Thursday, November 19, 2009


Version at BMCR home site
Michiel Arnoud Cor de Vaan, Etymological Dictionary of Latin and the Other Italic Languages (vol. 7 in the series "Leiden Indo-European Etymological Dictionary"). Leiden/Boston: Brill, 2008. Pp. xiii, 825. ISBN 9004167978. €229.00, $341.00.
Reviewed by Wolfgang David Cirilo de Melo, Universiteit van Gent

Table of Contents

There is no shortage of recent dictionaries of Latin, most of which contain a modest amount of etymological information, but de Vaan's book is the only purely etymological one. For the Sabellic languages we have of course Untermann's important Wörterbuch (2000), which provides a wealth of data and thorough discussion, but for Latin, before de Vaan's work appeared, one had to use Walde-Hofmann (1938) or Ernout-Meillet (1959). These older works are on the whole reliable, yet it is good to see an up-to-date dictionary fully adopting the laryngeal theory.

I have learnt much from reading de Vaan. I shall give just three examples. As an undergraduate I used to read the Miscellen or Gemischte Beiträge in various journals, but never understood why the first vowel in miscellus is long since the one in miscere is short; de Vaan lists the word under minor and derives it from *minuscellus > *minscellus, and now the quantity and the meaning make sense: the long vowel is the result of compensatory lengthening (loss of n before s) and the meaning is influenced by miscere. Second, the Appendix Probi contains an entry pauper mulier non paupera mulier. Here we can see how pauper adopted a more productive type of inflection (Italian povero). But the feminine paupera occurred already in Plautus (fr. xlvi Lindsay). I used to think that the Plautine form foreshadowed Romance developments, but de Vaan points out that the adjective was originally thematic and became athematic under the influence of its antonym diues. Hence the Plautine form may be a genuine archaism not connected with the later rethematization. Naturally I also learnt much about languages other than the Italic ones by working through de Vaan's book. In German a common insult is dämisch or dämlich "stupid", which most speakers derive from Dame "lady / woman". The correct etymology, however, is less misogynistic: the word is not connected with Dame, but ultimately has the same root as Latin temetum "alcoholic drink".

The structure of de Vaan's dictionary is clear and simple. After a brief list of abbreviations and introduction, the main part of the book consists of the dictionary entries, followed by a bibliography and useful indexes. The introduction outlines de Vaan's view of Indo-European, which is fairly orthodox; for instance, he believes in three laryngeals, whose places of articulation match those of the velar stops: palato-velar, pure velar, and labio-velar. He follows the traditional reconstruction of manner of articulation for the stops: there are voiced, voiceless, and voiced aspirated ones. After the Indo-European period, he accepts an Italo-Celtic and then a Proto-Italic stage and regards Venetic as an Italic language. It is useful to have de Vaan's outline of the major sound changes and the reconstructed sound system of Proto-Italic in the introduction.

The dictionary entries are systematic and easy to follow. The headword is followed by a rough translation, the declension or conjugation class, and other relevant information, such as first attestation or variant forms. This is followed by a section listing derivatives, again with first attestations. Next come the reconstructed Proto-Italic forms and the Italic cognates. After this we are presented with the Indo-European forms and cognates in non-Italic languages. Then we get a brief discussion and bibliography.

Reviewing a dictionary is very different from reviewing any other type of book. The reviewer is one of the few people who can reasonably be expected to read the entire work from cover to cover, while others are more likely to read only individual entries. And while the reader of a more general book will not be upset if one or two paragraphs contain mistakes because it is the whole that matters, people consulting a dictionary and finding a faulty entry will of course be upset because it is typically just one entry they need. For this reason the rest of my review is a list of suggested improvements, but it would be appropriate to offer first an overall assessment: de Vaan has produced a useful book fully incorporating recent research, and in many respects he replaces the older etymological dictionaries of Latin (though for the Sabellic languages I naturally prefer Untermann). The work is very reliable; if my list of corrections seems lengthy, one should not forget that the book is over eight-hundred pages long. However, I also have some points of criticism. I find it regrettable that de Vaan so rarely tells us his own opinions. For the most part we get only summaries of earlier research. As any etymologist freely admits, reconstructed forms can never be absolutely certain. But there are degrees of certainty. The etymology of equus, for example, is much more certain than that of abies. It would have been nice to have a clearer marking scheme for what is relatively certain, moderately certain, and completely uncertain. My final and most important point of criticism concerns the treatment of loanwords. All words that are definitely loanwords are excluded from the discussion. This is not a peculiarity of de Vaan's work: all dictionaries in the series follow the same procedure. However, the omission of all certain loanwords means that de Vaan's book cannot be a sufficient etymological work on its own.

But now it is time to turn to the finer points. My list of corrections begins with some obvious omissions and mistakes. Nux is feminine, not masculine, so it should be nux Abellana under Abella. Clam is listed as adverb. One would have liked to see at least a mention of its prepositional use in early Latin. The etymology of consulere advanced by de Vaan may well be correct, but *kom-sed- is dismissed too easily in view of considium in Plaut. Cas. 966. Under doleo we find the statement that the experiencer must originally have been expressed in the dative. This construction is actually well attested. Elutriare is listed under lauare as if it were a native formation, and the first two vowels are marked as long. But this rare word (attested in Laber. com. 151) is probably of Greek origin (ἔλυτρον) and the first two vowels are consequently short (the Laberius passage is inconclusive in this respect). Under facio I would have liked to see the form vhe:vhaked from the fibula Praenestina, even if de Vaan should consider the document a forgery. Under forceps de Vaan states that forfex is a by-form that arose by metathesis. This cannot be right. The form that arose by metathesis is forpex, and forfex is derived from this by long-distance assimilation or by association with agentive nouns in -fex. All three forms are feminine, but forfex is attested with masculine agreement in Vitr. 10. 2. 2, which makes the second explanation, that there was an association with agent nouns, more likely. The noun gerulifigulus is listed under gero as if it were securely attested. It is a textually problematic hapax legomenon (Plaut. Bacch. 381). Under hic, the nominative plural hisce, mostly used before vowels, is not even mentioned. It seems rather unlikely that in- is automatically lengthened before -gn- (see my comments below). Iuxta is normally said to have a short first vowel; de Vaan does not indicate vowel length here (or in iuxtim), but derives the word from *jougVsto-, in which case the first vowel ought to be long. Inscriptional evidence for long -a- in largus is mentioned, but, if this is accepted, should de Vaan not also mark the derivatives of the word as having a long vowel? Much the same can be said of forma and its derivatives. Liquidus is discussed under liqueo. I would have liked to see a note stating that the first syllable of liquidus occasionally scans as heavy in Lucretius (3. 427 liquidus umor aquai as hexameter ending). Muricidus, a hapax legomenon in Plautus (Epid. 333), is glossed by Paulus Diaconus as ignauus and stultus. We are clearly dealing with an insult, though the exact meaning remains unclear. We find this word under marceo in de Vaan's dictionary, who thinks that it belongs here for semantic reasons and that it could be a by-form of murcidus (his term is "corruption", but obviously an emendation to murcide (voc.) in the Plautus verse would be unmetrical). He could be right if the first three syllables are light, but the first and third syllables can also count as long, in which case the meaning could be "mouse-slayer" (an insult for a cowardly soldier?) or "wall-destroyer" (an insult for a thief, τοιχωρύχός, also perfossor parietum in Pseud. 980). Muscus is marked with a macron on the first vowel, muscosus is not; the length of the first vowel is unclear in either case. Under nasci de Vaan cites an inscription with nationu gratia "on account of giving birth"; the inscription is from Praeneste (CIL 14. 2863) and the spelling is archaic, with cratia rather than gratia. On noxit, which de Vaan regards as an s-present to noceo, I refer to my book on the early Latin verb; this is definitely not an s-present, but an aoristic formation. Pater is said to go back to a nursery form *pa, phonologically *pH2. Much as I like the laryngeal theory in its modern form, a phonological representation of babies' first babbling seems slightly over the top. Exactly the same could be said about atta "daddy" < *H2et-o-. As for penis, I agree with de Vaan that the most likely semantic development is from "tail" to "penis". However, here as elsewhere I have to take issue with his chronological considerations; he argues that his theory finds support in the fact that the meaning "tail" is attested as early as Naevius, while "penis" is not found before Catullus. Roman comedy deliberately avoided terms like "dick" or "cunt", so the mere date of first attestations does not mean anything. Under pluma de Vaan lists another Plautine hapax legomenon, plumatile, which he translates as "feathered" and scans with the two first syllables heavy. The word occurs in Epid. 233 in a list of women's dresses and stands in opposition to cumatile, which must be connected with κύμα. Since de Vaan's scansion leads to an impossible divided anapaest, it is better to scan the first two syllables as light and derive the word from πλύμα, as Duckworth (1940 ad loc.) does. His translation "watery or dishwatery" also brings out the joke. Under pungo we find a brief discussion of pugna and related words. Again it is claimed that before -gn- vowels are automatically long, a statement which is hard to maintain in view of Italian degno, segno < dignus, signum (not too much should be made of the spellings seignum in CIL 1^2. 42 and dIgne, with i longum, in CIL 6. 6314). Under quiris we read that the word might be a loanword with Sabellic connections; but in that case the labiovelar would be odd. Under saxum de Vaan lists Germanic words for "knife" as possible cognates, but then says that the connection between "rock" and "knife" is not straightforward. However, Latin saxum can of course also refer to a flint knife (as in the proverbial inter sacrum saxumque sto), so perhaps the connection is not far-fetched. Under se, it is wrongly claimed that Plautus still uses the accusative / ablative sed (he only uses med and ted). Under limus "oblique", an adverb sublimen is listed. Though this word occurs several times in Plautus manuscripts (e.g. Men. 992), we are probably dealing with nothing more than a corruption of adjectival forms of sublimis. Subuolturius is listed under uoltur as if it were a normal derivative, while in reality it is a nonce-formation punning on subaquilus (Plaut. Rud. 422). Under uerbera I miss a reference to the verb form uerberit in a lex regia, discussed in Szemerényi (1987). Since de Vaan distinguishes between vocalic u and consonantal v, it should be veruina rather than vervina (under veru), and conversely, volvi rather than volui (under volvo). The alternative form uotare for uetare is not a hapax, pace de Vaan. Vinum receives an Indo-European reconstruction and the fact that this might be a later loan (possibly from Semitic, where words of similar shape occur) is not even mentioned. The form ullo (Acc. trag. 293), a future perfect of ulcisci, should probably be restored to ulso; de Vaan comes up with an ad hoc sound change *-lks- > *-ls- > -ll-, but obviously l and s look rather similar in certain types of minuscule manuscripts.

The treatment of Faliscan data is sometimes less than satisfactory. Of course Bakkum's impressive treatment of Faliscan, published in 2009, was not yet out when de Vaan submitted his book, but some oddities point to a lack of knowledge of Faliscan. Sta MF 28,1 glossed as "(it) stands", and statuo MF 29, glossed as "I erect", are considered to be related with stare. Since in both cases these are the only words on the objects, the interpretation looks unlikely. It is clear that a dedicated object stands, the question is who set it up for whom. Probably we are dealing with abbreviated names. Tulom MF 68 certainly does not belong with tollere. The alleged meaning, "I set up", makes no sense on a one-word inscription, where we would expect the donor or the recipient; there are also morphological difficulties if the 1st sg. perfect ends in -ai (pe:parai EF 1), unless one resorts to the unlikely assumption that Faliscan preserved a separate aorist ending as well. Perhaps the inscription just means "of the Tulli" (with an old genitive plural in -om). It is odd that under unda, Faliscan umom "water vessel" EF 2 (< *ud-mo-) is not mentioned. Finally, the alleged Faliscan forms datu (under do), rected (under rego), sacru (under sacer), and uootum (under uoueo) all come from an inscription in the Faliscan alphabet (LF 214), but the language is clearly Latin; the Faliscan ending is -om or -o, not -um or -u, and long vowels are not written double in the Oscan style in Faliscan.

Venetic remains a language of which we know little. Several times de Vaan follows earlier literature in assigning meanings to words which in fact remain obscure. Atisteit (*Es 1222), under at and sto, is considered to have a prefix ati- and is glossed as adstat; the meaning of this word remains uncertain. Equally obscure is stati (Od 1), analysed by de Vaan as an instrumental singular meaning "weight". Whether Venetic poltos (Es 113) belongs with pellere, as alleged, is doubtful. A certain amount of scepticism, along the lines of Untermann (1980), would have done no harm.

In an etymological dictionary one expects correct indications of vowel length. By and large, de Vaan gets it right, but there are several unfortunate mistakes. Before -ns- and -nf-, and before -nct- and -nx-, vowel length is automatic because the nasal was lost and there was compensatory lengthening (Meiser 1998: 78); de Vaan omits macrons in around eighty cases. On the other hand, vowels are not automatically long before -gn- (see above), and while de Vaan does not always mark them as long here, he does so in around fifteen cases where they are probably short. Elsewhere, macrons are often omitted, which can lead to some confusion.3 Occasionally de Vaan marks a vowel that is short as long; thus the prefix re- always has a short vowel, so it should be réconcinnare4 (under concinnus), récusare (under causa), rélaxare (under laxus), and résumere (under emo). Similarly, de- is shortened before vowel, so we should have déambulare (under ambulo), déesse (under sum), déhortari (under horior), déorio (under haurio), and déosculari (under os "mouth"). The other cases where a short vowel is marked as long are the following (head words in brackets): conuóuere (uoueo), dáre (in several places), denté (dens), éra ("mistress", under ira), faucé (faux), latébricola, latébrosus (both under lateo), monétrix (moneo), nísi (ni), prófecto (facio), and prófiteri (fateor).

All remaining errors are of minor importance. The author's English is very good and almost free of mistakes, but Latin of the Empire is called Imperial Latin rather than "Empirical Latin" (p. 503). This new, important dictionary cannot be neglected by anyone interested in the history of words.


Bakkum, G. C. L. M. (2009), The Latin Dialect of the Ager Faliscus: 150 Years of Scholarship, 2 vols. (Amsterdam).

de Melo, W. D. C. (2007), The Early Latin Verb System: Archaic Forms in Plautus, Terence, and Beyond (Oxford).

Duckworth, G. E. (1940), T. Macci Plauti Epidicus: Edited with Critical Apparatus and Commentary, in Which Is Included the Work of the Late Arthur L. Wheeler (Princeton).

Ernout, A. and Meillet, A. (1959^4), Dictionnaire étymologique de la langue latine: Histoire des mots (Paris).

Pellegrini, G. B. and Prosdocimi, A. L. (1967), La lingua venetica, vol. 1: Le iscrizioni (Padua).

Szemerényi, O. S. L. (1987), "Si parentem puer verberit, ast olle plorassit", in Scripta Minora: Selected Essays in Indo-European, Greek, and Latin, vol. 2: Latin, ed. by P. Considine and J. T. Hooker (Innsbruck), 892-910.

Untermann, J. (1980), "Die venetische Sprache: Bericht und Besinnung", in Glotta 58: 281-317.

------ (2000), Wörterbuch des Oskisch-Umbrischen (Heidelberg).

Walde, A. and Hofmann, J. B. (1938^3), Lateinisches etymologisches Wörterbuch (2 vols. + reg.) (Heidelberg).


1.   The numbers are those in Bakkum (2009).
2.   The numbering system is that introduced by Pellegrini and Prosdocimi (1967).
3.   The correct forms are as follows (I leave the head words in brackets unmarked): âctiô, âctor, âctus, âctûtum (all four under ago), adârêscere (areo), adulêscentia, adulêscentiârî, adulêscentulus (all three under alo), afflîctim (fligere), ârdêre, ârdor, ârdus, ârfacere (all four under areo), ascrîptîuus (scribo), bilîx (licium), cânûtus (canus), capessitûrus (capio), catîllâre, catîllus (both under catina), cênâculum, cênâre, cênâticus, cênâtus (all four under cena), clâmôs (clamo), cômptiônâlis, cômptus (both under emo), cônâtum (conor), conciliâtrîx (calo), concubînâtus (cumbo), cônsentês (sum), corrêctor (rego), crâstinus (cras), creâtrîx (creo), dêlectâmentum (lacio), dêpudîcâre (pudeo), dîlêctus (lego), êbriâcus, êbriolâtus, êbriolus (all three under ebrius), êlêctilis (lego), êmptîcius, êmptor, êmpturîre, êmptus, exêmptiô (all five under emo), facit ârê (areo), farînârius (far), fautrîx (faueo), fictrîx (fingo), fugitîuârius (fugio), honôrârius, honôrâtus (both under honos), immâtûrus (maturus), indipîscî (endo), inêscâre (edo "eat"), labôrâre (labor), lâpsus (labo), lupîllum (lupus), mandûcâre (mando), mendîcâbulum (mendum), môrâtus, môrigerârî, môrigerâtiô, môrigerus (all four under mos), mûstêlînus, mûstricula (both under mus), nefâstus (fas), neglêctus (lego), nîxârî, nîxus (both under nitor), obuâgîre (uagio), ôrâculum, ôrâtiô, ôrâtor (all three under oro), ôsculentia, Ôstia (both under os "mouth"), pâstus (pascere), praefestînâtim (festino), praemâtûrus (maturus), prômptâre, prômptârius, prômptus (all three under emo), Pûblius (populus), quadrâgintâ (quattuor), quârticeps (-ceps), quîntânus (quinque), quînticeps (-ceps), quîntîlis (quinque), rêctâ, rêctor (both under rego), redêmptitâre, redêmptor (both under emo), rêgillus, rêgnâre, rêgnâtor, rêgnum (all four under rex), rôbôsem (robur), scîtâmenta (scire), scrôfipâscus (scrofa), sêmêstris (sex), struîx (struo), subrêctitâre (rego), sûmptiô, sûmptuôsus (both under emo), tâctus (tango), trilîx (licium), uîndêmia, uîndêmiâtor, uîndêmitor (all three under emo), ûtî (utor).
4.   For typographical reasons I use the acute here to mark a short vowel.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.