BMCR 2013.07.30

Italic Verse: A Study of the Poetic Remains of Old Latin, Faliscan, and Sabellic. Innsbrucker Beiträge zur Sprachwissenschaft, Bd 145

, Italic Verse: A Study of the Poetic Remains of Old Latin, Faliscan, and Sabellic. Innsbrucker Beiträge zur Sprachwissenschaft, Bd 145. Innsbruck: Institut für Sprachen und Literaturen der Universität Innsbruck Bereich Sprachwissenschaft, 2012. xxvii, 437. ISBN 9783851247312 €80.00.

[The table of contents appears at the end of this review.]

The riddle of the Saturnian may be solved at last. In this important book, Angelo Mercado proposes an accentual schema for Saturnian verse, one with just enough freedom to account for the existing lines, but not so loose as to turn all of Latin literature into a corpus of pseudo-Saturnians. He goes on to discuss similar verse forms in the other Italic languages and, briefly, to compare this Italic form with forms elsewhere in Indo-European. This is the fullest modern discussion of Saturnians and the most convincing. It will be of interest not only to scholars of early Latin but also to metrists and Indo-Europeanists.

Mercado concludes that the Saturnian line consists of two cola, each made up of two syllabo-tonic metra, basically trochaic and amphibrachic; acephaly and anaclasis, either of metra or of feet within the metra, are permissible variations (89). Thus the cardinal basic form of a Saturnian verse is:

‘⏑’⏑|⏑’⏑ || ⏑’⏑|⏑’⏑

as in the first line of the Epitaphium Naevii

Immortales mortales    si foret fas flere,
flerent divae Camenae    Naevium poetam;
itaque postquam est Orchi    traditus thesauro,
obliti sunt Romae    loquier lingua Latina.

though the type with “inversion” or “mis-match” in the opening of the second colon is more common, as in the second line of the same poem, scanned:

‘⏑’⏑|⏑’⏑ || ‘⏑⏑|⏑’⏑

Certain positions in the line, both stressed ones and unstressed, can be resolved. Hiatus is permissible, particularly between the two cola and between the two metra, but if two vowels coincide elsewhere, they will be elided.

What is new here is the precision of Mercado’s analysis. There have been dozens of attempts to determine the rules of Saturnians (surveyed by Mercado in chapter 2, 40-53). Many of them end up resorting to emendation, or postulate odd exceptions to the normal rules of stress or quantity in Latin, to account for a majority of the extant lines. Others are simply too vague, accepting almost any set of seven syllables (or so) followed by another six (or thereabouts) as a Saturnian line. Mercado’s schema is flexible enough to account for the attested variations, but not so loose that we can’t tell what is and is not a possible line.

To determine which constraints on the line are important, which variations cannot occur together, and so on, and to show that his schema is not impossibly vague, Mercado uses a handful of statistical tests, which he explains briskly in his introduction (13-14). He is generally good about explaining the implications of a statistical result in less technical language, so I expect readers with little or no background in statistics should still be able to grasp the main points.

Mercado says (45) that he began working on Saturnians after reading Parsons’s 1999 TAPA paper. Parsons proposes a quantitative analysis in which each colon is made up of two dipodes (thus four feet); each foot has a strong position followed by a weak one, and each metrical position is zero to two morae. Some weak positions may be unrealized (Parsons 124-125). Division into prosodic feet depends on accent, and a crucial part of Parsons’s argument is that the Saturnian developed while Latin had word-initial stress; as he puts it, “The shift to Plautine accentuation left the Saturnian meter irreconcilably at odds with the language’s new prosodic structure. In a very short time, its metrical system became unintelligible and was abandoned” (Parsons 135). Mercado published a paper in 2003 in which he “offered a favorable critique and exploration of Parsons’s theory” (45), but now finds Parsons’s schema not only too free but also too large, since its lines have 16 positions while almost no extant Saturnians have more than 14.

As Mercado points out, “the hypothesis that it was not syllable duration but rather prominence which governed Saturnian versification has been pursued by fewer investigators and has never gained any currency in the field” (48-49). Mercado revives an old, nearly forgotten accentual schema due to Rudolf Thurneysen (1885), who treated the Saturnian as a four-part line with two stresses in the first part, one in each of the other three, with a variety of permissible variations. Lindsay (1893) responded to Thurneysen’s work, though by the time of his Early Latin Verse (publ. 1922), he had given up the Saturnian as a lost cause (9-10). But neither Thurneysen nor Lindsay can make their relatively rigid accentual scheme work; as Mercado puts it, “Lindsay can only do violence to the text in order for it to fit his scheme” (51). Since the word-stresses simply are not always in the same places in the extant Saturnian lines, a more flexible scheme is required.

Mercado aims to start over from scratch, looking at the verses as they are and deducing the patterns from the entire extant corpus. He begins with a sub-group of lines where there are no textual problems, no possible elisions, and no “adjacent non-final light syllables” (57), that is, lines in which it is clear how many metrical positions there are; there are 31 such lines, including the first two of the Epitaphium Naevii. 1 He assumes the Plautine accent rules: in particular, a word ending ⏑⏑⏑⏒, such as capitibus, will be accented on the fourth-from-last syllable, cápitibus, not the third-from-last as in later classical Latin. This is plausible for the attested verses, since Livius Andronicus and Naevius are roughly contemporary with Plautus. “It would otherwise have to be assumed that the (pre- ?) Old Latin system was replaced wholesale by the Plautine one at almost the same time that Andronicus, Naevius, Plautus, and Ennius flourished” (55), he observes; even if Saturnians were being composed when Latin had word-initial stress, the poems that survive presumably use the stress patterns of their authors’ own time.

Given these accent rules, and the 31 unproblematic lines, Mercado counts up how many syllables there are, where the word breaks come, and where the stresses are, treating the first colon and the second separately. He then tests the resulting model against the rest of the corpus and determines where variation is permissible. The result is the model shown above. Next, in chapter 4, he works backward from there to figure out the detailed scansion rules, in particular when elision happens as opposed to hiatus, and when two syllables can be resolved into a single position. The exposition here is a bit circular, since in some places in chapter 3 Mercado has assumed the existence of the rules he will justify later; the argument does in fact hang together, but that may not be clear until after reading chapter 3, chapter 4, and chapter 3 again.

In the next few chapters, Mercado considers style in Saturnian verse, both within lines and in the construction of entire poems. He also looks at some difficult fragments and some texts that might or might not be Saturnians. While Naevius fr. 43 may be a “lost cause” (159), Mercado proposes emendations to the text or colometry of several others and rejects emendations in other cases, thus bringing another 19 fragments into line with his metrical schema. He determines that the Faliscan Cooks’ Dedication ( CIL I 2 364) is in fact in Saturnians (196-199) while several other epigraphical texts are not.

At this point one may wonder whether a flexible, stress-based schema is so flexible that any Latin text looks like Saturnians. Mercado uses two measures of flexibility, the “surprisal” of a metrical pattern and attempted metrical scansion of prose texts. Surprisal is a term from information theory which roughly measures the amount of information a given event contains. The more restrictive a metrical pattern is, the lower its surprisal value. Mercado calculates that the Latin dactylic hexameter has a surprisal of 0.50, the iambic trimeter 0.83 in tragedy and 1.05 in comedy, the trochaic septenarius 1.14; these can be taken as rough boundaries of plausibility (35-39). In other words, a proposed metrical pattern with a smaller surprisal than 0.5 is probably too restrictive to be usable by poets, but one with a value much larger than 1.14 may be too unrestrictive to describe anything.2 The surprisal of Mercado’s schema is 0.98, right between the values for tragic and comic iambic trimeters, and smaller than that of any of the competing proposals except Thurneysen’s (133).

Mercado’s second test is to see whether his schema can distinguish verse from prose. As he points out (205), it is possible to find accidental senarii or even hexameters in a paragraph of prose, but the bulk of a prose text does not conform to the familiar rules for the various quantitative meters. Phrases that sound like Saturnians are even easier to find in prose, but Mercado shows that they are generally not true Saturnians, because they do not observe the constraints related to colon combinations, acephaly, and anaclasis (207). In other words, the model does account for the essential differences between Saturnian verse and contemporary prose.

The second part of the book talks about verse forms in the other Italic languages, Faliscan, Umbrian, South Picene, and Oscan. None of these has Saturnians in the strict sense, though all except Umbrian have forms that are similar and possibly related. In the third part, Mercado broadens the discussion beyond Italic to Celtic, and Indo-European more generally. He argues that the seven-syllable Old Irish line may be similar to a single colon from a Saturnian, and that both may possibly be descended from “a Proto-Italo-Celtic trochaic-dactylic colon” (373). Germanic might profitably be brought in as well.

Mercado’s analysis of the Saturnian is the most straightforward and the most convincing of the dozens of models that have been proposed over the years. He is surely right to abandon the idea of quantitative analysis, and his schema makes sense of the extant texts. This will be the definitive treatment of Saturnians for a long time to come.3


Lindsay, Wallace M. “The Saturnian Metre.” AJP 14 (1893), 139-170, 305-334.
Lindsay, Wallace M. Early Latin Verse. Oxford: Clarendon Press, 1922.
Mahoney, Anne. “Alliteration in Saturnian Verse.” NECJ 28 (2001), 78-82.
Mercado, Angelo O. “A New Approach to Old Latin and Umbrian Poetic Meter.” In Proceedings of the Fourteenth Annual UCLA Indo-European Conference, ed. K. Jones-Bley, M. Huld, A. Della Volpe, M. Robbins Dexter. Journal of Indo-European Studies Monograph 47. Washington: Institute for the Study of Man, 2003; p. 188-219.
Parsons, Jed. “A New Approach to the Saturnian Verse and Its Relation to Latin Prosody.” TAPA 129 (1999), 117-137.
Probert, Philomen. “On the Prosody of Latin Enclitics.” In Oxford University Working Papers in Linguistics, Philology and Phonetics 7 (2002), ed. I. J. Hartmann and A. Willi, 181-206.
Thurneysen, Rudolf. Der Saturnier und sein Verhältniss zum späteren römischen Volksverse. Halle: M. Niemeyer, 1885.

Table of Contents

1. Introduction
Part I: Latin
2. The Problem of the Latin Saturnian
3. The Latin Saturnian Meter
4. Rules of Saturnian Scansion
5. Saturnian Stylistics and Metrical sStructure
6. Problematic Saturnians
7. Putative Saturnians
8. Saturnian Monocola
9. The Latin Trochaic and Dactylic Tetrapodies
Part II: Faliscan and Sabellic
10. The Problems in Faliscan and Sabellic
11. Faliscan Verse
12. Umbrian Verse
13. South Picene Verse
14. Oscan Verse
Part III: Comparisons
15. Typological Plausibility and Synchronic Reality
16. The Proto-Italic Metrical System?


1. Not the other two, though: the third line has a possible elision or prodelision, postquam ‘st, and the fourth has a possible resolution in loquier.

2. Surprisal is defined as the binary logarithm of the number of possible outcomes divided by the number of independent variables. To apply this to a metrical schema, call each position in the schema a variable. For example, for the dactylic hexameter, there are twelve positions, and six of them can vary: each biceps can be realized as two short syllables or as one long, and the final syllable can be long or short. The number of possible outcomes is the number of ways the line can be realized; for the hexameter, as each variable element has two possibilities, and they are independent of each other, the number of possible realizations of the hexameter is 2 × 2 × 2 × 2 × 2 × 2 = 64. So the surprisal of this meter is log(64)/12 = 0.5.

3. Two editorial problems should be noted. First, the discussion of accent and clisis, p. 120-127, is garbled, as if incompletely revised. At p. 120 we read “in the archaic Latin of the earliest Saturnian poets enclisis and proclisis preceded accentuation” (in other words, virumque is to be read as virúmque). But on p. 127 we have “I incline towards the ordering of accentuation before proclisis,” and the discussion in between (e.g. 121) has shown that accent must come before enclisis as well (giving vírumque). Second, two papers are missing from the bibliography, Mahoney (2001) and Probert (2002).