Razred za filološke znanosti HAZU

In this paper new etymologies of four Latin plant names are proposed. Lat. ferula ‘giant fennel’ is derived from *feselā, from the PIE root *b^hes- ‘to scrape’; Lat. lolium ‘darnel’ is argued to be a loanword from some Pre-PIE substrate and related to MW ller ‘darnel’; Lat. rumex ‘sorrel’ is derived from the PIE root *h₁rewg’- ‘belch, vomit’ and related to Lith. rūgštỹnė ‘sorrel’ (from the same root), and Lat. rūscus ‘butcher’s broom’ is compared to OIr. rúsc ‘bark’ and derived from the PIE root *(H)rewH- ‘to scrape, scratch, tear off’.

Keywords: Latin etymology, Proto-Indo-European, ferula, lolium, rumex, ruscus

1. Latin ferula and the related words

Lat. ferula ‘giant fennel; stiff, slender branch, rod, staff, stick’ (attested since Varro) does not have an etymology (WH I: 487, EM 230, DV: 214). However, a derivation from an Italic root *fes- and a connection with festūca ‘stalk, straw’ seems plausible both semantically and formally. Ferula would be derivable from *feselā, while festūca (attested since Plautus) is formed with the unproductive suffix -ūca-/-ūco- (Balles 2008: 57), attested also in mandūcus ‘glutton, greedy eater’, albūcus ‘asphodel (kind of onion), bulb of the asphodel’, verrūca ‘wart, steep place, excrescence’, aerūca ‘a kind of verdigris’, carrūca ‘cart’, ērūca ‘rocket, rucola’. However, like ferula, festūca does not have an etymology (WH I: 489, EM 231, DV: 216). We are thus left with the root fes- which can go back to PIE *d^hes-, *g^whes- and *b^hes-.

We propose a derivation from PIE *b^hes- ‘to scrape’ (Skt. bábhasti ‘chews’, bhásma- ‘ashes’, EWA II: 257, IEW 145f.,1 LIV s. v. *bhes-), cf. Germ. Besen ‘broom’ (OHG besmo), Eng. besom ‘a broom made of sticks or straws’ from the same root. For the semantic connection cf. also Germ. Gemeines Steckenkraut ‘Ferula communis’ (from Stecken ‘stick’). The leaves of giant fennel actually look like a broom, or other instruments for scraping or dusting (the floor). Therefore, the original meaning of ferula was probably ‘the thing that scrapes’ or ‘a broom’, and the transfer of meaning to the plant was metaphorical.

The nominal derivatives of PIE *b^hes- include the adjective *b^hoso- ‘naked’ (< ‘scraped’) (OCS bosъ ‘barefoot’, Lith. ba͂sas, OHG bar, perhaps also Arm. bok ‘barefoot’ (presumably from *b^hoso-g^wh₂o- ‘walking barefoot’, where *g^wh₂o- is the same root as in Skt. gā̍ti ‘goes’).

Possible cognates in Hittite are pašihaezi ‘rubs, squeezes, crushes’, paššila- (c.) ‘stone, pebble’, but note that Kloekhorst (EDHIL 650) doubts them, since the suffixes -iha- and -ila- are otherwise unattested.

However that may be, we believe that the etymological connection between Lat. ferula ‘giant fennel, rod, stick’ and festūca ‘stack straw’ is all but ascertained, and that the derivation of both words from the PIE root *b^hes- ‘to scrape’ is plausible.

2. Latin lolium and Middle Welsh ller

Latin lolium ‘kind of grass, darnel, Lolium temulentum’ is attested since Plautus. It does not have many derivatives (only denominal adjectives loliāceus and loliārius) and it does not have an etymology (DV: 348, EM 365). WH I: 819, followed by IEW (650), speculate that this word is somehow derived from the onomatopoetic root *lā- (Skt. rāyati ‘bellows’, Lat. lāmentum ‘lament’, OCS lajati ‘belch’), but this is both formally and semantically unconvincing.

In European languages, there are several words for ‘darnel’ that are similar to Lat. lolium, but they are clearly Latin loanwords, e.g. Germ. Lolch < OHG lolli (KL 525) and Croat. ljûlj (ERHJ 577). Proto-Germanic *terwōn ‘tares’ (ME tare ‘wild or cultivated vetch’, pl. ‘weeds growing in grain fields, darnel’, Du. tarwe ‘wheat’) is related to Skt. dūrvā- ‘a kind of grass, Panicum dactylon’ (EDPG 514), but it is formally too dissimilar for there to be any connection with Lat. lolium. In several European languages, the name of the plant Lolium temulentum is connected with adjectives meaning ‘drunk’ because of its intoxicating qualities (cf. Fr. ivraie2 < Lat. ebriācus, Eng. darnel, which is probably related to Walloon darne, derne ‘stunned, drunk’, cf. also Ukrainian (pažitnicja) p’janka, Macedonian pijanec from PSl. *pьjanъ ‘drunk’), but this does not help us elucidate the etymology of Lat. lolium.

However, the Latin word for ‘darnel’ has a hitherto undiscovered parallel in Celtic. We believe that Lat. lolium is related to MW ller pl. ‘darnel, cockle’ (singulative lleren f.). Since the meanings are identical, and the forms are rather similar, this hypothesis appears rather likely. However, the exact relationship between these two words is not immediately clear. GPC (s. v. ller) does not give any etymology of MW ller, and the word is not mentioned in EDPC.3 Was the Welsh word borrowed from Latin, like so many other words, or are we dealing with common inheritance? In principle, MW ller (attested since the 14^th century) could be derived from earlier *llel by dissimilation, but the vocalism of this word would still not match Lat. lolium. Welsh e cannot be derived from earlier *o by the Brittonic vowel affection rules, since we would expect *y (as in pont ‘bridge’ pl. pynt, or corn ‘horn’ pl. cyrn < *kornī, Lat. memoria ‘memory’ >> W myfyr ‘meditation’). Latin short *o is generally preserved in old Latin loanwords in Welsh, cf. Lat. corpus ‘body’ >> W corff, Lat. columna ‘column’ >> W colofn , etc. (Haarmann 1970: 136, Schrijver 1995: 258-260).4 Moreover, there are no indications that Lat. lolium was borrowed from Celtic, and even if it were, the problem of divergent vocalism would still remain. It is therefore more probable that both MW ller and Lat. lolium are inherited (or borrowed very early from some non-IE source), rather than that one of these words represents a borrowing (either from Latin into Brittonic, or from Celtic into Latin).

By comparing MW ller and Lat. lolium we can posit a proto-form *le/oL-, where *L stands for a liquid resonant (either *l or *r). However, there are no parallels in the other IE languages (except, as mentioned earlier, those words for ‘darnel’ that were borrowed from Latin, such as Croat. ljûlj). Hence, it appears probable that Italo-Celtic *le/oL- ‘darnel’ is an early borrowing from some pre-IE substratum language of Western Europe. This is all the more likely in view of the fact that a root with two liquids would contradict the rules of root-formation in PIE. Moreover, if Lat. lolium were inherited, it would have been dissimilated to *lorium by the “liquid dissimilation rule” (Meiser 1998: 127), cf. populāris ‘people’s, popular’ (from populus ‘people’) vs. amīcālis ‘friendly’ (from amīcus ‘friend’). The relationship of Lat. lolium to MW ller is reminiscent of the one between Lat. līlium ‘lily, Lilium candidum’ to Gr. leírion ‘id.’. These words are also of probable substratum origin (EDG 845, Chantraine 629 compares Coptic hrêri, hlêli, ‘lily’). Both in the case of lolium and līlium it seems that either there was an assimilation in Latin (*l...r > l...l), or the donor language had a sound which was adopted as -l- in Latin, but as -r- in another language.

Although it cannot be absolutely ruled out that MW ller ‘darnel’ was borrowed from Lat. lolium, and underwent subsequent irregular sound changes, we think it is much more probable that both Lat. lolium and MW ller were borrowed from some pre-Indo-European substratum language of Western Europe. Words for weeds and plants with little economic value are often of substratum origin.5

3. Lat. rumex ‘sorrel’

Lat. rumex ‘sorrel, Rumex patientia’ does not have an etymology. It is not mentioned in DV, and WH II: 450 implausibly connect it to Lat. rūta ‘a bitter herb, rue’ (which is a Greek loanword) and/or to OHG sūr ‘bitter’. EM 581 frankly say ‘sans étymologie’.

The word rumex is attested since Plautus (apponunt rumicem, brassicam ‘they add sorrel, cabbage’, Ps. 815). It is otherwise a rare word, but it appears in Pliny (Nat. hist. 11.18). Its reflexes in the Romance languages include Fr. ronce, It. romice and Sp. romaza. No derivatives of rumex are attested in Classical Latin, but some are attested in Romance languages, e.g. Fr. ronceux ‘surrounded by thorn-trees', roncière ‘place abundant in brambles’, etc.

The suffix -ix/-ex is very common in Latin plant names (cf., e.g. īlex ‘holm oak’, filix ‘fern’, cārex ‘rush, sedge’, cf. Balles 2008: 50, Matasović 2016), but the root rum- is not attested in other plant names in Latin.

In light of the fact that leaves of sorrel are very sour, the phytonym meaning ‘sorrel’ is derived from the adjectives meaning ‘sour’ in most European languages: cf., e.g., Eng. sorrel < OFr. surele from a Frankish adjective related to OHG sūr ‘sour’, Croat. kȉselica ‘sorrel’ vs. kȉseo ‘sour’, Fr. oseille < Lat. acidula; therefore, we propose to derive Lat. rumex from PIE *h₁rewg- ‘belch, vomit’. This PIE root is reflected in the Lithuanian name of sorrel, Lith. rūgštỹnė (Smoczyński 2007: 523), but also in the verb rū́gti ‘get sour’, Gr. ereúgomai ‘belch out, vomit’, Lat. irūgīre ‘roar’, ērūgere ‘disgorge noisily, belch’, Russ. rygát’ ‘belch’, Croat. rȉgati ‘belch’, Po. rzygać ‘throw up’, etc. (EDSIL 441f., IEW 867). The reconstruction of the palatalized *g’ in the root is somewhat problematic in light of the Balto-Slavic reflexes, but these could be due to secondary depalatalization, whereas Arm. orcam ‘belch’ (if it comes from *orucam < *erucam) points to the root *h₁rewg’- with the palatalized velar.

On the formal side, we could think of the derivation from PIE *h₁rug’-sm-ek-; the consonant group *-gsm- was reduced to -m- in Latin, e.g. in iūmentum ‘draught animal, beast of burden’ < *yowgs-men-to (from the root of iungo), cf. IOUXMENTA (CIL 1). A similar development occurred in lūna ‘moon’ < *lowksneh₂ (from the root of lūx ‘light’) and, probably, strēnuus ‘active, vigorous’ < *streg-sno-, cf. OIr. trén ‘strong’, Szemerényi 1989: 23f.). However, this development was associated with a compensatory lengthening before consonant groups involving *s and a nasal (e.g. in cōmis ‘friendly’ < cosmis), aēnus ‘made of bronze’ < *aesnos (cf. aes ‘bronze’), ēmergō ‘emerge’ < *eks-mezgō, (Meiser 1998: 118, Weiss 2010: 178), cf. also tēmō ‘pole or yoke-beam of a cart or plough’ < *tenksmen- (OHG dīhsala ‘thill’, DV 610). Therefore, we would expect PIE *h₁rug-sm-ek-s to develop as *rusmeks > *rūmex.6 However, short -u- in rumex is assured, e.g., by Virg. Mor. 73: fecundusque rumex malvaeque inulaeque virebant (“(Here) throve rich sorrow, mallows and elecampane”). It follows that either rumex cannot be derived from PIE *h₁rewg’-,7 with the suffix *-sm- (and the common phytonymic suffix -ex), or the original long *-ū- was for some reason shortened in the Latin word for ‘sorrel’.

The latter might indeed the case. Unattested *rūmex ‘sorrel’ could have developed to rumex by analogy with the homophonous noun rumex ‘a kind of spear or javelin, hunting spear’.8 The analogy may have been facilitated by the fact that leaves of sorrel are javelin-like in form, so the word for ‘sorrel’ and the word for ‘(a kind of) javelin’ may have been connected by folk etymology.

Lat. rumex ‘a kind of spear or javelin, hunting spear’ is a rare word, but it is attested in Lucilius (1315) and Aulus Gellius (10.25.2). Its origin is uncertain, but one might speculate that this word is derived from the same root as ruō ‘to rush, tumble down’ < PIE *h₁rew- ‘hurry, rush, strike’ (OIr. rúathar ‘assault’, Alb. ra [3sg. aor] ‘strike, hit’, Gr. oroúō ‘hurry’, EM 582f. According to De Vaan (DV 530), from this root we also have rūna ‘a kind of weapon’ (attested since Naevius). If rūna is from *h₁rew-neh₂, rūmex can be built to a parallel (but unattested) derivative *rūma < *h₁rew-meh₂. For the pattern, cf. caudex ‘tree-trunk’ from cauda ‘tail’ (DV 99).

Furthermore, the short -u- in rumex may be due to analogy with another word with which it was often associated, either in syntagms, or conceptually. One possibility would be the adjective ruber ‘red’ (with short -u-), since stems and flowers of certain kinds of sorrel (e.g. sheep’s sorrel, rumex acetosella, also known as “red sorrel”) are indeed red.

The least likely, but not utterly impossible, explanation of the short vowel in rumex, would be to assume that it was generalized from forms where the geminate was simplified by “mamilla rule” (Meiser 1998: 127), i.e. in pretonic positions (cf. the diminutive mamilla vs. the base word mamma ‘teat’ where the geminate was preserved after the accented syllable). Such forms could have been the diminutive *rumicilla, or the derivatives such as *rumicōsus (locus) ‘a place where sorrel grows’. Since rumex is a rare word in Latin text, it is not unusual that its derivatives are not attested, but their reflexes exist in Romance languages such as French (as mentioned above).

The most plausible hypothesis, perhaps, is to derive rumex from *rug-m-Vk-, i.e. without *-s- separating the velar from the nasal. Apparently the cluster -gm- is preserved in tegmen ‘cover’ and agmen ‘stream, crowd’ (Meiser 1998: 121); however, as these words are obvious derivatives from the verbal roots tegō ‘cover’ and agō ‘drive’, respectively, they may have restored the consonant -g- by analogy with the verbs (in tegmen the noun tegimen ‘cover’ could also have played a role). A different development of the cluster -gm- can be apparently be observed in flamma ‘flame’ (from the root of flagrō ‘burn’) and, perhaps, in ammentum (besides āmentum) ‘thong or loop’ (DV: 30). The geminates in those words can be due to littera-rule, in which case they can also be derived from *flag-sma and *ag-smen-tom respectively. Therefore, if tegmen and agmen have -g- restored by analogy, nothing stands in the way to deriving Lat. rumex ‘sorrel’ from *h₁rug-m-Vk- regularly.

Be that as it may, in light of the overwhelming number of semantic parallels, we believe that Lat. rumex ‘sorrel’ is a derivative of PIE *h₁rewg’- ‘belch, vomit’, just like Lith. rūgštỹnė ‘sorrel’ is.

4. Lat. rūscus ‘butcher’s broom’

Lat. rūscus [o m] ‘butcher’s broom, Ruscus aculeatus’ (also rūscum [o n]) is attested since Virgil;9 its derivative rūsceus ‘coloured like the berries of butcher’s broom’ is attested since Cato; there is also the adjective rūscārius (with a variant rustārius, which seems to be secondary) ‘of or for butcher’s broom’ (also attested since Cato). The word rūscus (rūscum) does not have an etymology (DV 530, WH II: 454f., EM 583. However, its exact formal equivalent is attested in Celtic: both Lat. rūscus and PCelt. *rūsko- (OIr. rúsc m. gl. cortex, ‘bark’, MW risgl, W rhisg m. ‘bark’, EDPC 317)10 can be derived from PIE *(H)ruH-s-ko-, probably from the root of ON rýja ‘shear (of sheep)’, OCS ryti ‘dig’, rъvati ‘tear’, Lat. ruo ‘rush down, tumble down’, ToB ru- ‘pull out’, Skt. rávati ‘wound’, Lith. ráuju, ráuti ‘pluck, weed out’ etc. (LIV 461, IEW 868). The form *(H)ruH-s-ko- seems to have been built by adding a nominal suffix *-ko-11 to a verbal s-stem (where *-s- may have originally been a present-stem forming suffix); such an s-stem is indeed found in Balto-Slavic, e. g. OCS rušiti ‘destroy, damage’ (also Croat. rȕšiti, Russ. rúšit’ ERHJ II: 313), Lith. rau֮sti ‘scratch, dig’ (with unexpected circumflex), rū́sas ‘a hole dug out for winter potatoes’ (Smoczyński 2007: 507), perhaps also ON rúst ‘chipped off pieces, ruins’ and ON rysja ‘to flay’ (EDPG 418).

The problem with this etymology is, admittedly, in the semantic connection between the plant name (Lat. rūscus) and the meaning of the alleged cognates (‘bark’ in Celtic and the verbal meanings of the root *(H)rewH- stretching from ‘dig, dig out’ and ‘tear’ (in Slavic) to ‘shear’ (in Old Norse), ‘pluck’ (in Lithuanian) and ‘pull out’ (in Tocharian). It is obvious that the root *(H)rewH- denoted a (potentially violent) physical action, and most, if not all, of the attested meanings can be derived from ‘to scratch, to tear off’ or ‘to sweep’. From this it is easily understandable that the meaning ‘bark’ could develop in the nominal derivative *rū-s-ko- in Celtic, as ‘bark’ is precisely that which is scratched, or torn off trees. The meaning of Lat. rūscus, on the other hand, becomes understandable if one considers the English name of the plant, ‘butcher’s broom’, as branches of Ruscus aculeatus were traditionally used as brooms (not only in butcher’s shops),12 i.e. for sweeping. Its sharp leaves can also be used for other similar physical actions, such as scratching or whipping, hence its Croatian names mètlika (from mètla ‘broom’) and jéževina (a derivative of jêž ‘hedgehog’).13 The original meaning of Lat. rūscus would have been ‘the sweeper’ or something similar, and, in that case, the meaning ‘butcher’s broom, Ruscus aculeus’ is metaphorical.

Beekes, Robert S. P. 2010.Etymological Dictionary of Greek, Leiden: Brill.

Derksen, Rick. 2008. Etymological Dictionary of the Slavic Inherited Lexicon, Leiden: Brill.

DV = De Vaan, Michiel. 2007. Etymological Dictionary of Latin and the other Italic Languages, Leiden: Brill.

EDPC = Matasović, Ranko. 2009. Etymological Dictionary of Proto-Celtic, Leiden: Brill.

EM = Ernout, A. & Meillet, A. 1967. Dictionnaire étymologique de la langue latine, Paris.

Haarmann, Harald. 1970. Der lateinische Lehnwortschatz im kymrischen, Bonn: Romanisches Seminar der Universität Bonn.

KL = Kluge, Friedrich. 1989. Etymologisches Wörterbuch der deutschen Sprache, Berlin: De Gruyter.

Kroonen, Guus. 2013. Etymological Dictionary of Proto-Germanic, Leiden: Brill.

Kloekhorst, Alwin. 2008.Etymological Dictionary of the Hittite Inherited Lexicon, Leiden: Brill.

Lausberg, Heinrich. 1967-1969. Romanische Sprachwissenschaft (3. Ausgabe), I-II. Berlin: De Gruyter.

LIV = H. Rixet alii. 2003. Lexikon der indogermanischen Verben, Wiesbaden: Reichert.

Orel, Vladimir. 1998. Albanian Etymological Dictionary, Leiden: Brill.

REW = Meyer-Lübke, Wilhelm. ⁵1972. Romanisches etymologisches Wörterbuch. Heidelberg: Winter.

Schrijver, Peter. 1995. Studies in British Celtic Historical Phonology. Amsterdam & Atlanta: Rodopi.

Smoczyński, Wojciech. 2007. Słownik etymologiczny języka litewskiego. Vilnius: Vilniaus Universitetas.

Szemerényi, Oswald. 1989. An den Quellen des lateinischen Wortschatzes. Innsbruck: Innsbrücker Studien zur Sprachwissenschaft.

Topalli, Kolec. 2017. Fjalor etimologjik i gjuhës shqipe. Tiranë: Qendra Studimeve Albanologjike.

Weiss, Michael. 2010.Outline of the Historical and Comparative Grammar of Latin, Ann Arbor: Beech Stave Press.

WH = A. Walde & J. B. Hofmann. 1938. Lateinisches etymologisches Wörterbuch, Heidelberg: Winter.

ABBREVIATIONS: Alb. = Albanian, Arm. = Armenian, Bret. = Breton, Croat. = Croatian, Du. = Dutch, Fr. = French, Germ. = German, Gr. = Greek, Hitt. = Hittite, Lat. = Latin, ME = Middle English, MLat. = Medieval Latin, MW = Middle Welsh, OCS = Old Church Slavonic, OFr. = Old French, OHG = Old High German, OIr. = Old Irish, ON = Old Norse, PCelt. = Proto-Celtic, Po. = Polish, PSl. = Proto-Slavic, Russ. = Russian, Skt. = Sanskrit, Sp. = Spanish, ToB = Tocharian B, W = Welsh.

Iz latinske fitonimije: etimologije fitonima ferula „divlji komorač, obična firula”, lolium „ljulj”, rumex „kiselica” i rūscus „veprina”