammatical tie between ἠτίμασεν and Ἀτρείδης is closer than merely that between verb with implicit subject and apposition. The prosodic realisation of the verse end maintains its character as metrical boundary, but as a pause the verse end is muted compared to the primary pause at position 3 in the line that follows. Highly interesting in this respect, and a fine example of the need to distinguish between “appositions” and “true appositions” is Iliad 9.337–341, where the denomination tail-constituent for the runovers in 338, 339, and 341 would make the emotional outburst of Achilles look as if he has lost his concentration:

τερπέσθω τὶ δὲ δεῖ πολεμιζέμεναι Τρώεσσιν
Ἀργείους τὶ δὲ λαὸν ἀνήγαγεν ἐνθάδ᾿ ἀγείρας
Ἀτρείδης ἦ οὐχ Ἑλένης ἕνεκ᾿ ἠυκόμοιο
ἦ μοῦνοι φιλέουσ᾿ ἀλόχους μερόπων ἀνθρώπων

Iliad 9.337–341

Let him take his pleasure! But why was there need to fight the Trojans | for the Greeks? Why did he gather the army and bring it here, | Atreus’s son? Was it not because of fair Helen? | Are we to understand that of mortal men the only ones who love their partners | are Atreus’s sons?

Phonological phrasing, however, points at the fact that Achilles has not. If stylised metrical units [
65] are indeed the compositional building blocks of Homeric syntax, as Bakker claims, the divergence between the audible pauses and metrical boundaries enables the phrases to surpass the metrical units. The stichic repetition of metrical units does not only allow, [66] but actually elicits “unhexametric clustering”. Syntactical continuation is prosodically determined.

5.2 Audible Punctuation as a Structuring Device

So far, attention has been given to the prosodic cause behind continuation in Homeric discourse by means of connectors, subordinating conjunctions, participles, appositions, and genitive case attributive elements. The examples I provide to illustrate the way prosody determines their value as means of continuation, have led to detailed considerations on a smaller scale. In this section, 5.2, I will present examples that illustrate preparatory effect of prosodic determination on continuation on a larger scale. To that end, I will apply the “audible pause” test to discourse markers and to hyperbaton. Hyperbaton has already been taken into account in 5.1.4, but will be studied as a separate means of continuation in the following section. Following hyperbaton, the location of discourse structuring elements will be considered. I will show that the location of these grammatical signals of continuation-in-preparation is in accordance with the semantic promise they hold out: the perceptible “chunking” of the narrative for a listening audience suits the comprehensible “chunking” in units of sense.

5.2.1 Hyperbaton

The ancient grammarians and the scholia treat ὑπερβατόν as the opposite of ἑξείης, ἑξῆς ‘in a regular grammatical sequence’. In poetry, the qualification hyperbaton may have far greater implications than in prose: the perception of the audience is not only affected by the interference of other words between those in hyperbaton. The strength of intervening pauses must be taken into account as well.

Hyperbaton as in Iliad 1.2 is not subject to phonetic disturbance. Even the possibility of a secondary pause is nullified through elision:

μυρί᾿ Ἀχαιοῖς ἄλγε᾿ ἔθηκε

Iliad 1.2

That caused countless pains for the Greeks

The occurrence of sandhi through elision ties the constituents in hyperbaton together in one phonological phrase.

Some more examples will illustrate my point. Frequently, word end at the bucolic dieresis occurs inside word groups that are regularly analysed as preposition groups featuring both hyperbaton and reversed word order, for example in Iliad 1.12 (θοὰς ἐπὶ :8 νῆας Ἀχαιῶν) and Iliad 1.25 (κρατερὸν δ᾿ ἐπὶ :8 μῦθον ἔτελλεν). The demarcating “strength” of the word end at position 8 depends on the analysis of the preposition. If ἐπὶ is identified as prepositive and proclitic, the word juncture at position 8 is not a phonetic-word juncture. As such, it does not allow for phonetic disruption. In Iliad 1.25, elided δ᾿ refutes phonetic word-end at position 7: the words κρατερὸν δ᾿ ἐπὶ μῦθον ἔτελλε are then phonologically tied together as one phrase. The hyperbaton is phonologically “configured”: the phonological realisation of the phrase containing hyperbaton keeps the words that make “sense” together, together.

Similar configuration of hyperbaton is found at the third foot caesura and the verse end. An example of the former is Iliad 8.180:

ἀλλ᾿ ὄτε κεν δὴ νηυσιν ἔπι γλαφυρῆισι γένωμαι

Iliad 8.180

But when I will have finally reached the curved ships

The combination of word end at position 5½ and phonological sandhi ties the constituents in hyperbaton together in one phonological phrase. The most famous example, of course, is Odyssey 1.1:

ἄνδρα μοι ἔννεπε Μοῦσα πολύτροπον ὃς μάλα πολλὰ

Odyssey 1.1

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

Iliad 1.1–2 is the best-known example of configuration at, or rather over, the verse end. This is not to say that the two words in hyperbaton are tied together in one minor or major phonological phrase: up to position 5 Iliad 1.1 is a single major phrase, and there is a second major phrase starting from position 5 and ending at position 3 in Iliad 1.2. The latter contains two minor phrases, one ending at and one starting from the verse end of Iliad 1.1. The verse end, however, at least does not contribute to nonconfigurationaly here:

μῆνιν ἄειδε θεά Πηληιάδεω Ἀχιλῆος

Iliad 1.1–2

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

Nonconfigurationality, the lack of hierarchical organisation in Homeric syntax, cannot explain the hyperbaton in the following two examples, nor defend an audible pause at position 7. At the same time, Iliad 1.12 and Iliad 1.25 do have audible pauses verse-internally and at verse end:

Ατρείδης :3 ὃ γὰρ ἦλθε θοὰς ἐπὶ νῆας Ἀχαιῶν :12

Iliad 1.12

The son of Atreus, for he had come to the swift ships of the Greeks

ἀλλὰ κακῶς ἀφίει :5 κρατερὸν δ᾿ ἐπὶ μῦθον ἔτελλεν :12

Iliad 1.25

But he sent him away disgracefully, and added a serious warning

The words in hyperbaton are closely tied together in one phonological phrase; so closely in fact, that the configuration is not at all that different from a ‘regular grammatical sequence’. Even if the words in hyperbaton are further apart, as they are in Odyssey 1.1 (ἄνδρα μοι ἔννεπε Μοῦσα πολύτροπον ‘tell me about the man of many ways, goddess’), the coherence of the phonological phrase points at the regularity of the configuration of the words in hyperbaton.

5.2.2 Discourse structuring elements

After examining the location of words in hyperbaton, I will present the location of discourse structuring elements as a final illustration of the way prosody furthers grammatisation in the Homeric epic. Discourse structuring elements can be divided into two groups. On the one hand, there are the elements that not only structure the discourse but the narrative as well. On the other hand, there are elements that announce, or actually themselves are changes in type of discourse (for example, a shift to an imperative, a question, or direct speech). In Homer, among the elements that also serve a function as narrative-structure devices, adverbial expressions indicating time [72] are prominent. For example, in Iliad 1.53–54 the adverbial expression ἐννῆμαρ μὲν ‘for a period of nine days’ seems to be preparatory:

ἐννῆμαρ μὲν ἀνὰ στρατὸν ὤιχετο κῆλα θεοῖο
τῆι δεκάτηι δ᾿ ἀγορήνδε καλέσσατο λαὸν Ἀχιλλεύς

Iliad 1.53–54

For a period of nine days the arrows of the god vexed the army, | but on the tenth Achilles called the soldiers to the place of assembly

The adverbial expression ἐννῆμαρ μὲν prepares the audience for a contrast with, or a further elaboration of, the description of the horrible plague in Iliad 1.42–52. It looks as if the adverbial expression is the start of some sort of list that specifies αἰεί in Iliad 1.52. Another adverb of time (τῆι δεκάτηι ‘on the tenth’) shows that Iliad 1.53 was the concluding remark: concluding to the description of the effects of the disease, and dismissing the topic of dying as foil for a new topic in line 54, Achilles. For the audience, the concluding character of Iliad 1.53 was not an eagerly awaited surprise. There is no primary pause at the verse end of Iliad 1.53. On the contrary: any audible break between Iliad 1.53 and Iliad 1.54 is hardly noticeable. The verse end of Iliad 1.53 is being muted, allowing the major phonological phrase to extend forwards to the nearest primary pause. Prosody does not suggest a clear break to underline the concluding character of Iliad 1.53.

A discourse structuring element of the second category is the insertion of direct speech. In Homeric narrative, direct speech is usually announced explicitly. The start of direct speech always coincides with the verse-initial foot of the hexameter. The use, however, of a verb denoting to speak, talk, shout, pray, et cetera at or near verse end, does not necessarily indicate an immediate transition to direct speech. Iliad 1.35–37 may serve as an example:

πολλὰ δ᾿ ἔπειτ᾿ ἀπάνευθε κιὼν ἠρᾶθ᾿ ὁ γεραιός
Ἀπόλλωνι ἄνακτι τὸν ἠύκομος τέκε Λητώ
κλῦθί μευ ἀργυρότοξ᾿ ὃς Χρύσην ἀμφιβέβηκας

Iliad 1.35–37

And then, having arrived at a great distance, the old man spoke in prayer | to lord Apollo, whom fair Leto gave birth to: | Please listen to me, Silverbow, who stands protectively over Chryse

Various discourse markers guide and help the audience in mentally processing what is being said. Their syntactical function and pragmatic effect are comparable to similar discourse markers and shifts in written language. Still, in written language, the same phenomena are commonly understood as marking distinct smaller and larger scale units as well, for example the sentence. The mapping, however, of above examples shows that in Homer they do not necessarily signal the start or the conclusion of what is called a “sentence” in written utterances. In the example Iliad 1.53–54, and similar cases, it is evident that there is neither need nor wish for a strong or emphatic pause to divide the two verses. Even for the start of direct speech, equation with the start of a new “sentence” is disputable.

5.3 Prose-like Audible Punctuation

5.3.1 Opportunities for grammatisation through prosody

Despite the paratactic character of its syntax, the Iliad and the Odyssey both show ample examples of syntactical coherence beyond the individual metrical colon, and beyond the single verse. Syntactical coherence in Homer may disregard all positions of frequent word end. Formulaic expressions fitting a metrical colon persist, but their semantic value is re-determined within larger scale syntactical and prosodic wholes. Grammatical governance straddles verse-internal boundaries, and grammatisation regularly straddles the metrical boundary between verses:

                    αὐτίκα δ᾿ ἔγνω
Παλλάδ᾿ Ἀθηναίην

Iliad 1.199–200

Immediately he recognised | Pallas Athena

Ἰδομενεὺς δ᾿ ἑτέρωθεν ἐνὶ Κρήτεσσι θεὸς ὣς

Iliad 3.230–231

And on the other side | stood | Idomeneus, like a god amidst the Cretans |

                    πὰρ δέ οἱ ἔστη
Ἀτρείδης Μενέλαος

Iliad 6.43–44

Next to him appeared | Atreus’s son Menelaus

                    οὐδὲ συβώτης
λήθετ᾿ ἄρ᾿ ἀθανάτων

Odyssey 14.420–421

And the swineherd did not | forget what was due to the immortals

ἦ τοι ἐγὼ καὶ Τηλέμαχος μνηστῆρας ἀγαυοὺς
σχήσομεν ἔντοσθεν μεγάρων

Odyssey 22.171–172

For our part Telemachus and I will keep the arrogant suitors | where they are within the walls of the great hall

αἴ κε ζὼν πέμψηις Σαρπηδόνα ὅνδε δόμονδε
φράζεο μή τις ἔπειτα θεῶν ἐθέληισι καὶ ἄλλος
πέμπειν ὃν φίλον υίὸν ἀπὸ κρατερῆς ὑσμίνης

Iliad 16.445–447

If you will let Sarpedon leave for his house unharmed | be careful then lest an other of the gods will want as well | to let his own son leave from the strenuous battle

Homeric syntax is still, to an extent, “loose”, but Homer phrases his narrative in grammatical clauses whose size is the poet’s choice. His appositional construction cannot in general be compared to the psychologically motivated nonconfigurationality of, for example, ballads from other centuries and places. [
76] The pattern of Homer’s grammatical clauses shows clear syntactical organisation, regularly spanning an entire verse, or several verses:

χερμαδίωι γὰρ βλῆτο παρὰ σφυρὸν ὀκριόεντι

Iliad 4.518

For he was hit by a jagged stone on the lower leg

οἳ δ᾿ ἀμφὶ Πρίαμον καὶ Πάνθοον ἠδὲ Θυμοίτην
Λάμπον τε Κλυτίον θ᾿ Ἱκετάονά τ᾿ ὄζον Ἄρηος

Iliad 3.146–147

Those around Priam, Panthous, Thumoites, | Lampus, Clytius, and Hicetaon, the fosterchild of Ares

τὸν μὲν ἄρ᾿ Αὐτόλυκός τε καὶ υἱέες Αὐτολύκοιο
εὖ ἰησάμενοι ἠδ᾿ ἀγλαὰ δῶρα πορόντες
καρπαλίμως χαίροντα φίλην χαίροντες ἔπεμπον
εἰς Ἰθάκην

Odyssey 19.459–462

Him Autolycus and the sons of Autolycus, | having cured him well and provided him with lavish gifts, | cheerfully sent, himself cheerful, quickly to his own | Ithaca

Metrical structure makes it harder to accept larger scale syntactical organisation. The repetitive metrical colometry, paratactic in its own way, persistently suggests syntactical parataxis. As the Parryan approach shows, parataxis is most easily accepted at the verse end. Parry saw the compositional technique as juxtaposing of whole-verse metrical units. Syntactically and semantically, however, the subsequent verse is often an addition to the previous, instead of a new, whole clause. There is an abundance of complete lines that cannot stand on their own as clauses. Two examples to illustrate this:

καλούς ἀμβροσίους ἐκ κράατος ἀθανάτοιο

Iliad 14.177

Beautiful, divine, down from the head of the immortal

ἐκ Κρήτης ἐς γουνὸν Ἀθηνάων ἱεράων

Odyssey 11.323

From Crete to the acropolis of sacred Athens

There are equally many lines that contain a complete clause, and a metrical colon that is an addition to the previous line (or a preparation for the line to follow):

νῦν δ᾿ οὐκ ἔσθ᾿ ὅς τις θάνατον φύγηι ὅν κε θεός γε
(Ἰλίου προπάροιθεν ἐμῆις ἐν χερσὶ βάληισι)

Iliad 21.103–104

But now, there will not be anyone who escapes death, whom a god | (delivers in my hands before the walls of Troy)

(αὐτὰρ ἐμὲ προέηκε Γερήνιος ἱππότα Νέστωρ)
τῶι ἅμα πομπὸν ἕπεσθαι ἐέλδετο γάρ σε ἰδέσθαι

Odyssey 4.161–162

(but me the horse-taming Gerenian Nestor sent forth) | to be his companion on the journey; for he longed to see you

At the same time, clauses start and end somewhere within the line, leaving some metrical space to be filled with constituents that join the line to the previous or subsequent line:

                    εἰς ὅ κε νεκροὺς
κείομεν ὕστερον αὖτε μαχησόμεθ᾿ εἰς ὅ κε δαίμων
ἄμμε διακρίνηι

Iliad 7.376–378

Till | we burn | the deceased; later we will do battle again until a god | puts us apart

          καὶ σὺ θεοῖς αἰειγενέτηισιν
εὔχεο καὶ δὲ σὲ φασὶ Διὸς κούρης Ἀφροδίτης

Iliad 20.104–106

And you, | pray | to the ever-living gods; indeed, they say that you | are descendent | from Zeus’s daughter Aphrodite

               οὐδ᾿ Ἐνοσίχθων
λήθετ᾿ ἀπειλάων τὰς ἀντιθέωι Ὀδυσῆι
πρῶτον ἐπηπείλησε

Odyssey 13.125–127

But Poseidon did not | forget the threats that against the godlike Odysseus | he had uttered in the past

If the metrical colometry, and juxtaposition of verses, is considered evidence or reflection of compositional parataxis (or paratactic composition), I argue that the identification of semantically hierarchical and syntactically dependent metrical phrases must lead to incompatibility of metrical and syntactical phrasing, even in the process of composition. In performance, there is audible metrical colometry, next to a different type of audible phrasing; but even in composition, meter and syntax seem to get in each other’s way. The metrical phrasing, when analysed as paratactic juxtaposition of repetitive metrical cola, suggests that the incompatibility of metrical and syntactical phrasing is best described as a stylistic feature. It seems to be a deliberate incompatibility for poetic effect. As such, the “anomaly” is regularly characterised as evidenced in syntax: the syntactical features of the “run-over” into the subsequent line determine the stylistic effect of the anomaly, as the metrical boundaries and their paratactic demarcation are taken for granted. Are we supposed then to assume a theoretical incompatibility in the process of composition that was not, and could not be, exploited to poetic effect in performance?

In the metrical approach, the poetic effect of the incompatibility may be refered to as a mismatch. [77] To avoid monotony in versification, the oral poet may choose to lessen the “natural” correspondence between the metrical and the syntactical phrasing. [78] Consider a passage like Iliad 1.428–431:

ὧς ἄρα φωνήσασ᾿ ἀπεβήσετο τὸν δ᾿ ἔλιπ᾿ αὐτοῦ
χωόμενον κατὰ θυμὸν ἐυζώνοιο γυναικός
τήν ῥα βίηι ἀέκοντος ἀπηύρων αὐτὰρ Ὀδυσσεὺς
ἐς Χρύσην ἵκανεν ἄγων ἱερὴν ἑκατόμβην

Iliad 1.428–431

Thus she spoke and went away. Him she left there | raging in his heart over the beautiful woman | whom they stole in spite of him unwilling. But Odysseus | reached Chryses, bringing sacred offering

At the verse end, the poet creates a deliberate “mismatch” between the repetitive metrical unities of the verses, and the grammatical clauses. Metrically, the transition 428–429 resembles 429–430. Both transitions have word end coinciding with verse end. Both feature an utterance that is perfectly understandable if the reader’s text were to stop, for example due to damage, at the verse end: the lines are grammatically complete. At both transitions, the verse end seems to demarcate a break in grammar [
79] that identifies the two lines, 428 and 429, as whole-line utterances. The terminology whole-line utterance points at the completeness of the grammatical and syntactical organisation within the single verse. It does not include any consideration for the perceptible phonological phrase. In verse 428 the grammatical clause τὸν δ᾿ ἔλιπ᾿ αὐτοῦ ‘him she left behind’ is “complete” at verse end; in 429 there is an expanded (κατὰ θυμὸν ‘in his heart’) verb form (χωόμενον ‘raging’) accompanied by a “causal” genitive (ἐυζώνοιο γυναικός ‘over the beautiful woman’). More interesting than the resemblances are the differences between the two end-stopped lines as indicated by phonology and phonetics.

In the transition 428–429, the verse end of 428, if observed, strengthens the hyperbaton (τὸν … χωόμενον); the transition 429–430 underlines the syntactical break between the main clause and what seems to be a subordinate, relative clause. The transition 430–431, however, more fundamentally questions the identification of any incompatibility. The reason for questioning the identification lies in the location of semantic domains, as reflected in phonetic realisation. The major phonological phrase, started by the minor phrase transitional constituent αὐτὰρ Ὀδυσσεὺς ‘but Odysseus’ introducing a grammatical clause, is not confined to the domain of the metrical unit of the single verse. This developing grammatical clause must be drawn into a syntactically coherent whole over the verse end of 430. The metrical colon αὐτὰρ Ὀδυσσεὺς cannot remain “on its own” and still function independently within a whole-line utterance, as τὸν δ᾿ ἔλιπ᾿ αὐτοῦ may. There are many more, similar instances, as shown in the appendix.

There appear to be only two ways out of the “mismatch”. The first is the one I have rejected in this study. It is a solution that is based on metrical analysis. It focuses on metrical colometry and on the equation of metrical and audible units. This solution accepts the incompatibility, and grants Homeric poetry numerous occurrences of verse-end enjambment, with variable “strength”. In this approach, the metrical unit that is the single verse is considered a persistently perceptible entity in performance, and given prominence in analysis accordingly. In other words, the single verse is seen as the basic unity for composition, and, possibly, the basic unity in performance. In the latter case, with the verse as the basic unit in performance, the single verse is the rhythmical unity as well. Still, enjambment is not necessarily considered a poetic device: the variation the poet aims at through incompatibility may be due to “rhythmic” variation. [80] My criticism centred around the notion that, if the single verse is seen as the basic unity for composition from a diachronic perspective, verse end enjambment is a mere acknowledgement of (theoretical) syntactical-metrical mismatch. [81] It is the recognition of something that is easily perceptible for readers, but possibly not so easily for a listening audience, whose only clue concerning the approach of line end may have been the gradually decreasing tempo of speech. This approach to the “mismatch” makes it more difficult to appreciate the poetic effectiveness of verse-end enjambment in oral performance. Still, poetic effect is widely and rather arbitrarily assumed. [82] But as acknowledgement of the incompatibility, such identification of enjambment takes as its point of departure the grammatical completeness within the single hexameter. Rhythmic variation would not take grammatical completion of the verse into account: it would first and foremost serve to indicate a coherence that has nothing to do with the metrical (and in performance: purely theoretical) “coherence” of the single line. The term “enjambment” to describe the phenomenon is, I think, particularly ill-chosen.

What the audience receives from the composing poet are “chunks”, “spurts of vocalisation”, “one-breath utterances”, that make the word groups they encompass appear as being closely tied together. The higher the rates of speech, the less demarcating the nonbreathing pauses are. At higher rates of speech, the one-breath utterance ties together smaller chunks that would appear as more strongly “prosodically characterised” in slow speech. The smaller chunks are still recognisable as such, but the phonetic disturbance at their demarcation is only slight. Secondary pauses facilitate both the start and the completion of such audible building blocks, the minor phonological phrases. Primary pauses indicate the boundaries of the major phrases into which the minor phrases are reorganised. When listening to an especially good composer, the audience might even be tricked into expecting more than what the minor phrase offers for the moment: the poet uses the audible pause effectively to create expectations-heightening enjambment. [84] But only the audible pause: the minor phrase is not framed by the repetitive metrical boundaries. Enjambment as the unavoidable acknowledgement of metrical boundaries [85] to poetic effect results in untimely prosodic disturbances. Untimely prosodic disturbances will merely hinder the audience’s perception. The one-breath utterance equals a prosodic unit, grouping together with other prosodic units to build larger wholes. Grammatical requirements do not determine Homeric discourse. Its movement allows grammatical requirements to have been met without stopping further minor phrases from attaching themselves. On the contrary, the divergence of syntax and prosody actually facilitates the continuing attachment of minor phrases, as connective phonetic words are tied into larger wholes together with the preceding phonological phrase. The metrical boundaries are in themselves often phonologically “designed” to mitigate the disturbance they cause to the audience’s perception. As secondary pauses, they are incorporated in the units framed by primary pauses.

5.3.2 Determination of grammatisation through prosody

In the previous chapters, it has become clear that perceptible prosodic phrasing supports a third way out that takes syntactical organisation as its starting point. The prosodic pattern built by primary and secondary pauses reveals a coincidence of hierarchical phonological and syntactical phrases. Metrical phrasing, however, remains normative in existing studies: especially verse-end enjambment and its poetic effectiveness are sought after. The Appendix to this study visualises the coincidence of hierarchical phonological and syntactical phrases, the patchwork of grammatical clauses and transitional constituents as the result of the mapping of phonetic disruption. As lead-in to the sample presented in the Appendix, I offer a finer-grained analysis of part of this sample, Odyssey 1.88–92, as a final example, both of the results to be expected from phonetic analysis, and of the confusion that may arise from fitting well established formulas into a narrative that is presented in grammatical clauses:

αὐτὰρ ἐγὼν Ἰθάκην ἐσελεύσομαι ὅφρα οἱ υἱὸν
μᾶλλον ἐποτρύνω καί οἱ μένος ἐν φρεσὶ θείω
εἰς ἀγορὴν καλέσαντα κάρη κομόωντας Ἀχαιοὺς
πᾶσι μνηστήρεσσιν ἀπειπέμεν οἳ τέ οἱ αἰεὶ
μῆλ᾿ ἁδινὰ σφάζουσι καὶ εἰλίποδας ἕλικας βοῦς

Odyssey 1.88–92

Now I shall go to Ithaca, that to his son | I may give more encouragement and put strength in the heart for him | to call to the place of assembly the long-haired Greeks | and to give notice to all suitors, who for him always | butcher the thick-thronging sheep and shambling, screw-horned cows

The whole-line utterance is not a persistent compositional device in these lines. The “rhetorical” punctuation in the edition of Stanford already suggests as much:

αὐτὰρ ἐγὼν Ἰθάκην ἐσελεύσομαι, ὅφρα οἱ υἱὸν
μᾶλλον ἐποτρύνω, καί οἱ μένος ἐν φρεσὶ θείω,
εἰς ἀγορὴν καλέσαντα κάρη κομόωντας Ἀχαιοὺς
πᾶσι μνηστήρεσσιν ἀπειπέμεν, οἳ τέ οἱ αἰεὶ
μῆλ᾿ ἁδινὰ σφάζουσι καὶ εἰλίποδας ἕλικας βοῦς.

Odyssey 1.88–92

Are all the printed comma’s more or less the same in the way the word end they mark is phonetically realised? Does printed punctuation indicate the primary or the secondary pause? Or no pause at all? Is there no third foot pause in 88, 89, 91 and 92? Is there no verse-internal pause at all in 90? Or does the inevitability of an audible third foot pause create some sort of meaningful verse-internal enjambment in 90? And what about verse-end enjambment? The verse end of 89 may coincide with the transition from one embedded predication to another. The verse end of 88 and 91 falls somewhere halfway the developing grammatical clause. Why would the clause οἱ υἱὸν μᾶλλον ἐποτρύνω signal the metrical boundary following υἱὸν? Are we supposed to forgive Homer for the fact that καλέσαντα is attracted into the construction of accusative and infinitive in stead of to οἱ precisely because “enjambment” is more than a theoretical acknowledgement? Is the participle phrase εἰς ἀγορὴν καλέσαντα κάρη κομόωντας Ἀχαιοὺς expanded into a whole line to grant verse-initial position to πᾶσι μνηστήρεσσιν so that any emphatic position for the clause’s verb ἀπειπέμεν would be further strengthened by the third foot caesura? What is the use of identifiable “enjambment” here?

This last question can be extended to other boundaries of metrical phrases like every third foot caesura:

αὐτὰρ ἐγὼν Ἰθάκην | ἐσελεύσομαι, ὅφρα οἱ υἱὸν ||
μᾶλλον ἐποτρύνω,| καί οἱ μένος ἐν φρεσὶ θείω, ||
εἰς ἀγορὴν καλέσαντα | κάρη κομόωντας Ἀχαιοὺς ||
πᾶσι μνηστήρεσσιν | ἀπειπέμεν, οἳ τέ οἱ αἰεὶ ||
μῆλ᾿ ἁδινὰ σφάζουσι | καὶ εἰλίποδας ἕλικας βοῦς. ||

Odyssey 1.88–92

The answer to the question does not necessarily have to be found in further sifting of the rhetorical punctuation: [
86] we have to abandon the alleged “paratactic” juxtaposition of metrical phrases. After all, rhetorical punctuation can be visualised with similar ease in prose:

αὐτὰρ ἐγὼν Ἰθάκην ἐσελεύσομαι, ὅφρα οἱ υἱὸν μᾶλλον ἐποτρύνω, καί οἱ μένος ἐν φρεσὶ θείω, εἰς ἀγορὴν καλέσαντα κάρη κομόωντας Ἀχαιοὺς πᾶσι μνηστήρεσσιν ἀπειπέμεν, οἳ τέ οἱ αἰεὶ μῆλ᾿ ἁδινὰ σφάζουσι καὶ εἰλίποδας ἕλικας βοῦς.

Odyssey 1. 88–92

Together with metrical demarcation, rhetorical punctuation can be visualised when the text is divided into the chunks of spoken language, Bakker’s “intonation units” (a–l) as discussed in chapter 2:

a        αὐτὰρ ἐγὼν Ἰθάκην
b        ἐσελεύσομαι,
c        ὅφρα οἱ υἱὸν
d        μᾶλλον ἐποτρύνω,
e        καί οἱ μένος ἐν φρεσὶ θείω,
f        εἰς ἀγορὴν καλέσαντα
g        κάρη κομόωντας Ἀχαιοὺς
h        πᾶσι μνηστήρεσσιν
i        ἀπειπέμεν,
j        οἳ τέ οἱ αἰεὶ
k        μῆλ᾿ ἁδινὰ σφάζουσι
l        καὶ εἰλίποδας ἕλικας βοῦς.

Odyssey 1.88–92

The syntactical structure, however, is clearly built despite the metrical phrasing. More precise, its likeness to spoken discourse becomes visible not in metrical, but in phonological phrasing. Applying the criteria for primary and secondary pauses, as I presented them in the introduction to this chapter, brings out the phonological phrasing. I will start by marking all the minor phonological phrases, regardless of the strength of the phonetic disruption that starts or ends the phrase. As pointed out, phonetic disruption stems from phonetic-word end in a syllable as opposed to the phonological segment:

i        αὐτὰρ ἐγὼν Ἰθάκην ἐσελεύσομαι, ὅφρα οἱ υἱὸν μᾶλλον ἐποτρύνω,
ii       καί οἱ μένος ἐν φρεσὶ θείω,
iii      εἰς ἀγορὴν
iv      καλέσαντα κάρη
v       κομόωντας Ἀχαιοὺς
vi      πᾶσι μνηστήρεσσιν ἀπειπέμεν, οἳ τέ
vii     οἱ αἰεὶ μῆλ᾿ ἁδινὰ σφάζουσι καὶ εἰλίποδας ἕλικας
viii    βοῦς.

Odyssey 1.88–92

A number of issues calls for extra attention. The first minor phrase is very long and includes what seems to be a complete subordinate clause that straddles verse end. The second minor phrase, καί οἱ μένος ἐν φρεσὶ θείω, is a complete and completed grammatical clause. The third, εἰς ἀγορὴν, appears to be isolated in the way one would expect an additive chunk to be isolated, just like the two phrases that follow. In the sixth minor phrase, the transitional constituent providing continuation is included. The seventh minor phrase is long again and encompasses a grammatical clause that straddles verse end. Finally, the eighth phrase, βοῦς, is remarkable. It consists of a single monosyllabic word that is thrown into relief by the phonetic realisation. [
87] Whereas word end in a heavy syllable always results in an anapaestic word end, phrasal metarrhythmisis to catalectic anapaests is twice in order here: at position 12 in Odyssey 1.90 Ἀχαιοὺς (AT[A]T), and at position 12 in Odyssey 1.92 βοῦς ([A]T).

Applying the two different pauses to the minor phrases leads to the following visualisation of the way they phonetically start and end (primary pause/major phrase boundary °, secondary pause ˙):

i         ˙αὐτὰρ ἐγὼν Ἰθάκην ἐσελεύσομαι, ὅφρα οἱ υἱὸν μᾶλλον ἐποτρύνω,°
ii        °καί οἱ μένος ἐν φρεσὶ θείω,˙
iii       ˙εἰς ἀγορὴν°
iv       °καλέσαντα κάρη°
v        °κομόωντας Ἀχαιοὺς°
vi       °πᾶσι μνηστήρεσσιν ἀπειπέμεν, οἳ τέ˙
vii      ˙οἱ αἰεὶ μῆλ᾿ ἁδινὰ σφάζουσι καὶ εἰλίποδας ἕλικας°
viii     °βοῦς.°

Odyssey 1.88–92

The primary pauses are the start and end of major phonological phrases. At normal rates of speech, minor phrases will be reorganised into subphrases of the major phrase. This leads to six major phrases:

  1. ˙αὐτὰρ ἐγὼν Ἰθάκην ἐσελεύσομαι, ὅφρα οἱ υἱὸν μᾶλλον ἐποτρύνω,°
  2. °καί οἱ μένος ἐν φρεσὶ θείω,˙εἰς ἀγορὴν°
  3. °καλέσαντα κάρη°
  4. °κομόωντας Ἀχαιοὺς°
  5. °πᾶσι μνηστήρεσσιν ἀπειπέμεν, οἳ τέ˙οἱ αἰεὶ μῆλ᾿ ἁδινὰ σφάζουσι καὶ εἰλίποδας ἕλικας°
  6. °βοῦς.°

Odyssey 1.88-92

The issue of the sentence and the clause (‘which of these two ought to be taken as the syntactical unit to judge the grammatical completeness of the whole-line utterance?’) is indeed no longer relevant in this approach. [
88] In my opinion, however, it may be still. It is noteworthy that some grammatical clauses are tied together in, and by, a phonological phrase. On the other hand, if primary pause disrupting a grammatical clause is an indicator for (verse-internal) “enjambment”, there is remarkably often “enjambment” in and around much-used formulaic expressions (κάρη κομόωντας Ἀχαιοὺς, ἕλικας βοῦς). Earlier, [89] I pointed out that phonological phrasing tends to lead to remarkable phonetic disruption in and around well established formulas, particularly noun-epithet combinations. Especially in these combinations, I am inclined to lessen the disruptions caused by the primary pause somewhat, as the usefulness of the formula may have been felt as more important than the slightly awkward phonetics of the word group in performance. At the same time, rhetorical punctuation suggests the boundaries of smaller scale syntactical units that cannot be made audible as units. Punctuation, however, does not always indicate the start and ending of smaller units that can be made audible. The smaller scale units contribute to audible phrase variation: they are not characterised metrically, but phonologically. They do not reflect the coherence of the metrical unit, but reorganise metrical word-types along the lines of a different type of coherence. At times this coherence, determined by prosody, evidences the demarcation of larger scale units in Homer as visualised by means of rhetorical punctuation. Then again, the example Odyssey 1.88–92 shows that we must remain careful not to attribute too much importance to larger scale demarcation in Homer. [90] Any larger scale demarcation I have made identifiable, is no more than the identification of major phonological phrases comprising of one or more minor phrases. Larger scale is not a division into sentences: it is a division into perceptible units of variable size and internal shape. The importance of meter suffers most from the phonological-syntactical phrasing. “Rhetorical colometry”, including the abundant (verse-internal) “enjambment”, defies not only the verse end: there are other metrical boundaries that are observed or straddled, depending on syntactical analysis. Pure “syntactical chunking” (κατὰ διαστολάς) is confined between phonological boundaries, but seems to make the repetitive pattern of Homeric metrical phrasing less relevant:

          αὐτὰρ ἐγὼν Ἰθάκην ἐσελεύσομαι, ὅφρα οἱ υἱὸν μᾶλλον ἐποτρύνω,
καί οἱ μένος ἐν φρεσὶ θείω,
          εἰς ἀγορὴν (…) καλέσαντα κάρη (…) κομόωντας Ἀχαιοὺς
πᾶσι μνηστήρεσσιν ἀπειπέμεν, οἳ τέ
          οἱ αἰεὶ μῆλ᾿ ἁδινὰ σφάζουσι καὶ εἰλίποδας ἕλικας (…) βοῦς.

This is as close to “prose-like” phrasing as Homer gets.

5.4 Conclusion

Combining the conclusions of chapters 1 to 5, I have shown that the syntactical contours in Homer are not reflected in metrical phrasing: not in performance, and not even in composition. The prosodic isolation of minor phonological phrases is not supported by semantic or syntactical isolation of the minor phrase. On the contrary: the minor phrase is both audibly and “sensibly” drawn into a larger scale phonological whole, sometimes with regard for “semantic hierarchy” within the resulting larger phrase. The prosodic realisation of the minor phrases’ boundaries has thus proven to be an incentive for grammatisation into coherent major phonological phrases. The major phrases demarcated by audible pauses evoke a perception of clausal variety and freedom that is to an extent reminiscent of prose composition.

I would have liked to add that prosodic analysis of the Homeric epic proves that prose-like phrase formation fully applied to the Iliad and the Odyssey. But it does not. Homer is not a continuous piece of prose; no one ever said it was. Dionysius of Halicarnassus describes the Homeric epic as “intermediate composition”. It stands between “austere composition” and prose. “Intermediate” composition has something of both: it features both the whole-line utterance of austere composition, and the more freely formed grammatical clause from prose. If there is any “mismatch” in Homeric composition, it is due to what I would label prose elements in the prosody of the Iliad and the Odyssey. These prose elements, again, show their basis in both prosody and syntax. They are reflections of grammatisation: syntactical development facilitated by prosody. Prosodic developments made it possible to extend the domain of the syntactical coherent phrase over, or despite, the boundaries of the metrical phrase. Eventually, even the variance and choice in phonetic demarcation, that suggested some sort of prosodic hierarchy, facilitated the further development of grammatical governance and syntactical hierarchy. At the same time, grammatisation implies that the prosodic modifications and adaptations are the motivation behind syntactical development. Syntax emerges in response to phonetic circumstances—or rather selected possibilities; syntax and prosody develop together. The Iliad and the Odyssey are our last remaining witnesses of the coincidence of developing syntax and framing phonetics.


[ back ] 1. In chapter 4, I have argued that the prepausal syllable before the primary pause is the μακρὰ τελεία, as meant by Dionysius of Halicarnassus On Literary Composition 17 & 20.

[ back ] 2. Depending on the type of transition at verse end, Higbie’s, as the most elaborate enjambment classification (1990), labels the transition a) clausal-internal enjambment, b) adding enjambment / clausal external enjambment, and c) unenjambed.

[ back ] 3. Koster 1953:51–52; Devine & Stephens 1978.

[ back ] 4. West 1982:36 states that position 3 doubles as a sense-pause in 7% of all verses, position 2 in 6%.

[ back ] 5. Cf. the localisation of δέ and τε at verse end; Maas 1962:87.

[ back ] 6. Cf. Denniston 1954:185.

[ back ] 7. Bakker 1997a:54–85.

[ back ] 8. Perhaps retaining some of its prepausal characterisation as a result of the realisation of the enclitic as a long segment, cf. the location of enclitics at verse end.

[ back ] 9. Ruijgh 1990; Devine & Stephens 1994:422–423; Bakker 1993:15–25; 1997b; 2005:92–113.

[ back ] 10. Cf. the remarks on extra-clausal constituents in Hellenistic hexameters in Van Raalte 1986:69–70.

[ back ] 11. Accordingly, enclitic postpositives like δέ and τε are rare but not impossible at verse end, cf. Maas 1962:87.

[ back ] 12. Cf. left-dislocation (Bakker 1990), staging (Bakker 1997b), and disambiguation (Bakker 2005:13).

[ back ] 13. I prefer to read, contra the manuscripts and contra West 1998, στέμμα τ᾿ ἔχων with Bentley, in accordance with Iliad 1.28. Leaf’s argument (1900–1902:4–5) that the participle ‘is subordinate to the preceding participles, indicating a detail, and not co-ordinate with λυσόμενός, expressing the main object of his journey’ cannot be held when dealing with paratactic style.

[ back ] 14. The difference is of course primarily due to καί not being postpositive.

[ back ] 15. Which raises questions on the τε in Iliad 1.7: is it preparatory to καί or clausal-resumptive following the verse end of line 6?

[ back ] 16. Cf. in addition the remarks on interrogative prepausal intonation in Devine & Stephens 1994:429–433.

[ back ] 17. Devine & Stephens 1994:323–326.

[ back ] 18. As I argued in 2.2.3.

[ back ] 19. Devine & Stephens 1994:309, 315–316, 353–357.

[ back ] 20. Devine & Stephens 1994:310.

[ back ] 21. Bakker 1993:15–25; 2005:11–12.

[ back ] 22. Denniston 1954:23–31.

[ back ] 23. In 6% of all verses in the sample of the Appendix.

[ back ] 24. Depending on the analysis of περὶ (adverb or preposition), the rhythmical disruption at the right branch of the phonetic word περὶ δὲ may be nullified due to the prepositive direction of περὶ δὲ.

[ back ] 25. On the subordinating conjunctions, see Chantraine 1953(II):235; on the usage of modes in subordinate clauses: 205, 235.

[ back ] 26. On the “looseness” of style, see Chantraine 1953(II):234.

[ back ] 27. Devine & Stephens 1994:303.

[ back ] 28. Cf. the use of the subjunctive with κε in a main clause like Iliad 1.324 ἐγὼ δέ κεν αὐτὸς ἕλωμαι.

[ back ] 29. Denniston 1954:32 notes that ἄρα is perhaps connected with Lithuanian ‘and’, and that it ‘denotes connexion (consequence or mere succession)’, only to conclude, when considering ‘the almost reckless profusion with which it is used’ (33), that ‘it is surely improbable that in Homer, where logical relationships are, broadly speaking, seldom emphasized or very precisely defined, the idea of connexion should thus be obtruded at every turn.’ Firmer ground can be found in Kühner & Gerth 1963(III):318: ‘Es dient zunächst, wie τε, zur Anreihung, und zwar zur Anreihung von Begriffen, die miteinander in einem gewissen natürlichen Zusammenhange stehen. Insofern aber dieser Zusammenhang sich bei den durch ἄρα verknüpften Sätzen vorwiegend als das Verhältnis von Grund und Folge darstellt, war die Partikel ganz besonders geeignet, sich allmählich aus einem anknüpfenden dann (ferner) einerseits, in ein folgerndes dann, denn (und so denn), anderseits in ein erklärendes denn (nämlich) umzuwandeln.’ (my italics).

[ back ] 30. Cf. the comment by the scholiast (bT): why does Hera support the Greeks and Achilles’s wrath?

[ back ] 31. Kirk 1985:59.

[ back ] 32. With elision turning the postpositive monosyllabic enclitic into a consonant-onset.

[ back ] 33. The closest example is only 8 lines away, in Iliad 1.65: εἴ ταρ ὅ γ᾿ εὐχωλῆς ἐπιμέμφεται ἠδ᾿ ἑκατόμβης. It is precisely the resemblance to a line like Iliad 1.57 that, to me, shows that Aristarchus’s reading εἴ ταρ is better than the vulgate εἴτ᾿ ἄρ᾿.

[ back ] 34. Leaf 1900–1902:8; Kirk 1985:59.

[ back ] 35. Cf. Iliad 1.81 εἴ περ γάρ τε χόλον γε καὶ αὐτῆμαρ καταπέψηι where γε is meant to emphasise χόλον, suggesting an awkward minor phrase εἴ περ γάρ τε (or εἴ περ, γάρ τε?). I find it an attractive idea in Erbse:1969 ad loc. (contra Kirk 1985), that Zenodotus athetised the line, which is a rhythmical monster.

[ back ] 36. Cf. Iliad 22.156.

[ back ] 37. Together the two verses 97–98 are suspect, or at least non-representative: their mismatch of phonological and syntactical realisations gives rise to suspicions concerning later development.

[ back ] 38. Cf. Chantraine 1953(II):346.

[ back ] 39. Chantraine 1953(II):232–235, 351–364.

[ back ] 40. That is how its meaning and location were interpreted by various authors of scholia (e.g. AT: ἤρξατο μὲν ἀπὸ μήνιδος ἐπείπερ αὕτη τοῖς πρακτικοῖς ὑπόθεσις γέγονεν), cf. Erbse 1969 ad loc.

[ back ] 41. Kirk 1985:52 ‘The wrath of which the goddess is to sing will persist throughout the entire poem and is to determine, in a sense, the fate of Troy; [. . . . .] its immediate beginning is the subject of book 1 which follows.’

[ back ] 42. Bakker 2005:5.

[ back ] 43. As the poetic purpose of enjambment is regularly described as emphasis on the word preceding or following the enjambment, cf. Edwards 1991:42–44; Bakker 2005:53–55.

[ back ] 44. Cf. Bakker 2005:6.

[ back ] 45. Bakker 2005:11.

[ back ] 46. A remarkable phrase, considering the further absence of elliptic answers to direct questions in Homer. When it comes to shape and position within the verse, the first hemistich of Iliad 1.9 resembles Bakker’s right-dislocation (Bakker 1990; 1997a:89–108; 1997b:293–297; 2005:12 n33; cf. Devine & Stephens 2000:143–144). On the other hand, the answer, especially since it is in nominative case, resembles the parenthetic position of the nominative case, i.c. a parenthesis to τίς in Iliad 1.8; cf. Bakker 1997a:198–200; 2005:13.

[ back ] 47. A remarkable case is the location under Wernicke’s law of Ἀχαιῶν with word end at position 8, see 4.6.3 above.

[ back ] 48. Bakker 1997b.

[ back ] 49. Cf. schol.: ὅρα δὲ τὰς ἐν ἀρχῇ ἀναφωνήσεις; cf. Erbse 1969:6.

[ back ] 50. For the participle as a ῥῆμα in Dionysius of Halicarnassus’s theory, see Ruijgh 1987:342n62 and further discussion in chapter 2.

[ back ] 51. Bakker 2005:54–55 points out that ‘a unit can straddle the verse-boundary’, but reserves the possibility of this ‘fugal effect’ to ‘some of the more emotional moments’.

[ back ] 52. Clark 1997:35–40.

[ back ] 53. Cf. the “clausal” division in Clark 1997:21–30.

[ back ] 54. Devine & Stephens 2000:115–118.

[ back ] 55. Prosodic considerations are not included in the treatment of the licensing and origin of hyperbaton in Devine and Stephens 2000:141–174.

[ back ] 56. Devine & Stephens 2000:142–148.

[ back ] 57. Chantraine 1953(II):12.

[ back ] 58. Devine & Stephens 2000:142.

[ back ] 59. In personal communication Joel Lidov points out that Iliad 1.36 is very likely a case of “free indirect discourse”. Change the dative in vocative, and Iliad 1.36 is the start of a formal prayer, to be continued and turned into a complete prayer in line 37. Cf. his comments in Lidov 1996:134 with references.

[ back ] 60. Bakker 1997b, especially 293–297.

[ back ] 61. In his 1990 study, Bakker used the terminology right-dislocation.

[ back ] 62. Bakker 2005:13; cf. 1997a:198–200.

[ back ] 63. The right-dislocation, or pragmatic function as tail-constituent, may soften the effect of the verse end enjambment, as may left-dislocation (cf. Bakker 1997b:303–303; 2005:53). In my view, however, it is the other way around: the strength of the pause separating the apposition from the “intonation unit” it is appositional to, defines the likeliness of identification as a true grammatical apposition.

[ back ] 64. Cf. the intonation units m and n in Bakker’s presentation of Iliad 1.1–7.

[ back ] 65. Bakker 1997a:146–155; 2005:48.

[ back ] 66. Bakker 1997b:303; cf. 2005:54–55.

[ back ] 67. See 2.1 (especially 2.1.1), and 5.1.1 above.

[ back ] 68. Devine & Stephens 2000 deal with hyperbaton in prose. Remarks on poetry and orality may be found on pages 202–203, 206–209.

[ back ] 69. Bakker 1997b.

[ back ] 70. Devine & Stephens 2000:69–70. Note the usage of μή … οὐ in Iliad 1.28 μή νύ τοι οὐ χραίσμηι σκῆπτρον καὶ στέμμα θεοῖο ‘surely your staff will not rescue you, nor the wreath of the god’ to negate the postponed verbal form.

[ back ] 71. Cf. Devine & Stephens 1994:417–418. Arguments against a pause in comparable instances in chapter 4.

[ back ] 72. Cf. the frequent mention of Ἠώς and Ἠέλιος; of course their appearance in the text suggests a good location for book division.

[ back ] 73. Though Iliad 1.36 may be an instance of free indirect discourse (cf. footnote 59 above). In that case line 36 supplies what Chryses will omit as a lead-in to his prayer: line 37 presents the conventional start of a prayer in direct quotation.

[ back ] 74. 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 ] 75. The sample in the Appendix provides the following statistics:

  Iliad     Odyssey    
  Clause end* “Enjambment” Other** Clause end* “Enjambment” Other**
Pos. 3 8.10% 9.30% 8.1% 7.50% 12.50% 13.75%
Pos. 4 0.00% 0.00% 0.00% 0.00% 1.25% 0.00%
Pos. 5 17.44% 5.81% 3.49% 6.25% 7.50% 3.75%
Pos. 7 11.63% 5.81% 3.49% 5.00% 10.00% 2.50%
Pos. 8 2.32% 0.00% 0.00% 0.00% 1.25% 2.50%
Pos. 9 2.32% 2.32% 1.20% 0.00% 3.75% 1.25%
Pos. 11 0.00% 0.00% 0.00% 0.00% 1.25% 0.00%
Pos. 12 12.79% 3.48% 2.32% 7.50% 5.00% 7.50%

Table 6: Demarcation due to the primary rhythmical pause as percentage of total number of primary pauses.
* Including the right branch demarcation of the predicatively used participle.
** The right branch demarcation of a transitional constituent or a cluster of transitional constituents.

[ back ] 76. Cf. Rubin 1995.

[ back ] 77. Devine & Stephens 1994:400.

[ back ] 78. Kirk 1966:113; Clark 1997:26.

[ back ] 79. Clark 1997:21–30; 2004.

[ back ] 80. Clark 1997:26.

[ back ] 81. If rhythmic variation was the aim of the poet and the effect of “enjambment”, the notion of “strengths” is misleading, as is the terminology “enjambment” itself.

[ back ] 82. Cf., for example, Kirk 1985, 1990 passim. For arguments to re-evaluate the poetic effectiveness of enjambment, see Edwards 1991:42–44; Bakker 2005:53–54.

[ back ] 83. For Basset 1938:150–154, verse-end enjambment does not carry any special poetic weight like e.g. emphasis. Barnes 1970 tries to narrow the usage of the terminology enjambment. The focus on syntactical coherence in Clayman & Van Nortwick 1977 and Clayman 1981 did not stop the terminology from being used to indicate the theoretical incompatibility. Cf. Grotjahn 1981; Clark 1994; Kahane 1997; Bakker 1997b.

[ back ] 84. The consequence of this observation may be that expectations-heightening enjambment is primarily to be found verse-internally. See for the statistics 5.3.

[ back ] 85. Clark 1997:28 ‘All runovers form enjambments’.

[ back ] 86. As Daitz 1991 suggests, or even in banning all printed punctuation from our texts.

[ back ] 87. Stanford 1950:218 comments: ‘perhaps H. deliberately intended to suggest the heavy movement of the cattle with this heavy monosyllabic ending.’

[ back ] 88. As already pointed out by Bakker 1997a:54–71; 1997b.

[ back ] 89. See pages 157, 199, and 214–215.

[ back ] 90. As mentioned in chapter 1, Bakker 1997b explains how the syntax of Homeric discourse is a “movement” due to continuation and addition. In antiquity, scholars describe the style of the Homeric narrative as “strung style” (Aristotle: λέξις εἰρομένη; Dionysius of Halicarnassus: λόγος εἰρόμενος) while pointing at the ease with which it keeps the narrative continuing.