“On Latin bōs”
Historische Sprachforschung 101.1 (1988): 127-137
[Reprinted with the permission of
the publisher, Vandenhoeck-Ruprecht]
Go to the bottom of this article for the NOTES.
On Latin bōs 1
Comparison among Greek, Latin and Germanic reflexes of PIE g w displays a systematic pattern of Greek [b], Latin [v] and Germanic [k] (Gothic [q]). The following examples are familiar enough (Pokorny 1959: 464, 468f., 482):
PIE Greek Latin Germanic
*g wiw- bios vivō GO qius 2
*g wm- bainō veniō GO qiman
< ban-yō AS cuman (E. come)
Yet, despite these established cognate sets, Latin deviates from the expected pattern in the lexeme for ‘cow,’ bōs:
*g w-oϜ- 3 boûs bōs AS kū
What is expected here is vōs, not bōs. What has happened? The traditional explanation, deeply rooted in the scholarly literature for at least a century (Brugmann 1888:322; Von Planta 1892:331; Buck 1904:94; Wright 1912:103, 184; Kieckers 1925:75; Muller 1926:215; Buck 1933:30, 130; Walde-Hofmann 1938:112; Kent 1940:94, 123; Kent 1946:47; Ernout 1953:60; Branderstein 1954:156; Palmer 1954:37; Frisk 1960:261; Solta 1974:47; Leumann 1977:151; Bammesberger 1984:35-6, 120), invokes borrowing from the so-called ‘p-dialects’ of Italic, Oscan and Umbrian. Thus, for example, the Latin lexemes ‘lup-us‘ and ‘pop-ina‘ must be Oscan-Umbrian loans into Latin, because the expected Latin reflexes (Leumann 1977:137, 156) would have been *luqu- or *luc- 4) and, by dissimilation, *quoqu- –> coqu 5) (and cf. Latin coqu- īna). In short the argument runs, since Latin shows loans from Oscan-Umbrian reflexes of PIE *k w, such as the attested ‘lupus’ and ‘popina’, Latin must have borrowed Oscan-Umbrian reflexes of PIE *g w as well. On the surface this assumption is not implausible, for it seems intuitively appropriate that a voiced segment should behave in a fashion analogous to that of its voiceless counterpart.
But is such an assumption warranted? The fact is that there is no attested form bōs from the extant material in any Oscan-Umbrian dialect.6 In fairness, however, it must be noted that the possibility for counter-evidence is naturally quite minuscule in such a small data-base as is afforded by the extant Oscan-Umbrian texts, and one should not axiomatically exclude an argumentum ex silentio. That is, given a larger corpus, the non-Latin bōs might well have been attested in some Oscan-Umbrian dialect. But even granting that it might have, why should the Romans have borrowed it?
Certainly the semantic field represented by the lexeme was well-known to the inhabitants of the peninsula from earliest times, as is suggested in the possible etymology of Italia itself (< *Ϝitul– ‘calf’ –> vitula), the ‘calf-land.’ 7) If a native Latin word for such a common concept as ‘cow’ had to be replaced by a lexeme from an outlying non-Latin dialect, why not so too with other common words that fit the same pattern of original *k w– (such as attested Oscan pid for Latin quid ‘what’ or Umbrian panta for Latin quanta ‘how great’) or *g w– (such as attested Oscan biv– for Latin viv– ‘live’ or Umbrian benust for Latin venerit ‘come’)? It seems quite arbitrary, not to say inexplicable. At least with a word like the non-Latin loan popīna one can argue for a type of etymological doublet to coquīna with somewhat different 8 semantic “coloring.” 9
Although such an account of the admittedly anomalous Latin bōs is phonologically as well as historically possible10 and even plausible, it wholly lacks pragmatic motivation. None of the scholars listed above has given serious consideration to what is in essence a sociolinguistic question: why borrow a word in the first place, and for no apparent reason?
Before considering this matter, I would like to return to some alternative analyses of bōs that breach the unanimous and united ranks drawn up above.
Tucker (1931:36), for example, apparently following Halsey (1893:177) at least in part, advanced a theory that the Latin bōs had nothing to do with the PIE lexeme for ‘cow.’ Rather, he saw it as a reflex of *gu- ‘roar, bellow,’ or, barring that possibility, of *beu- ‘swell out, be big.’ While admitting that it “is commonly assumed that the word is Osc.-Umb. (for a true Latin *vos)” Tucker goes on to assert that the “root which most naturally suggests itself is that of Sk. gu (roar, bellow)”. Even this derivation, however, has “no certainty,” nor, in fact, is there any “convincing proof” that either Sanskrit (gāus) or Balto-Slavic (e.g. Lett. gu°ws ‘cow’, ChSl. gove, do ‘ox’), and much less, then, the Latin, “contained a labio-velar initial.” In short, while “the usual derivation of bōs is specious & cannot be disproved, it is here held that the word is more prob. indigenous Latin < *bexu- *boxo– 11 & that [Greek] boûs is to be referred to the same root.”12
Tucker’s suggestion is both ingenious and, given the weight of scholarly consensus against him, bold. Nor can his phonology be faulted given the prevailing truths accepted about the PIE stop-system in 1931. As that system is currently interpreted by Indo-Europeanists (e.g., Emonds 1972, Hopper 1973, Gamkrelidze 1973, Miller 1977, Hopper 1977a, 1977b, Collinge 1985: 259-269), however, Tucker’s analysis must now encounter fatal objections. As early as Brugmann (1888:263) and Whitney (1889:18; and cf. Buck 1933:121) troublesome suspicions about the precise nature of certain aspects13 of the PIE stop system had surfaced. Only in the last decade or so has comparative material from non-Indo-European data made it possible on typological grounds to demonstrate the strong likelihood that the traditionally reconstructed PIE stop-system is incorrect. A consistent and long recognized problem has been in the so-called mediae aspiratae (voiced aspirated oral stops) and mediae (voiced oral stops). The lack of any substantial body of cognates in initial voiced labial makes typological sense only if the segment in question is reconstructed as a glottalic (Collinge 1985: 261-2) or ejective labial stop ([b’]) rather than plain voiced labial stop ([b]), precisely the segment which reveals a paradigm gap in those languages with attested ejective stop series (Hopper 1977a:43-4).
This revised picture of the PIE stop-system, then, creates the following ineluctable difficulty for Tucker’s interpretation of the origin of bōs. Working in the framework current in his day, he permits a Latin initial segment [b-] to be a reflex of an initial [*b-] in the proto-language, a derivation no longer tenable. His Latin bōs from PIE *bexu- is improbable on typological grounds, for any initial segment [b-] in Latin is in all likelihood the result either of some secondary phonological development with Latin (or Italic) or of direct borrowing, usually from Greek or some non-Latin Indo-European Italic dialect. In this connection it is significant to note that of all the entries under B in Glare (1982:222-246) not a single one is unequivocally14 a strictly Latin reflex of an Indo-European original.15 Items not marked as specifically dialectal or foreign borrowings are noted as etymologically “unkn(own), perh(aps), onomat(opoeic), dub(ius [= doubtful]),” and so forth. And this state of affairs is exactly what the new PIE stop-system would lead us to expect, as Whitney sensed long ago. Whatever else we may say about the origin of Latin bōs, we may be certain (on the current understanding) that it is not from any PIE form with initial *b-.
Although not putting forward as radically novel a proposal as Tucker, just a few years earlier Meillet (1928:101) likewise had expressed some doubts about the necessity of accepting the canonical explanation. Discussing the nature of “formes originellement rurales” (99) in the classical Latin lexicon, he admits that “le mot bōs est curieux.” His reasoning follows:
Comme le traitement latin de *g w initial est u, et le traitement osco-umbrien b, on admet que bōs serait emprunté à l’osco-umbrien, comme l’est certainement le mot popīna ‘cabaret’ en face de lat. coquīna (issu de quoquīna). Mais il n’est pas évident que *g w ait abouti à u consonne dans tous les parlers latins, et il n’est pas exclu que bōs soit pris des parlers du Latium où le traitement b aurait existé.
Thus, although Meillet wants to avoid the accepted practice of appealing to the notion of an Oscan-Umbrian loan for Latin bōs, his own explanation appears to be merely a variant on the standard opinion: bōs comes for him not from a rural non-Latin Italic dialect, but from some rural (sub-) dialect of Latin itself.
Like Tucker’s, this proposal is not without its ingenuity, but appears in the final analysis quite ad hoc as an explanation. It is not clear just what evidence Meillet has in mind when suggesting that *g w may (occasionally?) reflex as b rather than the attested u (v) in Latin (as opposed to Oscan or Umbrian), but he does offer a possible phonetic motivation for the adoption of this ‘allolexic’ form bōs from alleged (rural) Latin “b-speakers.” For the form bovis avoids the repetition of v internally that *vovis would necessitate. This essentially dissimilatory explanation, however, is less than compelling in view of such thoroughly attested standard and common Latin words as vīvus and voveō.
One further non-traditional view of bōs must be considered before we turn to our version of the solution. A few years before Meillet, Conway (1923:71) suggested, in general agreement with others, that “we must regard bōs… as a farmer’s word borrowed from one of the country dialects (e.g. Sabine).” What distinguishes Conway from the rest, however, is that he made an attempt to offer a non-algebraic, or societally motivated, explanation for this borrowing. As he goes on to say:
the pure Latin form would have been *vūs, plural *voves, which must have sounded very much like oves sheep, – a very undesirable confusion in the marketplace! No doubt this was one of the reason why the farmers’ pronunciation of the name was preferred to that originally current among the townsfolk of Rome.
Conway does not indicate what any of the other reasons may have been for the adoption of the rural form. But whatever the truth of the particular reason adduced, it has the great merit of going beyond description and abstract observations on obviously anomalous phonological mechanics to an attempt to locate a solution squarely in the social and pragmatic center of a living language.
A commonsensical conception is, however, in this instance less convincing in its execution. For it is unclear how there might have been confusion between oves ‘sheep’ and *voves ‘cattle.’ On the assumption that the initial sound in the latter is a labio-dental fricative, even though voiced, it might well be thought that pronunciation could be slurred,16 and so confused in utterance with oves. Available evidence on the phonetic value of v in Latin, however, points to a “bilabial, semivocalic articulation” (Allen 1978:41) for this segment long before it had begun to undergo fricativization. It thus had a phonetic value hardly likely to be confused with, in effect, an initial zero realization of bilabial in oves. Hence it is not in the least likely that bōs was required for the purpose of disambiguating possible confusion in what would have been clearly distinct and differentiating articulations in oves and *voves.17
Let us now return to some concluding thoughts about the reason for the unexpected Latin shape bōs as opposed to *vōs ‘cow.’ It appears essentially immaterial to the analysis being proposed here whether the word was an actual loan from an Italic p-dialect or not: maybe it was, maybe it wasn’t. Indeed, it may as well have been a Greek loan, for that matter, and a not unreasonable case can be made for such a supposition at least on linguistic grounds.18 What is material in this discussion, however, is the intolerable homonymic clash with the second person plural personal pronoun, vōs ‘you.’
Bynon (1977:186-9) addresses a similar if not identical difficulty in the development of French dialectal variants of coq ‘rooster’ in order to avoid confusion with the term gat ‘cat.’ Due to phonological developments localized in Gascony, an inevitable homonymy arose for the two etymologically distinct words that had to be disambiguated. Bynon notes
the importance of homonymic clash (collision homonymique) as a factor in lexical replacement. … when, on the basis of the phonological rules, two forms must have become homophonous in a dialect, one or other [sic] of them, sometimes even both, were likely to be replaced by an entirely new lexical form. The interesting point is that homophony is not always unacceptable, leading to lexical change. It is only if the homophonous forms are capable of occurring in identical contexts so that confusion will result that lexical replacement takes place. (186) [Bynon’s italics]
Bynon’s ‘conditioning factor’ for lexical replacement is precisely that which we have in the case of Latin bōs (and was the general motivation behind Conway’s analysis [see above]). It takes no great leap of the imagination to appreciate that homonymic clash in the following set of derivations could seriously have impaired communication in the barnyard, field and market-place:
‘cow’ *g w–oϜs *wes- ‘you’
Thus, no matter what the immediate origin of bōs for phonologically correct *vōs ‘cow,’ it must have proved an irresistible replacement to avoid confusion with vōs ‘you.’19 Since *vōs ‘cow’ is nowhere attested in Latin, one assumes that the lexical replacement must have taken place relatively early in the history of the language. Indeed, we do know in general that as early as the sixth or fifth century B.C. Greek was borrowed into Latin, as is evidenced by the use of triumphō (from t hriámbos ‘Bacchic song’) in the Carmen Arvale (Williams 1982:57) and of poena (from poinē) in the Tabulae XII. Further, if Roman New Comedy is any reflection of linguistic reality during the early Republic, it is worth noting in this connection that Greek loans in comedy are “used predominantly by slaves and characters of low social standing” (Maltby 1985:110), precisely the group that is likely to have worked in the field. In any event, whether a ‘rural’ Latin form, an Oscan-Umbrian form, or a Magna Graecia Greek form was the ultimate source is not ascertainable with complete confidence. Indeed, there is a fourth possibility. Although perhaps not as likely as one anchored in a surrounding sea of linguistic substitutables, it entails a simple shift, internally motivated, from bilabial stop to bilabial glide. For Latin, this would differentially phonemicize the initital segment in each word (cf. bellō ~ vellō, bīnī ~ vīnī, biduum ~ viduum, bolō ~ volō, etc.) and obviate confusion within the given semantic domain.
It remains to ask why Latin *vos ‘cow’ should have changed to bōs ‘cow’ (whatever the source if borrowed or precise phonetic mechanism if only internally motivated) rather than Latin vōs ‘you’ to *bōs ‘you’ (or any other term, for that matter), which would just as effectively also have obviated the homonymic clash. Again, one can only speculate. The latter possibility seems in fact less likely than what actually happened simply on the pragmatic grounds that it could have required a restructuring of the entire paradigm for the personal and possessive pronoun of the second person plural to something like *bostrī, *bōbīs, *bester, *bestrīs, etc. The first person plural forms might then themselves perhaps have come under some analogical20 pressure to restructure to something like *dostrī, *dōbīs, etc. If this is a possible scenario, it should be clear that such a change would have required restructuring of a major order; by contrast, the actual change called for no such radical reorganization at any level of the lexicon and thus, offering as it were the path of least resistance, won out.21
1) Without in any way implicating them in responsibility for what follows, I do nonetheless wish to thank my colleagues Professors Helena Dettmer and Mary Kuntz for reading a draft of this paper and offering valuable comment.
2) Cf. the modern English fossil ‘quick’ in the phrase ‘the quick’ and the dead.
3) I use the symbol Ϝ to indicate digamma.
4) < ‘*luk w – ’; cf. Greek ‘luk-os’, the velar k being the normal reflex of PIE k w in a Greek environment of [ + hi -bk + rd].
5) < ‘*pek w – ’; cf. Greek ‘péssō’ (< *pek w-yō), ‘-ss-‘/ ‘-tt-‘ being the normal sandhi for Greek juncture of [ + vel] + [γ].
6) Buck 1904:94) tacitly acknowledges this fact when he states that “L. bōs is borrowed from some O.-U. dialect” (my italics); Wright (1912:103) claims that “Lat. bōs for *vōs is an Umbrian-Samnitic loadworld.”
7) See Glare (1982:974) on italus.
8) Thus, the Latin word coquīna means ‘art of cookery,’ while the loan popīna refers to a ‘low-class eatery, bistro.’ See Glare 1982: ad loc.
9) Perhaps the situation is not entirely unlike that of the culinary (and sometimes also etymological) doublets found in English since the Norman invasion of a Latinate lexicon ranged itself alongside the earlier Germanic one. Thus, there is the barnyard domain on the one hand, and, on the other, that of the manor: cow ~ beef, swine ~ pork, meat ~ viand.
10) On the separate but vexed and seemingly irresolvable problem of accounting for the vocalism in the various reflexes of the PIE form, see Szemerényi 1956:186-90.
11) Superscripted x stands for anceps quantity.
12) When Tucker lists bo:s (page xxvi) as one of those words for which “the etymologies … are, so far as I am aware, either novel or based upon … (a) different … conception of the root” (page xxv), he would seem to be addressing, in the case of Latin bōs, his own suggested derivation from the root *bexu- *boxu- ‘swell out, be big,’ not mentioned by Halsey. For Halsey, already in 1893, clearly had derived bōs from a root *gu– ‘bellow’.
13) In particular, the plain voiced labial stop ([b]). Brugmann (1888:263) was moved to note that it “occurred more rarely in the Indg. prim. period than any of the other explosives …” and it “seems not to exist in suffixal elements.” Whitney, more dramatically, put it as follows: “Owing to the absence (or almost entire absence) of b in Indo-European, the Sanskrit b also is greatly exceeded in frequency by bh …”
14) But see Leumann 1977:158-9.
15) Even a transparently Indo-European form like bibō does not contain original initial b- in Latin. Presumably that b- represents an original reduplicating p(i)– on a root peH3 –> pō, with voicing of the root labial in (voiced) laryngeal environment. The reduplicating p- is then in turn voiced through regressive assimilation. See Leumann 1977:157; Bammesberger 1984:69-70.
16) Thus, the diachronic development of (pre-vocalic) initial digamma in Greek might be seen as a roughly parallel phenomenon.
17) The fact that a word-initial vowel (like [o] in oves, for example) has different formants than one preceded by a stop (whether voiced or voiceless [cf. Catford 1977:203]), would in any event have been sufficient to disambiguate the terms. That is, the o in oves would not have had the same acoustic property as the o in *voves, due to the latter’s o acquiring ‘spill-over’ formants from the preceding consonant. Even in antiquity it was recognized by theoreticians (e.g., Dionysius Thrax and Dionysius of Halicarnassus [Zirin 1970:44]) that one or more consonants in front of a word-initial vowel clearly ‘colored’ the vowel itself: “According to Dionysius of Halicarnassus the syllable ho of horos can be made longer by the addition of hr, i.e. hro, as in hrodos, while tro as in tropos and stro as in strop hos show even greater length, while remaining ‘short’ in the technical sense” (Zirin 1970:45). This observation by the ancients was surely founded on the actual experience of a variant acoustic perception due to ‘coloring’ (here through syllabic lengthening) of the word-initial vowel in question, and would appear to be equally applicable to Latin: “The basic Greek concept of syllabic quantity and its relation to duration does … apply to the Latin language to the extent that the prosodic structures involved are similar to those of Greek” (Zirin 1970:49). In both modern English (Wolfram and Johnson 1982:84-5) and, especially, German (Kenstowicz and Kisseberth 1979:141-2), it may be noted, there is a general tendency to obviate word-initial vowels in the phonetic output by inserting a pre-vocalic glottal stop to a vowel-initial surface form: (O –> ? / # _ V).
18) Here the assumption is that the Greek stem of the oblique cases (i.e. bo-) has become generalized. Indeed, it seems equally if not more reasonable to imagine Greek boós (sing. gen.) as possibly having undergone a(n erroneous) ‘latinate’ morphemic reanalysis as nominative singular to yield a stem bo-, which is then in turn subsequently reinterpreted analogically on the pattern of other third declension nouns for animals in the same semantic barnyard (e.g. mūs [stem mur-] or sūs [stem su-] would serve as paradigms for bōs relative to bo-, the latter perhaps further encouraged by the [oblique] pattern in ov-is [cf. b-ov-is]). In any event, the standard appeal to Italic borrowing fails to explain the source of the actual oblique forms of Latin bōs. Thus, if Umbrian bum is the source (cf. authorities cited above), as is often suggested, where did Latin bovem come from (or, for that matter, bovis etc.)? The standard appeal also has some difficulty accounting for actual ō [rather than ū] as the vowel in the Latin form (Ernout 1909:124-5). A hypothesis of Greek rather than Italic borrowing, though it appears neither more nor less likely, at least provides some rationale for the declensional system of the Latin word that we do in fact find.
19) In view of the important work on dialectology done by the French of the period (Bynon 1977:187), the following comment by Ernout (1909:124) is curious indeed: “Il est peu probable que bōs soit dû au souci d’éviter l’homonymie du pronom personnel vōs.”
20) That is, [v] –> [b] might have been generalizable at some level of the phonological system as homorganic (recall that both [b] and [v] are bilabial) [+ delayed release] –> [ – delayed release], a ‘rule’ (?) that might then have a mapped itself onto the identical change [n] –> [d].
21) A final point could be added in support of this conclusion (and I do so, with some hesitation). Although ‘psychological’ explanations are notoriously protean — especially so when we try to enter into the ‘psychology’ of a dead people who can no longer defend themselves against modern imputers – it might be argued that changing the community’s sign for an animal rather than that for a human would have presented less of a threat to the integrity of the very concept ‘human.’ In an almost iconic sense, in other words, meddling with the linguistic reality nōs-vōs would somehow at some level of awareness be sensed as an uncomfortable or ominous tinkering with the ontological reality ‘human.’ If so, one can no doubt readily appreciate why it was etymological *vōs ‘cow’ and not etymological vōs ‘you’ that was restructured.
Allen, W. Sidney, 1978: Vox Latina. Second Edition. Cambridge, Cambridge University Press.
Bammesberger, Alfred, 1984: Lateinische Sprachwissenschaft. Regensburg, Verlag Friedrich Pustet.
Brandenstein, Wilhelm, 1954: Griechische Sprachwissenschaft: Einleitung, Lautsystem, Etymologie. Berlin, Walter de Gruyter & Co.
Brugmann, Karl, 1888: Elements of the Comparative Grammar of the Indo-Germanic Languages. Vol 1: Introduction and Phonology [Translated by Joseph Wright]) New York, Westermann & Co.
Buck, Carl Darling, 1904: A Grammar of Oscan and Umbrian. Boston, Ginn & Company, Publishers.
–, 1933: Comparative Grammar of Greek and Latin. Chicago, The University of Chicago Press.
Bynon, Theodora, 1977: Historical Linguistics. Cambridge, Cambridge University Press.
Collinge, N.E., 1985: The Laws of Indo-European. Amsterdam/Philadelphia, John Benjamins Publishing Company.
Conway, R.S., 1923 (reprinted 1983): The Making of Latin: an Introduction to Latin, Greek and English Etymology. New York, Aristide D. Caratzas, Publisher.
Emonds, J., 1972: “A Reformulation of Grimm’s Law.” Contributions to generative Phonology. [ed. by Michael K. Brame], pages 108-22. Austin, University of Texas Press.
Ernout, A., 1953: Morphologie Historique du Latin. Paris, Librairie C. Klincksieck.
Frisk, Hjalmar, 1960: Griechisches Etymologisches Wörterbuch: Band I: A-Ko. Heidelberg, Carl Winter Universitätsverlag.
Gamkrelidze, T.; Ivanov, V., 1973: “Sprachtypologie und die Rekonstrucktion der gemeinindogermanischen Verschlüsse.” Phonetica 27.150-156.
Glare, P.G.W., 1982: Oxford Latin Dictionary. Oxford, Clarendon Press.
Halsey, Charles S., 1893 (reprinted 1983): An Etymology of Latin and Greek. New York, Aristide D. Caratzas, Publisher.
Hopper, Paul J., 1973: “Glottalized and murmured occlusives in Indo-European.” Glossa 7.141-166.
–, 1977: “Indo-European Consonantism and the New Look.” Orbis 26.57-78.
Kenstowicz, Michael; Kisseberth, Charles, 1979: Generative Phonology: Description and Theory. New York, Academic Press.
Kent, Roland G., 1940: The Sounds of Latin. Baltimore, Linguistic Society of America.
–,1946: The Forms of Latin: a Descriptive and Historical Morphology. Baltimore, Linguistic Society of America.
Kieckers, E., 1925: Historische griechische Grammatik: Lautlehre. Berlin, Walter de Gruyter & C.
Leumann, Manu, 1977: Lateinische Laut- und Formenlehre. Handbuch der Altertumswissenschaft 2.2.1. München, C.H. Beck’sche Verlagsbuchhandlung.
Maltby, Robert, 1985: “The Distribution of Greek Loan-Words in Terence.” CQ [n.s.] 35.110-123.
Meillet, A., 1928: Esquisse d’une Histoire de la Langue Latine. Paris, Librairie Hachette.
Miller, D. Gary, 1977: “Some Theoretical and Typological Implications of an IE Root Structure Constraint.” Journal of Indo-European Studies 5.31-40.
Muller, Frederik, 1926: Altitalisches Wörterbuch. Göttingen, Vandenhoeck & Ruprecht.
Palmer, L.R., 1954: The Latin Language. London, Faber and Faber Limited.
Planta, Robert von 1892: Grammatik der Oskisch-Umbrischen Dialekte. Erster Band. Straβburg, Verlag von Karl J. Trübner.
Pokorny, Julius, 1959: Indogermanisches Etymologisches Wörterbuch. I. Band. Bern und München, Francke Verlag.
Solta, Georg Renatus, 1974: Zur Stellung der lateinischen Sprache. Wien, Verlag der Österreichischen Akademie der Wissenschaften.
Szemerényi, O., 1956: “Latin re:s and the Indo-European long-diphthong stem nouns.” KZ 73.167-202.
Walde, A; Hoffman, J.B., 1938: Lateinisches etymologisches Wörterbuch. Erster Band. Heidelberg, Carl Winter’s Universitätsbuchhandlung.
Whitney, William Dwight, 1889 : Sanskrit Grammar. (15th issue  of the second edition ). Cambridge, Harvard University Press.
Williams, Gordon, 1982: “The Genesis of Poetry in Rome,” in The Cambridge History of Classical Literature: The Early Republic. Volume II, Part 1 (E.J. Kenney, editor), pages 53-59. Cambridge, Cambridge University Press.
Wolfram, Walt; Johnson, Robert, 1982: Phonological Analysis: Focus on American English. Washington, D.C., Center for Applied Linguistics.
Wright, Joseph, 1912: Comparative Grammar of the Greek Language. London, Oxford University Press.
Zirin, Ronald Andrew, 1970: The Phonological Basis of Latin Prosody. The Hague/Paris, Mouton.