Use the following persistent identifier: http://nrs.harvard.edu/urn-3:hul.ebook:CHS_BlankenborgR.Rhythm_without_Beat.2014.
3. Metrics and Phonology of Grammatical Clauses
3.1 The Phrasal Domain
3.1.1 Coherence and demarcation of the phrasal domain
The syllable structures are such that they correspond to the weight of the metrical positions they occupy. Prosodic syllabification is confined to the metrical unit of the verse: prosodic syllabification phenomena like shortening, liaison, and elision demonstrate the συνάφεια “juncture” within the verse. This “juncture” is broken at the metrical boundary of the verse-end: shortening, liaison, and elision do not occur there. Sometimes, the ‘juncture’ is broken at word-end within the verse. The συνάφεια indicates a type of coherence within the metrical unit of the individual verse. It is this coherence that facilitates the division of the verse into vowel-centred sounds that do not correspond with morphological or orthographic syllabification: the vowel-centred sounds are characterised prosodically, not orthographically. The prosodic “cutting up” of the line, however, can hardly be called “syllabification” as its clusters of “syllables” do not correspond with morphological lexemes, that is, the understandable and separable units of meaning (words and affixes). The orthographical syllables, a further subdivision of lexemes in single units of articulatory prominence,  do not signify quantity.  The units of quantity are phonemes, sounds that are formed regardless of morphology. It is better to leave the general terminology “syllable” behind and refer to the units of prosodic cutting up as “segments”. As phonology determines the exact shape of the segments, prosodic “syllabification” is best referred to as “phonological segmentation”. The segmentation itself is coherent within the verse, but cannot transgress the boundaries of the single line. As mentioned above, the phonemes of phonological segmentation are best referred to as segments. The terminology “syllable” is then only applicable to segments that are free from the influence of sandhi: liaison of consonants (or συνάφεια), and vowel coalescence (or συναλοιφή) through elision and shortening.
As was shown in the modification to the analysis of Iliad 1.1–16 in 2.2.3, elision looks like the coda to onset type of resyllabification, but is in fact reduction of a syllable to onset. In other words, elision creates a segment-initial consonant  . The elided vocalic sound is still present, though severely reduced: elision affects the pitch pattern in case of an elided accented syllable, and in case of elision of clitics with secondary accentuation:
In case of an elided monosyllable, the remaining consonant turns into an orthographic onset, that is, the syllable-initial consonant(s):
Hiatus, on the other hand, is a means of demarcating the phrasal domain. Following a heavy syllable containing a long word-final vowel or diphthong, hiatus can be avoided in the arsis by shortening: 
Apart from hiatus, demarcation of the phrasal domain is indicated by the simultaneous completion of both phonological segment and orthographical syllable. Demarcation of the domain is hence found at the appearance of phonetic word-final syllables, including not-shortened heavy phonetic word-final syllables containing a long vowel or a diphthong: 
Next to word-final light syllables on the arsis, another indication for demarcation is the syllable containing a short vowel on the foot’s first element:
δακρυόφιν πλῆσθεν θαλερὴ δέ οἱ ἔσχετο φωνή
The coherence and demarcation of the phrasal domain show that there is a second type of “prosodic unit”, next to the phonetic word: there is a phonological phrase that is larger than the metrical-colon appositive group, but not necessarily equal in size to the single hexametric verse. At the same time, this larger phrase seems to incorporate the, smaller size, phonetic words. The larger phrase resembles the major phonological phrase, the phonetic word the minor phonological phrase. If this comparison makes any sense for Homeric Greek, how do these two types of phonological phrases create one pattern of phrasal domains?
3.1.2 Reorganisation of the phrasal domain
Devine and Stephens identify the minor phrase as a prosodic unit that forms the domain for grammatical development because of its characterisation as a domain. In Greek, the combination of clisis and metrical phrasing is evidence of the coherence of the phonological domain, and hence, as Devine and Stephens point out, of some coherence within a syntactical domain. They point at the positions of frequent word end, the occurrence of sandhi, and the minor phrase boost of the accent (as evidenced in the musical settings of the Delphic hymns) as proof of the prosodic coherence within the minor phonological phrase. In most respects, the minor phrase in Greek works, I would say, like the intonation-unit metrical phrase in Homer as presented as the unit of Homeric discourse by Bakker. In chapter 2, I showed that both prosodic and syntactical clisis turns enclitics into “bidirectionals” as intonation-unit transitional constituents turn into sentential prepositive phonetic words. That way, the minor-phrase intonation unit is drawn into a larger whole (Devine and Stephens speak of ‘prosodic unit’), reflecting a higher level of grammatisation, and demarcated by audibly stronger disruptions at normal rates of speech. These larger scale units resemble what Devine and Stephens describe as major phonological phrases:
The demarcation of minor and major phonological phrases is hence a matter of both meter and apposition (for the minor phrase), and phonetics (for the audible phenomena demarcating the major phrase). In the remainder of this chapter, I will first analyse the metrical pattern of the grammatical clause and its intonation units. It is easy to identify and the easiest way to provide evidence for the prosodic determination of phrasal-domain reorganisation. The phonological pattern is a different story entirely; it provides the rationalisation of the possibilities and impossibilities of phrase termination, but phrase termination can not change the metrical surface structure. Phonology evidences the internal coherence of the phonological phrase and its demarcation. Phonetic clues for phrase termination are hence identifiable. Their identification requires a new approach of the audible pause that is partly my own contribution in this study. The description of Homeric syntax as a progressive tendency featuring grammatical clauses will finds its reflection in the patterning of audible pauses, discussed in detail in chapter 4. Before that, following the description of the grammatical clause’s metrical and phonological pattern, I will assess the clues from antiquity for the perception of Homeric discourse as a progressive tendency due to a series of grammatical clauses.
3.2 Meter and the Homeric Grammatical Clause
3.3 The Metrical Pattern of the Homeric Grammatical Clause
3.3.1 The start of the clause
The start of a grammatical clause after position 1 by means of a heavy segment at position 2 is not common. When used, it is often in combination with an elided particle at position 1, in a formula, or with an elided particle in a formula, like τὸν δ᾿ αὖτε προσέειπε ‘and to him, in turn, said’. In this formula, the sandhi due to elision makes the grammatical clause start at the beginning of the verse: another indication that clause start with a heavy segment at position 2 is avoided. The example Iliad 10.344 is clear enough, but looks like a modification of a prototype starting from 1 [+ ⏑ ⏑]. Leaf  suggests original ἀλλ᾿ ἐάωμεν πρῶτα, pointing at the irregular position of μιν, but retains the elision. It is difficult to determine whether or not reorganisation through elision postdates the modification of formulaic prototypes: both have the major-phrase hemistich as their domain.  It would be theoretically sound, I think, to correct to ἀλλὰ ἐῶμεν πρῶτα (as do Brandreth and Von Christ).
The start of a clause after position 2 can of course be preceded by a heavy syllable as in Iliad 10.203 ἀγχοῦ δ᾿ ἱστάμενος ἔπεα πρερόεντα προσηύδα ‘standing close to him he spoke the winged words’.
Similar to clauses starting after 1 [+ heavy segment], clauses starting after 3 [+ heavy segment] may show signs of modification.
Hilberg’s law is being observed. The phonetic-word boundary following enclitic τοι, is reversed by the prepositive character of οὔ τοι:
The particle τοι becomes bidirectional  (phonetic word [οὔ τοι πάντες]).
Contrary to what one may expect, the number of grammatical clauses starting with double short after position 5 (after the masculine third foot caesura) is small.
Clause start after position 6 is rare. Not surprisingly, as it falls between two positions of frequent word end, not being a position of frequent phonetic-word end itself. In Odyssey 10.79 there is no phonetic-word end due to shortening.  Phonetic word end at position 6 is not impossible, though. Since clause start with a heavy segment is regularly preceded by a postpositive, light syllable, clause start after position 6 is unlikely when position 6 is itself occupied by a heavy syllable. A heavy syllable on position 6 may be followed by elision of the particle, as in Odyssey 8.505:
The heavy syllable + elided particle is rare at this position throughout the Iliad and the Odyssey. With double-light at position 6, the same remarkably often applies: the phonetic word between positions 5 and 7 is kept intact.
Reorganisation is clearly visible in Odyssey 7.110: original πέρι γὰρ σφίσι δῶκεν Ἀθηνη is likely. Odyssey 5.146 shows reorganisation in the elision of δ᾿.
This phenomenon is common to the Iliad and the Odyssey. Phonologically, position 7 is straddled due to (geminate -μμ-) onset-to-coda resyllabification: δὲμ.μέ.γα.  Elision of the particle in Odyssey 5.146 is a clever echo of Iliad 1.5: Διὸς δ᾿ ἐτελείετο βουλή (including reference to a μῆνις). The elision is equally common:
ἑσταότες κατ᾿ ἀγῶνα πολὺς δ᾿ 7 [ὑπὸ κόμπος ὀρώρει
Clause start after an elided particle is common at this position, in this example allowing for a pronoun at position 7½. A combination like ὁ δ᾿ maintains the word end at position 7 and Hermann’s bridge: there is no room for another word end in the fourth foot, demarcating the extra-clausal constituent. When the transitional particle is given in full (not elided) at position 7½, a feminine fourth foot caesura is avoided by adding another postpositive particle or a pronoun:
The bucolic dieresis can easily be used to start a new clause. There is a clear preference for starting the grammatical clause after position 8 over starting after 7½. It is noteworthy that clause starts after position 8 are not as often preceded by preparatory pragmatic constituents (like theme) as starts after other positions. The clause start can be preceded by the transitional constituent καί, as in Odyssey 14.404:
Alternatively, the clause after position 8 starts with a verb form (including the participle), or with a proclitic connector:
The well-known continuation after position 8 αὐτὰρ ἔπειτα ‘but then’ is itself a transitional pragmatic constituent, to be compared with other pragmatic constituents that are intonation units fitting the metrical pattern after the bucolic dieresis. These observations strongly suggest that the bucolic dieresis can be realised as both a minor-phrase and a major-phrase boundary. Prepausal location of the word-final biceps on position 8 is suggested by punctuation  at the bucolic dieresis, but is not markedly realised in phonetics:
In Odyssey 15.82, the realisation of phonetic-word end is frustrated by shortening,  in 15.88 by elision. Realisation of phonetic-word end is equally frustrated by sandhi in formulaic expressions:
The realisation of sandhi over the metrical boundary frustrates phonetic word end. As such, the presence or absence of sandhi is a feature of phrase boundary. The bucolic dieresis must be demarcated by the minor-phrase boundary in order to be realised as a prosodic boundary. If there is no phonetic word end at the bucolic dieresis, there is no minor-phrase boundary:
οὔ τι καταισχύνοντα τεὸν γένος : 8 ([)ὡς ἀγορεύεις
Though frequent as a position of word end, position 9 is not often used to demarcate the phonological phrase. It is phonologically straddled by enclitics (ἡ :9 μὲν ἔπειτα / οἱ :9 δὲ δὴ ἄλλοι / παῖδας :9 τ᾿ ἄλοχόν τε), or proclitics (αἴ κ᾿ :9 ἐλεήσηι / ὡς :9 ἐτέτυκτο / ὣς :9 τὸ πάρος περ).
Odyssey 13.389 is not an example of clause start with a heavy segment after position 9; it is an example of position 9 as the demarcation of a phonological phrase. Clause start with a heavy segment after position 9 requires a spondaic verse with word end in the fifth foot. Split fifth-foot spondees are rare in Homer, let alone clause start with a heavy segment after position 9. If one reads formulaic ὄφρ᾿ ἐὺ εἰδῶ as ὄφρ᾿ εὖ εἰδῶ (Odyssey 1.174), one gets close, but the elision shows that in the dactylic hexameter the split spondee is reorganised to a metrical bridge in the fifth foot.
Just as after positions 7½ and (in some cases) 8, the start of a new clause after position 9½ is often preceded by a pragmatic constituent filling the metrical pattern of the minor phrase from the nearest phonetic-word end (in this case, at the bucolic dieresis φίλε :8 σοὶ). Compare the use of ὄφρα, or αὐτάρ in, for example, Odyssey 1.215:
Clause start after position 10 is infrequent thought not impossible. The actual clause start may be after 9½ with position 10 being filled with a clitic pronoun or a particle like καί, τοι or δή. Proper clause start after position 10 can be found in, for example, Iliad 18.358–359 with σεῖο at the close of the hexameter:
ἐξ αὐτῆς ἐγένοντο
Another example is Odyssey 7.69:
The prepositive character of καὶ is, as so often, further strengthened by shortening. The resulting single verb form ἔστιν remains elliptic and awkward: in his commentary, Hainsworth  suggests Plato Symposium 195b μετὰ δὲ νέων ἀεὶ ξύνεστί τε καὶ ἐστι as a parallel. Extension through particles may, however, postpone the true clause start until the start of the subsequent hexameter:
On the other hand, clause start after position 10 may even result in completion of that clause at position 12 in the same line:
Clause start after position 11 only seldom occurs in Homer, and the example quoted stands in brackets: it can only be accepted if the participle is seen as equivalent to a finite verb. Verse end may be in enclitic particles like τε, γε, περ, or in monosyllabic words like δῶ, κῆρ, φώς, βοῦς, χρῆ or Ζεύς, but verse end in a combination [transitional extra-clausals + monosyllabic] (for example: *ὁ δὲ Ζεύς || ) is not found. Iliad 1.128 τριπλῆι τετραπλῆι τ᾿ ἀποτίσομεν αἴ κέ ποθι Ζεύς ‘threefold, fourfold we will compensate you, if Zeus ever’ comes close, but of course the minor phrase contains at least [αἴ κέ ποθι Ζεύς]. The prosodic bridge that hampers monosyllabic ending of the hexameter renders phonetic-word end facilitating clause start from position 11 nearly impossible. Still, just as at position 9 there appears to be a possibility for a nearly straddled minor-phrase boundary:
[μή τί μευ ἠύτε παιδὸς ἀφαυροῦ πειρήτιζε
Clause-start after position 12 means: the start of a grammatical clause coincides with the start of a hexameter. As after other positions, clause start after position 12 may be anticipated by extra-clausal constituents in the preceding line:
[ἔνδον ἅπας πεπάλακτο
[ἐκ χειρῶν προέηκε
Both examples show bucolic anticipation, one of the “hooks” identified by Clark.  The anticipation is not in terms of the clause, as suggested by Clark 1997: the grammatical clause is yet to start with the next verse. The anticipation is syntactical only from the point of view of the major phrase. Phonologically, the anticipation itself is a minor phrase, with a minor-phrase demarcation at verse end.
3.3.2 The completion of the clause
πρὸς οἶκον Πηλῆος ἐγὼ δ᾿ ἐπὶ μείλια δώσω
πολλὰ μάλ᾿ ὅσσ᾿ οὔ πώ τις ἑῆι ἐπέδωκε θυγατρί
The verb form ἀγέσθω has a complete predicate frame in 146, but the frame is nonetheless extended into 147 by means of the satellite πρὸς οἶκον Πηλῆος. The true completion of the predicate frame is marked by an extra-clausal constituent with the pragmatic function theme: ἐγὼ δέ in 147. The description of πρὸς οἶκον Πηλῆος varies according to the theoretical point of view. If the verse end is acknowledged as a must-be observed metrical boundary, πρὸς οἶκον Πηλῆος is an example of run-over.  Taking a start from the whole-verse utterance, πρὸς οἶκον Πηλῆος stands in adding internal  enjambment: the addition is not a grammatical requirement, and ‘belongs to the same clause as the leading line’.  As an addition, it occupies a metrical colon, and forms a minor phonological phrase. Standing between the verse end and the start of a new syntactical unit, πρὸς οἶκον Πηλῆος may be described as a phrase. In performance, πρὸς οἶκον Πηλῆος may well be a single intonation unit. Both from the syntactical and the prosodic point of view, πρὸς οἶκον Πηλῆος demonstrates the open-ended character of the grammatical clause.
3.4 Phonology and the Homeric Grammatical Clause
3.4.1. Phonology and phonetics
3.4.2 Ancient terms
Not that he had to: Basset suggests that Dionysius might have cited Iliad 1.214  :
Furthermore, Dionysius distinguishes between word end that is masculine (following the foot’s thesis) and feminine (following the foot’s trochee). No specific terminology is being used for word end at the metron dieresis. As heavy syllables are preferred over light ones, so word end in a heavy syllable is better than word end in a light syllable. I argue that from these observations from antiquity another observation, metarrhythmisis, can be deduced.
Often, word end in a heavy syllable on the thesis brings out a different metrical pattern: that of the anapaest. The metrical shift is a prosodic reality on the level of the rhythmical word, that of metarrhythmisis (τὰ μέτρα μεταρρυθμίζεσθαι): the shift from one metrical pattern (dactyl) for phonetic words to another (anapaest).  One of the ancient ῥυθμικοί, Aristides Quintilianus, notes that the thesis often had a little extra duration compared to the arsis. He labels the dactyl ἀνάπαιστος ἀπὸ μείζονος (a maiore, cf. the ionicus), and the anapaest ἀνάπαιστος ἀπὸ ἐλάττονος (a minore).  This description makes it tempting to explain the dactylic-anapaest metarrhythmisis as a form of metrical anaclasis on word level.
In fact, Dionysius always chooses his examples of the different ῥυθμοί according to these principles.
3.4.3 The major phrase and the grammatical clause
Metrical structure (and my translation) suggests two grammatical clauses in asyndeton. The metrical boundary dividing the two grammatical clauses is the trochaic third foot caesura (τάχιστα :5½ πύλας). The metrical boundary dividing the two grammatical clauses is not a strong phonetic boundary: the phonological phrase develops well despite the third foot caesura. But then there is another (minor-phrase) phonetic boundary that is potentially much more disruptive: the dieresis after the first foot (θάπτέ με :2 ὅττι), realised after the postpositive of a metrical-colon phonetic word. As an audible disruption, this boundary softens both the asyndeton and the hiatus: θάπτέ με·2 ὅττι τάχιστα πύλας Ἀίδαο περήσω ‘bury me; that I may soonest pass the gates of Hades’s house’. Van Leeuwen, in his 1895 edition of the Iliad, already suggested this rhetorical punctuation; I think phonetics proves him right. Even in a line where grammatical structure seems to follow metrical colometry, phonological phrasing may point at another grammatical structure.
In the edition of Leaf, there is a full stop after κεῦθε, but a comma after ἐξαύδα. The use of the comma after ἐξαύδα connects the imperative in asyndeton to another imperative. This usage of punctuation is in line with Nicanor’s system of punctuation,  but it is not consistently applied. A full stop after an imperative as the first word of the line is not uncommon in Leaf’s edition (for example Iliad 4.29 (= 16.443, 22.181), Iliad 21.128, Iliad 22.365  ), but this edition consistently avoids two full stops within one line, that is, without counting a full stop at the verse end. As a result, the treatment of similar syntactical phenomena is not always equal: at times, it appears to be rather inconsistent.
Here, both imperative phrases are right-branch demarcated by a heavy element on the thesis. Line Iliad 1.363 consists of three complete grammatical clauses, as does Iliad 18.74. And, equally, Iliad 1.363 cannot be analysed as a phonological “threefolder”:  the heavy syllable on position 7 is subject to gliding,  the phonological liaison bridging the word division, hence weakening the demarcative effect of the word-final heavy syllable on the thesis. Contrary to what is suggested by written punctuation, it is difficult to find an example of repetitive phonological phrasing within a Homeric line. Can it be found in Iliad 22.450?
Here we have a directive, with a second one following in asyndeton. The first directive is not an imperative, but an adverb. Mostly this adverb is used with an imperative or a subjunctive in one clause (Iliad 7.350, Iliad 14.128). Here, however, combining the adverb and the imperative does not make sense, so Leaf and West, I think rightly, separate them into two separate clauses: the adverb is then used as an imperative  without an accompanying verb in the same clause (as in Iliad 13.481, Odyssey 8.307). What applies to a verb acting as an imperative, applies to an adverb acting as an imperative as well: if such an imperative forms a clause on its own and is followed by another clause in asyndeton, there is reason to consider the possibility of phonological phrases corresponding to each imperative.  Again, there is no repetitive pattern. Read δεῦτε·1½ δύω μοι ἕπεσθον·5½ ἴδωμ᾿ ὅτιν᾿ ἔργα τέτυκται and both imperatives are right-branch demarcated by word end that does not justify such harsh rhetorical punctuation. Alternatively, δεῦτε δύω·3 μοι ἕπεσθον ἴδωμ᾿ ὅτιν᾿ ἔργα τέτυκται creates an imperative phrase shaped as a metrical colon, but its phonological shape is not repeated within the line.
3.4.4 The phonology of verse end
ὧς ἔφατ᾿ Εὐρύπυλος βεβλημένος οἱ δὲ παρ᾿ αὐτὸν
The phonetic realisation of the metrical value of the preceding verse-final syllable -όν may have been to some extent metrically indifferent (“anceps”)—thought the deceleration of speech tempo towards verse end would probably have made its duration equal to that of a heavy syllable.  When introduced by a pragmatic constituent in the preceding line (for example αὐτὰρ ἔπειτα), clause start from the start of the hexameter still leaves room for a metrically light element ending the preparatory constituent, and ending the hexameter. The resulting metrical pattern – ⏑ ⏑ – ⏑ would be similar to the pattern preceding clause start after position 3½, and resemble clause start after the trochaic third foot caesura (5½). Still, of course, the clause about to start after position 3½ or 5½ starts as an ascending clause, with a single light segment. Descending clause start following a single light element would be a rhythmical impossibility  within the hexametric major phrase sequence. Starting a descending clause with a marked heavy element within the line often requires, as I showed in chapter 2, phonological sandhi at the right side of a preceding constituent-final segment. The descending clause is always preceded by either double-light or a heavy element, never by a single light element. Major phrases straddling the verse end obey to the same rules.
And as a word-final syllable [→ short vowel + consonant] before a consonant within the hexameter counts as heavy, so a verse-final syllable [→ short vowel + consonant] may count as heavy as well: τεῦχε κύνεσσιν = – ⏑ ⏑ – –. In such cases there is a true heavy syllable to satisfy the expectations of the verse-final metrical position. The strongest visible evidence of ambisyllabism in this occurrence lies in the movable nu at verse end. Moveable nu reflects the rhythmical indeterminability, rather than the avoidance of hiatus. 
Save for the verse-final syllable [naturā-long vowel + ς] at position 12, examples of superheavy word-end syllables within the grammatical clause are equally hard to find (compare the types 14 and 17 in footnote 66):
In fact, the licence that Wernicke’s law represents, the condition for spondaic word end, is similar to the condition for spondaic word end at verse end within a run-over grammatical clause. In other words, the metrical repetition within a grammatical clause is maintained despite the apparent “prosodic neutrality” of the verse-final syllable: it is not “prosodically neutral”, but rhythmically indeterminate.
|Type of enjambment in Higbie’s system:||Phonetic realisation:|
|* Adding internal enjambment||Phonetic-word boundary
(possibly) Minor-phrase boundary
|* Adding external enjambment||Minor-phrase boundary
(possibly) Major-phrase boundary
|* Clausal internal enjambment||Phonetic-word boundary
(possibly) Minor-phrase boundary
|* Clausal external enjambment||Minor-phrase boundary
(possibly) Major-phrase boundary
|* Necessary enjambment||Phonetic-word boundary
(possibly) Minor-phrase boundary
|* Violent enjambment||Phonetic-word boundary|
Higbie’s classification disregards that a) the realisation of phonetic-word boundary is not automatically perceptible as a pause, b) the realisation of minor phrase boundaries is in accordance with grammatical organisation,  and c) within the phonological phrase there is no “affective prosody”,  that is, no emphasis due to localisation. The first constituent of the Homeric hexameter following enjambment does not automatically receive emphasis, not even if it is, grammatically, a mot-en-rejet. Looking at Higbie’s classification, and from her point of view, I still find it advisable to change the label “violent”: it is the only label that does not refer to what the word(s) following the enjambment contribute to the grammatical (in)completeness of the line ending in enjambment at verse end.
3.5 The Phonology of Grammatisation
3.5.1 Grammatical coherence and phonological phrasing
παύσασθαι πολέμοιο δυσηχέος εἰς ὅ κε νεκροὺς
κείομεν ὕστερον αὖτε μαχησόμεθ᾿ εἰς ὅ κε δαίμων
ἄμμε διακρίνηι δώηι δ᾿ ἑτέροισί γε νίκην
Metrically, αἴ κ᾿ ἐθέλητε παύσασθαι πολέμοιο δυσηχέος, εἰς ὅ κε νεκροὺς κείομεν, and ὕστερον αὖτε μαχησόμεθ᾿ εἰς ὅ κε δαίμων ἄμμε διακρίνηι are “skewed”, but syntactically they are not. The syntactically coherent clauses may be demarcated as phonological major phrases: the verse end does not impede such analysis. The phonetic realisation of the verse-final element is not necessarily a phonological clue for audible termination. Some “subordination” of the metrical boundaries and subsequent “yielding” of metrical to phonological phrasing are the result of the mismatch of boundaries and cola with phonetic boundaries. I realise that this seems to be a circular argument, but it is not: it would be if, as Parry and his followers take for granted, the Homeric utterance is equal to the hexametric metrical phrase. Later hexametric Greek poetry shows that the equation of metrical and sense-phrase gradually developed into a poetic norm. Latin hexametric poetry accepted the norm, but realised that there were exceptions: in Latin prosody, however, such exceptions could not be imitated by means of a verse-final metrical foot, as there is no possibility for realisation of the verse-final syllable as rhythmically indeterminate. Furthermore, verse-final syllable [→ short vowel + consonant] in Latin is not metrically heavy.  Straddling the hexameter’s verse end with a phonological phrase in Latin requires resyllabification: in other words, elision at the end of the hypermetric verse. 
3.5.2 Homeric clausal structure as a prose-like element
Here, however, the elision points at phonological reorganisation. A better example would be Odyssey 10.(214–)215:
οὐρῆισιν μακρῆισι περισσαίνοντες ἀνέσταν
Here the extra-clausal constituent is separated from the phonologically coherent grammatical clause by the verse end as phonological-phrase boundary. But examples of transitional constituents (within the clause and extra-clausal) with the verse end as their right branch demarcation (type: αὐτὰρ ἔπειτα ||) are relatively rare.  It is far more common that the single hexameter contains (parts of) both a grammatical clause and one or more (extra-clausal) transitional constituents. As shown above, the (extra-clausal) transitional constituents are themselves sentential-prepositive minor phrases. As such, they are part of the larger scale unit of the major phrase, while retaining their prosodic characterisation of isolated chunks. When elided on their final syllable, the transitional constituents are reorganised into the major phrase.
3.5.3 The Homeric hexameter as prose-like rhythm
It rather looks like a heptameter. What makes it appear as a hexameter is the punctuation following υἱὺς. To Friedländer, the punctuation suggested that ‘παι[δί] seems to fall outside the verse’. Comparison with other metrical inscriptions suggests that his conclusion may have been precipitate.
The inscriptions vary ‘among hexameter, prose, and hybrid forms’  . In all examples cited the verse end may function as a linguistically defendable minor phrase boundary. Such a minor phrase boundary, however, is impossible in IG I2 689:
This inscription contains a formula familiar from prose dedications. Here, the run-over word ἀπαρχήν belongs to the predicate frame of ἀνέθηκε. Is there any reason to assume affective prosody due to verse-end enjambment in this inscription? Compare an inscription on the Chest of Cypselus:
Further consider the formation of grammatical clauses in IG I2 975:
The first line is not a hexameter: Γνάθωνος τόδε σημα is the start of a hexameter, and θέτο δ᾿ αὐτὸν ἀδελφὴ the completion, but in between there is an iamb missing. The metrical phrase between positions 5½ and 7 is not there. In Homer, this specific metrical phrase may be occupied by an extra-clausal constituent  . If a Homeric example like Iliad 10.164 is compared with IG I2 975, the inscription seems to disregard the need for maintaining podic structure  within the metrical period:
The Homeric example, by contrast, shows the way in which maintaining podic structure can be used to single out the phonetic demarcation between the extra-clausal constituent and the subsequent grammatical clause. In Homer, localisation of extra-clausals proves to be an antidote to stichic metrical repetition. In the inscription, such appliance of an antidote is not required. At the same time, the metrical period is disregarded: θέτο δ᾿ αὐτὸν ἀδελφὴ fills the metrical phrase between positions 7 and 12, and ἡλίθιον νοσηλεύσασα cannot be analysed as a dactylic metarrhythmisis  . The combination of prose-elements and dactylic verse-patterns appears to be old and persistent. Compare the “enjambed” dactylic (or anapaestic?) clausula to an otherwise prose inscription:
As in some of the inscriptions, the freedom in clause-formation in Homer may disregard the metrical boundary that is the verse end. The resulting phonological phrasing turns the period-final metrical colon into a minor phrase. The rhythm is still dactylic, but the metrical phrasing of the period, the verse, yields to the phonological phrasing:
πνοιῆι Ἀχιλλῆος πάλιν ἔτραπε
This slightly resembles prose rhythm in that the phrase structure is more important than the metrical structure in the experience of the verse. Of course, such prose rhythms are not predominant: there remain many verses that have a less subordinated verse end, either as a minor or major phrase boundary. With the possibility of prose-like rhythms on the one hand, and the natural chunking of spoken language on the other, it is theoretically unsound to grant emphasis to words based on their metrical localisation. Affective prosody (as emphasis) must stem from the mismatch of grammatical / syntactical phrasing and the phonetic realisation of positions of frequent word end.
3.6 Further Indications for Phonetic Demarcation
3.7 Conclusion: Good Homeric Poetry Takes a Certain Likeness to Prose
|Clause starts after position …||Iliad||Odyssey|
|1 [+⏑ ⏑]||0.00%||0.00%|
|1 [+ –]||0.89%||1.07%|
|3 [+⏑ ⏑]||2.68%||3.23%|
|3 [+ –]||8.93%||11.8%|
|5 [+ ]||1.78%||0.00%|
|5 [+ –]||9.82%||6.45%|
|7 [+ ]||5.36%||2.15%|
|7 [+ –]||1.78%||1.07%|
|9 [+ ]||3.57%||2.15%|
|9 [+ –]||0.00%||0.00%|
|11 [+ –]||0.00%||0.00%|
|Total number of clauses*||112||93|
Table 2: Percentages of clauses starting after various metrical positions
*Predicatively used participles have not been counted as individual grammatical clauses
|Total number of clauses*||112||93|
Table 3: Percentages of clauses ending on various metrical positions
*Predicatively used participles have not been counted as individual grammatical clauses
Table 4: Different realisations of the verse-final heavy element as percentage of the number of “enjambed” grammatical clauses.