Proto-Uralic

Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Luobbal Sámmol Sámmol Ánte (Ante Aikio) 1. Proto-Uralic 1.1 Introduction The Uralic languages form one of the most firmly established and thoroughly studied ancient languages families in the world. Uralic comparative linguistics is recognized as a highly advanced field of research, surpassed in breadth and depth of study only by the Indo-European family (Campbell 1998a: 164–165). The long and productive history of research has established a solid overall picture of the history of the language family, including a partial reconstruction of the Uralic proto-language described in this chapter. Despite the breadth of research, the reconstruction of Proto-Uralic remains much more fragmentary than that of Proto-Indo-European and many aspects of the proto- language remain poorly understood and open to debate. This is due not only to the shallow philological records but also to the uneven state of research between different levels of language: phonological and lexical reconstruction have received the most attention, whereas the comparative study of morphology has been less systematic and methodologically less advanced, and research in diachronic syntax has been scarce indeed. 1.2 The structure of the Uralic language family Comparative Uralic linguistics operates on two principal levels of reconstruction, which could be called the ‘shallow’ and the ‘deep’ level. On the shallow level the language family consists of nine obvious branches whose statuses as separate taxonomic entities are beyond doubt: 1) Saami, 2) Finnic, 3) Mordvin, 4) Mari, 5) Permic, 6) Mansi, 7) Khanty, 8) Hungarian, and 9) Samoyed. In terms of their original core areas these branches can be organized into a rough geographical continuum with main orientation along the east-west axis (Figure 1.1); Hungarian is a geographical outlier which despite its present location in the Pannonian basin is originally most closely associated with the Mansi and Khanty branches. The branches show far- reaching differences from each other on all levels of language, and no transitional 1  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   varieties between the branches exist. The nine branches form the primary level of linguistic reconstruction in the Uralic family. Eight of them consist of relatively closely related languages or linguistic varieties on the basis of which branch-level proto-languages can be reconstructed in much detail; here, too, the exception is Hungarian which is a single language. Figure 1.1 The branches of the Uralic family in an approximate geographical order along the east-west axis (slightly modified from Salminen 1999: 20) The deep level of comparative Uralic linguistics concerns the more remote genetic connections between the nine branches and the reconstruction of Proto-Uralic. Knowledge regarding the latter mainly derives from comparison of the intermediate branch-level proto-languages, and thus forms a secondary level of reconstruction. Consensus is lacking regarding the taxonomy of the Uralic languages on the deep level, and the hierarchy of genetic relationships between the nine branches is an issue of major disagreement. This state of affairs has implications for the reconstruction of Proto-Uralic. As there are conflicting views regarding what the primary branches are, there are also conflicting views regarding what exactly can be reconstructed to Proto- Uralic. According to a traditional model the taxonomic structure of the family can be described by a mostly binarily branching tree scheme in which the time depth of the intermediate subgroups tends to decrease toward the west (Figure 1.2). This model was the received view in Uralic studies until the 1980s and has been applied in standard works such as etymological dictionaries (e.g., Rédei 1988–1991) and the important studies on Uralic historical phonology by Janhunen (1981) and Sammallahti (1988). Since then, however, the model has met with increasing criticism (e.g., K. Häkkinen 1983; Salminen 2001, 2002), and at present it has few active proponents (see, however, Janhunen 2001b; 2009). 2  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Uralic Finno-Ugric Finno-Permic Ugric Finno-Volgaic Finno-Saamic Ob-Ugric Saami Finnic Mordvin Mari Permic Hungarian Mansi Khanty Samoyed   Figure 1.2. The taxonomical structure of the Uralic language family according to the view commonly held until the 1980s, but now widely contested The main problem in the traditional taxonomic scheme is that the intermediate nodes between Proto-Uralic and the nine low-level branches are not supported by sufficient evidence. Proponents of this model have invoked lexicostatistic arguments, but evidence in form of sound changes (Sammallahti 1988) remains both limited and ambiguous (Salminen 2002). On the other hand, various alternative models of Uralic taxonomy have been proposed (e.g., Michalove 2002; J. Häkkinen 2009; Syrjänen et al. 2013), but they are based on widely varying methods and data and none of them enjoys wide support. Such models may also include unorthodox subgroups, such as the grouping of “Ugric” and Samoyed together into an “East Uralic” node by J. Häkkinen (2009); these, however, are not backed up by more systematic or convincing evidence than the traditional subgroups (Zhivlov 2018). Therefore, the present results of Uralic taxonomic research hardly warrant other than an agnostic stance. The most crucial unresolved taxonomic question concerns the supposed primary divergence between Samoyed and a “Finno-Ugric” node comprising all the other Uralic languages. The low number of Uralic word stems in the Samoyed lexicon might indicate an early divergence from the rest of the family. However, the “deviant” 3  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   character of the Samoyed lexicon has also been exaggerated, and it partly stems from paucity of research and documentation. As the binary division of the family into Samoyed and ‘Finno-Ugric’ is under serious doubt and no well-argued alternative taxonomy exists, it remains unclear in which branches cognate linguistic material must be attested in order to qualify as Proto-Uralic. The present chapter will employ a somewhat ad hoc practical solution: a feature is considered definitely to derive from Proto-Uralic if the distance between its attested reflexes reaches at least either 1) from Samoyed to Permic, or 2) from “Ugric” to Mordvin. Thus, in the present framework the traditional concept of “Proto- Finno-Ugric” is essentially synonymous with Proto-Uralic. 1.3 Phonology 1.3.1 State of research Phonological reconstruction has for long formed the hard core of Uralic comparative linguistics, and therefore phonology is the most comprehensively reconstructed level of structure in Proto-Uralic. Research into the westernmost Uralic languages made major advances already well over a hundred years ago already; for instance, the complicated system of regular vowel correspondences between Finnic, Saami and Mordvin was in large part worked out by Genetz (1896). Since then, research slowly progressed towards the more eastern branches; highly influential works in the field include, e.g., E. Itkonen (1956) on Mari and Permic, Lytkin (1964) on Permic, Honti (1982) on Khanty and Mansi, and Janhunen (1976; 1977) on Samoyed. The present understanding of Proto-Uralic phonology is largely congruous with the groundbreaking studies by Janhunen (1981) and Sammallahti (1988); these two papers combined a synthesis of previous studies on historical phonology with a critical reappraisal of the etymological corpus (see 1.6.1), and for the first time outlined a coherent theory of Uralic historical phonology that integrated all branches of the family. The description of Proto-Uralic phonology that follows largely conforms to the picture established by Janhunen and Sammallahti, but revisions on some details have been made by subsequent research. 1.3.2 Phoneme inventory Eight vowel phonemes have been reconstructed to Proto-Uralic (see Table 1.1), which represents a “large vowel inventory” according to the WALS framework (Maddieson 2013a). The phonological oppositions between the vowels can be described in terms 4  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   of backness, height and roundedness. The sets of front and back vowels were symmetrical: each contained three cardinal vowels contrasting in height, and an additional non-cardinal vowel differentiated from the corresponding cardinal vowel by its value of roundedness. The last feature is typologically notable, as vowel oppositions based on roundedness are not very common. Moreover, front rounded vowels occur predominantly in languages of northern Eurasia, many of which also have vowel harmony (Maddieson 2013b), a feature also present in Proto-Uralic (see 1.3.3, 1.3.5).1   FRONT BACK CARDINAL NON - NON - CARDINAL CARDINAL CARDINAL CLOSE i   ü   i̮   u   MID e       o   OPEN ä       a   Table 1.1 The Proto-Uralic vowel inventory. The approximate phonetic qualities of most vowel phonemes can be quite unambiguously established, but the non-cardinal back vowel *i̮ remains unclear in this regard. Janhunen (1981) and Sammallahti (1988) interpret it as a close vowel (≈ IPA [ɯ] ~ [ɨ]), but it could instead have belonged to the series of mid vowels, and in terms of phonetic quality even open-mid realizations cannot be ruled out (≈ IPA [ɤ] ~ [ɘ] ~ [ʌ] ~ [ɜ]). This is suggested by the tendency of *i̮ to merge with *a in various branches, as well as by the fact that *i̮ appears as a substitute for foreign *a in a couple of old loanwords (e.g., *śi̮ ta ‘hundred’ and *pi̮ ŋka ‘psychedelic mushroom’ from Proto-Indo-Iranian *ćatam- ‘hundred’ and *bhanga- ‘a drug plant’). In any case, *i̮ can be described as a non-open unrounded back vowel. As regards the open back vowel *a, Janhunen (1981) and Sammallahti (1988) reconstruct it as a rounded vowel (*å ≈ IPA [ɒ]), but this solution appears less likely in typological terms and is not necessitated by the data. At any rate, roundedness did not function as a phonologically distinctive feature in the case of *a, so the question is inconsequential in regard to the phonological system. 1 For the reconstructed protolanguages, italicized FUT characters will be used; see chapter 6.7. 5  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   The theory of eight Proto-Uralic vowel phonemes has been very successful in explaining the complicated vowel correspondences between the Uralic branches. These correspondences are too complex to be discussed in detail here, but some general tendencies of vowel development can be noted. The Finnic branch is well- known for its conservative vowel system, and indeed, it appears almost bizarrely archaic that modern Finnic languages often preserve the quality of Proto-Uralic vowels as such (as in Finnish käsi ‘hand’, olka ‘shoulder’ and silmä ‘eye’ from PU *käti, *wolka and *śilmä, respectively). However, also Mordvin and Samoyed show relatively conservative vowel systems. Even Saami vowels derive highly consistently and regularly from Proto-Uralic, despite the fact that this branch has restructured its vowel system in a most radical way; the situation in Khanty and Mansi appears to be similar, even though the issue has been less studied. The development of vowels in Mari, Permic and Hungarian remains somewhat less clear: main outlines have been established, but numerous details beg further research. As regards the consonant inventory of Proto-Uralic, 16 consonant phonemes can be unequivocally established, and two more can be reconstructed with a fair degree of certainty (Table 1.2). Overall, the Proto-Uralic consonant system was fairly simple and represents a “moderately small” inventory according to the WALS classification (Maddieson 2013c). Voicing does not appear to have been a distinctive feature. In phonological terms, all obstruents were unvoiced and all sonorants were voiced; allophonic variation in voicing may have occurred, of course. 6  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   CORONAL DORSAL ALVEOLO-PALATAL POSTALVEOLAR ALVEOLAR PALATAL LABIAL VELAR STOP p t k OBSTRUENT AFFRICATE č SIBILANT s ś ?š NASAL m n ń ŋ LATERAL l SONORANT RHOTIC r GLIDE w j (UNCLEAR) d ď ?x Table 1.2 The Proto-Uralic consonant inventory. Each reconstructed consonant can be established as an independent phoneme on the basis of distinct patterns of regular sound correspondences between the daughter branches of Uralic. Table 1.3 shows the regular consonant correspondences in word- initial position and in intervocalic position following a stressed (first-syllable) vowel. Note that in most cases where multiple reflexes occur conditioning factors have been established; some minor conditioned developments that only concern highly specific environments have been ignored. For most consonant phonemes also their place and manner of articulation can be quite unambiguously reconstructed, but some unresolved questions regarding phonological status and phonetic realizations remain. 7  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   PU Saami Finnic Mordvin Mari Permic Hung. Mansi Khanty Samoyed *p   *p   *p   *p   *p   *p   f   *p   *p   *p   *t   *t   *t   *t,  *ť   *t   *t   t   *t   *t   *t   WORD-­‐INITIAL  CONSONANTS   *k   *k   *k   *k   *k   *k   k,  h   *k   *k   *k   *č   *c   *t   *č   *č   *č   ?   *š   *č   *č   *s   *s   *s   *s   *s   *s   Ø   *t   *ʟ   *t   *ś   *ć   *s   *ś   *š   *ś   s  <sz>   *s.  *š   *s   *s   *š   *s   *h   *č   *š   *š   Ø   *t   *ʟ   ?   *m   *m   *m   *m   *m   *m   m   *m   *m   *m   *n   *n   *n   *n,  *ń   *n   *n   n   *n   *n   *n   *ń   *ń   *n   *n,  *ń   *n   *ń   nʲ  <ny>   *ń   *ń   *ń   *ď   *ϑ   *t   *l,  *ľ   *l   *ľ   ?   *ľ   *j   *j   *l   *l   *l   *l,  *ľ   *l   *l   l   *l   *l   *j,  *l   *w   *v,  Ø   *v,  Ø   *v,  Ø   *w,  Ø   *v   v   *w   *w   *w   *j   *j,  Ø   *j,  Ø   *j,  Ø   *j,  Ø   *j   j,  Ø   *j   *j   *j   *p   *p   *p   *v   *w,  Ø   Ø   ?   *p   *p   *p   *t   *t   *t   *d,  *ď   *d   Ø   z   *t   *t   *t   *k   *k   *k   *v,  *j   Ø   Ø   v,  Ø   *ɣ,  *w   *ɣ   *k,  Ø   *č   *c   *t   *č   *č   *ǯ,  *ž   ?   *š   *č   *č   INTERVOCALIC  CONSONANTS   *s   *s   *s   *z   *z   *z   s  <sz>?   *t   *l   *t   *ś   *ć   *s   *ź   *ž   *ź   s  <sz>   *s,  *š   *s   *s   *š   *s   *h   *ž   *ž   *ž   ?   *t?   *l?   ?   *m   *m   *m   *m   *m   *m   m,  v   *m   *m   *m   *n   *n   *n   *n,  *ń   *n,  Ø   *n   n   *n   *n   *n   *ń   *ń   *n   *ń   *ń?   *ń   nʲ  <ny>   *ń   *ń   *ń   *ŋ   *ŋ   *v,  Ø   *ŋ   *ŋ,  Ø   *ŋ,  Ø   g,  Ø   *w,  *ŋk   *ŋ,  *ŋk   *ŋ   *d   *δ   *t   *d,  *ď   *d,  Ø   Ø   l   *l   *l   *r   *ď   *δ   *t   *d,  *ď   *d,  Ø   *ľ   ɟ  <gy>   *ľ   *j   *j   *l   *l   *l   *l,  *ľ   *l   *l   l   *l   *l   *l,  *j,  Ø   *r   *r   *r   *r,  *ŕ   *r   *r   r   *r   *r   *r   *w   *v   *v,  Ø   *v   Ø   Ø   v,  Ø   *w   *w,  *ɣ   Ø   *j   *j   *j,*i,Ø   *j   *j,  Ø   *j   j,  Ø   *j   *j   *j,  Ø   *x   *k   Ø   *j   Ø   Ø   v,  Ø   *ɣ,*w,*j   *ɣ,*w,*j   Ø   Table 1.3 The reflexes of Proto-Uralic word-initial and intervocalic consonants in the nine branches of the family. Question marks indicate cases where conclusive data is lacking. Three sibilants contrasting solely by their place of articulation are traditionally reconstructed for Proto-Uralic (*s : *ś : *š), but alternative interpretations recognizing two sibilants or even just one are also possible. The reconstruction of the postalveolar 8  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   sibilant *š remains problematic, as it is scarcely represented in the etymological material, and most of the proposed examples involve etymological problems such as deviant phonotactic patterns, irregular sound correspondences, and a distribution skewed toward the western part of the language family (Aikio 2015: 44–46). Janhunen (1981) and Sammallahti (1988) suggest that *š arose as a secondary innovation after the breaking up of Proto-Uralic, but nevertheless, a couple of well- behaved etymons with a wide distribution (such as *jäkši- ‘cold’, *šiŋir(i) ‘mouse’ and *kajši- ‘be sick’) suggest that *š nevertheless occurred as a low-frequency phoneme in Proto-Uralic. Another kind of problem is involved in the reconstruction of the alveolo-palatal sibilant *ś. While its status as an independent phoneme is not in doubt, its sibilant quality is less certain. Janhunen (2007a: 211) interprets it as a palatal stop; Zhivlov (2014: 114) reconstructs it as an alveolo-palatal affricate (*ć ≈ IPA [t͡ sʲ]), pointing out that its phonotactic distribution resembles that of the affricate *č rather than that of the sibilants *s and *š, and that Saami shows an affricate as its default reflex which could represent an archaism instead of an innovation. In support of this interpretation one can add that also a few other branches show occasional affricate reflexes, even though the default reflex is a sibilant: cf., e.g., Võro (Finnic) lats [lat͡ sʲ] ‘child’ < Proto-Uralic *li̮ pśi ‘cradle’ and dialectal Komi (Permic) [kod͡ zʲin] ‘dowry’ < PU *kaśa- ‘give (as a gift)’. The sound values of the two coronal consonants *d and *ď are particularly difficult to reconstruct, but at any rate they are not to be interpreted as voiced stops despite the notational practice deriving from Janhunen (1981) and Sammallahti (1988), and despite Janhunen’s (2007a: 212) suggestion that the two phonemes could have been “weak obstruents, which may or may not have involved a fricative pronunciation”. Judging from their phonotactic distribution, *d and *ď seem to have been sonorants: they occur in consonant clusters such as *dm, *dw and *ďw, but clusters of the type obstruent+sonorant seem to have been absent in Proto-Uralic (see 1.3.3). Traditional reconstructions have identified the two consonants as “spirants”, i.e. non-sibilant fricatives, and employed the symbols <δ> and <δ́>. As regards *d, it is indeed reflected as a voiced dental fricative (IPA [ð]) in several Saami varieties, but elsewhere its reflexes have merged with those of *t, *l or *r. The consonant *ď, on the other hand, appears to have been the palato-alveolar counterpart of *d: the two phonemes have merged in western branches (Saami, Finnic, Mordvin, Mari) but 9  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   elsewhere the reflexes of *ď are voiced alveolo-palatal and palatal consonants: an alveolo-palatal lateral [lʲ] (Permic, Mansi), a palatal glide [j] (Khanty, Samoyed), and a palatal stop [ɟ] (Hungarian). Considering its reflexes, *d might reasonably be reconstructed as a voiced dental fricative, but it seems impossible to apply this analysis to the alveolo-palatal consonant *ď. From the point of view of articulatory phonetics and phonological typology a “voiced palatalized dental fricative” (≈ IPA [ðʲ]) appears too implausible a consonant phoneme to be reconstructed, as has been noted by Janhunen (2007a: 212). Moreover, such a phoneme would also be an odd member in the Proto-Uralic consonant inventory, which features no other oppositions based on a palatal coarticulatory gesture – the phonemes *ś and *ń being properly described as alveolo- palatal and not as “palatalized” consonants. On the other hand, it is perhaps not necessary to assume that *d and *ď shared the same manner of articulation. In fact, one phonotactic constraint suggests that they did not belong in the same category in the phonological system: *ď occurred in word-initial position, whereas *d did not (see 1.3.3). In any case, the question of the phonetic quality of the two consonants remains unresolved. The most enigmatic member of the Proto-Uralic phoneme inventory is the consonant that Janhunen (1981) and Sammallahti (1988) mark with *x (the symbol indicates its unknown phonetic quality, and is not to be confused with IPA [x], the unvoiced velar fricative). *x is mostly either reflected as a glide (IPA [v], [β], [j], [ɣ]) or completely lost, the exception being Saami where *x has merged with the velar stop *k. Considering these reflexes, *x can be tentatively identified as a velar consonant, perhaps a voiced velar fricative or glide (≈ IPA [ɣ]). A more extensive discussion on the issue is presented by Janhunen (2007a) who classifies *x to the broad group of “laryngeal” consonants. It has also been claimed that in loanwords *x occurs as substitute for the phonetically likewise ambiguous Proto-Indo-European laryngeals (Koivulehto 1991), but this remains debatable (see 1.6.3). The most obscure feature of *x is not its sound value, however, but its status in the phonological system. According to Janhunen’s (1981) original proposal *x only occurred in two specific phonological environments: in word stems of the shape *CVxi- and *CVxCi-. The latter stem type was reconstructed by Janhunen for those words in which a long vowel (*V̄ ) in Finnic corresponds to a vowel sequence *Vǝ in Samoyed. For these stem types, Janhunen assumed the vocalization of *x in coda position: e.g., Estonian keel ‘tongue, language’ (< Proto-Finnic *kēli) ~ Nganasan 10  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   /ɕiǝdʲǝ/ ‘tongue’ (< Proto-Samoyed *käǝ(-)jǝ) would reflect Proto-Uralic *käxli. However, recent analysis has revealed major problems in this interpretation and presented alternative explanations for both Finnic long vowels and Samoyed vowel sequences (Aikio 2012b). The situation with stems of the shape *CVxi- is different: in this environment it indeed appears necessary to postulate *x as an independent phoneme. However, only about ten word stems of this type can be reliably reconstructed. Despite the scarcity of material, these kinds of stems clearly display a unique pattern of sound correspondence (see Table 1.3) which necessitates the reconstruction of *x as a phoneme distinct from other consonants. This result is odd, however: it is difficult to see why the occurrence of any consonant phoneme should be restricted to such a specific environment as V_i. Moreover, considering the marginal position of *x in the phonological system it is counterintuitive that most examples of it are core items of basic vocabulary (such as *ji̮ xi- ‘drink’, *mexi- ‘give/sell’, *toxi- ‘bring/give’, *wixi- ‘take, transport’, *mi̮ xi ‘earth’). One is tempted to speculate that the consonant *x is an artefact of reconstruction and the correspondences exhibited by stems currently reconstructed as *CVxi- could be accounted for in some other way that remains to be discovered. Another possibility is that *x also occurred in other phonological environments which have not yet been identified. It should be noted that traditional reconstructions also assume some additional alveolo-palatal phonemes, most commonly the lateral *ľ and the affricate *ć; in traditional reconstructions the latter is considered a phoneme distinct from the “sibilant” *ś, whose possible affricate quality was discussed above. However, the evidence supporting the reconstruction of these consonants is weak, and even Sammallahti’s (1988: 490–491) cautious view on the issue seems to be too optimistic: the relevant material seems to consist of dubious etymologies and misanalyzed instances of *ď and *ś. 1.3.3 Phonotactics Proto-Uralic had a relatively simple syllable structure which imposed strict limitations on the distribution of consonants and vowels. Word-initial syllables had the form (C)V(C): the syllable nucleus consisted of a single vowel phoneme, and a single consonant phoneme could appear in both the syllable onset and the coda position. Non-initial syllables had the form CV(C); sequences of two heterosyllabic vowels 11  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   were apparently not allowed. There is some evidence suggesting that the semivowel *j had an exceptional status in that it could occur between the nuclear vowel and the coda consonant, producing the syllable structure (C)VjC. The same has been claimed to hold for the semivowel *w (Janhunen 1982: 25), but no clear evidence seems to be found in support of this. As the syllable nucleus always consisted of a single vowel, neither long vowels nor diphthongs could occur. Homosyllabic sequences of the type Vj and Vw did occur (as in *ojwa ‘head’, *täwdi ‘full’), but such sequences cannot be phonologically analyzed as diphthongs: the synchronic reflexes of these postvocalic glides still function as consonants in Saami, participating in consonant gradation (see 7.?.?). Further restrictions on vowel distribution were connected with word stress. In Proto-Uralic primary stress fell invariably on the initial syllable of the word, which due to the lack of prefixes also always was the initial syllable of a word stem. Only in this syllable the full inventory of eight vowels (see 1.3.2) was in use, while the following non-initial syllables featured a restricted vowel inventory (with the exception of words containing more than one word stem, i.e. compounds). According to the standard reconstruction the restricted inventory comprised only three phonemes: the open vowels *a and *ä, and a non-open unrounded vowel that has been varyingly reconstructed as *i, *e and *ǝ; the first option is used in the present paper. Furthermore, the distribution of *a and *ä in non-initial syllables was governed by vowel harmony: *a occurred after initial-syllable back vowels, and *ä after initial- syllable front vowels (as in *pala ‘bit’, *pälä ‘half’). Vowel harmony also resulted in morphophonological alternations in the shape of suffixes containing an open vowel (see 1.3.5). Non-initial syllable *i, however, does not appear to have been affected by vowel harmony, as it freely occurred in both front and back harmonic stems (e.g., *süli ‘fathom’, *tuli ‘fire’). Janhunen (1981) and Sammallahti (1988) do reconstruct a harmonic pair *i / *i̮ in non-initial syllables, but this view seems to be motivated more by considerations of system symmetry than by actual comparative evidence. More recently, Janhunen (2000: 69) has doubted the presence of vowel harmony in Proto- Uralic, but the harmonic behaviour of the open vowels *a and *ä in both Finnic and Samoyed does not look like a coincidental parallel development. There is still one unresolved problem regarding vowel harmony. It is possible that even in initial syllables the vowel *i was neutral and could occur in back harmonic stems. This is suggested by a few etymologies displaying a correspondence 12  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   between Finnish i in a back-harmonic stem and a back vowel in certain other branches (e.g., Finnish nila ‘phloem’ ~ Erzya Mordvin nola ‘sapwood, bast’; Finnish kisko- ‘tear’ ~ Hungarian has-ad ‘bursts, cracks’). If the Finnic forms have preserved the original harmonic structure, this implies the reconstruction of back harmonic stems such as PU ?*ńila ‘phloem’ and ?*kiśka- ‘tear’. However, alternative solutions can also be envisaged, and the issue requires further scrutiny. The description above would imply that only a single vowel opposition (*a/*ä : *i) occurred in non-initial syllables, with the possible exception of stems with *i in the first syllable. Nevertheless, this picture must be an oversimplification, as there is evidence suggesting that also some other vowels occurred in non-initial syllables, albeit not as frequently. One possible trace of this is found in word stems where *a in the westernmost branches unexpectedly corresponds to a reduced vowel in Samoyed (as in Finnish maksa ~ Nganasan mitǝ ‘liver’); the same correspondence also occurs in some suffixal morphemes, e.g. in the ablative suffix (*-tA in the west vs. *-tǝ in Samoyed). In Janhunen’s (1981: 226–230) view the Samoyed forms have undergone secondary vowel reduction, but Zhivlov (2014: 117–121) has proposed that the correspondence reflects a distinct vowel phoneme, traces of which would also be found in Mari, Khanty and Hungarian. On the other hand, also correspondences between rounded vowels in non-initial syllables can be established, at least in derivatives: cf., e.g., North Saami doarru- ‘fight (verb)’ ~ Tundra Nenets taro- ‘wrestle’ < PU ?*toro-, derived from *tora ‘fight, quarrel (noun)’; the phenomenon is further discussed by Aikio (2015: 37–39). Some cases of non-initial syllable rounded vowels have been convincingly explained as deriving from an original unrounded vowel followed by the glide *w, but it is unlikely that all rounded vowels could be accounted for in the same way, considering the diverse sound correspondence patterns involved. Thus, further research is needed. As regards consonant distribution, Proto-Uralic syllable structure imposed three major restrictions on the combinatory properties of consonants: a) No consonant clusters occurred in word-initial position. b) In word-internal positions consonant clusters could consist of no more than two consonants (with the possible exception of clusters of type *jCC). c) No consonant clusters occurred in word-final position (with the possible exception of clusters of the type *jC). 13  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   There were also many more specific restrictions: for example, the consonants *ŋ, *r, *d and *x did not occur in word-initial position. Nearly all instances of final consonants involve suffixes, and the reconstruction of Proto-Uralic morphology remains quite incomplete (see section 1.4), but even so, there appears to be no clear evidence of the consonants *p, *č, *ś, *š, *ń, *d, *ď and *x in word-final position. With the exception of *p and *ś these consonants may have been banned from suffixal morphemes altogether (see 1.3.4). In word-internal positions geminates and many clusters of two consonants were allowed. A contrast between single and geminate consonants can be definitely established for the stops (*pp, *tt, *kk) and the postalveolar affricate (*čč). Concerning other geminates, at least *mm is suggested by Finnic evidence in a few stems (*amma- ‘ladle’, *ammi- ‘old, ancient’, *kämmin(i) ‘palm’, *kumma ‘shady, dark’). Reconstructed consonant clusters, in turn, can be divided into three major groups based on whether an obstruent or a sonorant appears as the first and the second member. The following categorization lists all clusters in each category that can with a high degree of certainty be reconstructed in at least one Proto-Uralic word stem: a) obstruent + obstruent: *pt, *kt, *čt, *st, *śt, *tk, *čk, *sk, *śk, *ps, *ks, *pś, *kś, *kš b) sonorant + obstruent: *mp, *mt, *nt, *ŋt, *ŋk, *nč, *ŋs, *mś, *ńś, *rp, *rt, *rk, *lt, *lk, *dk, *ďk, *jt, *jk, *jč, *jš c) sonorant + sonorant: *rm, *lm, *lŋ, *dm, *jm, *jn, *jŋ, *wn, *wŋ, *jr, *wl, *wd, *rw, *lw, *dw, *ďw, *jw, *rj, *lj, *wj Most of these clusters are found in monomorphemic word stems, but some can only be established on morpheme boundaries in derivatives: *čt (*mač-ta- ‘be able to’ ← *manči ‘capacity’), *st (*pis-tä- ‘stick, sting’ ← *pisi- ‘get stuck’), *lt (*kul-ta- ‘fish with a dragnet’ ← *kulki- ‘run, flow’), *mś (*ńim-śä ‘breast / milk’ ← *ńimi- ‘suck’), *jr (*koj-ra ‘male animal’ ← *koji ‘man, male’), *dm (*ad-ma ‘sleep (noun)’ ← *adi- ‘sleep (verb)’). The few plausible examples of three-consonant clusters of the type *jCC are all found in derivatives such as ?*tejm-tä- ‘soften’ ← *tejmi- ‘turn soft’. Some further clusters can be reconstructed for inflectional forms, e.g. *ln (*ül- nä above-LOC), *ms, *ns (*ojwa-m-sa head-ACC-POSS.3SG, *ojwa-n-sa head-GEN- POSS.3SG), *sn (*läs-nä ‘near-LOC’). 14  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   It is notable that there is no clear evidence of consonant clusters consisting of an obstruent followed by a sonorant within word stems (Janhunen 1982: 26). Some etymologies featuring such clusters have been proposed, but most of them involve severe phonological irregularities and other problems, and the ones that have a reasonable chance of being correct may involve clusters that secondarily arose through syncope in the individual branches (e.g., ?*kupla ~ ?*kupVla ~ *kum(p)Vla ‘bubble, bladder’). Moreover, even on morpheme boundaries such clusters appear to have been extremely rare; perhaps the only convincing example is Finnish läsnä ‘present, in attendance’ ~ Meadow Mari /liʃne/ ‘close, nearby’ ~ Tundra Nenets /jeʔnʲa/ ‘against, opposite to’ (< PU *läs-nä). Hence, clusters of this type could probably only occur across morpheme boundaries. Of course, the list above cannot come even close to a full inventory of consonant clusters that actually occurred in Proto-Uralic. However, it is rather difficult to determine whether the absence of particular clusters in the reconstruction is due to a true phonotactic constraint or an accidental gap in the limited material, and hence, the exact rules of consonant combining are difficult to reconstruct. It is reasonable to assume that we are dealing with a true phonotactic constraint when not only is the cluster absent in the Proto-Uralic corpus, but also its predictable reflexes in the lower- level proto-languages are either absent or clearly secondary (e.g., they only occur in loanwords). By such a criterion one can identify a few specific types of consonant clusters that were probably not allowed in Proto-Uralic, for example the following: ● obstruent + *p ● *w + labial consonant (*p or *m) ● nasal + heterorganic non-coronal stop (*p or *k) ● nasal + non-nasal sonorant 1.3.4 Morpheme structure In terms of their phonological structure, Proto-Uralic morphemes can be divided in three categories: content word stems, function word stems, and suffixes. Content word stems and function word stems showed a major structural difference: the former were always polysyllabic, whereas no such limitation was imposed on the latter. The canonical shape of monomorphemic content words was a disyllabic stem ending in a vowel: *(C)V(C)CV-. Furthermore, due to the phonotactic limitations of vowel 15  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   distribution, the stem-final vowels in the second syllable were mostly (or perhaps completely) limited to *a, *ä and *i (see 1.3.3). No structural differences between stems of different word classes (nouns vs. verbs) seem to have occurred. In contrast, most function word stems had a monosyllabic shape *(C)V-; less than 20 such stems have been reliably reconstructed. The following serve as examples of canonical stem shapes: Content word stems: *elä- ‘live’, *muna ‘egg’, *suxi- ‘row’, *śilmä ‘eye’, *i̮ kta- ‘hang’, *täwdi ‘full’ Function word stems: *me- 1PL personal pronoun, *ku- content interrogative, *tä- proximal demonstrative, *e- negative auxiliary This structural pattern is preserved basically intact in many Saami languages, where monosyllabic content word stems are still synchronically disallowed. All other daughter branches of Uralic, however, have developed content word stems of the shape (C)V-, (C)VV- or (C)VC(C)- through reductive sound changes. While the overwhelming majority of reconstructed monomorphemic content word stems display the canonical shape *(C)V(C)CV-, also other marginal stem shapes can be reconstructed. The most obvious cases involve noun stems of the shape *(C)V(C)CVw- (e.g., *käliw ‘brother- or sister-in-law’, *śi̮ lkaw ‘pole, rod’) and *(C)V(C)CVC(i)- (e.g., *wVdim(i) ‘marrow’, *jekin(i) ‘gums’, *šiŋir(i) ‘mouse’, *epik(i) ‘owl’); regarding the latter type, it is difficult to determine whether the final vowel *i originally was a part of the stem or an epenthetic vowel added before suffixes in inflected forms. Also trisyllabic noun stems ending in a low vowel did occur although rarely (e.g., *peŋärä ‘round / round object, wheel’). In regard to these non-canonical stem types nouns might also have differed from verbs: apparently, every single verb for which a non-canonical trisyllabic stem can be reconstructed is a semantically obscured derivative formed from a canonical stem of the type *(C)V(C)CV- (e.g., *pintäli- ‘defend, protect’ ← *pintä- ‘cover’; *puwali- ‘swell’ ← *puwa- ‘blow’). However, also many of the non-canonical noun stems correspond formally to known derivative types, and may thus have originally been bimorphemic. On the other hand, there is also scattered evidence suggesting that non-canonical noun stems may have been more common than is generally assumed. In some lexical items a canonical stem shape in one language or variety corresponds to a non-canonical 16  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   shape in another, as in the noun meaning ‘autumn’: cf. the canonical modern Finnish syksy, North Saami čakča, Erzya Mordvin śokś vs. the non-canonical archaic Finnish syys (< *sügüs) and Hungarian ősz /øːs/ (< *öɣös). In such cases the latter kinds of forms have usually been considered secondary, but it appears more natural to interpret the canonical forms as innovations that arose through syncope (e.g., PU ?*sükiś(i) ‘autumn’ > *sükśi). As regards suffixal morphemes, their phonological structure differed markedly from that of word stems. The details also are much less clear, as the reconstruction of Proto-Uralic morphology remains fragmentary, and exact sound correspondences for suffixes are more difficult to establish due to the massively reductive developments that non-initial syllables have undergone in most branches. In any case, most suffixes either consisted of a single consonant (*-C) or contained a single syllable (*-CV(C), *-CCV(C)). However, even some polysyllabic suffixes can be reconstructed, the longest being the trisyllabic negative participle suffix *-mAktAmA (see 1.4.5); most polysyllabic suffixes can be interpreted as originating in combinations of simpler suffixes. The vowels in suffixes were subject to the general phonotactic restrictions that applied to non-initial syllables (see 1.3.3): at least in most cases, the vowel was either *i or the open vowel (*a / *ä) whose frontness was further determined by vowel harmony (see 1.3.5). Also the consonant inventory in suffixes seems to have been restricted: the consonants *č, *š, *ń, *d, *ď, and *x are not known to have occurred in any suffix. 1.3.5 Morphophonology Proto-Uralic can be reconstructed as an agglutinating language with very little morphophonological alternation. Nevertheless, two morphophonological phenomena can be reconstructed. The first is vowel harmony, which governed the distribution of the open vowels *a and *ä in non-initial syllables (see 1.3.3). Due to the application of vowel harmony, suffixes containing an open vowel had front and back vocalic variants: cf., e.g., the allomorphs of the locative case suffix in *taka-na ‘behind’ and *ül-nä ‘on’. In such cases the underlying suffix can be written as *-nA, containing an archiphoneme *A whose phonemic realization was determined by the harmonic class of the stem it attaches to. The second morphophonological process is the formation of consonant stems. In word stems of the shape *(C)V(C)Ci- the stem-final *i could alternate with zero, 17  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   producing a stem allomorph of the shape *(C)VC-, called the “consonant stem”. The process was triggered by the addition of a suffix consisting of a full syllable (*- CV(C)), and it was productive in both inflection and derivation. An example of the former is provided by the local case forms of the relational noun stem *üli- ‘space above, upper surface’: LOC *ül-nä, ABL *ül-tä vs. DAT *üli-ŋ. Reconstructed derivatives based on consonant stems include causative verbs (e.g., *kan-ta- ‘transport, carry’ ← *kani- ‘go away’) and various deverbal nouns (e.g., *ad-ma ‘sleep (noun)’ ← *adi- ‘sleep (verb)’, *ńim-śä ‘breast / milk’ ← *ńimi- ‘suck’). While the phenomenon of consonant stem formation can be unambiguously reconstructed into Proto-Uralic, many details remain unclear. Janhunen (1982: 27) describes the process as purely phonologically conditioned, maintaining that the stem vowel was dropped in any suffixed form of the type *(C)VC-CV(C)-, provided that the phonotactic rules of consonant combination were not violated. However, the underlying rules have evidently been more complex. First, inflectional forms based on consonant stems actually appear to have contained consonant clusters which otherwise did not occur in the language; e.g., the cluster *sn occurred in *läs-nä (the locative form of the relational noun *läsi- ‘near / opposite to’) even though clusters consisting of an obstruent followed by a sonorant seem to have been forbidden within word stems (see 1.4.3.). Second, consonant stems have even been formed from some stems with consonant clusters (*(C)VCCi-), and in these cases concomitant cluster simplifications took place in order to avoid a three-consonant cluster not allowed by Proto-Uralic phonotactics (see 1.3.3). At least two rules of cluster simplification can be reconstructed for Proto-Uralic: ● n > Ø /_CC (e.g., *mač-ta- ‘be able to’ ← *manči ‘capacity’ + verbalizer *- tA-) ● k > Ø /C_C (e.g., *kul-ta- ‘fish with a dragnet’ ← *kulki- ‘run, flow’ + causative *-tA-) In individual branches examples of even other types of cluster simplifications are found; cf. e.g. Finnish lapsi ‘child’: PART las-ta (< PU *li̮ pśi ‘cradle’ : ABL ?*li̮ ś-ta). The Uralic origin of such alternations appears plausible, but this remains an unverified hypothesis; the phenomena associated with consonant stems require further comparative study. 18  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   1.4 Morphology 1.4.1 State of research The Uralic languages are well-known for their rich inflectional and derivational morphology, and thus, comparative morphological research has played an important role in Uralic historical linguistics since the very establishment of the language family. Nevertheless, the reconstruction of Proto-Uralic morphology remains quite sketchy. In fact, relatively little research has been conducted with the primary aim of reconstructing the system of Proto-Uralic morphology; a large part of research has been aimed at explaining the historical morphology of the individual branches or accounting for the origin of specific forms or categories of forms. The received paradigm of Proto-Uralic morphology is composed of research results that have accumulated over a long time. However, the reconstruction of Proto- Uralic morphology has not yet been systematically and critically reevaluated in the same way as phonological and lexical reconstruction were in the 1980s (see 1.3.1, 1.6.1). Moreover, information is scattered over a wide range of publications, the most readily accessible of which are general handbooks (Sinor 1988 and Abondolo 1998a, in addition to the present volume) and monograph-length introductions to individual branches (e.g., Korhonen 1981, Sammallahti 1998a on Saami; Bartens 2000, Csúcs 2005 on Permic). Rarely has there been any attempt to provide an integrated description of the Proto-Uralic morphological system as a whole, the most notable exception being the brief overview presented by Janhunen (1982: 27–38). It goes without saying that the time depth of the Uralic language family, combined with general lack of early documentation, causes major difficulties for the reconstruction of Proto-Uralic morphology. Nevertheless, Uralic historical morphology has probably relied too heavily on mere phonological correspondences between suffixes, without a systemic perspective on language change: the functions and semantics of morphemes have tended to play only a minor role, and the nature of grammatical morphemes as interacting components of a highly integrated system of morphosyntax has been largely ignored. Therefore, two misguided principles of explanation have become well-established. First, phonologically simple suffixes of the types -C and -CV tend to be viewed as historically related whenever they share the same shape (or merely the same consonant) and show whatever vague similarity of meaning or function. Second, phonologically more complex suffixes (-CCV, -CVC, etc.) are then claimed to consist of combinations of simple suffixes, with epenthetic 19  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   vowels occasionally added for phonotactic reasons. Little attention is usually paid to semantic, functional and typological questions connected with the alleged morphological developments. To take an example from verbal morphology, the 2SG imperative suffix *-k, the connegative suffix *-k and the present tense marker *-k- (see 1.4.4) are frequently conflated instead of reconstructing three homonymous suffixes, and occasionally even further connected with the Finno-Saamic derivational suffix *-(i)k(i) for deverbal nouns (a view most recently supported by Janhunen, forthcoming). The semantic and functional connections of these suffixes are far from transparent, however, and chance resemblance remains an evident possibility in a case where the compared forms consist of a single high-frequency consonant. Despite this, a further manifestation of this hypothetical PU “nominalizer” *-k has even been seen in the causative verb suffix *-k(-)tA- (cf. 1.4.5). A particularly curious manifestation of the principles discussed above is what Ylikoski (2001: 256) calls the “lative paradigm”: a persistent assumption that Proto- Uralic had several different directional (“lative”) case endings of the shape *-C (at least *-k, *-n, *-ń, *-ŋ, *-s, and *-j are commonly suggested, and even others have been proposed), and that particularly many morphologically complex suffixes have arisen by combining these “lative” endings both with other morphemes and with each other. Consequently, an unreasonably large part of nominal case endings, non-finite verb morphology and even derivational suffixes in the modern Uralic languages have been viewed as having its source in the alleged “lative” case endings. Many hypotheses of this kind were first suggested during the first half of the 20th century and then illegitimately canonized in later references, although they lacked detailed semantic and functional arguments and never were reevaluated on the basis of what is currently known about typology and change in morphology. More detailed critiques of this tradition are provided by Ylikoski (2011) and Aikio & Ylikoski (2016); the following discussion on Proto-Uralic morphology will ignore these kinds of speculations, despite their popularity in literature on Uralic historical morphology. 1.4.2 Word classes By morphological criteria one can distinguish two basic open word classes, nouns and verbs, in Proto-Uralic. Nouns were inflected for number, case and possessor person, whereas verbs were inflected for tense and mood, and they displayed subject agreement; there may also have been a separate definite conjugation displaying 20  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   agreement with a definite object. Nominal and verbal morphology is described in more detail in sections 1.4.3 and 1.4.4. In older research one frequently encounters claims that the distinction between nouns and verbs was somehow weakly developed in Proto-Uralic, a view that appears implausible in light of modern knowledge of linguistic typology. The issue is connected with the question of the so-called “nomenverbum” stems that appear to function both as nouns and intransitive verbs. Such stems occur in varying numbers in modern Uralic languages and can also be reconstructed into Proto-Uralic. However, they are rather few in number and their semantic scope is restricted: many are associated with body functions (e.g., PU *kuńśi- ‘urine; urinate’, *kusi- ‘cough’, *śülki- ‘spit’) even though also other types occur (e.g., PU *śoji- ‘sound’, *wuwa- ‘current; flow’). Instead of postulating a special “nomenverbum” word class, such cases are best analyzed as instances of cross-category homonymy of parallel lexical items in Proto-Uralic, exactly as in the case of their modern reflexes (e.g., PU *śülki ‘spit (noun)’ : *śülki- ‘spit (verb)’ > North Saami čolga : čolga-). That a handful of such pairs occurred in Proto-Uralic does not imply that the boundary between nouns and verbs in general was any more ambiguous than it is in the modern Uralic languages, contrary to Janhunen (forthcoming). No morphological criteria seem to support the reconstruction of open word classes other than nouns and verbs. In modern Uralic languages adjectives share most of their grammatical properties with nouns (with the partial exception of Samoyed where mixed systems are found, with one class of adjectives associated with nouns and another associated with verbs), but usually some morphological criteria for distinguishing a separate adjective class can be found, such as distinct inflected or derived forms for comparative, superlative, adverbial or adnominal functions. No such forms can be traced back to Proto-Uralic, however, so adjectives appear to have been originally inflected as regular nouns. Even so, adjectives obviously existed as a distinct syntactic and semantic category (see 1.5). It should be noted that Proto-Uralic had at least three derivational suffixes for forming adjectives: the proprietive adjective suffixes *-ŋa and *-ji and the caritive adjective suffix *-ktAmA (see 1.4.5). Excluding such derivatives, however, only very few reconstructed words can be unambiguously classified as adjectives; the clearest cases are PU *wuďi ‘new’, *küsi ‘thick’, *pid-kä ‘long / high’ and *i̮ ńa ‘tame’. Much more often, however, adjectives in one branch correspond etymologically to nouns in another: cf., e.g., Finnish nuori ‘young’ ~ Meadow Mari nørø ‘moist, wet; flexible’ ~ 21  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Moksha Mordvin nar ‘grass; cartilage’ ~ Tundra Nenets /nʲer/ ‘cartilage’ (< PU *ńi̮ ri); Finnish seppä ‘smith’ ~ North Saami čeahppi ‘skillful’ ~ Hungarian szép ‘beautiful’ (< PU *śeppä). Often also synchronic ambiguity is observed: e.g., Finnish nuori is not only an adjective (‘young’) but also a noun (‘young person’), and Finnish seppä is also dialectally attested as an adjective (‘skillful’). Straightforward pathways for shift of word class are created by the identical behaviour of nouns and adjectives in many syntactic constructions, such as in predicate noun and predicate adjective position (see 1.5). Numerals as well seem to have been inflected as regular nouns, but due to their special quantifying function they might have displayed syntactic idiosyncracies, as they do in many contemporary Uralic languages. A specific derivational suffix *-mtV for ordinal numerals existed already in Proto-Uralic. Non-numeral quantifiers were obviously present in Proto-Uralic, but none of them have been successfully reconstructed. Relational nouns, sometimes called ‘spatial nouns’ in Uralic linguistics, are a special subgroup of nouns with abstract, more or less grammaticalized functions. Their most important role was to function as postpositions and adverbs in their local case forms. Morphologically, however, Proto-Uralic relational nouns do not seem to have differed from other nouns: e.g., PU *ül-nä LOC ‘on’, *ül-tä ABL ‘off (from)’ and *üli-ŋ LAT ‘onto’ seem to be formally quite regular local case forms (see 1.4.3) of the relational noun *üli ‘space on or above’, but these word-forms were syntactically exceptional in that they functioned as postpositions. In modern Uralic languages, however, such case forms of relational nouns have tended to become lexicalized and opaque; sometimes they have even lost their status as phonologically independent words, being reduced to clitics and eventually grammaticalized as new case suffixes (see 1.4.3). That relational nouns originally resembled ordinary nouns can, however, be seen from several factors. On the one hand, Proto-Uralic relational nouns have functioned as bases of derivatives just like ordinary nouns have (cf., e.g., Finnish yltä- ‘reach’, ylevä ‘noble, sublime’, ylänkö ‘highlands’ ← PU *üli ‘space on or above’), and on the other, various word stems which appear to have been ordinary nouns in Proto-Uralic have secondarily developed into relational nouns in some Uralic languages. In fact, the number of stems reconstructible as relational nouns in Proto- Uralic is quite limited, the clearest cases being *üli- ‘space on or above’, *i̮ la- ‘space under or below’, *edi- ‘space in front of’, *taka- ‘space behind (on the other side of)’, *läsi- ‘space near or opposite to’, *müŋä- ‘space behind (one’s back)’. During the 22  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   independent development of the various Uralic branches many more relational nouns have developed from diverse sources, for example from the PU nouns *peŋä ‘end’, *mi̮ xi ‘earth’, *tüŋi ‘base’, and *muka ‘back (body part)’. Nearly all postpositions and adverbs that can be reconstructed for Proto-Uralic are case forms of relational nouns. However, rare examples of other types of adverbs are also known, e.g. Lule Saami mieddel(a) ‘past’ ~ East Mansi mæntǝl ‘along, through; during’ (< PU *mentälä(-n), apparently derived with an unclear suffix *-lA from the verb *mentä- ‘miss, cause to go’, which in turn is a causative derivative of *meni- ‘go’); but these may also involve lexicalized case forms of nouns (at least Lule Saami mieddela reflects a genitive form). Janhunen (1982: 28) maintains that no evidence exists for any separate group of indeclinable words in Proto-Uralic. However, as languages in general tend to have also lexicalized and opaque forms, also Proto-Uralic has probably had such even if they remain difficult to reconstruct. As regards closed word classes with grammatical functions, relatively little evidence survives. Personal pronouns are difficult to reconstruct (Abondolo 1998b: 24), and only a highly tentative reconstruction of singular and plural personal pronouns can be presented: ?*mi(-)nä 1SG, ?*ti(-)nä 2SG, ?*sV- (*sä-?) 3SG, ?*me. 1PL, ?*te 2PL, ?*se 3PL. The correspondences between personal pronouns are rife with irregularities, the background of some of which is not well understood. The most notable anomalies are the irregular initial *n instead of *t in 2nd person pronouns in Khanty and Mansi (cf. Kulonen 2001), the loss of initial *m in the 1SG pronoun in Mansi and Hungarian, and the development of the singular personal pronouns into the shape *mu(-)n 1SG : *tu(-)n 2SG : *su(-)n 3SG in Saami and Mordvin. Proto-Uralic *mun 1SG and *tun 2SG were formerly reconstructed for Samoyed as well (Janhunen 1981: 232–233), but it has recently become clear that the Samoyed pronouns reflect front vocalic forms (*min(ä)-, *tin(ä)-) after all. Dual personal pronouns cannot be reconstructed, but they most probably existed because dual number occurred as a category of both nominal and verbal morphology in Proto-Uralic (see 1.4.3, 1.4.4). Personal pronouns may originally have been indeclinable or at least showed a defective case declension; this is suggested, for instance, by the fact that languages such as Samoyed and Hungarian show supplementary oblique case forms of personal pronouns that have originated in word stems of diverse background (such as Tundra Nenets puxǝdᵊ ‘body’ and PU *keti ‘skin’ > ‘person’) (Janhunen, forthcoming). In addition to personal pronouns a couple of other pronouns can be established, the most unambiguous cases being *tä- (a proximal demonstrative), ?*to-/?*tu- (a 23  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   distal demonstrative), *mi- ‘what’, *ke- ‘who’, and *ku- (a general content interrogative). From these stems, various kinds of adverbs and other function words have later been derived, but the concrete formations mostly cannot be traced back to Proto-Uralic. Besides personal, demonstrative and interrogative pronouns, no other types of pronouns can apparently be reconstructed. The branches from Saami to Permic have a shared reflexive pronoun (e.g., North Saami ieš, Finnish itse, Komi at͡ sʲ-, asʲ-), but it developed through grammaticalization from a Proto-Uralic noun meaning ‘(shadow-)soul’ (cf. Irtysh Khanty is ‘shadow-soul’, North Mansi is ‘shadow, shadow-soul, ghost’). A couple of Proto-Uralic auxiliary verbs can be reconstructed. Clausal negation was expressed with a construction containing the negative auxiliary *e-, and negative imperative constructions were formed with the prohibitive auxiliary ?*älä- ~ ?*elä- (see 1.5); the latter might have been derived from the former with an unknown derivational suffix *-lA-. The verb *woli- ‘be’ is attested as a copula and an auxiliary in all branches except for Saami, and must thus have had various grammaticalized functions in Proto-Uralic already, but also the concrete lexical meanings ‘be alive’, ‘live’ and ‘dwell’ are marginally attested and may be more archaic. Certain irregular forms such as Võro um 3SG : ummaq 3PL (? < *wo-ma : *wo-ma-t) suggest that *woli- could have been derived from a monosyllabic primary auxiliary *wo- (see 1.3.4), but an irregular loss of *l could also be assumed, considering the special status and high frequency of the verb *woli-. A clearer example of a secondary monosyllabic auxiliary is (North) Saami (*)lea- ‘be’ and its cognates. They have been claimed to derive from a Proto-Uralic auxiliary *le- ‘be / become’, but more probably they represent a reduced and grammaticalized form of an originally bisyllabic content verb stem ?*lewV-, preserved in its more original sense in Meadow Mari lij- and Nganasan dʲiǝ- ‘give birth (of animals)’ (Aikio 2014: 23). No other function word classes (e.g. conjunctions, discourse particles, pro- sentences, interjections) can be reconstructed. This, however, does not mean that such word classes did not exist in Proto-Uralic. 1.4.3 Noun declension Proto-Uralic nouns were inflected for case, number and possession. The case system consisted of at least six cases, which can be divided into two groups. Three grammatical cases were used for marking core syntactic functions, in addition to 24  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   which there were three local cases. Yet two other adverbial cases may have existed in Proto-Uralic, but their reconstruction remains somewhat uncertain. (See Table 1.4.) SG DU PL NOM -Ø ?*-k(V) *-t ACC *-m GEN *-n }*-j LOC *-nA ABL *-tA LAT *-ŋ ?TRANS ?*-ksi ?CAR ?*-ktAk Table 1.4 The Proto-Uralic case suffixes. In terms of morphosyntactic alignment Proto-Uralic was highly probably a nominative-accusative language. The subject arguments of both intransitive and transitive verbs were in the nominative case, the singular form of which was morphologically unmarked. The accusative case in *-m can be reconstructed, but Proto-Uralic like most of its descendants probably had differential object marking, allowing also nominative objects (see section 1.5). The main function of the third grammatical case, the genitive in *-n, was to mark nouns appearing in adnominal modifier position, especially as possessive modifiers. The system of Proto-Uralic grammatical case endings survived basically intact in Saami, Mari and Samoyed; Finnic and Mordvin developed syncretism of the accusative and the genitive through a sound change *m > *n in word-final position. The local cases formed a tripartite system, with one case expressing a static location (the locative in *-nA) and two dynamic cases expressing movement away from a location (the ablative in *-tA) and to a location (the lative in *-ŋ). Unlike the system of grammatical cases, the system of original local cases is nowhere preserved in its entirety, even though some languages have retained one or two of these cases in a local function: e.g., Khanty has a locative case in *-nǝ (< PU *-nA), and Permic has both an inessive case in *-i̮ n (< PU *-nA) and an illative case in *-e̮ (with a lost final consonant, possibly PU *-ŋ). Vestiges of the complete set of three local cases are, however, found in sets of postpositions and adverbs that originated as local case forms 25  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   of relational nouns: cf. Erzya Mordvin alo, aldo, alov ~ Nganasan ŋilʲenu, ŋilʲeðǝ, ŋilʲe (< PU *i̮ l(a)-na ‘under-LOC’, *i̮ l(a)-ta ‘under-ABL’, *i̮ la-ŋ ‘under-LAT’). The Proto-Uralic local case endings have tended to develop more abstract grammatical functions or to lose their productivity and become confined to lexicalized forms. At the same time, new tripartite series of local cases have developed mostly by agglutination and grammaticalization of original postpositions. For example, in Saami and Finnic the locative *-nA developed into an essive case, the ablative *-tA developed into a partitive case (which became subsequently marginalized in Saami), and the lative case *-ŋ was lost in both branches. The primary local functions of these cases were, in turn, taken over by secondary case forms combining an element *-s(i)- with the primary local case endings (*-s-nA : *-s-tA : *-si-n < *-si-ŋ). These secondary local case endings appear to have partial cognate forms in Samoyed and to originate from Uralic postpositions (Ylikoski 2016b). Later Finnic also developed another tripartite set of “external” local cases (*-l-nA : *-l-tA : *-le-n < *-li-ŋ) via agglutination of the PU postpositions *ül-nä LOC ‘on’, *ül-tä ABL ‘off (from)’ and *üli-ŋ LAT ‘onto’, which in turn were local case forms of the relational noun *üli ‘place on or above’ (Aikio & Ylikoski 2016). The Proto-Uralic case system comprised at least the six cases described above, but there are also two other cases whose reconstruction appears plausible, though somewhat uncertain. A caritive (also called abessive or privative) case with the suffix *-ktAk is found in Saami, Finnic, Mari and Permic, and T. Itkonen (1992) has presented plausible arguments for the view that precisely the same suffix ultimately lies behind the phonologically anomalous Khanty *-lǝɣ and Mansi *-tǟl which function as both caritive case endings and caritive adjective derivatives. Thus, it is probable that *-ktAk functioned as a case ending in Proto-Uralic already (see Janhunen 1982: 31; 2014: 317). The ending must originally be bimorphemic, as the component *-ktA- also occurs in the derivational suffix *-ktAmA which forms caritive adjectives (and in Mordvin also functions as a caritive case suffix); in individual branches also other complex suffixes containing *-ktA- are attested. Moreover, *- ktAmA in turn acts as a component of the negative participle suffix *-mAktAmA. Hence, the caritive morpheme *-ktA is in any case of Uralic ancestry, although it remains somewhat unclear whether caritive was a true noun case (instead of a derivational category) in Proto-Uralic. The eighth possible Proto-Uralic case is the translative in *-ksi. This case is traditionally assumed to occur only in two western branches, Finnic and Mordvin, but 26  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Ylikoski (2017a) argues that the same suffix probably also underlies the Mari lative case ending -(e)ʃ (which despite its name is not really a directional local case). Moreover, according to Janhunen (1989b) the suffix is cognate with the Samoyed predestinative (also called destinative or benefactive) marker *-tǝ-. The Samoyed predestinative is a special category of noun declension that expresses a sense of ‘for, for the benefit of’ and usually occurs in combination with possessive suffixes. In objection to Janhunen’s proposal it has been pointed out that the uses of translative case forms and predestinative forms show notable differences in regard to argument structure (Siegl 2013a: 401–402). Nevertheless, one can also find far-reaching congruence between the use of a Samoyed predestinative form and the Finnic translative case in many quite prototypical constructions, for example: (1) Tundra Nenets (Tereščenko 1965: 291) nʲe nʲuːm nʲedᵊnta meᵊda woman child woman.DST.GEN.POSS.3SG take.3SG>SG (Finnish:) otti tytön vaimoksensa take.PAST.3SG girl.ACC wife.TRSL.POSS.3SG ‘he took the girl as a wife for him’ (2) Tundra Nenets (Tereščenko 1965: 688) tʲikiᵊ nʲem nʲebʲadǝnᵊ pæᵊrŋawǝsʲᵊ that woman.ACC mother.DST.GEN.POSS.1SG regard.PST.1SG>SG (Finnish:) luulin sitä naista äidikseni think.PST.1SG that.PART woman.PART mother.TRSL.POSS.1SG ‘I mistook that woman for my mother’ (3) Forest Enets (Siegl 2013: 386) uu biiðunʲ ebut soiða 2SG son-in-law.DST.GEN.POSS.1SG be.COND.GEN.POSS.2SG good.3SG (Finnish) sinä olisit hyvä vävykseni 2SG be.COND.2SG good son-in-law.TRSL.POSS.1SG ‘you would be a good son-in-law for me’ 27  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   It is notable that these examples do not only show an exact semantic and functional equivalence, but also a fully regular sound correspondence of suffixal morphology: Tundra Nenets -dᵊnta, -dǝnᵊ ~ Finnish -ksensa, -kseni (< PU *-ksi-n-sa, *-ksi-ni). In the last example even the word-form in its entirety matches: Forest Enets biiðunʲ ~ Finnish vävykseni (< PU *wäŋiwi-ksi-ni). Some further illustrative examples are discussed by Salminen (2014: 291–296) and Ylikoski (2017a). Thus, it is very difficult to avoid the conclusion that the predestinative and translative morphemes are cognate and derive from PU *-ksi. Lehtinen (2007: 67) goes as far as to say that “according to the present view” there were seven noun cases in Proto-Uralic, the translative in *-ksi being one of them, and also Janhunen (2014: 317) laconically postulates a “predestinative case” for the proto-language. Nevertheless, it remains somewhat unclear whether the Proto-Uralic *-ksi suffix was really a marker of case and not some other morphological category. Three grammatical numbers can be distinguished in Proto-Uralic: singular, dual, and plural. However, the categories of case and number seem to have been mutually almost completely exclusive in nominal declension. A full paradigm of 6–8 cases can only be established in the singular, whereas for plural only two distinct forms can be reconstructed, and neither of them is based on the singular case endings: the nominative plural in *-t, and another form with the suffix *-j (Honti 1997 argues for a different view). Regarding the former, it is worth noting that apparently *-t occurred as a plural marker also in possessive suffixes (see below) and subject personal endings of verbs (see 1.4.4). The latter suffix has been hypothesized to have been a “conjunctive” form used in both accusative and genitive functions (Janhunen 1982: 29). The actual status of this form remains unclear, however; the suffix *-j marks the genitive plural in Saami but the accusative plural in Samoyed. Elsewhere the suffix does not occur as the sole marker of any form; most branches have simply lost the suffix, whereas in Finnic it was generalized as a regular marker of nominal plurality that occurs in combination with case suffixes; Hungarian -i-, which marks plural possessa in conjunction with possessive suffixes, might be of the same origin. As regards dual number, it can be unambiguously reconstructed for Proto-Uralic possessive and verbal morphology, and is likely to have also been a subcategory of nominal number. However, the reconstruction of dual markers for nouns remains problematic. Dual as a grammatical subcategory has only been retained in Samoyed, Khanty, Mansi and Saami, and unambiguous reflexes of the nominal dual suffix *- k(V) are only found in Mansi (*-ɣ-), Khanty (*-ɣǝn-, *-ɣǝl-), and Samoyed (*-kǝń-). 28  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   The Samoyed form and the Khanty allomorph *-ɣǝn- seem to consists of two distinct plural markers, the nominal dual marker PU *-k(V) and the dual marker *-(j)n which occurs in possessive suffixes on nouns (see below) and subject personal endings on verbs (see 1.4.4). In Saami the use of dual number is restricted to definite persons, and nominal dual can be expressed by transparent yet fully grammaticalized compound forms (such as North Saami nieidda-guovttos girl.GENACC-DU ‘the two girls’, with a dual marker derived from the numeral guokte ‘two’). Thus, as clear material evidence of the dual suffix *-k(V) is limited to the Siberian branches, its reconstruction to Proto-Uralic remains somewhat hypothetical. However, it is not impossible that a trace of it survives in the now moribund Návuotna Sea Saami dialect of North Saami, in which dual predicate nouns and adjectives (with a lost final consonant, possibly PU *-k) were distinct from plural ones (in -t < PU *-t) (Larsen 2014: 219); the issue requires further research. Proto-Uralic nouns could also be declined for possessor person: a set of nine possessive suffixes, distinguishing three numbers and three persons, can be reconstructed. The possessive suffixes were attached after the suffix for case and number. The original suffix order is preserved in the lateral parts of the family – Saami, Finnic and Mordvin in the west, and Samoyed in the east – and the reverse order attested in Ugric and partially also in Permic and Mari represents an innovation (Nichols 1972). Minor morphophonological alternations can be reconstructed into Proto-Uralic in connection with the first person possessive suffixes beginning with *m: when attached after the accusative suffix *-m or the genitive suffix *-n, the sequence of two nasals was simplified: *m–m > *m; *n–m > *n (Janhunen 1982: 31). The endings of the possessive declension in the singular forms of the three grammatical cases are shown in Table 1.5. 29  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   NOM ACC GEN 1SG *-mi *-mi *-ni 2SG *-ti *-mti *-nti 3SG *-sA *-msA *-nsA 1DU *-mi(j)n *-mi(j)n *-ni(j)n 2DU *-ti(j)n *-mti(j)n *-nti(j)n 3DU *-sA(j)n *-msA(j)n *-nsA(j)n 1PL *-mAt *-mAt *-nAt 2PL *-tAt *-mtAt *-ntAt 3PL *-sAt *-msAt *-nsAt Table 1.5 The Proto-Uralic suffixes marking different possessor persons in the nominative, accusative and genitive singular. Possessor person marking in noun declension can be reconstructed in detail because the system happens to be preserved nearly unchanged in the western and eastern extremes of the language family, i.e., in Saami and Samoyed (Salminen 1996); in other branches the system has been more or less restructured. Even so, unclear issues remain, such as the marking of plural possessees: western branches (e.g., Saami and Mordvin) have a suffix *-n-, whereas Samoyed indicates *-j- in the same function. The possessive suffixes as well as the personal endings of verbs (see 1.4.4) display the same initial consonants (*m for 1st person, *t for 2nd person, *s for 3rd person) as the personal pronouns (see 1.4.2). Thus, it has often been assumed that the suffixes arose through agglutination of original pronouns. However, if such a grammaticalization process indeed has taken place, it was completed already well before the disintegration of Proto-Uralic. Yet another category sometimes postulated for Proto-Uralic noun morphology is the so-called predicative or nominal conjugation. While most Uralic languages allow nominal predication without a copula (see chapter 52), in Mordvin and Samoyed such predicate nouns can display subject person and number markers and even past tense markers identical to verbs. This feature has occasionally been interpreted as an archaism (Janhunen 1982: 38), but as the phenomenon is absent in all other branches of Uralic, it probably represents an innovation. 30  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   1.4.4 Verb conjugation While Proto-Uralic noun declension can be reconstructed in relatively good detail (see 1.4.3), the picture of verb conjugation is less clear. Numerous cognate morphemes in verb conjugation have been identified, but it is much more difficult to establish entire cognate forms with a definite combination of cognate suffixes. This is because the systems of verbal morphology in the various branches of Uralic show major discrepancies, of which at least three types can be distinguished: a) Presence vs. absence of a grammatical category: e.g., a distinction between the so-called objective conjugation (marking of the object on the verb) and subjective conjugation is present in five branches (Mordvin, Mansi, Khanty, Hungarian, Samoyed) but absent in four (Saami, Finnic, Mari, Permic). b) Presence vs. absence of overt marking of a grammatical category: the present tense is mostly unmarked, but a present-tense marker *-k- nevertheless occurs in Mansi and possible traces of it are also found elsewhere. c) Different phonological shapes and paradigmatic distributions of markers for a grammatical category: e.g, the past tense can be marked with either *-j- or *- ś(A)-, or even both; when both suffixes are present in the same branch, they show an idiosyncratic paradigmatic distribution. Such variation obviously poses major problems for reconstruction. In the present state of research it is possible to identify a common pool of ancient verb suffixes and to roughly outline the main features of Proto-Uralic verb conjugation, but not to reconstruct entire paradigms or even subsets of paradigms. Several issues remain unresolved, and for a number of details alternative solutions can be proposed. At any rate, it seems clear the morphological skeleton of Proto-Uralic finite verb forms consisted of the verb stem followed by two suffix slots – the first for a marker of tense or mood, and the latter for a subject person suffix. Tense and mood seem to have been mutually exclusive categories. The Proto-Uralic verb distinguished between at least two tenses, the present (or more exactly, the present-future) and the past tense. Most Uralic languages suggest that present tense forms were formed by directly attaching subject person suffixes to the verb stem, as in Skolt Saami mõõnâ-m, Finnish mene-n, Meadow Mari mij-em ‘I go’ (< *meni-m). On the other hand, a completely different situation is found in Samoyed, where a stem-extending element must always be attached to the verb stem 31  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   before the subject person suffixes; the tense in Samoyed most closely corresponding to the present tense in other Uralic languages is the so-called aorist, which is marked with suffixes of more or less unclear origin (Proto-Samoyed *-j-, *-ŋA- and *-ǝ-). While this feature of Samoyed is likely to represent an innovation (Janhunen 1998: 471–472), it is not so clear that the former type would represent the original system either, as in the Ob-Ugric languages also present tense suffixes are attested. The Khanty present tense marker -l- (as in East Khanty men-l-ǝm ‘I go’) is obviously innovative and originated in a derivational suffix (Honti 1998a: 346), but the Mansi present tense marker *-ɣ- (as in Konda Mansi min-ɔːm ~ mǝn-ɣ-ǝm, North Mansi min-eːɣ-ǝm ‘I go’ < ?*min-ǝj-(ǝ)ɣ-ǝm) may have a different origin, possibly reflecting the present tense marker *-k-. This marker is not found as a productive element of verb conjugation in other branches, but fossilized vestiges of it may survive in certain forms of the negative auxiliary in Mari and Permic, as well as in the otherwise unexplained geminate consonants of Finnic 1PL and 2PL person suffixes (see chapter 14.3.2). Therefore, the Proto-Uralic present tense was not necessarily unmarked. Two past tense markers, *-j- and *-ś(A)-, can apparently be reconstructed. Most branches seem to have generalized one of the two suffixes: in Mansi and Samoyed only *-ś(A)- is attested, whereas Saami, Finnic and Permic use *-j- as the default past tense marker. In Saami and Finnic, however, *-ś- is found in past tense forms of the negation auxiliary; this was probably the original situation in Mordvin, too, where third-person past tense forms with *-ś- appear to have secondarily developed to avoid homophony with present-tense forms (Zaicz 1998: 200). However, in two branches both suffixes play key roles in verbal conjugation, albeit with a completely different functional opposition. Mari uses *-j- in conjugation I verbs and *-ś- in conjugation II verbs; the former type mainly derives from Proto-Uralic verbs with stem-final *i and the latter from ones with stem-final *A. In East Khanty, on the other hand, the two markers encode different tense categories, even though synchronically *-j- is reflected as zero: cf., e.g., archaic Surgut Khanty /mǝnsǝm/ ‘I was going’ (< *meni-śV-m) vs. /mǝnǝm/ ‘I went’ (< *meni-j(i)-m); note that in modern language the distinction between the two past tenses has become obsolete. It is obviously quite difficult to reconstruct the original system behind the diffuse synchronic patterns, but it seems clear that both past tense markers must have been in one way or the other present in Proto-Uralic. The marker *-ś(A)- may ultimately have been grammaticalized from 32  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   some sort of participle suffix of the shape *-śA, which is attested in fossilized derivatives (Janhunen 1998: 472; see 1.4.5). As regards mood, the only category that can be completely unambiguously reconstructed is the imperative, which in the 2SG form was marked with the suffix *-k, as in PU *meni-k ‘go!’ (> South Saami mïnnh, Finnish mene(ʔ), Tundra Nenets /mʲinᵊʔ/, etc.). This suffix is homonymous with (and perhaps related to) the suffix *-k of the connegative form (see below) and the present-tense marker *-k- discussed above, as far as the latter can indeed be reconstructed into Proto-Uralic. Note that the 2SG imperative contained no subject person suffix, unlike the other finite verb forms. The reconstruction of other imperative forms is less clear; they may have evolved later, often on the basis of the 2SG form. Markers for other moods in the Uralic languages are highly heterogeneous, and they usually originate from either verbal derivative suffixes with an aspectual function or auxiliary verbs. Nevertheless, a mood marker possibly reconstructible for Proto- Uralic is *-ni-, attested in at least Finnic as a marker of potential mood and in Mari as a marker of desiderative mood (cf. Finnish men-ne-n ‘I might go’, Meadow Mari mijəә-ne-m ‘I want to go’ < *men(i)-ni-m). The same morpheme likely also occurs as a component of what appears to be a bimorphemic marker (?*-ni-kV-) of conditional mood in Mansi and Hungarian: cf., e.g., North Mansi min-nuw-ǝm ‘I would go’ < ?*meni-ni-kV-m. However, no trace of the mood marker *-ni- seems to be found in other branches, nor does it have any obvious cognate morphemes in other functions. Therefore, it is uncertain whether a mood marker *-ni- occurred in Proto-Uralic, and even if it did, its exact original function remains unclear. Regarding subject person suffixes, the Uralic languages are divided as to whether they possess one or two sets of such endings. Several branches possess a distinct set of suffixes that is generally used with transitive verbs and marking the presence of a topical/definite object (the so-called objective or definite conjugation), and another set of suffixes used with intransitive verbs and transitive verbs with a non- topical/indefinite object (the so-called subjective or indefinite conjugation); the exact criteria determining the choice of conjugation are complex and varying, however. The four easternmost branches, Samoyed, Khanty, Mansi, and Hungarian, have a system of objective conjugation encoding the presence (and, with the exception of Hungarian, also the number) of a definite object. In the west, Mordvin has an even more extensive system of objective conjugation which encodes both the person and the number of a definite object. On the other hand, in Saami, Finnic, Mari and Permic no 33  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   obvious traces of two separate conjugations seem to be found; the subject person suffixes in these branches show correspondences with the respective suffixes of the subjective conjugation in the other branches. The Mordvin system of objective conjugation differs so radically from the others that it is highly probably an innovation (Keresztes 1999), but the objective conjugation in the easternmost branches might nevertheless represent a Proto-Uralic archaism that was lost in the more western branches (Janhunen 1982: 35). On the other hand, it is equally possible that the Saami, Finnic, Mari and Permic branches possess a more archaic verb system, and the definite conjugation in the eastern branches is an areal innovation (Salminen 1996). Thus, one can alternatively reconstruct one or two sets of subject person suffixes for verbs, depending on whether definite conjugation is assumed to have existed in Proto-Uralic or not. Table 1.6 shows the two sets of endings: the default or subjective conjugation endings can be reconstructed in any case, whereas the objective conjugation endings remain more hypothetical. It is important to note that the personal endings that may have been used in the objective conjugation are identical with the possessive suffixes on nouns (see 1.4.3). CONJUGATION: PERSON: SG DU PL DEFAULT / 1 *-m *-mi(j)n *-mAt INDEFINITE? 2 ?*-t / ?*-n *-ti(j)n *-tAt 3 ?-Ø ?*-kA(j)n ?*-t DEFINITE? 1 ?*-mi ?*-mi(j)n ?*-mAt 2 ?*-ti ?*-ti(j)n ?*-tAt 3 ?*-sA ?*-sA(j)n ?*-sAt Table 1.6 The subject person suffixes of Proto-Uralic verbs. Some issues in the reconstruction of the default set of subject person suffixes also remain uncertain. The most unclear case is the 2SG category for which two endings occur, *-t and *-n. The four westernmost branches (Saami, Finnic, Mordvin and Mari) agree on *-t, whereas Khanty and Mansi point to *-n exclusively (and also show *n instead of *t in all second-person morphemes, including personal pronouns, which must be a secondary development). Samoyed has *-n in the indefinite conjugation. The Permic branch shows a most peculiar duality, with *-n occurring primarily in 34  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Komi and *-t in Udmurt. It is possible that both endings go back to Proto-Uralic in one way or the other. On the other hand, it is not unthinkable that the original 2SG ending was *-n, and the ending *-t secondarily emerged through analogy: the model would have been provided by the stop *t in the 2DU and 2PL endings, the 2SG possessive suffix *-ti (see 1.4.3), and the 2SG personal pronoun ?*ti(-)nä (see 1.4.2). As regards the third person suffixes in the default set, many Uralic languages have innovated secondary third person suffixes: cf., e.g., Finnish puree, (archaic) purevi 3SG ‘(s)he bites’, purevat 3PL ‘they bite’ (< *puri-pa, *puri-pa-t, with an active participle suffix *-pA), Erzya Mordvin porʲi 3SG ‘(s)he chews’, Konda Mansi pori 3SG ‘(s)he bites’ (< *puri-ja), Erzya Mordvin porʲitʲ 3PL ‘they bite’ (< *puri-ja-t; with a deverbal agent noun suffix *-jA) (see 1.4.5). Originally, however, third person verb forms were probably unmarked for person: the 3SG form apparently had a zero (-Ø) ending and the 3PL form merely contained the plural marker *-t (see 1.4.3). The same analysis applies to the 3DU form at least as far as Samoyed is concerned: the 3DU suffix in the indefinite conjugation, Proto-Samoyed *-kǝń, was homonymous with the dual marker of nouns. Apparently, this suffix can be analyzed as containing two separate dual markers, *-k(V) and *-(j)n (see 1.4.3). Nevertheless, in verb conjugation the combined suffix as a whole must date back to Proto-Uralic: it is quite obviously also reflected as the 3DU person suffix *-kān in western Saami languages, even though this suffix has been formerly regarded as a secondary element of obscure origin (Sammallahti 1998a: 79, 218) or a product of an analogical development (Korhonen 1981: 278–279). In addition to finite conjugation, several non-finite verb forms most probably existed in Proto-Uralic, considering their prominence in modern Uralic languages. Indeed, many derivational suffixes forming deverbal nouns (such as *-mA, *-pA, *-jA, *-ntA; see 1.4.5) are also attested as markers of non-finite forms. However, it is not clear whether these varying and diverse forms reflect categories of verb inflection (rather than deverbal derivation) in Proto-Uralic. The only non-finite form that has been unambiguously reconstructed is the connegative form used in clausal negation (see 1.5). As noted above, the connegative form was homonymous with the 2SG imperative, a feature preserved to the present day in many branches. 1.4.5 Word formation Word formation is the least systematically studied part of Proto-Uralic morphology. The synchronic systems of derivation and compounding of most branches of the 35  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   family are well described (a concise overview of derivation in Uralic languages is provided by Kiefer & Laakso 2014), but in-depth comparative study on the topic has been scarce. Lehtisalo (1936) remains the most thorough treatment of Proto-Uralic derivation in spite of being severely outdated, while modern knowledge on the topic remains scattered in diverse papers and monographs. The following list presents a total of 35 derivational suffixes that can highly probably be reconstructed to Proto-Uralic. For each one or two example formations are given together with their reflexes in at least two branches. Note that even if the suffixes are Proto-Uralic, the reconstructed derived lexemes may be quasi-cognates that were separately formed at later stages. A. Deverbal nouns (i) *-mA (general nominalizer): e.g., PU *elä- ‘live’ → *elämä > North Saami eallin, Finnish elämä ‘life’, Meadow Mari ilem ‘farmstead, settlement’, Nganasan nʲilɨmɨ ‘herd’. (ii) ?*-o / ?*-w (general nominalizer): e.g., PU *elä- ‘live’ → ?*elo/?*eläw > North Saami eallu ‘herd’, Finnish elo ‘life; crops; livestock; household property’, Nganasan nʲilu ‘life’. (iii) *-pA (active participle): e.g., PU *elä- ‘live’ → *eläpä > Finnish elävä PRS.PTCP ‘living, live’, Udmurt ulep ‘living, live, alive’, East Khanty jǝlǝw ‘new, fresh’. (iv) *-ntA (action noun): e.g., PU *elä- ‘live’ → *eläntä > South Saami jielede ‘life’, Olonetsian eländy ‘living, life; dwelling’, Tundra Nenets jilʲenʲa IPFV.PTCP ‘living’. (v) *-jA (agent noun): e.g., PU *sala- ‘steal’ → *salaja > Moksha Mordvin sɑlɑj, Nganasan tolɨǝ ‘thief’ (vi) *-śA (participle with unclear semantics): e.g., PU *kali- ‘die’ → *kaliśa > Erzya Mordvin kulozʲ PFV.PTCP ‘dead, deceased’, Konda Mansi kɔls ‘person, human being’ (< *‘mortal’), Tundra Nenets xasa ‘late, deceased’ (vii) *-kkAs(i) (inclinative adjective or agent noun): e.g., PU *sala- ‘steal’ → *salakkas(i) > Mator tælǝgæt ‘thief’; *sala- → *salanti- → *salantikkas(i) > North Saami suoládahkes ‘thievish’. (viii) *-mAktAmA (negative participle): e.g., PU *tumti- ‘know, recognize’ → *tumtimaktama > Finnish tuntematon ‘unknown’, Tundra Nenets 36  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   tumtǝwᵊdawe-jᵊ NEG.PTCP ‘unrecognized’. — A combination of *-mA (i) and *-ktAmA (xvii). B. Denominal nouns (ix) *-kA (unclear semantics, forms both nouns and adjectives): e.g., PU *ńi̮ ri ‘soft, tender’ → *ńi̮ rka > Hill Mari nørgəӛ, North Mansi nʲɑːrəәɣ ‘cartilage; young; tender’; PU *pidi ‘long / high’ → *pidkä > Finnish pitkä ‘long’, Tundra Nenets pʲirt͡ sʲa ‘high’. (x) *-kkA (unclear semantics, forms both nouns and adjectives): e.g., PU *pälä ‘side, half’ → *päläkkä > Hill Mari pelæk, East Khanty peːɭəәk ‘half’; PU *kuwa ‘long’ → *kuwakka > North Saami guhkki, Erzya Mordvin kuvɑkɑ ‘long’, Nganasan kuogu-nu ‘long ago’. (xi) *-kśi (relational animate noun): e.g., PU *käliw ‘brother- or sister-in-law’ → *käliwikśi-t NOM.PL > Finnish kälykset, Tundra Nenets sʲelǝsᵊʔ ‘in-laws (to each other)’ (Salminen 2014: 296–297). (xii) *-ńśA (collective animate noun): e.g., PU *ekä ‘uncle / father’ → *ekäńśä > North Saami egeš ‘uncle (father’s older brother) and nephew/niece’, Konda Mansi jeɣǝnʲsʲ ‘father and son/daughter’. (xiii) *-ksi (unclear semantics): e.g., PU *śepä ‘neck’ → *śepäksi > Erzya Mordvin sʲiveks ‘harness’, Meadow Mari ʃyjɑʃ ‘necklace; collar’, Komi ɕijɘs ‘horse collar’; PU *pučki ‘hollow stalk, tube’ → *pučkiksi > Komi pɨt͡ ʃɘs ‘inside, interior; cavity’, Nganasan /hytəәðəә/ ‘body, figure’. (xiv) *-wiksi (unclear semantics): e.g., PU *i̮ la ‘space under or below’ → *i̮ lawiksi > North Saami vuolus ‘lower part of fishing net’, Finnish alus ‘ship; bottom, underlay’, Komi ulɘs ‘chair’, Tundra Enets iruð ‘bottom; sole’. — A homonymous, potentially related deverbal noun suffix is attested in at least Saami, Finnic and Mordvin; this can be analyzed as a combination of ?*-w / *-o (ii) and *-ksi (xiii). (xv) *-ŋA (proprietive adjective): e.g., PU *wäki ‘strength’ → *wäkiŋä > Finnish väkevä, Erzya Mordvin vijev, East Khanty wøːkǝŋ ‘strong’. (xvi) *-ji (proprietive adjective): e.g., PU *kala ‘fish’ → *kalaji > North Saami guol'lái ‘rich in fish’, East Khanty kuːlɨ, Nganasan kolɨǝ ‘fish-’. (xvii) *-ktAmA (caritive adjective): e.g., PU *wäki ‘strength’ → *wäkiktämä > Skolt Saami viõǥǥte´m, Estonian väeti, Meadow Mari βijdǝme, Kazym Khanty weːwtɑm ‘weak’. — The suffix is bimorphemic: *-ktA- also occurs in the abessive case suffix (see 1.4.3). 37  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   (xviii) *-mpV (moderative or comparative adjective): e.g., PU *śoma ‘nice / good’ → *śomampV > Fi somempi ‘prettier, nicer’, Tundra Nenets sǝwɑmpo-jᵊ ‘rather good’. — Possibly a bimorphemic suffix consisting of the translative verb suffix *-mi- (xxiii) and the active pariciple suffix *-pA (iii) (Ramstedt 1917). (xix) *-mtV (ordinal numeral): e.g., PU *wij(i)t(t)i ‘five / ten’ → *wij(i)t(t)i-mtV > North Saami viđát, Finnish viides : viidente- ‘fifth’, Nganasan biimti-ǝ ‘tenth’. C. Denominal verbs (xx) *-tA- (general verbalizer): e.g., PU *suŋi ‘summer / thaw’ → *suŋita- > Finnish suvea- ‘get mild (of weather)’, East Khanty joŋǝt- ‘melt (of snow); come (of summer)’, Taz Selkup tɑŋɨrɨ- ‘spend the summer (somewhere)’; PU *nimi ‘name’ → *nimtä- > North Saami navdit, Meadow Mari lymdəә-, Tundra Nenets nʲumtʲe- ‘give a name’. — Presumably the same suffix as the deverbal causative suffix *-tA- (xxvii). (xxi) *-ji- (general verbalizer): e.g., PU *muna ‘egg’ → *munaji- > North Saami mon'ne-, Finnish muni- ‘lay eggs’; PU *nimi ‘name’ → *nimiji- > Izhma Komi ɲimjɨ-, Irtysh Khanty neːmǝj- ‘give a name’. (xxii) *-li- (general verbalizer): e.g., PU *kama ‘crust, rind’ → *kamali- > Meadow Mari komle- ‘cover, give a cover, bind’, Hungarian háml-ik ‘peel, flake off, come off in layers’; PU *lämi ‘broth, soup’ → *lämili- > Konda Mansi løæməәl- ‘cook soup’, Tundra Nenets jewǝl- ‘mix (fluids), dilute’. — Possibly the same suffix as the deverbal aspectual suffix *-li- (xxxi). (xxiii) *-mi- (transformative): e.g., PU *pidi- ‘long / high’ → *pidimi- > Finnish pitene- ‘become longer’, Tundra Nenets pʲirǝm- ‘become higher’. (xxiv) *-mtA- (factitive): e.g., PU *pidi- ‘long / high’ → *pidimtä- > Finnish pidentä- ‘make longer’, Tundra Nenets pʲirǝmtʲe- ‘make higher’. — A bimorphemic suffix consisting of *-mi- (xxiii) and *-tA- (xxvii). (xxv) *-mtAw- (transformative / stative?): PU *pidi ‘long / high’ → *pidimtäw- > Fin pidentyä ‘become longer’, NenT pʲirᵊmtʲo- ‘rise, tower (above), appear tall’. — A trimorphemic suffix consisting of *-mi- (xxiii), *-tA- (xxvii) and *-w- (xxx). (xxvi) ?*-o- (unclear semantics): e.g., PU *tora ‘quarrel, fight (N)’ → ?*toro- > North Saami doarru- ‘fight’, Tundra Nenets taro- ‘wrestle’. 38  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   D. Deverbal verbs: (xxvii) *-tA- (causative): e.g., PU *peji- ‘cook, be cooking (INTR)’ → *pejtä- > Erzya Mordvin pidʲe-, Tundra Nenets pʲirʲe- ‘cook (TR)’. — Cf. denominal *-tA- (xx). (xxviii)*-ptA- (causative): e.g., PU *sewi- ‘eat’ → *se(w)ptä- > Vote süttä- ‘set on fire’, Komi sot- ‘burn (TR)’, East Khanty læːwǝt- ‘feed; burn (TR)’. — Possibly a bimorphemic suffix consisting of an unclear morpheme *-p- and the causative suffix *-tA- (xxvii). (xxix) *-ktA- (causative): e.g., PU *kuma- ‘fall over’ → *kumakta- > North Saami gomih-, Meadow Mari kuməәktəә- ‘overturn, turn upside down (TR)’, Tundra Nenets xəәwᵊdɑ- ‘knock down’. — Possibly a bimorphemic suffix consisting of an unclear morpheme *-k- and the causative suffix *-tA- (xxvii). (xxx) *-w- (stative / automative passive): e.g., PU *kaďa- ‘leave (TR)’ → *kaďaw- > North Saami guđ'đo- ‘be left’, Estonian kadu- ‘disappear’, Tundra Nenets xɑjo- ‘stay’. (xxxi) *-li- (momentative / inchoative?): e.g., PU *puwa- ‘blow’ → *puwali- > Meadow Mari puɑl-, Far East Khanty poɣǝl- ‘blow (once); swell’, Nganasan hüolǝ- ‘blow’, Tym Selkup pul- ‘swell’. (xxxii) *-lta- (momentative?): e.g., PU *suxi- ‘row’ → *suxilta- > Meadow Mari ʃuɑltəә-, Kazym Khanty ɬɔːwǝɬt- ‘row’. — A bimorphemic suffix consisting of *-li- (xxxi) and *-tA- (xxvii). (xxxiii)*-nti- (frequentative / imperfective?): e.g., PU *peli- ‘be afraid’ → *pelinti- > North Saami balad-, Moksha Mordvin pelʲǝnʲdʲǝ- ‘be afraid (frequentative)’, Tundra Nenets pʲiːnǝ- ‘be afraid’. (xxxiv) *-kśi- (frequentative?): e.g., PU *ńali- ‘lick’ → *ńalikśi- > Finnish nuoleksi- ‘lick (repeatedly)’, Komi nʲulsʲɨ- ‘lick (repeatedly); lick oneself’, Kazym Khanty nʲoɬɛs- ‘lick, lick off’ (xxxv) *-ji- (unclear semantics): e.g., PU *woppi- ‘inspect, go and see’ → *woppiji- > Finnish oppi- ‘learn’, Nizyam Khanty wɔːpij- ‘spy; visit’, Konda Mansi wopj- ‘notice’. — Possibly the same suffix as denominal *- ji- (xxi). In general, the derivational suffixes reconstructed above represent rather well the types of derivatives that are also synchronically widely attested in Uralic. A notable 39  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   gap, however, is that no suffixes for forming adverbs have been reliably reconstructed, but some fossilized formations nevertheless suggest that such may have existed (cf., e.g., PU *mentälä(-n) ‘past / along’ ← *mentä- ‘miss, cause to go’ mentioned in 1.4.2). The exact semantic functions of many of the suffixes remain unclear, and further study is needed. Of course, some derivational suffixes may have been polyfunctional in the proto-language already. A case in point are the three denominal verb suffixes *-tA-, *-ji- and *-li-, for which hardly more than a general verbalizing function can be reconstructed (Laakso 1997). Besides derivation, compounding must also have been a highly productive method of word formation in Proto-Uralic. However, individual compound words can be reliably reconstructed for Proto-Uralic only if they have been retained in an opaque, monomorphemized form in all or nearly all branches, and very few attested compounds fulfill these conditions. Unsurprisingly, all known examples are nouns. A few words for body parts appear to derive from compounds with the word *li̮ wi ‘bone’ as their head. These examples suggest that endocentric compounds could be formed by simple conjunction of two noun stems, without any specific morphological processes: – South Saami boelve, Finnish polvi, Tundra Nenets puːliᵊ ‘knee’ < ?*po/ux/wi- ‘knee’ + *li̮ wi ‘bone’ (Aikio 2012: 230). – Komi ɕili ‘neck, withers’, East Khanty sæːwǝl ‘neck’ < *śepä ‘neck’ + *li̮ wi ‘bone’; the compound is preserved transparent in North Mansi sip-luw ‘neck’. – Meadow Mari oŋəәl-ɑʃ ‘lower jaw, chin’, Surgut Khanty uːŋǝɬ ‘jaw, jawbone’ < *aŋi ‘mouth’ + *li̮ wi ‘bone’. There is also one apparent case of an obscured compound with the relational noun stem as the head: the Uralic word for ‘armpit’ (Finnish kainalo, Udmurt kunul, North Mansi xɑːnl, Hungarian hónalj, etc.) probably consists of ?*ka/ojV-n (an otherwise unattested noun) and the relational noun *i̮ la ‘space under or below’. This example differs structurally from the previous type in that the modifier seems to be in the genitive case form (with the suffix *-n; see 1.4.3). Nearly all examples involve endocentric compounds. However, an archaic exoncentric compound is probably involved in Hungarian hajnal ‘dawn, daybreak’ and East Khanty kuːnʲǝl ‘red sky (at dawn or dusk)’, apparently consisting of *kaja-n ‘sun-GEN’ and *i̮ la ‘space under or below’ (cf. English sundown ‘sunset’), but due to 40  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   its limited distribution this does not qualify as a true Proto-Uralic item. Interestingly, the most reliably reconstructed Proto-Uralic compound word is neither endocentric nor exocentric. There is a cognate set of words meaning ‘mother-in-law’ (Finnish anoppi, Surgut Khanty ontǝp, North Mansi ɔːnip) and ‘father-in-law’ (Tundra Nenets ŋinǝbǝ, Nganasan ŋinǝbǝ) which reflects a copulative compound *i̮ na-i̮ ppi, consisting of PU *i̮ na- ‘mother-in-law’ (> North Saami vuoni) and *i̮ ppi ‘father-in-law’ (> North Saami vuohppa, Finnish appi, Hungarian ipa) (Janhunen 1981: 227–228). The original meaning of the compound was probably ‘parents-in-law’. Similar copulative compounds are widely attested in the Uralic languages (e.g., Komi ɑj-mɑm, Konda Mansi jeɣ-sʲyk, Nganasan dʲesɨ-nʲemɨ “father-mother” = ‘parents’), and this cross- linguistically less common compound type thus turns out to be a Proto-Uralic archaism. 1.5 Syntax Compared to other levels of language, not much is known about Uralic historical syntax. Very little work has been done on syntactic reconstruction even within low- level branches of the family, to say nothing of more remote time depths. In fact, much of what has been claimed regarding Proto-Uralic syntax is not based on any systematic research at all, but instead on educated guesswork or even speculation. This being the case, only some most general remarks on syntax will be made here. The following sketchy overview is essentially based on an extrapolation of those syntactic features that are most widespread and give the impression of being most archaic in the Uralic languages. As regards constituent order, Proto-Uralic was most obviously an SOV language with postpositions. The morphosyntactic alignment of Proto-Uralic was almost certainly of the nominative-accusative type, as is the case with all modern Uralic languages with the sole exception of East Khanty which makes limited use of an ergative-like construction (the ergative suffix derives from the Uralic locative case in *-nA; see 32.4.2.1 and 44.1). There have been attempts to uncover traces of an earlier ergative or split-ergative system in Uralic (e.g., de Smit 2014); the evidence is far from conclusive, but one cannot exclude the possibility that such a system indeed existed in Pre-Proto-Uralic and the nominative-accusative alignment of Proto-Uralic was an innovation. Even if this is the case, the East Khanty ergative construction is 41  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   best explained as a recent innovation; the claim that it is a vestige of more original ergative alignment in Uralic (Havas 2006) lacks substantiation. The core arguments were at least partly marked by case: subject arguments appeared in the nominative case, whereas direct objects of transitive verbs could take the accusative case (with the suffix *-m in the singular). However, many Uralic languages display a differential object marking with nominative objects alongside accusative ones (see 44.4.1 and 54.2.4.1). Finnic and Nenets indicate that objects of second-person imperative forms took the nominative case; this feature is, in fact, one of the very few morphosyntactic rules whose reconstruction to Proto-Uralic has been argued in detail (Janhunen 2002). As regards noun phrase structure, adnominal modifiers precede nouns in all Uralic languages, and the situation in Proto-Uralic was evidently the same. In morphological terms, adjectives were a subclass of nouns (see 1.4.2), but modifier adjectives most likely took the unmarked nominative singular form and displayed no agreement with the head noun in either case or number: e.g., *wuďi kota ‘new tent’ SG.NOM : *wuďi kota-m SG.ACC : *wuďi kota-t PL.NOM (> South Saami orre gåetie : orre gåetie-m : orre gåetie-h). In contrast with adjectives, true nouns prototypically took the genitive case (with the suffix *-n in the singular) when appearing as adnominal (possessor) modifiers. Postpositional complements were marked with the genitive case, and all known Proto-Uralic postpositions were actually local case forms of relational nouns (see 1.4.2). Thus, postpositional phrases like *kota-n taka-na [tent- GEN behind-LOC] ‘behind the tent’ and *jäŋi-n üli-ŋ [ice-GEN on-LAT] ‘onto the ice’ appear not to have differed formally from noun phrases like *śodka-n pesä-nä [goldeneye-GEN nest-LOC] ‘in the goldeneye’s nest’ and *emä-n süli-ŋ [mother-GEN lap-LAT] ‘into mother’s arms’. This parallelism is, however, not retained in today’s Uralic: the primary local case suffixes have tended to become lost or secondarily develop more abstract grammatical functions, which in turn has rendered the morphosyntactic structure of postpositional phrases opaque. Simple verbal predicates consisted of a single finite verb form, marked by conjugation for tense, mood, subject person, and possibly also the presence and number of a definite object (see 1.4.4). Complex predicates involving e.g. modals must also have occurred, but essentially nothing is known of their syntactic properties. Clausal negation was marked with a periphrastic construction that consisted of a finite form of the negative auxiliary *e-, which probably carried all marking of grammatical categories associated with verbs, followed by the lexical verb in the connegative form 42  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   with the suffix *-k. Negative imperatives were based on a similar periphrastic construction with the prohibitive auxiliary ?*älä- ~ ?*elä-. The later development of clausal negation in the Uralic languages shows a general trend of gradual reduction in the verbal nature of the negator. The Ugric languages have proceeded farthest on this path by innovating particle negation, but also in the more western Uralic languages one can observe a drift in the same direction: tense and mood marking has often shifted from the negative auxiliary to the lexical verb, and even subsystems of particle negation have been created by obliterating subject person marking in parts of the paradigm of the negative auxiliary. For instance, Estonian has generalized original NEG.3SG ei as a particle marking clausal negation, but still retains the prohibitive auxiliary as a verb with distinct subject person forms. Samoyed languages, in contrast, have more consistently retained the marking of grammatical categories on the negative auxiliary. Adjectives and noun phrases could act as nominal predicates without any preceding copula, as in *kota wuďi [tent new] (> South Saami gåetie orre ‘the tent is new’), *kota wara-n ül-nä [tent mountain-GEN on-LOC] (> South Saami gåetie vaerien nelnie ‘the tent is up on the mountain’). In order to mark tense, mood or negation, however, a construction with the copula *woli- ‘be’ had to be used. As mentioned in 1.4.3, in Mordvin and Samoyed nominal predicates take subject person and number markers and past tense markers identical to verbs, but this feature is probably a secondary innovation. In addition to the basic transitive, intransitive and nominal-predicate sentences, one can also reconstruct a more marginal agentless sentence type predicating natural phenomena, with the bare predicate verb in 3SG (e.g., Finnish sataa ~ Forest Nenets xɑːrɑː [rain.3SG] ‘it is raining’). Beyond this, the syntactic properties of other sentence types remain largely unknown. A case in point is predicative possession, which is expressed by tremendously diverse structures in the modern Uralic languages, including at least transitive, nominative, genitive, locational and adjectivized possessives; some of these types have emerged multiple times in the Uralic languages, as parallel innovations with distinct material sources. The picture is heterogeneous also in the case of several other sentence types, including comparatives, reflexives, reciprocals, causatives, passives, and polar questions; the morphosyntactic expression of these categories in Proto-Uralic thus remains unknown. 43  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   As regards subordinate clauses, the general view is that Proto-Uralic lacked both conjunctions and relative pronouns, and that both relative and complement clauses were built around non-finite nominalized verb forms (Janhunen 1982: 39). This view is in agreement with the generally archaic look of non-finite subordinate clauses and the clearly recent origin of most conjunctive words in the Uralic languages. However, the fact that no conjunctions or relative pronouns have been reconstructed to Proto- Uralic cannot, in itself, prove that such words did not exist at all. No concrete reconstructions of the structure of the assumed non-finite subordinate clauses have been presented either; the features of clause embedding in Proto-Uralic thus remain uncertain. 1.6 Lexicon 1.6.1 State of research Progress in Uralic lexical reconstruction has been intimately connected with advances in historical phonology. Prior to the 1980s etymological research was encumbered by the lack of a well-defined reconstruction of Proto-Uralic phonology. Therefore, sound correspondences were often dealt with in an imprecise and impressionistic manner, especially in the case of Khanty, Mansi, and Samoyed, whose phonological development remained the least well understood. The etymological dictionary Uralisches Etymologisches Wörterbuch (Rédei 1988–1991) presents a comprehensive synthesis of the results of this era of research, but does not reflect the major turn in the field in the 1980s. In Janhunen (1981) and Sammallahti’s (1988) studies on phonological and lexical reconstruction the focus shifted from search for new lexical comparisons to a critical reevaluation of proposed Uralic etymologies. The establishment of exact criteria of regular sound correspondence decreased the number of reliable reconstructs dramatically, as shown by a comparison of the numbers of etymologies accepted by Rédei & al. (1988) on the one hand, and Sammallahti (1988) on the other (note that both references operate with the traditional taxonomic distinction between “Uralic”, “Finno-Ugric” and “Finno-Permic” discussed in section 2): Uralic: Finno-Ugric: Finno-Permic: total: Rédei & al. (1988) 284 419 197 900 Sammallahti (1988) 124 267 142 533 44  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   The most striking feature of Janhunen and Sammallahti’s revised corpus of etymologies was that Samoyed appeared to share very little lexical material with the rest of Uralic. Janhunen (1998: 475) maintains that Samoyed only has about 150 underived word-stems of Uralic origin, which in his view supports the traditional binary division of Uralic into Samoyed and a “Finno-Ugric” branch. However, the established rigorous framework of regular sound correspondences has also greatly facilitated subsequent discovery of new etymologies, in the field of Samoyed etymology in particular (see, e.g., Aikio 2002; 2006a; forthcoming). The estimated number of reliable Uralic etymologies for Samoyed words might therefore be doubled to 300. Knowledge of the Proto-Uralic lexicon has significantly increased since the word-lists of Janhunen (1981) and Sammallahti (1988), and this new picture differs dramatically from that presented in Rédei & al. (1988). Unfortunately, however, there is no up-to-date etymological dictionary of Uralic. 1.6.2 Semantic overview The bulk of the reconstructed Proto-Uralic lexicon consists of monomorphemic word stems. Derivatives and compounds must have been numerous, but only very few of them can be reliably reconstructed (see 1.4.5). Applying the distributional criteria outlined in section 1.2 and reasonably strict criteria of phonological and semantic correspondence, approximately 500 Proto-Uralic lexemes can be reconstructed. The material is heavily concentrated in basic vocabulary, and obviously represents only a fraction of the actual lexicon of the proto-language. In fact, the bulk of the material can be classified in only a couple of broad semantic groups. About 120 lexemes are connected with body and mind. Roughly half of them are words for body parts (e.g., *ojwa ‘head’, *śilmä ‘eye’, *i̮ pti ‘hair on the head’, *käli ‘tongue’, *piŋi ‘tooth’, *künči ‘nail’, *keti ‘skin’, *mi̮ ksa ‘liver’, *śali ‘gut’, *weri ‘blood’). There are also words for physiological states (e.g., *elä- ‘live’, *kali- ‘die’, *tarki- ‘shiver’, *poči- ‘be sick or exhausted’), body functions (e.g., *kusi- ‘cough’, *kuńśi- ‘urinate’), sensory experiences (e.g., *näki- ‘see’, *kuwli- ‘hear’, *ipsä- ‘smell’), cognitive functions (e.g., *tumti- ‘know’) and mental states (e.g., *peli- ‘be afraid’). A surprisingly comprehensive set of “mouth verbs” can be reconstructed: *sewi- ‘eat’, *puri- ‘bite’, *soski- ‘chew’, *jürä- ‘gnaw’, *ji̮ xi- / *juxi- ‘drink’, *ńimi- ‘suck’, *ńali- ‘lick’, *ńäli- ‘swallow’, *pala- ‘eat up’, *appa- ‘devour, eat voraciously’, *puwa- ‘blow’, *śülki- ‘spit’. 45  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   About 100 lexemes are verbs for various elementary and universally known actions not covered by the previous category. These include, for instance, verbs for motion (e.g., *tuli- ‘come’, *meni- ‘go’, *läkti- ‘leave, go out’, *wi̮ lka- ‘come down’, *uji- ‘swim’), transferring objects (e.g., *wixi- ‘take, transport’, *toxi- ‘bring’, *kanta- ‘carry’, *kaďa- ‘leave (TR)’, *pi̮ ni- ‘put’), object manipulation (e.g., *panča- ‘open’, *ńükä- ‘pull out’, *peksä- ‘beat’, *puśa- ‘wring, squeeze’, *mośki- ‘wash’), and also a few social actions (e.g., *jupta- ‘speak, tell’, *kaśa- ‘give as a gift’, *i̮ śiw- ‘set up a camp’, *kätki- ‘tuck (a child) in the cradle’). About 120 lexemes are connected with the natural environment. These include words connected with the diurnal and seasonal cycles (e.g., *päjwä ‘day / warmth’, *eji ‘night’, *kaja ‘sun / dawn’, *sükiśi ‘autumn’, *tälwä ‘winter’, *i̮ di ‘year’), basic elements and objects of the natural world (e.g., *weti ‘water’, *lumi ‘snow’, *jäŋi ‘ice’, *kiwi ‘stone’), topographic terms (e.g., *tuwi ‘lake’, *jäŋkä ‘bog’, *li̮ nti ‘lowland’) and weather (e.g., *lupsa ‘dew’, *künti ‘fog’, *śada- ‘rain’, *purki ‘snowstorm’). Words for flora are few, and mainly include tree names (e.g., *kowsi ‘spruce’, *ďi̮ mi ‘bird-cherry’, *si̮ ksa ‘Siberian pine, Pinus sibirica’). The last word is relevant for narrowing down possible homeland hypotheses, as the tree grows widely in Siberia, but in Europe only in a limited region in the proximity of the central and northern parts of the Ural mountains. Also an alleged Proto-Uralic word for ‘Siberian fir, Abies sibirica’ (cf. Komi nʲɨv, Taz Selkup nʲulqɨ, etc.) has frequently featured in homeland theories, but it involves too many phonological irregularities to be reliably reconstructed (contra J. Häkkinen 2009: 36). As regards animals, there are general words connected with fish (e.g., *kala ‘fish’, *kuďi- ‘spawn’, *śi̮ mi ‘scales, fish skin’) and birds (e.g., *pesä ‘nest’, *muna ‘egg’, *tulka ‘feather’), but species names are few. Most frequent are bird names (e.g., *śäkśi ‘osprey’, *śodka ‘goldeneye’, *kurki / *ki̮ rki ‘crane’, *lunta ‘goose’, *epik(i) ‘eagle owl’, *ti̮ ktV / *tuktV ‘black-throated loon’), in addition to which there are five words for mammal species (*ńoma(-la) ‘hare’, *ńukiś(i) ‘marten / sable’, *ora(-pa) ‘squirrel’, *śijil(i) ‘hedgehog’, *šiŋir(i) ‘mouse’), three fish names (*särki ‘roach / ruffe’, *säwni / *sewni ‘ide’, *totki ‘tench’), and a couple of words for reptiles and insects (e.g. *küji ‘snake’, *täji ‘louse’). Interestingly, most words denote species that must have been of little or no economic significance; words for animal species that were important for subsistence were apparently more easily replaced. The three broad semantic groups outlined above cover about two thirds of the reconstructed lexicon. On the other hand, some parts of basic vocabulary remain 46  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   poorly known, kinship terms being a case in point. Words for close blood kin mostly remain unclear: the word for ‘mother’ was *emä / *ämä, but words for ‘father’, ‘brother’ or ‘sister’ cannot be reliably reconstructed. Some cognate sets in this sphere involve semantic variation which makes the original meaning difficult to determine: cf. *čečä ‘uncle / paternal uncle / maternal uncle’, *koska ‘maternal aunt / grandmother’, *ekä ‘father / paternal uncle / grandfather’. Interestingly, words for in- laws seem to have been better preserved: cf. *i̮ na ‘mother-in-law’, *i̮ ppi ‘father-in- law’, *mińä ‘daughter-in-law’, *wäŋiw ‘son-in-law’, *ańi ‘sister-in-law’, *käliw ‘sister- or brother-in-law’, *nataw ‘sister- or brother-in-law’. Besides kinship terminology very little vocabulary connected with social relationships can be reconstructed. Malicious or hostile relationships are evidenced by *sala- ‘steal’, *tora ‘quarrel, fight’ and *śoďa ‘warfare’. No vocabulary pertaining to any kind of social organization is known. There are about 50 lexemes broadly connected to material culture and technology. This vocabulary offers little basis for linguistic paleontology, as it merely testifies of a level of technology that by far predates Proto-Uralic. The most extensive subgroups are formed by words connected to use of fire (*tuli ‘fire’, *śüďi ‘charcoal’, *küči- ‘smolder’, *kupsa- ‘go out (of fire)’, *peji- ‘cook’, *japśi- / *jepśi- ‘roast on a spit’, *lämi ‘broth, soup’) and bows (*ji̮ ŋsi / *joŋsi ‘bow’, *jänti(ŋi) ‘bowstring’, *ńi̮ li ‘arrow’, *lewi- ‘shoot’). Some other notable words in this sphere are *äjmä ‘needle’, *pura ‘drill’, *ďimä ‘glue’ and *pata ‘clay pot’ (for discussion on the last word see Kallio 2006a: 5–6). Means of transportation included skis (*suksi) and boats (cf. *suxi- ‘row’, *tukta ‘seat in a boat’; no word for ‘boat’ can be reconstructed), possibly also sleds (?*śona). Fishing methods included use of nets (cf. *kala- ‘fish with a net’, *kulta- ‘fish with a dragnet’, *pado ‘fish weir’). Evidence of immaterial culture is very limited, but a couple of lexical items can be seen as pointing to a shamanistic system of beliefs and practices. The concept of soul dualism, widely attested among Uralic peoples, likely dates back to Proto-Uralic, as at least two words for ‘soul’ can be reconstructed. Samoyed and Saami share the word *wajŋi ‘breath; soul, spirit’, which was probably used in reference to the ‘breath-soul’, believed to be bound to the living body and only leaving at the moment of death (note that Permic and Ugric languages share another word for ‘breath’ and ‘soul’, *lewli(w), but it may have originally meant ‘vapor’). The ‘breath-soul’ is distinct from the ‘shadow-soul’ thought to be able to leave the body, as during dreaming, unconsciousness or a shaman’s spirit journey. The word for the latter was 47  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   *eśi / *iśi / *ićći (phonological reconstruction is difficult); the meaning ‘shadow-soul’ is only preserved in Khanty and Mansi, whereas more western languages grammaticalized the word as a reflexive pronoun (‘my soul’ > ‘myself’, etc.). The Proto-Uralic word for shaman (?*nojta) is probably preserved in North Saami noaidi ‘shaman’, Finnish noita ‘witch; shaman’, and Tavda Mansi /nɛjt/ ‘shaman’, even though this etymology is somewhat uncertain. Some kind of shamanistic practice was referred to by the verb *jada-, but the cognates show much semantic variation (cf. Erzya Mordvin jɑdɑ- ‘conjure, do magic, bewitch’, East Khanty jɔːl- ‘tell fortunes, shamanize’, Ket Selkup tʲɑːrəә- ‘curse; quarrel’). The verb *kixi- ‘court (of birds)’ has also the widely attested parallel meaning ‘sing a shamanistic song’, suggesting that it referred to states of both sexual and spiritual excitement already in Proto-Uralic. Shamans’ altered states of consciousness may have been induced by use of psychedelic mushrooms (presumably the fly agaric, Amanita muscaria), referred to by the Indo-Iranian loanword *pi̮ ŋka (see 6.3). The most significant conclusions regarding Proto-Uralic society can, however, be based on what is absent in the reconstructed lexicon. Most evident is the complete lack of any vocabulary connected with agriculture, which indicates that Proto-Uralic was spoken in a hunter-gatherer speech community. Contrary to this communis opinio, J. Häkkinen (2009: 28–30) maintains that four such lexical items could be reconstructed to Proto-Uralic after all, but the material does not stand critical scrutiny: the alleged cognate sets for ‘sheep’, ‘wheat / barley’ and ‘flour’ suffer from phonological irregularities and limited distribution, whereas the suggested word for ‘butter’ (*waji) seems to have originally meant ‘grease’. An issue of more disagreement is whether any vocabulary connected to metallurgy can be traced back to Proto-Uralic. It has often been claimed that a word meaning ‘metal’ or ‘copper’ can be reconstructed on the basis of forms such as North Saami veaiki, Finnish vaski ‘copper, bronze’, Hungarian vas, and Nganasan basa ‘iron’ (e.g., Kallio 2006: 6–8). However, also this etymology involves many phonological irregularities, which suggests it is a “Wanderwort” that spread after the disintegration of the proto- language (Aikio 2015: 42–43). The numeral system forms an interesting and much-discussed part of the Proto- Uralic lexicon (see, e.g., Honti 1993; Janhunen 2000). One can identify cognate numerals up to seven, even though they pose some difficulties for phonological reconstruction: *ükti / *äkti ‘1’, *kVkta / *kVktä ‘2’, *kolmi / *kulmi / *kurmi ‘3’, *neljä ‘4’, *wij(i)t(t)i ‘5’, *kuw(V)t(t)i ‘6’, *śäjśimä / *śä(j)ććimä ‘7’. These numerals 48  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   are retained in the “Finno-Ugric” languages, with the exception that Khanty and Hungarian replaced ‘1’ with a stem of unknown origin (*eďi) and all Ugric languages replaced ‘7’ with Indo-Iranian loanwords (e.g., Hungarian hét). Higher numerals are etymologically more diverse, but *luka ‘10’ (with a morphologically opaque relationship to the verb *luki- ‘count’) can be reconstructed on the basis of Saami, Mari and Mansi. Moreover, all “Finno-Ugric” languages share the Indo-Iranian loanword *śi̮ ta ‘hundred’ (see 1.6.3). In Samoyed, few cognates for the numerals are found, which according to Janhunen (2000: 60–61) implies that “the numeral system in the Proto-Uralic stage was still incipient”. Nevertheless, Samoyed retains three Uralic numerals: Proto- Samoyed *kitä ‘2’, *wüǝt ‘10’ (cognate with Finno-Ugric numerals for ‘5’), and *säjʔwǝ ‘7’. The Uralic origin of the last has remained a matter of dispute, but its strikingly regular correspondence with Proto-Finnic *säiccemä ‘7’ can hardly be due to chance – especially as the alleged alternative etymology connecting *säjʔwǝ to Tocharian or other Indo-European numerals for ‘7’ (Janhunen 1983: 119; Napoľskikh 2001: 373) has turned out to be phonologically untenable (Kallio 2004: 131–132). Thus, Proto-Uralic must have had a well-developed system of low numerals after all, as one can hardly imagine that a language with a defective numeral system nevertheless possessed words for ‘seven’ and ‘five’ (or ‘ten’). The innovative Samoyed numerals show phonological structures quite foreign to Uralic vocabulary (Proto-Samoyed *nakur ‘3’, *tättǝ ‘4’, *sǝmpǝläŋkǝ ‘5’, *mǝ̑ktut ‘6’), which suggests that they may have been borrowed from an unknown source. 1.6.3 External connections The more remote etymologies of most Proto-Uralic words remain entirely unknown, as they appear to be monomorphemic and nothing is known of the further genetic relationships of the family. Existing theories of long-range genetic connections are generally based on evidence of poor quality; for example, the untenability of attempts to relate Uralic to the “Nostratic” macrofamily is demonstrated in ample detail by Campbell (1998b). For a fraction of Proto-Uralic word stems, however, an Indo-European loan etymology has been proposed. It is unanimously recognized that the Uralic family came to contact with Indo-European already in an early stage of its development, and some scholars maintain that the contacts reflected by the oldest loanwords took place between the respective proto-languages of the two families already. The etymological 49  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   material evoked in support of the latter theory (e.g., Koivulehto 1991; 2001) is by no means free of problems, however (Campbell & Garrett 1993). Nevertheless, there certainly are a few striking lexical parallels, and some of them even include apparent reflexes of reconstructed Indo-European laryngeal phonemes in Uralic; for example: PU *aja- ‘drive; flee’ ~ PIE *h2aǵ- (cf. Sanskrit ájati ‘drives’) PU *kaja ‘dawn / sun’ ~ PIE *h2ay-en/r- (cf. Avestan aiiarǝ ‘day’) PU *kulki- ‘go, run, flow’ ~ PIE *kwelh1-e- (cf. Sanskrit cárati ‘moves, walks’) PU *teki- ‘do; put’ ~ PIE *dheh1- (cf. Sanskrit dádhāti ‘puts’) PU *toxi- ‘bring ~ PIE *doh3- ‘give’ (cf. Sanskrit dádāti ‘gives’) PU *weti ‘water’ ~ PIE *wed-en/r- (cf. Hittite wedār ‘water’) How such parallels are to be interpreted remains a matter of controversy. The number of plausible etymological comparisons is quite small, and as far as they are correct in the first place, the words in question might also have been borrowed into an early Uralic dialect continuum that postdated the disintegration and spread of Proto-Uralic. This interpretation is especially attractive because Proto-Uralic and Proto-Indo- European represented language typologies so radically different that they are simply unlikely to have originated as neighbouring languages in the same linguistic area (Janhunen 2001a); instead, in typological terms (Proto-)Uralic is strikingly close to the so-called Altaic languages and one can even speak of a “Ural-Altaic” typological zone spanning across northern Eurasia (Janhunen 2007b). Notably, however, Proto- Uralic shares no evident material parallels with Altaic language families, making the Ural-Altaic theory invalid in the genetic sense. Another possibility is to assume that Proto-Indo-European words were not borrowed directly into Uralic, but mediated by unknown intermediate languages. Such a hypothesis could also better explain those apparent lexical parallels which involve a less precise phonological match, such as PU *nimi ‘name’ ~ PIE *h3noh3men- ‘name’. However, also chance similarities probably feature in the scarce material, especially as the more complex phoneme system of Proto-Indo-European can be represented in Uralic by way of various simplifying sound substitutions, which makes possible matches easier to find. A particularly likely case of a coincidental correspondence is PU *koki- ‘check, go and see’, which Koivulehto (1991: 44–47) considers a loan from PIE *h3okw-(ye/o-) ‘see’ (cf. Ancient Greek ὄπωπα ‘watch, observe, view’, ὄσσοµαι ‘look, forebode’); but because he derives PU *k from as many as nine distinct Indo- 50  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   European consonant phonemes (*k, *g, *gh, *kw, *gw, *gwh, *h1, *h2, *h3), a very large number of theoretical forms could be matched with a Uralic word stem with two velar stops. While only few plausible lexical parallels point to a level of reconstruction corresponding to Proto-Indo-European, a rather large number of old Indo-Iranian loanwords is found in the Uralic languages. Most such loans clearly postdate Proto- Uralic, but a small number of apparent Indo-Iranian loans are technically reconstructible into Proto-Uralic. The clearest cases are the following: PU *ora ‘awl’ ~ PII *ārā (cf. Sanskrit ā́ rā ‘goad, awl’) PU *ki̮ nta ‘tree trunk’ ~ PII *skandha- (cf. Sanskrit skandhá- ‘shoulder; tree trunk; large branch’) PU *mekši ‘bee’ ~ PII *mekš- (cf. Sanskrit mákṣ- ‘fly, bee’) PU *meti ‘honey’ ~ PII *medhu- (cf. Sanskrit mádhu- ‘honey’) PU *pi̮ ŋka ‘psychedelic mushroom’ ~ PII *bhangā- (cf. Sanskrit bhaṅgā́ - ‘hemp; an intoxicating drink made of hemp’) PU *śarwi ‘antler, horn’ ~ PII *ćrwā- (cf. Avestan sruuā- ‘horn; nail’) PU *śi̮ ta ‘hundred’ ~ PII *ćatam- (cf. Sanskrit śatám ‘hundred’) These loanwords do not occur in Samoyed languages, which in the traditional taxonomic framework (see 1.2) was seen as evidence for the view that “Finno-Ugric” came into contact with Indo-Iranian after the Samoyed branch had split off. On the other hand, in the context of his revised taxonomy, J. Häkkinen (2009: 20–25) views such loans as implying that Proto-Uralic disintegrated only after the adoption of the earliest Indo-Iranian loanwords. Both interpretations are unconvincing, however. The great majority of the Indo-Iranian loanwords show phonological irregularities or limited distribution within Uralic, and thus must have spread between already distinct dialects or languages after the disintegration of Proto-Uralic. The few exceptions to this general pattern can simply be interpreted as words that became adjusted to the sound correspondences occurring in native vocabulary as they spread in an early Uralic dialect continuum; analogous phenomena occur in the sound correspondences displayed by Nordic and Finnic loanwords in Saami languages, for instance (Aikio 2006b; 2007). Summing up, there are obviously some very old Indo-European and specifically Indo-Iranian loanwords in Uralic, but they are unlikely to reflect a direct contact of 51  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Proto-Uralic with Proto-Indo-European or its descendant Proto-Indo-Iranian. However, the two language families must have come into contact at a very early phase of their evolution, soon after Proto-Uralic disintegrated into a dialect continuum. More exact dating of the onset of these contacts remains difficult, but at any rate the material confirms the time depth of the Uralic family as roughly analogous to that of Indo-European. Proto-Uralic also shows some notable lexical similarities with Yukaghir, a Siberian language family that was formerly spread at least as far as the Lena River in the west but today survives only as two moribund languages in the basin of the Kolyma River of the Russian Far East. The number of promising Uralic-Yukaghir lexical parallels is small, hardly more than 30 words, but the material does include quite appealing similarities in basic vocabulary. The following examples are among the most striking (regarding Yukaghir reconstructions and lexical data, see Nikolaeva 2006): PU *aŋi ‘mouth, opening’ ~ PYuk *aŋa ‘mouth’ PU *emä / *ämä ~ PYuk *eme ‘mother’ PU *i̮ la- ~ PYuk *āl- ‘place under or below’ PU *kälä- ‘wade / rise’ ~ PYuk *kile- ‘wade’ PU *käliw ‘brother- or sister-in-law’ ~ PYuk *keľ- ‘brother-in-law’ PU *kani- ‘go away’ ~ PYuk *qon- ‘go’ PU *koji ‘male, man, husband’ ~ PYuk *köj ‘fellow, boy, young man’ PU *mälki ~ PYuk *mel- ‘breast’ PU *nimi ~ PYuk *ńim / *nim ‘name’ PU *ńali- ~ PYuk *ńel- ‘lick’ PU *pidi- ‘long / high’ ~ PYuk *puδe ‘place on or above’, *puδe-nmē- ‘tall, high’ PU *pi̮ ni- ‘put’ ~ PYuk *pöń- / *peń- ‘put; leave’ PU *sala- ~ PYuk *olo- ‘steal’ PU *sula- ~ PYuk *aľ- ‘melt, thaw’ PU *wanča(w) ~ PYuk *wonč- ‘root’ PU *wixi- ‘take, transport’ ~ PYuk *weɣ- ‘lead, carry’ Such lexical similarities have been seen as evidence of distant genetic relationship (see e.g. Collinder 1940), but newer research has also pursued other 52  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   explanations. A majority of Uralic-Yukaghir etymologies proposed earlier are quite evidently erroneous, but the material also includes some obvious borrowings from Uralic to Yukaghir. This view was first argued in detail by Rédei (1999), and further corroborated by J. Häkkinen (2012a, 2012b) and Aikio (2014). Phonological criteria suggest that the earliest source of Uralic loans in Yukaghir was Proto-Samoyed or Pre-Proto-Samoyed. Borrowing could also account for similarities in basic vocabulary. While this hypothesis is debatable, it would be even more problematic to interpret the words as genetic cognates when no system of regular sound correspondences has been established between Proto-Uralic and Proto-Yukaghir, nor has any clear evidence of shared morphology been presented (Aikio 2014; Rédei 1999). Last, one can mention the generally overlooked issue that Proto-Uralic also shows some interesting lexical similarities with Eskimo-Aleut. The idea of an “Eskimo-Uralic” genetic relationship has only been developed by some individual scholars, most notably Bergsland (1959) and Seefloth (2000), and mostly ignored by the mainstream of Uralic linguistics. It is interesting, however, that in an unbiased inspection the potential lexical correspondences between Proto-Uralic and Proto- Eskimo-Aleut do not seem to be any less promising than those between Proto-Uralic and Proto-Indo-European or Proto-Yukaghir. One could reasonably present something in the range of 20–30 comparanda, some of which even show possible vestiges of regular but phonetically opaque correspondences. An example of this is the hypothetical correspondence between PU intervocalic *-l- and Proto-Eskimo *-t, Aleut -c, which in turn may display a fossilized unproductive alternation with *-l- before a derivational suffix (regarding Eskimo-Aleut reconstructions and lexical data see Fortescue, Jacobson & Kaplan 1994): PU *i̮ la- ‘place under or below’ ~ PEsk *at(ǝ)- ‘down’, Aleut ac- ‘lower part, place below, bottom’; PEsk *alaq ‘sole’ (no Aleut cognate) PU *elä- ‘live’ ~ PEsk *ǝt(ǝ)- ‘be’ (no Aleut cognate) PU *tuli- ‘come’ ~ PEsk *tut- ‘arrive, land’ (no Aleut cognate); PEsk *tulaɣ-, Aleut cala- ‘land’ The most surprising correspondence involves matching pairs of homonyms that can, quite plausibly, be reconstructed into both Proto-Uralic and Proto-Eskimo-Aleut: 53  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   PU *kuda ‘morning, dawn’ ~ PEsk *qilaɣ- ‘sky’, Aleut qila-ẋ ‘morning, dawn’ PU *kuda- ‘weave’ (no Ugric or Samoyed cognates) ~ PEsk *qilaɣ- ‘knit, weave’ (no Aleut cognate) The meanings ‘weave’ and ‘morning, dawn’ appear to be completely unrelated, which implies that these must be instances of coincidental homonymy. It appears rather improbable that a correspondence like this was produced by chance, considering that lexical homonymy is a rare phenomenon, and only extremely few homonyms have even been reconstructed in the two proto-languages in the first place. The example strongly suggests that there has been some kind of historical connection between the two language families, but more exact conclusions cannot be drawn in the present state of research. The existence of such parallels suggests that future research should evaluate the possible connections of Proto-Uralic to Proto-Eskimo-Aleut as thoroughly as those to Proto-Indo-European and Proto-Yukaghir. 54  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   References Abondolo, Daniel (ed.) 1998a. The Uralic languages. London & New York: Routledge. Abondolo, Daniel 1998b. Introduction. — Daniel Abondolo (ed.), The Uralic languages. London & New York: Routledge. Pp. 1–42. Aikio, Ante 2002. New and old Samoyed etymologies. — Finnisch-Ugrische Forschungen 57: 9–57. Aikio, Ante 2006a. New and old Samoyed etymologies (part 2). — Finnisch-Ugrische Forschungen 59: 5–34. Aikio, Ante 2006b. On Germanic-Saami contacts and Saami prehistory. — Journal de la Société Finno-Ougrienne 91: 9–55. Aikio, Ante 2007. Etymological Nativization of Loanwords: a Case Study of Saami and Finnish. — Ida Toivonen & Diane Nelson (eds.), Saami linguistics. Amsterdam & Philadelphia: John Benjamins. Pp. 17–52. Aikio 2012 = Luobbal Sámmol Sámmol Ánte (Ante Aikio) 2012. On Finnic long vowels, Samoyed vowel sequences, and Proto-Uralic *x. — Tiina Hyytiäinen, Lotta Jalava, Janne Saarikivi & Erika Sandman (eds.), Per Urales ad Orientem: Iter polyphonicum multilingue. Festskrift tillägnad Juha Janhunen på hans sextioårsdag den 12 februari 2012. Mémoires de la Société Finno-Ougrienne 264: 227–250. Aikio 2014 = Luobbal Sámmol Sámmol Ánte (Ante Aikio) 2014. The Uralic- Yukaghir lexical correspondences: genetic inheritance, language contact or chance resemblance? — Finnisch-Ugrische Forschungen 62: 7–76. Aikio, Ante 2015. The Finnic ‘secondary e-stems’ and Proto-Uralic vocalism. — Journal de la Société Finno-Ougrienne 95: 25–66. Aikio, Ante (forthcoming). New and old Samoyed etymologies (part 3). Aikio, Ante & Ylikoski, Jussi 2016. The origin of the Finnic l-cases. — Fenno- Ugrica Suecana Nova Series 15: 59–158. <http://www.su.se/polopoly_fs/1.356087.1510165998!/menu/standard/file/FUS2 016-15_06c-JY_AA-Final.pdf>, accessed 12/15/2017. Bartens, Raija 2000. Permiläisten kielten rakenne ja kehitys. Mémoires de la Société Finno-Ougrienne 238. Bergsland, Knut 1959. The Eskimo-Uralic hypothesis. — Journal de la Société Finno-Ougrienne 61: 1–29. Campbell, Lyle 1998. Nostratic: a personal assessment. — Joseph D. Salmons & Brian D. Joseph (eds.), Nostratic: sifting the evidence. Current Issues in Linguistic Theory 142. Amsterdam & Philadelphia: John Benjamins. Pp. 107– 152. Campbell, Lyle 1998. Historical linguistics: an introduction. Cambridge, Massachusetts: The MIT Press. 55  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Campbell, Lyle & Garrett, Andrew 1993. Review: Jorma Koivulehto, Uralische Evidenz für die Laryngaltheorie (1991). — Language 69: 832–836. Collinder, Björn 1940. Jukagirisch und Uralisch. Uppsala: Almqvist & Wiksell. Csúcs, Sándor 2005. Die Rekonstruktion der permischen Grundsprache. Budapest: Akadémiai kiadó. de Smit, Merlijn 2014. Proto-Uralic ergativity reconsidered. — Finnisch-Ugrische Mitteilungen 38: 1–34. Fortescue, Micahel & Jacobson, Steven & Kaplan, Lawrence 1994. Comparative Eskimo Dictionary with Aleut Cognates. Alaska Native Language Center Research Paper Number 9. Fairbanks: University of Alaska Fairbanks. Genetz, Arvid 1896: Ensi tavuun vokaalit suomen, lapin ja mordvan kaksi- ja useampitavuisissa sanoissa. Helsinki: Suomalaisen kirjallisuuden seura. Häkkinen, Jaakko 2009. Kantauralin ajoitus ja paikannus: perustelut puntarissa. — Journal de la Société Finno-Ougrienne 92: 9–56. Häkkinen, Jaakko 2012a. Early contacts between Uralic and Yukaghir. — Tiina Hyytiäinen, Lotta Jalava, Janne Saarikivi & Erika Sandman (eds.), Per Urales ad Orientem: Iter polyphonicum multilingue. Festskrift tillägnad Juha Janhunen på hans sextioårsdag den 12 februari 2012. Mémoires de la Société Finno- Ougrienne 264: 227–250. Häkkinen, Jaakko 2012b. Uralic-Yukaghir wordlist. Appendix to: Early contacts between Uralic and Yukaghir. <http://www.elisanet.fi/alkupera/UralicYukaghirWordlist.pdf>, accessed 1/17/2018. Häkkinen, Kaisa 1983. Suomen kielen vanhimmasta sanastosta ja sen tutkimisesta. Suomalais-ugrilaisten kielten etymologisen tutkimuksen perusteita ja metodiikkaa. Turun yliopiston suomalaisen ja yleisen kielitieteen laitoksen julkaisuja 17. Havas, Ferenc 2006. Die ergativität und die uralischen Sprachen. — Finnisch- Ugrische Forschungen 59: ... Helimski, Eugene 1995. Proto-Uralic gradation: Continuation and traces. — ... (eds.), Congressus Octavus Internationalis Fenno-Ugristarum I: Orationes plenariae et conspectus quinquennales. Jyväskylä: Moderatores. Pp. 17-51. Honti, László 1982. Geschichte des obugrischen Vokalismus der ersten Silbe. Budapest: Akdemiai kiadó. Honti, László 1993. Die Grundzahlwörter der uralischen Sprachen. Budapest: Akadémiai kiadó. Honti, László 1997. Numerusprobleme (Ein Erkundungszug durch den Dschungel der uralischen Numeri). — Finnisch-Ugrische Forschungen 54: ... Honti, László 1998. ObUgrian. — Daniel Abondolo (ed.), The Uralic languages. London & New York: Routledge. Pp. 327–357. 56  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Itkonen, Erkki 1956. Zur Geschichte des Vokalismus der ersten Silbe im Tscheremissischen und in den permischen Sprachen. — Finnisch-Ugrische Forschungen 31: 145–345. Itkonen, Terho 1992. Ugrilaisten kielten karitiivista. — Pál Deréky, Timothy Riese, Marianne Sz. Bakró-Nagy & Péter Hajdú (eds.), Festschrift für Károly Rédei zum 60. Geburtstag. Studia Uralica 6: 221–237. Janhunen, Juha 1976. Adalékok az északi-szamojéd hangtörténethez: vokalizmus, az első szótagi magánhangzók. — Néprajz és nyelvtudomány XIX–XX: 165–188. Janhunen, Juha 1977. Samojedischer Wortschatz. Gemeinsamojedische Etymologien. Castrenianumin toimitteita 17. Janhunen, Juha 1981. Uralilaisen kantakielen sanastosta. — Journal de la Société Finno-Ougrienne 77: 219–274. Janhunen, Juha 1982. On the Structure of Proto-Uralic. — Finnisch-Ugrische Forschungen : 23–42. Janhunen, Juha 1983. On early Indo-European–Samoyed contacts. — Juha Janhunen, Anneli Peräniitty & Seppo Suhonen (eds.), Symposium saeculare Societatis Fenno-Ugricae. Mémoires de la Société Finno-Ougrienne 185: 115–127. Janhunen, Juha 1989. Samojedin predesinatiivisen deklinaation alkuperästä. — Journal de la Société Finno-Ougrienne 82: 298–301. Janhunen, Juha 1998. Samoyedic. — Daniel Abondolo (ed.), The Uralic languages. London & New York: Routledge. Pp. 457–479. Janhunen, Juha 2000. Reconstructing Pre-Proto-Uralic typology spanning the millennia of linguistic evolution. – Congressus Nonus Internationalis Fenno- Ugristarum I: Orationes plenariae & Orationes publicae. Tartu. Pp. 59–76. Janhunen, Juha 2001a. Indo-Uralic and Ural-Altaic: On the diachronic implications of areal typology. — Christian Carpelan & Asko Parpola & Petteri Koskikallio (eds.): Early Contacts between Uralic and Indo-European: Linguistic and Archaeological Considerations. Mémoires de la Société Finno-Ougrienne 242: 207–220. Janhunen, Juha 2001b. On the paradigms of Uralic comparative studies. — Finnisch- Ugrische Forschungen 56: ... Janhunen, Juha 2002. The Nenets imperative sentence and its background. — Finnisch-Ugrische Mitteilungen 24/25: 71–85. Janhunen, Juha 2007a. The primary laryngeal in Uralic and beyond. — Jussi Ylikoski & Ante Aikio (eds.), Sámit, sánit, sátnehámit: Riepmočála Pekka Sammallahtii miessemánu 21. beaivve 2007. Mémoires de la Société Finno-Ougrienne 253: 203–227. Janhunen, Juha 2007b. Typological expansion in the Ural-Altaic belt. — Incontri Linguistici 30: 71–83. 57  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Janhunen, Juha 2009. Proto-Uralic — what, where, and when? — Jussi Ylikoski (ed.), The Quasquicentennial of the Finno-Ugrian Society. Mémoires de la Société Finno-Ougrienne 258: 57–78. Janhunen, Juha 2014. Ural-Altaic: The Polygenetic Origins of Nominal Morphology in the Transeurasian Zone. — Martine Robbeets & Walter Bisang (eds.), Paradigm change in the Transeurasian languages and beyond. Studies in Language Companion Series 161. Amsterdam & Philadelphia: John Benjamins. Pp. 311–336. Janhunen, Juha (forthcoming). Grammaticalization in Uralic as viewed from a general Eurasian perspective. Kallio, Petri 2004. Tocharian loanwords in Samoyed? — Irma Hyvärinen & Petri Kallio & Jarmo Korhonen (eds.), Etymologie, Entlehnungen und Entwicklungen: Festschrift für Jorma Koivulehto zum 70. Geburtstag,. Mémoires de la Société Néophilologique de Helsinki 63: 129–137. Kallio, Petri 2006. Suomen kantakielten absoluuttista kronologiaa. — Virittäjä 110: 2–25. Keresztes, László 1999. Development of Mordvin definite conjugation. — Mémoires de la Société Finno-Ougrienne 233. Kiefer, Ferenc & Laakso, Johanna 2014. Uralic. — Rochelle Lieber & Pavol Štekauer (eds.), The Oxford Handbook of Derivational Morphology. Koivulehto, Jorma 1991. Uralische Evidenz für die Laryngaltheorie. Wien: Verlag der österreichischen Akademie der Wissenschaften. Koivulehto, Jorma 2001. The Earliest Contacts between Indo-European and Uralic Speakers in the Light of Lexical Loans. — Christian Carpelan & Asko Parpola & Petteri Koskikallio (eds.): Early Contacts between Uralic and Indo-European: Linguistic and Archaeological Considerations. Mémoires de la Société Finno- Ougrienne 242: 235–263. Korhonen, Mikko 1981. Johdatus lapin kielen historiaan. Helsinki: Suomalaisen kirjallisuuden seura. Kulonen, Ulla-Maija 2001. Zum n-Element der zweiten Personen besonders im Obugrischen. — Finnisch-Ugrische Forschungen 56: ... Laakso, Johanna 1997. On verbalizing nouns in Uralic. — Finnisch-Ugrische Forschungen 54: ... Larsen, Anders 2014. “Mearrasámiid birra” ja eará čállosat – “Om sjøsamene” og andre skrifter. Ed. by Ivar Bjørklund & Harald Gaski. Lehtinen, Tapani 2007. Kielen vuosituhannet: suomen kielen kehitys kantauralista varhaissuomeen. Helsinki: Suomalaisen kirjallisuuden seura. Lytkin 1964 = В. И. Лыткин 1964. Исторический вокализм пермских языков. Москва: издательство “Наука”. 58  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Maddieson, Ian 2013a. Vowel Quality Inventories. — Matthew Dryer & Martin Haspelmath (eds.), The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. <http://wals.info/chapter/2>, accessed on 1/17/2018. Maddieson, Ian 2013b. Front Rounded Vowels. — Matthew Dryer & Martin Haspelmath (eds.), The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. <http://wals.info/chapter/11>, accessed on 1/17/2018. Maddieson, Ian 2013c. Consonant Inventories. — Matthew Dryer & Martin Haspelmath (eds.), The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. <http://wals.info/chapter/1>, accessed on 1/17/2018. Michalove, Peter 2002. The classification of the Uralic languages: Lexical evidence from Finno-Ugric. — Finnisch-Ugrische Forschungen 57: 58–67. Napoľskikh, Vladimir 2001. Tocharisch-Uralische Berührungen: Sprache und Archäologie. — Christian Carpelan & Asko Parpola & Petteri Koskikallio (eds.): Early Contacts between Uralic and Indo-European: Linguistic and Archaeological Considerations. Mémoires de la Société Finno-Ougrienne 242: 367–383. Nichols, Johanna 1972. Suffix ordering in Proto-Uralic. — Lingua 32: 227–238. Nikolaeva, Irina 2006. A Historical Dictionary of Yukaghir. Berlin & New York: Mouton de Gruyter. Ramstedt, G. J. 1917. Suomalais-ugrilaisen komparatiivin syntyperä. — Virittäjä 21: 37–39. Rédei, Károly 1988–1991. Uralisches etymologisches Wörterbuch, Band I–III. Unter Mitarbeit von Marianne Bakró-Nagy, Sándor Csúcs, István Erdélyi, László Honti, Éva Korenchy, Éva K. Sal und Edit Vértes. Wiesbaden: Otto Harrassowitz. Rédei, Károly 1999. Zu den uralisch-jukagirischen Sprachkontakten. — Finnisch- Ugrische Forschuingen 55: 1–58. Salminen, Tapani 1996. A commentary address to László Honti’s “Zur Morphotaktik und Morphosyntax der uralischen/finnisch-ugrischen Grundsprache”. — ... (ed.), Congressus Octavus Internationalis Fenno-Ugristarum  VIII, Pars VIII: 25– 27. Online version: <http://www.helsinki.fi/~tasalmin/cifu8.html>, accessed 1/17/2018. Salminen, Tapani 1999. Euroopan kielet muinoin ja nykyisin. — Paul Fogelberg (ed.), Pohjan poluilla. Suomalaisten juuret nykytutkimuksen mukaan. Bidrag till kännedom av Finlands natur och folk 153: 13–26. Salminen, Tapani 2001. The rise of the Finno-Ugric language family. Christian Carpelan & Asko Parpola & Petteri Koskikallio (eds.): Early Contacts between 59  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Uralic and Indo-European: Linguistic and Archaeological Considerations. Mémoires de la Société Finno-Ougrienne 242: 385–396. Online version: <http://www.helsinki.fi/~tasalmin/tvarminne.html>, accessed 1/17/2018. Salminen, Tapani 2002. Problems in the taxonomy of the Uralic languages in the light of modern comparative studies. — А. Е. Кибрик (ed.), Лингвистический беспредел: сборник статей к 70-летию А. И. Кузнецовой. Москва: Издательство Московского университета. Pp. 44–55. Online version: <http://www.helsinki.fi/~tasalmin/kuzn.html>, accessed 1/17/2018. Salminen, Tapani 2014. Suomalais-samojedilaisia muotovertailuja. — Nobufumi Inaba, Jorma Luutonen, Arja Hamari & Elina Ahola (eds.), Juuret marin murteissa, latvus yltää Uraliin. Juhlakirja Sirkka Saarisen 60-vuotispäiväksi 21.12.2014. Mémoires de la Société Finno-Ougrienne 270: 289–300. Sammallahti, Pekka 1988. Historical phonology of the Uralic languages with special reference to Samoyed, Ugric and Permic. — Denis Sinor (ed.), The Uralic languages. Description, history and foreign influences. Leiden & New York & København & Köln: E.J. Brill. Pp. 478–554. Sammallahti, Pekka 1998. The Saami Languages: an Introduction. Kárášjohka: Davvi Girji OS. Seefloth, Uwe 2000. Die Entstehung polypersonaler Paradigmen im Uralo- Sibirischen. — Zentralasiatische Studien 30: 163–191. Siegl, Florian 2013. Materials on Forest Enets, an Indigenous Language of Northern Siberia. Mémoires de la Société Finno-Ougrienne 267. Sinor, Denis (ed.) 1988. The Uralic languages. Description, history and foreign influences. Leiden & New York & København & Köln: E.J. Brill. Syrjänen & al. 2013 = Syrjänen, Kaj & Honkola, Terhi & Korhonen, Kalle & Lehtinen, Jyri & Vesakoski, Outi & Wahlberg, Niklas 2013. Shedding more light on language classification using basic vocabulies and phylogenetic methods. A case study of Uralic. — Diachronica 30:3: 323–352. Tereshchenko 1965 = Терещенко, Н. М. 1965. Ненетско-русский словарь. Москва: издательство “Советская энциклопедия”. Ylikoski, Jussi 2011. A survey of the origins of directional case suffixes in European Uralic. Ylikoski, Jussi 2016. The origins of the western Uralic s-cases revisited: historiographical, functional-typological and Samoyedic perspectives. Ylikoski, Jussi (forthcoming). On the tracks of the Proto-Uralic suffix *-ksi – a new but old perspective on the origin of the Mari lative. Zaicz, Gábor 1998. Mordva. — Daniel Abondolo (ed.), The Uralic languages. London & New York: Routledge. Pp. 184–218. Zhivlov, Mikhail 2014. Studies in Uralic Vocalism III. — Journal of Language Relationship / Вопросы языкового родства 12: 113–148. 60  Luobbal Sámmol Sámmol Ánte (Ante Aikio): Proto-Uralic. — To appear in: Marianne Bakró-Nagy, Johanna Laakso & Elena Skribnik (eds.), The Oxford Guide to the Uralic Languages. Oxford University Press.   Zhivlov 2018 = Живлов, М. А. Историческая фонетика и внутренняя классификация уральских языков. XIII традиционные чтения памяти С. А. Старостина, Институт восточных культур и античности РГГУ. Mar 22–23, 2018. <http://www.academia.edu/38048089/Историческая_фонетика_и_внутренняя _классификация_уральских_языков>, accessed 1/21/2019. 61