Poetic Diction

proceedings of the British Academy, 93.
21-93
Poetic Diction, Poetic Discourse

and the Poetic Register
R. G. G. COLEMAN
Summary. A number of distinctive characteristicscan be iden-

tified in the language used by Latin poets. To start with the
lexicon, most of the words commonly cited as instances of
poetic diction - ensis; fessus, meare, d e , -que. . . -que etc. -
are demonstrably archaic, having been displaced in the prose
register. Archaic too are certain grammatical forms found in
poetry - e.g. auldi, gen. pl. superum, agier, conticuere - and
syntactic constructions like the use of simple cases for pre-
I .
positional phrases and of infinitives instead of the clausal
structures of classical prose. Poets in all languages exploit the
linguistic resources of past as well as present, but this facility
is especially prominent where, as in Latin, the genre traditions
positively encouraged imitatio. Some of the syntactic character-
istics are influenced wholly or partly by Greek, as are other
ingredients of the poetic register. The classical quantitative
metres, derived from Greek, dictated the rhythmic pattern of
the Latin words. Greek loan words and especially proper
names - Chaoniae, Corydon, Pyrrha, Tempe, Theseus,
Z e p h y m etc. -brought exotic tones to the aural texture, often
enhanced by Greek case forms. They also brought an allusive
richness to their contexts. However, the most impressive charac-
teristics after the metre were not dependent on foreign
intrusion: the creation of imagery, often as an essential feature
of a poetic argument, and the tropes of semantic transfer -
metaphor, metonymy, synecdoche - were frequently deployed
through common words. In fact no words were too prosaic to
appear in even the highest poetic contexts, always assuming
their metricality. Native too are the aural figures of alliteration
and assonance, and the exploitation of word-order variation for
0 "he British Academy 1999.
Copyright © British Academy 1999 – all rights reserved

22 R. G. G. Coleman
pragmatic effects. Many of these characteristics can be found

occasionally in literary prose; it is their frequency and accumu-
lation that define the poetic register. Nor is this register merely
a code for translating prose discourse into poetic form. It is the
vehicle for distinctively poetic modes of argument. This is why
we can hope to recover what a poem meant to the author who
conceived it, including the occasional ambiguities and ambi-
valences that are in the text and not merely imposed by our
own ignorance, incompetence or cultural preoccupations, only
if we have some notion of what the poetic register was in detail
and what its relationship was to contemporary prose usage. To
attempt to articulate such a notion is the purpose of this essay.
INTRODUCTION
1.1. ON A NARROW DEFINITIONpoetic diction can be viewed simply as the

words that are used exclusively or primarily in poetry or used in meanings
that are not normal to them in prose discourse. A broader definition would
include also phonetic characteristics, the patterns of sound that are created
in the choice and organization of the vocabulary, most conspicuously those
dictated by metrical conventions, and grammatical distinctions exhibited
in the choice of morphological forms and syntactic structures.’ In general
the distinguishing characteristics are not the presence or absence of a
particular linguistic phenomenon but its relative frequency in comparison
with prose.
1.2. The dividing line between the registers is often hard to draw. Some
poetry can be classified as such only from its metrical form. Aristotle’s
judgement on Empedocles may have been too severe - 0 6 6 2 ~62 K O L V ~ V
E)UTLV O p < p q K a i E p e S o ~ h eTA+
i T A p k ~ p o v 616
, T ~ p2v
V T O L ~ T ~Sirtaiov
~ V Kaheiv,
T ~ V 62 $ ~ a ~ o h d y op&lhhov
v 9 T O L ~ T + J (PO. 1447b) - but there are certainly
passages in the Sicilian philosopher, as in his illustrious Latin admirer
Lucretius, that can only be described as versified prose? However, the
important qualification that Aristotle apparently made to his judgement
on Empedocles, - Sr~vdsr e p ; T ~ V$poEu~vyiyovev, pera$opqnKds r e dv K a i
70:s ~ X A O L S
70:s m p i T O L ~ T L K ~&uciypauL
~V xpLjpevos (up. D.L. 8.57) - applies
far more aptly to Lucretius. A didactic poem that is concerned primarily
with conveying information and expounding basic scientific or philo-
The importance of all areas of linguistic usage in defining the poetic register is well stressed
by Janssen (1941: 14,35).
4
* clear distinction is drawn in Sanskrit between scientific treatises in verse and true poetry
kdvyam.

POETIC DICTION AND THE POETIC REGISTER 23
sophica] theory simply cannot operate, like Virgil’s Georgics, at a

consistently high poetic level. The effective style must be objective and
univocal, not metaphoric, allusive and subjective. Coleridge was right to
insist3 that the fundamental distinction is between poetry and science, not
poetry and prose.
13.To take an example from English, the couplet True Wit is Nature
to advantage dress’d, / What oft was thought, but ne’er so well express’d
(pope, Essay on Criticism 297-8) may seem little more than a versified
alternative to Tme Wit is Nature dressed to advantage, What was often
thought, but never expressed so well, as Johnson’s paraphrase Wit is that
whichhas been often thought, but was never before so well expressed4clearly
reveals. The sentiment is not especially poetic and the only properties of
the poetic register are the use of oft and ne’er for often and never, the
reminder in the spelling of dress’d and express’d that the archaic variant
of syllabic -ed was still available to poets as in learned beside unlearn’d
(ibid. 327), and the placing of the two participles in phrase-hal position,
a poetic order in itself but here used to secure the rhyme that is essential
to the verse form. There are many couplets like this in Pope but few
readers would subscribe to Matthew Amold’s description5of Pope and
Dryden as ‘classics of our prose’.
1.4. On the other side we can point to many passages of high literary
prose that have more in common with the poetic register. The links are
explicit in the view expressed by Antonius in Cic. De or. 2.34, qui enim
cantus moderata oratione dulcior inueniri potest? quod carmen artificiosa
uerborum conclusione aptius? and Quintilian’s famous description of his-
toriography as proxima poetis et quodam modo carmen solutum (10.1.31),
following the line taken by ‘Antonius’ again in the discussion of historia
at De or. 2.51-64. The comparison is of course stylistic, but it applies also
to the subject matter of historiography, which has replaced epic as the
medium in which the myths and ideologies of a society are projected on
to that society’s past by some of its major creative writers6
1.5. It is interesting that Aristotle (up. D.L. 8.57 again) described Empe-
docles as the inventor of { V T O ~ L K ~in~ , view of the modern tendency to
The most comprehensive exposition is in Biographia Literaria, especially chapter 14.

Lives of the English Poets: Life of Cowley (World’s Classics edn., Oxford 1952) 1.13.
The Study of Poetry. Essays in Criticism, Second Series (London, 1905) 42. The remark
must be seen in the context of Amold’s general argument that poetry was replacing religion
and philosophy as the complement to science, continuing the viewpoint held by Words-
worth and Coleridge.
For poetry as the intellectual nucleus of prehistoric society see Meillet (1965: 121-3) and
Watkins (1982 1989). This nucleic status is recoverable in a number of poetic words
and phrases in Latin, Greek, Sanskrit, Old Enghsh etc. that can be traced back to Proto-
Indo-European. See the detailed discussions in Schmitt (1967) and Watkins (1982,1989).

24 R. G. G. Coleman
polarize ~ V T O P L K ?and
~ T O L T ~ L K Tand
~ both with 4vatohoyia (cf. the Aristotle
citation in 31.2). In fact many important discussions of the ingredients of
the poetic register are found in works on rhetoric, modern as well as
ancient. Nor should we be surprised, since poetry and prose alike share
the three classic aims of linguistic communication, to be pursued severally
or in combination - ut doceat, ut moueat, ut delectet.
1.6. Even outside the literary registers the boundaries between prose
and poetry are often hard to draw. In many cultures the rhetoric of religious
prayer, secular proclamations etc. have a strongly poetic character. In Latin
the Carmen Fratrum Arualium (CZL 122)is clearly composed in verse triads
of some sort, but the corrupt text prevents useful analysis. No less clearly
the prayer at the lustratio agri cited by Cat0 (Agr. 141) is in prose: Mars
pater, te precor quaesoque uti sies uolens propitius mihi domo familiaeque
nostrae . . . uti tu morbos uisos inuisosque uiduertatem uastitudinemque cala-
mitates intemperiasque prohibessis defendas auerruncesque, utique tu fruges
+menta uineta uirgultaque grandire beneque euenire sins . . . This rhythmic
and formulaic prose, enriched with alliteration, assonance, rhyme and ety-
mological word-play (cf. 4416-18), can be paralleled in the ritual passages
of the Umbrian Iguvine Tables, and so belongs to the Italic liturgical tra-
dition. The rhythmic prose of the Christian Te Deum, composed in the fifth
century AD belongs to the same tradition?
1.7. The same very rhythmic character is found in political rituals,
recalling the close connection between the religious and the secular in
earlier societies. For instance the Formula Patris Patrati for declaration of ,
war (Livy 1.32.7ff): audi Zuppiter et tu Zane Quirine deique omnes caelestes
uosque terrestres uosque inferni audite. ego uos testor populum illum iniu-
stum esse neque ius persoluere. . . si ego iniuste impieque illos homines
illasque res dedier mihi exposco, tum patriae compotem me numquam siris
esse.8 It is not surprising to find such productions classed as carmina in
classical Latin.
1.8. It is worth reminding ourselves at this point that in the linguistic
characterization of the elevated registers that we have been looking at
both stand well apart from the sermo cottidianus of the mass of the
population. This is not so much because the characteristic features of
ordinary conversation - false starts, nonce mispronunciations, abrupt and
’Traditionally attributed to St Ambrose and St Augustine, it is probably the work of Bishop

Niceta of Remesiana or a contemporary. For the rhetorical shaping of Christian collects see
Coleman (1987 45-7).
See Gordon Williams (Kenney-Clausen (1982 53-5)). The use of curmen of the lbelve
Tables, however, more likely refers to its mode of recitation by schoolboys: the fragments
show plenty of formulaic composition but nothing rhythmic enough to parallel the versified
laws of certain early Germanic societies.

ungrammatical transitions, anacolutha, rambling pleonasm and banal re-

petition, not to mention mere noise, the Latin equivalent of ‘er.. . um . . .
ah. . .’ - have not survived. Even in a living language such phenomena,
ruthlessly exposed in oral recording, are usually edited out of transcripts.
It is rather that even the most realistic passages of Roman comedy, say,
are alfeady creative adaptations of what such transcripts would have
looked Iike. In fact we do not possess a single example of real everyday
conversation from any language before the invention of the phonograph.
Even the most vulgar Pompeian graffiti were after all written by literates
and subject to conventional literacy pressures.
1.9. The gap is considerable. Every sample of a language at whatever
level of performance, whether it is intended for ephemeral communication
at the oral level or for permanent accessibility in written form, belongs
ultimately to the same linguistic system and must be taken account of in
any description of the language claiming to be comprehensive. There is a
fashion among some linguists for giving more attention to the ephemeral
or casual discourse as being more representative of the way most native
speakers use their language. Telephone conversations then become more
important than contemporary works of literature. This view is mistaken if
we are concerned to explore the full communicative resources available
in the phonology, grammar and lexicon of the language. We can properly
lament the absence of transcripts from the conversation of Latin farmers
or artisans, but the surviving text of the Aeneid yields far more information
about the character and potentialities of the Latin lang~age.~ The higher
literary registers may well be deviations from the norms of ordinary written
Latin and of the spoken Latin that we can only infer from it, but they are
ultimately much more revealing to philologists and linguists.
2.1.’ Any attempt to define the poetic register and the nature of poetic
diction runs into a number of practical problems. Apart from the differ-
ences between poetic and prose discourse, which is the chief concern of
this essay, there are for instance the variations between the poetic genres
themselves that are stressed by ancient theorists. Sua cuique proposito lex,
SUUS decor est, nec comoedia in cothurnos adsurgit nec contra tragoedia
socco ingreditur, writes Quintilian (10.2.22), summarizing the doctrine of
stylistic decorum, 76 ~ p i w o v ,already alluded to in Plato’s Laws 7OOa-b.
The classic statement for later ages was Horace’s ( A 2 73-98). The conven-
Thoughout this essay examples are taken mostly from the central classical authors, Lucre-
tius and Catullus, Propertius and Ovid and above all Virgil and Horace, who as a group
represent the Latin poetic register at its richest.

R. G. G. Coleman
tions were not of course totally rigid for all times, but as with, say, the
sonnet form or the nineteenth-century symphony, innovations were gener-
ally made in the context of adherence to what had gone before.
2.2. Poetic discourse is of course an occasional mode of communication.
It is not the way poets talk or write all the time but the product of a
consciously creative process that is activated only on particular occasions.
Yet there must always be a considerable input from their own normal and
largely unconscious linguistic habits. If we have no other representative of
their idiolectal practice, there is a very real danger of confusing these
linguistic habits with the linguistic peculiarities of the poetic register in
which they are composing.
2.3. Sometimes we have the chance to compare a poet’s Latinity in
different literary registers, e.g. Seneca’s tragedies, philosophical treatises
and letters. Again there may be an opportunity to compare contemporaries
or near contemporaries writing in the same genre, as for instance the
Augustan elegists, where we are also able to make comparisons between
Amores and Metamorphoses. Where material is more sparse the problems
become more acute. We have perhaps good reason to infer that Lucretius
was an archaist, Catullus an innovator, but what is the norm against which
these classifications are made?
2.4. Even more vexing is the situation in the early decades of the
second century BC, a formative period for the Latin literary register. We
have the comedies of Plautus and the fragments of Ennius’ work in divers
genres, many of them cited by later authors precisely because they exhibit
linguistic eccentricity or provide a precedent for some anomalous usage
in a classical author. (It is striking how much less strange most of the
citations made by literary commentators, antiquarians or orators turn out
to be.) But there is nothing much beside, not even in prose, except for a
few inscriptions, including the Senatus consultum de Bacchanalibus, which
certainly deserves a place in the history of literary Latin as the most
elaborate piece of oratio obliqua before Caesar. Non-epigraphic prose is
represented by Cato’s De agri cultura, which hardly counts as literature,
and the fragments of history and oratory by various authors (as well as
the material cited in §§1.6,7) which certainly do. But the latter have been
too corrupted in the manuscript tradition and by ancient editorial revision
to provide as finn a basis for comparison as is needed. So there is no real
touchstone by which to distinguish the early Latin poetic register from
early Latin as a whole. What sort of a picture would we have of Augustan
Latin, say, if all that survived were the elegies of Propertius, some frag-
ments of Horace and a handful of inscriptions?
2.5. The diachronic problems are of course even more imposing. For
the history of English or French we have a sufficiently large and varied

body Of material to be able to plot changes in the phonology, grammar and

lexicon with some degree of confidence. But for Latin the first surviving
attestation of a phenomenon must generally be interpreted as an inno-
vation only with the utmost caution. Conversely what is found at an early
date but is then not attested for several centuries can be classed at its
reappearance as an archaism only on the assumption that its intervening
absence from the records is accidental. Finally what is distinctly poetic at
one period may not have been so at an earlier or later stage in the language.
Much of what follows must be read with these important qualifications in
mind.
2.6. One area of linguistic behaviour that does not impinge upon the
literary registers is that of dialect. There is no trace of a Latin Robert
Burns, let alone anything like the ancient Greek association of regional
dialects, artificial and selective to be sure, with specific literary genres.
Sappho's Aeolic, the Doric of Alcman and Theocritus, the Ionic of Archi-
lochus etc. not only characterized their own poetry but the subsequent
tradition of the respective genres as well. Popular regional poetry there
must have been in Latin, but no record of it survives, and ambitious poets
from the regions came to Rome for their education and wrote in the senno
urbanus of the capital.'O Whatever Pollio meant by patuuinitas (Quint.
1.5.56), the peculiarities of Livy's magnificent prose owe nothing to Padua.
Dialect forms like Lucilius' Cecilius pretor (1130) are satiric and merely
underline the disdainful attitude of urbanitus to rusticitas. Most of the
major Latin authors from Plautus and Ennius onwards were Italians or
provincials, not Romans, but in literature as in so much else in the Roman
culture centralization came early and remained, being receptive to external
iduence only from Greek.
SOUNDS AND PATTERNS OF SOUND
3.1. The first and most clear-cut definition of the poetic register is
phonetic." Poetry was distinguished from prose by the recurrence of
regular rhythmic patterns that marked the boundaries of verses and
groups of verses, and influenced the choice of word order within the
verse.
l0Cicero's reference to Cordubae natis poetis, pingue quiddam sonantibus atque peregrinum
(Arch. 26) seems to be a comment on regional pronunciation rather than on departure from
urbanitas in the Iberian Latin poets' habits of composition.
The relevance of the sound of Latin to all aspects of poetry and literary prose is abundantly
demonstrated in Wilkinson (1963).

28 R. G. G. Coleman
3.2. Oratorical prose was also highly rhythmical. The classical clausulae,
marking the cadences of clauses and sentences, have been well researched,
but one can often detect rhythmic patterns extending over large sections
of the sentence, which can hardly be coincidental. Cicero cites a sentence
from Crassus in which the numerus appears to be non quaesitus: ‘nam ubi
lubido dominatur, innocentiae leue praesidium est’ (Or. 219). The sequence
of third paeon, dactyl and trochee (an echo here of the hexameter cadence
mischievously hinting at the licentia of poets?), two trochees, dactyl and
choriamb is deploying aural effects redolent more of lyric poetry than of
any prose idiom. Or consider the rhythmic structure of one of Cicero’s
youthful tricola: nam commoditati ingenium, grauitati aetas, libertati
tempora sunt impediment0 (ROSC.Am. 9), with its sequence of spondee,
anapaest and choriamb, anapaest and spondee, two spondees, choriamb,
cretic and spondee. Phrases like uersus . . . propemodum and numeros
quosdam were not lightly used to describe the best prose style (De or.
3.173).
3.3. The effect of such rhythmic sequences delivered in the mannered
style that characterized public oratory must have been comparable to
that of a poetic recitation (cf. Cic. Or. 55-60, Quint. 10.1.16-17). Never-
theless the fundamental division remains: verse rhythms, as Cicero
explicitly recognized (Or. 195), fell into finite, regularly repetitive
patterns.
Al. From Plautus and Ennius onwards the literary metres were all Greek.
Although syllabic weight, based on vowel length and consonontal content,
was a phonological property of Latin as of every human language, its use
as the basis of metre was adopted along with the verse forms themselves
from the Greek tradition. The binary classification into light and heavy
syllables was an oversimplification and a matter of convention, as indeed
was the placing of syllable divisions; cf. pa-tris with a light initial, pat-ris
with a heavy (see $6.2) but only lap-sus.
4.2. The only difference between the two languages that had a bearing
on versification was that of accent. The Greek tonal accent was of no
metrical significance, though of course it provided a melodic line, both in
poetry and prose, that was foreign to Latin. Against this the Latin stress
accent did have metrical implications, since it had some effect upon vowel
length (see §§10-11) and, more importantly, set up a dynamic pattern that
corresponded only partially and accidentally to the pattern of quantitative
rhythm. Thus we can represent a line like Tityre tu patulae recubans sub
tegmine fagi as

To this typical interaction between stress and quantity we must return

(94.6). For the present we need only note this as a new element in the
adopted Greek metres, just as the quantitative basis probably was in Latin
versification.
I4 .3. The native verse forms seem to have been based, like the Germanic
ones, on a fixed number of syllables and word-stresses in each verse. Thus
the Saturnian verse would have been divided by a caesura into two parts,
a seven-syllable group with three stresses and a six-syllable group with
two stresses, or something of the sort;13e.g. Livius Andronicus’ uirum mihi
Carntnu insece uersutum and tbpper citi ad utdis utnimus Circai, where it
is assumed that the initial syllable stress had already been replaced by ‘the
rule of the penultimate’, and Naevius’ nbctu Troiad exibunt cupitibus
o p t r h , alongside the earliest surviving Scipio epitaph (CZL 129),hbnc oino
pldrume cosentiont R[omane]etc.
I 4.4. It was presumably in Saturnian verses that the clurorum uirorum
luudes recited at banquets before Cato’s lifetime (Cic. Brut. 75, Varro up.
Non. S.V. ussu uoce) were handed down, to find their way via the annalists
into the pages of Livy.14 How long Saturnians survived in epigraphic elogiu
we do not know, but the rejection of the old metre for the elegiac couplet
on the epitaph for Scipio Hispanus (CZL I%, after 139 BC) may be
significant. The replacement of Saturnians by the dactylic hexameter for
narrative poetry in the Annales of Ennius was certainly decisive.
I 4.5. In fact the abandonment of a stress-based system of versification
for a quantitative one was facilitated by accidental similarities in the stress
patterns of the two major verse forms concerned. For many of Ennius’
h a m e t e r s are characterized by a similar stress pattern to that assumed
l2 The choice of musical notation does not imply that Latin poetry was normally sung or
chanted, though there is evidence that some pieces were sung - e.g. the distinction between
the metrically freer cantica and the more strict diuerbia in Plautine comedy - and some
maybe even mimed (Don. Vim Virg. 26). The musical notation has been used here because
the interaction of stress patterns and temporal units is for modern European readers more
familiar in music than in poetry.
l3 Basically this analysis derives from Lmdsay (1893). Fixed syllable verses occur i n Gatha-
Avestan, combined with set quantitative patterns in Vedic and Aeolic Greek, with set stress
patterns in Germanic. For further discussion of the Saturnian metre with different conclusions
see Pasquali-Timpanaro (1981).
Niebuhr’s hypothesis has never quite recovered from Schwegler’s criticisms. See Momi-
&no (1957) for a modem critical but sympathetic discussion. The certified parallels in other
Indo-European cultures and the absence of any other obvious vehicle for the stones of
Romulus, Horatius etc. to have survived into the histonographical tradition have ensured
that the hypothesis retains its sympathizers.

30 R. G. G. Coleman
for the Sat~rnian;‘~ e.g. Mzisae quae ptdibus mrignum puksatis Oltimpum
and transnliuit cita per ttneras caliginis auras and tcilia tum mtmorat lricru-
mans, exttrrita sbmno (Ann. 1, 18, 35 Sk). The increased number of
syllables in the dactylic hexameter (the three examples cited have respect-
ively 14, 15 and 16) made a limit of five stresses untenable and the
frequency of a penthemimeral main caesura altered the distribution of
stresses between the two ‘halves’ of the verse. Nevertheless the stress
patterns that were a distinctive feature of the native poesy remained the
distinguishing feature of Latin hexameters against their Greek models.
4.6. The interaction between word stress and syllabic quantity in the
Latin hexameter yields three possibilities, for which there would be no
precedent in Greek. First a homodyne pattern, in which the stress coincides
with the fixed heavy syllable of the foot, as in Mzisae and pukslitis
Olrimpum. Homodyne rhythms are frequent in the first and fourth feet of
classical hexameters, almost invariable in the fifth and sixth.l6 Indeed in
some of the more vulgar epigraphic verses that survive it is only the
concluding rhythm to each line L -
- that confirms the composer’s
metrical intentions. Secondly a heterodyne pattern, in which the stress
falls on the second or third syllable of the foot, as in quae ptdibus mcignurn.
Heterodynes were most frequent in the second and third feet, as here.
Thirdly a neutral pattern, which has either no stress at all within the foot,
as in magnum puksatis, or two stresses, as in arma uinimque ccino.
47. Homodynes obviously underlined the time signature of the metre,
heterodynes obscured it. Virgil, perhaps uniquely, seems to have employed
varying patterns of distribution of the three possibilities for expressive
effect. Thus in impius hatc tam crilta noucilia miles habtbit? ( E . 1.70) the
six homodynes, four of them dactyls, seem expressive of agitated despair.
By contrast in the well known onomatopoeia of quridrupedcinte pzitrem
sbnitu qucitit ungula ccimpum ( A . 8.596) the prevalent dactylic rhythm,
strongly marked by the first and fifth-foot homodynes, is blurred by the
neutral second foot and the heterodyne third and fourth, and the aural
effect is - perhaps appropriately - more confused. All this presupposes
of course retention of the normal Latin accent in the recitation of verse.l7
For a partial anticipation of this view see Bartalucci (1968), who however sees the simi-
larities as due to the influence of one metre upon the other.
l6 For the hypothesis that in Indo-European poetics the latter part of the verse was more
regular metrically than the earlier part see Lotz’s discussion of metric typology in Sebeok
(1960: 135-48, esp. 136).
For an argument in favour of a metrical ictus of some sort distinct from word accent, based
on the possibility of resolving the second but not the first heavy syllable of each hexameter
foot except the last, see Allen (1978: 924). For a general discussion of Latin hexameter
rhythm see Allen (1973: 335-59).

5.1. very few Latin lexemes are unmetrical in all their grammatical forms,
if one does not confine eligibility to dactylic metres. Pacuvius’ line Nerei
,.epandirostrum incuruiceruicum pecus is a regular trochaic tetrameter,
though the two adjectival compounds censured by Quintilian (1.5.67) are
ineligible for dactylic verse. Because dactylic metres, employed in hexa-
meter and elegiac poems, were so prominent in classical literature, a large
number of words that could in no other sense be considered unpoetic
were excluded by their syllabic composition. Thus cognitio and uoluptas
are acceptable in all their cases, notio and suauitas in none.
5.2. Frequently it was only particular grammatical forms that were
excluded. While audimus and audit were acceptable, audiunt was not; nor
was audio without clumsy elision into a following light syllable. Tribrachs
were more easily elided, e.g. agere, but were intractable in prefinal position,
e.g. celeritas.
53. Various expedients were adopted to overcome this problem. One
was to employ a synonym.18Where the meaning is very close, nothing is
lost in the process. The well established functional convergence between
the originally distinct action nouns in -tio and -tus enabled Lucretius to
use for instance iniectus in place of iniectio animi to translate Epicurus’
h p ~ hT+Sj Gtavolas without creating any semantic problems. In contrast to
the classical and post-classical periods -tu- had been the more prominent
in early Latin. So the metrically acceptable variant also contributed to the
archaic tone that the poet consistently sought in promoting this very
unRoman philosophy and projecting a high seriousness that distanced his
work from the nugae of the contemporary poetae noui.
5.4. Again nothing is lost when a cretic rhythm is avoided by a tmesis,
which recalls the precompounding stage of the language, e.g. inque pediri
(Lucr, 3.484) or inque salutatam linquo (Virg. A. 9.288), where the archaic
effect is enhanced by the presence of the uncompounded finite verb, and
inque Zigatus / cedebat (A. 10.794-5), where the figure is combined with
the old meaning of cedere. On the other hand the replacement of imperator
by the archaic induperator imports an archaic tone that the poet may not
want. Characteristically Juvenal exploits the satiric possibilities against
Domitian (unnamed of course) in Sat. 4.28-9, qualis tunc epulas ipsum
gluttisse putamus / induperatorem, combining the mock-heroic grandilo-
quence of the archaic form with the distinctly subliterary verb gluttire.
5.5. Substitution may however blur a semantic distinction, as when the
intractable drbdra ‘trees’ is replaced by the plesionym arbusta, plural of
the collective arbustum ‘a cluster of trees’. The peculiarly Roman conno-
tations of imperator are lost when it is replaced by the less specific ductor
’”See Leumann (1959 147-8).

32 R. G. G. Coleman
or d w , and there is again a loss of precision. We may contrast the nice

use of the more generic word in Lucretius’ ductores Danaum (1.86) to
recall Aavacjv 7jy+opcs (Iliad 11.816).
5.6. Even clearer examples are to be found in Lucretius’ use of maxi-
mitas (2.498) to replace magnitudo, where a comparative implication ‘the
state of being largest’ is introduced irrelevantly. The intensive force of
the -tare suffix (cf. dicto, dictito with dico) is weakened by its use to
maintain metricality, e.g. nominitamus (Lucr. 3.352; cf. CIL 121221).The
problem was of course particularly troublesome for a poet confronted
with the highly nominalking terminology of philosophical discourse and
compelled into neologisms like differitas for differentia and uariantia for
uarietas (and uariatio), neither of which however poses the same semantic I
queries. (For unmetrical inflections see 9913.4, 14.1-3).
5.7. The use of tropes like metaphor and metonymy for metrical pur-
poses can gain advantage from the semantic imprecision (see 9929, 32).
Thus the metonymy of thalamus, thalami for nuptiae (Virg. A. 4.550, Ov.
M. 6.700) offers the poet both an evocative image and at the same time
the exotic colour of a Greek loan-word.
6.1. Phonetic expedients are sometimes adopted to cope with the

unmetrical. The most radical, employed almost exclusively with proper
names, is to alter a vowel length. Thus ItZlfam has its initial vowel length-
ened in Virg. A. 1.2 etc., following Callimachean precedent (H. 3.58), a5
Catullus had already done with ftulorum (1.5). It is worth noting that
Virgil (A. 3.185) chooses the long vowel in ftalu regna to secure the plural
phrase where the normal italum regnum could have been accommodated
in a hexameter. The similar device in Greek, e.g. Z t ~ c X l t i a v(Theocr. Id.
7.40) beside L?K&~o& (Hom. Od. 20.383) would have been a reassuring
precedent for Latin poets. So we h d Sicelides and Siculis in Virg. E. 4.1,
2.21. The impossibility of accommodating the cretic Scipio or any of its
cases is overcome by the use of the Greek patronymic suffix, noster Scipi-
adas19 (Lucil. 1139), geminos . . . Scipiadas (Virg. A. 6.843), which of course
has the additional effect of associating the family with the Homeric heroes,
Pollio can appear in dactylic verse only in the nominative, and then only
by a harsh elision (for unattested substitution of the glide i see 99.2), e.g.
l9 -as is strange. -666s (or -665 in some early W. Greek inscriptions) would be the regular
Greek form outside Attic-Ionic, and if the suffix became known first from Italian Doric
dialects, it would have been early enough to have undergone assimilation to Latin -ada‘. We
should expect therefore either Scipiades (metrical) or Scipiada‘ (unmetrical). On Greek
inflection see 6115,27.

pollio et ipse, Pollio et incipient (Virg. E. 3.86, 4.12), or by shortening the

final vowel, Polli6 (Hor. S. 1.10.42;cf C. 2.1.14).
6.2. There is no evidence for any specifically poetic pronunciations
apart from those just noted, though we may infer from the accounts of
oratorical delivery supplied by Cicero (Or. 57) and Quintilian (1.8.2; cf.
33.3) that poetic verses were also delivered with distinctly artful and
unprosaic intonations. Synizesis and vowel contraction, as in d&is for d&
zis, and ddicere for adiicere (cf. Manil. 4.44 with Gellius 4.17.1-8) are likely
to have ’begun in colloquial usage. Some instances however, like Zdem
(Lucr. 4.’744), a u r z (Virg. A. 1.698), are likely to have been inspired as
much by Greek models. In fact the great majority of phonetic devices for
overcoming unmetricality can be directly linked to known facts of ordinary
Latin speech. Even the variation in the treatment of ‘mute plus liquid’
clusters between, for instance, the normal ublu-crem and the occasional
uolzic-res (Virg. A. 11.858,4.525; see Quintilian 1.5.28) may be a reflection
of cohtemporary speech variants rather than a contrast between current
and archaic metrical conventions. Elision, for instance, was certainly a
feature of the spoken language, though in educated usage the elided
syllabld was rarely totally omitted.2O
7.1. Examples occur of pronunciations that were obsolete in current speech

being preserved in poetry. This is merely a special case of the general
distinbshing characteristic of the poetic register, that all earlier poetry
is, or can be made by poet or reader, contemporary poetry. The language
of earlier poets is thus a reservoir on which each new generation of poets
can draw.
7.2 The early Latin treatment of final -s following a short vowel is
instructive.21Before c. 240 BC it was regularly omitted in writing before a
following initial consonant, e.g. Cornelio L. f on the earliest surviving
Scipio epitaph, CZL 1%. All the available evidence indicates that the
consonant was retained where a vowel followed, - a situation complemen-
tary to that of -m, which was lost before a vowel and retained, at least as
a nasal adjunct to the vowel, before a consonant. Examples certainly occur
in the comic poets at verse ends, a well-known location for archaisms, e.g.
occidisti(s) me (Pl. Ba. 313), tempu(s) fert (Ter. Ad. 839).= However, it is
likely that the phenomenon also occurred elsewhere in the verse, but
2o The whole subject of elision in Latin poetry is fully treated by Soubiran (1966).
*’ For further examples and discussion see Leumann (1977: 227-8).
’’See Lindsay (1922: 126-135).

R. G. G. Colemun
because of the range of permissible metrical variants certification is

impossible.
7.3. Final s is consistently written in inscriptions from the end of the
third century BC onwards, assisted no doubt in literary texts and official
epigraphy by Greek models It continues to be written throughout the
classical and post-classical periods. Its restoration in pronunciation, though
guaranteed for educated speech by classical verse, may not have been
universal in the dialects. Cicero (Or. 161) describes its omission as ium
subrusticum but olim politius, citing Ennius’ omnibu(s) princeps.
7.4. This last example is typical of Ennius’ Annules, e.g. suuuis homo
fucundu(s) suo contentu(s) beutus (280 Sk), though there are a few excep-
tions, e.g. uoluuit uuis. simul(87 Sk), Luurentis terra (30 Sk). The omissions
in lines like Lucilius’ uitu illu dignu(s) locoque (cited in Cic. Or. 161),
lateruli(s) dolor, certissimu(s) nuntiu(s) mortis (cited by Marius Victorinus
6.217 K) are unlikely at this date to be reflections of colloquial urban
usage and are better taken as echoes of Ennian epic.
7.5. In poetry of the late republican period omission is very rare.
Catullus’ unique dabi(s) supplicium, a special case with its successive
sibilants, occurs at the end of some invective against Gellius (116.8) for
rejecting the poet’s neoteric interests. The piquancy of the usage here
comes from the disdain of the poetue noui for such archaisms (cf. Or. 161
again).23As befitting the epic style of didactic poetry, -s is occasionally
omitted by Lucretius, e.g. ex omnibu(s) rebus, infuntibu(s) puruis (1.159,
186)’ and by Cicero, e.g. lustrutu(s) nitore, Aquiloni(s) locutue (Arut. 92,
97). However, neoteric prejudice prevailed and even Virgil’s passion for
archaism and for specific echoes of Ennius did not induce him to employ
what was now felt to be altogether too subrusticum for poetic usage.
8. Another group of archaisms involves the revival of long vowels in final

syllables. Like the omission of -s this had the effect not of accommodating
unmetrical words but of enabling metrically acceptable words to stand in
places where it would not otherwise have been possible to place them. So
Ennius could write (Ann. 108 Sk) o pater, o genitor, o sunguen dis
oriundum, taking advantage of the original long vowel, probably a recent
archaism at this date, to place the word immediately before a vowel. The
traditional label ‘lengthening under ictus’ misclassifies the phenomenon,
which is rooted in the history of the language, with the poeticu licentiu of
ftulium etc. (96.1), which is not. There is moreover no evidence that ictus
was a phonetic reality in classical antiquity, whatever its role may have
The frequency of qui sit in Cicero and other prose writers may represent qui(s) sit or qursir.

been in kiter conventions of reading quantitative verse (see 04.7 fin.) Some
of the instances are in any event not 'under ictus'. Thus Ennius has essLit
induperator and ponebat ante salutem (Ann. 78,364 Sk). The lengthening
is frequent in Plautus, e.g. non uxbr eram and alienum ardt, incultum
(Asin. 927, 874). The archaism survives into the classical poets, including
Lucretius and Virgil, who has many instances, e.g. aberdt. ipsae (E. 1.38)
uidet, bminesne and paubr et plurima and umittebat oculusque (A. 1.308,
2.369,5,853). Outside the epic tradition the variant may be used to evoke
Antiquity, as in Propertius' description of early Rome (4.1.17) nulli cura
@it extemos quuerere diuos. Here fuit < fueit (cf. fuit < fuet), but some
instances of lengthening are merely analogical, as manusque sinit. hinc and
fruter,Terit?o quae (A. 10.433, 12.883), unless we envisage the vowel being
checked by final -t before a pause.
9.1. Sometimes advantage is taken of free variants available in the ordinary

spoken language, in particular the ambivalence of i and U, which may be
consonantal or vocalic. So Virgil's abiete for 2fbiZtZ and Lauintaque for
Laulnldque' ( A . 2.16, 1.2) to accommodate otherwise unmetrical words,
Ennius' autum and insidbntes (Ann. 89,425 Sk) and Horace's principtum
for princi'pticm (C. 3.6.6 Alcaics), to enable particular placings in the verse.
Emius has qzuzttgor for quattibr (Ann. 88 Sk), Catullus has the archaisms
soliiit and p e w l i i e n t (2.13, 95.6) for the usual solyit, peruolpnt, which
are of course not in themselves unmetrical. Lucretius uses tenyia for
unmetrical tZniiih. (4.66) and conversely siiadent (4.1157), Vir@ gengu
for gZnU (A. 5.432), Horace siliiae for silyae (Epod. 13.2). Palatalized
forms in Romance like grace, prezzo (<gratia, pretium) imply a shift from
ti to tt to ti, which is attested in isolated vulgar forms from the imperial
period, and was already acceptable in educated speech by the late fifth
century, as we can see from manuscript and epigraphic confusions of
e.g. condicio and conditio, set beside statements by grammarians such as
Papirianus (up. Cassiod. 7.216 K).
9.2. It is surprising that, given the precedent of Ennius' insid-htes and
V&gd's abiete, the consonantal variant was not used to overcome the
unmetricality of oratio, auditio etc. (cf. 05.1, 3). It may be that ora40 etc.
in classical times had still a hint of vulgarism, and the few instances of
comparable phenomena perhaps confirm this, e.g. from Pompeii o4osis
locus hic non est. discede, morator (CZL IV.813) and the spellings ActZuni,
ClaudZanum beside digredlens, ~ucZO."~Whatever the reason, occasional
24 See Vaanlnen (1966: 34-6).

36 R. G. G. Coleman
substitutions of - forv v - (abiete) and of - - for - v v (principium) were

acceptable, but - - for - v - (Pollio, otiosi) was not.
10.1. One particular example of the introduction of current habits of

pronunciation has attracted much scholarly attention: the so-called iambic
shortening by which unstressed final vowels are regularly shortened in
bene, duo, ego, modo, nisi, scio, uolo etc., often also in fero, mihi, uide,
ubi. In fact it is verse texts that enable us to plot the distribution and
history of this and related phenomena in Latin, though of course the
earliest attestations provide only a terminus ante quem for the respective
changes in ordinary speech (c€ 02.5). Instances abound in the classical
poets. In early Latin verse the phenomenon is even more widely attested,
e.g. domi (Pl. Mil. 192), ab6 dar6 lupii (Ter. Ph. 59, Ad. 311, Eun. 832)
and sicuti, quasi (Enn. Ann., 522, 542 Sk).
10.2. Phonetically this change belongs with the general reduction of
unaccented vowels that is typical of languages with a strong stress accent.
In Latin this reduction had led at a time when the stress apparently fell
on initial syllables to such changes as *konfdkiont > conficiunt, *exkaidet >
excidit, *obklaussom > *occlz.isum. Such shortening affected both pre- as
well as post-accentual long vowels, resulting for instance in an initial
tribrach in amicitiam (Pl. Mere. 846). That the tendency continued into
post-classical Latin is confirmed by Consentius (5.392 K), who like St
Augustine (D.Chr. 4.10.24) regards iirhtor as uitium Afrorum along with
the lengthenings in piper, pl’ces. It is not surprising to find so many
examples of breuis breuians in the comic poets, apud quos, nisi quod
uersiculi sunt, nihil est aliud cottidiani dissimile sermonis (Cic. Or. 67), but
its occurrence in the tragedies of E n n i u ~shows
~ ~ how close poetic discourse
in general was to common speech.
10.3. This phonetic shortening of long vowels was extended however
to the treatment of heavy syllables adjacent to the accent as metrically
light. Thus the light syllables underlined in & &e orationem (Enn. tr.
258J), skn& u&t (Pl. MO. 952, Cu. 268), h& (tr. 415) uii&thtes, gii&
nhbunt (Pl. Am. 939, Mi. 1091). These extensions cannot be attributed to
any plausible phonetic The only way to lighten such syllables
would have been to reduce the consonant cluster following the vowel, e.g.
*ades or *uolutates.
10.4 In fact Ritschl proposed just such an explanation for pyrrhic
uelint, feriint etc., citing dedrot, dedro from the Pisaurian dialect (CZL
25 The tragic fragments are cited from Jocelyn’s edition (1969), as those of the Annales are
from Skutsch’s (1986: 59-61).
26 As Lindsay (1922 7) already observed.

12378, 379). Other early forms like dedron from Rome and coruueron,
dedel.0 from Praeneste (CZL 1230, 59, 61) might also be cited, but the
restoration of -nt at Rome and the survival of the nasal into Romance
justifies Lindsay’s scepticism. In any case Ritschl’s hypothesis would not
account for anapaests like bdu2s qlli (Pl. Ps.812) or pyrrhics like bdnrs in
Terenoe’S ex Gruecis bonis Lutinus (Eun. 8), which is surely calculated to
suggest the vulgarism of Latin-speaking burburi. Even a shortening of the
long vowel in -3, -rs would still have left bou2s qui, bonB Lutinus, where
we should have to assume the same loss of -s as after an original short
vowel ($7.24) or some other equally unattested development. Praenestine
sueq. e& for sueisque eisdem (CZL 1262) is inconclusive as to the length
of the monophthong, and in any case we should need attestations from
Rome to account for the massive literary incidence. The extension is much
more likely to have been an artificial one, metrically motivated, and as
such never belonged to the spoken language and did not survive into the
classical poetic register.
(10.5. At this point it is perhaps worth raising the question whether
some of the instances of breuis breuiuns may in fact conceal peculiarities
in early Latin pronunciation, for which we do have other evidence. Thus
-
duus secum (Rud. 129) may be scanned not 6 - - but - - with dg for
dli, ilk qui (Rud. 1240) not --
- but - -, with apocope of -e. Sometimes
an archaic form may have disappeared in the tradition. Thus potest fieri
prosus (Trin. 730) may well conceal pote fieri with ellipse of est as in Aul.
309?7 However, the volume of hard evidence is too large and its range,
including all the pre-classical dramatic poets, too extensive to be greatly
whittled away by alternative hypotheses. The shortening of final long
vowels in dissyllabic words with which we began undoubtedly reflects
ordinary speech habits: b e d , dud, egd etc. were standard in classical poetry.
Where the word belonged to or was in its usage associated with a paradigm,
the length was retained: thus adverbial citd, mod6 but ablative cito, modo;
b e d but probe; putit ‘for instance’, cuu2 SLS ‘look out please’, all with
specialized senses that detached them from the verbal paradigms to which
they had originally belonged.
10.6. Sometimes poets exploit the availability of alternative forms, the
one current, the other archaic, for metrical convenience or for expressive
ekfect (cf. $12). m e latter is seen in Virgil’s uak uuZ2 (E. 3.79), the
*‘C. F. W. Miiller (1869) based his Iambenkiirzungsgesetz largely on metrical analysis, as did
most of his successors, including even 0. Skutsch (1934). It is surely time that the vast body
of data assembled by Miiller and the scholars who subsequently refined and modified the
initial account was re-examined in the light of modem research both in early Latin metrics
and more particularly in Latin historical linguistics. For a detailed examination of the relevant
phonology see Allen (1973 179-99).

38 R. G. G. Coleman
former in Lucretius’ idque sibi solum per se sapit et sibi gaudet (3.145) and
Propertius’ prora cubile mihi seu mihi puppis erit (2.26.34), both rather
laboured lines, Catullus’ tepcfaciet beside madgfient (64.360, 369, Ovid’s
liquefaciunt but liqu2fiunt (M. 7.161, Pont 1.2.55).
10.7. The extension of -6 beyond verbs of iambic shape like ago, fero,
scio is again to be seen as reflecting current speech. Thus we find in place
-
of cretic words with L -, impossible in dactylic verse (see 06.1), nescid
(Cat. 85.2), dixer6, menti6 (Hor. S. 1.4.104, 93), desin6 (‘rib. 2.6.41), PoZliEi
and Scipi6 (Ov. Ars 3.410); even for spondaic words, L -,find6 (Prop.
3.9.39, tolZ6, est6 (Ov. Am. 3.2.26, Tr. 4.3.72); and thence for spondaic
ends to longer words imag6, soluend6 (Sen. Ag. 874, Oed. 942), properab6
(Stat. Th. 2.342). The range of occurrence goes far beyond the genres in
which we should expect colloquialism. A striking instance of the early
adoption of a very colloquial pronunciation into the tradition of high
poetry is the use of uidZn. The form reflects udksne > uidkn > utd2n)
and is already attested in comedy, e.g. P1. Mil. 219. It is used in highly
excited passages by Catullus (61.77, 62.8) and by Virgil in A . 6.779, where
the colloquial tone is enhanced by the indicative verb in the dependent
ut- clause. Virgil’s context is Anchises’ solemn address to his son, and
Servius notes the usage as Ennian, adding the interesting comment adeo
eiw est immutata natura ut iam ubique breuis inueniatur. This situation
cannot have come about by the influence of Ennius and Virgil. It is
simply that between Virgil and Servius the boundary dividing literary from
colloquial had shifted, and it was no longer necessary, as it had been for
classical poets, to just@ the presence of colloquialisms in elevated contexts
from ancient precedent. The use of a subjunctive in the ut clause by
Tibullus (2.1.26) and Silius (12.713) suggests a desire to tone down the
colloquialism. Finally, what is characteristic of the poetic register is not so
much the introduction of the shortened forms but the retention side by
side with them of the older forms, ego, mihi etc. and of course can0 etc.,
where the analogy of other 1st sg. forms retarded the spread of -6; a
reminder that in the poetic register nothing is ever obsolete.
11.1. Syncope of unstressed vowels was a feature of the spoken language

at all periods, with or without reduction of the resultant consonant clusters.
Prehistorically *dyiiugai > bigae, *bpifakina > opificina (Pl. Mil. 880) >
o f i c i a , *pbsino > pdnd etc. In imperial Vulgar Latin dhminw > domnus
(perhaps already in Plautus, Cas. 722) uiridis > uirdis, comparare > com-
prare etc. Even in educated usage new examples continued to appear:
Augustus preferred caldw to calidus as being less otiosum (Quint. 1.6.19),
and caldu became the regular form in Vulgar Latin, whence caldo and

chaud. Quintilian himself preferred auducter to the unsyncopated form,

which he regarded as molestksima diligentiae peruersitas (1.6.17 and 19),
and the exclamatory ualde (< ualide P1. Am. 1062) was normal in classical
Latin alongside the adjective ualidus. The grammarian Caper (7.108K)
regarded balneae as more correct than balineae, which was in fact the
older form.
11.2. The availability of syncopated and unsyncopated forms, whether
or not they were both current outside the poetic register, again provided
useful metrical variants. Thus Plautus has colloquial ardos (Pers. 266)
beside midi (Rud. 574), and surpuit (Capt. 760), which must have been
formed in the period of initial stress, from *sbbrapuet, like captibus from
c a p i t i b , if this is the correct form at Enn. Ann. 511 Sk; conversely archaic
dexteram beside dexrras (Merc. 149, 969,an alternation that survives into
the classical poetic register, e.g. Horace’s dextera and the usual dextrii (C.
1.2.3, 2.17.29).
1l.3. This variation not surprisingly became confused with that between
forms with and without an anaptyctic vowel, e.g. again from Plautus per-
iculo and populi beside the older periclum and poplo (Capt. 687, St. 492,
As. 617, Ps. 125). Variants like saeclum and saeculum, uinclum and uin-
culum are normal in classical poetry, the anaptyctic forms being obviously
favomed in the nom.-acc. plural. The archaic tone of saecla, however,
justifiesits choice in Virg. E. 4.46 in preference to the metrically acceptable
saecula.
1l.4. Shortened forms like dites beside diuite (Prop. 3.4.1, 3.5.4) are
due not to syncope but to the loss of w between like vowels and subsequent
contraction. The phenomenon, well attested in Plautus, was especially
frequent in the w- perfects, audissem, complerunt etc., and was analogically
extended to give Plautine amarit, noram etc. The forms were convenient
for poets, e.g. admorunt in Virg. A. 4.367, nossem with audissem in Tib.
1.10.11-12. The latter, like complerunt, imply derivation from -grunt rather
than -aunt (see 9014.34). In prose such forms were accepted by Cicero
( O r 157) and treated as the norm by Quintilian, who recommends uitauisse
etc. only for compositio (1.6.17, 9.4.59). So here again the poets are in line
with general trends in the language.
11.5. The acceptability in literary language of the shorter variants, of
whatever origin, enabled the poets to adopt them occasionally without
incurring charges of colloquialism. In satire of course occasional echoes
of sermo cottidianus were appropriate. So we find Horatian caldior (S.
1.3.53), for ccilldtor, soldum (S. 2.5.65) and even surpite (S. 2.3.283), which
like Lucretius’ surpere (2.314) extends syncopation beyond the perfect
forms to which Plautus seems to have confined it. Only calidibr is metrically
impossible; the rest would be positionally restricted. This last consideration

40 R. G. G. Colemun
accounts for the rare examples in Horace's odes, e.g. lumnue (2.2.2). There
are a few where the unsyncopated form would be metrically impossible.
Thus surpuerut (4.13.20) and, much more remarkable puertiue (1.36.Q
which cannot be motivated phonetically from pu'Zritfue and would have to
be an otherwise unattested archaism, reflecting pkrftfue!This is surely out
of the question, and the form must be put down to the poet's boldness in
innovation. Nothing quite comparable is to be found in other classical
poets, who follow established precedent.
ll.6. Forms like posta for positu (Lucr. 1.1059), which were metrically
convenient rather than necessary, originated in colloquial Latin (e.g. ,
expostus in Cato, Agr. 151.2), where the influence of the noun postis may
have assisted the verbal syncope. They were regular in Vulgar Latin, as
the Romance reflexes imposto, compote etc. show. However, they had
entered the poetic tradition as early as Ennius, whose precedent for Virgil's
repostum (A. 1.26), where for once the full form is indeed unmetrical, is
reported by Servius. Lucretius also has repostu (1.35) along with many
other syncopated compound forms that are used by Virgil. The latter often
serve to emphasize an alliteration. Thus in plucidu compostus puce and
impostu Typhoeo and expostaque ponto (A. 1.249,9.716,10.694), -pos- falls
in a homodyne position.
ll.7. Factors other than mere metrical convenience can also be dis-
cerned elsewhere; e.g. in Puridis direxti telu (A. 6.57), where the longer
variant direxisti would be uneconomical of metrical space and phonetically
less harsh. The importance of tradition is clear in Virgil's choice of porgite
(A. 8.274). Servius on 1.26 again reports Ennian precedent, and the form
occurs in Cicero's Arateu 211. The established status of -postus is illustrated
further by Propertius' impostu (4.2.29) alongside older and more widely
established shortenings like duxti and consumpsti (1.3.37)'" and by Silver
epic examples like repostum (Val. macc. 2.286), impostu, expostus (Stat.
Th. 1.227, 5.551). Rare examples of apparent extensions are to be found,
like Statius' replictue (Silv. 4.9.29) in place of the unmetrical replicitue or
the alternative innovation replicutue. The classical reluctance to admit what
was so prominent a feature of Vulgar Latin, except where archaic p r e
cedent within the genre conferred respectability, was thus strictly
maintained.
12.1. Poets were not conspicuous innovators in pronunciation or mor-

phology, being content to adopt selectively what was already current in
Again likely to have begun at the time of initial stress: dobistei rather than dowhtei,
unless -tei was older than -istei in sigmatic perfects.

educated use. However, they did take full advantage of archaic gram-
matical forms which, though rare or even obsolete outside poetry, were
established in the poetic tradition and so not alien to cultivated readers.29
Sometimes an archaism provided a briefer form. Thus deum for deorum
in the urgent sequence of angry questions .. . quare / templa munt antiqua
deum? (Hor. S. 2.2.104), with the archaic form juxtaposed to antiqua, or
the even more archaic diuom in tuis flexus Venerisque gratae / uocibus
diuom pater (Hor. C. 4.6.21-2), which contributes to the heroic context
not only the old case form but also the archaic form of the root, which
adds to the Ennian alliterative effect.
123.Some examples suggest that the poet had Greek -ov in mind, as
magnanimum heroum (Virg. A. 6.307) and Danaum Euboico litore mille
rates (Prop. 2.26.38), where the toponymic adjective following V1k-n in
the previous line provides a strong Hellenic tone,30while the archaism of
a locatival ablative without a preposition and the metonymy of rates
form a rich poetic context for the genitive. The use of the old genitive
forms in technical terms of law and the trades, e.g. fabrum, iugerum,
sestertium, socium, well illustrates the parallels that are possible between
the subliterary registers of the language and that of poetry.
lZ.3. The a-stem gen. pl. -drum was as ancient as thematic -um. Hence
forms like caelicolum (Enn. Ann. 445 Sk, Virg. A. 3.21) and agricolum
(Lucr. 4.586) must be analogical, like the converse -difor -as. In proper
names the influence of Greek -Qv must again be admitted, as in Lucretius'
Aeneadum genetrix hominum diuomque uoluptas, where the forms Aene-
a d ~ andm ~diuomque
~ reflect the association of heroic Greece and ancient
Rome. The appropriate Greek resonance of Virgil's optume Graiugenum
(A.8.127) in Aeneas' address to Evander contrasts with the more native
Latin Graiugenarum (Lucr. 1.477).
12.4. At other times the chosen form may be longer, more weighty,
like the archaic gen. sg. -di. The inherited -as,which it replaced, survives
in the legal phrase pater familias and as an archaism already in Livius'
escas (poet. 12M), Naevius' Terras (FPL 8B) and Ennius' uias32along with
ithe ancient nom. sg. aquilii (Ann. 430, 139 Sk). In fact -di was already
29 For archaic morphology see Leumann (1959 143-5).

In view of the attested Greek variants 'OXLUC~S and 03h~&k and the recharacterization of
hwppr, acc. sg. of hqzs (&cpeds), VIixCn may be a genuine Greek form in spite of Housman
(Diggle-Goodyear (1972 2. 834-5)).
31 The early Latin nom. sg. would have been Aened?, d archaizing Aeacidn (Enn. Ann. 475
Sk).
32 aurus for aurae, reported by Servius as an anh'qua Iectio at A. 11.801, turns up in the ninth-
century Bemensis d and may be due not to Virgil's archaizing passion but to the interference
of auras (799).

42 R. G. G. Coleman
archaic by the early second century. Monosyllabic -ai (usually written -ae
in our MSS) is much the more frequent in Ennius, who tends to keep the
dissyllabic form for verse ends, e.g. the aptly toilsome siluai frondosai and
in the wholly spondaic line olli respondit rex Albai Longai, where the form
of the pronoun adds to the antique tone (Ann. 179, 31 Sk). Plautus has -
d i in parodic phrases like filiai nuptiis (Aul. 295), and magnai rei publicai
gratia (Mil. 103).
US.Lucretius has many instances, again especially at verse ends
(animal; materidi etc.), and the overall total in fact exceeds that for -ae.
But Virgil uses the variant more freely, if much less frequently. Thus in A.
3.354, aulai medio libabant pocula Bacchi, the opening archaic genitive
dependent on an unprepositioned locatival ablative (cf. the normal classical
idiom in media aula), the metonymic Bacchi for uini, appropriate to
the strict sense of libabant with which it is linked by assonance, and the
contrasting heterodyne and homodyne patterns in the otherwise metrically
symmetrical halves of the verse all combine to celebrate this emotional
high point in Aeneas' narrative. The genitival archaisms both occur
together in the description of 'hrnus' army, diues equom, diues pictai uestis
et auri ( A . 9.26). The use of the genitive rather than ablative case with
diues, the archaic forms of the two genitives and the stately spondaic
rhythms provide tones worthy of the heroic scene. The use of -di declines
in imperial epic. When it does occur, it is probably a Virgilism. 8
13.1. The interaction of i-stem and consonant stem nouns and adjectives
had begun prehistorically with the transfer of i-stem nom. pl. -es and dat.-
abl. pl. -ibos. However, in the historical period it is generally at the expense
of i-stems, with *mentim, *mentl; menti;r replaced by mentem, mente,
mentes.
13.2. The encroachment of imbre on imbri, parte on parti etc., as well
as reverse confusions like couentionid for *couentione,are attested in early
Latin (cf. P1. MO. 142, Ter. HT. 57, CZL 1381). In the first century BC,
where -e was spreading rapidly in educated usage, retention of the older
forms provided useful metrical alternatives, e.g. fin& fin2 (Lucr. 1.978,
4.627), ignl; ign2 and currentl; rubent2 (Prop. 1.9.17, 3.5.36, 4.5.12, 3.10.2);
cf. caelestl; caelest2 (Ov. Pont. 3.5.53, M. 15.743). In only a few instances
were the newer forms metrically difficult - celere for instance required
elision into a heavy syllable - but the availability of options once more
enabled greater variety in placing the word, varying the proportion of long
vowels and creating patterns of assonance, e.g. the opening juxtaposition
in Sigea ignifreta lata relucent, and the repetition of tiin cristasque rubentis/
excipiam sorti (Virg. A . 2.312,9.270-1). For increased syllabic weight innov-

ative -t was used in maiori (Luc. 7.162), and capiti (Virg. E. 6.16) avoids
capite, which would have needed elision.
13.3. That the distinction between k and the exilis uox of f mattered
to the educated Latin ear is plain from Probus’ discussion of the relative
merits of urbts and urbcs, turrim and turrem and Gellius’ reflections
thereon (N.A. 13.21). Because such pairs are not metrically distinct, we
have no way of recovering their distribution from the manuscript tradition,
which was inevitably corrupted by the phonetic confusion between the
I
reflexes of classical i and e in the late Empire. Appeal to Virgil’s autograph

is made by Probus to show that the poet made his choices arbitrio consili-
oque usus au*. So we have the assonance of urbisne inuisere, Caesar (G.
1.25) but the appropriately pinguior (‘fuller’) uox in centum urbes habitant
magnm (A.3.106). Similar observations are reported on turrim and securim
(A. 2.460, 224), and Gellius cites the alternation between tres and trk in
successive lines (A. 10.350-1).
13A Finally the difficulties posed by gen. pl. -ium following a heavy
syllable, as in present participles and many i-stem nouns and adjectives.
These were overcome in the participle by reviving the consonant stem
forms that made up the original paradigm of the -nt- participle, the i-stems
having been introduced from the remodelled feminine *ferenti (cf. Skt
bharantr and G k +kpouoa <*bherontia). Virgil’s cadentum (A. 10.674) and
the remarkable sequence including the verse-final rhymes ruentum,
parenturn ( A . 11.8867) would have seemed distinctly archaic, especially
if the older spelling lacrumantum is authentic. Later but still pre-classical
is the extension of consonant-stem genitives from participles to i-stem
adjectives, especially when used substantivally like caelestum (Enn. var.
23V: cf. Virg. A. 7.432) and agrestum Virg. G. 1.10). The metrically useful
forms, though not so likely to have had an archaic tone, since confusion
of the two paradigms was clearly widespread in the colloquial register,
nevertheless distinguish poetic discourse from literary prose usage, which
here, as sometimes elsewhere, distanced itself from the colloquial more
sharply than poetic usage felt the need consistently to do.
14T.Older forms of the imperfect tense like scibas (Enn. tr. 27W), insan-
ibat (Ter. Ph. 642) were useful to dactylic poets avoiding tribrachs and
cretics, e.g. stabilibat (Enn. Ann. 42 Sk), audibat (Ov. E 3.507), in place of
the more recent forms stabiliebat, audiebat which were preferred in prose.
Now and then the archaic form is preferred to a metrically possible alterna-
tive for positional and phonetic reasons. Thus in Virg. A. 8.436, certatim
squumis serpentum auroque polibant, the verb form contributes along with
the metrically unavoidable serpenturn to the aural effect of a verse in

44 R. G. G. Coleman
which the only dactyl is aurOquZpZllibant with auroque in hendiadys with

squamis. This ability to make a virtue of metrical necessity can also be
seen in Cat. 64.319, uellera uirgati custodibant calathisci with its dignified
central spondees.
14.2. Of the three classical variants of the third person plural perfect,
inherited -Ere, innovative -2runt (<*-is-ont) and -grunt, a conflation of the
two, the first is notably more frequent in poetry than in prose. It is in fact
the oldest of the three forms, but it was its trochaic ending that enabled
it to overcome the prejudice reported by Cicero (Or. 157) against a form
wrongly believed to be shortened from -aunt. The frequency at all periods
of the shortened forms cessarunt, norunt, which must have replaced ces-
saue'runt, nouErunt, not the forms in -Erunt, suggests that the former were
much more frequent than is often supposed. They are certainly the starting
point for the Romance forms, It. -arono etc. Even in Ciceronian clausulae
the short penult sometimes offers a better rhythm; e.g. mortem daldZrunt,
rnbu8runt respdndZ0, de mknihk iimfs2runt (Cic. Cael. 24, 27, 64). In
dramatic verse, -Zrunt can be certified only at the end of a verse 01;
hemistich, e.g. merubunt (Pl. MO. 281), emZrunt (Ter. Eun. 20), but there
are many other positions in which pZpUZrunt, dWrunt etc. would be as
acceptable as the -Erunt forms that are usually assumed.33
14.3. The heavy final syllable of most perfect stems severely restricted
the incidence of -Zrunt in dactylic metres. Thus Virgil has tiilZrunt, mix-
ciiZruntque,stZtZruntque ( E . 4.61, G. 3.283, A . 2.774) but implErunt, nbrunt
(E. 6.48, A . 6.641) for impZEuZrunt, nbuZrunt. The shortened forms were
supported by implesti, nosse etc. (see 911.4). In Lucretius, combined with
the small number of -Zrunt forms, they outnumber -Erunt. They are also
notably more frequent in the Metamorphoses than in the Aeneid.
14.4. However, -Ere, which is the most frequent of the three in Lucretius,
remains the distinctively poetic form in classical Latin. Virgil uses it in all
positions even where it is not metrically necessary; e.g. at the end of A.
1.398, et coetu cinxere polum cantusque dedere, where the rhyme with
cinxere is clearly important, and in the famous opening to Book 2, Contic-
uere omnes intentique ora tenebant, where the light syllable is actually
elided to usher in the rhythmic gravity of the central section of the verse.
145. Finally an archaism which, though not frequent, is remarkably
persistent in all genres of classical poetry, the medio-passive infinitive -ier.
Its origin is probably the same as that of -E,% with which it is in allomorphic
33See F'ye (1963) for further discussion and relevant statistics.

74viz.a gerundial *-i;: > *uortiz (cf. Skt v g y i ) uorti or uortie+r. The -1 form could
also reflect -ei (cf. agrwith Skt nir-aje),but there are no examples before monophthongization
to test this, and it would not account for agier etc.

variation in early Latin. The tendency for -ier to be placed at the end of
the verse in republican drama suggests that it was already archaic. In
classical verse it offered useful metrical alternatives to iambic and pyrrhic
endings. In practice the first of these, exemplified in, say, *@er uentis, is
rarely if ever attested, the second widely, e.g. nitier (Cat. 61.68), accingier
(Virg. A. 4.493), labier (Hor. Epist. 2.1.94), spargier (Hor. C. 4.11.8), all
followed by a vowel.
15.1. One notable difference between Latin poetry and literary prose is
the higher frequency of Greek sounds. This is true of all poetic genres,
even when the subject matter is not specifically Greek, though the pleasure
that was got by educated Latin speakers from the exotic sounds cannot
be dissociated from the cultural associations of the words that contained
them (see 0027,28).
15.2. The introduction of Greek sounds into Latin by Greek speakers
was of course, like their inability to cope satisfactorily with distinctive
Latin sounds such as [k"], [w] and [f], a mark of incompetence (cf. Quint.
1.4.14). As in other societies, unless the speaker belonged to one of the
prestigious artistic or learned professions, the defect was a social handicap,
looked down upon by educated native speakers. By contrast the occasional
deliberate injection of a distinctively Greek word, sound or idiom into the
discourse of an educated Latin speaker, provided it was kept to conven-
tionaFlimits -one thinks, for instance, of the contrast between the poetical
and epistolary Cicero and the Cicero of the formal literary prose - was
viewed as the natural outcome of what since the second century BC had
been a self-consciously bilingual culture. This bilingualism accepted after
son18 resistance the introduction of Greek technical terms into the learned
vocabularies of rhetoric, grammar, philosophy, science and technology,35but
the &lux there was much more a prose phenomenon and a relatively late
one, as the pages of the A d Herennium, Celsus and Vitruvius show.
15.3. Quintilian extols at some length (12.10.27-34) the aesthetic superi-
ority of Greek y and . ~ , 3the
~ two peregrinae litterae, and the aspirated stops
ph, th, ch. (It is not clear whether his omission of the aspirated allophone
of r is significant or not.) Instead Latin has the horridae litterae, f, w (the
Aeolica littera!), qu and final -m. Of &$vpos he writes si nostris litteris
scribentur, surdum quiddam et barbarum eficient.
95 See Coleman (1989).

" Greek v was originally [U] everywhere and remained so in some dialects until the spread
of [y] from Attic-Ionic in the Hellenistic period. 5 was originally [dz]or [zd], perhaps both,
depending on the dialect, but in Hellenistic Greek had become [z:] generally.

46 R. G. G. Coleman
15.4. In early Latin the convention seems to have been, as in English

until recent times, to naturalize all foreign words, including proper
Thus *ampora < 6p$opia (acc.), reflected in the diminutive ampulla beside
classical amphora, carta < x d p r ~ sand mama < @.Sa, tumba < T+/~os, tus
< O h , in which the peregrinae litterae were never restored. This practice
seems to have continued in Vulgar Latin, where e.g. colpus < d a + o s ,
balneus < fiahavciov, spada (cf. class. spatha, but dim. spatula) < umdOT.
15.5. Traces of the early Latin convention are discernible in the MS
tradition of Plautus and Ennius. Thus sonam (PO. 1008 ms. A), carmidi
(Trin. 744 A), sumbolum (Ps. 716 A), Sibulla (Ps. 25 A), scema (Am. 117
ms. J), baratrum (Ba. 149 P). From Ennius Andromaca (tr. 102 J test.),
cartis (Ann. 458 Sk), poinice (tr. 258 J test.), tyeste (tr. 307 J test.) maceriis
for maceris (Ann. 519 Sk) < *macaireis. It is probable that before c. 150
BC all Latin poets wrote (and pronounced) Ampitruo, numpa, Smurna,
teatrum, Tuestes.
15.6. The first datable use of the digraphic representation of Greek
aspirates occurs in Achaia and the unetymological triumpham beside
Corinto on Mummius’ dedication of 145 BC (CIL I* 626). At what date )I
and z were introduced we do not know, but it is likely that they were
systematically pronounced and written by the neoteric poets in charta,
nympha, Rhodope, tympanum, zephyrus, Zmyma etc.
15.7. It was not just the individual Greek sounds in cithara, cyathus)
delphinus, rhythmus, lychnus etc. that brought exotic colour. There were
also the unLatin combinations of familiar Latin sounds, as in the diphthong
in Eurus, the vowel sequences in ac?~Chaos, Trciilus, the initial clusters in
psallere, Xanthus, the medials in cycnus, Lesbia, the final syllables of
Actaebn, Corydan.
15.8. The exotic effect was further enhanced by the increasing retention
in poetry of Greek case forms in Greek nouns, which also provided
metrical variants, e.g. the accusatives lampad& (Lucr. 2.25), lampada (Virg.
A . 6.587) and nominative lampad2s (Ov. Ep. 14.25). Ennius had already
adopted acc. ait(h)era and nom. Aiacida (Ann. 545, 167 Sk), but it was
probably Accius who established the practice, at any rate in proper nouns
according to Varro, who also notes (L. 10.70) the currency of mixed
declension and complete Latinization, which was the norm in common
nouns of Greek origin. Catullus has for instance in 64 and 66 VOC. Theseu,
acc. Amphitritcn, Minaa, nom. pl. Nereid&, gen. Locridos Arsinocs and the
variant datives Minoidi and Thetidl; P e G and Pelt%, where the metrically
necessary but unLatin contraction gives the latter form a distinctly Greek
quality. The usage was maintained and extended in Augustan and later
37 For a comprehensive account of the whole subject see B i d e (1987).

poetry, Lucretius’ Tityon (3.984), and cinnamon and Patroclon in Proper-

tius (3.118, 2.8.33).
15.9. ‘Ihe effects of this Greek colouring were recognized explicitly by
the ancients themselves. Cicero, having stated that poets are governed
by sound rather than sense, cites as an instance a verse that is locorum
splendidis nominibus illuminatus; the text is uncertain but it clearly con-
tained Helles, Tmolum and Tauricos (Or. 163). Other examples are
discussed in 0027-28.
16.1. The phonetic of alliteration, assonance and rhyme are a

well-known feature of Latin poetry, though by no means exclusive to the
register. The most famous of all Latin alliterations was after all Caesar’s
ueni uidi uici, reported by the Elder Seneca (Suus. 2.22), which shows that
it was typical of conversational epigram, though the figure is conspicuously
absent from the Commentarii. All three figures can have expressive func-
tions in poetry, as in ritual texts and even legal formulae, though with the
possible exception of onomatopoeia, it is often impossible to dissociate
interpretation of any such phonetic effects from the perceived meaning of
the words themselves. Examples can be noted and analysed precisely and
objectively, but anything beyond that must be to some extent subjective
and inlluenced by the reader’s own linguistic culture.39
162. Alliteration, even more than the other two, also has a powerful
cohesive effect in binding together words that the author wishes to
associate closely, whether or not they are adjacent or belong together
syntactically. In Horace, C. 2.3.1-4, aequam memento rebus in arduis /
seruare mentem, non secus in bonis / ab insolenti temperatam / laetitia,
moriture Delli, the alliterative link between memento and mentem, with
its characteristically Horatian etymological word-play, and the unnerving
moriture contribute significantly to the effect of the exhortation.
16.3. Alliteration is a striking feature of early Latin literary compo-
sition, no doubt assisted by the initial stress accent. This coincidence of
alliteration with stress is also seen in Old English poetry with alliteration
of two or three stressed syllables in each line: e.g. oft Scyld Sc@ng sceap-
ena preatum / monegum m@g@m meodosetla oftcah (Beowulf4-5) ‘Often
Scyld son of Scef from bands of enemies, from many races, the mead-
benches took away’.4oWe find in prayers formulae like quod felix faustum
For more detailed discussion see HofmannSzantyr (1965 699-721).

39 For some salutary remarks on the significance of alliterations see Goodyear (1972 336-41).
4o Whatever rules there may have been, comparable to those in Germanic and Welsh alliter-
ative. poetry, have disappeared with the lost texts. For an unsuccessful attempt to recover
such rules in extant Latin poetry see Evans (1921).

48 R. G. G. Coleman
fortunatum siet (Cic. Diu. 1.102) and pastores pecuaque salua seruassis
(Cato Agr. 141), and in dedicatory inscriptions various combinations of
donum with datum, donatum and dedicatum, which show that the figure
survived in the religious register. In early poetry Naevius has eorum sectam
sequontur multi mortales (FPL 6B), Ennius Apollo puerum primus Priamo
qui foret / postilla natus temptare tollere;/ eum esse exitium Troiae, pestem
Pergamo. (tr. 5 9 4 1 J ) and the virtuoso sequence machina multa minax
minitatur maxima muris (Ann. 620 Sk).
16.4. There is nothing on this scale in Greek poetry, nothing either in
surviving Latin prose, though some remarkable alliterations can be found.
For instance Cicero, Tusc. 1.118-19: tum incideret in mortis malum sempi-
ternum; portum potius paratum nobis et perfugium putemus. quo utinam
uelis passis peruehi liceat! sin reflantibus uentis reiciemur . . . . This highly
alliterative and sustained nautical metaphor comes at the emotionally
charged conclusion of a long exposition. Even Nepos sometimes indulges,
e.g. illum ait Magnesiae morbo mortuum neque negat fuisse famam
uenenum sua sponte sumpsisse (Them. 10.4). The two examples are signifi-
cant in not being drawn from high oratory; a reminder of the stylistic
common property shared between poetic discourse and literary prose
generally.
16.5. The archaizing Lucretius was of course greatly addicted to allitera-
tion. As a result the functional power of the figure is greatly diminished.
Catullus by contrast sets the appropriately archaic tone for his epic theme
of Peleus and Thetis with Peliaco . . . prognatae . . . pinus, Neptuni nasse
and fluctus et fines (64.1-3); but the figure is reserved in the personal,
poems purely for expressive effects, as in lepidum nouum libellum and
pumice expolitum (1.1-2), senum seueriorum (5.2), fidem. . . foedere far;
lendos (76.3~l).4~ The contribution of cedente carina and languida . . .
Zitoribus to Propertius’ depiction of the deserted heroine (1.3.1-2) is all
the stronger because he alliterates so infrequently. Similarly in 2.22.1-2,
scis here mi multas pariter placuisse puellas; / x i s mihi, Demophoon, multa
uenire mala, the alliteration of m, p, 1 combined with the repetition of scis
and the colloquial here and mi convey a vividly dramatic effect. Finally in
Virgil’s et pro purpureo poenas dat Scylla capillo (G. 1.405) the juxta-
position of the visual image of crime and its punishment is reinforced by
the same alliteration which unites the syntactic unit of epithet and noun,
underlined by the grammatically conditioned homeoteleuton. As often in
poetry one feature accompanies others.
I
41For the relatively high frequency of alliteration in the dialogue love poems, Catullus 45
and Horae C. 3.9 see Wilkinson (1963 26-7).

17. Assonance is also employed for expressive and cohesive purposes. It

is often combined with alliteration, as in Lucretius’ linquitur ut merito
maternurn nomen adepta / terra sit, e terra quoniam sunt cuncta creata
(5.795-6); with its characteristic word-play between maternum and terra,
and the use of recurrent patterns of linguistic elementa to reflect the
@hes@ of cosmic elernent~.~’ In C. 3.4.69-71 Horace gives a didactic,
pedantic, emphasis to the mythological exemplum of uis consili
expers, which mole ruit sua, by the repetition of te, ti in testis mearum
centintanus Gyges 1 sententiarum. This is continued in the repeated t
and i through the rest of the stanza to the h a 1 domitus sagitta. The
opening stanza of C. 4.13.1-4 is even more striking, with the repeated
exilitasl,of 1; r and C expressing contempt for the hapless Lyce. The use
of assonance merits attention, even if interpretation of it must often
remain speculative.
18.1. Rhyme has been an important structural feature of Western

vernacular poetry since the Middle Ages and has a strong cohesive role
in such varied genres as the terza rima o f Dante’s Divina Commedia, the
heroic, couplet of Chaucer’s Canterbury Tales and the sonnet forms of
Petrarch and Ronsard. The need to provide the specified rhyme can be
constricting and the use of the figure has declined in modern poetry. The
earliest surviving pieces of rhymed Latin verse are Christian hymns, though
in most of the pre-medieval hymns the rhymes, even when frequent, are
not systematic. Moreover we cannot be sure of pronunciation. In Fortu-
natus’ Vexilla Regis was it assumed that congregations would rhyme
prodeunt with mysterium, as they would uestigia, gratia, hostia? In Insular
Latin, where pronunciation was more conservative, we have in fact some
of the earliest systematic rhymes, as in the Lorica attributed to Gildas and
St Columba’s Altus Prosator (Anal. Hymn. 51.262 and 216).
18.2. The two areas where rhyme is to be found in pre-Christian Latin
are in popular incantations, where the correspondence was systematic and
obviously magical, and in high rhetorical prose, where homeoteleuton or
sequential rhyme was occasional and calculated for special effect. Thus
terra pestum teneto. salus hic maneto (Var. R. 1.2.27), and est igitur
haec, iudices, non scripta sed nata lex, quam non didicimus, accepimus,
legimus, uerum ex natura ipsa arripuimus, hausimus, expressimus, ad
quam non docti sed facti, non instituti sed imbuti sumus (Cic. Mil. lO),
where the rhythmic structure also reveals careful composition.
18.3. Such examples indicate an awareness of rhyme as a possible
42 See Friedluder (1941) for the exposition of this idea genemy in Lucretius.

50 R. G. G. Coleman
stylistic device, and, given the obvious frequency of good rhymes offered
by the grammatical morphology of a highly inflected language, it is perhaps
surprising that the possibilities were not taken up and developed in the
educated poetic tradition. One reason may have been that in contrast to
alliteration and assonance, which can affect lexically as well as grammati-
cally significant syllables, rhyme affects only the inflections, unless it is
dissyllabic, as in Chaucer, or even more intricate, as in the fantastic hexa-
meter couplets of the twelfth-century Bernard of Cluny’s De contemptu
mundiz4’ Hora nouissima, tempora pessima sunt, uigilemus. / ecce minaciter
imminet arbiter ille supremus.
18.4. The great majority of verse-end rhymes in classical poetry are
unsystematic. Many seem to have no significance, which is surprising, given
that the poet’s choice was unconstrained. Now and then however we can
see some significance, as in Ovid’s amnis harundinibus limosas obsite
ripas, / ad dominam propero. siste parumper aquas (Am. 3.6.1-2). Here
the rhyming links the exhortation closely to the relevant descriptive phrase.
Even the echo in the imperative siste of the rare vocative obsite may also
be calculated. Again in Horace, A.E! 99-100, non satis est pulchra esse
poemata. dulcia sunto. / e t quocumque uolent animum auditoris agunto, the
archaizing legalistic imperatives, unexpected in this context, are empha-
sized by their position, and the rhyme is supported both by the assonance
of pul- dul- and perhaps uol- and by the alliteration in both
verses.
18.5. There is good evidence for the occasional cultivation of internal
rhyme with the syllable before the principal caesura. This is particularly
notable in the Propertian elegiac pentameter, and can be seen as an easy
extension of a tendency to place two concordant members of a noun
phrase in these positions.44Thus in 1.1we find suis . . . ocellis (l), castas. . .
puellas (9,nullo. . . consilio (6) with the grammatically comparable
nullis. . . Cupidinibus (2), constantis. . .fastus (3) and impositis . . .pedibus
(4). An exceptional sequence, but there are plenty of similar individual
verses, and even a few dissyllabic rhymes like Tyrrhena . . . harena (1.8a.11).
Now and then two weakly rhymed words are not even in grammatical
concord, as irasci . . . tibi (1.5.8) and urgenti . . . dedi (4.3.12), though these
are at least syntactically linked.
18.6. In the well known sequence in Horace, C. 1.1.6-10, the rhyming
words in terrarum dominos euehit ad deos (6), which are also linked by
43Which includes the Name of the Rose line: nunc ubi Regulus? aut ubi Rornulus? aut ubi
Remus? / stat rosa pristina nomine, nomina nuda tenemus.
For a detailed classificationof the data on juxtaposed homeoteleuton see Shackleton Bailey
(1994).

alliteration, are syntactically unconne~ted?~ as they are also, for instance,

in 1.36.12, neu morem in Salium sit requies pedum. But the rest of the
passage shows strict grammatical concord from mobilium . . . Quiritium
(7) to Libycis . . . areis (lO), with only the alliterative certat tergeminis
tollere honoribus ( 8 ) unrhymed. Elsewhere in Horace internal rhymes of
this kind occur often enough to suggest intention, e.g. in C. 1.22
uenenatis.. . saginis (3), silua.. . Sabina (9), curis.. . expeditis (11) etc.,
in contrast to the concordant but unrhymed Syrtis . . . aestuosas (5). Ovid
uses such rhymes more sparingly after early instances like Ep. 1.6, which
has two rhyming and two non-rhyming concords. Perhaps the figure itself
was codsidered too naive and monotonous for fastidious poets, though
Martial still has examples now and then of both rhyming and unrhyming
concords.
18.7. In dactylic hexameters, while internal rhyme is frequent enough
for obvious grammatical reasons, it seldom marks the two ‘halves’ of the
verse. Examples are therefore the more noteworthy, as in Virg. A . 4.652-8,
uixi et quem dederat cursum Fortuna peregi / et nunc magna mei sub terras
ibit imago, / urbem praeclaram statui, mea moenia uidi, / ulta uirum poenas
inimico a fratre recepi, / felix, heu nimium felix, si litora tantum / numquam
Dardaniae tetigissent nostra carinae. Here five rhyming first person verbs
are placed one in verse-initial, three in verse-final position and one before
a principal caesura, as is the rhyming first person pronoun mei. The queen’s
proud statement of her own achievements, thus powerfully emphasized,
contrasts poignantly with the closing couplet in which she is the passive
victim, and there is no first person verb, only the alien and hostile teti-
gissent.
THE LEXICON
19.1. The lexical ingredients of the poetic register - the choice of words
and the way they are employed - are clearly important for its definition.
Indeed poetic diction often consists of words that in origin belonged to
ordinary prosaic usage but because of their adoption and reiteration in
poetic contexts acquired a distinctive status.
19.2. Words like formosus, mollis and tener were all thoroughly at
home, say, in rustic or horticultural contexts -porcus formosus, asparagi
molles, tenerae gallinae etc. But in poetry of the Callimachean connection
mollis and tener were polarized with durus and seuerus over a wide meta-
phoric range (Prop. 3.1.19-20, Ov. Am. 2.1.3-4 etc.) and through their
4s See Nisbet-Hubbard ad loc. For rhyme in the Odes see Skutsch (1964).

52 R. G. G. Coleman
frequent occurrence in such contexts acquired like formosus connotations

that in turn coloured new contexts. of erotic poetry they
certainly became, but the continued use of tener and mollis outside the
genre in their older more prosaic denotations hamper their classification
as poetic diction.
19.3. The case of formosus is more complicated. It is well known that,
whereas Virgil often uses the adjective in the Eclogues (e.g. 1.5, 2.1, 3.79),
where it is appropriately molle atque facetum, he uses it only once in the
Georgics, in its old sense, formonsd’ iuuenca ‘a well shaped heifer’ (3.219),
and never in the Aeneid. Did0 is always pulcherrima, never formonsa. The
ancient pulcher, used as a near synonym of formosus in Cic. Fam. 9.14.4,
nihil est. . . uirtute formosius, nihil pulchrius, nihil amabilius, retains an
elevated tone in both classical verse and prose usage. Bellus, originally the
diminutive of bonus (or rather *duenos), was used by Plautus and indeed
Cicero to mean both ‘fine’ and ‘pretty’, but it never quite rivalled formosw
in erotic poetry. Its lexical profile includes Catullus, Lucretius, Tibullus,
Horace Sat., Persius and Martial, which certainly does not mark it as
‘unpoetisch’, unless we confine ‘poetisch’ to the highest poetical genres.
However, the two words prevailed in Vulgar Latin and are reflected in
Romance, beau, hermoso etc. Pulcher disappeared without trace from the
spoken language.
19.4. Words may of course rise and fall in literary as well as social
status over the years and the emergence of a particular word in Vulgar
Latin at one period is no guarantee of its literary humility at another.48
19.5. Sometimes ancient testimony on the status of a word raises more
questions than it answers. When Andromache says, referring to Ascaniqs,
ecquid . . . et pater Aeneas et auunculus excitat Hector? (A.3.342-3), Servius
says of auunculus:humiliter dictum in heroic0 carmine. The word is certainly
rare in poetry, but is this merely because poets did not normally talk about
uncles? Perhaps Virgil is being bolder than he seems in introducing a homely
colloquial tone into epic dialogue? Equally infrequent in poetry is patruw.
It is used satirically by Catullus (74) of Gellius’ debauchery and also occurs
in the Sibyl’s conversation with Charon ( A . 6.402), casta licet patrui seruet
Proserpina limen. The fact that Pluto, who had raped Froserpine, was &e
brother of her father Jupiter, is bitterly underlined by the juxtaposition of
46 See Delatte (1967), who records tener as the most frequent adjective in Tibullus, fornosus
and mollis in Propertius.
47 Presumed to be the original form of this as of other adjectives in -0su.s. Virgil may have
been exceptional in retaining -ns-. For the change of Vns to Vs generally see M e n (1978:
28-9,654).
48 The point was well made by Ernout (1947) in reviewing Axelson (1945), who in this as in
other matters did not always see the significance of the patterns in his own statistics.

with casta and its alliteration with Proserpina. Like auunculus the
noun is used with characteristic semantic precision, as was not perhaps
the case in Servius’ day, when the use of auunculus of paternal as well as
maternal uncles may already have been consigning patruus to oblivion.
20.1. The rise and fall of words needs to be plotted with some care. Thus
@rare and lacrimare certainly ousted flere in Vulgar Latin, but both are
well attested in classical poetry. The start of Ovid’s lament for ‘Tibullus,
Memnbrta si mater, mater plorauit Achillen (Am. 3.9) is surely a place for
solemnity; nor does the chiastic structure of the line, with the Greek
accusatives of Homeric names at either end, seem an appropriate setting
for the unpoetic or colloquial. As for lacrimare, its membership of the
lexical group lacrima, lacrimosus ensured its survival in all registers of
the language.
2Q.2. When Juno complains to Aeolus (Virg. A . 1.68) that her enemy
sails the Tyrrhenian Sea Ilium in Italiam portans uictosque penatis, the
verb has its normal classical sense of ‘carrying a burden’, and its subsequent
semantic expansion to replace its hyperonym ferre in Vulgar Latin is
irrelevant. Again, although (delfessus was replaced in Vulgar Latin by
lassm, little can be inferred from this for classical poetry. Ovid prefers
fessus in his elegies, lassus in the Metamorphoses, but this is not merely a
matter of idiolectal whim. Virgil in the flower simile describing the death
of Euryalus has ( A . 9.436-7) languescit moriens lassoue papauera col10 /
demisere caput. The preference for lasso over the metrically identical fesso
may have been motivated by the resultant la- assonance, which binds
together two central words in the simile. However, given the long tradition
of the flower simile from Iliad 8.306-8 and Sappho 10% LP, it is unlikely
that Virgil would have been so motivated, if it had meant introducing a
word with distinctly subliterary associations.
’ 20.3. In contrast to fera, the noun bestia is rare in poetry. It is sporadi-
cally reflected in Romance (e.g. Fr. biche) and must therefore have become
established at some time in Vulgar Latin. Its occurrence in B.Afr. 81.1 but
not in Caesar might suggest an early date, but it is used by Livy and often
by Cicero; so its literary acceptability is secure for the classical period.
Not vulgar then, but perhaps not poetical either. This would make its rare
occurrence in Catullus 69.8 all the more striking: hunc metuunt omnes
neque mirum; nam mala ualde est / bestia nec quicum bella puella cubet.
The syntax and the vocabulary - ualde, bella - are colloquial but not
vulgar, and bestia, with its contrastive alliteration with bella, is certainly
vituperative (cf. mala tu es bestia in P1. Ba. 55), but again not necessarily
vuigar.

54 R. G. G. Coleman
20.4. It is of course rare that a lexical profile is homogeneous, wholly

poetic or wholly prosaic, and there are many other cases where the details
of a heterogeneous profile are illuminating. Consider for instance the word
l a b r ~ m ?a~shortened form of lambrum (cE $11.4). The formation type is
ancient, being paralleled in cribrum, delubrum, uelabrum etc. but it is not
productive in the historical period. The unshortened form occurs once,
where Lucretius warns against lingering in hot baths - si calidis. %
..
cunctere lauabris (6.799). In this, its original sense, it seems to have been
replaced by lauatio (originally ‘the act of washing’; cf. 032.1) and especially
by the Greek loan-word balineum, whence Romance bagno, bain etc. In
Cic. Fam. 14.20 labrum is clearly something that can be placed inside a
balineum, presumably a ‘tub’ or ‘basin’. In agriculture labrum is used of
a tub for mixing oil (Cato, Agr. 66.2) or making wine (Virg. G. 2.6, Col.
12.15.3). This certainly looks prosaic, and the impression is confirmed by
the diminutive labellum (Col. 12.38.3), which is also used of a decorative
bowl on a tomb (Cic. Leg. 2.66) and survives in Italian dialect words for
‘trough’ and ‘coffin’. Nothing very poetic here either.
20.5. However, the word is twice found in the Aeneid. At 8.22 it appears
in the simile of turbulent water in a cauldron, adapted from A.R. 3.756fL
where the corresponding word h&s is unmarked as to register. The dis-
tinctly unepic Latin word and the image of which it forms part are perhaps
not inappropriate to the unheroic predicament of a very indecisive Aeneas.
Again in 12.417, labris splendentibus introduces a homely detail into the
lofty heroic diction, reminding us that the high and mighty have humble
domestic objects around them, even if they are grander versions than ours.
The poetic effect in both instances depends precisely on the unpoetic
status of the word, for which plenty of parallels can be found in the homely
details of other epic similes.
20.6. Similarly in Ovid’s use of the plural labra of the bath of Diana
in Ep. 21.180, E 4.761 and Zb. 479, a reference to the goddess’ ‘bath-tub’
is deliberately irreverent, and even the trochaic plural seems belittling. In
the very next line of Fmti there is a reference to Faunus medio cum premit
arua die. The image of the god applying pressure to the croplands is
distinctly satiric, whether or not there is an allusion to his midday sexual
activities. There may even be a parody in labra Dianae of the labrum
Venerium ‘teazel’, a plant described by Pliny (Nut. 25.171) as in flumine
nmcentem. Diana, behaving in the myth like an excessively prim maiden
lady, is treated like one. There is a lot going on here, but the presence of
labrum, while it is essential to the witty poetic effect, does so again
49 What follows is developed out of an attempt to answer a query from Professor Kenney.

precisely because it is not in itself poetic. Nor does it acquire the status
of poetic diction as a result of its contribution here.
20.7. The most remarkable case in Latin of a sequence of prosaic words
combining to create a powerful poetic effect comes from Catullus. Not
one of the bawdy pieces sprinkled with coarse vocabulary - Ameana
defutata and the rest - which were designed to shock conventional sensi-
bilities, but the famous epigram, odi et amo. quare id facium fortasse
repiris. /nescio sed fieri sentio et excrucior (85).Every word here is prosaic
and there is not a single trope or figure to enrich their sheer ordinariness.
Contrast for instance a couplet like Propertius’ in me nostra Venus noctes
exercet amarus et nullo uucuus tempore deft Amor (1.1.33-4). Sustained
plainness of diction characterizes all Catullus’ epigrams, but this one is an
extreme case.
20.8. There is an almost unbelievable density of verbalization - nine
finite verbs and an infinitive in fourteen words. All the finite verbs except
one are in the present tense, all except one in the first person. Only one
is a subjunctive, and the mood there is purely rule-governed. The opening
paradox odi et umo is compounded by the choice of the ergative facium
to denote a passive state, a choice that is not unparalleled (cf. Hor. S.
1.1.64, 94);’ though it is here promptly ‘corrected’ by the passive fieri,
with the two contrasted verbs in alliteration with the prosaic adverb for-
tasse,5l which introduces the only tentative note in the couplet, referring
however not to the poet but to the reader. The progression from the
opening paradox through the puzzled question and his puzzled answer to
the reiteration of the emotion and the pain caused by it, is articulated
in the starkest and most austere lexical material. Poetic diction is clearly
not essential for a powerful poetic effect: series iuncturuque certainly a ~ e . 5 ~
Nevertheless it is doubtful whether this power could be sustained through
a poem much longer than an epigram solely by such unenriched prosaic
vocabulary, however artfully arranged the words might be.
20.9. Once again a walk-on part in a highly charged poetic context
does not confer the status of poetic diction. The fourteen words in Catullus’
epigram revert thereafter to their former prosaic status Similarly with
Horace’s odi profunum uulgus et arceo, which acquire their full metaphoric
sense only when we have read the rest of the stanza. The individual words
” On ut scias me recte ualere quod te inuicem fecisse cupio at Vindolanda see Adams (1995~:
123).
” The grammarian Cledonius noted (5.66 K) that forsan and forsitan are both more poetic
than fortasse, which certainly never occurs in the more elevated passages of Virgil and Ovid.
’* Axelson (1945: 98) by labelling idoneus, ordimre, praesidium etc. as unpoetic fails to
account for their successful appearances in poetry, most notably Horatian lyric. As Ernout
(1947: 69) and Marouzeau (1949, 1954 passim) insisted, context is of critical importance.

56 R. G. G. Coleman
retain no particular poetic charge; their effect hereafter is recoverable only

in a context that alludes specifically to this verse. In fact Horace constantly
uses plain prosaic phrases within richly elaborated contexts as an instru-
ment of poetic argument. The summary duke et decorum est pro patria
mon (C. 3.2.13) derives its impact from the vivid images of bravery
and violence that surround it. The homely iam dudum upud me est
(C. 3.29.5) is carefully contrasted with the grand and luxurious imagery
that precedes it. Poetic effects do not depend upon consistency of lexical
texture.
21.1. We often find that a pair of synonyms is divided between literary

prose and the higher genres of poetry but both words occur in the less
elevated poetic contexts. Consider for instance the ‘poetic’ words ensis,
letum, coniunx and amnis. The ancient noun ensis was synonymous with
gladius (Quint. 10.1.11), which supplanted it in general’usage before the
classical period and was in its turn later supplanted in Vulgar Latin ’by
the Greek loan-word spatha. For letum, a nobler synonym of the ancient
mors, Varro cites (L. 7.42) an old legal formula ollus let0 datus est, though
the legal register normally has mors. No less ancient than mors is uxor,
which is again regular in legal texts, though coniunx is generally preferred
in sepulchral inscriptions. While umnis, frequent in early Latin, is rare in
classical prose - only once even in the archaist Sallust - it is frequent
in Livy, another instance of historia proxima poetis. Of course gladius,
mors, uxor and flumen are often found in the higher poetic genres; the
significant fact is rather the very rare occurrence of ensis, letum, coniunx
and amnis in It is not irrelevant either to note that all four ‘poetic’
words disappeared eventually from the spoken language and have left,no
direct traces in Romance.
21.2. There are many other instances of such synonymic pairs: c€
pulcher and formosus (013.3), uirgo and puella (026.1), meare and ire,
celsus and altus, natus andjilius etc. The effect of a combination of two
such elevated synonyms can be seen in the solemn words of Virgil’s Sibyl,
Trbius Aenem pietcite insignis et armis / ad genitorem imas Erebi desckndit
ad ambras (A. 6.403-4). The identical quantitative rhythm of the two
verses produces an incantatory effect, which avoids monotony by the
different patterns of homodynes (indicated by the accents). The Greek
sound of TrciTzZs Am&% is balanced by the gravity of genitorem i m a , in
I
53For further discussion of the ‘poetic lexicon’ see Marouzeau (1962: 193-8), Leumann (1959:
155) and for Indo-European Schmitt (1967) and Watkins (1954 193-8).

to patrem and the unmetrical infimas. Another instance of the

great poet making a virtue of necessity.
22.1. The retention of archaisms, words that have become obsolete in

ordinary usage, is a feature of the poetic lexicon that has already emerged
in the preceding paragraphs. uerba a uetustate repetita, says Quintilian
(1.6.39), adferunt orationi maiestatem aliquam non sine delectatione. Not
only do they have auctoritatem antiquitatis but also quia intermissa sunt,
gratiam nouitati similem parant. In 8.3.24, writing of the dignitas conferred
by antiquitas, he continues eoque ornament0 acerrimi iudicii I! Vergilius
unice est usus; a good example incidentally of the close relation between
rhetoric and poetic in the ancient theorists.
222. Quintilian’s examples of Virgil’s archaisms include moerus, which
was presumably not merely a relic of the old orthography moirus” but
also intended to carry with it the archaic pronunciation and so underline
the connection with moenia. Other words cited with approvalssare olli, the
older equivalent of illi, also found in Lucretius and rarely on official
inscriptions, quianam, attested outside the Aeneid only in Naevius, Ennius
and Accius, and pone, which as an adverb is found in Virgil, Propertius and
Silver epic as well as Suetonius and Apuleius, but as a preposition has far
too wide and heterogeneous a lexical profile generically and diachronically
to even qualify as an archaism. Indeed pone occurs in a passage (Rhet.
Her. 4.14) illustrating the cottidianus sermo!
I 223. Not all archaisms were approved. In 1.6.40 Quintilian refers dis-
paragingly to words ab ultimis et iam oblitteratis repetita temporibus, citing
va?.ious examples (beside those in 8.3.25-7), which end with Saliorum
carmina uix sacerdotibus suis satis intellecta. These were still to be heard
in his own day and unlike the revival of old words with which we are
principally concerned here, they represent an unbroken tradition, pre-
served but progressively garbled over the centuries. As for the other words
cited, it is not clear whether the objection is to attempts to revive them
or to the idea of doing so. The first, topper, is not attested after Accius and
Coelius Antipater, antegerio which nemo nisi ambitiosus utetur (8.3.25), is
not attested at all, exanclare, well attested in early Latin, is not found
between Cic. Luc. 108 and Apul. Met. 1.16, and prosapia, described by
Which may indeed be what Quintilian wrote; d the MS variants mus, mis. m o e m is due
to Ribbeck.
55 One of them, porricere, Haupt’s correction of pollicerent, is problematic, since after Plautus,
although it appears in a proverb quoted in Cic. An. 5.18.1, inter mesa et porrecta, and in a
conjecture for porrigit in Var. R. 1.29.3, there are no certain poetic or indeed literary
attestations at all.

58 R. G. G. Coleman
Quintilian in 8.3.26 as insukrum, is used by Cicero ( E m . 39) with the

explicit label uetus uerbum and later by Suetonius and Apuleius. It is also
used by Sallust (lug. 85.10), whose taste for archaizing was notorious (Suet.
Aug. 86.3 Gramm. 10).
22.4. There is no evidence in any case that any of these words was
revived by any of the classical poets. What is important, however, is that
for poets there is no such thing as an obsolete word. Any word can be
resurrected for a specific poetic purpose, though to be effective it must
obviously be intelligible and its traditional connotations recognizable to
the educated reader. It looks as if there was already in classical times
uncertainty about the meaning of topper, as of caluitur and pedem struit
in the B e l v e Tables. Such doubts would certainly have been a deterrent
to revival.
23.1. Archaisms rarely appear in isolation from other ingredients of the

elevated style. Take Catullus 64.35-6, deseritur C i e r o ~linquunt
,~~ Pthiotica
Tempe / Crannonisque domos ac moenia Larisaea. The archaism linquunt
which like other uncompounded forms, cedere, Jidere, gradi, solari etc.,
was a particular feature of the higher poetic genres, is surrounded by the
exotic sounds of toponyms appropriate to the narrative and to Achilles,
who was born in Phthia. The strange coupling of Phthia with Tempe is
implied in Callimachus, H. 4.105,112, and the chiastic structure of 36 with
its concluding double spondee is very much in the grand style.
23.2. In Virgil, A. 1.254, Jupiter’s reassurances to Venus are introduced
with appropriate majesty: olli subridens hominum sator atque deorum. The
verse contains the archaism olli and the poetic use of sator as a synonym
for pater. In the speech itself we find fabor enim, quando haec te curu
remordet, / longius et uoluens fatorum arcana mouebo (261-2), with etymo-
logical word-play between fata and the archaic verb fari. In 2.54 fata deum,
the passive participle of the middle verb, itself an archaism, is combined
with the archaic genitive plural.
233. Ovid, M. 14.806, has casside in place of the usual classical word
galea?’ Unlike Virgil Ovid does not employ archaic words very much,
even in this epic-scale work. However, the whole sentence is laden with
archaism: posita cum casside Mauors / talibus adfatur diuumque homin-
umque parentem. Though not itself archaic, the presence of cum in the
56 Meineke’s correction of Scyros, the form which lies behind the actual manuscript readings,
is adopted by most modem editors.

57 As Leumann (1959 143) noted, observing Ovid’s reduction in archaic vocabulary, compared
with Virgd and Horace.

opening phrase reminds readers of the comitative sense out of which

the absolute construction developed. Straightforward archaisms are posita
for deposita, Mauors for Mars, adfatur for adloquitur, diuum for deorum
and the inherited -que.. . -que, the retention of which in poetry was no
doubt assisted by Greek TE . . . TE. Finally there is the descriptive phrase
substituted for a proper name that was such a feature of Latin poetic
discouISe, though not always as unambiguous as it is here and in Virgil,
A . 1.254 above. All these details combine to distance the sentence from
normal prose usage, e.g. galea deposita, Mars patrem deorum hominumque
allocutus est.
24.1. Some poetic paraphrases are metrically constrained, like ter quattuor,
bis senos for duodecim (Em. Ann. 88 Sk, Virg. E. 1.43) and other numeral
substitutes. No such problem was posed by caseus, but cheese, like uncles,
seldom needed to be mentioned in poetry. An exception is supplied by
Virgil.1 In E. 1.34, pinguis et ingratae premeretur caseus urbi, the perennial
down-to-earth complaint of the farmer against the city, an important motif
of the poem, places the prosaic word in a carefully crafted alliterative
‘golden’ line. Later in the poem in an inviting rustic bill of fare we find a
more conventional poetic periphrasis, mitia poma, / castaneae molles et
pressi copia lactis (80-1). The contexts dictate the variation.
24.2. At the start of Anchises’ cosmological lecture (A. 6.725) lucen-
temque globum lunae Titaniaque astra, the periphrasis for solem, itself of
course not unmetrical, not only counterbalances the description of luna
but also alludes in a demythologizing context to the older religious view
that made Helios the son of the Titan Hyperion. This kind of periphrasis
is an important didactic vehicle in the Georgics, continually associating
the old mythology with science as complementary elements in a unified
world view. So we get fratris radiis obnoxia . . . Luna and Tithoni croceum
linqwns Aurora cubile (G. 1.396,447). These periphrases contribute more
than embellishment to their context (see 0329-30).
25.1. Various lexical formants can be singled out as to some extent charac-
teristic of the poetic register. The first is the diminutive suffix, which has
an interesting role in the history of the lexicon as a Always an
ingredient of the colloquial register, some diminutives came to replace
their base forms in late Vulgar Latin, e.g. agnellus, masculus, uetulus.
Others, like anulus, calculus had already replaced their base forms in early
s8 See Ernout (1947 67-9).

60 R. G. G. Coleman ~
Latin. In the case of uitulus, ancilla the bases themselves had disappeared,
though anculus is reported by Festus without attribution. Others like bellus,
flagellum, osculum and ampulla had become distanced from their bases
It is doubtful whether any of these should be classed as live diminutives
in the classical period. The diminutive form could be used in both a literal
sense -castellum,sacellum, tabella etc. -and an emotive sense, connoting
affection or contempt. The latter is a feature of satiric writing, e.g.
deteriore . . .forma muliercula (Lucr. 4.1279) and furtiuae . . . aurum /pelli-
culae (Juv. 1.11) of the golden fleece. Its colloquial associations kept it
from the higher genres of poetry and of prose. It may be that the epic
preference for uirgo dates from a time when puella was still felt to be the
diminutive of an obsolescent puera. ~
25.2. Lucretius has examples of diminutives used referentially; parti-

cula, tantulus etc. were obviously useful for discussing atomic theory. In
2.1534, nec singillatim corpuscula quaeque uaporis / sed complexa meant
inter se conque globata, we have besides corpllscula the old adverb based
on the diminutive of singuli and the archaic tmesis of conque globata, both
metrically necessary variants, and the epic archaism meant: in all a typical
sample of the poet’s purest didactic style. Sometimes the diminutive form
has technical connotations, as uitis . . . nouellas (Virg. E. 3.11); cf. uineas
nouellas (Var. R. 1.31.1). On the other hand both literal and emotive
senses seem appropriate to many of the Bucolic contexts e.g. gemellos
(1.14), agelli (9.3), capellae (10.77).
25.3. The purely emotive use was especially associated with the dis-
course of love and already a target for satire in Plautus (As. 666-8).
However Catullus does not restrict diminutives to personal erotic like
turgiduli. . . ocelli (3.18) or erotic satire like rosea ista labella (80.1). He
has a number of them in poem 6 4 Ariadne on the shore maestis Minois
ocellis prospicit (60-l), the grim detail in the picture of the Parcae,
laneaque aridulis haerebant morsa labellis (316). This neoteric initiative in
bringing elements of the lower genre into at least the more abbreviated
forms of epic composition was not developed by the poet of the Aeneid
or his successors in the epic genre. This may be due not to stylistic
disdain - Virgil was content to include labrum and auunculus - but to
the fact that epic has few occasions when diminutives, whether literal or
emotive, would be appropriate. Where they do seem appropriate, Virgil
does not hold back. The poignancy of Dido’s lament in A . 4.328-9 is
deepened by the diminutive: si quis mihi paruulus aula / luderet Aeneas . . .I.
26.1. If the diminutive suffix was fed from colloquial usage, the use of
compound words has different orientations. Quintilian, after quoting Pacu-

Vius, repandirostrum incuruiceruicum pecus, concludes his discussion of

sed res tota magis Graecos decet, nobis minus succedit
(1.570).‘Ihis observation accords with Livy’s comment on the word andro-
gyni (27.11.5), uulgus, ut pleraque, faciliore ad duplicanda uerba Graeco
SermOne appellat. Latin was certainly not a compounding language compar-
able to Greek, Sanskrit or modem German, but the process was already
established in common Italic. Compounding simple verbs with preverbs, as
in aduenio, anticipare circumferre,praeterire, was an ancient and expanding
formative process.
2612,Compound nouns and adjectives reflecting various kinds of predi-
-
cate determinatives like princeps (c€ primum capere), possessives like
nundinum (c€ nouem dies) etc. - already existed in Latins9and provided
a precedent on which Greek influence could expand. Thus in the areas of
trade ?andcraft, law, politics, war and religion we find, for instance, sed@-
care, artifa, auspex, biennium, iudex, mancipium, sacerdos and sestertius,
as well as some, like hospes, manifestus and uindex, whose analysis was
obscured by the passage of time. Semantically transparent compounding
of this sort remained a property of technical registers, where it received
additional stimulus from Greek models, especially in philosophy, science
and linguistic studiesmThus Celsus, a purist in Latinity, willingly employed
Greek loan-words like cataplasma, emplastrum, habrotonum and leth-
argicus as well as calques like auripigmentum, exulcerare, febricitare.
26.3. Many poetic compounds are clearly formed on Greek models. In
epic and tragedy Homer, Pindar and the choral odes of tragedy were
influential, and the spread of compounding into Hellenistic epigram, most
notably Meleager, accounts for its presence in Augustan elegy. Genre is
important. Horace (AR 93-98) sees sesquipedalia uerba as the norm for
tragedy, from which a Telephus or Peleus might depart si curat cor spec-
tanfistetigisse querela, while on the other hand the tumidum os of an angry
Chremes represents a departure from the norms of comedy. The frequency
of compounds in the surviving fragments of tragedy - beniuolentia,miser-
icordia, uitisator (Accius), caprigenus, grandaeuitas (Pacuvius) etc., most
of ithem metrically impossible for epic (see 95.1), confirm the realism of
Horace’s precept.
26.4. Many of these early compounds were calques on Greek originals.
Thus Ennius’ altitonam is modelled on & , h p p ~ p i ~ q frusifer
s, on Kaprro+6pos.
Sometimes the relation is more complex. Thus magnanimus in a Plautine
parody of the elevated style (Am. 212) seems to be modelled on pqdOvpos
59 See Bader (1962). More recently there has been intensive study of Latin word formation,
especially in France, reflected in the numerous writings of E Biville, M. Fruyt and C. Kircher.
On which see Coleman (1989).

R. G. G. Coleman
but with a native stimulus in the phrase magno animo; Ennius’ caelicolae
on o;pav[wves, assisted like Naevius’ siluicola by native incola; Lucilius’
grandaeuus certainly on paKpalwv. Livius however chose uersutus for wohd-
T ~ O ~ Orather
S than multimodus (cf. multimodis, the substantival ablative
used adverbially).
26.5. The process of compounding in poetry became cumulative. Lucre-
tius follows earlier poets in using frondifer (Naev.), laetificus (Enn.), Lucifer
(Acc.) etc., but has also the earliest attestations of suauidicus, montiuugus,
siluifragus, turicremus etc. Though first attestations can be misleading (cf.
$2.5), a diachronic pattern of retention and innovation is unmistakable.
Thus omnipotens is first attested in Ennius and Plautus and may well
be independent of w a y ~ p i ~ rignipotens
)~, does not appear before VirgiI,
auricomus, calqued on X ~ U U O K ~ ~ is~ ) first
S , found in Virgil, aurifuus not
before Prudentius; suauisonus, calqued on rjSd+wvos, is found as early as
Naevius, dulcisonus not until Sidonius Apollinaris. Because of the domi-
nation of dactylic metres Livius’ quinquertib, calqued on wivTadhos, and
inzimrgdre as well as eccentricities like Catullus’ lasarpicifer were not taken
up. In general the most productive classes were the determinatives, with
verbal second components - laniger, armisonus, belligerens etc. - and,
some way behind, the possessives - longaeuus, aequanimus etc.
26.6. Sometimes, inevitably, the compound becomes a clichC and its
semantic distinctiveness is eroded.6l Accius’ quadrupedantum sonipedum
(tr. 603) is the first attestation of the calque on Hesiod’s Kavapjwovs. The
aural image is clearly prominent, as it is in Virgil’s stat sonipes ac frena
ferox spumantia mandit ( A . 4.135). But in Valerius Flaccus’ quemque suus
sonipes . . .portat (1.431) the word has become just an anapaestic synonym
for equus, a piece of ‘poetic diction’ in the pejorative sense of the phrase.
The same is even more true of the choriambic quadrupedes.
26.7. The poetic associations of compounding can be confirmed
indirectly from two sources. The first is its occurrence in Plautus, who like
Aristophanes employs it for comic and specifically parodic purposes. Some
examples, like magnanimus, caelipotens, are simply lifted from contexts of
high seriousness and survive in that tradition; others, like multibiba, unoc-
ulus, turpilucricupidus, reflecting the formation of nicknames like the
cognomen Crassipes, do not outlast their immediate comic role.
26.8. The second source is the effect of Occurrences of known poetic
compounds in prose contexts. Thus suauiloquens is attested first in Ennius’
praise of Cethegus (Ann. 304 Sk) and recurs appropriately in Lucretius’
suauiloquenti carmine Pierio (1.945-6). However, it is also used once in
classical prose by Cicero, who cites the Ennian passage for Cethegus’
61 What follows is developed from Ernout (1947: 56).

suauiloquentiam at Brut. 58; in Rep. 5.11, a certain suauiloquens iucunditas

is attributed to Menelaus. Far from being diminished here, the poetic
connotationsare especially apt for Cicero’s context: an epic epithet for an
epic hero. Another example: in Tusc. 5.79, non montiuagos atque siluestris
cursus lustrationesque patiuntur? Cicero takes over montiuagus from Luc-
retius, who uses it, as Cicero does, in analogies between human and
animal behaviour (1.404, 2.597). But the poetic tone is again stylistically
appropriate, since the context is the emotional conclusion of the Stoic
argument, and the whole clause forms a rhythmic and grandiloquent finale
to a tricolon. Finally ueridicus, which first occurs in Lucr. 6.6 and Cat.
64.306, is again taken up by Cicero, ueridicae uoces ex occult0 missae esse
dicuntur (Diu. l.lOl), and then by Livy, ueridica interpres deum (1.7.10),
where the poetical tones are appropriate respectively to the antique tale
of Faunus’ prophecies and to Evander’s account of Hercules and Cacus.
Subsequent attestations in the elder Pliny and Martial suggest that the
word soon lost its high poetic status, but its effectiveness in Cicero and
Livy still depends upon that status.
26.9. Compounds were also, as we noted in 026.2, a feature of technical
registers. Lucretius, who was cited for poetic compounds, also exemplifies
the technical usage, as one would expect of a truly didactic poet. Thus ex
alienigenis rebus constare (1.865), genitales auctifcique motus (2.571-2),
sensifer unde oritur. . . mutus (3.272), modis multis multangula (4.654).
The patterns of formation are identical with those in 026.5, and this is true
for the scientific, technological and learned compounds generally, as
witness the following, taken from the pages of Pliny’s Natural History:
aquifolia, the hybrid aurichalcus, internodium, lapicidinae, multiformis, pro-
pottio, saxifragurn, triangulus, unicolor and uitifer. In both registers the
intention is the same, to present an object or concept in clear and eco-
nomical terms. The difference is that the technical compounds, recurring
in contexts where their effectiveness depends upon univocal precision, do
nat have and must not have the evocations and allusiveness that come
from the succession of emotive contexts in which poetical compounds
recur and on which their effectiveness depends.
27.1 We have already noted ($15) the phonetic contribution of Greek

words to Latin poetic contexts, but of no less importance is their semantic
input to these contexts.
27.2. Macrobius notes (Sat. 6.4.17) lychnus as a graecism used by
Ennius, Lucretius and Virgil. Attestations in Statius confirm the poetic
profile for the word. But in contrast to lucema, lantema etc. it is used of
hanging lamps, and as the regular word for this luxury item of furniture it

64 R. G. G. Coleman
is found in Cicero, Cael. 67. The association with grandeur and wealth
gives the word its passport into epic, but it is not in itself poetic. Virgil's
fondness for Greek words for drinking vessels is also noted by Macrobius,
who cites (Sat. 5.21.1) cantharus, carchesium, scyphus, but what these words
tell us is not that the poet preferred Greek names to Latin ones but that
he preferred the imagistic detail of Greek utensils. In this he was merely
extending the precedent of amphora, cadus, crater etc.
27.3. Many Greek words had in fact become so familiar in Latin as
to have almost surrendered their foreign connotations. Thus balineum,
bracchium, carta, machina, mama and purpureus had undergone Latin
sound changes, ampulla, gubemator and spatula had acquired Latin suf-
fixes. The entry status of aer, barbaricus, corona, dracuma, ostreum and1
stola, all attested already in Ennius, cannot be determined, but they cer-
tainly have no specifically poetic connotations in classical Latin. When
Horace, characterizing spring, writes trahuntque siccas machinae carinas
(C. 1.4.2) it is precisely the unpoetic character of machinae, characteristi-
cally juxtaposed with the poetic synecdoche of carinas in an alliterative
sequence that injects the workaday detail into the image. By contrast in
C. 1.1.29-34 the counterpoint of Greek and Latin words brilliantly enacts
the synthesis of the two poetic cultures that Horace is proclaiming as his
vocation. Thus the hederae that crown learned brows place him among dis
superis, the Nympharum . . . cum Satyris chori are what secemunt populo,
Euterpe does not constrain the Latin tibias nor Polyhymnia refuse to tune
Lesbbum.. . barbiton. The hybrid phrase lyricis uatibus (35) sums it all
up, and the self-mocking deference of the conclusion does nothing to
destroy the effect.
27.4. Apart from Greek words there is little evidence of direct bor-
rowing into the poetic register of words from other foreign sources. There
is ancient testimony to Italic origins for casczu, crepusculum, dirus and
famulus, to which we can add bos and the very unpoetic multa; from
dialectal Latin testis, uerna and possibly sol. None of these is likely to have
entered Latin through poetry.
I
28.1. It is of course proper nouns and their derivative epithets that are
most often used for poetic effect, above all those that relate to mythological
incidents and geographical locations. i
28.2. Propertius boasts (1.9.5) that not even Chaoniae . . . columbae can
surpass his power to prophesy in amorous matters. The phrase also occurs
in Virgil, E. 9.13, and refers with conventional obliquity to the oak-tree
cult of Zeus at Dodona and the oracular columbae ( ~ ~ k t d associated
6 ~ ~ )
with it. But columbae are after all Veneris dominae uolucres, mea turbu

(3.3.31);so they have a special relevance, the poet would have us believe,
to erotic prophecy. Learning and wit are the characteristics of such mytho-
l o g i d allusions
28.3. In the much discussed opening similes of Propertius, 1.3, the three
sleeping beauties are given identifymg epithets. The local epithet Cnosia
(3), juxtaposed with dese& recalls the treachery to her family, now pun-
ished by the treacherous Thesea carina (1); Andromede (4), with the exotic
colour of the Greek nominative, needs no further specification and the
patronymic epithet Cepheia alludes to the story of her sacrifice; Edonis
( 5 ) identifies the third woman as a member of the Thracian tribe famous
for Bacchant worship, or as a Thessalian devotee behaving like one beside
the river, Apidano (6). In each instance the actual descriptive details look
back)to events before the sleep - desertis. . . litoribus, libera. . . cotibus,
fessa choreis -so that the triad form a poetic argument fashioned in myth
and imagery but highly relevant to the following autobiographical scene.
This use of a group of mythological exempla, already employed in 1.2.15-24
occurs elsewhere in Propertius, e.g. 1.13.21-24.
2&4. An interesting example of a Graecism used apparently to distance
the familiar comes in Vkg. A. 8.72, tuque o Thybri tu0 genitor cum flumine
sancfo, a line which we learn from Macrobius (Sat. 6.1.12) is adapted from
Ennius’ teque, pater Tiberine, tu0 cum Jtumine sancto (Ann. 26 Sk). Having
referred to the god of this Etrusco-Latin river as d e w . . . Tiberinus (31),
Virgil subsequently has the god identifymg himself as caeruleus Thybris,
caelo gratzksimus amnis (64). The Hellenized version of the name (for the
more usual Greek @6p&ts) is set between a very poetical colour adjective
and an allusion to its supposed etymology. The exotic form of the name
is anhanced by the choice of the Greek vocative case, adding the majestic
genitor in place of pater, which was the normal address even in Latin
prayars. The tribulations of Aeneas, Laornedontius heros (18), like those
of Priam, Laomedontiades (158), are the penance that Laomedon’s
descendants must pay for his impious crime (cf. 4.542). Now suddenly a
potentially hostile river god with a Hellenized name announces the pres-
ence of friendly Greeks at the site of Rome and an offer to conduct the
harassed Trojans to unexpected allies Thybris is after all propitious in its
distancing.
28.5. Similarly significant is the epithet in Horace’s C. 3.29.1. Tyrrhena
regum progenies has a grandeur above the more matter-of-fact address,
Maecenas atauis edite regibus, in 1.1.1. Again the Etrusco-Greek form is
preferred to Tuscus or Etruscus, which is used of the mare Tyrrhenum
later in the ode. Its transfer from regum to progenies gives the phrase a
more characterizingforce compared with 1.1.1, but the form of the epithet
well characterizes the Hellenophile Etruscan. Appropriate too is the sybar-

66 R. G. G. Coleman
itic Greek detail of pressa tuis balanus capillis (4), /3Mavos amidst the
homely Latin hospitality promised. The warning that follows against
the perils of fashionable lhsculum is wittily presented in the learned
allusion of Telegoni iuga parricidae (S), which we can set beside the
Circaea.. . moenia of Epod. 1.30. Three illustrations in two stanzas of
the allusive potentialities of Greek words in Latin contexts.
28.6. A frequent and colourful use of Greek names, whether of natural
phenomena or of mythological persons, is to form geographical peri-
phrases. In Catullus, 64.3, the Greeks sail not to Colchis but Phasidos ad
fluctus et fines Aeeteos. The river Phasis, provided with a Greek case form,
and the derivative noun from the name of Medea’s father, with its unLatin
spondaic vocalism, provide appropriate exotic colour and enclose an old-
fashioned alliterative phrase in which fluctus ‘waves’ suggests a substantial
flumen. The periphrasis adds grauitas to the remote oriental region, but
before readers can appreciate this, they do need to know what the refer-
ences are. There is thus a loss in accessibility.
28.7. Catullus at the end of 66 uses the form OdrBn, echoing Callima-
chus’ ‘Qaplwv (H. 3.265), appropriately, since the poem is an adaptation
of the Greek poet’s Coma Berenices. The short iota provides respectable
precedent for modifications of OriOn, which is not in itself metrically
difficult. So Virgil has nimbosus &fOn (A. 1.535), reminding us of the
importance of the heavenly bodies as clock, compass and calendar in
the premodern world. Much that strikes us as esoteric astronomical
learning would have been less unfamiliar to ancient readers, Quint. 1.4.4
notwithstanding. Similarly the identification of the constellation with the
mythical hunter giant would have been familiar to educated readers; and
Horace can cite as an exemplum of uis consili expers (C. 3.4.65) notus et
integrae / temptator Orion Dianae (7G1). The ancient myths provided an
accessible store of paradigms of good and ill, appeal to which was a
characteristic of Greek and thence Latin poetry. Accessibility could be
endangered by the tendency to make the allusions more and more oblique.
But Greek words certainly had far more than the ornamental effect noted
in 015.3.
28.8. Not all allusive epithets are mythological. In C. 3.5. Horace pre-
sents Regulus as an exemplum of self-sacrificing patriotism, following
Pindar’s paradigmatic citation of the exploits of Greek mythical heroes.
The poem ends with the depiction of the Roman hero in happier times
tendens Venefranos in agros / aut Lacedaemonium Tarentum. It is not
known whether Regulus had ancestral estates near the Samnite town and
what associations he had with Tarentum (nearby Brundisium had been the
site of his earlier consular triumph). But Lacedaemonium alludes to
the Spartan foundation that Tarentum had long claimed. It evokes the

Greek state with which traditional Romans preferred to identify their

own civic ideals6’ specified by Cicero, Flacc. 62-3, but only after his own
Athenhl preferences. Moreover, if Livy’s belief (38.17.12) that contem-
porary Tarentines retained little of the proverbial Spartan qualities was
widely,shared, the phrase mirrors summarily the decline from Regulus’
Rome to the Rome castigated in this and other Horatian ‘laureate’
odes.
28.9. The next ode provides an important example of a non-Greek
allusive epithet. Horace contrasts with the degenerate Augustans (3.6.17f€,
45ff.) those who had made Rome great, rusticoncm mascula militum /
proles, Sabellis docta ligonibus / uersare glaebas et seuerae / matris ad
arbitrium recisos /portare fustis (37-41). The homely unglamorous imagery,
created out of very prosaic vocabulary, evokes a way of life remote from
Augustan urbanitas and luxuria; but it is rural Italy, not rural Latium that
is extolled. The synecdoche in Sabellis may have a subversive edge when
used to extol traditionally ‘Roman’ virtues. For the epithet refers strictly
to non-latin speaking, especially Oscan, regions of Italy (cf. Sabellus ager
in Liv. 8.1.7). We can compare his fellow Italian’s pubem. . . Sabellam (G.
2.167).
29.1. The allusiveness that we have observed constantly in the preceding

paragraphs depends upon readers learned enough especially in Greek
poetry to respond to the references. This is why it is particularly associated
with the higher genres of epic, didactic and public lyric. By contrast
metaphor and other lexical tropes of semantic transfer,63which are, with
the creation of imagery, the most powerful and distinctive of the poet’s
conimunicative vehicles, yield their effect to anyone who knows the lan-
guage and has imagination. Various tropes employing semantic shifts away
from the current meaning of a word are widely attested in poetry. The
most prolific is metaphor, which, at least since Aristotle’s Poetics64 has
been recognized as the most important of all the poet’s verbal skills.
Aristotle also remarks in Rh. 3.2.5 that everyone uses metaphor as well
as,literal meaning - K d p l O V KaL( ~d O ~ K E ~ O-
V in ordinary discourse, resorting
to,metaphors in order to repair deficiencies in the lexicon but thereafter
See Rawson (1969: 99-106).

This area of lexical usage has been much discussed in recent years both by literary critics
and theorists and by linguists; e.g. Nowottny (1962), Henry (1971), Ortony (1979), Lakoff and
Johnson (1980), Levinson (1983 esp. 14762).
FA The most important Graeco-Roman discussions of metaphor that survive are Arist. PO.
22.1458b-59a, Rh. 3.2.6-15, 1404b-l405b, Cic. De or. 3.155-69, Longin. Sublim. 32.1-7. The
examples cited and the discussions of them all repay close attention.

68 R. G. G. Coleman
retaining them for ornamentation. To quote Cicero (De or. 3.155), uerbi
translatio instituta est inopiae causa, frequentata dele~tationis.~~
29.2. The ornamental function of metaphor, which gives pleasure to
readers, does not displace other poetic functions. By suggesting a similarity
between dissimilars, especially between the less familiar, and the more
familiar, authors are able to convey to their readers new perceptions of
the world and to move them by associations thus made, while delighting
their imagination. Aristotle stresses that metaphor is of great effect in
both prose and poetry (Rh. 3.2.7), and Cicero, whose concern is the
education of orators, nevertheless chooses examples of metaphor and other
semantic transfers from poetic as well as prose writings. I
29.3. The use of metaphor to supplete a limited vocabulary is well

illustrated in the linguistic habits of children and remains, as Aristotle
noted, a feature of adult discourse in all languages. Along with neologisms
and loan-words metaphor and other transfer tropes provide the tools for
dealing with new objects, experiences and concepts and are the major
vehicle of semantic change in the lexicon. Sometimes the shift is permanent
and the original sense is lost. The Romance reflexes of caput are almost
wholly metaphorical, cap0 di famiglia, Cap0 dell0 Stato, chef de cuisine,
chef d’oeuvre etc., while the original meaning has been replaced by reflexes
of testa, the slang use of which in turn completely ousted its older meanings.
There are numerous parallels in Latin, e.g. animus *‘breath, wind’, audax
*‘insatiable, eager’, scire *‘to cut’ etc. I
29.4. It is not always easy to ascertain the direction of metaphoric shift
without an abundance of documentation for the periods in question. Of
the three instances of rustic metaphop adduced by Cicero (De or. 3.155),
gemmare uitis, luxuriem esse in herbis and laetas segetes two are misleading.
For gemma originally meant ‘bud’, as in Cato, Agr. 42, and it is ‘gem’ that
is the metaphoric extension, first attested for us in Cicero’s own Verr. 4.39,
The original meaning of l a e m was ‘lush, sleek, in good condition’, as in
Cato’s contrast between agro laeto and agro sicco (Agr. 61.2), so that the
more familiar meaning is again the metaphoric one.
29.5. This is also true of felix, cognate with fecundus, felare, femina and
meaning ‘fruitful’, as in Cato’s definition (Festus Sl), felices arbores quae
fructum ferunt. Only luxuries really belongs here, being used of plants
running wild and of uncontrolled growth. Virgil’s si luxuria foliorum exub-
erat umbra (G. 1.191) exploits the extension to human behaviour as a term
of moral disapproval. The tree is squandering its resources; the harvest of
For an important discussion of De or. 3.155-68 see Fantham (1972 176-80).

cf. the rustic metaphors like Y&Y T& 4.~6 and &p& C ~ V U Ccited
, in the Homeric Scholia
(Iliad 23.598, ed. H. Erbse).

nuts will suffer. It is characteristic of a doctus poeta in any language to

exploit semantic layers in the history of a word. Virgil surely had in mind
too Dido’s childlessness when he gave her the epithet infelix (c€ 4.68 with
595 etc.). In such words the original meaning may lie dormant for a period,
to be reactivated by a poet confident in the knowledge that readers will
be familiar enough with earlier literature to appreciate that what to the
uninitiated can only appear as a metaphor is in fact a semantic reversion.
29.6. Many other words retain their original sense while acquiring
various metaphoric uses permanently. Thus OE heafod has retained its
anatomid meaning along with a host of extensions, which fill over
eighteen columns of the Oxford English Dictionary. Similarly in Latin we
have clarus ‘loud’ etc., comprehendere ‘to take hold of etc., uirfus, ‘manli-
ness’ etc.
29.7. Such metaphors affect the semantic fields of words in a permanent
way. For this reason they are conspicuous in scientific and technical regis-
tersjNew inventions and discoveries demand new linguistic resources. One
has only to think of the vocabulary of modem computer technology -
gateway, menu, mouse, sofhuare, virus, to back, go down etc. Latin architects
gave new metaphoric senses to existing words, like ordinatio, membrum,
principium, reticulatus, surdus. What is characteristic of scientific meta-
phors is their permanence and univocality. Once the new meaning has
been assigned, it remains purely referential within the register; there is no
ornamental function, no contextual variation and no emotive accumulation
from previous occurrences. This univocality is far removed from literary
metaphor, which is very often a nonce phenomenon; no one since Macbeth
has called life a walking shadow except with reference to the famous
soliloquy.
B.8. There are on the other hand many poetic metaphors that, once
invented, are taken up and extended by later authors. Macrobius (Sat.
6.4.3) cites Ennius’ et Tiberis flumen. . . uomit in mure salsum (Ann. 453
Ski) and Virgil‘s mane salutantum totis uomit aedibus undam (G. 2.462).
The image of a river god spewing water into the sea is certainly not
ornamental, and the later poet has exploited the unattractive image in a
context of repugnance, using a chiastic word order around uomit to
enhance the image. Here, unlike most reworked metaphors, the poetic
effect does not depend very much on the reader’s recognition of the
original source.
I
>
30.1. Scientific and poetic metaphors naturally appear side by side in
didactic poetry. To the scientific category belongs Lucretius’ elementa
(1.827), calqued on m o q c i a . The aptness of the metaphor, for which Cicero

70 R. G. G. Coleman
(Ac. 1.26) seems to claim the credit, is underlined by the use of the word
in its literal sense ‘letters’in the course of an analogy with the atoms out of
which things are composed (1.197). Another Lucretian calque is simulacra
(1.1060) for dSwha, the Epicurean term for the images given off by objects,
which account among other things for simulucra in the ordinary sense,
‘ghosts’ (1.123). The use of materies for 137 is especially appropriate to
Epicurean metaphysics, given its usual sense ‘building timber’ and the
etymological connection with mater (see 330.6).
30.2. More poetic is, for instance, the use of lacessere of colliding atoms
(2.137), a personification, since the word was originally used of challenging
to a contest. A particularly Latin source of such metaphors is the legal
and political registers. Thus Horace declares (C. 3.29.54-6) his choice of
Poverty rather than dependence on Fortune in the legal terminology of gift
repayments, resign0 quae dedit (sc. Fortuna), and marriage contract, pro-
bamque / Pauperiem sine dote quaero.
30.3. But some of the most striking examples occur once again in
Lucretius. The concept of natural law is presented anthropomorphically
(1.76-7) in terms of the definition of constitutional power: refert nobis. . ./
.. .finita potestas denique quoique / quanam sit ratione (cf. infinita potestas
granted to Pompey: Cic. Agr. 2.33). This metaphor, combined with the
qu alliteration that is not uncommon in legal phraseology, is followed
immediately by another, taken from land tenure,. . . atque alte terminus
haerens. Nature assigns powers and fixes their limits. The two metaphors
come towards the end of a passage (63-79) rich in metaphor - the
crushing weight of the monster Religio, Epicurus’ breach of the urta.t/
Naturae. . . portarum claustra and advance beyond the fimmantia moenia
mundi, the outcome of his victory, which nos exaequat . . . caelo, and much
else. The powerful appeal to imagination and emotion well illustrates how
much more than mere ornament metaphor can bring to a poetic - and a
proselytizing - context.
30.4. There are references elsewhere to foedera Naturai (1.586),
Nature’s treaties, the pacts that define the relations between different parts
of the universe. The use of concilium to depict the combinations of atoms
to form sensible objects was no doubt inspired by Epicurus’ dldpompa, but
the Latin term has much stronger political and constitutional connotations
The term is introduced in a passage (1.182-3) that itself illustrates the
difficulties than can arise in identifying a metaphor: primordia quae geni-
tali / concilio possent arceri. The verb ordiri is used by Pliny (Nut, 11.80)
of a spider weaving its web, and this may be the original sense of the verb,
which would make Pacuvius’ machinam ordiris nouam (tr. 379) a bolder
metaphor than it seems. It would also make Lucretius’ primordia and
exordia not ‘initial particles’ but ‘threads’ from which the world’s fibres

are woven. P b y again (Nut. 7.61) has the phrase projluuium genitde of
menstruation,SO it is possible that a notion of biological reproduction was
prominent in Lucretius’ use of genitalis. The notion that atoms could be
held together in a ‘generative council’ is certainly strange, but no more so
perhaps’ than, say, creatrix Natura (1.629) or daeaizla Tellus (1.7). Stoics
could indeed define Natura as that which contineat mundum omnem
eumque tueatur (Cic. Nut. 2.29) and so provide a context for Nature’s
creativity in a literal sense. But its introduction even metaphorically in an
Epicurean poem seems bold.
305. The personification of Natura belongs with a number of other
metaphors that exploit traditional religious concepts. Especially notable is
the reference to Terra, who has earned her maternum nomen when she
genus. . . creauit / humanum and animal. . . fudit / omne quod in magnis
bacchatur montibu(s) passim (5.821-5). In an author not given to using
Greek words bacchatur, like daedala Tellus, has a distancing effect, but
the emotive power of Mater Terra, like that of Pater Aether, with whom
she is coupled in 1.250-1, is great and helps to conceal the awkward gap
in the rational argument between inanimate and animate modes of being.
The poet again comes to the philosopher’s rescue.
31.1. Personification is of course a form of metaphor more easily available

in languages that assign animate gender to inanimate objects or abstrac-
tions. Hence also the easy deification of X p p o v l a , Fortuna etc. Timol; Minae
and Cura are all personified in Horace, C. 3.1.37&, where their appearance
constitutes the poet’s warning to the dominus.. . terrae fastidiosus: sed
Timor et Minae / scandunt eodem quo dominus neque / decedit aerata
triremi et / p o s t equitem sedet atra Cura. Fear is personified also in Virg.
A. 9.719, atrum. . . Timorem, and Phobos son of Ares already appears in
Homer and Hesiod (Zl. 4.440, Theog. 934). Timor like Cura resides in the
mind of the rich owner, the Minae, which are personified here perhaps for
the first time, inspire that fear from without. So Timor et Minae is almost
a hendiadys.
31.2. The personification of Cura itself is found elsewhere in Augustan
poetry, e.g. Virg. A. 6.274, ultrices.. . Curae. However, the metaphorical
animation of Anxiety already occurs in Theognis 729, @ p o v ~ B ~ .s.. m c p h
X O L K ~ &ovaat,
X which is echoed by Horace in C. 2.16.11, Curas laqueata
clrcum tecta u0Iantes.6~Later in the same ode (21-2) there is an interesting
anticipation of the present passage: scandit aeram uitwsa nauis / Cura nee
67 Here, as usual, the commentary by Nisbet and Hubbard (1978) is illuminating.

72 R. G. G. Coleman
turmas equitum relinquit, where uitiosa suggests both the diseased mental
state and, within the trope itself, the corrosive effect on the metal.
31.3. The eerie presence of the trio is adumbrated in just a few descrip-
tive details. The vigorous scandunt mocks fustidiosus, and this is
emphasized by the repetition of dominus. More static images are presented
in neque decedit.. . e t . . . sedet.. . . Both triremi and equitem are usually
taken to refer to the rich man’s expensive recreations, sailing and horse-
riding. This may well be true of equitem, though Horace is perhaps
exploiting the ambiguity of the word to make the point that those in the
top income bracket are as much at risk as cavalry officers. The nautical
image seems more distinctly military: triremes were usually warships and
aerata ‘armoured’is a familiar epithet of naues longae (e.g. Caes. Civ. 2.3.1;
cf. ratem aeratam, attributed to Naevius in Var. L. 7.23). The verb decedere
has strong military connotations also, of retreat (Caes. Civ. 1.71), retire-
ment from active service (Liv. 41.10.7) and even desertion (Cic. Sen. 73).
Cura is continually on active service aboard expensive warships, whatever
purposes they are being used for. Finally atra, in implicit visual contrast
with aerata, yet invisible to the knight himself, suggests, as nigra would
not, a baleful presence uninterrupted and unseen. Timor, Mime and atra
Cura enclose the whole passage.
32.1 Of the other tropes of semantic transfer metonymy and synecdoche

in particular merit attention here. Metonymy like metaphor plays +in
important role in the lexical history of a language, as for instance ago,
ciuitas, dies and res all illustrate. It is most frequently seen in the shift
from ‘verbal action’ to ‘concrete effect’ in the meaning of nouns like
comitium, legio, natura, where the original meaning was displaced, apd
gaudium, oratio, cultura, where it was not. The semantic shift was particu-
larly clear in the plural forms (as was the separate and less remarkable
shift from generic or collective to specific or individual in nix, fnrmenfurn,
aes etc.).
32.2. In literary usage the effects of this trope are less permanent, as
in the examples cited by Cicero (De OK 3.167), curia for senatus, campus
for comitia, arma and tela for bellum, toga for pax. This last example is
attested in a notorious line from his own poetry, cedant arma rogue, cgn-
cedut laurea linguae (cited in 08 1.77), with the combination of two very
Roman metonymic images in toga and lauream. The contents of Cicerp’s
list illustrate once more the common property shared by poets and orators.
The reading linguae is preserved in Quint. 11.1.24.The tradition of De officiis itself strongly
favours luudi. e \

32i3;Metonymy, like metaphor, has many uses for a poet, whether he

is writing in a high or low genre. In Horace’s Graecia barbariae lento
conlisa duello (Epist. 1.2.7) the opening words give a sense of deper-
sonalized contestants that Graeci and barbari would not (c€ Graecia but
barbaris in Cic. Ofi3.99). The already current use of barbaria in the sense
of ‘barbarity’ adds an emotive element to the traditional interpretation of
the war, whose epic status is marked by the archaic duello. In Graecia
capta ferum uictorem cepit et artis / intulit ugresti Latio (Epist. 2.1.156-7)
the contrast is between geographical possession, which comes from military
victoe and cultural conquest, which is made more threatening and per-
vasive by the replacement of Grueci by the more impersonal Graecia.
Latio for Lutinis enhances the play on the two senses of agrestis, also an
instance of metonymy.
32.4. One of the most frequent poetic metonymies is the substitution
of divine names for the object or activity of which they were in traditional
religibn the patrons. Cicero (De O K 3.167) again instances Mars for bellum,
Ceres for fruges, Liber for uinum, Neptunus for mare. A characteristic
Oviaan example is quis Venerisfamulae conubia liber inire /. . . uelit? (A.
2.7.21-2). The witty juxtaposition of famulae with the goddess’ name, used
metonymically though it is, is compounded by conubia and liber. For
although Lucretius’ conubia ad Veneris partusque ferarum (3.776), where
the metonymy identifies Venus not with the one who inspires amor but
with amor itself, offers precedent for the use of conubium for concubitus,
the presence of famula and liber recalls that conubium in the strict sense
was a legal impossibility between slave and freeborn. The adynaton would
no doubt have struck Corinna as an irrelevance.
33.1. Synecdoche too is common to all registers. For instance the use of
caput of one’s person or personal status is well established in legal termin-
ology and in prose generally, and need not be a calque on Koipa, K E + a h 7 j .
But poetry shows much bolder and more extensive use of the figure.
Nautical terminology shows a remarkable diversity of synecdoche in
poetry. Varro defines ratis as being used ubi plures mali aut mseres iuncti
aqtla ducuntur (L. 7.23). In the sense of ‘raft’ it is distinguished from nauis
in Cic. Vex 5.5, but Ennius already has it by metonymy in the latter
sense in ratibusque . . . fremebadimber Neptuni (Ann. 515 Sk); whence it
passes to classical poetry, e.g. pandas rutibus posuere carinas (Virg. G.
2.445). Horace already has carina ‘keel’ by synecdoche for nauis in C.
1.35.8, quicumque Bithyna lacessit / Carpathium pelagus carina, where the
Greek loan-word pelagus ‘sea’, first attested in Pacuvius and Lucilius, also

14 R G. G. Colemun
contributes to the alliterative effect and even perhaps to the sense of

geographical distance.
33.2. Another synecdoche is the use of puppis for nauis (cf. Quint.
8.6.20). This is already attested in Cat. 64.6, uuda salsa cita decurrere puppi,
which is also one of the earliest attestations of U& ‘shallows’, again a
synecdoche for ‘sea’. Virgil has nautica pinus (E. 4.38) for ‘ship’, where
the direction of the trope is clarified by the epithet, as it is by the context
in quos, . . infesta ducebat in aequora pinu (A. 10.205-6). In the next line
arbor for ‘oar’ is similarly disambiguated centenaque arbore fructum /
uerberat adsurgens. In A. 5.504 the reference to a mast is explicit: sagitta .. .
uduersi . . . infigifur arbore muli. In Ov. M. 11.476 arbore is similarly disam-
biguated by the following malo; but Valerius Flaccus’ cekior arbore ponfus
(1.496) no longer needs more than a general contextual support.
33.3. To revert to words used by trope for ‘the sea’: Cic. Arat. 67 seems
to be the earliest attestation of aequor without explicit specification. There
is a well-known ambiguity in Virgd’s omne tibi stratum silet aequor (E.
9.57), and the context alone disambiguates aequora in A. 10.206. The need
for specification was increased in A. 2.780, m t u m maris aequor arandum,
by the metaphoric use of arare, but it is less severe in A. 5.158, longa
sulcant w d a salsa carina, where a comparable metaphoric verb is combined
with two unambiguously synecdochic nouns. The colour noun caerulurn
(<*cael-ul-) is already used of ‘the sea’ by Cicero in FPL 29.3B, nemo
haec u m q w m est transuectus caerula cursu. The rise of this rival metaphoric
use partly explains Lucretius’ wish to specify the original meaning in
1.1090,et solis flammam per caeli caerula pusci, echoing Enn. Ann. 48 Sk.
When Silius writes sulcarunt caerula puppes (15.239) the tropic character
of the statement is less striking inasmuch as all three of the constituents
were already well established in their transferred meanings (see 029.8). It
should be noted again (cf. 029.6) that all the words cited continued to be
used in their literal sense in poetry as well as prose. Virgil could still write
an Ennian portrait of Octavian, cum patribus populoque, penatibus et
magnis dis, /stuns celsa in puppi (A. 8.680), with puppis used in its original
sense.
34.L What has been said by a great critic about metaphor applies equally
to all the tropes of semantic transfer that we have just been considering:
‘the inimitable mark of the poet is his ability to control the realization of
a metaphor to the precise degree appropriate in a given The
E R. Leavis (1948. 77).

creative mind that perceives the relevant similarity between dissimilars

also creates the context within which that similarity is sufficiently clarified.
3d2. An important consequence of the various types of semantic
transfer that we have been surveying was that poets had a large store of
synonyms for many common words.7O For mare we could add to the
examples cited already altum, fretum, gurges, pontus, profundum; for nauis
there are prora, remus, trabs, uelum. For aqua poets could use latex, liquor,
lymph@,urnor, for amor there are aestus, ardor, cura, furor, ignis etc., for
arnica we find cura, domina, era, lux, puella, uita. Many of these are
ordinary words, used in poetry sometimes with their ordinary prose mean-
ings, sometimes with special poetic meanings, like puppzk (033.3). Such too
are cedere, mortalis, mucro, ponere (for deponere). Alongside these
instances of occasional poetic diction are those words which are virtually
exclusive to poetry, such as ensis, fari, letum, meare, olle, -que. . . -que, or
unprehed forms like fessus, gradi, linquere.
34.3. Last among semantic transfers comes the substitution of plural
forms of nouns for singulars. 713 2v TOM&TOL& is recognized by Aristotle
(Rh. 3.6.4) as a feature of poetic discourse, but there is no reason to think
that the Latin usage is much indebted to GreekJl Some distinctions need
to be made: first between plurals that generally have a semantic oppo-
sition to singulars, e.g. thalami v. thalamus, and those that do not, e.g.
nuptiae (see 05.8); second between thalami, thalamus and niues, nix or
ratione$ ratio (see 032.1). Which leaves us with the difficult question of
what the distinction between thalami and thalamus etc. actually was. There
are various possibilities: an aggrandizement of the concept signified by the
singular, the recognition of a plurality of components or adjuncts in
the singular concept, an expressively heavier or lighter texture in the sound
of the plural form. But confronted with a typical list like aequora, astra,
otia, regna, saecla, no one would have much confidence in assigning mean-
ings distinct from aequo5 astrum etc. except in the relatively few instances
in which the neuter plurals appear to be collectives. Two things are certain
however: first that none of these plurals became permanent synonyms of
their singulars (in contrast to vulg. gaudia, folia etc.), occurrences are
occasional and selective; second that the trope provided metrically useful
variants (see 006.1, 11.3).
35.1. The poet's other major lexical tool is the creation of images and
imagery, not only as a source of pleasure for the reader but also as an
70 See KroU (1924 264-5), Leumann (1959 155).

7' On the poetic plural see Lijfstedt (1942 27-65, esp. 38ff.), Marouzeau (1962 221-3).

76 R. G. G. Coleman
integral part of the poetic argument itself.72 Unlike tropes, which c m

acquire the status of poetic diction if they became part of the literary
convention, imagery rarely depends on allusive vocabulary and requires
no contextual support to clarify its meaning. The vividness and emotive
power of Livy’s description of the destruction of Alba (1.29) or of the
Battle of Trasimene (22.4-6) show that imagery also has an important
function in hist~riography~~ that again makes it proxima poetis.
35.2. Horae, as we saw in 031.1, often presents his imagery with
remarkable brevity. Important here is the complementary distribution of
contrasting descriptive details. Thus in suluitur acris hiems (C. 1.4.1) the
verb implies the adjectives dura and rigida. Conversely acris implies mite-
scit not soluitur; cf. frigora mitescunt Zephyri (4.7.9). So what we get,
elliptically presented, is soluitur et mitescit acris rigidaque hiems.
35.3. The complementary distribution of epithets is well employed in
C. 2.3.9-14, which begins with the contrast between the (dark green) pinus
ingens and the (slender but thickly leaved) alba. . . populus. The trees are
then personified in forming a partnership to provide hospitality, u m b r m
hospitalem consociare amant. Personification extends to the busy stream
lympha fuga^,'^ which laborat. . . trepidare, so that what might have been
a second component of the locus amoenus becomes instead a symbol of
futile activity in a transitory world. The unease thus awakened is continued
into the conventional symposiastic imagery, where the visual focus flores
amoenae rosae is described ominously as nimium breuis, thus becoming
another symbol of transitoriness, which links up with the closing image of
sororum fila trium atra. The vocabulary of the two stanzas is rich and
variegated and the imagery that it creates in effect constitutes the
argument. 8
35A. Almost the whole argument of C. 1.9 is conducted through a
sequence of images, beginning with images of winter -visual in candidum
Soracte and siluae laborantes, tactile in gelu . . . acuto, and concluding with
the spring season of human life, dulcis amores and choreas, campus et
areae, lenes.. . sub noctem susurri and g r a m puellae risus, where the
imagery is more nominalized and the epithets more subjective. The link
between the real winter and the metaphoric spring is provided by dunec
uirenti canities abest / morosa (17-18), where canities morosa looks back
to candidum Soracte, making it into a symbol of old age, while uirentz
72 For the interrelationship between imagery and metaphor see Silk (1974) and also Fanthm
(1972), whose main concern is with Plautus, Terence and Cicero.
71 Here, as always, the literary characteristics of historiography must be taken to include
prose fiction. For imagery in literary dialogue and oratory see Fantham (1972 115-75).
74 For the ‘disjunctiveness’,as Postgate called it, exemplified in laborat lymphafugax rrepidare
riuo for riuus lympha fugace trepidare laborat, see Nisbet-Hubbard ad loc. ,

looks back to siluae that are however not laborantes, and forward to youth
(cf. uirentis.. . Chiae in 4.13.6) that is not content with the sedentary
pleasures af focus and merum (5 and 8). All this abundant imagery sur-
rounds the central message of the ode, quid sit futurum cras fuse quaerere
et / quem Fors dierum cumque dabit lucro / adpone (13-14), presented as
an old-fashioned and prosaic exhortation, in which the only trope is a
metapha from accounting.
36.1. Similes are founded on images and so on the selection and deploy-
ment of lexical meanings. This is true even of brief examples, like the
description of Pindar monte decurrens uelut amnis, imbres / quem super
notas aluere ripas, which is followed by the metaphoric feruet inmensusque
ruitprofundo / Pindants ore (C. 4.2.5-8, on which stanza see further 045.3).
Horace ,describes himself apis Matinae more modoque / grata carpentis
thyma per laborem (ibid. 27-9) in humble contrast to Pindar, Dircaeum . . .
cycnum (25). This time the accessibility of the imagery is limited by the
allusions We need to know what Matinae and Dircaeus refer to, what
the bee is doing in Tibur, and above all that Pindar used the bee com-
parison (e.g. E! 10.53-4) of his own conception of poetic composition, so
that Horace is not after all as self-deprecatory as he appears.
W Vhgil’s epic similes provide a kind of poetic commentary on the
context in which they are set, and every specific part of the image tells.
There’is nothing comparable to what we sometimes find in Homeric
similks, where a detail is included which, while it may bring the image into
familiar focus for the hearer, does not always relate easily to the context
and may even become bizarre if we attempt to relate it. A famous example
is the end of the simile describing the slaying of Sarpedon, who falls like
an oak 42 nirvs /3hoOprj, rrjv ~’017pca~ T ~ K T O V E Sbv6prs I E‘[hapov ~ ~ ~ K K E O O L
VE~~KZU ylLiiov &at (Zl. 16.4834).

36.3. Vhgil by contrast will sometimes sacrifice realism within the simile
to gain a contextual point. Aeneas’ reaction to his first sight of Dido (A.
1.496-7) is depicted in a simile intended to recall Hom. Od. 6.102-9.75 But
whereas Nausicaa seems to Odysseus like the maiden goddess Artemis at
play with the nymphs, Dido is compared to the regal Diana coming in
procession to one of her great cult-centres. Her mother is present in both
similes. Homer’s rirVer S i re +piva AVT; is very apt; Virgil’s Latonae
taciturn pertemptant gaudia pectus is passing strange. The image of the
silent heart and the unsettling temptation must apply not to Latona within
the simile but to Aeneas outside it.
’’h r an ancient view of the two similes see Valerius Probus up. Gell. 9.9.12.

78 R G. G. Coleman
36.4 Imagery is important to Lu~retius’~

not only for the relief it offers
to the austere doctrinal exposition. It is also appropriate that the exponent
of a materialist philosophy should appeal to the senses in order to instruct
his readers as well as to delight them and appeal to their emotions. This
skill shows itself especially in places where the rational argument becomes
a bit thin, though we cannot be sure whether the change in discourse was
calculated or unconscious. For instance in 2.308-32 the poet is expounding
the doctrine that though the atoms are constantly in motion - like the
motes in the sunbeam that helped him over a similar problem in 125-8 -
the objects formed from them appear at rest. non est mirabik, the poet
says, pre-empting the reader’s reaction. omnis enim longe nostris ab sen-
sibus infra / primorum natura iacet. So the epistemological criterion of
sensation has to be rejected here, to be superseded not by iniectzu animi
but by two elaborate and highly poetic analogies, summarized in 322 and
332, before we are moved rapidly on to the next topic. The two quasi-
similes have been deservedly lauded for their vivid concentrated imagery
and contrasting details - collis and campus, lanigerae.. . pecudes and
magnae legiones, pabula laeta and belli simulacra, herbae gemmantes rore
recenti, agni . . . blande . . . coruscant and aere renidescit tellus, reptant and
circumuolitant.But the brilliance and familiarity (or rather familiarizing-
for how many of Lucretius’ readers had looked down on two armies o,n
manoeuvre?) of the imagery beguile us into forgetting that this is analogy,
not proof Atoms are like t h q if they are like this. The mellis dulcis
flauusque liquor no longer serves to disguise the absinthia taetra (1.936-8):
it conceals its inefficacy. Similes are no mere ornaments
k
SYNTAX
37.1. The meanings of individual words - their lexical stems and gram-
matical inflections - are fully articulated only in syntactic combinations,
and the repeated combinations in turn affect the meanings of words. The
syntax used by poets, including syntactic tropes, is therefore essential to
the definition of the poetic register. So too is the ordering of the words,
though in a highly inflected language this belongs not to syntax, as it does in
Enghsh, but to pragmatics and in particular stylistics. A few representative
phenomena will be discussed in the paragraphs that follow.
372. Among case uses the extension of direct object accusatives is
frequent. Cicero has canere, already old-fashioned in the sense of ‘to sing’,
with clarorum uirorum l a d e s (Tusc. 4.3) but not with claros uiros; yet
76 See West (1969) for a detailed treatment of the subject.

Lucretius writes cur. . . non alias alii quoque res cecinere poetae? (5.326-7).
Whence Horace extends to Liberum, ~ u s m Veneremque . . . canebat (C.
1.32.9), Virgil to arma uirumque cano. The verb ardere is used of emotional
states with the instrumental ablative specifying the emotion, e.g.
militibu.. . studio pugnae ardentibus (Caes. Civ. 3.90.3), and in poetry
also the object of love, e.g. arsisse Bathyllo / Anacreonta (Hor. Epod. 14.9).
But Terence, who also writes amore ardeo (Eun. 72)’ already has a direct
object in hanc ardere coepit perdite (Ph. 82). So Virgil’s ardebat Alexin (E.
2.1) is not as innovative as it might seem. Quintilian (9.3.17) takes
Tyrrhenum nauigat aequor (Virg. A. 1.67) as a Graecism. That the accus-
ative need not be perlative, as we might surmise from, say, m e . . .
nauigasse. . . per infesta latrociniis litora (Sen. Ben. 7.15.1)’ is clear from
the passive use in etiamsi nauigari posset Oceanus (Sen. Suus. 1.8).
37.3. The internal accusative function might be thought, on the basis
of the equation of duke ridentem (Cat. 51.5) with Sappho’s ychaluas I ~ ~ ~ o E v ,
to owe something to Greek influence. But the adverbs multum, parum
( < * p a w m ) and dulcius etc. together with ‘cognate’ usages like noxiam
noxit from the Twelve Tables guarantee a native origin, even if, as else-
where, Greek influence helped to maintain what might otherwise have
been a non-productive or even obsolescent usage. An interesting inter-
action of external and internal uses is Virgil’s nee uox hominem sonat (A.
1.328). For humanum with sonat would be internal; cf horrendum sonuere
(A. 9.732). However, hominem with indicat or monstrat would be external.
The syntax here, as in ardebat Alexin, reflects a shift in the semantic
orientation of the verb itself.
37.4. The pursuit of economy leads to a reduction of prepositional
phrases. Most often this results in archaism~~~ like Virgil’s Ztaliam . . . Laui-
niaqzze uenit / litora (A. 1.2-3) and the even more antique ibimus Afros
(E. 1.64), which is followed immediately by the incongruously epic rapidum
cretae ueniemus Oaxen. Not surprisingly the ablative, being a syncretic
case, offers divers examples. Thus from Propertius uaga muscosis flumina
fusa iugis (2.19.30) where the ablative indicates separation, multis decus
artibus (1.4.13) origin, contactum nullis ante Cupidinibus (1.1.1) agent, in
f a d an instance of dative-ablative indeterminacy, illa meo carus donasset
funere crinis (1.17.21) location, and medius docta cuspide Bacchus erit
(2.30.38) accompaniment. A remarkable haul, but although Propertius like
H o m e is notably bold in his case usage, each of the examples can be
paralleled widely from other classical poets. The ablative of comparison is
also employed more frequently and extended more boldly than in prose
e.g. turpior et saecli uiuere luxuria (Prop. 1.16.12), and inuidiaque maior
For the development of prepositional syntax in Latin see Coleman (1991).

80 R G. G. Coleman
urbis relinquam (Hor. C. 2.20.4-5), which clearly echoes Callimachus’

Kptuooova ,!?autcavITs (Ep. 21.4).
37.5. A poetic rival to ibit Afros for ibit ad Afros is ibit Afris. The
dative had always shared with the accusative the semantic function of
allativity, the distinction being between subjective involvement, implying
animatedness, and purely physical direction. Thus Panthoiden iterum Orco /
demissum (Hor. C. 1.28.10-ll), where the dative suggests that the obscure
underworld deity, not the underworld itself, is being referred to; c€ the ’
more physical missos ad Orcum (C. 3.4.75). Virgil’s it clamor caelo ( A
5.451; c€ tollitur in caelum clamor in 12.462) indicates that Heaven, which
in traditional culture was not after all an unpopulated region, is moved by
the shout. I
37.6. A number of distinctive uses of the genitive can be observed in

poetic discourse. Tkis had originally been the normal case of dependency
with adjectives as with nouns, but was steadily encroached upon by the
instrumental ablative. Phrases like tempus edax rerum (Ov. M. 15.234),
laeta laborum.. . Sidonia Did0 (Virg. A. 11.73-4; c€ laetus Eois / Eurus
equis in 2.417), and aeui maturus Acestes (5.73; c€ animo maturus et aeuo
in Ov.M. 8.617) are all conservative, if not archaizing.
37.7. Especially interesting is the famous integer uitae scelerisque purus
(Hor. C. 1.22.1), where the two genitives chiastically juxtaposed have
different semantic relations with the head adjectives of their grammatically
parallel phrases. For whereas integer uitue implies cui uita integra esf,
scelerisque purus does not imply cui scelus purum est, and the alternative
constructions are integer uita (instrumental) and a scelere purus (ablatival).
Nothing anomalous here: genitives of reference need not always denote
the same kind of reference. But the second phrase disambiguates the first,
which taken by itself would more likely refer to physical health than moral
goodness. Such clarification is an important function of double or multiple
descriptive phrases (c€ the similar role of the Propertian similes in 028,3).
Finally Virgd’s ereptue uirginis ira (A. 2.413) is not as strange as it has
sometimes been made out to be, since the genitive is the regular depen-
dency case in the nominalization of all predicative complements, whatever
their case; ira + genitive beside irasci + dative is paralleled in pairs like
inuidia, inuidere and usus, uti beside timor, timere.
37.8. Some uses of the genitive are influenced by Greek. While iustit-
iuene prius mirer belline laborum (Virg. A. 11.126) can be placed with the
genitive of reference that is found with pudet, piget etc., cf de impudentia
singulari . . . sunt qui mirentur (Cic. Ver. 2.1.6), this specific example is best
seen as an extension of the native idiom under the intluence of 0aupil;w
+ genitive. Horace’s desine mollium / tandem querelarum (C.2.9.17-18) is
modelled on the ablatival uses of the Greek syncretic genitive, and may

have au ad hominem point, given Valgius’ neoteric attachment^^^ The

Same is also probable in eripa te morae from the Maecenas ode, C. 3.29.5;
cf. the ablatival phrase in aegrum eripere de periculo (Vitruv. 1.1.15).
However, a datival interpretation is just possible, with morae personified,
as if Cicero had written huic me timori rather than the more normal hunc
mihi timorem eripe (Cat. 1.18). Lastly qua pauper aquae Daunus ugrestium/
regnuuit populomm (Hor. C. 3.30.11-12), which begins with a straight-
forward genitive of reference. The genitive with regnuuit could be partitive,
‘someI peoples’, but it is more likely modelled on the Greek ablatival
genitive of comparison with verbs of ruling, a syntactic reflection of the
fusioniof Italian and Greek in the h a 1 quatrain of the ode. What is clear
is that the Graecisms are no mere ornamental affectation.
38.1. ,In,the Latin verb the middle voice is reflected in deponents like
Zoquor, utor and semi-deponents like conjido, gaudeo, in none of which
does ‘it$havea distinctive meaning. A semantic distinction can however be
identified in certain uses of the passive of active verbs like induor, luuor,
mutol; reuerfor, uetor. The co-existence of doublets like urbitro/-or,
assentio/-or, comperio/-or suggests that the loss of systematic distinction
was recent.
W.All this provided a platform, if an obsolescent one, from which
to launch a revival of the middle voice. At what date the revival began is
uncertain. The earliest instance^'^ are from Plautus and Ennius: cingitur.
cerfeexpedit se (Am. 308) and indutum . . . pullam (Men. 511-12) could be
native, though the absence from early prose - unless the isolated togue
purfe caput ueluti (Cato, Orig. 1.18) is genuine - and its rareness in
classical prose is then very strange. In Ennius’ succincti cordu machaeris
(Ann 519 Sk) the presence of the Greek loan-word perhaps reduces the
probability that this is a mere Latin archaism,80while the transitive use, in
contrast to cingitur and to Ennius’ own succincti gludiis (Ann. 426 Sk),
which is a normal passive, confirms that this is a middle. But thereafter
the,construction is rare before the Augustans, e.g. Catullus’ non contectu
Zeui uelutum pectus amictu, where the enclosing word order enacts the
meaning, and lactentis uinctu pupillas (64.64-5). More striking examples
appear in the Eclogues (see 038.3).
38.3. The area of the revival was very circumscribed. The overwhelming
majority of instances are in poetry. Livy’s uirgines Zongam indutue uestem
On which see Nisbet-Hubbard (1977: 135, 148).

’’For a collection of examples see Kiihner-Stegmann (19552.1.288-90).
g,For the difficulties of classification here see Coleman (1975).

82 R. G. G. Coleman
(27.37.13), a rare prose example, can be set beside Cicero’s more typical
soccos quibus indutus esset (De or. 3.127), no less clearly a passive. More
over, the use was restricted both lexically and morphologically. Most of
the examples are with verbs of covering and putting on or removing
clothing and armour, e.g. suras euincta cothumo (Virg. E. 7.32), inutik
ferrum cingitur (Virg. A. 2.51&1) and even Horace’s alliterative laeuo
suspensi loculos tabulamque lacerto (S. 1.6.74). Conspicuous among the
exceptions are Ovid’s oculos in humum deiecta modestos (Am. 3.6.67) and
suffunditur ora rubore (M. 1.484), Virgil’s injlatum hesterno uenas, ut
semper, laccho (E. 6.15), where the Greek tone is reinforced by the
metonymy chosen for uino, and inscripti nomina regum.. . flores (E.
3.106-7), with its allusion to Greek mythology, in which the middle use
contrasts with the passive in Cicero’s sepulcrum inimico nomine inscriptum
(Dom. 100).
38.4 The examples cited have illustrated the morphological restriction.
Most Occurrences are in the perfect participle used in a descriptive rather
than narrative sense. Latin had no perfect active participle except in the
deponents, which in phrases like talia clara uoce locutus provide a syntactic
model for many of the middle uses that have just been cited. There is a
growing tendency to more frequent use of present tense finite forms.
Ovid’s suffunditur is already an extension from Virgil’s sums
(A. 1.228),
Virgil’s loricam induitur, coupled with a passive fidoque accingitur ense
(A. 7.640), an extension from indutus. In A . 11.6 he writes fulgentia induit
arma with a clear semantic distinction from the middle induitur. I
38.5. Another construction, again poetic in occurrence, is syntactically

parallel but semantically distinct. It is represented clearly in Lucretius’
percussi membra timore (5.1223), where the participle must be understood
not as a middle, as in percwsae pectora matres (Virg. A. 11.877), but as a
passive, with the accusative in its perlative sense, indicating the area within
which a state or prolonged action occurs, a usage already found, it seems,
in Enn. Ann. 310 Sk, percukri pectora Poeni 81 and of course more fami-
liar in the ‘accusative of duration in space or time’. It is thus comparable
to other poetic constructions like tremit artus (Lucr. 3.489) beside the more
normal construction exemplified in et corde et genibus tremit (Hor. “C.
1.23.8). The revival and extension of both accusative uses, assuming that
they had native precedent, was clearly influenced by Greek, as are
adnominal uses such as cetera Graius (Virg. A. 3.594) andflaua comas
(Ov. M . 9.307). Sometimes analysis is uncertain. Thus sensus deperditus
*l See also Skutsch’s note on fossari corpora telis (583), where the text and context are less
secure.

OmneS (Prop. 1.3.11) may refer either to total loss (middle) of the senses
or destruction (passive) over all the senses.
39.1. A number of uses of the infinitive are characteristic of the poetic

register. In dependence on adjectives as in celerem sequi (Hor. C. 1.15.18)
the infinitive is older than the gerundival constructions which replaced it
in classical prose; cf. auidus consul belli gerundi (Sall.lug. 35.3) with auidi
cornmitterepugnum (Ov. M.5.79, and in the predicative complement sum
defessus quaerere (Pl. Epid. 197) with defessus sum ambulando (Ter. Ad.
713). The infinitival construction is Indo-European and the infinitive had
originally kept to the active form, being like all verbal nouns unmarked
for voice, but adjustment is sometimes made for voice, as in niueus uideri
(Hor. C. 4.2.59). A comparison between felix et ponere uitem (Virg. G.
1.284) and felix uobis corrumpendis fuit (Liv. 3.17.2) indicates the eco-
nomical value to poets of the infinitive, which was used similarly but
independently in Vulgar Latin.
392. In dependence on nouns the same two constructions were in
rivalry already in Plautus, e.g. tempus est subducere hinc me (As. 912) but
ternpust udeundi (Trin. 432); tantus amor casus cognoscere nostros (Virg.
A. 2.10) but amor sceleratus habendi (Ov. M . 1.131). In pudor est quaedam
coepisse priorem (Ov. Ars 1.705) the adjective is in agreement with a non-
existent subject, a situation that has fascinated generations of grammarians.
Strictly speaking, only the gerundial construction is adnominal, ‘the time of
approaching exists’, the infinitive is a datival complement, ‘the time exists
for approaching’. In the last example the infinitival phrase is a predicative
complement, ‘the shame is to begin’.
40.1. Greek influence also helped to re-establish the purely aspectual

distinction between infinitives in the complements of certain verbs. Its
native credentials are guaranteed by neiquis eorum Bacanal hubuise uelet
(CIL 1581) and nequid emisse uellet (Cato, Agr. 5.4). The revival in poetry
clearly had something to do with the metrically useful trochaic ending, but
a firm aspectual sense can be seen in Horace’s tendentes opaco / Pelion
inposuisse Olympo (C. 3.4.51-2) and Propertius’ ergo uelocem potuit domu-
isse puellam (1.1.15), both in contexts rich in Greek mythological allusion.
The occurrence of this usage at Vindolandap2illustrates once more the
cras quid uelis nos fecisse rogo domine praecipias, Tab. Vindol. ii 505 in Bowman and
Thomas (1996: 324) and p. 8 above.

84 R. G. G. Coleman
possibility of correspondence between poetic and colloquial usage against

literary prose (see 013.4).
40.2. In contrast to these usages, which became established as perma-
nent additions to the poetic register, there are other Graecisms that remain
extremely rare. The plain infinitive complement in quo ire dixeram (Pl.
Cap. 194) and quae conuenere. . . / fatetur transtulisse atque usum (not
usus esse!)pro suis (Ter. An. 13-14), may be native to Latin. But Catullq’
phaselus ille . . . ait fuisse nauium celerrimus (4.1) is a specially motivated
Greek intrusion, perhaps intended to characterize the old boat as an
immigrant from the Greek-speaking East. Horace’s uxor inuicti Zouis esse
nescis (C. 3.27.73) addressed by Venus to Europa may also be intended to
give localized Greek colour. This could apply to Androgeos in Virg. A.
2.377, sensit medios delapsus in hostes, and to Penelope in Prop. 2.9-7,
uisura et quamuis numquam speraret Vlixen.
4l.l. Another economically motivated preference is for participial syntax

as an alternative to subordinate clauses. Here the morphological poverty
of the Latin participial system and the predominantly adjectival function of
the present participle in early Latin reduce the probability of native
precedent. Catullus is the lirst poet to use participles extensively in his
syntax (there are many in the narrative parts of 64,seven in lines 1-10
alone), at a time when they were little employed in prose. Cicero, si@-
cantly, has participial constructions more often in his poetry than in his
prose. The Augustan poets followed Catullus’ lead, and the usage then
spread through Livy into the conventions of prose writing. All this confirms
a Greek stimulus.
41.2. As in Greek, the syntactic conciseness is purchased at the cost of
semantic precision. The distinction between a state described and action
narrated is blurred, and the relation between subordinate constituent and
principal clause, whether it is descriptive (viz. relative), temporal, con-
ditional, causal or concessive, is left unspecified unless some clarifying
adverb is attached to the participle, which reduces the gain in conciseness.
In diua quibus retinens in summis urbibus arces / ipsa leui fecit uolitantem
flumine currum (Cat. 64.8-9) the uolitantem phrase is descriptive but reti-
nens could be taken temporally, like aspirans and implorata in 68.64-5.
41.3. Participles are much less prominent in the Aeneid than they are
in the Peleus and Thetis and are deployed with more subtle diversity.
They are an important ingredient in Horace’s concentrated style, and the
ambiguities mentioned above are exploited semantically by the poet. C.
4.7 is by his own standards very short on participles, but decrescentia (3)
and interitura (10) are both impressive. The pentasyllabic present participle

with its1 opening e-sounds evokes the more leisurely flow of the spring
fiver and with praetereunt becomes, like the river in 3.29.33-41, symbolic
of growth and decay and the relentless passage of time, which is also
reiterated in interitura. NO image this time: the participle draws on the
unusual iimagery of uer proterit aestas, to which it is an ominous appendage,
whether we take it as descriptive, concessive or both. Sometimes the
participles come in force, as in 3.2.6-9, illum ex moenibus hosticis matrona
bellantis tyranni prospiciens et adulta uirgo suspiret, where the second is
temporal, the first and third descriptive, and the compression achieved by
them is important in focusing the image. The various expressive possi-
bilities of the choice between participle and clause are explored to full
effect by the Augustan poets and by their imperial successors, in prose as
well as verse.
42.1. Syntactic dislocations of various kinds are widely used by the Latin
poets. Hypallage is perhaps the most frequent. We have already noted
Tyrrhena regum progenies (028.5). Also from Horace is obliuioso leuia
Massico / ciboria exple (C. 2.7.21-2), an easy transfer of an epithet that is
itself rare and means simply ‘full of forgetfulness’. Virgil’s saeuae memorem
Zunonis ob iram (A. 1.4), seems like a reciprocal transfer or enallage from
the metrically parallel but phonetically inferior saeuam memoris Iunonis ob
iram. The transfer gains semantically too: saeua becomes the characterizing
epithet for Juno, and the wrath is given its specific motivation. A similar
enallage occurs with comparable effect in the famous ibant obscuri sola
sub nocte (A. 6.268). Hypallage is not unknown in prose, e.g. Cic. Man.
22, eorum (sc. membrorum) collectio dispersa; but it is primarily poetic.
422. More violent dislocations can be seen in Prop. 2.26.18, qui, puto,
Arioniam uexerat ante lyram, where the grammatical form, in contrast to
the prosaic Ariona lyram ferentem, makes Arion subordinate to his lyre,
a witty conceit underlined by puto. The same quasi-satiric effect can be
seen in 3.2.19, Pyramidum sumptus ad sidera ducti, a bizarre dislocation
of the prosaic Pyramides sumptuosae ad sidera ductae, which would still
retain the central alliteration and the hyperbolic ad sidera. Juvenal in Sat.
1.1&11 has unde alius furtiuae deuehat aurum /pelliculae, for the prosaic
aliw pelliculam auream furtiue deuehat. The casual alius, the choice of the
diminutive pelliculae, made subordinate grammatically to aurum, which is
thus highlighted, and the transfer of furtiue into the object phrase all
conspire to reduce the heroic legend to an anonymous act of larceny. Such
dislocations are the converse to hendiadys; but like it they are seldom
mbre word games.

86 R. G. G. Coleman
42.3. Hendiadys itself, in replacing a head-plus-dependent noun phrase

by a co-ordinate one, invites the reader to contemplate an object and its
material component as distinct and equal items. In qualem pateris libamus
et auro (G. 2.192) Virgil describes wine offered in sacrifice. The hendiadys
highlights, as pateris aureis could not, the visual image of colour and
wealth. In clausae tenebris et carcere caeco (A. 6.734) Virgil reinforces the
disjunction by the alliterative link back across tenebris to clausae. ‘Ihe
semantic extension of caecus, itself perhaps originally the result of hypal-
lage, was as old as Accius, nocte caeca (tr. 32) and too widespread by now
to be distinctive. Some alleged examples can be analysed more satisfac-
torily in other ways. In Remulus’ taunt, non pudet obsidione iterum
ualloque teneri? (A. 9.598), ualloque adds the physical object that
accompanied the siege, and it is hard to envisage a unitary noun phrase
that would convey this. Again in Horace’s oppida publico /sump& iubentes
et deorum /templa nouo decorare s u o (C. 2.15.18-20) the two instrumental
phrases have different semantic relations with the infinitive, which once
again a single noun phrase of the head-plus-dependent type would not be
able to represent.
42.4. Oxymoron perhaps belongs here also, since it breaches the con-
ventions of syntactic collocation between lexical items. Again examples
occur in prose, such as absentes adsunt et egentes abundant. . . (Cic. Am.
23), emphasizing the paradoxical character of friendship; but the figure is
more characteristic of poetry. In Horace’s quid uelit et possit rerum con-
cordia discors (Epist. 1.12.19) the noun phrase aptly indicates the unity,of
disparates that was the goal of philosophical systems and by combining it
with a verbal pair appropriate rather to an animate subject seems !to
suggest that this goal has an existence independent of its proponents In
Horace’s famous description of Hypermestra as splendide me& (C.
3.11.35) the contrast with her impiue sisters and periurum parentem under-
lines the moral dilemma, resolved in a way that may itself be liable to the
charge of impious treachery but still ensured that she would be in omne
uirgo nobilis aeuom. Sometimes an oxymoron, once created, comes to be
exploited in prose as well as verse; cE cruda deo uiridisque senectus, used
of Charon in Virg. A. 6.304, with senem sed mehercules uiridem animo et
uigentem (Sen. Ep. 66.1) and other imperial examples
I’
43.1. It is well known that Livy’s prose style represents a distinct move
towards poetic usage, especially in lexicon and syntax (see, e.g. $38.3).
Conversely Augustan poetry introduces into its syntax something associ-
ated primarily with prose style, the extended complex sentence. Among

Virgdian exampless3is A. 1.305-9 (1) at pius Aeneas (2) per

noclem p * ~ 1 (3) ut primum lux alma data est (1) exire
l ~ uoluens,
locosque/explorare nouos (4) quas uento accesserit oras, I ( 5 ) qui teneant -
(6) nam incdta uidet - (7) hominesne feraene I (1) quaerere constituit
sociisque (8) exacta (1) referre.
43.2. There are eight constituents@in this narrative sentence, and the
structure is quite elaborate. It is a pure period in that no terminal point
is possible until the end; the principal clause (1) weaves through the
sentence from the start to the finish at pius Aeneas exire . . . explorare . . .
quaerere constituit sociisque . . . referre. The participial constituent (2),
which directly dependent on (l), is semantically a past imperfective
i s 1
participle (‘having been turning over. . .’) a semantic distinction for which
the finite verb system offers no exponent either (‘when he had been
turning over.. .’). The group of indirect questions (4) (5) (7) is dependent
not loosdy on locosque explorare nouos, as it seems to be at first, but
firmly on the delayed quaerere. It shows moreover an internal incoherence.
For qui teneant ( 5 ) is effectively replaced by hominesne feraene (7) with
teneant understood from the former. Since nam is not a subordinating
conjunction, its constituent (6) must be treated as an aside. This incoher-
ence aptly reflects the hero’s bewilderment. The structural profile of the
whole sentence can be represented thus:
It is very reminiscent of a prose period and even the word order is

surprisingly prosaic. But what is remarkable is that the entire event from
Aeneas’ sleepless night to his report back to his men is reported in a
single five-verse sentence, recalling in fact many of Livy’s narrative periods.
43.3. Horace’s Cleopatra ode (C. 1.37), consists of three sentences,each
longer than its predecessor: 1-4, 2-12, 12-32. The last of these offers a
sharp contrast to the Virgilian example. It is very complex but not strictly
periodic, since numerous stopping points would have been possible -after
ab ignibus (13), Caesar (16), adurgens (17), columbas (18), Haemoniae (20)
and fatale monstrum (21), which is indeed taken by some editors to mark
the end of a sentence; after oras (24), seuero (26), serpentes (27), uenenum
(28), ferocior (29), invidem (30) and finally triumph0 (32). The effect is of
a long succession of narrative or descriptive details being added by way
83 See Norden (1903 377-90) for some aspects of Vigil’s practice not covered here.
&1 For the method of analysis employed here see Coleman (1995).

88 R. G. G. Coleman
of qualification to what has gone before, so that after the principal clause,
comprising sed minuit furorem. . . mentemque . . . redegit Caesar (12-M),
the focus of the sentence is moved steadily away via the predator-prey
simile to the image of the noble queen - generosius /perire quaerens nec
muliebriter / expauit ensem (21-3) - the final epic noun here like the enic
simile in 17-20 adding heroic status to the events - uoltu sereno -..
fortis . . . ferocior . . . non humilis mulier (26-32). The radical shift of focus
effected by this constant series of additions anticipates Tacitus. By contrast
the continual addition of new material to the famous complex sentence.in
C. 4.2.5-24 (see 0 36.1) has a cumulative effect after the initial principal
clause (5-8). The choice of words in 1.31 is much more poetic than Virgil$,
and the word order, though conditioned to a great extent by the metre, is
highly effective, beginning with the emphatic predicate-subject order in
12-16. The use of complex structures may have been inspired by classical
prose, but the poets exploited the innovation to various expressive effects.
44.1. Apostrophe, by replacing third person by second person narration,

is a calculated intrusion by the poet into the impersonal context, in order
to suggest his own emotional involvement and invite the reader to respond
accordingly. It is especially appropriate to the more highly charged contexts
of poetry. Thus Laodamia’s desperation is highlighted by Catullus’ direct
address in quo tibi tum casu pulcherrima Laodamia, / ereptum est uita
dulcius atque anima / coniugium (68.105-7). In Prop. 3.2.7-8, within a triad
of mythological musicians, Polyphemus is highlighted for the erotic
success of his carmina by a remarkable apostrophe: quin etiam, Polypheme,
fera Galatea sub Aetna / ad tua rorantis carmina flexit equos. The ironic
humour of this misinterpretation of the tale is compounded by the mock-
heroic connotations of the concluding equestrian image.
44.2 The figure is of course not unknown to early epic e.g. already
Hom. Od. 14.55, T ~ G’dpc~fidpcvos
V w p o o i ~ ~ EGpaLc
s, avficjra, but Virgil uses
it for various more emotive purposes, e.g. to elevate the Campanian chief-
tain in A. 7.733, nec tu carminibus nostris indictus abibis, / Oebale,
preceding an explicit prosopographical account clearly significant i for
Italian legend and an Italian poet but now largely obscure. The most
famous instance is of course the powerful and self-fulfilling promise, Fortu-
nati ambo! si quid mea carmina possunt . . . (9.446-50), following the death
of Nisus and Euryalus. Apostrophe is of course a familiar figure in oratoW
Quintilian (9.2.38) cites examples from Cicero’s speeches, including itme
based on personification, uos iam, Albani tumuli atque luci, uos, inquam,
imploro atque testor, uosque Albanorum obrutae arae (Mil. 85). The differ-
ence between the two registers is chiefly one of frequency.

45.1. o n e of the most notable differences from prose is word order vari-
ation, in part metrically conditioned, though disrupted word orders are by
no means unknown in literary prose e.g. breuis a natura nobis uita data est
(tic. phil. 14.32).In a highly inflected language word order has no syntactic
function, as it has in English, but it does have an important pragmatic role,
adding emphasis to a word or phrase by distancing it from words with
which it is grammatically linked, by juxtaposing it with words that are
not grammatically connected but which help to define its meaning more
specifically or by any major deviation from the patterns of ordinary prose
writing, e.g. uita breuis a natura nobis data est or fluctus uastos ad litora
uoluunt. In uastos uoluunt ad litorafluctus (Virg. A. 1.86) uastos, in allitera-
tion with uoluunt with which it forms a menacing pair, is held in suspense
till fluctus at the end. The wind-driven waves from all sides converge ad
litera. In aureus et foliis et lento uimine ramus (6.137) epithet and noun are
separated, with the colour image emphatically first and the noun coming as
a revelation at the end. In ingentes Rutulae spectabit caedis aceruos (10.245)
the definition of the appalling image awaits the two final nouns and the
chiastic word order enclosing the verb enacts the vivid sense that its
subject, the personified crastina lux (244), will be surrounded on all sides.
453. A couplet taken almost at random from Propertius illustrates the
effects that variation from the prosaic order can contribute: his tum blandi-
tiis furtiua per antra puellae / oscula siluicolis empta dedere uiris (3.13.334)
v. turn puellae oscula furtiua his blanditiis empta uiris siluicolis per antra
dedere. The hypallage of furfiua,placed in a metrically prominent position,
in any case distances the latter phrase from prose. The reproach of empta is
emphasized in the separation from its noun by the syntactically ambiguous
siluicolis (with empta or dedere?) and in the oxymoronic juxtaposition with
dedere. The lofty compound adjective itself is a witty reminder that the
passage began with the Golden Age image of felix agrestum quondam
paccrta iuuentus (25).
45.3. The severe metrical constraints of the Horatian lyric metres clearly
entailed distortions of the normal Latin word order, but the poet shows
again and again an unparalleled skill in turning the necessities to relevant
poetic effects. Take the river simile in 4.2.5-8: monte decurrens uelut amnis,
imbres / quem super notas aluere ripas, / feruet immensusque ruit profundo /
Pindarus ore. If we rewrite the stanza in a normal prosaic order, e.g. Pind-
arm, auelut amnis monte decurrens, quem imbres super notas ripas aluere,
feruet immensusque ore profundo ruit, the contrast is very marked. It is
worth noting incidentally that, as with most good poetry, rearrangement
into a prose order leaves us with a very unprosaic piece of prose.
45.4. The noun amnis is attested predominantly in poetry, ex or de
montibus would be more normal in prose (Livy has amnis diuersis ex

90 R. G. G. Colemun
AZpibus decurrentes in 21.31.4), and the particular metaphoric use of h e r e

is not found in prose before Tacitus. But the order contributes to the
poetic effect. The delaying of ueZut gives the powerful initial image promin-
ence before its reference is specified, and the preposing of imbres to its
relative clause juxtaposes it emphatically with amnis. The fact that feruet
immensusque ruit and profundo do not refer to amnis is held back till the
last verse, where their metaphoric reference is at last made clear and
the aptness of the simile confirmed. The only notable alliteration
reinforces the link between Pindarus and profundo.
CONCLUSION
46.1. This long but far from exhaustive surveys can now be summarized
and its argument brought together. An attempt has been made to
describe and illustrate poetic diction, which in both the narrow and the
wider sense (01.1) essentially defines the poetic register of the language,
that is to say the form of Latin in which poetic discourse was conducted.
Literary prose, being more selectively rationalized, establishes a linguistic
norm against which we can plot the characteristics of the poetic register
as a series of deviations. The task of plotting ought to have led us to a
body of exclusive criteria, but in practice the number of linguistic
phenomena that are found exclusively in poetry is small, if far from
negligible. This is perhaps not so surprising, given that no dividing line
can be drawn between poetic and prose subjects. It is worth recalling for
instance that Virgil wrote out a preliminary prose version of the Aeneid
(Don. Vitu 23) and that many manuscripts of Horace’s odes offer titles for
most of the poems that include standard rhetorical categories, such
as prugmatice (1.1, 3.30), puruenetice (1.9, 4.7), exprobrutio (l.25),
inuectio (1.23). Not all of them are helpful, but the use of them is
revealing.%
46.2. Nevertheless it is possible to point to linguistic usages that are
more prominent in poetry and this is what I have done in 503-45. The
most fundamental distinguishing characteristic of Latin poetry is metrical
form ( 0 3 4 ) . Of less importance at the phonetic level, since it is their
frequency not their presence that is distinctive, are Greek sounds
and Greek collocations of Latin sounds (015). More infrequent are
85For a longer treatment of relevant phenomena see HofmannSzantyr (1965: 685-858).

86The rhetorical classifications of ancient poetry have been taken up, developed and modified
by Cairns (1972). Some of his classes correspond with those in the Horatian manuscripts
e.g. propemptice (1.3), hymnus (1.21); many do not.

phonetic archaisms like anirndi (07-8) and innovations, which are virtu-
ally confined to words that are otherwise metrically intractable but never
distort the basic phonology of the language (05-11): rnetri gratia is not
laissez-faire.
463. In the area of grammatical morphology archaic inflections are
sometimes retained (012). This illustrates an important difference between
poetry and prose. Whereas literary prose was closely in touch with current
educated speech and archaism tended to be disapproved,8’ poetry was less
at the mercy of linguistic change. In a literate society the written poetry
that survived from the past was not only constantly accessible - this was
equally true of prose - but it also exercised a formative influence on the
work of individual poets, who consciously imitated it as a deliberate signal
of allegiance to a specific tradition.= This was quite distinct from the
largely unconscious and impersonal influence that past states of a language
must always have upon its present character. Literary works that acquired
a classic status within their genres obviously had the greatest influence
upon individual poets. This can at different periods be true also of prose
genres, as illustrated in the influence of Cicero on patristic writing from
Minucius Felix and Lactantius onwards,89or of The Authorized Version of
the English Bible on subsequent prose literature in English.
46.4. This timeless status of poetry applies even more to the lexicon,
where it not only signals adherence to a tradition but also extends the range
of synonyms available to the poet (0021-3). Innovations were certainly
acceptable in this area, principally loan-words from Greek (0027-8) and
the formation of compound words, scrupulously modelled on existing
patterns of Latin word formation even though their inspiration almost
invariably came from a particular Greek word (026).
46.5. However, the two most important features of the poetic lexicon,
not only in Latin, do not need to depart from ordinary vocabulary. The
first is the use of semantic transference, as in metonymy (032), synecdoche
(033) and above all metaphor (029-31). These are largely responsible for
the creation of poetic diction in the narrow sense (034.2). The second is
imagery, which is often created out of very unpoetic vocabulary, not least in
similes (0035-6). A word may have the status of poetic diction temporarily
conferred on it by the context of poetic discourse in which it is set, as for
instance the metaphoric praesidiurn of Maecenas in Hor. C. 1.1.1 or the
hyperbolic centurn clauibus to characterize the miser against his wasteful
For the rationalizing process by which classical prose usage was formed see Neumann
(1968).
For comprehensive treatment of this important subject see Williams (1968). Thill(1979).
89 rite intluence of the teachings of Romani amtor eloquii is pervasive i n De doctrina
Chrirriana, though Augustine never names him there.

92 R. G. G. Coleman
heir in the magnificent final stanza of C. 2.14. Other words become estab-
lished as permanent elements of poetic diction by successive imitations of
an original trope - carina, freturn etc. (003-34).
46.6. It is in syntax that the contrast with the rationalizing processes
of prose is most striking. Both archaism and innovation are again at work
here, and the placing of a given construction on the diachronic axis is
often difficult. An archaic construction, purged from literary prose in
favour of a more rational rival, may be identical with a vulgarism similarly
rejected by literary prose (039). The poetic and vulgar registers after all
share a concern for concise and vivid expression which will end up in the
same place, whether it is a preference for metaphor or for infinitival
constructions. Finally while the introduction of complex sentence structure
into poetry (043) shows interaction with literary prose that is remote
from vulgar usage, the various forms of syntactic dislocation (042) and
manipulation of word order (045) mark off poetry from both vulgarism
and prose literature.
46.7. The poetic register thus contained far more than poetic diction
even in the wider sense of the term. A poet may, as we saw in 320.7-9,
choose to confer the status of poetic diction on words that in other contexts
of occurrence are thoroughly prosaic, simply by placing them in an appro-
priate setting. Thus the plain ‘unpoetic’ vocabulary of the famous Wher’ere
you walk quatrain from Pope’s Pastorals, 2.73-6, is charged with poetic
power by the thoroughly traditional use of the trope known since Ruskin
as the pathetic fallacygoConversely the poetic effect may depend precisely
on the words retaining their prosaic connotations. Only the context of the
poetic discourse itself can enable us to decide. For it is not just the presence
of this or that linguistic item that is definitive, but rather the texture of a
whole passage, formed from the accumulation of other ingredients sum-
marized in these concluding paragraphs. This is why, in many of the
examples cited to illustrate a particular feature of the poetic register in
the course of this essay, other features of the context not relevant to the
point immediately under discussion have been noted. Moreover, because
the poetic register is not just a set of procedures for translating prose into
verse and so embellishing an argument otherwise conceived, but a vehicle
for deploying its own kind of argument, the reason why a particular feature
or group of features is there has been constantly sought. For they always
form an integral part of poetic discourse. This is why the definition of
poetic diction in the narrow sense does not take us beyond a circumscribed
and relatively small area of the lexicon; but the definition of the poetic
90The phonetic and syntactic composition of which is finely analysed by Nowottny (1%2:
11-12).

register takes Us into the entire concept of what a poem is and what it is
created to do.9'
p1I am grateful to the two editors and to Professor H. D. Jocelyn, who have greatly improved
this chapter by their learned and trenchant criticisms and by their corrections of downright
errors, and would no doubt have improved it even more, if I had allowed them to do so.

Bibliography
Adamietz, J. (ed.) (1986), Die romische Satire (GrundriB der Literaturgeschichten
nach Gattungen) (Darmstadt).
Adams, J. N. (1971), ‘A type of hyperbaton in Latin prose’, PCPhS 1 7 1-16.
-(1976), ‘A typological approach to Latin word-order’, Zndogermanische For-
schungen 81: 70-99.
-(1980a), ‘Latin words for woman and wife’, Glotta 5 0 23455.
-(1980b), ‘Anatomical terminology in Latin epic’, BZCS 27: 50-62.
-(1982a), The Latin Sexual Vocabulary (London).
-(19826), ‘Anatomical terms transferred from animals to humans in Latin’,
Zndogermanische Forschungen 87: 90-109.
-(1983), ‘Words for “prostitute” in Latin’, RhM 126 321-58.
-(1992), ‘Iteration of compound verb with simplex in Latin prose’, Eikasmos
3: 295-8.
-(1994a), ‘Wackernagel’s law and the position of unstressed personal pro-
nouns in Classical Latin’, TPhS 9 2 103-78
-(1994b), Wackernagel’s Law and the Placement of the Copula esse in
Classical Latin (Cambridge Philological Society, Suppl. vol. 18) (Cambridge).
-(1995a), ‘The language of the Vindolanda writing tablets: an interim report’,
JRS 85: 86134.
-(1995b), Pelagonius and Latin Veterinary Terminology in the Roman Empire
(Studies in Ancient Medicine, 11) (Leiden).
Allen, W. S. (1973), Accent and Rhythm. Prosodic Features of Latin and Greek: a
Study in Theory and Reconstruction (Cambridge).
-(1978,2nd ed.), Vox Latina. A Guide to the Pronunciation of Classical Latin
(Cambridge).
Alfonso, S., Cipriani, G., Fedeli, P., Mazzini, I., Tedeschi, A. (1990), Zlpoeta elegiac0
e il viaggio d’amore (Scrinia, 3) (Sari).
Anderson, R. D., Parsons, P. J. and Nisbet, R. G. M. (1979), ‘Elegiacs by Gallus
from Qasr Ibrim’, JRS 6 9 125-55.
Anderson, W. S. (1956; 1964; 1970; 1981), ‘Recent Work in Roman Satire’, ClW
5 0 3340; ClW 57: 293-301; 343-8; ClW 63: 181-94; 199; 217-22; ClW 75:
273-99.
-(1961), ‘Venusina lucerna: the Horatian model for Juvenal’, TAPA 52: 1-12.
(Reprinted in Anderson (1982) 103-14.)
-(1962), ‘The Programs of Juvenal’s Later Books’, CPh 57: 145-60.
(Reprinted in Anderson (1982), 277-92.)
-(1982), Essays on Roman Satire (Princeton).
AndrC, J. (1949), Etude sur les termes de couleur duns la langue latine (Paris).
-(1967), Les noms d’oiseaux en latin (Paris).

416 BIBLIOGRAPHY
-(1980), ‘Deux remarques sur le volume du mot latin’, RPh 5 4 7-18.

-(1987), Etre mtdecin a Rome (Realia) (Paris).
-(1991), L e vocabulaire latin de l’anatomie (Paris).
Arens, J. C. (1950), ‘-fer and -ger: their extraordinary preponderance among com-
pounds in Roman poetry’, Mnemosyne4 3: 241-62.
Argenio, I. (1963), ‘I grecismi in Lucilio’, CRSt 11: 5-17.
Artymowicz, A. (1909), ‘Der Wechsel von et und que zu Beginn lateinischer
daktylischer Verse von Ennius bis Corippus’, Wiener Studien 31: 38-81.
Atherton, C. (1996), ‘What every grammarian knows?’, C Q NS 46: 239-60.
Austin, R. G. (ed.) (1964), P. Vergili Maronis Aeneidos Liber Secundus (Oxford).
-(ed.) (1971), P. Vergili Maronis Aeneidos Liber Primus (Oxford).
-(ed.) (1977), E! Vergili Maronis Aeneidos Liber Sextus (Oxford).
Avotins, I. (1980), ‘Alexander of Aphrodisias on vision in the atomists’, C Q NS 3 0
429-54.
Axelson, B. (1945), Unpoetische Worter. Ein Beitrag zur Kenntnis der lateinischen
Dichtersprache (Lund).
Bader, E (1962), L a formation des compos& nominaux du latin (Paris).
Baehrens, E. (ed.) (1885), Catulli Veronensis liber (Leipzig).
Baehrens, W. A. (1912), Beitrage zur lateinischen Syntax. Philologus, Suppl. 12
(Leipzig).
Bagnall, R. S. (1993), Egypt in Late Antiquity (Princeton).
Bailey, C. (ed.) (1947, corr. ed. 1949,3 vols), Titi Lucreti Cari de rerum natura libri
sex (Oxford).
Baratin, M. (1989), L a naissance de la syntaxe a Rome (Paris).
Barnes, J., Mignucci, M. (edd.) (1988), Matter and Metaphysics (Naples).
Bartalucci, A. (1968), ‘La sperimentazione enniana dell’esametro e la tecnica del
saturnio’, SCO 17: 99-122.
Bauer, C. E (1933), The Latin Perfect Endings ‘-ere’ and ‘-erunt’ (Ling. Soc.
America, Language Diss. 13) (Philadelphia).
Beck, M. (1996), Die Epistulae Heroidum XVZIZ und XZX des Corpus Ovidianum.
Echtheitskritische Untersuchungen (Paderborn).
Bell, A. J. (1923). The Latin Dual and Poetic Diction (London and Toronto).
Benediktson, D. T. (1977), ‘Vocabulary analysis and the generic classification of
literature’, Phoenix 31: 341-8.
Bennett, C. E. (1910), Syntax of Early Latin, Vol. Z-The Verb (Boston).
Bentley, R. (ed.) (1711), Q. Horatius Flaccus (Cambridge).
Benz, L., Stark, E., Vogt-Spira, G. (edd.) (1995), Plautus und die Tradition des
Stegreifspiels. Festgabe fiir E. Lefkvre zum 60. Geburtstag (Tubingen).
Berkowitz, L. and Brunner, “h. E (1968), Index Lucilianus (Hildesheim).
Binder, G. (ed.) (1988), Saeculum Augusturn ZZ (Wege der Forschung 512)
(Darmstadt).
Biville, E (1987), Graphie et prononciation des mots grecs en latin (Paris).
-(1989), (ed.) ‘Grec et latin: contacts linguistiques et crkation lexicale. Pour
une typologie des hellknismes lexicaux du latin’, in Lavency and LongrCe
(1989), 29-40.
-(1990), Les emprunts d u latin au grec: approche phonktique vol. I
(Bibliothkque de l’information grammaticale, 19) (Louvain-Paris).

BIBLIOGRAPHY 417
Blase, H. (1903), ‘Tempora und Modi’, in G. Landgraf (ed.), Historische Grammatik

der lateinischen Sprache. 3. Band Syntax des einfachen Satzes (Leipzig).
Bloch, H. (1940), ‘L. Calpurnius Piso Caesoninus in Samothrace and Hercu-
laneum’, MA 44: 485-93.
Blilmel, W. (1979), ‘Zur historischen Morphosyntax der Verbalabstrakta im Latei-
Nschen’, Glotta 57: 77-125.
Boetticher, G. (1830), Lexicon Taciteum (Berlin).
Boldt, H. (1884), De liberiore linguae Graecae et Latinae collocatione verborum
(Diss. Gottingen).
Bollack, J. (1965-69), Emptdocle (3 vols; Paris).
Bomer, E (1951), Review of Axelson (1945), Gnomon 2 3 166-8.
-(1952), ‘Excudent alii. . .’, Hermes 8 0 117-23.
-(1957), ‘Beitrage zum Verstandnis der augusteischen Dichtersprache’, Gym-
nasium 64:1-21.
-(1965), ‘Eine Stileigentiimlichkeit Vergils: Vertauschen der Pradikate’,
Hermes 93: 130-1.
_ _ (1967), ‘Ovid met. I 39’, Gymnasium 7 4 223-6.
-(1969), Z? Ovidius Naso. Metamorphosen. Buch 1-111 (Heidelberg).
-(1976), Z? Ovidius Naso Metamorphosen. Buch IV-V (Heidelberg).
-(1982), Z? Ovidius Naso Metarnorphosen. Buch XII-XI11 (Heidelberg).
Bonjour, M. (1984), ‘Cicero nauticus’, in R. Chevallier (ed.), Prtsence de Cictron,
9-19 (Collection Caesarodunum 19 bis) (Pans).
Bonner, S. E (1949), Roman Declamation in the Late Republic and Early Empire
(Liverpool).
Booth, J. (1981), ‘Aspects of Ovid’s language’, in H. Temporini (ed.), A N R W 11.31.4
2686-700 (Berlin-New York).
_ _ (ed.) (1991), Ovid. The Second Book of Amores (Warminster).
Bourgeois, I? (1940), ‘L‘hellCnisme, pro&dC &expression dans les GCorgiques’,
REL 18: 73-94.
Bowman, A. K., Thomas, J. D. and Adams, J. N. (1990), ‘Two letters from Vindo-
landa’, Britannia 21: 33-52.
Bowman, A. K. and Thomas, J. D., with contributions by Adams, J. N. (1994), The
Vindolanda Writing-Tablets (Tabulae Vindolandenses ZZ) (London).
Bowman, A. K. and Thomas, J. D. (1996), ‘New writing tablets from Vindolanda’,
Britannia 27: 299-328.
Bowra, C. M. (1952), Heroic Poetry (London).
Bramble, J. C. (1974), Persius and the Programmatic Satire (Cambridge).
-(1982a), ‘Martial and Juvenal’, in Kenney and Clausen (1982), 101-27.
-(1982b), ‘Lucan’, in Kenney and Clausen (1982), 533-57.
Braund, S. H. (1989a), ‘City and country in Roman satire’, in Braund (1989b),
23-48.
-(ed.) (1989b), Satire and Society in Ancient Rome (Exeter Studies in History,
23) (Exeter).
-(1992a), Roman Verse Satire (Greece and Rome New Surveys in the Classics,
23) (Oxford).
-(1992b), Lucan, Civil War, translated with introduction and notes (Oxford).
-(ed.) (1996), Juvenal, Satires, Book Z (Cambridge).

418 BIBLIOGRAPHY
Brenous, J. (1895), Etude sur les helltnismes dans la syntaxe latine (Paris).
van Brock, N. (1961), Recherches sur le vocabulaire mkdical du grec ancien (Etudes
et Commentaires, 41) (Paris).
Brown, R. D. (1987), Lucretius on Love and Sex: a Commentary on De Rerum
Natura 1030-1287, with Prolegomena, Text and Translation (Columbia
studies in the classical tradition, 15) (Leiden).
Burger, R. (1911), ‘Beitrage zur Elegantia Tibulls’ in XAPZTEZ. Friedrich Leo
371-94 (Berlin).
BrunCr, E. A. (1863), ‘De ordine et temporibus carminum Valerii Catulli’, Acta
Soc. Scient. Fennicae 7: 599-657.
Bulow-Jacobsen, A., Cuvigny, H. and Fournet, J.-L. (1994), ‘The identification of
Myos Hormos. New papyrological evidence’, BZFAO 94: 2742.
Burnyeat, M. E (1978), ‘The upside-down back-to-front sceptic of Lucretius IV
472’, Philologus 122: 197-206.
Cairns, E (1972), Generic Composition in Greek and Roman Poetry (Edinburgh).
-(1979), Tibullus: a Hellenistic Poet at Rome (Cambridge).
-(1983), ‘Propertius 1.4 and 1.5 and the ‘Gallus’ of the Monobiblos’, PLLS
4 61-104.
-(1984), ‘The etymology of militia in Roman elegy’ in Apophoreta philologica
Emmanueli Fernandez-Galiano a sodalibus oblata 2.211-22 (Madrid).
-(ed.) (1986), Papers of the Liverpool Latin Seminar 5, 1985 (Liverpool).
-(1986), ‘Stile e contenuti di Tibullo e di Properzio’ in Atti del Convegno
Znternazionale di Studi su Albio Tibullo 49-50. (Rome).
Callebat, L. (1974), ‘Le vocabulaire de l’hydraulique dans le livre VI11 du De
architectura de Vitruve’, RPh 48: 313-29.
-(1982), ‘La prose du De Architectura de Vitruve’, in H. Temporini (ed.),
ANRW 11.30.1: 696-722 (Berlin).
-(ed.) (1995), Latin vulgaire, latin tardif:ZV Actes du 4‘colloque international
sur le latin vulgaire et tardif. Caen, 2-5 septembre 1994 (Hildesheim, Zurich,
New York).
Campanile, E. (1985), art. ‘grecismi’,in Enciclopedia Virgiliana ii.805-7 (Rome).
Casali, S. (ed.) (1995), l? Ovidii Nasonis Heroidum Epistula ZX. Deianira Herculi
(Florence).
Caspari, F. (1908), De ratione quae inter Vergilium et Lucanum intercedat quaes-
tiones selectae (Diss. Leipzig).
Cavenaile, R. (1958), Corpus Papyrorum Latinarum (Wiesbaden).
Ckbe, J.-F! (1966), La caricature et la parodie dans le monde romain, (Bibl. des
Ecoles franqaises d’Ath5nes et de Rome, 206) (Paris).
Charpin, E (ed.) (1978, 1979, 1991), Lucilius, Satires. Texte ttabli, traduit et annott
(Paris).
Christ, W. (1879, 2nd ed.), Metrik der Griechen und Romer (Leipzig).
Christes, J. (1971), Der friihe Lucilius. Rekonstruktion und Znterpretation des XXVZ.
Buches sowie von Teilen des XXX. Buches (Heidelberg).
-(1972), ‘Lucilius. Ein Bericht uber die Forschung seit E Marx (1904/5)’, in
H. Temporini (ed.), ANRW 1.2. 1182-1239 (Berlin).
-(1986), ‘Lucilius’, in Adamietz (1986), 57-122.
Cichorius, C. (1908), Untersuchungen zu Lucilius (Berlin).

BIBLIOGRAPHY 419
Clark, S. B. (1908), ‘The authorship and the date of the double letters in Ovid’s
Heroides’, HSCPh 19: 121-55.
Coffey, M. (1989, 2nd ed.), Roman Satire (Bristol).
Coleman, R. G. G. (1977) ‘Greek influence on Latin syntax’, TPhS 1975: 101-56.
-(1987), ‘Vulgar Latin and the diversity of Christian Latin’, in J. Herman
(ed.), Latin vulgaire-latin tardif 37-52. (TZibingen).
-(1989), ‘The formation of specialized vocabularies in grammar and rhetoric:
winners and losers’, in Lavency and LongrCe (1989: 77-89).
-(1991), ‘Latin prepositional syntax in Indo-European perspective’, in
Coleman (ed.), New Studies in Latin Linguistics 323-38 (Amsterdam).
___ (1995), ‘Complex sentence structure in Livy’, in D. LongrCe (ed.), De Vsu.
Etudes de syntaxe latine offertes en hommage a Marius Lavency, 71-84
(Louvain-la-Neuve).
Collinge, N. E. (1962), ‘Medical terms and clinical attitudes in the tragedians’,
BZCS 9: 43-7.
Conrad, C. (1965), ‘Word order in Latin epic from Ennius to Virgil’, HSCPh 69:
194-258.
Conte, G. B. (1970), ‘Ennio e Lucano’, Maia 22: 132-8.
Contino, S. (ed.) (1988), A . Cornelii Celsi, De medicine liber VZZZ (Bologna).
Coppel, B. (1976), review of Ross (1%9), Gnomon 48: 559-66.
Cordier, A. (1943), ‘La langue poCtique B Rome’, M h o r i a l des ttudes
latines . . .offert U J. Marouzeau 80-92 (Paris).
Courtney, E. (1965), ‘Ovidian and non-Ovidian Heroides’, BZCS 12: 63-6.
-(1980), A Commentary on the Satires of Juvenal (London).
-(ed.) (1993), The Fragmentary Latin Poets (Oxford).
Cutt, T. (1936), Meter and Diction in Catullus’ Hendecasyllabics (Diss. Chicago).
Dagron, G. (1969), ‘Aux origines de la civilisation byzantine: langue de culture et
langue &&at’, Rev. Hist. 241: 23ff.
Daube, D. (1956), Forms of Roman Legislation (Oxford).
De Decker, J. (1913), Juvenalis Declamans (Ghent).
Deichgraber, K. (ed.) (1935), Hippokrates iiber Entstehung und Aufbau des
menschlichen Korpers, ZIcpL‘ uaprtdv mit einem sprachwissenschaftlichenBeitrag
von Eduard Schwyzer (Leipzig-Berlin).
-(1971), Aretaeus von Kappadozien als medizinischer Schriftsteller (Abh. d.
Sachs. Akad. d. Wiss. zu Leipzig, Philo1.-hist. Kl., 63, 3) (Berlin).
D’Elia, S. (1961), ‘Echi del “de officiis” nell’ “Ars amatoria” ovidiana’, in Atti del
I congr. int. di studi ciceroniani, ii. 12740 (Rome).
Della Corte, M. (1958), ‘Le iscrizioni di Ercolano’, Rendiconti della Accademia di
Archeologia, Lettere e Belle Arti, n.s. 33: 239-308 (Naples).
Delatte, K. (1%7), ‘Keywords and poetic themes in Propertius and Tibullus’,
R E L O 3: 31-79.
Delz, J. (ed.) (1987), Sili Ztalici Punica (Stuttgart).
De Meo, C (1983), Lingue tecniche del latino (Testi e manuali per l’insegnamento
universitario del latino 16) (Bologna).
Denniston, J. D. (1952), Greek Prose Style (Oxford).
Deufert, M. (1996), Pseudo-Lukrezisches im Lukrez. Die unechten Verse in Lukre-

420 BIBLIOGRAPHY
zens ‘De rerum natura’. Untersuchungen zur antiken Literatur und Geschichte
48 (Berlin and New York).
Diggle, J. (1972), ‘Ouidiana’, PCPS NS 18: 31-41.
Diggle, J. and Goodyear, E R. D. (edd.) (1972), The Classical Papers of A. E.
Housman (Cambridge).
Dingel, J. (1997), Kommentar zum 9. Buch der Aeneis Vergils (Heidelberg).
Dionisotti, A. C. (1995), ‘Hellenismus’ in 0. Weijers (ed.), Vocabulary of Teaching
and Research Between Middle Ages and Renaissance (Civicima. Etudes sur le
vocabulaire intellectuel du Moyen Age 8) (Turnhout).
Dominguez Dominguez, J. E and Martin Rodriguez, A. M. (1993), ‘Dare con
infinitivo en latin clasico’, Cuadernos de filologia clasica, 4 9-22.
Dover, K. J. (1963), ‘The Poetry of Archilochus’, in Entretiens sur I’Antiquitk
classique 10: 183-212 (Geneva).
-(1968, corrected reprint of 1960 ed.), Greek Word Order (Cambridge).
Draeger, A. (1882, 3rd ed.), Uber Syntax und Stil des Tacitus (Leipzig).
Drexler, H. (1967), Einjiihrung in die romische Metrik (Darmstadt).
Dubois, J. (1966), ‘Les problbmes du vocabulaire technique’, Cahiers de lexicologie
9: 103-12.
Dumortier, J. (1935), L e vocabulaire mkdical d’Eschyle et les tcrits hippocratiques
(Paris).
Easterling, P. E. (ed.) (1982), Sophocles, Trachiniae (Cambridge).
Eich, M. (1925), De praepositionum collocatione apud poetas Latinos inde ab
Ovidio (Diss. Bonn).
Eklund, S. (1970), Theperiphrastic, completive andfinite use of the presentparticiple
in Latin. With special regard to translation of Christian texts in Greek up to 600
A.D. (Acta Universitatis Upsaliensis. Studia Latina Upsaliensia, 5) (Uppsala).
Elder, J. P. (1962), ‘Tibullus: Tersus atque Elegans’ in J. P. Sullivan (ed.) Critical
Essays on Roman Literature: Elegy and Lyric, 65-106. (London).
Eliot, T. S. (1932 [1917]), ‘Tradition and the individual talent’, in Selected Essays,
13-22 (London).
Ellis, R. (1876; 2nd ed. 1889), A Commentary on Catullus (Oxford).
Erbse, H. (1953), ‘Homerscholien und hellenistische Glossare bei Apollonios
Rhodios’, Hermes 81: 163-96.
Ernout, A. (1946), ‘Infinitif grec et gtrondif latin’, Philologica (Paris).
-(1947), ‘Le vocabulaire pottique’, rev. of Axelson (1945), RPh 21: 55-70
(= 19576: 66-86).
-(1956), ‘VENVS, VENA, CVPIDO’, RPh 3 0 7-27 (= 1957b: 87-111).
-(1957a), ‘MEWS - TIMOR. Les formes en -us et en -os (-or) du latin’,
in 19.576: 7-56
-(1957b), Philologica I1 (Paris).
Ernout, A. and Meillet, A. (1959; 4th ed.), Dictionnaire ktymologique de la langue
latine. Histoire des mots, augment6 d’additions et de corrections nouvelles par
J. Andrt (Paris).
Ernout, A. and Thomas, E (1953), Syntaxe latine (Paris).
Evans, W. J. (1921), Allitteratio Latina (London).
Evrard-Gillis, J. (1976), La rkcurrence lexicale dans l’oeuvre de Catulle: ttude
stylistique (Paris).

BIBLIOGRAPHY 421
Fantham, E. (1972), Comparative Studies in Republican Latin Imagery (Toronto).

Farrell, J. (1991), Virgil’s ‘Georgics’ and the Traditions of Ancient Epic (New York
and Oxford).
Fedeli, P. (ed.) (1965), Properzio, Elegie libro N: Test0 critico e cornmento (Bari).
-(ed.) (1980), Sesto Properzio, I1 prim0 libro delle Elegie: Introduzione, testo
critico e commento (Florence).
-(ed.) (1985), Properzio, I1 libro terzo delle Elegie: Introduzione, testo e
cornmento (Ban).
Ferguson, J. (1987), A Prosopography to the Poems of Juvenal (Brussels).
Fiske, G. C. (1919), ‘The plain style in the Scipionic Circle’, in Studies in Honor
of Ch. Forster Smith (Madison).
-(1920), Lucilius and Horace. A Study in the Classical Theory of Imitation
(University of Wisconsin Studies in Language and Literature 7) (Madison).
Fitch, J. G. (1981), ‘Sense-pauses and relative dating in Seneca, Sophocles and
Shakespeare’, A J P 102: 289-307.
Flashar, H. (ed.) (1971), Antike Medizin (Wege der Forschung 221) (Darmstadt).
Fluck, H.-R. (1980), Fachsprachen: Einfiihrung und Bibliographie (Munich).
Flury, P. (1968), Liebe und Liebessprache bei Menander, Plautus und Terenz
(Heidelberg).
-(1990), ‘Beitrage aus der Thesaurus-Arbeit,XXV occurrere’, M H 47: 225-6.
Fordyce, C. J. (ed.) (1961; repr. with corrections and additional notes 1973), Ca-
tullus: a Commentary (Oxford).
-(ed.) (1977), l? Vergili Maronis Aeneidos libri 7-8, with a commentary ed.
by John D. Christie (Oxford).
Fraenkel, E. (1922), Plautinisches im Plautus (Philologische Untersuchungen 28)
(Berlin)
-(1928), Iktus und Akzent i m lateinischen Sprechvers (Berlin).
-(1960 = transl. of [1922] with addenda), Elementi plautini in Plauto
(Florence).
-(1968), Leseproben aus Reden Ciceros und Catos (Rome).
Freudenburg, K. (1993), The Walking Muse: Horace on the Theory of Satire
(Princeton).
Friedlfinder, P. (1941), ‘Pattern of sound and atomistic theory in Lucretius’, A J P
6 2 16-34.
Gaisser, J. H. (1993), Catullus and his Renaissance Readers (Oxford).
Gardner-Chloros, P. (1991), Language Selection and Switching in Strasbourg
(Oxford).
Garvie, A. F. (ed.) (1986), Aeschylus, Choephori, with Introduction and Commen-
tary (Oxford).
Geymonat, M. (ed.) (1973), I? Vergili Maronis Opera (Turin).
Gianfrotta, P. A. (1987), art. ‘Navis’, in Enciclopedia Virgiliana iii. 6 7 W (Rome).
Gigante, M. (1981), Scetticismo e epicureismo (Naples).
Gigon, 0. (1978), ‘Lukrez und Enniu’, in Entretiens sur l’Antiquit6 classique 2 4
167-91 (Geneva).
Godwin, J. (ed.) (1986), Lucretius: ‘De Rerum Natura’ N (Warminster).
-(ed.) (1991), Lucretius: ‘De Rerum Natura’ V I (Warminster).
Gow, A. S. F. (1931), ‘Diminutives in Augustan Poetry’, CQ 2 6 150-7.

422 BIBLIOGRAPHY
Goodyear, E R. D. (ed.) (1972), The Annals of Tacitus, I: Annals 1.1-54,

(Cambridge Classical Texts and Commentaries 15) (Cambridge).
-(ed.) (1981), The Annals of Tacitus, IZ: Annals 1.55-81 and Annals 2,
(Cambridge Classical Texts and Commentaries 23) (Cambridge).
Goold, G. F! (1974), Interpreting Catullus (London).
-(1983), Catullus, edited with introduction, translation and notes (London)
-(1990), Propertius (Cambridge, Mass.).
Gorler, W. (1982), ‘Beobachtungen zu Vergils Syntax’, Wiirzburger Jahrbiicher 8:
69-81.
-(1984), ‘Zum Virtus-Fragment des Lucilius (1326-1338 Marx) und zur Ge-
schichte der stoischen Guterlehre’, Hermes 12: 445-68.
-(1985), art. ‘Eneide, 6. La lingua’, in Enciclopedia Virgiliana ii. 262-78
(Rome).
Gransden, K. W. (ed.) (1991) Virgil Aeneid Book XI (Cambridge).
Gratwick, A. S. (1982), ‘The Satires of Ennius and Lucilius’, in Kenney and Clausen
(1982), 156-71.
Griffin, J. (1985), Latin Poets and Roman Life (London) (pp. 1-31 = JRS 66 [1976],
87-105).
Grilli, A. (1978), ‘Ennius podager’, RFIC 106: 34-8.
Groeber, G. (1884), ‘Vulgarlateinische Substrate romanischer Worter’, A L L 1:
204-54.
Guilbert, L. (1965), La formation du vocabulaire de l’aviation (Paris).
Haffter, H. (1934), Untersuchungen zur altlateinischen Dichtersprache (Problemata,
10) (Berlin).
-(1956), ‘Zum Problem der uberlangen Wortformen im Lateinischen’, WSt
69: 363-71.
Hahn, E. A. (1958), ‘Vergil’s linguistic treatment of divine beings, part II’, TAPA
89 237-53.
Hakamies, R. (1951), Etude sur l’origine et l’ivolution du diminutive latin et sa
survie dans les langues romanes (Helsinki).
Halm, C. (ed.) (1863), Rhetores Latini Minores (Leipzig).
Handford, S. A. (1947), The Latin Subjunctive. Its Usage and Development from
Plautus to Terence (London).
Hanslik, R. (1969), art. ‘Lucilius’, in Der kleine Pauly, vol. I11 (Stuttgart).
Hanssen, J. S. T. (1951), Latin Diminutives: a Semantic Study (Bergen).
Hardie, F! R. (ed.) (1994), Virgil, Aeneid, Book ZX (Cambridge).
Harrison, E. L. (1960), ‘Neglected hyperbole in Juvenal’, C R NS 1 0 99-101.
Harrison, S. J. (ed.) (1990), Oxford Readings in Vergil’s Aeneid (Oxford).
-(ed.) (1991), Vergil, Aeneid 10, with introduction, translation, and commen-
tary (Oxford).
Hartung, H. J. (1970), Ciceros Methode bei der Ubersetzung griechischer philosophi-
scher Termini (Diss. Hamburg).
Haupt, M. (1841), Observationes Criticae (Leipzig) (= 1875: 73-142).
-(1875), Opuscula I (Leipzig).
Heck, B. (1950), Die Anordnung der Gedichte des C. Valerius Catullus (Diss.
Tubingen).
Henry, A. (1971), Mktonymie et mktaphore (Paris).

BIBLIOGRAPHY 423
Herescu, N. I. (1960), La poksie latine. Etude des structures phoniques (Paris).

Heraeus, W. (1937), Kleine Schriften (Heidelberg).
Hermann, G. (1796), De metris poetarum Graecorum et Romanorum libri IZI
(Leipzig).
_ _ (1816), Elementa doctrinae metricae (Leipzig).
Hettrich, H. (1988), Untersuchungen zur Hypotaxe i m Vedischen (Berlin - New
York).
___ (1990), Der Agens in passivischen Satzen altindogermanischer Sprachen
(NAWG, 1. Philologisch-historische Klasse, Nr.2) (Gottingen).
Heusch, H. (1954), Das Archaische in der Sprache Catulls (Diss. Bonn).
Heurgon, J. (1959), Lucilius (Paris).
Heyne, C G. and Wagner, G. P. E. (edd.) (1830-33, 4th edn.), I? Virgili Maronis
opera. (Leipzig).
Highet, G. (1951), ‘Juvenal’s Bookcase’, A J P 72: 369-94.
_ _ (1954), Juvenal the Satirist. A Study (Oxford).
Hillen, M. (1989), Studien zur Dichtersprache Senecas. Abundanz. Explikativer
Ablativ. Hypallage (Untersuchungen zur antiken Literatur und Geschichte 32)
(Berlin - New York).
Hinds, S. E. (1987), ‘Language at breaking point: Lucretius 1.452’, C Q NS 37:
450-3.
Hofmann, J. B. (1951), Lateinische Umgangssprache. 3. Auflage (Heidelberg).
Hofmann, J. B. and Szantyr, A. (1965), Lateinische Syntax und Stilistik. (Handbuch
der Altertumswissenschaft I1 2.2) (Munich).
Holford-Strevens, L. (1988), Aulus Gellius (London).
Hollis, A. S. (ed.) (1977), Ovid, Ars Amatoria Book I, edited with an introduction
and commentary (Oxford).
Horsfall, N. (1971), ‘Numanus Regulus Ethnography and propaganda in Aen.
IX.598f.’, Latomus 3 0 1108-16 (= Hamson (1990: 127-44)).
-(1981), ‘Some problems of titulature in Roman literary history’, BICS 28:
103-11.
Housman, A. E. (1907), ‘Luciliana’, C Q 1: 51-74,148-59. (= Diggle and Goodyear
(1972) ii.662-97.)
Hunter, R. L. (ed.) (1989), Apollonius of Rhodes, Argonautica Book ZZZ
(Cambridge).
Hupe, C. (1871), De genere dicendi C. Valerii Catulli Veronensis. Pars I (Diss.
Munster).
Hutchinson, G. 0. (1988), Hellenistic Poetry (Oxford).
-(1993), Latin Literature from Seneca to Juvenal. A Critical Study (Oxford).
Ilberg, J. (1907), ‘A. Cornelius Celsus und die Medizin in Rom’, Neue Jahrbiicher
19: 377-412 (= Flashar (1971), 308-60).
Jacobson, H. (1974), Ovid’s Heroides (hinceton, N.J.).
Jal, A. (1861), Virgilius nauticus Etudes sur la marine antique (Paris).
Janni, P. (1967), ‘Due note omeriche’, QUCC 3: 7-30.
Janni, P. and Mazzini, I. (edd.) (1991), L a traduzione dei classici greci e latini
in Ztalia oggi. Problemi, prospettive, iniziative editoriali (Atti del Convegno
Nazionale, Macerata, 20-22 aprile 1989) (Macerata).
Janson, T. (1979), Mechanisms of Language Change in Latin (Stockholm).

424 BIBLIOGRAPHY
Janssen, H. H. (1941), De kenmerken der romeinsche dichtertaal (Nijmegen -

Utrecht).
Jenkyns, R. (1982), Three Classical Poets: Sappho, Catullus and Juvenal (London).
Jocelyn, H. D. (ed.) (1969a), The Tragedies of Ennius: the fragments edited with an
introduction and commentary (Cambridge).
-(1969b), ‘The fragments of Ennius’ Scenic Scripts’, A C 38: 181-217.
-(1971), ‘The Tragedies of Ennius’, Entretiens sur I’Antiquitk classique 17:
41-95 (Geneva).
-(1972), ‘The Poems of Quintus Ennius’, in H. Temporini (ed.), A N R W
1.2.987-1026 (Satires and minor works: 10226) (Berlin).
-(1977), ‘Ennius, Sat. 6-7 Vahlen’, RFZC 105: 131-51.
-(1979), ‘Catullus 58 and Ausonius, Ep. 71’, LCM 4 87-91.
-(1980), ‘Marcello Zicari and the poems of C. Valerius Catullus’, R P L 3:
55-72.
-(1986), ‘The new chapters of the ninth book of Celsus’ Artes’, P L L S 5:
299-336 (Liverpool).
-(1995), “zko Features of the Style of Catullus’ Phalaecian Epigrams’, Sileno
21: 63-82.
Jouanna, J. (1970), review of Lanata (1968), R E G 83: 254-7.
Jouanna, J. and Demont, P. (1981), ‘Le sens d’ ix&p chez Homkre (Iliade V, w. 340
et 416) et Eschyle (Agamemnon, v. 1480) en relation avec les emplois du mot
dans la Collection hippocratique’, R E A 83: 197-209.
Kaimio, J. (1979), The Romans and the Greek Language (CommentationesHuman.
Litterarum Soc. Scient. Fenn. 64)(Helsinki-Helsingfors).
Kaster, R. A. (ed.) (1995), C. Suetonius Tranquillus, De Grammaticis et Rhetoribus,
edited with a translation, introduction and commentary (Oxford).
Kenney, E. J. (1958), ‘Nequitiaepoeta’, in N. I. Herescu (ed.), Ovidiana. Recherches
sur Ovide, 201-9 (Paris).
-(1962), ‘The First Satire of Juvenal’, PCPS NS 8: 29-40.
-(1963), ‘Juvenal: Satirist or Rhetorician?’, Latomus 2 2 704-20.
-(ed.) (1971), Lucretius De Rerum Natura Book ZZZ (Cambridge).
-(1979), “Ibo disputed passages in the Heroides’, C Q NS 2 9 394-431.
-(ed.) (1996), Ovid Heroides XVZ-XXZ (Cambridge).
Kenney, E. J. and Clausen, W. V. (edd.) (1982), The Cambridge History of Classical
Literature, ii, Latin Literature (Cambridge).
Kingsley, P. (1995), Ancient Philosophy, Mystery, and Magic: Empedocles and
Pythagorean Tradition (Oxford).
Knoche, U. (1982; 4th ed.), Die romische Satire (Gottingen).
Knox, P. E. (1986), ‘Ovid’s Metamorphoses and the traditions of Augustan poetry’,
PCPS Suppl. 11 (Cambridge).
-(ed.) (1995) Ovid Heroides. Select Epistles (Cambridge).
Koch, P. (1995), ‘Latin vulgaire et traits universels de l’oral’, in Callebat (1995:
125-44).
Korfmacher, W. Ch. (1935), ‘Grecizing in Lucilian Satire’, CJ 3 0 453-62.
Korzeniewski, D. (ed.) (1970), Die romische Satire (Wege der Forschung 238)
(Darmstadt).
Krenkel, W. (ed.) (1970 2 vols), Lucilius, Satiren. Lateinisch und Deutsch (Leiden).

BIBLIOGRAPHY 425
Krause, H. (1878), De Vergilii usurpatione infinitivi (Diss. Halle).

Kroll, W. (1912), ‘Der lateinische Relativsatz’, Glotta 3: 1-18.
_ _ (1913) (repr. 1958), M. Tullii Ciceronis Orator. A l s Ersatz der Ausgabe von
Otto Jahn. Erklart von W. K. (Berlin).
_ _ (1924), Studien zum Verstundnis der romischen Literatur (Stuttgart).
-(1925), 3rd ed., repr. 1969, Die wissenschaftliche Syntax i m lateinischen
Unterricht (Dublin).
-(1929), 2nd ed., 1st ed. 1922, reprinted with addenda, 1968), C. Valerius
Catullus (Stuttgart).
Kudlien, E (1963), Untersuchungen zu Aretaios von Kappadokien (Maim).
Kiihner, R. and Stegmann, C. (edd.) (1955; 3rd ed. by A. Thierfelder, 2 vols),
Ausfuhrliche Grammatik der lateinischen Sprache (Darmstadt).
Labate, M. (1984), L‘arte di farsi amare. Modelli culturali e progetto didascalico
nell’elegia ovidiana. (Biblioteca di ‘Materiali e discussioni per l’analisi dei Testi
classici’, 2) (Pisa).
Lachmann, K. (1848), ‘De Ovidii epistulis’, Prooemium indicis lectionum aestiv-
arum a. 1848 = Kleinere Schrijien zur classichen Philologie, ed. J. Vahlen, 56-61
(Berlin).
Lakoff, G. and Johnson, M. (1980), Metaphors We Live By (Chicago).
Lakoff, G. and Tbmer, M. (1989), More than Cool Reason: a Field Guide to Poetic
Metaphor (Chicago).
Lanata, G. (1966), ‘Sul linguaggio amoroso di Saffo’, QUCC 2: 63-79.
-(1968), ‘Linguaggio scientific0 e linguaggio poetico. Note a1 lessico del De
morbo sacro’, QUCC 5: 22-36.
Landgraf, G. (1898), ‘Der Accusativ der Beziehung (determinationis)’, A L L 10:
209-24.
-(1914, 2nd ed.), Kommentar zu Ciceros Rede Pro Sex. Roscio Amerino
(Leipzig-Berlin)
Langslow, D. R. (1989), ‘Latin technical language: synonyms and Greek words in
Latin medical terminology’, TPhS 87: 33-53.
-(1991), ‘The development of Latin medical terminology: some working
hypotheses’, PCPS NS 37: 106-30.
La Penna, A. (1951), ‘Note sul linguaggio erotic0 dell’elegia latino’, Maia 4
187-209.
-(1956a), review of Heusch (1954), Gnomon 28: 2914.
-(1956b), ‘Problemi di stile catulliano’, Maia 8: 141-60.
Lateiner, D. (1977), ‘Obscenity in Catullus’, Ramus 6: 15-32.
Lausberg, M. (1990), ‘Epos und Lehrgedicht. Ein Gattungsvergleich am Beispiel
von Lucans Schlangenkatalog’, Wurzburger Jahrbucher 16: 173-203.
Lavency, M. and LongrBe, D. (edd.) (1989), Actes du Ir. Colloque de Linguistique
latine (Louvain-la-Neuve/ BorzCe, 31 March4 April 1989) (Cahiers de l’lnstitut
de linguistique de Louvain 15.14) (Louvain-la-Neuve).
Leavis, E R. (1948,2nd ed.), Education and the University, a sketch for an ‘English
School’ (London).
Lebreton, J. (1901), Etudes sur le langage et la grammaire de Cictron (Paris).
Lee, A.G. (1975), Tibullus: Elegies (Cambridge).

426 BIBLIOGRAPHY
-(ed.) (1990), The Poems of Catullus, Edited with an Introduction, Translation

and Brief Notes (Oxford).
Lehmann, C. (1979), ‘Der Relativsatz vom Indogermanischen bis zum Italieni-
schen. Eine Etude in diachroner syntaktischer Typologie’, Die Sprache 25:
1-25.
-(1984), Der Relativsatz (Tlibingen).
Lehmann, Y. (1982), ‘Varron et la medecine’, in Sabbah (1982), 67-72.
Leishman, J. B. (1956), Translating Horace (Oxford).
Lelikvre, E J. (1958), ‘Parody in Juvenal and T. S. Eliot’, CPh 53: 22-6.
Leo, F. (1896), Analecta Plautina de figuris sermonis I (Progr. Gottingen) =
Fraenkel, E. [ed.], [1960] Friedrich Leo. Ausgewiihlte kleine Schriften. Erster
Band Zur romischen Literatur des Zeitalters der Republik 71-122 (Rome).
-(1906), ‘review of Lucilii carminum reliquiae ed. Marx, vol. 1-11, GGA:
837-61 (= Fraenkel, E. [ed.] [1960], Friedrich Leo. Ausgewiihlte kleine Schriften.
Erster Band Zur romischen Literatur des Zeitalters der Republik: 221-247
(Rome)).
-(1967), Geschichte der romischen Literatur. Erster Band: Die archaische
Literatur. Im Anhang: ‘Die romische Poesie in der SullaNschen Zeit’
(Darmstadt) (= Unveranderter Nachdruck der Ausgabe Berlin 1913).
Leumann, M. (1947), ‘Die lateinische Dichtersprache’, M H 4 116-39 = Kleine
Schriften (Zurich-Stuttgart 1959) 131-56 = Lunelli (1980) 131-78.
-(1950), Homerische Worter (Basel).
-(1977, 6th ed.), Lateinische Laut- und Formenlehre (Handbuch der Altert-
umswissenschaft I1 2.1) (Munich).
Levinson, S. C. (1983), Pragmatics (Cambridge).
LCvy, C. (1992), ‘CicCron crCateur du vocabulaire latin de la connaissance: essai
de synthkse’, in La langue latine, langue de la philosophie (Ecole fraqaise de
Rome, 161) (Rome).
Lewis, N. (1959), Samothrace, the Ancient Literary Sources (London).
Leyhausen, J. (1893), Helenae et Herus epistulae Ovidii non sunt (Diss. Halle).
Linde, P. (1923), ‘Die Stellung des Verbs in der lateinischen Prosa’, Glotta 12:
153-78.
Lindsay, W. M. (1893), ‘The Saturnian metre’, R I P 14: 139-70, 305-34.
-(1907), Syntax of Plautus (Oxford).
-(ed.) (1913; repr. Hildesheim 1978), Sexti Pompeii Festi de verborum signifi-
catu quae supersunt cum Pauli epitome (Leipzig).
-(1922), Early Latin Verse (Oxford).
Linse, E. (1891), De li Ovidio Nasone vocabulorum inventore (Progr. Dortmund).
Lijfstedt, B. (1990), ‘Notizen zu Sprache und Text von Celsus, De medicina’, M H
47: 60-2.
Lijfstedt, E. (1911), Philologischer Kommentar zur ‘Peregrinatio Aetheriae’. Unter-
suchungen zur Geschichte der lateinischen Sprache (Uppsala).
-(1928 [vol. 11; 1933 [vol. 21; 1942 [2nd ed. of vol. l]),Syntactica. Studien und
Beitriige zur historhchen Syntax des Lateins I & IZ (Lund).
-(1959), Late Latin (Oslo).
Lohmann, A. (1915), De Graecismorum Vergiliano usu quaestiones selectae (Diss.
Munster).

BIBLIOGRAPHY 427
Long, A. A., and Sedley, D. N. (1987, 2 vols.), The Hellenistic Philosophers

(Cambridge).
Lot, E (1946), ‘La langue du commandement dam les armCes romaines’, in Mk-
langes dkdiks U la mkmoire de E Grat (Paris).
Luck-Huyse, K. (1996), Der Traum vom Fliegen in der Antike (Palingenesia 62)
(Stuttgart).
Lunelli, A. (ed.) (1980,2nd ed.), La lingua poetica latina (contains Italian versions
of Janssen (1941) and Leumann (1947) with updated bibliography and
annotations) (Bologna).
Lyne, R. 0. A. M. (1980), The Latin Love Poets:from Catullus to Horace (Oxford).
-(1989), Words and the Poet: Characteristic Techniques of Style in Vergil’s
Aeneid (Oxford).
McGlynn, P. (1963, 2 vols), Lexicon Terentianum (Glasgow).
McKeown, J. C. (ed.) (1987), Ovid: Amores. Text, Prolegomena and Commentary
(Liverpool).
McKie, D. (1984), ‘The horrible and ultimate Britons: Catullus, 11.11’, PCPS NS
3 0 74-8.
Madvig, J. N. (ed.) (1869), M. Tullii Ciceronis De finibus bonorum et malorum.
2nd ed. (Copenhagen).
Maltby, R. (1991), A Lexicon of Ancient Latin Etymologies (Leeds).
-(1993), ‘The Limits of Etymologising’, Aevum Antiquum 6 257-75.
Marache, R. (1964), ‘RhCtorique et humour chez JuvCnal’, in Renard and Schilling
(1964), 474-8.
Marganne, M.-H. (1993), L’ophtalmologie dans 1’Egypte grkco-romaine d’aprss les
papyrus littkraires grecs (Studies in Ancient Medicine, 8) (Leiden).
Marichal, R. (1992), Les ostraca de B u Njem (SupplCments de ‘Libya Antiqua’ 7)
(Tripoli).
Mariner, S. (1963), ‘Traiectus lora (Virg. En. I1 273)’, Estudios Clasicos 7: 107-19.
Mariotti, I. (1954), ‘I grecismi di Lucilio’, Stud. Urb. 28: 357-86.
-(1960), Studi Luciliani (Florence).
Mariotti, S. (1991, 2nd ed.), Lezioni su Ennio (Urbino).
Marouzeau, J. (1907), Place du pronom personnel sujet en latin (Paris).
-(1922), L‘ordre des mots dans la phrase latine, I: Les groupes nominaux
(Paris).
-(1949a), L‘ordre des mots dans la phrase latine, IIZ: Les articulations de
l’knonck (Paris).
(1949b), Quelques aspects de la formation du latin litte‘raire (Collection
linguistique 53) (Paris).
-(1962; 4th ed.), Traitk de stylistique latine (Paris).
Marshall, l? K. (ed.) (1968, 2 vols), A. Gellii Noctes Atticae (Oxford).
Martyn, J. R. C. (1979), ‘Juvenal’s Wit’, Grazer Beitrage 8 219-38.
Man, E (1882), Studia Luciliana. Diss. Bonn.
-(ed.) (1904, 1905), C. Lucilii carminum reliquiae. Vol. prius: Prolegomena,
testimonia, Fasti Luciliani, carminum reliquiae, indices, Vol. posterius: Com-
mentarius (Leipzig).
-(1909), ‘Die Beziehungen des Altlateins zum Spatlatein’, NJb. 4 d. class.
Alterturn: 434-48.

428 BIBLIOGRAPHY
-(ed.) (1915), A . Cornelii Celsi quae supersunt (CML, i; Leipzig-Berlin).

Mason, H. A. (1963), ‘Is Juvenal a Classic?’, in Sullivan (1963), 93-176.
Maurach, G. (1975), ‘Ovid, Met. I, 48 und die Figur der “Umkehrung” ’, Hermes
103: 47946.
Mayer, R. G. (1983), ‘Catullus’ divorce’, CQ 33: 297-8.
-(ed.) (1994), Horace, Epistles, Book Z (Cambridge).
Mazzini, I. (1988), ‘La medicina nella letteratura latina. I. Osservazioni e proposte
interpretative su passi di Lucilio, Lucrezio, Catullo e Orazio’, Aufdus 4 45-73.
-(1990), ‘I1 folle da amore’, in Alfonso et al. (1990), 3943.
-(1991a), ‘La medicina nella letteratura latina. 11. Esegesi e traduzione di
Horat. Epod. 11, 15-16 e Od. I 13,4-5’, in Janni and Mazzini (1991), 99-114.
-(1991b), ‘I1 lessico medico latino antico: caratteri e strumenti della sua
differenziazione’, in Sabbah (1991), 175-85.
-(1992a), ‘La medicina nella letteratura latina. 111. Plauto: conoscenze
mediche, situazione e istituzioni sanitarie, proposte esegetiche’, in Mazzini
(1992b), 67-113.
-(ed.) (1992b), Civilta materiale e letteratura ne1 mondo antico (Atti del
Seminario di Studio, Macerata, 28-29 giugno 1991) (Macerata).
Meillet, A. (1965; 7th ed.), A p e r p d’une histoire de la langue grecque (Paris).
Menitre, P. (1858), Etudes rntdicales sur les po2tes latins (Paris).
Mette, H. J. (1956), rev. of E. V. Marmorale, L‘ultimo Catullo, Gnomon 2 8 34-8
(part repr. in R. Heine (ed.) [1975] Catull [Wege der Forschung 308, Darm-
stadt]: 19-23).
Meyer, W. (1889), ‘Caesur im Hendekasyllabus’, SB Bayr. Ak., phi1osoph.-philol.
und hist. Cl. 2: 208-27.
Migliorini, P. (199O), La terminologia medica come strumento espressivo della satira
di Persio (Quaderni di Anazetesis 2) (Pistoia).
Mignot, X. (1969), Les verbes dtnominatifs latins (Paris).
Miller, H.W. (1944), ‘Medical terminology in tragedy’, TAPA 75: 156-67.
-(1945), ‘Aristophanes and medical language’, TAPA 76: 74-84.
Mohler, S. L. (1948), ‘Sails and Oars in the Aeneid’, TAPA 79: 46-62.
Momigliano, A. (1957), ‘Perizonius, Niebuhr and the character of the early Roman
tradition’, JRS 47: 104-14.
Morford, M. I? 0. (1972), ‘A Note on Juvenal 6.627-61’, CPh 67: 198.
Mras, K. (1927/28), ‘Randbemerkungen zu Lucilius’ Satiren’, WS 4 6 78-84.
Mudry, Ph. (1982), La prtface du De medicina de Celse: Texte, traduction et
commentaire (Bibliotheca Helvetica Romana 19) (Rome).
Muhmelt, M. (1965), Griechische Grammatik in der Vergilerklarung, (Zetemata
37) (Munich). >
Muller, C. E W. (1869), Plautinische Prosodie (Berlin).
(1908), Syntax des Nominativs und Akkusativs im Lateinischen (Leipzig and
Berlin).
Muller, C.W., Sier, K. and Werner, J. (edd.) (1992), Zum Umgang mit fremden
Sprachen in der griechisch-romischenAntike (Palingenesia 3 6 Kolloquium der
Fachrichtungen Klassische Philologie der Universitaten Leipzig und Saar-
briicken am 21. und 22. November 1989 in Saarbriicken) (Stuttgart).

BIBLIOGRAPHY 429
Miiller, H. M. (1980), Erotische Motive in der griechischen Dichtung bis auf Euri-
pides (Hamburg).
Miiller, K. (ed.) (1975), 1: Lucreti Cari: De rerum natura libri sex (Zurich).
Miinscher, K. (1921), ‘Metrische Beitrage 11. Erstarrte Formen im Versbau der
Aiolier’, Hermes 5 6 66-103.
Munari, E (1971), ‘Textkritisches zu mittellateinischen Dichtern’ in Coseriu, E.
and Stempel, W.-D. (edd.) Festschrift f i r Harri Meier zum 65. Geburtstag
(Munich).
Murgatroyd, F’. (1980), Tibullus Z: A Commentary (Pietermaritzburg).
-(1994), Tibullus: Elegies ZZ (Oxford).
Myers, R. and Ormsby, R. J. (1970), Catullus The Complete Poems for Modern
Readers (New York).
Myers-Scotton, C. (1993), Duelling Languages. Grammatical Structure in Code-
switching (Oxford).
Mynors, R. A. B. (ed.) (1958), C. Valerii Catulli Carmina (Oxford).
-(ed.) (1990), Virgil, Georgia, edited with an introduction and commentary
(Oxford).
Nagle, B. R. (1980), The Poetics of Exile: Program and Polemic in the Tristia and
Epistulae ex Ponto of Ovid. (Collection Latomus, 170) (Brussels).
Naiditch, P. G. (1988), ‘Three notes on “Housman and Ennius”’ Housman Society
Journal 1 4 46-9.
Naylor, H. D. (1922), Horace, Odes and Epodes: A Study in Poetic Word-Order
(Cambridge).
Neue, E and Wagener, C. (1892-1905; 3rd ed.), Formenlehre der lateinischen
Sprache (Berlin).
Neumann, G. (1968), ‘Sprachnormung im klassischen Latein’, Sprache der Ge-
gegenwart 2: 88-97.
Newman, J. K. (1990), Roman Catullus and the Modification of the Alexandrian
Sensibility (Hildesheim).
Nilsson, N.-0. (1952), Metrische Stildifferenzen in den Satiren des Horaz
(Stockholm).
Nisbet, R. G. M. (1978), ‘Notes on the text of Catullus’, PCPS NS 2 4 92-115
(=1995: 76-100).
-(1995), S. J. Harrison (ed.), Collected Papers on Latin Literature (Oxford).
Nisbet, R. G. M. and Hubbard, M. (1970), A Commentary on Horace: Odes Book
1 (Oxford).
-(1978), A Commentary on Horace: Odes Book ZZ (Oxford).
Norden, E. (ed.) (1903; 1957, repr. of 2nd ed., 1915), l? Vergilius Maro, Aeneis
Buch VZ (Leipzig and Stuttgart).
-(ed.) (1910), Einleitung in die Altertumswissenschaft (Berlin).
Nowottny, W. (1962), The Language Poets Use (London).
Nutton, V. (1993), ‘Roman medicine: tradition, confrontation, assimilation’, in H.
Temporini (ed.), ANRW, 11.37: 1,49-78 (Berlin).
Onnerfors, A. (1963), Zn Medicinam Plinii studia philologica (Lunds Univ.
hsskrift. N.F. Avd. 1. Bd 55, Nr 5) (Lund).
_ _ (1989), ‘Dare und AurdAuricula im Spatlatein’, Symb. Osl. 64: 130-57.

430 BIBLIOGRAPHY
-(1993), ‘Das medizinische Latein von Celsus bis Cassius Felix’, in H. Tempo-
rini (ed.), A N R W 11.37: 1, 227-392 (Berlin).
Ortony, A. (1979), Metaphor and Thought (Cambridge).
Paganelli, D. (1961), Properce: Eligies (Paris).
Page, D. L. (1936), review of Dumortier (1935), C R 50: 17-18.
Palmer, A. (ed.), (1898) P Ovidi Nasonis Heroides with the Greek translation of
Planudes [Completed by L. C. Purser.] (Oxford).
Palmer, L. R. (1954), The Latin Language (London).
Paludan, E. (1941), ‘The development of the Latin elegy’, ClMed 4 204-29.
Pascucci, G. (1961), ‘consens, praesens, absens’, SZFC 33: 1-61.
Pasquali, G. (1981), Preistoria della poesia romana: con un saggio introduttivo di
Sebastian0 Timpanaro (Florence).
Patzer, H. (1955), ‘Zum Sprachstil des neoterischen Hexameters’, M H 1 2 77-95.
Pearce, T. E. V. (1966), ‘The enclosing word order in the Latin hexameter’ CQ NS
1 6 140-71; 298-320.
Peppler, C. W. (1910), ‘The termination -KOS, as used by Aristophanes for comic
effect’, R I P 31: 4 2 W .
Peter, H. (1901), Der Brief in der romischen Literatur (Leipzig).
Petersmann, H. (1986), ‘Der Begriff satura und die Entstehung der Gattung’, in
Adamietz (1986), 7-24.
-(1989), ‘Die Urbanisierung des romischen Reiches im Lichte der lateinischen
Sprache’, Glotta 9 6 406-28.
_ _ (1992), ‘Vulgarlateinisches aus Byzanz’ in Muller, C. W. et al. (1992), 219-31.
-(1995a), ‘Soziale und lokale Aspekte in der Vulgarsprache Petrons’, in
Callebat (1995), 53347.
-(1995b), ‘Zur mundlichen Charakterisierung des Fremden in der Komodie
des Plautus’, in Benz et al. (1995), 123-36.
-(forthcoming), ‘Language and style as means of characterization in the
comedies of Plautus’, Papers of the Leeds International Latin Seminar.
Phillips, J. H. (1984), ‘Lucretius and the (Hippocratic) On Breaths: Addenda’, in
Sabbah (1984), 83-5.
Pigeaud, J. (1980), ‘La physiologie de Lucrkce’, R E L 58: 176200.
-(1982), ‘Virgile et la mCdecine. Quelques rCflexions sur l’utilisation de la
pensCe physiologique dans les GCorgiques’, Helmantica 33: 539-60.
-(1988), ‘Die Medizin in der Lehrdichtung des Lukrez und des Vergil’, in
Binder (1988), 216-39.
Pinkster, H. (1987), ‘The pragmatic motivation for the use of subject pronouns in
Latin: the case of Petronius’, in Etudes de linguistique gintrale et de linguistique
lathe offertes en hommage a Guy Serbat, 369-79 (Paris).
Pinotti, F’. (ed.) (1988), Publio Ovidio Nasone, Remedia Amoris (Edizioni e sag@
universitari di filologia classica, 39) (Bologna).
Platnauer, M. (1951), Latin Elegiac Verse (Cambridge).
Ploen, H. (1882), De copiae verborum differentiis inter varia poesis Romanae
antiquiorb genera intercedentibus (Diss. Strasbourg).
Poncelet, R. (1957), Ciciron traducteur de Platon. L’expression de la penske c o n -
plexe en latin classique (Pans).

BIBLIOGRAPHY 431
Powell, J. G. F. (1987), ‘The farrago of Juvenal 1.86 reconsidered’, in Whitby,

Hardie and Whitby (1987).
-(ed.) (1988), Cicero: Cat0 Maior De Senectute (Cambridge).
-(199%) ‘Cicero’s translations from Greek’, in Powell (1995b), 273-300.
___ (ed.) (1995b), Cicero the Philosopher (Oxford).
Puelma Piwonka, M . (1949), Lucilius und Kallimachos. Zur Geschichte einer
Gattung der hellenistisch-romischen Poesie (Frankfurt am Main).
Pye, D. W. (1963), ‘Latin 3rd plural perfect indicative active - Its endings in verse
usage’, TPhS: 1-27
Radermacher, L. (1951), Artium Scriptores (Reste der voraristotelischen Rhetorik).
(Osterr. Akademie der Wissenschaften, phi1.-hist. Kl., Sitzungsberichte, 227.
Bd., 3. Abh.) (Vienna).
Ramage, E. S. (1957), Urbanitas, rusticitas, peregrinitas: the Roman view of proper
Latin (Cincinnati).
Rand, E. K. (1925), Ovid and his Influence (London, Calcutta, Sydney).
Rawson, E. D. (1969), The Spartan Tradition in European Thought (Oxford).
-(1985), Intellectual Life in the Late Roman Republic (London).
Reichenkron, G. (1961), ‘Zur rdmischen Kommandosprache bei byzantinischen
Schriftstellern’, Byz. Zeitschr. 5 4 18-27
Reitzenstein, R. (1893), Epigramm und Skolion. Ein Beitrag zur Geschichte der
alexandrinischen Dichtung (Giessen).
(1907), art. ‘Epigramm’, R E 6.1: 71-111.
-(1912), Zur Sprache der lateinischen Erotik (Sitzungsb. d. Heidelberger Ak.
d. Wiss., Phi1.-hist. Kl., 12. Abh.) (Heidelberg).
Renard, M., and Schilling, R. (edd.) (1964), Hommages U Jean Bayet, (Collection
Latomus 70) (Brussels).
Riemann, 0. (1885; 2nd ed.), Etudes sur la langue et la grammaire de Tite-Live
(Paris).
Risch, E. (1984), Gerundivum und Gerundium. Gebrauch i m klassischen und
iilteren Latein. Entstehung und Vorgeschichte (Berlin-New York).
Risselada, R (1993), Imperatives and Other Directive Expressions in Latin
(Amsterdam).
Roby, H. J. (18%), A grammar of the Latin language from Plautus to Suetonius.
Part I1 Syntax (London).
Romaine, S. (1995; ed. 1, 1989), Bilingualism (Oxford).
Romano, A. C. (1979), Irony in Juvenal (Hildesheim and New York).
Ronconi, A. (1938), ‘Stile e lingua di Catullo’, A & R 111 6 139-56 (= 1950: 23-47).
-(1939), ‘Allitterazionee stile in Catullo’, Stud. Urb. 13B: 1-77 (= 1953: 9-82 =
1971: 11-86).
-(19&), ‘Per la storia del diminutive latino. Studi esegetici e stilistici’, Stud.
Urb. 14B: 1-45 (= 1953: 107-50 = 1971: 87-130).
-(194Ob), ‘Atteggiamenti e forme della parodia catulliana’, A & R 111 8:
141-58 (= 1953: 193-212 = 1971: 173-92).
-(1950), Da Lucrezio a Tacit0 (Messina-Florence).
-(1971; ed. 1, 1953), Studi catulliani (Bari-Brescia).
van Rooy, C. A. (1965), Studies in Classical Satire and Related Literary Theory
(Leiden).

432 BIBLIOGRAPHY
Rosler, W. (1989), ‘Typenhauser bei Aischylos?’, in Schuller et al. (1989), 109-14.

Ross, D. 0. (1969), Style and Tradition in Catullus (Cambridge, Mass.)
Rothstein, M. (1966; 3rd ed.), Sextus Propertim: Elegien (Dublin - Zurich).
Rudd, N. (1960), ‘Horace on the origins of satura’, Phoenix 1 4 36-44.
-(1986), Themes in Roman Satire (London).
Ruijgh, C. J. (1957), L‘tltment achten dans la langue tpique (Assen).
Sabbah, G. (ed.) (1982), Mtdecins et mtdecine dans l‘antiquitt (Centre Jean
Palerne: MCmoires, iii) (Saint-Etienne).
-(ed.) (1984), Textes mtdicaux latins antiques (Centre Jean Palerne:
Mtmoires, v) (Saint-Etienne).
-(ed.) (1991), Le latin mtdical. La constitution d’un langage scientifique
(Centre Jean Palerne: MCmoires, x) (Saint-Etienne).
Safarewicz, J. (1965), ‘Uwagi o jezyku Lucyliusza’, Eos 55: 96-105.
Sager, J. C., Dungworth, D. and McDonald, P. E (1980), English Special Languages:
Principles and Practice in Science and Technology (Wiesbaden).
de Saint-Denis, E. (1935), Le r8le de la mer dans la potsie latine (Paris).
-(1965), Essais sur le rire et le sourire des Latins (Paris).
Schaublin, C. (1988), ‘Housman and Ennius’, Housman Society Journal 1 4 42-5.
Schawaller, D. (1987), ‘Semantische Wortspiele in Ovids Metamorphosen und
Heroides’, Griizer Beitriige 1 4 199-214.
Scherer, A. (1963), ‘Die Sprache des Archilochos’, in Entretiens sur I’Antiquitt
classique 1 0 89-107 (Geneva).
Schmid, P. (1964), ‘JuvCnal. Essai d’une dCfinition stylistique’. RCsumC, in REL 4 2
57-9.
Schmid, W. and Stahlin, 0. (1929), Geschichte der griechischen Literatur, I : i
(Munich).
Schmidt, B. (ed.) (1887), C. Valeri Catulli Veronensis carmina (Leipzig).
-(1914), ‘Die Lebenszeit Catulls und die Herausgabe seiner Gedichte’, RhM
69: 267-83.
Schmidt, E.A. (1977), ‘Lucilius kritisiert Ennius und andere Dichter. Zu Lucilius
fr. 148 Man’, MH 3 4 122-9.
-(1985), Catull (Heidelberg).
Schmitt, R. (1967), Dichtung und Dichtersprache im indogermanischer Zeit
(Wiesbaden).
Scholte, A. (ed.) (1933), Publii Ovidii Nasonis Ex Ponto Liber Prirnm commentario
exegetic0 instructus (Amersfoort).
Scholz, U.W. (1986a), ‘Der fruhe Lucilius und Horaz’, Hermes 114: 335-65.
-(1986b), ‘Die satura des Q. Ennius’, in Adamietz (1986), 25-53.
Schreiber, G. (1917), De Lucili syntaxi (Diss. Greifswald).
Schunke, E. (1906), De traiectione coniunctionum et pronominis relativi apud poetas
Latinos (Diss. Kiel).
Schuller, W., Hoepfner, W. and Schwandner, E. L. (edd.) (1989), Demokratie und
Architektur: Der hippodamische Stadtebau und die Entstehung der Demokratie
(Konstanzer Symposion vom 17. bis 19. Juli 1987) (Munich).
Schulze, K. F! (1920), ‘Bericht uber die Literatur zu Catullus fur die Jahre
1905-1920’, Bursians Jahresb. 183: 1-72.
Schuster, M. (1948), art. ‘(123) C. Valerius Catullus’, RE 11.7.2: 2353-410.

BIBLIOGRAPHY 433
___ (ed.) (1949), Catulli Veronensis liber (Leipzig).

Schweizer, H. J. (1967), Vergil und Italien (Aarau).
Sconocchia, S. (ed.) (1983), Scribonii Largi Compositiones (Leipzig).
-(1993), ‘L’opera di Scribonio Largo e la letteratura medica latina del 1. sec.
d. C.’, in H. Temporini (ed.), A N R W 11.37: 1, 843-922. (Berlin).
Scott (Ryberg), I. G. (1927), The Grand Style in the Satires of Juvenal (Smith
College Classical Studies 8) (Northampton, Mass).
Sebeok, T. A. (ed.) (1960), Style in Language (Cambridge, Mass.).
Sedley, D. N. (1988), ‘Epicurean anti-reductionism’,in Barnes and Mignucci (1988),
295-327.
_ _ (1989), ‘The proems of Empedocles and Lucretius’, G R B S 30: 269-96.
-(1992) ‘Sextus Empiricus and the atomist criteria of truth’, Elenchos 13:
21-56.
Segal, C. (1990), Lucretius on Death and Anxiety (Princeton).
Segebade, J. (1895), Vergil als Seemann. Ein Beitrag zur Erklarung und Wiirdigung
des Dichters. Progr.d.Gymn. (Oldenburg).
Shackleton Bailey, D. R. (ed.) (1965), Cicero’s Letters to Atticus I1 58-54 B.C.
4 6 9 3 (Books IIZ and W ) (Cambridge).
-(ed.) (1977), Cicero: Epistulae ad Familiares. I 62-47 B.C. (Cambridge).
-(1992), ‘Homoeoteleuton in non-dactylic Latin verse’, RFZC 120 67-71.
_ _ (1994), Homoeoteleuton in Latin Dactylic Verse (Stuttgart-Leipzig).
Sharrock, A. R. (1994), Seduction and Repetition in Ovidk Ars Amatoria 2
(Oxford).
Shipley,E W. (1911), ‘The heroic clausula in Cicero and Quintilian’, CPh 6 410-18.
Silk, M. S. (1974), Interaction in Poetic Imagery with Special Reference to Early
Greek Poetry (Cambridge).
Simpson, E l? (1879), Select Poems of Catullus (London).
Skutsch, E (1892), Plautinisches und Romanisches Studien zur plautinischen Pros-
odie (Leipzig).
Skutsch, 0. (1934), Prosodische und metrische Gesetze dei Zambenkiirzung
(Forschungen z. griech. U. latein. Grammatik 10) (Gottingen).
(1964), ‘Rhyme in Horace’, BZCS 11: 73-8.
-(1969), ‘Metrical variations and some textual problems in Catullus, BZCS
1 6 3843.
-(1976), ‘Notes on Catullus’, BZCS 23: 18-22.
-(1980), ‘Catullus 58.4-5’, LCM 5: 21.
-(1985), The ‘Annals’ of Quintus Ennius edited with Introduction and Com-
mentary (Oxford).
Smith, K. F. (1913), The Elegies of Albius Tibullus (New York).
Smith, W. S. (ed.) (1989), ‘Heroic models for the sordid present: Juvenal’s view of
tragedy’, in H. Temporini (ed.), A N R W 11.33.1: 811-23 (Berlin).
Soubiran, J. (1966), L‘klision duns la poksie latine (Paris).
Spies, A. (1930), Militat omnis amans (Diss. mbingen).
von Staden, H. (1989), Herophilus: the Art of Medicine in Early Alexandria
(Cambridge).
Stevens, E. B. (1953), ‘Uses of hyperbaton in Latin poetry’, ClW 4 6 200-5.
Sullivan, J. P. (ed.) (1963), Critical Essays on Roman Literature: Satire (London).

434 BIBLIOGRAPHY
Summers, W. C. (1910), Select Letters of Seneca edited with introductions and

explanatory notes (London).
Svennung, J. (1935), Untersuchungen zu Palladius und zur lateinischen Fach- und
Volkssprache (Uppsala).
-(1945), Catulls Bildersprache. Vergleichende Stilstudien I (Uppsala Universi-
tets Arsskrift 3) (Uppsala-Leipzig).
Swanson, D. C. (1962), A Formal Analysis of Lucretius’ Vocabulary (Minneapolis).
Syndikus, H. P. (1984), Catull. Eine Interpretation. Erster Teil. Die kleinen Gedichte
(1-60) (Darmstadt).
Terzaghi, N. (ed.) (1934, 2nd ed.), Lucilio, (lbrin) (Repr. Hildesheim, New York
1979).
_ _ (ed.) (1966), Saturarum Reliquiae (Florence).
Thierfelder, A. (1955), ‘De morbo hepatiario’, RhM 98: 190-2.
Thill, A. (1979), Alter ab illo. Recherches sur l’imitation duns la potsie personnelle
a l’kpoque Augustkenne (Paris).
Thomas, R. E (ed.) (1988, 2 vols), Virgil, Georgics (Cambridge).
Thomson, D. F. S. (ed.) (1978), C a t u l l u A Critical Edition. Edited and Introduced
(Chapel Hill).
Tovar, A. (1969), ‘Lucilio y e1 lath de Espaiia’, in Studi linguistici in onore de t!
Pisani, ii.1019-32 (Brescia).
Townend, G. B. (1973), ‘The literary substrata to Juvenal’s satires’, JRS 63: 14-0.
Tracy, V A. (1971), ‘The authenticity of Heroides 16-21’, CJ 66: 328-30.
Trshkle, H. (1960), Die Sprachkunst des Properz und die Tradition der lateinischen
Dichtersprache (Hermes Einzelschriften 15) (Wiesbaden).
-(1967a), ‘Ausdrucksfulle bei Catull’, Philologus 111: 198-211.
-(1967b), ‘Neoterische Kleinigkeiten’, MH 24: 87-103.
-(1981), ‘Catullprobleme’,M H 3 8 245-58.
Traina, A. (1975), ‘Orazio e Catullo’ in Poeti latini (e neolatini). Note e saggi
jilologici: 253-75 (Bologna).
Untermann, J. (1971), ‘Entwiirfe zu einer Enniusgrammatik’, Entretiens de la Fon-
dation Hardt 17: 209-51 (Geneva).
-(1977), ‘Zur semantischen Organisation des lateinischen Wortschatzes’,
Gymnasium 8 4 313-39.
Vaananen, V. (1966, 3rd ed.), L e Latin vulgaire des inscriptions pomptiennes
(Berlin).
Vairel-Carron, H. (1975), Exclamation. Ordre et dtfense (Paris).
Van Sickle, J. B. (1968), ‘About form and feeling in Catullus 65’, TAPA 99:487-508.
Vechner, D. (1610, ed. 1, Frankfurt; ed. 2 Strasburg 1630; ed. 3 Leipzig 1680 ed.
4 Gotha 1733 (Heusinger)), Hellenolexia.
Vessey, D. W. T. C. (1969), ‘Notes on Ovid, Heroides 9’, CQ NS 19: 349-61.
Vetter, E. (1953), Handbuch der italischen Dialekte, I. Band: Texte mit Erklarung,
Glossen, Worterverzeichnis (Heidelberg).
Vollmer, E (1923), Romische Metrik, in A. Gercke and E. Norden (edd.), Einleitung
in die Altertumswissenschaft. I. Band 8. Heft (Leipizig & Berlin).
Wackernagel, J. (1892), ‘iiber ein Gesetz der indogermanischen Wortstellung’,
Zndogermanische Forschungen 1:333-436 (= Kleine Schriften (1955) i. 1-104
(Gottingen)).

BIBLIOGRAPHY 435
_ _ (1926 [vol. 11, 1928 [vol. 2]), Vorlesungen iiber Syntax (Basel).
Walde, A. and Hofmann, J. B. (193C1956, 2 vols), Lateinisches etyrnologisches
Worterbuch (Heidelberg).
Waszink, J. H. (1971), ‘Problems concerning the Satura of Ennius’, Entretiens sur
Z’Antiquitt! classique 17: 97-147. (Geneva).
Watkins, C. W. (1982), ‘Aspects of Indo-European poetics’, in E. C. PolomC (ed.),
The Indo-Europeans in the fourth and third millenia, 104-20 (Ann Arbor).
-(1989), ‘New parameters in historical linguistics, philology and cultural
history’, Language 65: 783-99.
-(1995), How to Kill a Dragon. Aspects of Indo-European Poetics (New York
- Oxford).
Watson, P. (1983), ‘Puella and Virago’, Glotta 61: 119-43.
-(1985), ‘Axelson revisited: the selection of vocabulary in Latin poetry’, C Q
NS 35: 4 3 U 8 .
Weinreich, 0. (1959), ‘Catull c. 60’, Hermes 87: 75-90.
_ _ (1960), Catull. Liebesgedichte und sonstige Dichtungen (Hamburg).
___ (1962; 2nd ed.), Romische Satiren (Zurich und Stuttgart).
Weis, R. (1992), ‘Zur Kenntnis des Griechischen im Rom der republikanischen
Zeit’, in Muller, C. W. et al. (1992), 137-42.
Weise, E 0. (1882), Die griechishen Wiirter in Latein (repr. 1964 Leipzig).
Wellmann, M. (1931), Hippokratesglossare (Quellen und Studien zur Geschichte
der Naturwissenschaften und der Medizin, 2) (Berlin).
West, D. A. (1969), Imagery and Poetry of Lucretius (Edinburgh).
West, M. L. (1982), Greek Metre (Oxford).
Westphal, R. (1867), Catulls Gedichte in ihrem geschichtlichen Zusammenhange
(Breslau).
Whitby, M., Hardie, P., and Whitby, M. (edd.) (1987), Homo Viator. Classical
Essays for John Bramble (Bristol).
Wiesen, D. S. (1989), ‘The verbal basis for Juvenal’s satiric vision’, in H. Temporini
(ed.), A N R W 11.33.1: 708-33 (Berlin).
Wifstrand, A. (1933), Von Kallimachos zu Nonnos (Lund).
Wilamowitz-Moellendorff,U. von (1898), ‘De uersu Phalaeceo’ in Mklanges Henri
Weil (Paris), 449-61 (revised in 1921: 137-53).
_ _ (1921), Griechische Verskunst (Berlin).
Wilhelm, E (1925), ‘Zu Ovid Ex Ponto 1,3’, Philologus 81: 155-67.
Wilkinson, L. P. (1959), ‘The language of Virgd and Homer’, C Q NS 9: 181-92.
-(1963), Golden Latin Artistry (Cambridge).
Williams, G. W. (1968), Tradition and Originality in Roman Poetry (Oxford).
Williams, R. D. (ed.) (1960), I? Vergili Maronis Aeneidos Liber Quintus
(Oxford).
Wills, J. (1996), Repetition in Latin Poetry. Figures of Allusion (Oxford).
Winterbottom, M. (1977a), ‘A Celtic hyperbaton?’, The Bulletin of the Board of
Celtic Studies 27: 207-12.
_ _ (1977b), ‘Aldhelm’s prose style and its origins’, Anglo-Saxon England 6
50-1.
Wiseman, T. P. (1969), Catullan Questions (Leicester).
-(1974), Cinna the Poet, and Other Roman Essays (Leicester).

436 BIBLIOGRAPHY
-(1979), ‘Onwhat Catullus doesn’t say’, Latin Teaching 35 n. 6: 11-15.

Wolfflin, E. (1882), ‘Uber die Aufgaben der lateinischen Lexikographie’, RhM 37:
83-121.
-(1885), ‘Das adverbielle cetera, alia, omnia’, A L L 2: 90-9.
-(1886), ‘Der substantivierte Infinitiv’, A L L 3: 70-91.
Wyke, M. (1989), ‘Mistress and metaphor in Augustan elegy’, Helios 16: 25-47.
Zanker, G. (1987), Realism in Alexandrian Poetry: a Literature and its Audience
(London-Sydney-Wolfeboro, NH).
Zichri, M. (1964), ‘Some metrical and prosodical features of Catullus’ poetry’,
Phoenix 18: 193-205 (= 1978 203-19).
-(1978), Scritti catulliani (Urbino).
Zwierlein, 0. (1986), Kritischer Kommentar zu den Tragodien Senecas (Akad. d.
Wiss. U. d. Literatur Mainz, Abhandlungen der geistes- und sozialwissenschaftli-
chen Klasse, Einzelveroffentlichung 6) (Wiesbaden).

Poetic Diction

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Poetic Diction

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Poetic Diction

Uploaded by

Copyright:

Available Formats

proceedings of the British Academy, 93.

Poetic Diction, Poetic Discourse

Summary. A number of distinctive characteristicscan be iden-

0 "he British Academy 1999.

Copyright © British Academy 1999 – all rights reserved

pragmatic effects. Many of these characteristics can be found

1.1. ON A NARROW DEFINITIONpoetic diction can be viewed simply as the

Copyright © British Academy 1999 – all rights reserved

sophica] theory simply cannot operate, like Virgil’s Georgics, at a

The most comprehensive exposition is in Biographia Literaria, especially chapter 14.

Copyright © British Academy 1999 – all rights reserved

’Traditionally attributed to St Ambrose and St Augustine, it is probably the work of Bishop

Copyright © British Academy 1999 – all rights reserved

ungrammatical transitions, anacolutha, rambling pleonasm and banal re-

Copyright © British Academy 1999 – all rights reserved

Copyright © British Academy 1999 – all rights reserved

body Of material to be able to plot changes in the phonology, grammar and

SOUNDS AND PATTERNS OF SOUND

Copyright © British Academy 1999 – all rights reserved

Copyright © British Academy 1999 – all rights reserved

To this typical interaction between stress and quantity we must return

Copyright © British Academy 1999 – all rights reserved

Copyright © British Academy 1999 – all rights reserved

’”See Leumann (1959 147-8).

Copyright © British Academy 1999 – all rights reserved

or d w , and there is again a loss of precision. We may contrast the nice

6.1. Phonetic expedients are sometimes adopted to cope with the

Copyright © British Academy 1999 – all rights reserved

pollio et ipse, Pollio et incipient (Virg. E. 3.86, 4.12), or by shortening the

7.1. Examples occur of pronunciations that were obsolete in current speech

Copyright © British Academy 1999 – all rights reserved

because of the range of permissible metrical variants certification is

8. Another group of archaisms involves the revival of long vowels in final

Copyright © British Academy 1999 – all rights reserved

9.1. Sometimes advantage is taken of free variants available in the ordinary

24 See Vaanlnen (1966: 34-6).

Copyright © British Academy 1999 – all rights reserved

substitutions of - forv v - (abiete) and of - - for - v v (principium) were

10.1. One particular example of the introduction of current habits of

Copyright © British Academy 1999 – all rights reserved

Copyright © British Academy 1999 – all rights reserved

11.1. Syncope of unstressed vowels was a feature of the spoken language

Copyright © British Academy 1999 – all rights reserved

chaud. Quintilian himself preferred auducter to the unsyncopated form,

Copyright © British Academy 1999 – all rights reserved

12.1. Poets were not conspicuous innovators in pronunciation or mor-

Copyright © British Academy 1999 – all rights reserved

29 For archaic morphology see Leumann (1959 143-5).

Copyright © British Academy 1999 – all rights reserved

Copyright © British Academy 1999 – all rights reserved

reflexes of classical i and e in the late Empire. Appeal to Virgil’s autograph

Copyright © British Academy 1999 – all rights reserved

which the only dactyl is aurOquZpZllibant with auroque in hendiadys with

33See F'ye (1963) for further discussion and relevant statistics.

Copyright © British Academy 1999 – all rights reserved

95 See Coleman (1989).

Copyright © British Academy 1999 – all rights reserved