<<

New and Old Samoyed Etymologies

Ante Aikio

New and Old Samoyed Etymologies*

The paper discusses etymological equations between Samoyedic and Finno-Ugric . The presentation includes both new etymologies and comments on previously presented comparisons that have been ignored or rejected in the strictest treatments of Uralic historical phonology. A total of 28 Samoyedic word families are subjected to etymological analysis, and several other etymologies are briefly commented upon. It is argued that even in a framework of relatively strict sound laws, the number of Proto- Uralic etymologies with Samoyedic cognates turns out to be higher than the most critical estimates presented earlier.

0. Introduction

The constitute the least studied branch of the Uralic family. Undoubtedly, the only spoken today that still have not been sufficiently documented belong to the Samoyedic branch. However, Samoyedic obviously occupies one of the key positions in comparative reconstruction in Uralic. According to the traditional view, the primary genealogical distinction in Uralic is between Samoyedic and all the other Uralic languages, i.. the Finno-Ugric branch. Even if this traditional dichotomy of the Uralic languages is not accepted (and perhaps it should not be, as the linguistic arguments supporting it are not particularly strong), it is clear that because of its geographically peripheral location, Samoyedic still plays a more important role in the reconstruction of Proto- Uralic than many of the central branches (see e.. Korhonen 1986, esp. 216–217). For a long time Proto-Finno-Ugric and Proto-Uralic were mainly ap- proached from the perspective of the western periphery of the . While extralinguistic factors may have contributed to some extent to such an approach, mostly it was due to the fact that the necessary low-level comparative research in the field of the eastern Uralic languages had not been done. But after Proto-Samoyed had been reconstructed in detail, a phonologically systematic between the western and eastern

9 Ante Aikio peripheries (Finnic-Samic-Mordvinic and Samoyedic, respectively) became possible, and a clearer picture of Proto-Uralic phonology began to emerge (note especially Janhunen 1981). This advance was largely achieved through a phonologically critical re-evaluation of the etymological corpus that led to a significant decrease in the number of Proto-Uralic etymons that were considered reliably reconstructed (see Sammallahti 1988: 479). This was to be expected, because only after the sound laws leading from Proto-Uralic to Proto-Samoyed and the present-day Samoyed languages had been recon- structed in detail were there exact phonological criteria for etymological cognateness between Samoyedic and the western branches of Uralic. The purpose of this paper is twofold. First, my goal is simply to etymologize certain Samoyedic words, because relatively little etymological basic research has yet been done in the field of Samoyedic linguistics. Secondly, my intention is to examine the predictive power of the recon- structed sound laws leading from Proto-Uralic to Proto-Samoyed. I intend to show that on the basis of the proposed sound laws one has not only been able to explain the data on which they are based, but even managed to predict a substantial amount of previously unnoticed etymological equations. I will discuss Proto-Uralic etymologies which had not been noticed up to the time of the publication of Janhunen 1981; all the cases produced here are new, except etymologies 1.6., 1.11. and 1.19., which were originally proposed by Daniel Abondolo (1996). Additionally, I shall present new arguments supporting previously proposed etymologies that were either rejected or (possibly accidentally) ignored in Janhunen 1981. The phonological treatment of the etymologies is based on the sound laws established by Janhunen (..) and Sammallahti (1988), unless otherwise specified. I operate with the traditional concept of a Finno-Ugric proto- language (PFU) as distinct from Proto-Uralic (PU), despite its controversial status. However, Sammallahti’ reconstructions of Proto-Finno-Permic and Proto-Finno-Volgaic are considered identical with PFU, as the assumed phonological innovations defining these proto-languages are minor. When established PU or PFU etymologies are discussed, a reference to Uralisches Etymologisches Wörterbuch (UEW) is made, where cognates can be found, but the reconstructed forms do not always accord with those given in this dictionary. Proto-Samoyed reconstructions are based on Janhunen’s Samojedischer Wortschatz (SW), except for the non-initial syllable , in the reconstruction of which I subscribe to the views of Eugen Helimski (see e.g. 1997: 70).

10 New and Old Samoyed Etymologies

Only Tundra Nenets forms are given for Nenets, and these are transcribed according to the phonological transcription employed by Tapani Salminen (see e.g. 1997: 31–35). Mator words are transcribed according to Helimski 1997. Nganasan, Enets, Selkup and Kamass words are transcribed according to the source used, except that unnecessary diacritics that have no phonemic relevance have been omitted.

1. New Uralic etymologies

1.1. *cana- ‘to rub, wear out (tr.)’ [> Nenets tanga-], *cano- ‘to wear out (intr.)’ [> Nenets tango-, Kamass tan5o-] (SW: 151) < PU *cav na- ‘to strike, beat, rub’

Reflexes of the PS etymon are attested only in Nenets and Kamass. The intransitive *cano- is a derivative. There are many other Proto-Samoyed verb pairs with the stem correspondence *a / *e (transitive) ~ *o / *ö (intransitive), e.g. PS *ita-{ ‘to hang (tr.)’ ~ *itö-{ ‘to hang (intr.)’, *kaje- ‘to leave (tr.)’ ~ *kajo- ‘to remain’ (cf. SW: 25–26, 58). The labial vowel goes back to the PU passive-reflexive suffix *, although it is unclear why it is reflected as *ö also in some back vocalic stems, which is in contradiction with . This suffix has cognates in at least Samic, Finnic, Mordvin, Mansi and Hungarian (Kulonen 1989: 51–53). There are also pairs of transitive and intransitive in Mari which differ in their stem vowels, e.g. woze- ‘to drop’ ~ woza- ‘to fall’. The intransitive verbs may have also originally contained the same suffix. Note especially Mari kode- ‘to leave (tr.)’ ~ koda- ‘to remain’, which correspond to Sami guod∞ di-∞ ‘to leave’ ~ gud∞ do-∞ ‘to be left (passive)’ and PS *kaje- ~ *kajo- (< PU *kad'a- ‘to leave’, *kad'a-w- ‘to be left, to remain’) (but cf. Kangasmaa-Minn 1998: 234–235).1 According to SW, the Proto-Samoyed reconstruction is *t1ana-, where *t1 can represent either PS * or *c. These consonants have retained their opposition only in Selkup, where reflexes of this word are not encountered. However, PS *c may be assumed here, because the word has a counterpart in PFU *cav na- ‘to strike, beat’, which is attested in Mordvin cavo- ‘to strike’, Mari co7 nge- ‘to beat’, (Vasjugan) cv og-, (Konda) cov nx- and Mansi (Mid-Konda) se58 nkØ - ‘to strike, kick’ (UEW: 53–54). The comparison is phonologically almost flawless. On the basis of Mordvin, Mari and Mansi PFU *a can be reconstructed in the first syllable; only the Khanty vowel

11 Ante Aikio

(Proto-Khanty *5o pro *a5 ) is irregular. The illabial vowel in Mansi can be explained by a transition of labiality to the following velar consonants (Proto-Mansi *sv ı{‰ nØ Ø - < *sv u5 nk-). The predictable reflex of PFU *a is Proto- Mansi *u5 . The comparison is also semantically natural. A parallel is found in English strike, which derives from Old English strıcan‰ ‘to move, stroke, rub’. English stroke and German Streich ‘blow, stroke’, streichen ‘to spread, smear, stroke’ belong to the same Germanic word family. (Klein 1966–1967 s.v., Kluge 1995 s.v.) Another word worth noting is Finnish lyöttää ‘to strike (causative); to rub, chafe’ (SSA s.v. lyödä).

1.2. PS *jqtq ‘block of wood’ [> Nenets yødo ‘block of wood (e.g. under a boat, or for beating something); pole (in a trap)’2, Enets (der.) дёго' дёзозту" ‘pole in a trap’ (дёго' ‘trap’), Selkup catv i{ ‘beater, cudgel’] (SW: 36) < PU *luta ‘block of wood? (some simple wooden tool)’

The reflexes of PS *jqtq display considerable heterogeneity in meaning, but all of them refer to some type of simple wooden tool. The word family is of Uralic origin: it is cognate with Finnic-Samic *luta, which is reflected in North Sami lohti ‘wedge’ (in also ‘plug’) and Finnish luta ‘a split twig with which the corners of a bark container are fastened; a wooden or bone stick with a sharp flat edge carved at the other end, used for removing birch bark’. The comparison is phonologically acceptable: PU *u > PS * (in *a-stems) and PU *- > PS *- are regular sound laws. The only irregularity is the sporadic reduction of the stem-final vowel (*a > *q). But this does not pose a problem, as there are several other examples of such reduction in Samoyedic, e.g. PU *kuma- ‘to fall over’ > PS *kqmq- (see Janhunen 1981: 226–230 for a discussion on the reduction phenomenon). The Sami words with the meanings ‘wedge’ and ‘plug’ come especially close semantically to the Samoyedic word family. A wedge is a simple wooden tool, and the semantic derivation ‘block of wood (used as a tool)’ > ‘wedge’ is fully conceivable. The meanings of the Finnish word are appar- ently secondary, but they are not particularly far from the assumed original meaning, either. The development was probably ‘block of wood’ > ‘piece of wood’ > ‘pin, stick’ etc. It is also worth noting that the shape of the instrument used for removing birch bark resembles a wedge. An almost exact parallel for the semantic development of Finnish luta can also be presented: Mordvin salgo7 ‘pin, stick; staff; a tool for removing linden bark (= Finnish luta)’ ~

12 New and Old Samoyed Etymologies

Finnish salko ‘long pole, rod etc.’, Sami cuolgguv ‘lever, crowbar; a bar for pushing nets under the ice’.

1.3. PS *jempq ‘clothes’ [> Nganasan де˙хя ‘clothes’, (der.) 'embitqs{i ‘to dress, get dressed’, Nenets (der.) yempøq- id., Selkup (der.?) t'empq- ‘to dress (tr.), (der.) t'empqt- ‘to get dressed’] (SW: 42–43; Helimski 1983: 210) < PU *lämpi ‘warm(th)’

PS *jempq is cognate with a Finnic-Mordvin *lämpi, which is reflected in Finnish lämmin ‘warm’, lämpö ‘warmth’ and Mordvin l'embe ‘warm, warmth’ (cf. UEW: 685). The comparison is phonetically flawless: PU *l- > PS *j- and PU *ä > PS *e are regular sound laws. The etymology presupposes a semantic shift ‘warmth’ > ‘clothes’ in Samoyedic, which is quite plausible. The most important function of clothing in a (sub) environment is to keep their wearer warm. A similar polysemy is encountered in biktas ‘(noun) something that warms, e.g. a piece of clothing, boiling water; (adjective) warming’ (Schlachter 1958 s.v.). Its North Sami cognate is bivttas ‘piece of clothing’. The word is a nominal derivative of biktit ‘to keep warm (: clothes)’, which in turn is originally a causative of bivvat ‘to keep warm (intr.)’. The meaning ‘clothing’ is also attested in the deverbal noun bivvu ‘winter clothing’.

1.4. PS *kajkq ‘spirit, god; idol’ [> Nganasan koika ‘idol’, Nenets xæxo ‘spirit, god’, Enets kaha, kiho ‘spirit, god; idol; helping spirit of a shaman’, Selkup kaga5 ‘deceased’, Mator (der.?) kajgo ‘idol’] (SW: 51; Helimski 1997: 262; Katzschmann & Pusztay 1978: 90–91; Janhunen 1998: 476) < PU *kod'ka ‘spirit, ghost’

This word is attested in all Samoyedic idioms except Kamass, but no etymology has been proposed for it. There is, however, a certain Finno-Ugric word family that is etymologically connected with it, namely Komi kul' ‘devil, evil spirit, water spirit’, Udmurt kil{ ' ‘plague, typhoid; disease-causing spirit’, Mansi kul' ‘devil, forest spirit’ and Hungarian hagy-máz ‘typhoid, severe fever (with hallucinations)’. These words have previously been compared to Finnish koljo ‘a large man or animal, a mythical giant, disease- causing spirit, devil etc.’ and its correspondents in other (Karelian Koljoi ‘a mythical creature’, Estonian koll ‘ghost, gnome’). UEW (173) considers the comparison between these words and Finnish koljo

13 Ante Aikio certain. The attitude of SSA (s.v.) is more reserved: the comparison is marked with a question mark, and the Hungarian cognate is not mentioned at all. Apparently Finnish koljo is of another origin; an adequate Indo-European loan etymology can be pointed out. It has been compared to a Germanic word family whose meaning is ‘the Underworld’ (cf. LÄGLOS s.v. koljo). A phonetically appropriate loan original is Pre-Germanic *kolja-5 > Proto- Germanic *haljo-5 ‘the Underworld’ (> English hell, German Hölle etc.) A semantic parallel occurs in Finnish mana ‘the Underworld’ ~ South Sami moenesje ‘good or evil spirit, disease-causing spirit; a supernatural foretoken of disease or death’ and Finnish perkele ‘the Devil’ ~ Estonian põrgu ‘hell’ (SSA s.v.). Old Norse Hel ‘goddess of death’ from Proto-Germanic *haljo-5 is also to be noted in this context. The comparison of Finnish koljo to its putative Finno-Ugric cognates can be discarded. It is geographically impossible to assume that a Pre-Germanic loan word would occur in Mansi, and the loan cannot be older than this, i.e. Proto-Indo-European. The word had an initial palatal stop *k-| in PIE, which would have been substituted with *s7 or *c7 in PFU, not *k. The semantics of the Finnish word also suggest a specifically Germanic loan original, because the word does not refer to supernatural phenomena in the other branches of Indo-European. The PIE root is *kel-| ‘to cover, hide’ (IEW: 553–554). As the Finnish word does not seem to be cognate with the Permic and Ugric words, the reconstructed PFU form must also be reassessed. The and Mansi show the palatalized lateral l', which has been derived from the PFU *lj required by Finnic. Since the Finnic word is of another origin, there is no need to reconstruct this cluster. Instead, the PFU spirant *d' can be assumed, which has regularly developed into l' in the aforementioned languages. The advantage of this explanation is that it better accords with the proposed Hungarian cognate hagy-máz. PFU *d' > Hungarian gy is quite regular, whereas the predictable outcome of PFU *lj is rather Hungarian l, cf. fül ‘ear’ < PFU *peljä (UEW: 370). The Hungarian cognate is slightly uncertain, though, as it might also have the same origin as Khanty (Vakh) kant7 '- ‘to be ill’ and Mansi (Lower Konda) xos ‘stomach disease’ (UEW: 860). But the sound correspondences in the latter comparison are not entirely clear. Reconstructing PFU *d' for this word is further supported by Proto- Samoyed *kajkq, which can be derived from PU *kod'ka. PU *d' > PS *j and PU *o > PS *a before an *a in the second syllable are regular developments. An original *a-stem must be assumed, since otherwise PU *k would not have

14 New and Old Samoyed Etymologies been retained in Samoyedic (see 1.13.). This word has also sporadically developed into an *q-stem in Samoyedic (cf. 1.2.). Note that Samoyedic also requires reconstructing PU *o in the first syllable, as PU *a would have changed to PS *ä before a tautosyllabic palatalized consonant (see 2.1., 2.2.) The PU form *kod'ka presupposed by Samoyedic fits the Finno-Ugric cognates quite well, even though they display no evidence for the *k in inlaut. In Permic and Hungarian this is only predictable, as in this kind of context PU *k has regularly changed to zero in these languages, in Hungarian via Proto-Ugric (PUg) *g. However, one would expect PUg *g to be retained in Mansi. But *g has sporadically changed to zero in Mansi, too, a development which seems to concern the PUg cluster *d'g without exception: cf. PFU *tŒd'ka ‘tip’ > PUg *tŒd'ga > Mansi (Pelymka) tal'k (-k is a suffix), PFU *so7 d'ka ‘duck’ > PUg *sad'ga > Mansi (Pelymka) sel85 ' (UEW: 482, 533–534), PU *käd'wä ‘female’ > PUg *käd'gä > Mansi (Pelymka) k5ol' (see 1.6). It is possible that PUg *g > Mansi Ø is actually a regular development in this particular context. The vowel correspondences are quite regular. The Permic vowels do not correspond to each other regularly, but the illabial vowel in Udmurt most probably results from a sporadic development as in many other words (see E. Itkonen 1954: 317). The Proto-Permic form may be reconstructed as *kul3 ', which regularly gives Komi kul' and Udmurt kil{ ' through illabialization. Proto-Permic *u3 goes back to the PFU vowel combination *o–a, which accords perfectly with the Ugric cognates. The Mansi dialectal variants (see Munkácsi – Kálmán 1986: 222) point to Proto-Mansi *kul5 ', and Proto-Mansi *u5 regularly goes back to PUg *a–a and PFU *o–a. The Samoyedic *kajkq word family can be connected phonetically quite regularly with the Permic and Ugric words that have previously been considered cognate with Finnish koljo. Semantically, the etymology is self- evident and requires no special proof. All the cognates refer to the spiritual world. The meanings ‘typhoid’, ‘severe fever’ etc. also belong to this semantic field, since it was a common belief that diseases are caused by evil spirits or other supernatural phenomena.

1.5. PS *karpq ‘fire?’ [> Nenets xarpo ‘aurora borealis’ (in Forest Nenets ‘a large fire, e.g. when a piece of land is set on fire to drive away the mosquitos’), Enets kabo ‘aurora borealis’], der. *karpq-- ‘to blaze’ [> Nenets xarpør-] (Lehtisalo 1956: 167; Tereshchenko 1965: 751; Katzschmann & Pusztay 1978: 79) < PU *korpi- ‘to blaze, scorch etc.’

15 Ante Aikio

Tundra Nenets and Enets share a word for ‘aurora borealis’, the original shape of which can be reconstructed as PS *karpq. The meaning of this word appears to be secondary, as suggested both by the Forest Nenets cognate and by the verbal derivative xarpør- ‘to blaze’. The original meaning of the word is apparently connected with ‘burning, blazing’ in general, even though exact semantic reconstruction is difficult. The analysis is further confirmed by a Finno-Ugric word family which the Samoyedic words are etymologically linked to. A PFU form *korpi- has been reconstructed on the basis of Sami guorbat ‘to be scorched’, Finnish (der.) korventaa ‘to singe, scorch’, Mordvin kurva- ‘to singe’. Selkup kura-5 ‘to scorch (wood)’ has previously been suggested as an uncertain Samoyedic cognate (UEW: 186), but the comparison is not phonologically satisfying. But the correspondence PFU *korpi- ~ PS *karpq is predictable, as the shift PU *o > PS *a regularly took place before consonant clusters. The compari- son is also semantically straightforward, as the meanings attested in Forest Nenets and in the Nenets derivative xarpør- are very close to the Finno-Ugric cognates.

1.6. PS *kejm(e) ‘female, mare’ [> Nenets (der.) syibyako, Enets se5 ‘female’, Selkup sümJ , Kamass sv ujmu, Mator kım‰ ‘mare’] ~ *kejwe [> Mator kejbe ‘mare’] (SW: 66–67; Helimski 1986: 127–128; 1997: 272) < PU *käd'wä ‘female’

This Samoyedic word family has with hesitation been compared to kiõvv‹ ‘reindeer cow’ and Livonian kev5 ‘mare’ (UEW: 152). The Sami and Livonian words are indeed cognate, but as regards Samoyedic, such a comparison is unacceptable. The protoform presupposed by the Livonian and Sami words is Finnic-Samic *keewi, and the PS consonant cluster *jm or *jw cannot correspond to the Sami and Livonian single *w. Furthermore, a Finnic-Samic long vowel regularly corresponds to a vowel sequence in Samoyedic. Another, more convincing etymology has recently been proposed for the Samoyedic words. Abondolo (1996: 68) suggests that they might be cognate with a Finno-Ugric word that means ‘female’ that has certain reflexes in Samic and -Ugric: North Sami (obsolete) gádfi∞ ‘female ermine’, Skolt Sami kä/ptt ‘female (of a small fur animal)’, Khanty (Vasjugan) köjqn, (Tremjugan) kœo3 jqg, Mansi (Pelymka) k5ol' ‘female’. Hungarian hölgy ‘dame;

16 New and Old Samoyed Etymologies

(archaic) bride, beloved; ermine’ might also be of the same origin. However, as regards Samoyedic, Abondolo considers the etymology uncertain because of the irregular consonant correspondence. The PFU form has been reconstructed as *käd'wä (Sammallahti 1988: 545) or back vocalic *kad'wa (UEW: 116). The former reconstruction is better argued, because the Sami and Ob-Ugric reflexes clearly point to an original PFU *ä. The only reason for reconstructing *a is Hungarian hölgy, since PFU *k gives Hungarian in anlaut only before PFU back vowels. But the putative Hungarian cognate is phonetically unclear. It presupposes the developments PFU *ä (or *a) > Hung. ö and PFU *d' > ?*l' > Hung. lgy, which are quite irregular. The PFU reconstruction cannot be based on Hungarian because of phonetic obscurities. It is not even certain that the Hungarian word belongs in this connection at all; Sammallahti (1988: 545) and Collinder (1977: 98) consider it uncertain. In addition, the other cognates display some phonetic irregularities, too, no doubt due to the affective nature of the word. The Vasjugan of Khanty shows an irregular (pro g) as a reflex of PFU *w, and in Mansi *w has strangely changed to zero (one would expect g here, too, but cf. 1.4). Another Samoyed cognate for PFU *käd'wä has also been recently suggested, namely PS *ki ‘marten’ (Helimski 1991: 263). However, this suggestion must be rejected, as it presupposes a wholly irregular develop- ment from PU *d'w to PS Ø. Note that the vowel correspondence is not satisfactory, either, as PU *ä regularly shifted to PS *e, not *i. Instead, Abondolo’s equation of PS *kejme ‘female, mare’ with PFU *käd'wä is quite convincing. The developments PU *ä > PS *e and PU *d' > PS *j are regular, the only problem being the correspondence PFU *w ~ PS *. But as both are labial consonants, the comparison can be accepted, since the word in question is affective and thus susceptible to irregular sound changes. Further- more, another possible etymology which shows instability between postconsonantal *m and *w has also been pointed out: PU *pilmi- ‘to darken’ ~ PFU *pilwi ‘cloud’ (Kulonen 1995: 90–91). The etymology treated here is still further reinforced by the Mator cognate kejbe, which can be traced to PS *kejwe rather than *kejme, cf. PS *qjwa ‘head’ > Mator ajba, but PS *ejme ‘needle’ > Mator ime (Helimski 1997: 201, 253). Mator kejbe thus reflects PU *käd'wä ‘female’ quite regularly. It is possible that kejbe has preserved the original *w, and the *m in Mator kım‰ and the other Samoyed languages is secondary.

17 Ante Aikio

1.7. PS *ker- ‘to dress, put on (clothes, a piece of clothing)’ [> Selkup ser-v , Kamass sv er-5 , Mator ker-] ~ *kerq- [> Nganasan ser7 q-, Nenets syerø-, Enets seri-] (SW: 68; Helimski 1997: 275) < PU *käri- ‘to wrap, tie, thread’

The common Samoyedic word for ‘dressing’ has two variants, South Samoyedic *ker- and North Samoyedic *kerq-. The first variant probably represents the original form, and the latter contains a binding vowel that has been analogically added to the stem. No etymology for the word family has been proposed. Proto-Samoyed *ker- derives regularly from PU *käri-, and compares thus perfectly to the reconstructed PFU verb *käri- ‘to wrap, tie, thread’. This has been retained as an independent verb only in Mari: kerä- ‘to thread (beads, a needle)’. Other languages show only derivatives of this root: Mordvin (Erzya) kerks7 ‘bunch; string (of pearls)’, kerksa-7 ‘to thread’, (Moksha) kärks ‘wreath’, kärksa- ‘to wind’, Komi kert-8 , Udmurt kert- ‘to tie’, Komi ger8 ed8 , Udmurt gerd ‘knot’ (with irregular g-), Khanty (Vasjugan) kerq, Mansi (Sosva) k Ø ar5 qk ‘bunch’. UEW (139–140) proposes other Ob- Ugric cognates, Khanty (Vasjugan) kir- ‘to put shoes on’, Mansi (Sosva) ker-5 ‘to weave’, but these words do not show regular vowel correspondence with the Finno-Permic word family. The Ob-Ugric words for ‘bunch’ have traditionally been considered cognate with Finnish kerä and Mordvin kire7 ‘ball (of yarn etc.)’ (UEW: 147). However, because their vowel goes regularly back to PFU *ä, not *e, they are more likely to belong to the etymology treated here (cf. Sammallahti 1988: 543).3 Komi kert-8 and Udmurt kert- have also been considered cognate with korttâd∞ ‘to bind’ (E. Itkonen 1954: 178, Sammallahti 1988: 552), but the vowel corre- spondence is not regular. The etymology suggested by UEW follows regular sound laws and is thus better. Inari Sami korttâd∞ is an Indo-European loan word (Koivulehto 1988: 41–42). The etymology proposed here presupposes a shift of meaning from ‘wrap, bind’ or ‘thread’ to ‘dress’. Both are quite natural. As regards ‘threading’, cf. Finnish pujottaa ‘to thread, slip’, pujottaa (vaatteet päälle) ‘to put (clothes on)’ and pukea ‘to dress, put on; to thread, slip’. The relation ‘wrap’ ~ ‘put on, wear clothes’ is found in Old Norse verja ‘to cover, wrap; (rarely) to dress’ ~ English wear (Buck 1965: 393). Many pieces of clothing are “wrapped” on, e.g. shoelaces and a cloak. Some reflexes of PU *käri- also show the meaning ‘to tie’, which can be compared with German anziehen ‘to

18 New and Old Samoyed Etymologies dress’ ← ziehen ‘to pull’ ~ English tie (Kluge 1995 s.v. ziehen, Klein 1966– 1967 s.v. tie). Finnic and Samic cognates can balso be added to this etymology. Finnish kääriä ‘to wrap’ and Ter Sami kı‰ /rred ‘to wrap, wind up’ have been considered cognate (SSA s.v.), but the further origin of these words is unknown. However, they can be derived from PU *käri-, even though the comparison poses slight phonetic difficulties. As regards Finnish, it requires an irregular lengthening of the first syllable low vowel, which is, though, a common phenomenon before the consonant r: cf. e.g. Finnish käärme ~ kärme ‘snake’. The stem-final i is unusual, but it might be a derivative suffix (< *-j-). The Ter Sami word is not entirely regular either, as its stem vowel does not reflect an original *i. It is possible to derive the Finnish and Sami words from PU *käri-, but this presumption requires slightly irregular vowel development in both Finnic and Samic. As regards Samic, the comparison is strengthened by South Sami gïeres ‘noose in a lasso (for catching reindeer); a piece of antler through which the lasso runs’ and gïerestidh ‘to catch with a lasso’ (Bergsland–Magga 1993 s.v.). The same words are found in (Grundström 1946–1954 s.v. kieras, kierastit) and in Ume Sami there is a derivative giarastahteet ‘to catch with a lasso; to tie, bind’ (Schlachter 1958 s.v.). All these words can be explained as derivatives of a Proto-Sami verb *kiere-{ . UEW (148) connects the Samic word family with Finnish keri ‘frame etc.’, Hungarian körül ‘around’ etc. (< PFU *keri), but this etymology is not acceptable, because Samic *ie goes back to PFU *ä, not *e. Instead, the sound correspondence with PFU *käri- is regular and the match is also semantically convincing. The meaning ‘to tie’ encountered in Ume Sami is also found in Permic, and a shift from ‘to tie, wrap’ to ‘to catch with a lasso’ is also quite natural: cf. English lace and lasso, which both ultimately derive from Latin laqueus ‘noose, snare’, the former via French, the latter via Spanish (Klein 1966–1967 s.v.). The North Sami verb gierrat ‘to lace up shoes’ is especially noteworthy in this context, because it, too, goes regularly back to Proto-Sami *kiere-8 . It is encountered with this meaning only as a relict in one phrase in a story about cáhkalakkisv , a small, human-like creature that is said to live in springs or holes in the ground. According to the story, this creature could be caught by placing a shoe and laces outside his home. When cáhkalakkisv found them, he put both his feet in the same shoe and started lacing it up while saying “gieran, gieran, de ain cuolbmadanv ” (‘I lace up, I lace up, then I tie again’). (For one version of the story, see E. Itkonen 1960 : 7.) This word has been

19 Ante Aikio considered a loan from Finnish kääriä (o.c. 112), but the vowel correspond- ence suggests otherwise: borrowing from Finnish would have resulted in Sami **gearrat. The word gierrat has been otherwise preserved in North Sami, too, though its meaning has changed. Today, it is known with the meaning ‘to charm, enchant’. This word has also been considered a derivative of gieris ‘dear, loved’ (Nielsen 1979 s.v. gierrât, gieres), which is of Scandinavian origin (Qvigstad 1893: 169), but the comparison to the verbs meaning ‘to wrap, tie’, etc. is more credible. There are numerous semantic parallels, cf. Finnish kiehtoa ‘to charm; to wind’ and such idioms as kietoa pauloihinsa ‘to ensnare in one’s toils’, literally ‘to lace up with one’s shoelaces’.

1.8. PS *ketq- ‘to wrap up’ [> Nenets syedø- ‘to wrap up a child, lay a child in its cradle’, (der.) syed oq ‘a blanket made of reindeer skin, with which a child is tucked up in a cradle’, Mator kedq- ‘to wrap up’] (Lehtisalo 1956: 441; Helimski 1997: 271) < PU *kätki- ‘to wrap up (a child?), to lay a child in its cradle’

The PS verb *ketq- has certain reflexes only in Nenets and Mator. The Nenets cognate syedø- exhibits a specialized meaning ‘to wrap up a child, lay a child in its cradle’, whereas the meaning of the Mator cognate kedq- has been recorded as ‘wrapping up’ in general. No further etymology has been suggested for the Samoyedic words, but a connection between them and a Finnic-Samic etymon can be established: Proto-Finno-Samic *kätki- > Sami gietka- ‘to wrap up a child, lay a child in its cradle’, (der.) gietkka : gietkam- ‘cradle’, Finnish kätke- ‘to hide (tr.)’, (der.) kätkyt ‘cradle’ (SSA s.v. kätkeä, kätkyt). The sound correspondence between the items is regular, as long as one assumes that Nenets sy- and Mator k- go back to PS *k- in this word (theoretically also PS *setq- would be possible, PS *s- > Mator k- being a regular development before non-low front vowels [Helimski 1997: 79]). PU *ä > PS *e and PU *tk > PS *t are regular sound changes. The etymology is also semantically convincing, as the meanings of the Nenets and Sami cognates are identical. In Finnic the basic verb has developed a more abstract meaning ‘to hide’, but the derivative kätkyt ‘cradle’ reveals that this meaning must be secondary. It has also been suggested that the Finnic verb might rather be cognate with Mordvin kekse-v ‘to hide’ and thus of different origin than the derivative kätkyt (UEW: 649). However, this suggestion is implausible, because a sound correspondence Finnish tk ~ Mordvin ksv would be irregular.

20 New and Old Samoyed Etymologies

1.9. PS *kqnse- ‘to cool down’ [> Nenets (derivatives) xønco ‘cool’, xøncyim- ‘to cool down’, Selkup kazi-v , Kamass kqnzq-] (SW: 52) < PU *kän7 sä-7 ‘to freeze’

The PS verb *kqnse- ‘to cool down’ can be compared to an etymon that has reflexes in Mari and Permic: PFU *kän7 sä-7 ‘to freeze’ > Mari kize-v ‘to freeze; to catch a cold’, Komi (der.) ke8 7z{id ‘cold, frost’, ke8 7zal- ‘to cool down, get cold’, Udmurt (der.) keDvzeg ‘fever’ (UEW: 648). The original form of these words has been reconstructed as *käc7 •- (o.c.) or *käs7 •- (Sammallahti 1988: 552). The reconstruction of PFU *s7 is slightly problematic, as it predictably gives Proto-Permic *z7 rather than *7z. The *c7 is a better possibility, but Permic *7z can also be derived from an earlier cluster *n7 s7 . The denasalization of this cluster in Mari is also predictable, cf. PU *kun7 si-7 ‘to urinate’ > Mari käqvza- (UEW: 210). The proto-form of the Mari and Permic words can thus be reconstructed as *kän7 sä-7 , which allows establishing a connection with PS *kqnse-. The semantics of the etymology are self-evident, and the consonant correspond- ence is quite regular. However, as regards the first syllable vowel, the comparison is slightly irregular. PFU *ä predicts PS *e, not *q. The PS reduced vowel *q normally presupposes PU *u, but on the other hand, the second syllable *e suggests original front vocalism. However, despite the slight irregularity, the etymology can be judged reasonably secure. A Uralic comparison with a slightly irregular vowel correspondence can be accepted, when the consonant correspondences are regular and the etymology is semantically obvious (cf. the discussion and other examples in Janhunen 1981: 252–264). It seems probable that the Samoyedic reduced vowel is secondary, so the PU form can be reconstructed as *kän7 sä-7 on the basis of Mari and Permic. A word further worth noting in connection with PS *kqnse- is PS *kqnt•- ‘to freeze’ (SW: 53), which has a remarkable resemblance in both form and meaning to the words discussed here. However, it seems that these items cannot easily be linked with each other. An assumption of parallel derivatives of a Pre-Samoyedic root *kqn(•)- would be problematic, as there is no such verbal suffix as PS *-se-. As the words appear to be unrelated, one can suggest that the irregular vowel development *e > *q in PS *kqnse- may have resulted from the influence of the semantically close verb *kqnt•-.

21 Ante Aikio

1.10. PS *mqjq- ‘to become happy’ [> Nenets møyø-] < PU *muja- ‘to become happy; happiness, smile’

The Nenets verb møyø- has no cognates elsewhere in Samoyedic, but it would go regularly back to PS *mqjq-. This word can be compared to a Samic word family, whose underived root is reflected in North Sami modji, Inari Sami moje ‘smile’. Various derivatives of this word are attested in most , e.g. North Sami moddját ‘to smile’, South Sami mujjies ‘smiling’, Skolt Sami mojjâm ‘smile’. The Proto-Sami form must be reconstructed as *moje5 , which reflects an earlier form *muja-. The Samic word family thus fits perfectly as a cognate of the Nenets verb møyø-. PS *mqjq- can be derived from PU *muja- via the regular shift *u > *q, and the irregular but common reduction of the stem vowel (see 1.2.). The lack of cognates elsewhere in Samoyedic and the rest of Uralic gives the comparison a somewhat hypothetical nature, but the etymology is both semantically and phonologically straightforward. Moreover, one must keep in mind that the words for ‘happiness’, ‘being happy’ do not belong to the most stable part of the lexicon, but on the contrary are quite frequently replaced by innovative expressions. Therefore, it is not entirely unexpected that the assumed PU word *muja has been so poorly preserved in the Uralic languages. A new Proto-Samoyed word comparison in the same semantic field can also be established. Enets edde ‘joy’ (Katzschmann & Pusztay 1978: 72) can be connected to Selkup antal-5 ‘to enjoy’ (Erdélyi 1970: 21) and their proto- form can be reconstructed as PS *änt•. However, the further origin of this word is unclear.

1.11. PS *mqja ‘earth, land, terrain’ [> Nganasan mou ‘earth’, Selkup maciv ‘forest, tundra’, Kamass mija ‘mountain’, Mator bijä ‘mountain’] (SW: 85; Helimski 1997: 218) < PU *mud'a ‘land, earth, soil’

The Nganasan word mou ‘earth’ has, independently of its cognates in South Samoyedic, often been cited as the cognate of Finnish maa, Komi mu ‘earth, ground’, etc. in etymological standard literature (SSA s.v. maa; UEW: 263). On the other hand, the Selkup cognate maciv has been cautiously compared to Inari Sami myejdi∞ ‘forest; hunt of wild reindeer in the winter’ (UEW: 263). Neither of these comparisons has, however, been accepted by Janhunen

22 New and Old Samoyed Etymologies

(1981) and Sammallahti (1988: 546), the obvious reason being the lack of regular sound correspondence. The PFU word for ‘earth’ must be recon- structed as *mixi{ , which predicts a PS cognate of the form **me8 , not *mqja. The comparison between the Selkup and the Inari Sami words is phonologically no more satisfying, as an unusual PU consonant cluster *jd' would have to be reconstructed to account for the Sami representation, and the correspondence of the first syllable vowel would remain irregular. Recently, a more plausible etymology for the Samoyedic word family has been suggested. Abondolo (1996: 28–29) has proposed a connection with a group of words in the western Uralic languages, whose reflexes mean ‘mud’ on one hand (Sami mod∞ di∞ , Finnish muta) and ‘earth, land’ on the other (Mordvin moda). Abondolo’s etymology is convincing both phonologically and semantically. The Proto-Uralic form can be reconstructed as *mud'a, and the Samoyedic form derives from this via the regular sound changes PU *u > PS *q (in *a-stems) and PU *d' > PS *j. The original meaning may be deduced as ‘land, earth’, which has been preserved in Nganasan and Mordvin; both the Finnic-Samic meaning ‘mud’ and the meanings ‘moun- tain’, ‘tundra’ etc. in South Samoyedic appear to be secondary. An alternative explanation of origin has also been proposed for the Finnic, Samic and Mordvin words: It has been suggested that they are borrowings from Proto-Germanic *mud(∞ d)a-∞ ‘mud’ (> English mud etc.) (LÄGLOS s.v. muta). However, it is worth noting that the distribution of the assumed loan original is confined to the Germanic languages, as no plausible cognates elsewhere from Indo-European have been attested to (cf. Kluge 1995 s.v. Moder, Klein 1966–1967 s.v. mud, IEW: 276–277). In addition to this, the Germanic word shows irregular variation between a single and geminate spirant, which points to loan origin (Petri Kallio: personal communication). It seems that the Germanic words are best explained as an early borrowing from Uralic, more precisely the predecessor of Samic and Finnic, as they show the same semantic innovation ‘earth’ > ‘soil’ > ‘mud’. Old Uralic loans in the Germanic languages are not very commonplace, but some cases have been pointed out (see e.g. Kallio 2001).

1.12. PS *nimse7 ‘breast, milk’ [> Selkup nips7 q, neps7 q ‘breast, milk’] (SW: 110; Helimski 1983: 129) < PU *nim-7 sä7 ‘breast, teat’ ← *nimi-7 ‘to suck’

The reflexes of PS *nimse7 are apparently confined to Selkup. The word is a

23 Ante Aikio derivative consisting of the verb root *nim-7 ‘to suck’ and the nominal suffix *-sa / *-se. There are several other Proto-Samoyed deverbal nouns derived with this suffix that look fairly old, e.g. PS *qmsa ‘meat, food’ ← *qm- ‘to eat’, *kaqsa ‘man’ (< ‘mortal’) ← *kaq- ‘to die’ (SW: 15, 61). The verb root *nim-7 has widespread reflexes in Samoyedic, and another derivative for ‘breast’ is also based on the same root: PS *nim-me7 (SW: 111). Even though the Selkup item is quite transparently explicable as a derivative, the derivation seems to have taken place already in Proto-Uralic. There is a Finnic-Samic word family which has a remarkable resemblance to the Selkup word in both form and meaning: Finnish nisä ‘teat’ ~ North Sami njizv ziv id. < Finnic-Samic *ni7 n7 sä7 . The irregular -i- in the Sami reflex (instead of regular **njazv ziv ) has developed due to the influence of the surrounding palatalized consonants. The Proto-Uralic form of the verb has traditionally been reconstructed as *imi- on the basis of Finnish imeä, Old Hungarian em- ‘to suck’ etc., and the initial nasal *n-7 in Sami (North Sami njammat), Komi (nimav-7 ) and Samoyedic has been considered a late expressive addition. However, it appears that a Proto-Uralic status has to be assigned to both of the variants (cf. Janhunen 1981: 256, who suggests this possibility). This is strongly supported by Samoyedic, where the variants show distinct reflexes, as Helimski (1986: 124–125) has pointed out: PS *nim-7 ‘to suck’ < PU *nimi-7 , PS *qm- id. < PU *imi-. The development PU *i > PS *q in the latter item seems to be regular in initial position. A previously known example of this sound law is PS *qptq ‘smell’ < PU *ipsi (cf. Janhunen 1981: 253–254, who regards the develop- ment as irregular). A third case recently etymologized by Helimski (1999b) is PU *itä- ‘to appear, come in sight’ > PS *qte- ~ *qtq- (SW: 16); cf. Finnish itä- ‘to sprout’, itä- ‘east’ and Khanty et-5 ‘to appear, come in sight’, etc. (cf. UEW: 85). A cognate in Mansi can also be added to this etymology: (Pelymka etc.) ätqr ‘clear (of weather, sky)’ (Munkácsi – Kálmán 1986). The Mansi form is a derivative and apparently formally identical to Nenets ngødyer ‘clearing (in forest)’ (Lehtisalo 1956: 13). A slight problem is involved in the sound correspondence PS *ms ~ Finnic-Samic *n7 s7 , as no certain parallels for this can at present be pointed out. Nevertheless, it seems fairly obvious that an assimilation *ms7 > *n7 s7 has taken place in Finnic-Samic *ni7 n7 sä7 . Due to the phonological and semantic close- ness of the assumed cognates, the etymology looks plausible despite the ad hoc assumption of a sound change *ms7 > *n7 s7 . Moreover, the assimilation does not have to be a regular development, as it may also have resulted from

24 New and Old Samoyed Etymologies the influence of the phonetically and semantically close PU word *sü7 n7 sä7 ‘breast’ (> PS *sünsq, North Sami civ zv ziv etc.; Sammallahti 1988: 540). According to SSA, Finnish nisä and its Sami cognate are probably of onomatopoetic origin. It is true that the words in question have an expressive character, and the same concerns the reflexes of the basic root *nimi-7 ~ *imi-. However, expressive words can also be ancient, and expressiveness does not necessarily imply onomatopoetic origin. There is thus no obstacle to explain- ing the words treated here as quite regular derivatives of PU *nimi-7 ‘to suck’. A somewhat similar case is the PFU word *nälmä7 ‘tongue, mouth’, which has been explained as an obscured deverbal noun based on PFU *neeli-7 (< *näxli-7 ) ‘to swallow’ (Janhunen 1992).

1.13. PS *pur ‘smoke; blizzard’ (derivatives) [> Nenets pur oq ‘smoke fire against mosquitos; haze’, Selkup purqi{ ‘smoke’, purqat5 ‘blizzard’, purqal'ti-{ ‘to get covered, blown over (by snow, sand etc.)’] (SW: 131; Erdélyi 1970: 196; SSA s.v. purku) < PU *purki ‘smoke, spray, blizzard’, verb. *purki- ‘to smoke, spray, whirl’

Proto-Samoyed *pur is attested only in derivatives in Nenets and Selkup. Kamass ber ‘smoke; dust’ is also doubtingly compared with these words in SW, but it can hardly be of the same origin, because Kamass initial goes back to PS *w, and the vowel correspondence is not regular either. No etymology has been proposed for the Nenets and Selkup words with the meaning ‘smoke’. They can, however, be linked to certain parallel deriva- tives in Selkup and further with PFU *purki ‘smoke; blizzard; to whirl’, with widely attested reflexes in Finno-Ugric. The PFU reconstruction *purki is reflected regularly in at least Sami borga ‘snowstorm, blizzard; (in compounds) spray (of water, snow, sand, etc.)’, borgat ‘to storm with snow’, Finnish purku ‘snowstorm; snowdrift’, Mari purga- ‘to storm with snow’, (der.) purgäqvz ‘snowdrift’, Komi pira{ ‘snowstorm’, Khanty (Vasjugan) pork3 i{ ‘smoke’ and Mansi (Tremjugan) porke{5 ‘snowstorm’ (SSA s.v. purku; cf. UEW: 406, where the Ob-Ugric cognates are unnecessarily regarded uncertain). There also appears to be a variant with an *a-stem that is reflected in Sami borgit ‘to smoke’ (and probably also borgi ‘shedding hair of reindeer’), Finnish purkaa ‘to drift (snow) etc.’ (and probably also purkaa ‘to take apart’), Mordvin purga- ‘to splash’ (on the semantics, see Kulonen 1995: 91). Furthermore, there are

25 Ante Aikio proposed cognates that must be considered uncertain for either phonological or semantic reasons: Mordvin (Moksha) porf ‘snowstorm’, Udmurt pura- ‘to glow (coals)’, Hungarian forr ‘to boil’. Some of the uncertain cognates might be later, descriptive formations, but this fact does not, of course, hinder reconstructing PFU *purki. Both UEW and SSA cite Nenets paranteyo ‘snowdrift’ and Selkup purqat5 ‘snowstorm’ as uncertain Samoyed cognates. These words do not corre- spond regularly to each other, and the proposed Nenets cognate is therefore rejectable. However, Selkup purqat5 corresponds quite well to the word purqi{ ‘smoke’ and Nenets pur oq, and is thus to be taken as a separate derivative of the same PS root *pur-. This word is not found in Erdélyi 1970, but a parallel derivative purqal'ti-{ ‘to get covered, blown over by snow, sand, etc.’ is included. Erdélyi lists it separately from purqi{ , but it is of the same etymology. PS *pur corresponds quite regularly to PFU *purki. According to the traditional explanation (e.g. Sammallahti 1979: 42), the loss of the stop *k in Samoyedic before a PU *i in the second syllable is regular. Janhunen (1981) and Sammallahti (1988) have abandoned this view due to the lack of conclusive etymologies. However, there are quite a few examples: PU *n7 irki{ ‘cartilage’ > PS *n7 er8 , PU *selki-v ‘to fly’ > PS *tij-, PU *ulki ‘pole’ > PS *uj, PU *kurki ‘crane’ > PS *kqrö (and Finnish kurki; Sami and Mordvin reflect another variant *karki), PU *kulki- ‘to go, flow’ > PS *kqj-, and probably also PU *koki- ‘to see, find’ > PS *ko-, even though this etymology is not entirely regular, as the assumed Finnic-Samic cognates do not show the expected development PU *o > PFU *u /_Ci (cf. Finnish kokea ‘to experience; to examine nets’, Inari Sami kuohâd∞ ‘to examine nets’ < Finnic-Samic *koki-; note also the Indo-European loan etymology suggested by Koivulehto [1991: 44–47]). The correspondence PFU *k ~ PS Ø was discovered anew by the Indo-Europeanist Schrijver (1997: 298) on the basis of the etymologies *n7 irki{ , *kulki-, *kurki and *selki-v . Schrijver tentatively reconstructed PU *lx and *rx for these items, but it is easier to assume a development PU *k > PS Ø before a high vowel, as suggested by Sammallahti (personal communication). This etymology is also semantically satisfactory. The meaning ‘smoke’ is also attested in the Khanty cognate and in Sami borgit, and Selkup also shows the meaning ‘snowstorm’ that is dominant in Finno-Ugric. The connecting factor between ‘smoke’ and ‘snowstorm, blizzard’ is ‘pouring, whirling in the air’: cf. Finnish suitsuta ‘to smoke, pour; to whirl about (e.g. snow)’ (SSA s.v. suitsu) and Sami soica ‘thick driving snow; thick smoke’

26 New and Old Samoyed Etymologies

(< Finnish). The somewhat descriptive nature of the words does not hinder the etymology, as all the cognates point to a uniform proto-form as regards the consonants and the first syllable vowel. Descriptiveness might also account for the irregular variation of the second syllable vowel (*purki ~ *purka), which seems to have led to semantic differentiation in Samic and Finnic. There are also many other descriptive words with a similar phono- logical shape *pVr(C)- which are obviously of much later origin, e.g. Sami bohrra ‘cloud of dust or ash’, borsutv ‘to spurt, splash, foam’, Finnish pyry ‘blizzard’ and purskahtaa ‘to spurt out’.

1.14. PS *sajq- ‘(to wage) war’ (derivatives) [> Nenets sayuo ~ sayuwo ‘soldier, warrior, enemy’, sayuodør- ‘to wage war’, sayonor- id., sayuodørma ‘war’, etc.] (Lehtisalo 1956: 402; Tereshchenko 1965: 42) < PU *so7 d'a ‘war’

The Nenets word family consists of parallel derivatives, whose original Pre- Nenets root can be reconstructed as *sajq-. The original meaning can be deduced as ‘war’, ‘to wage war’ etc. The word apparently has no cognates elsewhere in Samoyedic. However, it can be connected to the following Finno-Ugric word family: Finnish sota ‘war’, Mordvin sudo-7 ‘to curse’, Mari suv de- id. (UEW: 777). The PU form can be reconstructed as *so7 d'a (in the absence of a Samoyed cognate also the reconstructions *so7 da and *sota7 would be possible). The comparison is phonologically regular: PU *s7 > PS *s, PU *o > PS *a (in original a-stems) and PU *d' > PS *j are regular sound laws. The reduction of the second syllable vowel (*a > *q) is irregular, but there are numerous parallel cases (see 1.2.). The etymology is also semantically flawless: the meaning ‘war’ is attested both in Finnic and in Nenets. In Mordvin and Mari a semantic shift ‘to wage war’ > ‘to curse’ has to be assumed. No problem is involved in reconstructing a PU word for ‘war’, as warfare is an ancient human activity, and also hunter-gatherer cultures all over the world have waged war (see e.g. Keeley 1996).

1.15. PS *salq- ‘to flash, lighten’ [> Nenets (der.) salowø- ‘to flash, shimmer’, Mator salq- ‘to lighten’, (der.) salqmqr- id.] (SW: 135; Helimski 1997: 334) < PU *sala-7 ‘to flash, lighten’

A Finno-Ugric verb meaning ‘to flash, lighten’ is reconstructed on the basis

27 Ante Aikio of Ob-Ugric and Finnic: Khanty (Vakh-Vasjugan) sal-3 ‘to flash, lighten’, Mansi (Pelymka) sel-{5 ‘to lighten’, and Finnish (nominal derivative) salama ‘lightning’ (UEW: 459). No cognates from other branches of Uralic have been suggested. One can, however, connect here Mator salq- ‘to lighten’ and its derivative counterparts in Nenets (e.g. salowø- ‘to flash’). The word is not attested in other Samoyedic languages, but on the basis of Nenets and Mator the Proto-Samoyed form may be reconstructed as *salq-. Semantically the Uralic etymology is self-evident, but the vowel corre- spondences require closer scrutiny. The Samoyedic second syllable *q has arisen through the irregular, but common reduction of the stem vowel (see 1.2.). However, a more substantial problem concerns the correspondence between the Samoyedic and Finno-Ugric first syllable vowels. The Ob-Ugric reflexes are Proto-Khanty *sal- and Proto-Mansi *sıl-{‰ (< Proto-Ob-Ugric *sıl-{‰ ). The Mansi form points to PFU *i{ rather than *a; the regular reflex of PFU *a is Proto-Ob-Ugric *u5 (> Proto-Khanty *a5 , Proto-Mansi *u5 ). The PFU form *s7 ila-{ presupposed by Mansi does not accord with the Samoyedic representation, as PU *i{ would be preserved as such in Samoyedic: cf. e.g. PU *ila{ ‘under, below’ > PS *il{ q. In this particular case it can be shown that the vowel *i{ is, in all likelihood, a later Ob-Ugric development. Even though Ob-Ugric *ı{‰ generally goes back to PFU *i{ , in certain cases the reflex Proto-(Finno-)Ugric *a has been irregularly illabialized in Ob-Ugric (or perhaps even later in Proto-Mansi; the evidence from Khanty is ambiguous). One such case, the Mansi reflex of PU *cav na- ‘to beat, rub’, was already referred to above (see 1.1.). A phonologically more exact parallel to the etymology treated here is Mansi (Pelymka) set{5 ‘hundred’ < Proto-Mansi *sv ıt{‰ < PFU *sata.7 The original quality of the PFU vowel has been proved to be *a both by the cognate in Komi (so7 ‘100’ < PFU *sata7 instead of predictable **su7 < PFU **s7 ita{ ) and the Proto-Aryan loan original (*satam-7 ‘100’ < Proto-Indo-European *k7 ømtom-). Curiously, there seem to be even more examples of the illabialization of *a after PFU initial *7s-. A fairly certain case is PFU *so7 dka / *so7 d'ka ‘pochard’ > Proto-Ugric *sad'ga > Proto-Mansi *vs{‰ıl' (cf. Sammallahti 1988: 549). The parallels presented here also have implications for a PFU etymon for ‘pole, rod’, which can now be reconstructed as *7salka instead of *s7 ilka{ (cf. o.c.) and thus derived from Proto-Indo-European *g7 halgho- ‘pole, rod’, as Koivulehto (1983: 113) has suggested. Without the phonological parallels, the represen- tation in Mansi would have to be regarded as an argument against the loan etymology.

28 New and Old Samoyed Etymologies

It is worth noting that even though Proto-Mansi *ı{‰ generally seems to be an irregular reflex of Proto-(Finno-)Ugric *a, there appear to be no clear counterexamples to this shift after initial *7s-. It is perhaps possible that the vowel development in this particular context is actually regular, even though there seems to be no clear phonetic motivation for such a shift. Whatever the case, it can be asserted that the illabial vowel in Mansi sel-{5 ‘to lighten’ is secondary, and the word is thus cognate with the Samoyedic word family.

1.16. PS *tajwa- ‘to arrive, reach; to bring’ [> Nganasan tojbu- ‘to take, transport, deliver’, Nenets tæwø- ‘to arrive; to reach, catch up with; to bring’, Enets toe- ‘to reach’, ? Selkup (der.?) turi-{ ‘to end’ (? < *tajwa-ra-), Kamass tu- ‘to arrive, to reach (the destination); to get on, make progress; to appear’, Mator tajbq- (an with a resultative meaning)] (SW: 146; Helimski 1997: 349) < PU *tolwa- ‘? to bring / take’

This word family is represented in all Samoyed languages, except perhaps for Selkup, whose correspondent is somewhat uncertain. The PS form of the verb has been reconstructed as *tajwa-, with both a transitive meaning (‘to bring’) and an intransitive meaning (‘to arrive’). This word family has often been considered possibly cognate with certain Finno-Ugric words, namely Finnish saa- ‘to get’, Mordvin saje- ‘to take’, Mari soa-v ‘to come, arrive’, Komi su- ‘to arrive, reach; to find, meet’ (UEW: 429–430). However, the comparison is phonologically impossible, because the Finno-Ugric words derive from PFU *sixi-{ , which shows no regular sound correspondence with PS *tajwa- at all, except for the initial consonant. The proposed etymology can thus be rejected. PS *tajwa- can, however, be shown to derive from Proto-Uralic. At the opposite periphery of the Uralic family, in Samic, there is a verb which suits perfectly as a cognate for the Samoyedic word family. This word can be reconstructed as Proto-Sami *toalvo-5 ‘to take (somewhere), to lead’, cf. North Sami doalvut. The word is attested in Western Samic and Inari Sami, but in Skolt, Kildin and Ter Sami reflexes are missing. In South Sami the word exhibits a different meaning: dåalvodh ‘to follow, pursue, chase’. Semantically, the comparison with Samic is quite self-evident, despite the fact that the attested transitive meanings in Samoyedic and Samic are, in a sense, opposite: ‘to bring’ vs. ‘to take’. However, it is well-known that such

29 Ante Aikio concepts as ‘bringing’ and ‘taking’, ‘coming’ and ‘going’ etc. are very close to each other, the difference between them being one of perspective (move- ment towards vs. away from the point of reference). Furthermore, as the Nganasan cognate has the meaning ‘to take, transport’, the comparison with Samic poses no problem. Note also the Nenets meaning ‘to reach, catch up with’, which is quite near the South Sami meaning ‘to follow, pursue’. As regards phonology, the assumed developments PU *tolwa- > PS *tajwa-, Proto-Sami *toalvo-5 follow regular sound laws. The second syllable labial vowel in Samic can be analyzed as a suffix. The only certain cognates seem to be found in Samic and Samoyedic. However, this does not impede an otherwise regular Uralic etymology. There is even another, generally accepted etymology with cognates in Samic and Samoyedic only, PU *koska ‘aunt, grandmother’ (UEW: 189). Besides, somewhat uncertain correspondents can be pointed out in two other branches: Mari tola- ‘to come’ and Hungarian talál ‘to find, meet etc.’. Both of them have traditionally been regarded as reflexes of PFU *tuli- ‘to come’ (< PU *toli-), but their first syllable vowels do not regularly go back to PFU *u, and thus better match PU *tolwa-. However, etymologically PU tolwa- may ultimately be linked with PU *toli- ‘to come’, as the words appear to correspond quite well in both form and meaning. The former could perhaps be analyzed as a derivative (PU *tol-wa-). However, this assumption remains uncertain, as the assumed suffixal element *wa cannot be reliably identified.

1.17. PS *tense- ‘to stop, calm down, lie down, stay still’ [> Nenets (derivatives) tyencyer-, tyencyena-] (Lehtisalo 1956: 507; Tereshchenko 1965: 649) < PU *san7 sa-7 ~ *sän7 sä-7 ‘to stand (still)’

The Nenets word family consists of parallel derivatives, on the basis of which a Proto-Nenets verb *t'en7 se-7 can be reconstructed. The word has no cognates elsewhere in Samoyedic, but the PS form could be regularly reconstructed as PS *tense- (in principle also *tenke- would be possible). This form can be compared with the PFU verb *san7 sa-7 ‘to stand’ (UEW: 431). The PFU reconstruction is based on Samic, Finnic, and Ob-Ugric cognates: cf. North Sami cuov zv zutv , saisa- (Finnish seisoa is irregular), Khanty (Vakh) li{ nt7 '-, Mansi (Pelymka) tun7 s-7 . The word also appears in several other languages, but in an irregular form: Mordvin st7 'a- ‘to stand up’, Mari siv n7 ce-7 ‘to sit; to stand’, Komi si7z- ‘to set oneself down’.

30 New and Old Samoyed Etymologies

The consonant correspondence between the compared items is regular. Samoyedic shows the expected shifts PU *s > PS *t and PU *s7 > PS *s. The reconstruction of initial *s- in the Finno-Ugric item is based on Ob- Ugric and Permic; the Sami initial c-v (< *s-7 ) has resulted from an assimilation caused by the *s7 in inlaut (cf. the identical development in North Sami cakv cav ‘autumn’ < *sük7 si7 < PFU *süksi7 ; UEW: 443). There are, however, slight problems involved in the vowel correspondence. The Samoyed form presup- poses PU *sän7 sä-7 , whereas the reconstructed PFU form displays back vocalism. But the comparison can still be accepted, as it appears that an ancient variation between front and back vocalic forms is reflected in the cognates. The reconstruction of a PU variant form *sän7 sä-7 is supported by Mari siv n7 ce-7 and Komi si7z-, which can be derived regularly from the form *sän7 sä-7 . The vowel development is the same as in PFU *sä7 snä7 ‘woodpecker’ > Mari siv stev , Komi and Udmurt si7 z7 (UEW: 772). Semantically the comparison is quite natural. The Nenets words show the meanings ‘to stop, calm down, be still etc.’, which comes fairly close to the meanings ‘to sit down, set oneself down’ in Mari and Komi. The meaning ‘to stop’ also occurs in derivatives in Finnic and Samic, e.g. North Sami cuov cv cahitv , Finnish seisahtaa. Many parallels for similar semantic shifts are known elsewhere, e.g. English stay < Old French ester ‘to be; to stand’ < Latin stare5 ‘to stand’ (Klein 1966–1967 s.v.).

1.18. PS *terq ‘full; contents’ [> Nganansan t'erq ‘interior’, Nenets tyero ‘something inside something else: e.g. liquor in a glass, items in a sledge, inhabitant of a land’, (der.) tyerøq- ‘to be full’] ~ *tirq [> Selkup tiri{ ‘entirity’, der. tiril{ ' ‘full’, tirimp{ i-{ ‘to become full’] (SW: 158) < PU *täwdi- ‘full; to become full’

The Selkup word has, independently from its cognates in Nganasan and Nenets, been compared to Komi and Udmurt tir{ ‘full’ and further with Sami darrat ‘to become hard and stiff’ and Finnish tyrmä ‘unconsciousness etc.’, tyrehtyä ‘to cease (e.g. bleeding)’ (UEW: 524; SSA s.v. tyrmätä). Both SSA and UEW consider the etymology certain, except for the Finnish words. The proposed Sami and Finnish cognates are, however, semantically question- able. But connecting the Permic words here seems possible in principle; the PS form *tirq presupposed by Selkup could be connected with the Permic words by assuming a PU form *türä and sporadic reduction of the stem vowel in Samoyedic. This etymology must, however, be rejected, as it does not take

31 Ante Aikio the Nganasan and Nenets cognates into account (whose vowel *e does not correspond well to the Permic forms), and because a more plausible etymol- ogy can be presented. The Samoyedic words are cognate with PFU *täwdi- ‘full; to become full’, which has certain reflexes in all Finno-Ugric language branches except Mordvin and Mari. This etymology was cautiously sug- gested earlier by Paasonen (1917: 79), but he considered a connection with Komi and Udmurt tir{ more plausible. Two etymologies have been proposed for Finnish täysi ‘full’. Skolt Sami tiiudâs ‘full’, Komi dela8 , Udmurt dol-dol (reduplicated) ‘entirely’, Khanty (Vasjugan) tel, Mansi (Tremjugan) täwl ‘full’, and Hungarian tel- ‘to be filled’, tele ‘full’ have been considered its cognates, and the PFU form has been reconstructed as *täwdi. On the other hand, it has been supposed that täysi is a derivative of a root *täwi-, which is reflected in Sami dievva ‘full’, dievvat ‘to become full’. UEW (518) considers the Permic and Ugric words cognate, but regards the Sami and Finnish cognates uncertain (as well as Mari tic7 ‘full’, which is phonetically obscure). SSA (s.v. täysi) also considers the Permic and Ugric words cognate, but takes the view that the Samic and Finnic words are of another origin. Collinder (1977: 132) considers the Finno-Ugric etymology certain, whereas T. Itkonen (1970: 12–13) finds the comparison of the Finnic-Samic words and their assumed cognates problematic and suggests retrogradical derivation (see below). Referring to Itkonen, Koivulehto (1976: 266–268) has proposed a Germanic loan etymology for the Finnic- Samic root *täwi. Let me first argue that PFU *täwdi ‘full’ can indeed be reconstructed. The Ugric and Permic cognates can be derived regularly from this form, though a reconstruction of *tälki would work just as well. As regards Finnish, the first alternative is obviously better, because täysi is the predictable reflex of PFU *täwdi. This explanation also suits Skolt Sami tiiudâs. The word contains a secondary adjective suffix -s like e.g. oodâs∞ ‘new’ (< PFU *wud'i). In addition to this, a homonymous verbal root is attested in Skolt tiuddâd ‘to become full’. But if Finnish täysi is derived from the aforementioned root *täwi-, which North Sami dievva, dievvat presuppose, one must assume a rather complicated retrogradical formation, since there is no such nominal suffix as *-ti. Finnish täysi would have been abstracted from the verb täytyä ‘to be necessary, have to; (archaic) to be filled, become full’ according to the analogy of such pairs as yksi ‘one’ : yhtyä ‘to join, unite (intr.)’, whereas täytyä would be a derivative of **täytää ‘to fill’, which in turn would derive from Finnic-Samic *täw-tä- (> North Sami deavdit ‘to fill’). This explana-

32 New and Old Samoyed Etymologies tion is not very credible, because there is no such verb as **täytää in Finnish or in any other Finnic language. Instead, there is täyttää ‘to fill’, whose morphological structure is rather täyt-tä- than täy-ttä-. The verb täytyä is thus a derivative of the noun täysi (cf. yhtyä ← yksi above). The only problem in reconstructing PFU *täwdi is that it does not offer any satisfactory explanation for North Sami dievva, though this has also been considered retrogradical. According to Sammallahti (1998: 238), dievva derives retrogradically from a verb that is reflected by Skolt Sami tieu/dded ‘to fill’ (this would be North Sami **dievdit). This explanation is not very credible either, nor is it necessary, as an adequate Germanic loan etymology has been proposed for the Sami word. Pre-Samic *täwi is probably a loan from Proto-Germanic *tæwia-5 (> Anglo-Saxon æl-tæwe5 ‘all good, excellent, entire, sound, healthful, perfect, honest’), as Koivulehto (1976: 266–268) has proposed. The loan etymology is further supported by Sami deavis ‘truthful, reliable, accurate’ (< Pre-Samic *täw-äs). However, it is unnecessary to derive Finnish täysi and Skolt Sami tiiudâs from this root, because they reflect the PFU root *täwdi- quite regularly. The (*)ie in Skolt tieu/dded (pro *ea, cf. North Sami deavdit < *täw-tä-) has probably been influenced by tiiudâs, which is of PFU origin. Reconstructing PFU *täwdi ‘full’ is thus fully justified. The rare cluster *wd does not hinder this etymology, even though this has been assumed (o.c.). The same cluster occurs in at least PFU *käwdi ‘rope’ (cf. UEW: 135, SSA s.v. köysi).4 It should also be noted that a unique consonant cluster has been reconstructed for many commonly accepted PFU or PU etymologies (Sammallahti 1988: 492–494), so even if there were no other examples of the cluster *wd, this would not be a reasonable argument against the validity of the etymology. This reconstruction is supported by Samoyedic, as PS *terq can be derived regularly from PU *täwdi. The developments PU *ä > PS *e and PU *d > PS *r are regular. Only the loss of preconsonantal *w in Samoyedic needs closer scrutiny. Such a development seems fully possible, although there are no certain examples of PU clusters of the type *wC. There are two things that support the assumption of this sound change. First, the fact that the second syllable vowel did not disappear in Samoyedic through apocope suggests that the first syllable was originally closed (cf. PU *nüdi ‘shaft’ > PS *nir). Secondly, Proto-Samoyed apparently did not have consonant clusters of the type *wC; there are none in the SW material. There are a couple of PFU etymologies for which this kind of cluster must be assumed (e.g.

33 Ante Aikio

*lewli ‘breath, spirit’, *säwni ‘ide’, see UEW: 247–248, 437), and even though these words do not have Samoyed cognates, there is no reason to assume that this cluster type was a PFU innovation. On these grounds, it is natural to assume a development PU *w > PS Ø in preconsonantal position. One somewhat uncertain parallel can be pointed out, PU ?*suwd'a ‘finger’ > PS ?*tqja (Sammallahti 1988: 540; cf. Janhunen 1981: 224, who reconstructs PU *sud'a). The only reflex pointing to *w in this etymology is Sami cuvdiv ‘index finger’, which Sammallahti (o.c.) has added to the etymology. However, the Sami cognate is problematic because of its initial cv (normally the reflex of PU *s7 ), which on the other hand might be due to sporadic palatal assimilation caused by the *d' in inlaut; cf. the assimilation before *s7 in Sami cuov zv zutv ‘to stand’ < *sa7 n7 sa-7 < PFU *san7 sa-7 and cakv cav ‘autumn’ < *sük7 si7 < PFU *süksi7 (UEW: 431–433). Proto-Samoyed *terq can be considered a phonetically regular reflex of PU *täwdi. The Selkup cognate points to PS *i, but this irregularity is internal to Samoyedic and does not hinder the etymology. There are also a couple of other examples of sporadic PS *e > Selkup i, e.g. Selkup tipi{ ‘nail, wedge’ < PS *tepq, cf. Nenets tyebo (SW: 157–158). Semantically the etymology is straightforward, as the meaning ‘full’ is attested both in Nenets and in Selkup. The meanings ‘interior’, ‘contents’ etc. are quite transparently derivable from ‘full’ via ‘filling’.

1.19. PS *time ‘tooth’ [> Nganasan t'imi, Nenets tyibya, Enets t'‰ı, Selkup timi{ , Mator time] (SW: 163; Helimski 1997: 360) < PU *sew(i)-mä ← *sewi- ‘to eat’

Almost all Uralic languages exhibit reflexes of the word *pini ‘tooth’ (UEW: 382). The only exceptions are found in the western and eastern peripheries of the language family, Samic and Samoyedic. The Sami word for ‘tooth’ is bátni, which is of unknown origin. The Samoyed languages show reflexes of PS *time ‘tooth’. The distributional criteria suggest that both the Samic and Samoyedic words for ‘tooth’ are of secondary origin and have replaced the original word *pini in these language branches. Abondolo (1996: 57–58) has suggested that Samoyedic *time can be explained as a secondary formation, whose components are, however, of Uralic origin: it is an obscured derivative consisting of a verb root *ti- and the nominal suffix *me. He connects the assumed verb root with PFU *sewi- ‘to eat’ (> Finnish syödä, Mordvin sevems, Hungarian eszik etc.) (UEW: 440).

34 New and Old Samoyed Etymologies

The exact morphological correspondent of PS *time is thus the Finnish third syömä-; cf. also the obscured derivative seimi ~ soimi ‘manger’ (Koivulehto 1996: 329–330). Abondolo’s etymology has recently been supported by Janhunen (2000: 73). The etymology is a very plausible one, but several additional remarks can be made. The framework of the Uralic vowel diachrony that Abondolo operates with differs considerably from that employed by Janhunen (1981) and Sammallahti (1988), but nevertheless Abondolo’s etymology of PS *time can be easily transferred to Janhunen’s model of Samoyedic historical phonology. PU *s > PS *t and PU *e > PS *i are regular sound laws, and the loss of PU *w does not pose a problem either. If the derivative was originally based on the consonant stem (PU *sew-mä), the preconsonantal *w has regularly disappeared in Samoyedic, as in PS *terq ‘full’ (see 1.18.). But PS *time can just as easily be derived from the form *sewi-mä, which has the vocalic stem. The development PU *w > PS Ø can also be considered regular before *i. There are two etymologies in support of this: PU *sowi7 ‘throat, mouth’ > PS *so, PFU *suwi7 and PU *liwi{ ‘bone’ > PS *le8 , PFU *luwi. Janhunen (1981: 261–262) and Sammallahti (1988: 538, 540) reconstruct PU * for these words, but this interpretation is problematic. As regards PU *sowi7 , South Sami tjovve ‘throat’ points unambiguously to PU *w. This word cannot be a loan from Finnic as Janhunen (1981: 262) assumes because of its initial affricate. The sound change *s7 > *c7 is one of the earliest innovations in Samic (Sammallahti 1998: 190), and if the word had been borrowed after the change, it would show the sibilant s- or sj- in South Sami. PU *liwi{ ‘bone’ has no reflex in Samic, but Mordvin lovazav points quite clearly to an original *w. Mordvin o goes back to Pre-Mordvin short *u, but before *x the vowels were combinatorily lengthened in Pre-Mordvin (cf. Sammallahti 1988: 523). Additionally, *x has developed into Mordvin j also after a , cf. PFU *toxi- ‘to bring’ > Mordvin tuje-, PFU *sixi-{ ‘to arrive’ > Mordvin saje- (UEW: 429–430, 529). The reconstruction of PU *w also accounts for the labialisation (*liwi{ > *luwi) in Finno-Ugric. Janhunen (1981: 253) has proposed that PU *w was regularly preserved in Samoyedic also before PU *i. This suggestion is based on one etymology only, PU *kawi ‘ear’ > PS *kaw. This etymology does not show regular vowel correspondence and has a very restricted distribution: the only assumed Finno-Ugric cognate is Finnish korva ‘ear’ (according to Janhunen < *kowra < PU *kaw-ra). The probative value of the etymology is thus weak, and in the light of two counterexamples it must be rejected. Note that PS *tiw ‘lung’ is

35 Ante Aikio not a counterexample to the sound law proposed here, as it actually does show the development *w > Ø: PU *tewiw > *tew > PS *tiw (cf. Janhunen 1981: 258).5 According to the etymology treated here, the semantic motivation of PS *time is ‘tooth’ = ‘something used for eating’. This is quite natural, and a parallel occurs in Indo-European: the PIE word for tooth, *h1dont-, has been explained as a present of *h1ed- ‘to eat’ (IEW: 287–289, Kluge 1995 s.v. Zahn). The Uralic languages have at least two other words for body parts that are derived with the suffix *mA: PS *nim-me7 ‘breast’ ← *nim-7 ‘to suck’ (SW: 110–111) and PFU *näl-mä7 ‘tongue, mouth’, which is an obscured derivative of PFU *neeli-7 (< *näxli-7 ) ‘to swallow’ (Janhunen 1992).6 Cf. also Samic-Finnic-Mordvin *waj-ma ‘breath’ (in Samic ‘heart’) ← PU *wajni- ‘to breathe’ (cf. UEW: 552–553, 809–810). There are also three other Uralic words for body parts ending in *-mA, PU *silmä7 ‘eye’, PFU *n7 irma{ ‘hip, groin’ and *kulma ‘eyebrow’ (UEW: 201, 312, 479). These may also have originally been derivatives, but this cannot be proved, as their putative basic roots have disappeared without a trace. The suffix *mA is rarely used to form a nomen instrumenti, but some examples are known: North Sami cohkunv ‘comb’ ← cohkutv ‘to comb’, soallun ‘toothpick’ ← soallut ‘to pick teeth’, Finnish vasama ‘blunt-headed arrow’ ← *wasa-7 ‘to shoot’ (SSA s.v.), Mordvin kojme ‘spade’ ← PU *kajwa- ‘to dig’ (see 2.2.), Ob-Ugric *wu5 cv qm ‘weir’ ← *wu5 c-v ‘to fish’ (Honti 1982: 191).

1.20. PS *wanc•- ‘to sneak near’ [> Selkup квончи-] ~ (der.) *wanc•l- [> Selkup kental-8 , Mator mandql-] ~ *wanc•r- [> Enets bador- ‘to hunt wild reindeer’, Mator (der.) mandqra ‘wolf’] (Helimski 1997: 300) < PU *wanca-v ‘to go slowly and cautiously’

This Samoyed word family is a hunting term, whose original Proto-Samoyed meaning was ‘to sneak (near)’. This meaning is preserved in Selkup and Mator, but in Enets it secondarily broadened to ‘hunting wild reindeer’ in general. In Mator also a taboo-motivated circumlocution for ‘wolf’ (< ‘sneaker’) has been derived from this verb. This word family is not included in SW, and it was reconstructed to Proto-Samoyed by Helimski. No further etymology has been proposed for it. PS *wanc•- considerably resembles in both form and meaning the reconstructed PFU verb *wanca-v , whose reflexes mean ‘to walk’ on the one

36 New and Old Samoyed Etymologies hand (in Samic, e.g. North Sami vázzit) and ‘to cross a river (wading, or by a bridge)’ on the other (Mari wonvzem, Komi vuvz-, Udmurt vi{ vz{i-, Khanty (Vasjugan) un. c-v , Mansi (Pelymka) wuns-v ). The sound correspondence is perfect except for the second syllable vowel, which cannot be reliably reconstructed in Samoyedic. The cognateness of the Finno-Ugric words is generally accepted (UEW: 557). Collinder (1977: 135) has doubted whether the Sami word belongs here, but there is no reason for such doubt whatsoever. Phonologically the etymology is convincing, although a few details must be commented on. On the basis of Sami and Mari a proto-form *wanca-v can be reconstructed. In Samic one must assume an irregular development PFU *a > Proto-Sami *a5 (pro *uo), but other examples of such a development are known, e.g. Sami gáskit ‘to bite’ < PFU *kacka-v (UEW: 641). The Permic vowels do not correspond to each other regularly, but the illabial vowel in Udmurt is secondary (see E. Itkonen 1954: 317 and cf. 1.4.), and the Proto-Permic form can be reconstructed as *vu3 vz¥ı-. Proto-Permic *u3 goes normally back to PFU *o or *i{ , whereas the regular reflex of PFU *a is Proto-Permic *u. However, the Permic vocalism poses no problem because an identical vowel develop- ment is encountered in the homonymous PU noun *wancav ‘root’ (> Proto- Permic *vu3 vz¥ı > Komi vuzv , Udmurt vi{ zv i{ ) (UEW: 548–549) and the near- homonymous PFU verb *panca-v ‘to open’ (> Proto-Permic *pu3 vz¥ı- > Komi puvz- ‘to roll up (e.g. sleeves)’, Udmurt puzalt-v ‘to roll up, turn (up), bend, fold’) (Abondolo 1996: 101; cf. UEW: 352, which does not mention the Permic cognates). These three examples suggest that the vowel correspond- ence is actually regular in this particular context. There are no counterexamples, as PFU *a is not reflected as Proto-Permic *u in any word with an initial . The development is also phonetically natural, as it presupposes a labialization *a > *o after labial consonants in Pre-Permic. Additional examples might be Finnish vanka ‘handle’ ~ Proto-Permic *vug3 (> Komi, Udmurt vug) and Finnish paras ‘best’ ~ Proto-Permic *pur3 ‘good’ (> Komi pur, Udmurt bur); the Aryan loan etymologies proposed for these words (SSA s.v. vanko; Koivulehto 1999: 229) require an original *a rather than an *i{ . According to Sammallahti (1988: 500), a similar labialization after labial consonants has taken place in Proto-Ugric, too. Semantically Sami vázzit ‘to walk’ comes especially close to the Samoyed cognates, whose primary meaning is ‘to sneak (near)’. Sneaking is slow and cautious walking, and quick movement on foot (hurrying, running etc.) does not belong to the semantic field covered by the Sami verb either. Old Indic

37 Ante Aikio sárpati ‘to sneak, crawl, go, walk’ (IEW: 912) and Old Norse skrı‰ “a ‘to sneak’ ~ German schreiten ‘to pace, walk’ (Buck 1965: 685) serve as parallels. A similar shift is currently also taking place in Finnish hiippailla, hipsiä ‘to sneak; (colloquial) to go somewhere (on foot)’. The meaning ‘to walk’ is attested only in Samic, whereas the cognates elsewhere in Finno-Ugric mean ‘to cross a river (wading, or by a bridge)’. Here a shift ‘to walk slowly and cautiously’ > ‘to cross’ may be assumed. A somewhat similar semantic development has taken place in English wade < Old English wadan ‘to go, advance; to wade’ (Klein 1966–1967 s.v.) Wading across a river requires cautious movement, especially in rapids, which are typical crossing points because of their shallowness. The same has also applied to crossing by a bridge in the era of unsteady bridges. Formerly bridges were usually narrow, more or less temporary constructions with no railings, built only of a couple of large tree trunks.

1.21. PS *e-8 ‘to drink’ (only in derivatives): PS *e-r-8 ‘to drink’ [> Nenets nger-, Selkup ör- ‘to become intoxicated, get drunk’, Mator er-8 ], PS *e-k8 ql- [> Nenets ngexøl- ‘to drink with one gulp’, Enets ihora- ‘to drink exces- sively’] (SW: 21–22; Helimski 1997: 238) < PU *jixi-{ ‘to drink’

The Proto-Samoyed word for ‘drinking’ has been reconstructed as *er-8 . Janhunen (SW: 21–22) also cautiously compares this verb with PS *ek8 ql- ‘to drink avidly’, which has reflexes in Nenets and Enets, and suggests that the word contains the PS augmentative suffix *-r-. This analysis is in full accordance with Samoyedic derivational patterns, and is thus quite plausible. A PS verb *e-8 ‘to drink’ can be reconstructed, even though it has been preserved only in the two parallel derivatives *er-8 and *ek8 ql-. The reflexes of PS *er-8 have previously been compared to Mordvin ired7 'e- ‘to get drunk’ and Komi ir{ e8 sv ‘weak beer’. UEW (85) considers the compari- son certain and reconstructs a PU verb *ir•- ‘to drink’. However, this etymology can be discarded, because it is incompatible with the morphologi- cal structure of the Samoyedic forms, and because the vowel correspondence is not regular. As PS roots of the shape *(C)V- regularly correspond to PFU *(C)Vxi-, the reconstructed PS verb *e-8 has a fairly obvious point of comparison in the PFU verb for ‘drinking’, which has been reconstructed as *juxi-. Certain reflexes of this etymon are found quite widely: Sami juhkat, Finnish juoda,

38 New and Old Samoyed Etymologies

Mari jüä-, Komi and Udmurt ju-, Hungarian iszik (UEW: 542). The compari- son is not entirely regular; problems are involved both in the reconstruction of the initial consonant (PFU *j ~ PS Ø) and of the first syllable vowel (PFU *u ~ PS *e8 ). However, both of these irregularities can be accounted for, and the comparison can be considered fairly secure. As regards the initial consonant, a sporadic loss of *j before the high unrounded vowel *i{ must have taken place in Samoyedic. A convincing parallel for this irregular development is found in the PU word *ji{ nsi ‘bow’ > PSam *(j)int{ q, where the initial *j has been lost in all Samoyed languages except Nganasan (d'intq ‘bow’; cf. Nenets ngino, Selkup int{ i{ etc.). There are two somewhat similar cases of irregular loss of the glide *w-. PS *ü- ‘to pull, drag’ (SW: 30–31) must derive from an earlier form *wü-, because, as Helimski (1999b) has pointed out, it must be cognate with the PFU verb *wixi- ~ *wexi- ‘to take, transport’. Another fairly certain case of loss of *w- occurs in PS *aqj- (~ *aq- ~ *äqj-) ‘to be’ (SW: 16–17) < PU *waxli-. The PFU form of this verb should be reconstructed as *wooli- rather than *woli-, as almost all the cognates show the reflex of PFU *oo in the first syllable; the short vowel in Finnish olla ‘to be’ must be secondary. Despite the many phonological irregularities involved in this comparison, it can be considered correct, as it seems unlikely that the close resemblance of the Samoyedic and Finno-Ugric forms would be due to mere chance, and because the existential verb is susceptible to irregular sound changes due to its high frequency. The illabial vowel in PS *e-8 poses another problem for the comparison with PFU *juxi-. However, it must be noted that there are established etymologies where PS *i{ / *e8 irregularly corresponds to PFU *u: PS *kins{ •kajq ‘star’ ~ PFU *kun7 s7 •, PS *le8 ‘bone’ ~ PFU *luwi, PS *kij{ ‘moon’ ~ PFU *kuCi (?*kuwi, ?*kuxi, ?*kuni) (Janhunen 1981: 261–263). The most natural explanation for these cases is a sporadic labialization *i{ > *u in Finno- Ugric. Furthermore, as regards PFU *juxi-, the labial vowel *u can be shown to be, in all probability, an even later and more western innovation. The only cognates unambiguously pointing to PFU *u are found in Samic and Finnic. The irregular rounded in Mari jüä- can also derive from *u, but the Permic and Hungarian cognates actually show regular reflexes of PFU *i{ : Hungarian iszik and Proto-Permic *ju-3 > Komi, Udmurt ju-. The regular reflex of PFU *u is illabial *ı¥{ in Permic, but the Permic evidence for PFU *i{ is not absolutely certain, as there are a couple of cases in which PFU *u has been irregularly retained as labial *u3 in Permic (e.g. Komi and Udmurt uj- ‘to swim’ < PU *uji-; UEW: 103). In Hungarian the regular reflex of *u is (*)a

39 Ante Aikio before PU *x. The Hungarian i- in iszik was formerly regarded as the result of a secondary development caused by the word-initial glide *j-, but the Samoyedic form *e-8 suggests that it is rather an archaic retention of the PU vowel *i{ in this particular word. The PU word *ji{ nsi ‘bow’ is a similar case in this respect, too, as here Hungarian has also retained the illabial vowel (íj ‘bow’), whereas the other branches, including Ob-Ugric, show a sporadic labialization of the vowel (*ji{ nsi > *jonsi). Despite the substantial irregularities involved in the comparison between PFU *jixi-{ / *juxi- and PS *e-8 , the reconstruction of a PU verb *jixi-{ ‘to drink’ can be considered justified. As the PFU etymon is well-established and its reflexes are widely attested, the addition of PS *e-8 as a Samoyedic cognate to this etymology can be accepted in spite of the sporadic loss of *j- it presupposes. As regards the vowel correspondence, it must be observed that it is actually the westernmost (“Finno-Volgaic”) cognates that show an irregular development, not Samoyedic.7

2. Arguments for new approaches to previously proposed Uralic etymologies

2.1. PS *äs•l- ‘to step over’ [> Selkup asel-5 ] (Sammallahti 1979: 26) < PU *askili-7 ‘(to) step’

The etymon has widely attested indisputable reflexes in Finno-Ugric (see UEW: 19), and the Selkup cognate has also usually been regarded as certain. But Sammallahti (1979) marked the comparison with a question mark, later rejecting it (1988: 542), as did Janhunen (1981). The reason for the rejection was probably the vowel correspondence PFU *a ~ PS *ä, which Sammallahti (1979) suggested might be due to palatalization before *s7 in Samoyedic. This explanation is indeed correct, and the correspondence is actually regular: the shift PU *a > PS *ä took place before a tautosyllabic . The same development occurs in PS *käjwa ‘spade’ < PU *kajwa (see 2.2.) and PS *wäjn ‘breath, spirit’ < PU *wajni; the latter etymology is accepted by Janhunen (1981: 256) and Sammallahti (1988: 541) also. There are no counterexamples to the sound law proposed here, as long as one maintains that the shift took place before PU *o was lowered and merged with PU *a under certain conditions in Samoyedic. Thus e.g. PS *masq- ‘to wash’ < PU *moski-7 (Janhunen 1981: 222) has PS *a, not *ä.8 This

40 New and Old Samoyed Etymologies explanation is not resisted by PS *kas- < PU *kasi-7 ‘to donate’ (o.c. 43), as the development may have taken place before the apocope in PS. Further- more, it must be noted that PU *kasi-7 has been preserved in Samoyedic only in the nominal derivative *kas-o ‘gift, sacrifice’ (SW: 61), which has an open first syllable. The etymology treated here shows that the change PU *a > PS *ä took place before PU *k regularly disappeared in clusters with obstruents in PS and the first syllable became open. The lateral *l has normally been preserved also before a PU high vowel on the border of the second and third syllables, as shown by PU *kunili7 ‘tear’ > PS *kq7nqlq (Janhunen 1981: 255).

2.2. PS *käjwa ‘spade’ [> Nganasan kajbu, Enets sea, Nenets syíwa, Kamass ko5 ] (SW: 63) < PU *kajwa- ‘to dig; spade’

SKES (s.v. kaivaa) mentions the Samoyedic *käjwa word family in connec- tion with the Finnish verb kaivaa ‘to dig’, but considers the comparison uncertain and offers an Altaic loan etymology (cf. Mongol qajibi ‘oar’ etc.) as an alternative. Sammallahti (1979: 43) also considered the etymology uncertain, on the same grounds as SKES. Janhunen (1981) rejected the comparison, and Sammallahti (1988: 552) agreed with him; likewise, the proposed Samoyed cognates are no longer given in SSA (s.v. kaivaa). The loan etymology is, however, unnecessary, because the Samoyed words can be connected according to regular sound laws with PFU *kajwa- ‘to dig; to throw’, which is widely attested in Finno-Ugric. The correspondence PFU *kajwa- ~ PS *käjwa is predictable. The change *a > *ä took place in Samoyedic before a tautosyllabic palatal consonant (cf. 2.1.). PFU *kajwa- has certain reflexes in all Finno- except Ob-Ugric: Sami goaivut ‘to dig, scoop, ladle’, goaivu ‘spade’, Finnish kaivaa ‘to dig’, Mordvin kaja- ‘to throw away’, (der.) (Erzya) kojme, (Moksha) kajmä ‘spade’, Mari koe- ‘to dig’, (der.) kol'mo ‘spade’, Komi koj- ‘to throw water on the sauna stove; to pour, shake’, Udmurt (der.) kujal- ‘to throw away’, Hungarian (der.) hajít ‘to throw’ (cf. UEW: 116–117, 170–171, SSA s.v. kaivaa). It has been doubted whether all these words are of the same origin. According to SSA the Hungarian cognate is uncertain, and Mordvin kojme is not mentioned at all. UEW groups the words under two roots, namely PFU *kaja- ‘to throw’ (> Mordvin kaja-, Permic, Hungarian) and *kojwa- ‘to dig’ (> Sami, ?Finnish, Mordvin kojme, Mari, ?Samoyedic), and

41 Ante Aikio denies any etymological connection between these on semantic grounds (sic). The view taken by UEW is unacceptable, because all these words can quite easily be derived from one proto-form. The first syllable vowel *o which UEW reconstructs for the latter root apparently on the basis of Samic, is impossible. All the other languages point towards an original *a, and the Sami vowel must be regarded irregular. Also Erzya Mordvin o in the derivative kojme is obviously secondary, cf. Moksha kajmä; Erzya o could not even theoretically support the reconstruction given by UEW, as PFU *o gives Mordvin u before an *a in the second syllable. The cluster *jw that must be reconstructed on the basis of Sami, Finnish and Samoyed, also matches well with the other cognates. In Mordvin, Permic and Hungarian a postconsonantal *w has regularly disappeared, cf. PFU *tälwä ‘winter’ > Mordvin t'el'e, Komi tel8 , Udmurt tol, Hungarian tél (UEW: 516). The Mordvin consonantism in the derivative kojme < *kajwa-ma has an exact parallel in kujme ‘basket’ < *kojwa-ma ← PFU *kojwa ‘birch’. The devel- opment *jw > Ø in Mari is also predictable, cf. Mari kue ‘birch’ < *kojwa. (UEW: 169–170.) Semantically the etymology is quite transparent, despite the opposite view taken by UEW. The connection between ‘digging’ and ‘throwing’ is trivial – the intermediate stage between them has most probably been ‘scooping, ladling’, which is attested in Sami. The nominality of the Samoyed cognates does not hinder the etymology, as it is well known that the PU vocabulary contained many nomenverba. It seems that the word has retained its original nomenverbum character in Samic, but this is not entirely certain, as the -u in goaivu ‘spade’ might also be a nominal suffix.

2.3. PS *kqr ‘morning’ [> Selkup qar, Kamass (der.) karqld'en ‘tomorrow’] (UEW: 193) < PU *kudi ‘morning’

A Uralic etymology based on various words for ‘morning’ and ‘tomorrow’ has been widely quoted in etymological standard literature (e.g. Collinder 1977: 36; UEW: 193). On the Finno-Ugric side the cognates are restricted to Ugric: Hungarian holnap ‘tomorrow’ (nap ‘day’), Khanty (Vakh) kolt3 qg id., Mansi (Tremjugan) kol ‘morning’. Many Samoyedic words have been grouped under this etymology. However, they belong to as many as three distinct word families which have mistakenly been considered cognate: 1)

42 New and Old Samoyed Etymologies

Nenets xúwoh ‘in the morning’, Mator kuman5 ‘tomorrow’ < PS (?) *kuqmq- (SW: 77), 2) Nganasan kiduatu. , Enets kiduduo ‘morning’, which are deriva- tives based on the PS verb *kite- ‘to wake up’ (SW: 72), and 3) the Selkup and Kamass items deriving from PS *kqr (see above). The suggested etymology is impossible because it lumps together three Samoyedic word families which cannot, on phonological grounds, have anything to do with each other. However, when the word families 1) and 2) are correctly discarded from the etymology, the remaining comparison between the family 3) and the Ugric words is quite satisfying. The Ugric vowel correspondences point to PUg *u3 < PFU / PU *u. This fits perfectly with PS *q, which is the regular reflex of *u in PU open syllables. The correspondence Ugric l ~ PS *r points to an earlier *d. The PU form can thus be regularly reconstructed as *kudi.

2.4. PS *kiwe ‘snake, worm’ [> Nenets syibya ‘larva under the skin of reindeer’, Selkup süv ‘snake’, (compound) ütsiv ‘leech’ (üt ‘water’)] (SW: 72) < PU *küji-wä ‘snake’ ← *küji id. (? ~ *kiji)

This etymology has been considered certain in standard etymological literature (UEW: 154–155, SSA s.v. kyy). However, Sammallahti (1979: 30) questioned the comparison between Finno-Ugric and Samoyedic on the grounds that PFU *j ~ PS *w is not a regular correspondence. Presumably, this is the reason why Janhunen (1981) and Sammallahti (1988: 544) reject the etymology. The Finno-Ugric comparison has, however, never been doubted, even though there is some slight phonetic inconsistency in the cognates. Finnish kyy and Udmurt kij{ point to an original PFU *ü, Mari kiskv äq and Hungarian kígyó to PFU *i. The Mordvin variants with u (kuj, guj, kju) are wholly irregular. The original front vowel can be traced in Mordvin kijov. The variation is probably due to affect. The Samoyed cognates have been rejected on erroneous grounds, be- cause PS *w in the word *kiwe does not reflect PU *j. Instead, only PS *ki- represents the original root, and *we must be regarded as a suffix (< PU *küji- wä). The suffix *wä is scarcely attested, but there is at least one clear example: PFU *koj-wa ‘birch’ < PU *koxji > PS *koqj ‘birch’ (Janhunen 1981: 241). According to Janhunen (o.c. 259), PU *j was retained in Samoyedic also before a PU high vowel in the second syllable, and thus the development PU *küji > PS *ki would not be possible. This assumption is, however, based on one etymology only, PS *tqj ‘birch bark’ < PU *t•ji, which

43 Ante Aikio shows irregular vowel correspondence and a very restricted distribution: the only Finno-Ugric cognates are found in Permic. The probative value of this etymology can be questioned, because there is a clear counterexample: PU *uji- ‘to swim’ > PS *u-. This etymology is considered irregular by Janhunen (o.c. 260–261), but the comparison in question is not nearly as problematic as the aforementioned etymology *t•ji. Furthermore, two other slightly problematic comparisons may support the assumed loss of PU intervocalic *j before a high vowel. PS *kora ‘reindeer bull’ probably derives from PU *koji-ra (cf. PFU *koji, *koji-ra ‘male’); however, the comparison is compli- cated by the unexpected retention of *o in Finno-Ugric (one would expect PFU **kuji on the grounds of the sound law PU *o > PFU *u /_Ci). An irregular cognate of PFU *pajiw ‘willow’ might be involved in PS *pewq ‘willow bark’ (SW: 122) (? < Pre-Samoyedic *päjiw•). The Uralic etymology of PS *kiwe is further reinforced by the fact that the word appears as a secondary derivative in some other languages, too. Mari kiskv äq goes back to *kiji-skä. The exact correspondent of PS *kiwe is Mordvin kijov, where -v reflects the derivative *wä. The retention of *w suggests that a low vowel originally followed it, as a word-final *w has disappeared without a trace in Mordvin: cf. PU *wäniw ‘son-in-law’ > Mordvin ov, PFU *kintaw{ ‘tree stump’ > Mordvin kando (Sammallahti 1988: 541, 543). The second syllable labial vowel in Hungarian kígyó can also be explained with the suffix *wä: the development has been kígyó < *kij•wu < Proto-Ugric *kıj¥ ı¥ gä < PU *kijiwä. Proto-Ugric *g has had similar effects on the develop- ment of non-initial syllable vowels in Hungarian in general (cf. Papp 1968: 130–131).

2.5. PS *tqnta- ‘to teach, to accustom to’ [> Nenets (der.) tønara-, Enets tadda-], *tqntö- ‘to learn, to get used to’ [> Nenets tønyo-, Selkup tanti-{ ] (SW: 147) < PU *tun-ta- ‘to teach, to accustom to’, a causative of *tuni- ‘to learn, to get used to’

UEW (537) connects the Samoyedic words with Mordvin tonado-, Mari tunema- ‘to learn, to get used to’, Mordvin tonavto-, Mari tunäqkte- ‘to teach, to accustom to’, Hungarian tanul- ‘to learn’, tanít ‘to teach’. SKES (s.v. tottua) regards only the Mordvin and Mari words as certain cognates. Sammallahti (1979) and Janhunen (1981) do not accept the Uralic etymol- ogy; Sammallahti (1988: 550), however, connects the Mordvin, Mari and

44 New and Old Samoyed Etymologies

Hungarian words and reconstructs PFU *toni- ‘get used to’ as their proto- form. The Samoyed words go back to PS *tqnta-, whose meaning is transitive ‘to teach, accustom to’; in addition to this, there is a passive-reflexive derivative *tqntö- ‘to learn’. The form *tqnta- goes regularly back to PU *tunta-. This can be explained as a causative of a PU verb *tuni-; the derivative is based on the consonant stem, cf. PS *kan- ‘to go away’ < PU *kani-, PS *kanta- ‘to carry, transport’ < PU *kan-ta- (Janhunen 1981: 221, 231). The Finno-Ugric cognates, which are also derivatives, can be derived from the form *tuni-; the PFU *o reconstructed by Sammallahti is not necessary. The form *tuni- accords better with the Mordvin and Mari reflexes. Hungarian a < *u is apparently irregular, but not unheard of: cf. Hung. halad- ‘to go, proceed’ < PU *kulki-, Hung. tat ‘stern (in ship)’ < PU *tukta (UEW: 198, 534).

2.6. PS *wankq ‘hole, cave, den’ [> Nganasan banka, Nenets wangko, Enets baggo, Selkup (der.) kongqr] (SW: 171–172) < PU *wonki id.

Cognates in Samic, Finnic and Ob-Ugric have been proposed for the Samoyedic words: Skolt Sami vuâg/g ‘cave, den’, Finnish onkalo ‘hollow, cave’, Khanty (Vakh) wa3 nkq ‘cave, hole’, Mansi (Tavda) wanka5 ‘hole, grave’. This etymology is also considered certain by UEW (583). SSA takes a more reserved attitude, but considers it likely that the Ob-Ugric and Samoyed words are of the same origin. For the Finnic words SSA also proposes alternative Samic cognates, Inari Sami uággi ‘creek, inlet’, Lule Sami oagge ‘deep place in a river below rapids’. Sammallahti (1979) and Janhunen (1981) have rejected the comparison as regards Samoyedic. Sammallahti (1988: 551), however, considers the Samic, Finnic and Ob- Ugric words cognate and reconstructs PFU *wonki as their original form. PS *wankq and PFU *wonki can be derived from PU *wonki according to regular sound laws. The vowel *o has regularly been lowered to PS *a before consonant clusters. Only the retention of PU *k in Samoyedic needs commenting on, because normally it has disappeared before a PU high vowel (see 1.13.). But a loss is not to be expected in this case, as the intermediate stage in the loss was probably spirantization *k > *g and, from an articulatory point of view, it is only natural that this did not take place in the cluster *nk. The nasal n already produces a complete velar stricture, and thus a change

45 Ante Aikio

*nk > *ng would be an articulatory complication. In Ob-Ugric the cluster *nk is also generally retained even though PU *k was spirantized in all other voiced contexts. The Samic and Finnic cognates may also be considered certain. Skolt Sami vuâg/g goes back to Proto-Sami *vuonko5 , which derives regularly from PU *wonki, except for the second syllable labial vowel which might be a suffix. Likewise, although the second syllable a in Finnish onkalo does not regularly derive from PU *i, this does not impede the etymology which is otherwise regular. It is also possible that the vowel is a part of the suffix, cf. Finnish vartalo ‘body’ ← varsi ‘shaft’. The Karelian correspondent onkura ‘(noun) hollow, cave; (adj.) hollow, concave’ exhibits a different suffix (cf. Selkup kongqr), and has apparently acquired a descriptive character. The alternative Samic cognates mentioned in SSA belong to this connec- tion, too, although in a circuitous way. They are loans from Finnish (eastern dialect) and Karelian vonka ~ vonkka ‘river bend, deep in a river (e.g. below rapids); cove, bayhead; a hole with water in it; a cleft in rock created by a whirlpool etc.’ (SSA, SKES s.v.). The v- in initium is left without a substituent in Samic, because the sequences *voa- and *vo- were phonotactically illegal in Proto-Sami. Finnish and Karelian vonka is in turn an obvious loanword, too, as the initial sequence vo- is unetymological in Finnish.9 Proto-Sami *vuonko5 ‘cave, den’ fits perfectly as a loan original. For the semantics cf. Finnish luoma ‘brook’, luomanne ‘depression, gently sloping shore’, luomi ‘a river bank with promontories; cave, hollow’ ~ Estonian loom ‘river bend’ (cf. SSA s.v. luoma1). Since many other Samic loanwords in Finnish and Karelian have already been demonstrated (see Koponen 1996 and the literature mentioned there), borrowing from Samic to Finnic can be assumed naturally. There is also one other Samic loanword with the initial sequence vo- in Karelian and the eastern of Finnish, vorva ‘alert; light sleeper’ (cf. SKES s.v.).

2.7. PS *jürq- ‘to get lost; (>) to forget’ [> Nenets yurø- ‘to forget’, Enets (der.) jurota- id., Selkup üri-{ ‘to get lost’, (der.) ürqi-{ ‘to lose one’s way’, Kamass d'ur- ‘to get lost, disappear’, (der.) d'urzv u- ‘to lose one’s way’] (SW: 50) < PU *juri- ‘to go round, spin, (>) to get lost’

The Samoyedic word family can be compared to Khanty (Tremjugan) jor3 qglq- ‘to forget’, Mansi (Pelymka) jorl-5 id. < Proto-Ob-Ugric *jor3 qgl-

46 New and Old Samoyed Etymologies

(Honti 1982: 114) and Udmurt j{irom{i- ‘to go astray, lose one’s way’, as suggested in UEW (108–109). The Ob-Ugric and Udmurt verbs are deriva- tives. The Uralic etymology is further supported by a previously unnoticed cognate in Samic: the underived verb root is preserved in North Sami jorrat ‘to go round, spin; to fall over; to bustle about; to veer (wind); etc.’. The Sami word has a wide range of obscured derivatives, which also suggests that the word must be of relatively high age. There are at least three derivatives based on the consonant stem. The adjectives jorbbas, jorbat ‘round’ (< *jur-p-ita, *jur-p-isv ) contain the nominal suffix *-pa and an additional adjective suffix. The verb jorgut ‘to turn (tr., frequentative)’ probably contains the nominal suffix *-ka and the suffix *-o-, which creates denominal verbs (cf. e.g. Sami noaidut ‘to conjure’ ← noaidi ‘shaman’). The word jorbmi ‘whirlpool; deep place in water’ (< *jur-ma) is also most naturally explained as a deverbal noun based on the verb jorrat. In etymologi- cal literature the word jorbmi has often been erroneously compared to a Samoyedic word family with the meaning ‘deep’ (PS *jore; SW: 47) and certain other words (UEW: 105), but this etymology must be rejected for phonological reasons. The Sami cognates suggest that the meanings ‘to get lost’ and ‘to forget’ attested elsewhere in Uralic must have developed from the meaning ‘to go round, spin’. The semantic derivation is natural, and there are numerous parallels. According to Buck (1965: 1185–1186), Indo-European “words for ‘error, mistake’ are most commonly derived from vbs. meaning ‘wander about’ or the like”. Example cases include e.g. Latin error ← errare ‘to wander, go astray; to err, mistake’, Dutch dwalen ‘to err; to wander’, and bhrama- ‘to wander about; to be confused, mistaken’. The analysis is confirmed by several derivatives in Samic, which show the same semantic development as the Samoyedic cognates. E.g., the verbs jorbbodit ‘to become round’ (← jorbbas ‘round’) and jorggiidit ‘to begin to turn’ (← jorgut ‘to turn’) are also used in the meanings ‘to become perplexed or bewildered’ and ‘to lose one’s way’. Phonologically the comparison between Samic, Permic and Ob-Ugric is regular. The Sami word goes back to PFU *juri-, and Proto-Ob-Ugric *o3 and Udmurt i{ are regular reflexes of PFU *u. As regards Samoyedic, the comparison with PFU *juri- is not entirely regular; one expects PU *u to be retained unchanged in an open syllable before second syllable *q. However, the comparison can be accepted, as the etymology is semantically satisfying

47 Ante Aikio and the irregular PS *ü can quite plausibly be accounted for. The reason for the sporadic fronting of the vowel was probably the palatalizing effect of the word-initial glide *j. Moreover, it is possible that the irregular development *ju- > *jü- had not yet taken place at the Proto-Samoyedic stage, but only occurred later in Selkup. The evidence from other languages is ambiguous, as it is not possible to distinguish the reflexes of PS *ju- and *jü- in Nenets, Enets or Kamass: cf. PS *ju- ‘to become warm’ > Nenets yu-, Enets ju-, Kamass d'u- (SW: 47). The reconstructed PU verb *juri- is probably also reflected in Hungarian jár ‘to go, walk about, etc.’. A connection between Sami jorrat and Hungar- ian jár has been previously suggested (UEW: 102). The semantic develop- ment has been ‘to spin, go round’ > ‘to walk about’ (> ‘to go’). The vowel development has been slightly irregular in Hungarian, too. One expects Hungarian o as the reflex of PU *u. However, there are examples of irregular lowering of *u to a in Hungarian (see 2.5.).

3. Summary and discussion

I have above presented twenty new Uralic etymologies for Samoyedic words (including the cases *täjwa ‘nail, wedge’ and *wäjk ‘shoulder’ only briefly commented on in note 8) and presented additional evidence supporting the Uralic origin of the Samoyedic words for ‘tooth’, ‘female, mare’ and ‘earth, land’ recently suggested by Daniel Abondolo. It has also been shown that nine of the previously proposed etymologies which were not included in Janhunen’s (1981) corpus of Proto-Uralic etymons can be safely traced back to Proto-Uralic (including the words *peksä- ‘to beat’ and *waxli- ‘to be’ commented on in 1.21. and note 8). The word PU *lip{ si7 ‘cradle’ > PS *jeps8 q ~ *leps8 q can also be added to the list of rehabilitated etymologies (see UEW: 260; the etymology is not included in Janhunen 1981, but it was later accepted by Janhunen (2000: 69)). Moreover, the sound law PU *k > PS Ø /_i allows the addition of five items to the reconstructed PU lexicon (see 1.13.). New credible Uralic etymologies have also been proposed by other researchers after Janhunen’s (1981) paper: PU *ikta-{ ‘to hang’ > PS *ita-{ (Sammallahti 1988: 536), PU *tanti- ‘ground; to tread’ > PS *tantq- (Rédei 1998), PU *edi- ‘front’ > PS *ir- (Helimski 1986: 135–136), PU *läsi- ‘near’ > PS *jet- (Helimski 1999a), PU *jupta- ‘to speak, tell’ > PS *jqptq-, PU

48 New and Old Samoyed Etymologies

*wixi- ‘to take’ > PS *ü- ‘to pull, drag’, PU *läsä-7 ‘to arrange, prepare, cover etc.’ > PS *jese- ‘to cover the tent’, PU *itä- ‘to appear, come in sight’ > PS *qte- (Helimski 1999b; see 1.12. for parallels on the development *i > *q). The total number of Uralic etymologies that can be added to the corpus is thus 46. The number is substantial, considering that Janhunen 1981 lists only 140 Samoyedic lexical items having certain cognates elsewhere in Uralic. The new etymologies mean an increase of 33 % in the reconstructed PU lexicon (or 31 %, if one rejects the etymologies presented for PS *kaw ‘ear’ and *tqj ‘birch bark’ in Janhunen 1981; see 1.19. and 2.4. for critical remarks). The increase in the number of etymologies is actually a typical example of how progress in comparative reconstruction is achieved. Historical pho- nology and etymology are dependent on each other and thus in constant interaction. The phonology of a proto-language is first reconstructed on the basis of the phonologically and semantically most transparent etymologies (and thus the most likely ones to be correct), a process in which some of the etymologies may have to be discarded; then, the validity of the semantically and phonologically non-transparent correspondences is reassessed and the emerging new etymologies are evaluated on the grounds of the proposed sound laws, the new cases possibly allowing the correction of sound laws or the postulation of new ones. Considering that the reconstructed Samoyedic sound laws have predicted 46 etymologies so far, it is quite obvious that no major reinterpretation of the correspondences is needed anymore. Nevertheless, some minor modifica- tions can be made on the basis of the new etymologies. The following sound laws can be added to the ones leading from Proto-Uralic to Proto-Samoyed as reconstructed by Janhunen (1981). Sound law 4) is based on Sammallahti’s reinterpretation of the correspondence noticed by Schrijver (see 1.13.). Note that sound laws 2) and 3) contradict Janhunen’s view that PU *w and *j were retained in PS before PU high vowels (see 1.19., 2.4. for details).

1) PU *w > PS Ø preconsonantally (see 1.18., 1.19.) 2) PU *w > PS Ø intervocalically before *i (see 1.19.) 3) PU *j > PS Ø intervocalically before *i (see 2.4.) 4) PU *k > PS Ø intervocalically (?) and after a liquid, if followed by *i (see 1.13.; cf. 2.6) 5) PU *a > PS *ä before a tautosyllabic palatal(ized) con- sonant (see 2.1., 2.2.) 6) PU *i > PS *q word initially (see 1.12.)

49 Ante Aikio

Sound law 5), which explains the occurrence of Proto-Samoyed *ä in certain items of Proto-Uralic origin, is of special interest. The solution proposed here is not a completely satisfactory one and it seems that the conditioning factors of the shift *a > *ä may have to be somewhat revised in the future. There are certain Proto-Samoyed roots which show an *ä instead of the expected *a, and the solution proposed here does not account for these cases: PS *täkq- ‘behind’ ~ Finnic-Samic *taka- (UEW: 506–507), PS *kämq ‘scale’ ~ PFU *kama ‘skin’ (UEW: 121–122), PS *äntq ‘blade’ ~ PFU ? *anti ‘spear’ (UEW: 342), PS *jäpsq- ‘roasting spit; to roast’ ~ PFU ? *japsi-7 (> Proto- Permic *jus7 ; the other Finno-Ugric cognates display irregular vowel corre- spondence; UEW: 94), PS *pätq- ‘to pot’ (SW: 118) ~ PFU *pata ‘pot’ (UEW: 358), PS *än ‘mouth’ ~ PFU *ani ‘opening, mouth’ (UEW: 11), PS *nä7 nq- ‘to take off (clothes)’ ~ PFU *ana- ‘to open, take off’ (a new etymology; cf. SW: 106–107; UEW: 11).10 The last item shows an irregular prothetic nasal in anlaut. This is a fairly common phenomenon in Samoyedic, especially in words containing a nasal in inlaut: cf. PS *äncq ~ *näncq ‘belly’ (SW: 20), *ejme ~ *nejme ‘needle’ (SW: 22), *ä ~ *nä7 ‘friend’ (SW: 106; Helimski 1997: 208) etc. A more complicated, but nevertheless possible case of the correspondence PFU *a ~ PS *ä may be involved in Finnic-Mordvin *koori ‘bark, skin’ (? < *kaxri) (UEW: 184) ~ ? Ob-Ugric *kur-5 ‘to skin; to shear’ (Honti 1982: 155–156) ~ ? PS *kä(q)r ‘skin’ (SW: 64–65). It is noteworthy that most of the cases with the unexpected *ä involve an adjacent . However, it is not possible to assume that this is the sole conditioning factor: cf. PU *kala ‘fish’ > PS *kale, PU *cav na- ‘to beat, rub’ > PS *cana- etc. The reduction (and possible subsequent loss) of the second syllable vowel is also a potential conditioning factor, but this explanation is still contradicted by PS *kan- ‘to go away’ < PU *kani- and PS *salq- ‘to lighten’ < PU *sala-7 (instead of PS **kän-, **sälq-). In any case, the unexplained cases of the correspondence PS *ä ~ PFU *a are too numerous to be categorically dismissed. Yet, the Samoyedic vowel develop- ment remains to be explained. The increased number of Uralic etymologies in the corpus of the Samoyedic lexicon normalizes to a certain extent the relationship between Samoyedic and Finno-Ugric. The traditionally assumed genealogical dichotomy of the Uralic family between Samoyedic and Finno-Ugric has often been supported with lexical criteria, but it seems that the apparently small number of cognate lexemes common to Samoyedic and the various branches of Finno-Ugric partially resulted from research history. The lexica of most Samoyedic

50 New and Old Samoyed Etymologies languages have not been exhaustively documented, and the material has not been as thoroughly investigated as in other Uralic languages. However, the most remarkable gap in lexical correspondences between Samoyedic and Finno-Ugric is not found in the total number of shared lexemes, but rather in some individual words. There are suspiciously many lexemes that have cognates in all or almost all branches of Finno-Ugric, yet lack one in Samoyedic, e.g. PFU *käti ‘hand, arm’, *weri ‘blood’, *widimi ‘marrow’, *jäni ‘ice’, *süksi7 ‘autumn’ (cf. PS *uta ‘hand’, *kem8 ‘blood’, *kajma ‘marrow’, *ser8 ‘ice’, *erö8 ‘autumn’, all of unknown origin). The most striking case is perhaps the numeral system: Samoyedic and Finno- Ugric have only one certain cognate numeral with identical meaning, namely ‘two’ (the cognate of PFU *wiiti ‘five’ survives in Samoyedic in the meaning ‘ten’). The equation of PFU ?*se7 simi7 ‘seven’ and PS *sejtwq (? << *sejtmq << *sejtqmq) ‘id.’ seems also plausible, but the phonological irregularities involved in this comparison have not yet been solved (cf. the Tocharian loan etymology suggested for the Samoyedic item by Janhunen (1983), which according to Kallio (forthcoming) must be rejected for phonological reasons). While the numerals 3–6 are cognate in all the Finno-Ugric languages (PFU *kolmi, *neljä7 , *wiiti, *kutti), Samoyedic shows a different set (PS *näkur, *tettq, *sqmpqlenkq, *mqktut). Janhunen (2000: 60–61) suggests that the PU numeral system did not contain these items at all and that they were later created in Samoyedic and Finno-Ugric independently of each other. But this is not the only possible explanation of the situation. First, at least the Samoyedic numerals ‘3’, ‘4’ and ‘6’ can hardly be offered an autochthonous etymology, unless some sort of Urschöpfung is invoked (according to Janhunen (1998: 476), ‘3’ and ‘6’ are possibly “derivative[s] with unidentified elements”, which only confirms the lack of such an explanation). Secondly, it is difficult to see how the PU numeral system could only have contained words for ‘2’, ‘5 or 10’ and ‘7’. On these grounds, it seems more likely that Samoyedic replaced the Uralic numerals with loanwords from an unknown language. This hypothesis is supported by the fact that some of the presumably innovative PS numerals show an obviously non-Uralic phonotactic structure (at least *näkur, *mqktut, probably also *tettq, as the PS geminate stops are unetymological). In general, the Proto-Samoyed lexicon as reconstructed by Janhunen (SW) contains so many roots of unknown origin that it seems reasonable to assume that Samoyedic has replaced much of its inherited vocabulary with borrowings from at least one disappeared non-Uralic language. The same

51 Ante Aikio kind of situation is quite clearly traceable in the opposite periphery of Uralic, too: as much as 43 % of the Proto-Sami lexicon is of unknown origin (Lehtiranta 1989: 8). The low amount of cognate lexemes between Finno- Ugric and Samoyedic has often been used as an argument in favor of the traditional dichotomy of the Uralic family (e.g. Janhunen 1981: 219). However, the argument is not conclusive because the rate of vocabulary loss in languages need not be constant, and on the other hand, family-internal borrowings cannot always be distinguished from cognate items. For this reason, lexical correspondences alone are weak criteria for genealogical classification. It can be concluded that the small number of cognate lexemes between Samoyedic and the branches of Finno-Ugric has probably resulted both from research history and from extensive replacement of Uralic vocabulary in Samoyedic. A deeper genetic dichotomy between Samoyedic and Finno- Ugric is a less likely explanation, when taking into account that the morpho- logical and phonological isoglosses supporting such a classification are almost non-existent and that there are also innovations common to Samoyedic and Ugric: these branches show a similar reorganization of the sibilant system (PU *s7 > *s, while PU *sv , *s > *s (later *s > *t in Mansi and Samoyedic)) and the lowering of *o to *a under certain conditions (Sammallahti (1988: 484, 500) states the conditions for Ugric and Samoyedic somewhat differently, but they are essentially identical). Recent developments in research into early Indo-European loanwords in Uralic have also normalized the position of Samoyedic in the Uralic language family. Previously it was supposed that (Proto-)Indo-European loanwords only occur in Finno-Ugric and not in Samoyedic, but this view does not hold anymore: Proto-Indo-European loanwords were already adopted into Proto- Uralic (see e.g. Koivulehto 1991, 1999).11 While the later strata of Indo- European loanwords seem to show a strong western bias, this picture may also have been influenced by research history. Janhunen (1983) suggests that at least some borrowings from Aryan or Iranic have been independently adopted into Proto-Samoyed, and it would be necessary to thoroughly investigate the Samoyedic lexicon to determine if Pre-Samoyedic has adopted Indo-European loanwords independently of the more western branches of Uralic. There are some promising cases, e.g. PS *wata- ‘to grow (tr.)’ (SW: 172) ? < Pre-Samoyedic *waksa- < Proto-Aryan *waks-v ‘to grow (intr.)’, *waks-aya-v ‘to grow (tr.)’ (IEW: 84–85). However, a substantial amount of such etymologies would have to be presented to warrant a conclusion.

52 New and Old Samoyed Etymologies

Notes

* I wish to thank Petri Kallio, Pekka Sammallahti and Juha Janhunen for their valuable remarks on an earlier version of this paper. 1 Janhunen (1981: 228) derives the labial vowel in Samoyedic passive-reflexive verbs from a suffix *j, and referring to Janhunen, Kulonen (1989: 53–54) suggests that a Proto-Uralic passive suffix *j can be reconstructed. This explanation is not compat- ible with the Proto-Samoyed non-initial syllable vowel system as reconstructed by Helimski. As the element *j in Sami passive suffixes has also been explained otherwise (Sammallahti 1998: 84–85), there is no evidence for a PU passive suffix of the shape *j. 2 Lehtisalo (1956: 103) also gives the translation ‘Nähbrett’, the exact meaning of which is unclear in the light of the given usage examples. 3 Instead, Finnish kerä is etymologically linked with kerätä ‘to gather, collect’ (Häkkinen 1987 s.v.), and the opposite view taken by SSA (s.v.) is not justifiable (cf. also UEW: 149, where kerätä is incorrectly linked with kerjätä ‘to beg’). Further- more, the whole word family is probably of Indo-European origin, cf. PIE *ger(e)- ‘to gather, collect’ (IEW: 382–383). 4 The Finnic cognates of PU *käwdi (Finnish köysi, Karelian keysi, Estonian köis etc.) presuppose PFU *kewdi, but the representation is best explained as irregular, as all the other Finno-Ugric cognates point unambiguously to PFU *ä in the first syllable. 5 Another possible counterexample to the assumed development *w > Ø in Samoyedic should be mentioned here. It is possible that Selkup cü-v ‘to shoot’ (Erdélyi 1970: 243) derives from PS *jiw- and further PU *lewi- ‘to throw, shoot’; cf. Finnish lyödä ‘to strike’, Hungarian lo2 (löv-) ‘to shoot’ etc. < PFU *lewi- (UEW: 247; Sammallahti 1988: 545 reconstructs PFU *lexi-). However, the phonological details of this etymology are not clear, as the word has not been preserved in the other Samoyedic languages. The PFU form of the verb could also be reconstructed as *lüwi- or *lüxi-, in which case the vowel correspondence would have a parallel in PFU *üwä ‘belt’ (> Sami avvi, Finnish vyö, Hungarian öv) ~ Selkup cüv id. (? < PS *jiw) (SSA s.v. vyö; cf. UEW: 575, which incorrectly connects the Finno-Ugric words with the reflexes of PS *wine ‘strap’ (sic!)). However, the PS reconstruction of the word for ‘belt’ is highly problematic (see SW: 102; cf. Janhunen 1981: 260, who reconstructs PS *ni7 qj ~ *jiqj on the basis of Nganasan and Enets, and derives this form from a problematic PU reconstruct *üxji). Note that Selkup -ü- cannot be derived from PS *ü in these two cases, because even though PS *j- generally became c-v in Selkup, it was lost before PS *ü (see SW: 49–50 for examples), and thus PS *jü- would give Selkup **ü- instead of cü-v . A reconstruction of PS *jiw- ‘to shoot’ and *jiw(e) ‘belt’ would account for the representation in Selkup: cf. Selkup süv ‘snake’ < PS *kiwe (SW: 72). However, the equation of Selkup cü-v and PFU *lewi- must be regarded as tentative as long as no clear solution to the phonetic difficulties can be pointed out. 6 Note also North Sami njálggi-s, South Sami njaelkie ‘tasty, sweet’ etc. < Proto-Sami *n7 alk5 e5 < *näl-kä7 , which can be explained as a verbal adjective based on PU *näxli-7 . The derivative is also attested in Finnic, but with a different meaning: Finnish nälkä ‘hunger’ etc. (Sami nealgi ‘hunger’ is a Finnic loanword). (Cf. SSA s.v. and Sammallahti 1998: 256, 257.)

53 Ante Aikio

7 An Indo-European loan etymology has also been proposed to PFU *jixi-{ (*juxi-) ‘to drink’. Koivulehto (1991: 17–18) has, with some reservations, suggested that this word derives from Pre-Aryan *gu-7 g7 hew- / *gu-7 g7 hu- ‘to pour (especially a drink offering)’. As for the semantics, it should be noted that there are many derivatives of this word in Finnic which suggest that the original meaning was rather ‘to flow’, ‘to pour’, etc. (cf. SSA s.v. juoma, juopa, juosta, juottaa). This loan etymology does not accord with the new evidence from Samoyedic, according to which the original PU form must be reconstructed as *jixi-{ . Koivulehto (personal communication) points out that the only possibility to save the loan etymology would be to assume a very early borrowing form the oldest reconstructed Indo-European form *g7 hi-g7 hu-, where the assimilation *i > *u in the reduplicated syllable and Grassman’s law (the loss of aspiration in a stop that precedes another aspirated stop) had not yet taken place. In this case, one has to assume that PU *j- was substituted for PIE initial *g7 h- in very early borrowings, because PU *x which was the substitute in intervocalic position (cf. PU *wexi- ~ *wixi- ‘to take, transport’ < PIE *weg7 h-) did not occur word initially. PU *ji{ - instead of *ji- can be explained by the fact that PU *ji- is not attested, and may thus have been phonotactically illegal (unless its absence is an accidental gap in the reconstruction). This is, as Koivulehto admits, a rather hypothetical explanation, because no parallel examples of the substitution are known. 8 There seem to be two exceptions. PS *täjwa ‘nail, wedge’ (> Nenets tyíwa, Enets tea, Mator täjbä) (SW: 154; Helimski 1997: 353–354) is probably cognate with Mordvin tulo, Komi and Udmurt tul ‘nail, wedge’ (cf. UEW: 797–798). The PU form can be reconstructed as *tolwa, and an irregular development PU *o > *a >> PS *ä before the tautosyllabic palatal *j (< PU *l) must be assumed. Note the homonymous PU verb *tolwa- ‘to bring / take’ which shows regular development in Samoyedic (see 1.16.). A parallel case for the irregular development apparently occurs in PS *wäjk ‘shoul- der’ (> Selkup qeq8 ), (der.) *wäjk-kq ‘neck’ (SW: 173), which can be compared to PFU *wolka ‘shoulder’ (UEW: 581). For the irregular loss of the final low vowel in Samoyedic, cf. PS *pqn- ‘to weave’ (? < *pqnq-) << PU *puna- (cf. Janhunen 1981: 258) and PS *pit- ‘to tan’ (? < *pitq-) << PU *peksä- ‘to beat’ (UEW: 368–369; this etymology has been rejected by Janhunen (1981) and Sammallahti (1988: 553), even though the irregularity is quite small). 9 Finnish voi ‘butter’ < PFU *wooji (UEW: 578–579) is an apparent exception, but it has been explained otherwise (Korhonen 1981: 131). 10 Note that PFU (PU?) *ani ‘mouth, opening’ and *ana- ‘to open, take off’ are obviously etymologically linked, even though their morphological relationship is unclear. Similar but reverse cases of such correlative derivatives occur in PFU *luki- ‘to count’ ~ *luka ‘ten’ and PFU *puni-7 ‘to twist’ ~ *puna7 ‘twist’ (possible irregular cognates in Samoyedic: PS *pqj- ‘to twist’, *pqjq ‘bent, steep’; SW: 112–113, 115), perhaps also PFU *suki-7 ‘to sharpen’ (> Sami cohkatv , Mordvin suvoms7 ) ? ~ PFU *suka7 ‘sharp ?’ (> ‘comb’, cf. Finnish suka ‘currycomb’). It is noteworthy that the Samic lexicon contains a large number of cases of this type, e.g. North Sami roggi ‘hole, pit’ ~ rogga- ‘to dig’, caggi ‘prop, support’ ~ cagga- ‘to prop up’, and this model of word-formation is still semi-productive at least in North Sami. The origin of the correlative derivatives in Samic is still unresolved. The cases *ani/a- and *luki/a- suggest that the phenomenon may be a direct inheritance from Proto-Uralic; however,

54 New and Old Samoyed Etymologies

a thorough examination of its possible traces elsewhere in Uralic would be needed. 11 A new Proto-Indo-European borrowing the distribution of which reaches Samoyedic can be tentatively proposed: PU *kaja ‘?sun, ?brightness, ?dawn’ (SSA s.v. kajastaa; cf. UEW: 167, where the Samoyed cognates are incorrectly connected with Finnish

koi ‘dawn’ etc.) can be derived from PIE *h2ay-en- / *h2ay-er- ‘day, morning’ (IEW: 12; Kluge 1995 s.v. eher), cf. Greek eri-, aiiar- ‘day’, Gothic air ‘early’. The Indo-European heteroclitic has no reflex in Uralic, as expected on account of the

already known loanwords: cf. Finnic-Samic-Mordvin *kesä ‘summer’ < PIE *h1es-

en- / h1es-er- ‘summer, autumn’, Finnic-Mordvin *uni ‘dream’ < PIE *on-en- / *on- er- id., PU *weti ‘water’ < PIE *wed-en- / *wed-er- id. (Koivulehto 1991: 36–40, 107– 108; 1999: 210).

References

Abondolo, Daniel 1996: Vowel rotation in Uralic – obug[r]ocentric evidence. School of Slavonic and East European Studies Occasional Papers 31. London. –––– (ed.) 1998: The Uralic languages. London – New York. Bergsland, Knut – Mattsson Magga, Lajla 1993: Åarjelsaemien-daaroen baakoegærja – Sydsamisk-norsk ordbok. [Lakselv.] Buck, Carl 1965: A dictionary of selected synonyms in the principal Indo-European languages. Chicago – London. Collinder, Björn 1977: Fenno-Ugric vocabulary. Second, revised edition. Ham- burg. Erdélyi, István 1970: Selkupisches Wörterverzeichnis – Tas-Dialekt. The Hague. Grundström, Harald 1946–1954: Lulelappisches Wörterbuch. Skrifter utgivna genom dialekt- och folkminnesarkivet i Uppsala, ser. C:1. Helimski, Eugene 1983: The Language of the First Selkup Books. Studia Uralo- Altaica 22. –––– 1986: Etymologica 1–48 (материалы по этимологии маторско-тайгийско- карагасского языка). – Nyelvtudományi Közlemények 88: 119–143. –––– 1991: On the interaction of Mator with Turkic, Mongolic and Tungusic: a rejoinder. – JSFOu 83: 257–267. –––– 1997: Die Matorische Sprache. Studia Uralo-Altaica 41. Szeged. –––– 1999a: Ural. *läs “bei, in der Nähe” im Samojedischen (und im Ungarischen?). – Finnisch-Ugrische Mitteilungen 21/22: 77–81. –––– 1999b: Basic Vocabulary in PU and PFU – Remarks on Etymology and Reconstruction. A handout delivered at the Budapest Uralic Workshop 2: Etymology, September 1999, organized by the Nyelvtudományi Intézet, Budapest. Honti, László 1982: Geschichte des obugrischen Vokalismus der ersten Silbe. Budapest. Häkkinen, Kaisa 1987: Nykysuomen sanakirja 6: Etymologinen sanakirja. Por- voo – Helsinki – Juva. IEW = Pokorny, Julius 1959: Indogermanisches etymologisches Wörterbuch. Bern – München. Itkonen, Erkki 1954: Zur Geschichte des Vokalismus der ersten Silbe im

55 Ante Aikio

Tscheremissischen und in den permischen Sprachen. – FUF 31: 149–345. –––– 1960: Lappische Chrestomathie mit grammatikalischem Abriss und Wörter- verzeichnis. Hilfsmittel für das Studium der finnisch-ugrischen Sprachen VII. Hel- sinki. Itkonen, Terho 1970: Etsimisen ja löytämisen alalta. – Sananjalka 12: 5–17. [Reprinted in: Terho Itkonen (1993), Aloja ja aiheita – valikoima kolmen kymmen- luvun tutkielmia. MSFOu 216: 71–83.] Janhunen, Juha 1981: Uralilaisen kantakielen sanastosta. – JSFOu 77: 219–274. –––– 1983: On early Indo-European–Samoyed contacts. – MSFOu 185: 115–127. –––– 1992: Petrified verbal nouns in early Uralic. – In: Pál Deréky, Marianne Sz. Bakró-Nagy, Timothy Riese & Péter Hajdú (eds.), Festschrift für Károly Rédei zum 60. Geburtstag – Emlékkönyv Rédei Károly 60. születésnapjára. Studia Uralica 6: 239–244. Wien – Budapest. –––– 1998: Samoyedic. – In: Daniel Abondolo (ed.) 1998: 457–479. –––– 2000: Reconstructing Pre-Proto-Uralic typology spanning the millennia of linguistic evolution. – Congressus Nonus Internationalis Fennougristarum I: 59–76. Kallio, Petri 2001: Germanic ‘maggot’ and ‘moth’. – In: Dirk Boutkan & Arend Quak (eds.), Language Contact – Substratum, Superstratum, Adstratum in Germanic Languages. Amsterdamer Beiträge zur älteren Germanistik 54, p. 117–122. –––– (forthcoming): Prehistoric Contacts between Indo-European and Uralic. (The oral version of this paper was read at the 13th Annual UCLA Indo-European Conference, University of California at Los Angeles, 9.11.2001.) Kangasmaa-Minn, Eeva 1998: Mari. – In: Daniel Abondolo (ed.) 1998: 219– 248. Katzschmann, Michael – Pusztay, János 1978: Jenissej-Samojedisches (Enzisches) Wörterverzeichnis. Fenno-Ugrica 5. Hamburg. Keeley, Lawrence H. 1996: War before civilization. New York. Klein, Ernest 1966–1967: A Comprehensive Etymological Dictionary of the . Amsterdam – London – New York. Kluge, Friedrich 1995: Etymologisches Wörterbuch der Deutschen Sprache. Bearbeitet von Elmar Seebold. 23., erweiterte Auflage. Berlin – New York. Koivulehto, Jorma 1976: Vanhimmista germaanisista lainakosketuksista ja nii- den ikäämisestä II. – Virittäjä 80: 247–290. –––– 1983: Suomalaisten maahanmuutto indoeurooppalaisten lainasanojen valossa. – JSFOu 78: 107–132. –––– 1988: Lapin ja itämerensuomen suhteesta – ieur. -Tr- yhtymän korvautuminen lainoissa. – Virittäjä 92: 26–51. –––– 1991: Uralische Evidenz für die Laryngaltheorie. Wien: Verlag der österreichischen Akademie der Wissenschaften. –––– 1996: Kuoleman ja elämän sanoja. – Virittäjä 100: 322–339. –––– 1999: Varhaiset indoeurooppalaiskontaktit: aika ja paikka lainasanojen valossa. – In: Paul Fogelberg (ed.), Pohjan poluilla. Suomalaisten juuret nykytutkimuksen mukaan. Bidrag till Kännedom av Natur och Folk 153: 207–236. Koponen, Eino 1996: Lappische Lehnwörter im Finnischen und Karelischen. – In: Lars-Gunnar Larsson (ed.), Lapponica et Uralica. 100 Jahre finnisch-ugrischer Unterricht an der Universität Uppsala. Vorträge am Jubiläumssymposium 20.–23.

56 New and Old Samoyed Etymologies

April 1994. Studia Uralica Upsaliensia: 83–98. Korhonen, Mikko 1981: Johdatus lapin kielen historiaan. Helsinki. –––– 1986: History of the Uralic languages and the principle of lateral areas. – FUF 47: 156–162. [Reprinted in: Tapani Salminen (ed.) 1996, Typological and historical studies by Mikko Korhonen – a memorial volume published on the 60th anniversary of his birth. MSFOu 223: 213–218.] Kulonen, Ulla-Maija 1989: The passive in Ob-Ugrian. MSFOu 203. –––– 1995: Etymologie, Lautgeschichte, Herkunft. – Congressus Octavus Internationalis Fennougristarum 5: 88–91. Lehtiranta, Juhani 1989: Yhteissaamelainen sanasto. MSFOu 200. Lehtisalo, Toivo 1956: Juraksamojedisches Wörterbuch. LSFU XIII. LÄGLOS = A. D. Kylstra, Sirkka-Liisa Hahmo, Tette Hofstra & Osmo Nikkilä 1991–: Lexikon der älteren germanischen Lehnwörter in den ostseefinnischen Sprachen. Amsterdam – Atlanta. Munkácsi – Kálmán 1986 = Wogulisches Wörterbuch. Gesammelt von Bernát Munkácsi. Geordnet, bearbeitet und herausgegeben von Béla Kálmán. Budapest. Nielsen, Konrad 1979 [1932–1962]: Lappisk (samisk) ordbok. Oslo. Paasonen, Heikki 1917: Beiträge zur finnischugrisch-samojedischen Laut- geschichte. Budapest. Papp, István 1968: Unkarin kielen historia. Tietolipas 54. Helsinki. Qvigstad, J. K. 1893: Nordische Lehnwörter im Lappischen. Christiania. Rédei, Károly 1998: Suomen tanner. – In: Riho Grünthal & Johanna Laakso (eds.), Oekeeta asijoo. Commentationes Fenno-Ugricae in honorem Seppo Suhonen sexagenerii 16.V.1998. MSFOu 228: 384–385. Salminen, Tapani 1997: Tundra Nenets . MSFOu 227. –––– 1998: A Morphological Dictionary of Tundra Nenets. LSFU XXVI. Sammallahti, Pekka 1979: Über die Laut- und Morphemstruktur der Uralischen Grundsprache. – FUF 43: 22–66. –––– 1988: Historical phonology of the Uralic languages with special reference to Samoyed, Ugric, and Permic. – In: Denis Sinor (ed.), The Uralic languages – description, history, and foreign influences, 478–554. Leiden. –––– 1998: The Saami languages – an introduction. Kárásjohka.v Schrijver, Peter 1997: Animal, vegetable and mineral: some western European substratum words. – In: Alexander Lubotsky (ed.), Sound law and analogy. Papers in honor of Robert S.P. Beekes on the occasion of his 60th birthday. Leiden Studies in Indo-European 9: 293–316. Amsterdam – Atlanta. Schlachter, Wolfgang 1958: Wörterbuch des Waldlappendialekts von Malå und Texte zur Ethnographie. LSFU XIV. SKES = Suomen kielen etymologinen sanakirja. LSFU XII. 1955–1981. SSA = Suomen sanojen alkuperä. Etymologinen sanakirja. Helsinki. 1992–2000. SW = Janhunen, Juha 1977: Samojedischer Wortschatz. Gemeinsamojedische Etymologien. Castrenianumin toimitteita 17. Tereshchenko 1965 = Терещенко, Н. М.: Ненецко-русский словарь. Москва. UEW = Rédei, Károly 1988: Uralisches Etymologisches Wörterbuch. Wiesbaden.

57