BMCR 2007.12.21

La metrica di Plauto e di Terenzio

, La metrica di Plauto e Terenzio. Ludus philologiae ; 16. Urbino: QuattroVenti, 2007. xiii, 550 pages ; 24 cm.. ISBN 9788839207944. €54.00.

Table of Contents

Cesare Questa’s book on Plautine and Terentian metre is the fruit of over half a century’s work. With his unrivalled knowledge of manuscripts, especially manuscripts of Plautus, and of the history of metrical studies, Questa has created a true masterpiece that can be read with profit by beginners and experts alike.

The introduction to this work is short and concise and outlines what is to follow. The remainder falls into two parts, one on prosody and one on metre. Each of these parts is divided into several sections. The sections of the first part deal with (i) pre-classical quantities, (ii) loss of word-final phonemes, (iii) prodelision, intervocalic phonemes, syncope and anaptyxis, and the treatment of muta cum liquida sequences, (iv) pre-classical morphology, (v) iambic shortening, (vi) shortening through enclisis, and (vii) what happens if two vowels are adjacent to each other. The second part treats (i) “first-order” rules, that is, rules which apply to poets such as Horace and Seneca as well, (ii) “second-order” rules specific to early Latin, (iii) iambic and trochaic metres, (iv) cretic and bacchiac metres, (v) anapaests, (vi) the uersus Reiziani, and (vii) lyric metres.

The discussion of pre-classical quantities is straightforward. Plautus and Terence still have the final geminate consonants that arose from various assimilations, for example miless (from *milet-s) with a heavy final syllable in Aul. 528. Classical Latin has kept hocc (from *hod-ke), but in words of more than one syllable the final geminates have been simplified. In final syllables ending in single consonants Plautus has preserved inherited long vowels, for instance in aggrediâr in Persa 788, but by the first century BC such long vowels had been shortened unless the word was a monosyllable ending in -s, -r, -l ( rês, fûr, sôl, but dat with a short vowel1) or a polysyllable ending in -s ( amâs, but amor with a short vowel). Terence also has clear cases of inherited long vowels, but shortening is beginning to take place and in Ad. 453 we have to scan audiret with a light final syllable.

On one minor point I have to disagree with Questa. On p. 17 he states that the perfect indicative ending -it contains a long vowel.2 This is only partly correct. The ending continues the old, reinforced perfect ending *-eiti, but also the old aorist ending *-et. The former becomes -ît by regular sound change, the latter becomes -id (short vowel) and, by analogy to other third person forms, -it (also short vowel). Both forms co-exist in early Latin. On the Cista Ficoroni (CIL I 2 561) we find both dedit (perfect ending, long vowel) and fecid (aorist ending, short vowel). Admittedly, in And. 682 we have concrepuît and nothing proves the existence of light -it for Plautus or Terence. But in Cas. 669 and Merc. 357 the bacchiac rhythm becomes clearer if we scan dixit and extrusit with light final syllables.

There are three word-final phonemes which are lost relatively frequently: short -e, -d, and -s. Short -e is generally elided before vocalic onsets, but a number of words also lose this -e before consonants, either obligatorily ( nempe) or at least frequently ( unde). It seems that the longer forms are beginning to be restored before consonantal onsets in Plautus’ time. Final -d is kept after short vowels. After long vowels it has been lost entirely in polysyllables and almost entirely in monosyllables, where it was somewhat better preserved before vowels. Plautus still has med and ted, which are metrically convenient antevocalic variants of me and te, also common in Plautus (before both vowels and consonants) and the only possible forms in Terence. Some time before Plautus hau was the pre-consonantal variant of haud, the pre-vocalic allomorph; but again Plautus begins to restore the longer forms before consonants. It is interesting to contrast final -s with final -d. Final -s is kept after long vowels, but can be lost after short ones — the exact opposite of final -d. Final -s can only be dropped before consonants, where final -d was also lost more easily than before vowels.

In the third section Questa looks at a range of phenomena, among them prodelision. Prodelision of es and est does not really involve vowel loss. Rather, we are dealing with enclitic forms that never contained vowels to begin with ( s and st). On p. 48, Questa argues that bonust involves loss of the final -s of the adjective before the two forms can coalesce. The fact that the last syllable of what in editions is written suppromu’s ( Mil. 825) counts as long is then explained by saying that the enclitic form is ss rather than s (p. 40). I am not convinced that prodelision really involves loss of the final -s of the noun or adjective. The absence of *illicst and the like does not have to be explained by saying that final -c cannot be lost; an equally good explanation is that Latin does not permit -kst as a final consonant cluster. Enclitic ss does indeed exist, but a better proof than suppromu’s is dicacula’s ( Asin. 511) with a heavy final syllable.

Questa goes on to discuss the treatment of intervocalic -h- and the semivowel -u-, syncope and anaptyxis, and muta cum liquida sequences, which do not make position in Plautus and Terence (but must have done so earlier, as *genatrix became genetrix, while *genator became genitor 3).

The section on pre-classical morphology contains many important observations. Nouns in -ius for example have the genitive singular in -i, but the nominative plural and the locative singular in -ii. Questa does not explain why this should be the case, but it is easy enough to see that the genitive goes back to *-ii, while the nominative and locative go back to *-ioi and *-iei respectively. Vowels of the same quality contract more easily, hence the contracted genitive in Plautus, but the still disyllabic nominative and locative. Also of interest is the observation that in the third declension the ablative singular ending of the i -stems, -i, spreads to the consonant stems, but not the other way round. What I found most useful in this section was the detailed discussion of how pronouns scan in Plautus and Terence and how Greek loanwords are treated.

The fifth and perhaps most exciting section of this part of the book is about iambic shortening. Questa states that the breuis and the breuianda must form a single element, that the breuianda must be an unstressed syllable, and that there must not be word end between the two syllables, unless the breuis is a monosyllable. Many exceptions to these rules only appear to be exceptions; talentum scanned as an anapaest, for instance, has simply preserved the Greek accent position.4 However, a few difficult cases remain, for example sagitta scanned as a tribrach.

Since the breuianda is so often a closed heavy syllable rather than an open syllable with a long vowel, Questa concludes that iambic shortening, while having a basis in everyday speech in cases like the pyrrhichic adverb modo, is mostly a metrical rather than a linguistic phenomenon (p. 144—7). We are dealing with a metrical licence rather than a truly phonetic shortening.

School books usually teach that clitics such as -que lead to an accent shift to the syllable immediately preceding the clitic. The fact that dedistin can scan as an anapaest is in perfect agreement with this rule. However, there are also cases of bonoque and the like scanning as tribrachs (p. 90), and if accented syllables cannot undergo iambic shortening, these are disconcerting for the generally accepted theory of accent shift, at least for the archaic period.

In Bacch. 592, negato stands before esse. The last syllable of the imperative is elided and -ga- counts as a short syllable. On p. 98, Questa argues that this is not problematic because the imperative has become disyllabic through elision, and disyllabic words are accented on their first syllable, so -ga- can undergo iambic shortening. I am not happy with such claims. Elsewhere, in fact (see above) Questa says that elided syllables are not really lost. Evidence in favour of this is presented on p. 218: the law of Hermann-Lachmann states that a disyllabic element cannot coincide with the end of a polysyllabic word; as is well known, the type perficer(e) omnes, with -fice- as a disyllabic element, is legitimate; this shows that perficere does not really lose its final vowel entirely. Now if elision does not lead to true vowel loss, Questa’s accent shift is impossible, and iambic shortening of negat(o) becomes problematic if accented syllables cannot be affected.

At this point it may be worth adding a few remarks concerning Questa’s statements about the Latin accent. Like most modern scholars he does not believe in the existence of a verse accent or ictus; rhythm is simply the regulated sequence of short and long syllables. While I fully agree with him on this, I am taken aback when he contrasts the Latin “musical” accent with the “expiratory” accent of most modern European languages. Surely things are slightly more complex.5 A pitch-accent language is one in which pitch contours can be lexically distinctive, as in Greek φώς“man” versus φῶς“light”. Pitch can be combined with a greater degree of loudness, as in Swedish, but this does not have to be the case, as in Japanese. A stress-accent language is one in which pitch contours can possibly distinguish between sentence types (for example questions versus statements), but not between lexemes. There are no languages in which such a stress accent is not combined with changes in pitch. English is a good example: anyone who has ever looked at a spectrogram will know that the stressed syllables are not just louder, but clearly marked by pitch changes. Languages which emphasize syllables through pitch changes alone always have some lexical minimal pairs that only differ in pitch contours. Since Latin has no such minimal pairs, it must have been a stress accent language which combines loudness with pitch changes.

But let us continue with the sixth section. Like iambic shortening, shortening through enclisis is a phenomenon that has not yet been well explained. The enclitics which can — but do not have to — cause this shortening are quidem and forms of the indefinite / interrogative pronoun quis. Thus, Plautus can scan si quidem as an anapaest or as a cretic. Questa points out that only syllables ending in long vowels can undergo shortening through enclisis, but not closed syllables. Consequently, when hicquidem scans as an anapaest, we are presumably dealing with a form of hic that is lacking the particle -c(e).

The section discussing vowels adjacent to each other is full of interesting observations, particularly on hiatus. Hiatus after a caesura or diaeresis or when there is a change of speaker is not uncommon in Plautus, but much rarer in Terence. It is also worth noting — and was in fact already noticed by Bentley — that a monosyllable in prosodic hiatus, that is with shortening, can only form the first part of a divided element, not the second.

I shall present the second part of the book in less detail. The most important “first-order” rules are the laws of Ritschl and of Hermann-Lachmann. Questa summarizes them neatly on p. 221: a disyllabic element may not be formed from a word beginning before the element and ending inside it or together with it. These laws need not be observed in the second element of any verse and in the tenth element of iambic or trochaic verses with middle diaeresis (in effect the second element after the diaeresis, which in many respects behaves like verse-end). However, if the laws are not followed here, the rhythm gets disturbed and the poet has to compensate by giving the following element a prototypical realization if possible: if the following element is anceps, it is mostly realized as a short syllable, more rarely as a long one, but never as two short syllables; if the following element is a longum, it is mostly realized by a long syllable and rarely by a sequence of two short syllables. The laws of Ritschl and of Hermann-Lachmann do not apply in anapaestic metres and cola Reiziana. Here we get the law of Fraenkel-Thierfelder-Skutsch: a divided elementum biceps or anceps does not allow for a following disyllabic longum. The rule is different, the rationale the same: if the longum does not get realized in a prototypical way, it is difficult to recover the rhythm after the disturbance caused by the divided element.

There are not many “second-order” rules. In Plautus, but no longer in Terence, we find the so-called loci Jacobsohniani. These are the eighth elements of iambic senarii and the third and eleventh elements of trochaic septenarii. Here the longum can be realized by a long syllable in hiatus or by a short syllable, with or without hiatus. A longum can be realized in the same way before a pause. A pause can be a break in the verse, for example the middle diaeresis of an iambic octonarius, or a change of speaker, or a strong syntactic break. Terence prefers long syllables without hiatus in such cases, except when there is a change of speaker.

In the third section, Questa looks at iambic and trochaic metres and their breaks. He describes the law of Bentley-Luchs: a polysyllabic word can end with the last longum only if the preceding anceps is long or disyllabic. The rationale behind this rule is clear. If it is not observed, listeners might perceive the fifth foot of an iambic senarius as the end of the verse. There are two exceptions to the rule: the anceps does not have to be long or disyllabic if the polysyllabic word ending with the last longum also fills or forms part of the preceding disyllabic longum; and the anceps does not have to be heavy or disyllabic if the last two feet form a “metrical word”, for instance ad hunc diem. The rationale behind the exceptions is that in these cases listeners cannot perceive the fifth foot as line-end: the last longum is rarely disyllabic, so if the eighth element is disyllabic, the tenth element will not be perceived as line end even if preceded by a light syllable; and “metrical words” are not normally divided between lines. Questa examines Meyer’s law as well: word end is avoided in the fourth and eighth elements of the iambic senarius (and in the corresponding places of other iambo-trochaic metres) if the preceding elements are heavy or disyllabic. The reason for this rule is the same as the reason for the law of Spengel and Meyer discussed in the next paragraph.

The fourth section discusses cretic and bacchiac verses and cola with their breaks. The rule of Spengel and Meyer prohibits word-end of a polysyllable after the third and ninth elements in cretic tetrameters and after the fifth and eleventh elements in bacchiac tetrameters if the preceding elementa ancipitia are heavy or disyllabic. The reason for this rule is that in a cretic tetrameter the second and eighth elements are prototypically short, and the same holds for the fourth and tenth elements in bacchiac tetrameters; but word-end makes what precedes it sound somewhat prolonged, and if the anceps is already long, it would become extra-heavy and the rhythm would be lost.

The fifth section deals with anapaestic metres and the sixth with uersus Reiziani and cola Reiziana. The last section discusses Aeolic, Ionic, and other lyrical metres.

Questa’s book will undoubtedly be the standard reference work on Plautine and Terentian metre for many years to come. No one with a serious interest in these authors can afford to neglect this book. What makes it even better is that Questa does not just list rules, but gives their raisons d’être, and that he often tells us how the rules were discovered. Because of detailed indexes and a useful table of contents the book is easy to consult. It is beautifully produced and contains few misprints in the Italian (though there are more in the French and German quotations). The presentation of the Latin is clear and functional and contains very few errors in proportion to the length of the book.6 I will not put it on my bookshelf; it will remain on my desk as an indispensable tool.


1. Pace Questa’s suggestive remarks on p. 18, footnote 1, I hesitate to scan dat with a short vowel in Plautus. Metre allows both scansions, but if Plautus has not yet shortened final long vowels, we must have a long one here, despite the short vowel in the passive datur. Dare has to some extent preserved old ablaut distinctions, as the long vowel in das and the short one in datis show, even though there was an adjustment in vowel colour, otherwise we would have *dôs (compare ce-do“give here”) instead of dâs. The active indicative singular has the full grade, resulting in -â- the active indicative plural and the passive indicative of all persons have the zero grade resulting in short -a-.

2. Note, however, that the possibility of a short vowel is mentioned on p. 18, footnote 1.

3. Compare also capere, acceptus with a closed second syllable, and accipio with an open one.

4. Note also the secondary functional distinction between the personal name Philíppus, with a heavy penultimate, and the currency Phílippus, where the Greek accent position has been preserved and the penultimate is light.

5. Compare Ladd, D. R. (1996), Intonational Phonology (Cambridge).

6. As such errors can be quite confusing, especially for the novice, I list them here. On p. 87 ( Trin 833), we are dealing with an 8, not an 7. On p. 94 ( Curc. 268), I agree with Questa’s scansion of uelint as pyrrhichic, but I would not classify the second syllable as one that is normally long because of a long vowel; the vowel was shortened prehistorically by Osthoff’s law and the reason why the syllable is long is that the vowel is followed by two consonants. On p. 112 ( Cist. 71), faci(em) est should be faciem (e)st. Also on p. 112 ( Poen. 265), turb(a) est should be turba (e)st. On p. 114 ( Capt. 370), uortat should be uortar. On p. 141 ( Stich. 526), there should not be a hiatus between me and exilem. On p. 155 ( Curc. 703), decrere ought to be decrero. On p. 163 ( Phorm. 643), die should be dic. On p. 194 ( Hec. 1), Hecyr(a) est should be Hecyra (e)st. On p. 229 ( Mil. 1353), Questa scans ite cito as three elements, the second of which is ci- and the third of which is -to, which means that ite loses its final vowel. While this is possible, I would prefer ite to form the first two elements and cito, with iambic shortening, the third. On p. 230 ( Stich. 526), aegritudinem should be aegritudinum. On p. 232 ( Phorm. 566), the violation against the law of Hermann and Lachmann disappears if we give up the odd middle diaeresis separating preposition and noun and instead accept a break after the tenth element; in that case the eighth element is in i-, the ninth -tine-, and the tenth re. On p. 284 ( Truc. 757), I am suspicious of atque before a consonant, with -que counting as long because it is in a locus Jacobsohnianus; probably Gulielmus was right in adding ea after atque. On p. 300 ( Trin. 837), the metre is an 8, not an 7. On p. 314 ( Mil. 82), ascultare should be auscultare. On p. 324 ( Persa 854), the metre is not ia 4, but its catalectic variant. On p. 330 ( Men. 599), prim(um) est ought to be primum (e)st. On p. 335 ( Mil. 499), patrocinamini should be latrocinamini. Also on p. 335 ( Hec. 426), I fail to understand why there should not be a straightforward caesura after the fifth element. On p. 336 ( Asin. 16), natum meum should be natum tuae. Also on p. 336 ( Eun. 934), uidentur should be uidetur. On p. 337 ( Hec. 19), studios(e) ne should be studiose n(e). On p. 339 ( Curc. 378), poscunt should be poscant. On p. 350 ( Amph. 197), there is an obvious break after illi, not an “incisione latente” (there is no elision). On p. 352 ( Ad. 948), uos should be nos. On p. 355 ( Amph. 1068), orrore should be horrore. On p. 365 ( Hec. 561), consilium should be consilio. On p. 369 ( Epid. 526), si quid should be siquid (pyrrhichic scansion). On p. 372 ( Phorm. 647) petit should be petat. On p. 376 ( Pseud. 457), ad hunc diem should be ad hunc modum. On p. 385 ( Eun. 316), curuatura should be curatura. On p. 389 ( Persa 614), Questa’s scansion seems impossible; I take it that the fifth element is tib(i) i-, the sixth -bi-, and the seventh -dem, in which case I need to invoke neither Meyer’s law nor the exception of the following monosyllable. On p. 395 (n. 36, Mil. 511), da te should be de te. On p. 409 ( Capt. 1024), nebul(am) should be nebulam in hiatus. On p. 421 ( Asin. 134), the first syllable of acerrumum is marked as light, but the vowel is clearly long. Also on p. 421 ( Pseud. 1303), the second syllable of Massici is marked as long, but of course it is short. On p. 430 ( Aul. 120), it is perhaps better to read med and to analyse the verse as ba 4; Questa reads me and analyses the line as ba 2 followed by ba c, but the following lines are clearly tetrameters. On p. 433 (footnote), the first element of the last foot of ba 4 is not the ninth, but the tenth element. On p. 447 ( Stich. 26), facias should be faciat. On p. 449 ( Cas. 875), dedecus has been marked with a light first syllable and a heavy last one; it should of course be the other way round. On p. 456 ( Cas. 226), ubicumqu(e) est should be ubicumque (e)st. On p. 462 ( Rud. 196), mod(um) est should be modum (e)st. On p. 475 ( Amph. 165), mitter(e) hoc potuit should be mittere potuit, and the first syllable of potuit should not have been marked as long. On p. 476 ( Ad. 610 a), the penultimate syllable of improuiso should be marked as long, just like the other syllables. On p. 477 ( Rud. 945), the last syllable of retrahis has wrongly been marked as long. On p. 486 ( Ad. 611a), the second element is only less rhythmical if we read certum siet with Questa and the majority of the manuscripts; but some also have certus siem, and this reading, adopted by Kauer-Lindsay, may well be preferable.