The Prosody of Latin S Impura Consonant Clusters in the Waltharius


The Prosody of Latin S Impura Consonant Clusters in the Waltharius

By Blake Lopez     


Comprising the single most comprehensive account of the exploits of Germanic legendary hero Walther of Aquitaine, the Waltharius is a ninth or tenth-century CE Latin epic poem whose nearly 1500 dactylic hexameters offer a goldmine for the study of prosodic developments in post-Classical Latin poetic meter. In 1992, Edoardo D’Angelo tapped many of these veins in his Indagini sulla tecnica versificatoria nell’esametro del Waltharius, where he adroitly discusses and categorizes many instances of innovative vowel lengthening within the prosody of the poem.2 In his monograph, he stops short, however, of addressing one specific area of prosodic peculiarity: certain consonant sequences in the Waltharius fail to lengthen by position a preceding short vowel. The anomalous consonant sequences that are subject to this phenomenon of vowel-lengthening failure, as earlier commenters Rudolf Peiper,3 Hermann Althof,4 and Karl Strecker5 have noted, take the form of z (most probably representing an affricate) and of s impura consonant clusters. Throughout the Waltharius, s impura consonant clusters (sp, st, and sc) are consistently subject to vowel-lengthening failure when the cluster is in word-initial position, while z, which occurs only word-internally, is subject to vowel-lengthening failure only sporadically.

In this paper, we will investigate one type of these consonant sequences, namely, s impura clusters, and their prosody within the metrics of the poem. More specifically, we will examine the phonetic environments in the Walthariusin which s impura clusters are subject to vowel-lengthening failure and contextualize the Waltharius Poet’s tendencies in this regard among those of earlier and more contemporary Latin poets, all of whom treat these clusters prosodically in a manner that is far from consistent. Finally, in light of the scholarly uncertainty as to whether the Waltharius Poet was a speaker of an Old High Germanic or Romance vernacular, we will argue that the Waltharius Poet’s prosodic treatment of s impura clusters may be connected to a Germanic (or at least non-Romance) vernacular.

As mentioned earlier, the Waltharius’ word-initial s impura clusters consistently fail to lengthen by position a preceding short vowel—in this case, a final short vowel of a preceding word. This consistent vowel-lengthening failure was first noted by Peiper and later confirmed by Althof. S impura clusters are traditionally defined as sp, st, and sc(optionally followed by r in accordance with the mute + liquid rule), but individual occurrences of word-initial su (1311) and squ (482) are also notably subject to vowel-lengthening failure in the Waltharius.6 Even the poem’s word-internal s impura clusters are occasionally subject to vowel-lengthening failure, e.g., pugnamque restaurunt (193) and eripuit semispātam (1390).7, 8

All the above examples of non-lengthening s impura clusters come from the body of the Waltharius.Unfortunately, no definite occurrences of a word-initial s impura cluster preceded by a short vowel are found within the prologue, whose authorship is contested. As a result, this prosodic peculiarity cannot be used to determine whether the authors of the prologue and the poem proper are one and the same. The nearest example is the spondee-dactyl sequence in the phrase sancto spiramine (7): the final o of sancto is a long vowel, but throughout the Waltharius, long final o’s may be scanned as either long or short (including in the prologue—cf. promo [14]).9 The final o of sancto may, therefore, be a metrically short vowel which the word-initial s impura cluster of spiramine lengthens by position. This lengthening in the prologue would contrast with the consistent vowel-lengthening failure in the rest of the poem and would thus evidence a difference in authorship between the prologue and the poem proper.

One pre-Classical and Classical Latin prosodic phenomenon from which the Waltharius poet’s treatment of word-initial s impura clusters must be distinguished is word-final s weakening, wherein a word-final pre-Classical and Classical Latin /s/ may be weakened to [h]. This development allowed pre-Classical Latin poets to have a word-final s fail to lengthen by position a preceding short vowel; however, the latest use of this metrical device is found in Catullus during the 1st century BCE.10 While Dag Norberg notes two post-Classical users of this metrical device,11 neither is the Waltharius Poet; to the contrary, the Waltharius Poet’s non-lengthening s impura clusters occur only word-initially or -internally, never word-finally as in pre-Classical final s weakening, so the Waltharius Poet’s non-lengthening treatment of s is critically distinctive from this other sort of final s weakening.

In any case, this consistent vowel-lengthening failure is actually more consistent than any prosodic treatment of word-initial s impura clusters in Classical poetry. Vergil’s Aeneid, for instance, has word-initial s impura clusters both lengthen and fail to lengthen a preceding short vowel: cf. long Brontesque Steropesque (VIII 425) vs. short ponite. spes… (XI 209). D. S. Raven designates cases of lengthening in such instances as “studied imitations of Greek lines,” while arguing that in more prosaic contexts, Classical poets seem to have generally avoided putting short vowels in such a position in the first place, “for it seems to have been equally repugnant to the Roman poets to allow [short vowels] to remain short in such a position.”12

This inconsistent treatment by Classical poets of short vowels preceding word-initial s impura clusters seems entirely natural in view of what Rodney Sampson terms the “exceptional” status of word-initial s impura clusters in Latin phonotactic structure, wherein word-initial consonant clusters were limited “almost exclusively” to the following two types (illustrated in the following table from Sampson):13


(a) obstruent + liquid e.g. PL ENUS ‘full,’ GR ANDIS ‘great’
(b) s + voiceless plosive (+ liquid) e.g. ST O ‘I stand,’ STR AMEN ‘straw’


S is thus the only Latin consonant which may precede a voiceless plosive to form a type (b) (i.e. s impura) word-initial consonant cluster. Contributing to this “exceptional” status of word-initial s impura clusters is the fact that such clusters violate the Sonority Sequencing Principle, which states that consonant clusters are supposed to begin with the most sonorant consonant followed by less sonorant consonants, e.g., a voiceless stop may be followed by a voiceless fricative, but Latin s impura clusters reverse the order.14 It is, therefore, unsurprising that these “exceptional” clusters could be treated exceptionally in Classical Latin metrics by being occasionally subject to vowel-lengthening failure.

It appears that an uncertainty that began among the Classical poets regarding the prosodic treatment of word-initial s impura clusters only increased as time went on. Certain post-Classical poets—for instance, Sedulius—consistently have word-initial s impura clusters subject to vowel-lengthening failure, while other post-Classical poets—for instance, Aldhelm—even have word-internal s impura clusters subject to vowel-lengthening failure,15 as also occurs a few times within the Waltharius. Such tendencies that post-Classical Latin poets exhibited in composing their own verses could even color these poets’ readings of earlier, Classical meter: for instance, in his De arte metrica, Bede erroneously scans a line of the Aeneid as containing a word-internal s impura cluster subject to vowel-lengthening failure.16 Given that vernacular phonotactics appear to be responsible for word-initial s impura clusters’ vowel-lengthening failure in Classical Latin poetry, might vernacular phonotactics again be responsible for the same in post-Classical Latin poetry?

Both Aldhelm and Bede, for whom s impura clusters could be subject to vowel-lengthening failure, were speakers of the Germanic language Old English, a vernacular in which word-initial s impura clusters existed,17 as opposed to the Romance vernaculars of the Middle Ages, which generally lacked word-initial s impura clusters. These Germanic speakers’ vernacular contained word-initial s impura clusters, and their Latin poetry had word-initial s impura clusters subject to vowel-lengthening failure; the Classical Latin poets also spoke a vernacular containing word-initial s impura clusters and had word-initial s impura clusters occasionally subject to vowel-lengthening failure. Extrapolating from this example, it may be suggested that a Latin poet whose vernacular contained word-initial s impura clusters may have an s impura cluster subject to vowel-lengthening failure. In contrast to Classical Latin and the Germanic languages, then, the medieval Romance vernaculars, which generally lacked word-initial s impura clusters, form a convenient point of comparison for our suggestion.

Within the vernacular varieties of Romance during the Middle Ages, word-initial s impura consonant clusters had, in fact, already been lost centuries earlier (probably by the end of the 1st century CE) by the Vulgar Latin process of i-prothesis, wherein an epenthetic short i (> close e by later developments) was attached prothetically to a Latin word beginning with an s impura cluster, e.g., Latin scripta > Vulgar Latin iscripta > Spanish escrita.18 By this sound change, all former word-initial s impura clusters in Latin were now transformed into word-internal s impura clusters in the Romance vernaculars, and these word-internal clusters, as Classical Latin prosody attests, consistently lengthened by position a preceding short vowel.

Whereas the Classical Latin poets’ treatment of word-initial s impura clusters was marked by inconsistency, the transformation within the Romance vernacular of all s impura clusters to the word-internal type may be partly responsible for the tendencies of certain post-Classical Latin poets to attribute to word-initial s impura clusters the exact same consistency of lengthening as word-internal s impura clusters within their metrics. Among such poets is Venantius Fortunatus,19 born in the Romance-speaking region of Veneto, Italy. Entertaining this possibility, we could then set up an opposition between Romance-speaking Latin poets on the one hand, in whose Latin prosody all s impura clusters lengthen a preceding short vowel, and non-Romance- (e.g., Germanic-) speaking Latin poets on the other hand, in whose Latin prosody an s impura cluster does not necessarily lengthen a preceding short vowel, assuming of course that the non-Romance vernacular in question indeed allows word-initial s impura clusters.

While a poet’s vernacular phonotactics may indeed shape the prosodic behavior of their Latin s impura clusters, it is important to note that not every Latin poet exhibits behavior that fits neatly into the proposed opposition of Romance vs. non-Romance vernacular. Norberg, for instance, describes word-initial s impura clusters which lengthen a preceding short vowel as characteristic to Sedulius, writing in Latin in the 5th century;20 this prosodic treatment of word-initial s impura clusters might lead us to assume that Sedulius’ vernacular contains no word-initial s impura clusters, but if his vernacular is Old Irish, as his Irish name suggests, this assumption would in fact be incorrect.21 Furthermore, despite Aldhelm and Bede’s tendency to have even word-internal s impura clusters subject to vowel-lengthening failure, and the posited source for this allowance being the phonotactics of their Germanic (Old English) vernacular, the two poets do on occasion have word-initial s impura clusters lengthen by position a preceding short vowel.22

While counterexamples such as these immediately above assuredly complicate the proposed Romance vs. non-Romance generalization, these counterexamples by no means dismantle it entirely: such poets’ deviation from expected prosodic tendencies—especially in the case of Aldhelm and Bede, who deviate in such instances even from their own established precedent—may be due merely to imitation of either Romance-speaking authors like Venantius Fortunatus or the occasional Vergilian lines in which a word-initial s impura cluster lengthens by position a preceding short vowel—one of Raven’s “studied imitations of Greek lines.” Keeping both this generalization as well as the generalization’s limitations in mind, we return to our discussion of the Waltharius.

Though our generalization, namely that the prosodic treatment of Latin s impura clusters varies according to the Latin poet’s vernacular, is imperfect, applying this general principle to the Waltharius may help determine whether the Waltharius Poet’s vernacular was Romance or Germanic. Within the Waltharius Poet’s metrics, as earlier noted, word-initial s impura clusters are consistently subject to vowel-lengthening failure; this prosodic quality of the Waltharius Poet, as we have just suggested, may be inconsistent with a Romance vernacular and may rather suggest a Germanic vernacular. More specifically, exactly the same kind of word-initial s impura clusters exist in Old High German—a commonly proposed vernacular for a Germanic-speaking Waltharius Poet— as in Classical Latin,23 which, as we have discussed, may, in fact, form the grounds for such clusters’ vowel-lengthening failure in the Waltharius Poet’s Latin poetry. Given, then, that this prosodic quality of the Waltharius Poet’s metrics may be connected to an Old High Germanic vernacular, evidence may thus be adduced that the vernacular of the Waltharius Poet was Old High German as opposed to Romance.

While the argument for a Germanic rather than Romance vernacular of the Waltharius Poet on the basis of s impura clusters is supported in this paper by metrical and linguistic evidence, it is necessarily limited and best considered within the context of separate evidence for a Germanic vernacular of the Waltharius Poet, such as Althof’s identification of the kinship between alliteration in the Waltharius and that of Old High German poetry.24 In any case, the evidence of non-lengthening consonant sequences within the Waltharius, as we have demonstrated with this paper in the specific case of s impura clusters, provides a fertile ground for augmenting such argumentation on a metrical and linguistic basis. In particular, more investigation could be done in the future into the behavior of z in the Waltharius: in the poem, z may or may not be subject to vowel-lengthening failure,25 a practice which may perhaps be connected with the polysemy of the grapheme <z> within Old High German orthography (indicating either a single fricative or the affricate /ts/).26 In any case, this present paper demonstrates the immense insight into Latin historical phonology and metrics that can be gained not just from the Waltharius but from post-Classical Latin poetry as a whole.


Blake Alexander Lopez is a senior at Harvard College concentrating in Classics and Linguistics, with a secondary field in Medieval Studies and a language citation in German. Blake specializes in Latin philology and historical phonology, with a particular interest in the development of Latin into the Romance languages.



  1. I owe a great thanks to Jan Ziolkowski, Arthur Kingsley Porter Professor of Medieval Latin in Harvard University’s Department of the Classics, for whose course an early version of this paper was submitted and whose deeply insightful edits were instrumental in bringing this paper to its present form.
  2. D’Angelo 1992, 95-8.
  3. Peiper 1873, xxxix.
  4. Althof 1899, 52.
  5. Strecker 1951, 17.
  6. Althof 1899, 52.
  7. Peiper 1873, xxxix.
  8. All Waltharius text in this paper follows Strecker 1951.
  9. Althof 1899, 51.
  10. Allen 36-7, 1989.
  11. Norberg et al. 2012, 2.
  12. Raven 1965, 24-5.
  13. Sampson 2009, 42-45.
  14. Francois 2010, 413.
  15. Norberg et al. 2012, 2.
  16. Heikkinen 2012, 30.
  17. Checked in the Dictionary of Old English Web Corpus: . Accessed on 1 May 2022.
  18. Herman 35, 2000.
  19. Orchard 1994, 77.
  20. Norberg et al. 2012, 2.
  21. Checked in An Electronic Dictionary of the Irish Language: . Accessed on 1 May 2022.
  22. Orchard 1994, 77.
  23. Braune and Heidermanns 2004, 385-7.
  24. Althof 1899, 56-7.
  25. Strecker 1951, 17.
  26. Braune and Heidermanns 2004, 152.


Allen, W. Sidney. 1989. Vox Latina : a Guide to the Pronunciation of Classical Latin. 2nd ed., revised August 1988. Cambridge [Eng.]; New York: Cambridge University Press.

Althof, Hermann. 1899. Waltharii poesis: das Waltharilied Ekkehards I. von St. Gallen. Erster Teil. Germany: Dieterich.

Braune, Wilhelm, and Frank Heidermanns. 2018. Althochdeutsche Grammatik I. 16. Auflage. Vol. Bd. 5.1. Berlin: De Gruyter.

D’Angelo, Edoardo. 1992. Indagini sulla tecnica versificatoria nell’esametro del Waltharius. Catania: Centro di studi sull’antico cristianesimo.

Francois, Alexandre. 2010. “Phonotactics and the Prestopped Velar Lateral of Hiw: Resolving the Ambiguity of a Complex Segment.” Phonology 27 (3): 393–434.

Heikkinen, Seppo. 2012. “The Christianisation of Latin Metre: A Study of Bede’s De arte metrica.” Phd diss., University of Helsinki.

Herman, József., and Roger Wright. 2000. Vulgar Latin. University Park, Pa.: Pennsylvania State University Press.

Norberg, Dag Ludvig, Grant C. Roti, Jacqueline de La Chappelle Skulby, and Jan Ziolkowski. 2012. An Introduction to the Study of Medieval Latin Versification. Washington: The Catholic University of America Press.

Orchard, Andy. 1994. The Poetic Art of Aldhelm. Cambridge, England; New York: Cambridge University Press.

Peiper, Rudolf. 1873. Ekkehardi primi Waltharius. Berolini.

Raven, D. S. 1965. Latin Metre. London: Faber and Faber.

Sampson, Rodney. 2010. “The Latin Background.” In Vowel Prosthesis in Romance: A Diachronic Study. Oxford: Oxford University Press, 2009. Oxford Scholarship Online. doi: 10.1093/acprof:oso/9780199541157.003.0003.

Strecker, Karl. 1951. Monumenta Germaniae Historica: Poetae Latini Medii Aevi VI, 1: Nachträge zu den Poetae Aevi Carolini. Weimar.



Dictionary of Old English Web Corpus. 2004. Ann Arbor, Michigan: University of Michigan, Humanities Text Initiative. Accessed 1 May 2022.

An Electronic Dictionary of the Irish Language. 2019. Dublin: Royal Irish Academy, 1913-1976. Accessed 1 May 2022.