Printed version: Tapani Salminen: A reappraisal of M. Alexander Castrn’s Forest Nenets records. — Remota relata: essays on the history of Oriental studies in honour of Harry Haln = сборник научных статей по истории востоковедения к шестидесятилетию Г. Т. Галена. Edited by Juha Janhunen and Asko Parpola. Studia Orientalia 97; Helsinki: Finnish Oriental Society, 2003. 263–277.

[Note on transcription: the short vowel in Tundra Nenets ( in the original) has been given another symbol (ə); and the digraphs for the fricolateral and the velar stop (lh and ng in the original) have been replaced with monographs (ł and ŋ respectively); the short vowels in Forest Nenets ( ) remain to be denoted in this manner rather than by a breve over the corresponding characters of the long vowels (a i u).]

[Corrected misprints: Jwsyi Jwqsyi; xəq- xq-; denoted denotes; jaŋkł jŋkł; nədm nŋkudm; a common Nenets words a common Nenets word; ŋəmtiq ŋəmteq; Tundra Nenets meaning Tundra Nenets word meaning; in stead of instead of; ksye ksyeq; inadvertedly inadvertently.]

A reappraisal of M. Alexander Castrn’s Forest Nenets records

by Tapani Salminen

Among the first of the many highlights of M. Alexander Castrn’s great Siberian expedition (1845–1849) was his scientific discovery of the Forest Nenets people and language in the summer of 1845. It turned out to have significant consequences for Castrn’s thinking of comparative linguistics and Siberian ethnohistory, and his records of the language continue to be of utmost importance to Samoyedological studies. More exactly, Castrn’s material shows the existence of two distinct varieties of Forest Nenets, and the bulk of this article is devoted to determining the nature of their differences and their background.

The progress of Castrn’s expedition is accounted in detail in Castrn (1856), and his encounters with the Forest Nenets are described in the section titled “Reise von Samarowa nach Surgut” which includes a travel account (62–88) as well as two letters, one to Castrn’s supervisor, Councillor of State A. J. Sjgren (88–90), and the other to his friend, Assessor F. J. Rabbe (90–92). In the summer of 1845, Castrn was mainly engaged in Khanty studies, but he was even more intrigued about the possibility of contacting Forest Nenets people, known at the time as Kondinsk or Kazym Samoyeds, who he had heard about on his first expedition to Siberia. He finally succeeded in meeting and, albeit briefly, working with a few Forest Nenets after he had left Samarova at the beginning of July and, for a month, travelled along the waterways of the Ob to various directions between Samarova and Silyarskoy.

Silyarskoy was a small village on the upper Ob, and when leaving Samarova, Castrn’s idea was to make a trip to the local annual market, reportedly frequented by Forest Nenets. He decided, however, to give up this plan because the turnout was going to be very low due to famine caused by exceptional flooding in that summer. Instead, Castrn made by river the 70 verst journey to the village of Toropkova, also called Skripunova, which is situated slightly off the road between Samarova and Silyarskoy. In Toropkova he met six Forest Nenets belonging to the Jwqsyi (“Jewschi”) family, and he wrote to Sjgren soon after his arrival on the 4th (16th) of July in 1845 as follows: “the matter is namely that I have here, at the mouth of the upper Ob, made an unexpected discovery of a peculiar small Samoyed tribe, which has a dialect that differs greatly from those of other Samoyeds” (Castrn 1856: 89). From the data available to him, Castrn inferred, correctly as we can now tell, that the Samoyeds of Kazym and Agan river basins belong to the same people, and succeeded thus in defining the north-western and south-eastern boundaries of the Forest Nenets language area.

Castrn did not stay in Toropkova for long but made an excursion to the Khanty village of Chebakova on the upper Ob where he stayed for a couple of weeks. From there he returned to Toropkova to focus on Forest Nenets, but the actual time he could devote to intensive fieldwork was quite short, perhaps less than a week, as he also made a trip to Silyarskoy, and from there a 10 verst journey to the nearby Baly yurts, already located in Surgut division of Tobolsk gouvernement. In Baly, he met two Forest Nenets of the Nyetyu (“Nitschu”) family from Lyamin Sor region. There again, the time he had for work with informants was shorter than he would have desired. In his letter to Sjgren, Castrn expressly stated that Forest Nenets required an in-depth study, but because of his commission to Khanty he had much less time than was necessary for that purpose.

In his letter to Rabbe, dated in Chebakova on the 25th of July (6th of August), Castrn focuses on the significance of Forest Nenets to comparative linguistics, saying that “I am, however, now happy and satisfied, when I see various theories of mine being confirmed”, namely because “through a small previously unknown Samoyed tribe”, it is now possible to establish “a practically unbroken chain of Samoyed peoples from Archangel and Mezen to the Baykal region” (Castrn 1856: 91). Furthermore, Castrn (1856: 67) points out that beyond the Samoyed context, Forest Nenets “serves the purpose of highlighting the affinity between the Samoyed and Finnish languages to the greatest extent”, and his article on the topic published next year (Castrn 1846; in German: Castrn 1856: 68–77) was only made possible through the impetus provided by his experience with Forest Nenets. This article, while concise and restricted to lexical material, is the first treatment demonstrating the close connection between Samoyed, Finnish and other languages already known to be related to Finnish, and hence the existence of the Uralic language family, employing a methodology that is still valid today, and presenting many ingenious etymologies that constitute the cornerstone of comparative Uralic studies. Castrn also recognized that “the language of the recently discovered Samoyeds conveys, through a number of letter [= sound] changes and other features found in it, such a close affinity between Finnish and Samoyed, that the latter language, even if it cannot be regarded as a member of the Finnic language family, it will pass at least for the next closely related tribe” (Castrn 1856: 91) and concludes that a common linguistic origin is the only plausible explanation.

Alongside his comparisons within Uralic, Castrn noted various similarities to Turkic, Mongolic and Tungusic, and concluded that the original home of all these language families was in the Altay region. It must be kept in mind, though, that Castrn never claimed to have shown the relatedness of Uralic and the so-called Altaic languages in the same manner as he demonstrated the membership of the Samoyed branch in Uralic.

As Castrn’s more general achievements will be duly discussed in a forthcoming scholarly biography, the rest of this article deals with the factual Forest Nenets material collected by him in the course of his fieldwork at the two localities mentioned above, Toropkova and Baly. Castrn’s Forest Nenets records comprise 700 basic words, of which only 50 or so were included in the Samoyed vocabulary edited by Schiefner (Castrn 1855). Besides vocabulary, Castrn only wrote down a small number of paradigms, and it is therefore understandable why so few grammatical features typical of Forest Nenets are attested in the Samoyed grammar (Castrn 1854).

As Lehtisalo points out in his introduction to the publication of Castrn’s Forest Nenets records (Castrn 1960), the Baly material is practically identical with his and other fieldworkers’ later records of Forest Nenets. By contrast, there are striking discrepancies between the Toropkova material and what is otherwise known of Forest Nenets, and Lehtisalo consequently assumed that the western dialect of Forest Nenets spoken in the Toropkova region had changed substantially in the course of the 69 years that had elapsed between Castrn’s visit to the area in 1845 and his own, much more extensive fieldwork in 1914 (Castrn 1960: 263).

Notably, all of the peculiar features of the variety of Forest Nenets that Castrn first encountered in Toropkova are such that they make it more similar to the closely related, yet clearly distinct Tundra Nenets language. In theory, there are three possible scenarios about the origin of such a variety. Firstly, we might be dealing with a phenomenon with ancient roots, more exactly, the people in Toropkova would have spoken a remnant idiom that had preserved features common with Tundra Nenets that were replaced by innovations in (other) varieties of Forest Nenets. Secondly, the Toropkova dialect could represent a genuine variety of Forest Nenets that had, however, undergone significant Tundra Nenets influence, in other words, it would have emerged through later areal connections. Thirdly, given Castrn’s fluency in Tundra Nenets deriving from his first great expedition, it would have been possible that the Forest Nenets speakers he met in Toropkova tried to accommodate their speech accordingly, creating on the spot a variety that differed from their native dialect exactly in those features that now appear so perplexing.

The first scenario of an original intermediate variety can be rejected easily, as the peculiarities of the Toropkova material reflect reasonably late Tundra Nenets innovations that were, consequently, absent in common Nenets. The Toropkova material also contains examples of all distinctively Forest Nenets developments in its phonology and morphology. The second scenario of a contact-induced dialect is much more plausible, and we may think that it is the one implicitly supported by Lehtisalo. There is, however, little historical foundation for assuming Tundra Nenets influence in the Toropkova region, and, more to the point, the factual material exhibits an untypical pattern of variation concerning exactly those word-forms that display Tundra Nenets characteristics, including anomalous or hypercorrect formations. It appears therefore that the third scenario involving essentially a kind of foreigners’ talk is the best explanation to the emergence of the Toropkova material. In what follows, an attempt is made to explain the specific features recorded by Castrn in Toropkova as a restricted set of phonological and morphological substitutions carried out by speakers who in all other contexts would have used a Forest Nenets dialect basically lacking such features.

At first sight, the number of ‘Tundraisms’ in Toropkova material is impressive, as seen in the following table contrasting selected word-forms from modern Tundra Nenets (in phonological transcription), Castrn’s data (marked by underlining) from Toropkova and Baly, and the currently spoken Pur dialect of Forest Nenets (again in phonological transcription; represent short vowels contrasting with long a i u, and j rather than y denotes the palatal glide; otherwise the system is similar to the Tundra Nenets one):

gloss Tundra Nenets Toropkova Baly Pur
‘bridge’ pul pul pur
‘five’ səmplyaŋk sambljang hamprjang xmpłyaŋk
‘giant’ syudbya šudobe *šutpe sytpya
‘I will help him’ nyadaŋkuw njadanguu *njadanangam nyatanŋam
‘I tore it’ ŋədqŋaw ˜adangau ˜atngam ŋtqŋam

The word-forms marked by asterisk in the Baly column are unattested but reconstructed on the basis of comparable forms to facilitate comparison. A recent sound change ł < r has occurred in the Pur dialect, which means that, disregarding the trivial differences in transcription, the Baly forms contrast with Pur forms only in the presence of a voiced consonant in the stem for ‘help’, which, as explained below, can be an allophonic feature not warranting a connection with Tundra Nenets. Consequently, the Baly material is almost entirely consistent with what has later been recorded from Forest Nenets dialects, and it will be touched only marginally below. The following material marked by underlining belongs therefore to the Toropkova corpus, unless otherwise noted. Besides the eastern Pur dialect, a slightly archaic variety of the western Lyamin dialect illustrates the Forest Nenets language as recorded after Castrn.

The weakening, involving voicing, of intervocalic obstruents is a regular feature of Tundra Nenets (TN), and a similar phenomenon is widely attested in the Toropkova material but is generally absent in modern Forest Nenets, e.g. ˜ooba ~ Pur ŋopa ~ TN ŋoba ‘mitten’, wad ~ Pur wata ~ TN wada ‘word’, kooba ~ Pur kopa ~ TN xoba ‘fur’. Unlike in Tundra Nenets, however, the Toropkova material shows no signs of phonological contrast between strong and weak obstruents, which makes the voicing a purely phonetic phenomenon. Similar examples can be found sporadically in the Baly material as well, mainly when the consonant follows a stressed long vowel, e.g. koodorŋat ~ Lyamin koturŋat ~ TN xodorŋadm ‘I am coughing’, nad ~ Pur natu ~ TN nado ‘younger brother of the spouse’, but more often Castrn transcribes a voiceless obstruent from Baly as opposed to a voiced one from Toropkova, e.g. Toropkova naadam ~ Baly naatam ~ Pur natm ~ TN nadm ‘mucus of the nose, snot’. Since voicing is marginally possible in the above mentioned context in modern Forest Nenets, it is only the extent of this allophonic rule as reflected by the Toropkova material that is genuinely reminiscent of Tundra Nenets. In the case of obstruents following a nasal consonant, a similar obstruent weakening occurs, but it is allophonic both in the Toropkova material and in Tundra Nenets, e.g. ˜amz ~ Pur ŋmsa ~ TN ŋəmca (where the consonant cluster is often phonetically [mz]) ‘meat’. This phenomenon is also found in words lacking an exact cognate in Tundra Nenets, e.g. koondeu ~ Lyamin kontw ‘ptarmigan’. Crucially, though, there are counterexamples to, that is, variation in all types of obstruent weakening, for instance, transcriptions such as ˜ams ‘meat’ and ˜oopa ‘mitten’ are present in the Toropkova material alongside those quoted above. This variation is entirely random, and it is therefore a pure coincidence that the nominative singular of ‘hand’ is transcribed as ute while the genitive singular is given as ude, but such pairs may have triggered rumours about the existense of consonant gradation in Forest Nenets.

Another inconsistency in Castrn’s notation, although a highly trivial one, is that he occasionally and inadvertently leaves out the symbol for the initial velar nasal, yielding variation like ute ~ ˜ute = ŋta ‘hand’.

By contrast, it is tempting to regard the rare instances of r instead of expected t (or d) as hypercorrect formations that have emerged when the Forest Nenets have tried to approach the native Tundra Nenets pronunciation of d as a fricative. Two word families exemplify this possible development, namely jarangau ~ Baly jatangam ~ Pur jtaŋam ‘I shot it’ : durative jarambiu ~ Pur jtapyim ‘I am shooting it’ (but future jadanguu ‘I will shoot it’) ~ TN yəda- ‘to shoot’, and taaribe ~ taasibe [= corrected to] taadibe ~ Baly tatsbe ~ Pur tatyipya ~ TN tadyebya ‘shaman’. A possible third instance is found in the 3rd person plural personal suffix in the objective conjugation attached to the verb tana- ‘to drive (animals)’ transcribed as r while elsewhere the suffix is given as du (~ Lyamin tu ~ Pur tuŋ ~ TN doh). Whatever the background of this notation is, at least we can be sure that there is no connection to any ancient Samoyed d ~ r variation, as suggested by Lehtisalo in Castrn (1960: 264) in the context of the word for ‘shaman’.

Another apparent Tundra Nenets feature in the Toropkova material is the replacement of initial *k, otherwise fully preserved in Forest Nenets, with h. This phenomenon is, however, much more sparsely attested than the weakening of obstruents, and in practically all cases, Castrn has added a k on top of the h that he had originally transcribed, e.g. haeu kaeu ~ Pur kw ‘side’. It may be concluded that no regular sound change like the one known from the history of Tundra Nenets has occurred in Toropkova. This is corroborated by a comparison with a genuine sound change in Forest Nenets dialects, namely the change of initial *s to x, which was still in progress both in Toropkova and Baly at the time of Castrn’s visit, as evident from Castrn (1960: 289–291) showing initial s and h in free variation, but without any confusion of this h with the one appearing instead of *k. Notably, *s is reflected as s more often in Toropkova, while the Baly data mostly contains h, which may reflect the fact that in Tundra Nenets initial *s is preserved. Variation is particularly common though, e.g. sangat ~ hangat ~ Pur xqŋat ‘I want’, and many common words appear only with h, e.g. haem ~ Pur xm ‘eye’.

As noted below, the most conspicuous ‘Tundraisms’ appear in word-forms where an expected *m is replaced with u (= w) in verbal personal and nominal possessive suffixes, in which w also occurs in Tundra Nenets as the regular outcome of intervocalic *m. There are numerous examples showing that the preservation of intervocalic *m was the rule in Toropkova in the same way as in all Forest Nenets dialects, for instance, in the forms corresponding to Pur ŋmoł- ‘eat’, ŋmł ‘food’, ŋamat ‘waist’, pyemaq ‘boots’, xma ‘hat’, the Tundra Nenets cognates of which show an equally regular w. Besides the inflectional forms treated separately below, an unexpected u is only found in a couple of prosecutive case forms, e.g. jideuana ~ Pur jetimna ~ TN yedeywəna ‘new’ (but japtamana ~ Pur japtamna ~ TN yabtawna ‘thin’), and in two isolated nouns, kaasauwa ~ Pur kasama ~ TN xasawa ‘man’, and kaewa ~ Pur kma ~ TN xwa ‘bone marrow’. Surprisingly, on page 287 of Castrn (mscr.) there are both kaeuwa and kaema apparently recorded from Baly, but more probably the one with w belongs to Castrn’s secondary remarks referring to Tundra Nenets, as on page 287 of the manuscript there is only Baly kaema.

The distribution of liquid consonants is strikingly different in Forest Nenets in comparison with Tundra Nenets, which in this respect is archaic. Essentially, *l and *r merged in early Forest Nenets yielding r, which then changed back to l in a small number of environments, which have not been worked out exactly as yet. As a result, all possible correspondences between Tundra Nenets and Forest Nenets liquids occur. A subsequent change that has by now reached all Forest Nenets dialects is ł (fricolateral) < r (vibrant; still recorded in older material from Lyamin), reflected in the following Pur material. In the Toropkova material, as seen in Castrn (1960: 277–278), initial l ~ r occur in free variation, and the diacritic marking employed by Castrn in a few instances (but ignored here) seems to indicate that the Toropkova l is often, if not generally, to be read as ł; this is also the case of the word pul ‘bridge’ quoted in the table above. In other positions, there is also variation in the marking of r (> ł), but, crucially, the reflects are consistent with the distribution of liquids in Forest Nenets, e.g. (a) FN ł ~ TN l: dative ыr ~ Pur ŋł ~ TN ŋilh : locative ыrna ~ Pur ŋłna ~ TN ŋilna ‘under’; (b) FN ł ~ TN r: juol ~ Pur jołya ~ TN yorya ‘deep’, salj ~ Pur xałyu ~ TN saryo ‘rain’, penšel ~ Pur pyensył ~ TN pyencyr ‘drum’; (c) FN l ~ TN l: juolš (also in Baly) ~ Pur jolsy ~ TN yolcyh ‘measure, time’; (d) FN l ~ TN r: salmik ~ Baly helmik ~ Pur xlmyiqk ‘sable’ ~ TN (dial.) sərmyik ‘animal’.

Another sound change creating complex correspondence patterns between the Nenets languages is the development of intervocalic *ny (palatal nasal) to j (palatal glide) in Forest Nenets, Castrn’s records showing, again, exclusively Forest Nenets reflexes, e.g. ˜ai ~ Pur ŋaj ~ TN ŋanyh ‘but’, tjj ~ Pur tyaju ~ TN tyanyo ‘little’, wijku ~ Pur wyjaqku ~ TN yinyako ‘strap’. The same goes with the other sound changes that characterize Forest Nenets, notably (*lt >) *ll > nr > , e.g. pilnju ~ Baly pirnju (the apparent metathesis in these records seems a notational technicality of Castrn’s) ~ Pur pynłyu ~ TN pyilyo ‘gadfly’, and *y > ye, e.g. wik ~ wik ~ Pur wyeqk ~ TN yk ‘neck’, nien ~ Pur nyen ~ TN nyn ‘my friends’, nieš ~ nieše ~ Pur nyesya ~ TN nysya ‘father’, še’u ~ Pur syeqw ~ TN syqw ‘seven’. Forest Nenets retentions such as the preservation of final *ŋ, e.g. wing ~ Lyamin wiŋ ~ TN wh ‘tundra’, tang ~ Pur ~ TN təh ‘summer’, or the preservation of initial *wy, e.g. weab ~ weap ~ Pur wyap ~ TN yab ‘luck’ are invariably reflected in Castrn’s material.

While Tundra Nenets has largely preserved Proto-Nenets vowels, the Forest Nenets vowel system has gone through a major restructuring, after which there are now six long (full, tense) and four short (reduced, lax) vowels in stressed syllables. In monosyllabic word-forms the contrast in vowel quantity is neutralized in favour of the shorter vowels, and this is also shown by Castrn’s records, which differ from the Tundra Nenets forms that would also be identical with the respective Proto-Nenets ones: ˜u ~ Pur ŋ ~ TN ŋo ‘paint, dye’, ˜u ~ Pur ŋ ~ TN ŋo ‘island’ (cf. ˜u ~ Pur ŋ ~ TN ŋu ‘pole’), mu ~ Pur m ~ TN mo ‘twig’, pu ~ Pur p ~ TN po ‘year’. Also unstressed vowels got restructured in Forest Nenets, and this is quite consistently reflected in the Toropkova material, but the relevant examples are not diagnostic because of similar phonetic tendencies in eastern Tundra Nenets dialects.

In addition to the replacement of the vibrant r with the fricolateral ł, and the change of initial s to x, discussed above, every known phonological feature that has resulted from a recent sound change or analogy in western Forest Nenets, represented by the Lyamin dialect, can be seen in the Toropkova material. They include, firstly, the loss of syllable-final nasals in non-initial syllables, e.g. arkkaboi ~ Lyamin ŋarkapoj ~ Pur ŋałkampoj ~ TN ŋarkampoy ‘rather big’, niejak ~ Lyamin nyejak ~ Pur nyejaŋk ~ TN nyenyaŋk ‘mosquito’, ude ~ Lyamin ŋta ~ Pur ŋtaŋ ~ TN ŋudah ‘hand’ (genitive). Secondly, the conjugation of the so-called alteration stem verbs may follow the pattern of vowel stem verbs, e.g. xeamčungat (x is an odd symbol for ŋ) ~ Lyamin ŋamtyuŋat ~ Pur ŋimtit ~ TN ŋamtidm ‘I sit’, maijibiungat ~ Lyamin mjpyoŋat ~ Pur mjpyit ~ TN məympyidm ‘I rejoice’. The third dialectal phenomenon is the generalization of secondary vowel stems in the nominative singulars of consonant stem nouns, e.g. koot ~ Lyamin kot ~ Pur kq ~ TN xoq ‘cough’, njem ~ Lyamin nym ~ Pur nym ~ TN nyim ‘name’, ˜ut ~ Lyamin ŋt ~ Pur ŋq ~ TN ŋuq ‘trace’, tjamdt ~ Lyamin tyamtt ~ Pur tyamtiq ~ TN tyamteq ‘frog’, wiktat ~ Lyamin wyeqktt ~ Pur wyeqktq ~ TN ykədq ‘collar’, (reflexive marker:) tjiijet ~ Lyamin tyijt ~ Pur tyijq ~ TN tyyq ‘it flew away’.

There are further phonetic phenomena characteristic of Forest Nenets dialects, and they are well represented in Castrn’s records, namely the lowering of short high vowels before a syllable with a schwa, e.g. njem = nym ‘name’, ˜n = ŋn ‘bow’, tjem = tym ‘tooth’, tjen = tyn ‘granary’, and the affricatization of ty, e.g. piče ~ pitje ~ pitje = pytya ‘nest’.

There is in fact only one major sound change that has generally taken place in Forest Nenets but seems to be less advanced or partially absent in the Toropkova as well as Baly material, namely metaphony. This term refers to a regular change of stressed non-high vowels to high vowels when followed by a syllable with a high vowel in it, for instance, Forest Nenets nyiŋu ‘chin’ derives from a form that is still preserved as its Tundra Nenets cognate nyaŋu id. Castrn’s records of this particular word are niengu ~ njengu, and they indeed seem to indicate at least a shift from the original low vowel, but many other examples show the low vowel remaining intact, among them amdit ~ Pur ŋimtit ‘I sit’, kapui ~ Pur kpuj ‘lungs’, kasj ~ Pur kisuj ~ TN xasuy ‘dry’, sab ~ Pur xpu ~ TN səbu ‘bladder’. But there are also many words exhibiting variation, e.g. ˜ači ~ ˜etsi ~ Pur ŋtyi ~ TN ŋədyi ‘it is visible’, xa’nji ~ xenjui ~ sanjuj ~ Pur xqnyuj ~ TN səqnyuy ‘wet’, hartji ~ kerše ~ Baly karši ~ Pur krsyi ~ TN xərcyi ‘buttocks’, the metaphonic vowel being marked by a symbol for a mid vowel, which is also the case for several words without variation, including ‘chin’ quoted above as well as helk ~ Pur xlyiqk ~ TN səlyik ‘elbow’, and tti ~ Pur tiqti ~ TN tati ‘younger wife’. To complete the confusing picture, there are word-forms with a high vowel recorded by Castrn, e.g. pыritje ~ pыlitje ~ Pur płyitye- ~ TN pəryidye- ‘[to be] black’, piliku ~ Pur płyiqku ~ TN pəryiko ‘black’. Lehtisalo’s comment on the status of metaphony in Castrn (1960: 265) is simply that metaphonic forms are not as common as nowadays. A slightly more analytic standpoint may be reached when certain other records are taken into account: there happen to be a large number of forms with an original high vowel in the first syllable, retained by all varieties of both Nenets languages, nevertheless transcribed as low by Castrn, e.g. kaetje ~ Baly kače ~ Pur ktya ~ TN xidya ‘cup’, witngar ~ Pur wyt ŋłi ‘beaver’ (cf. TN yidh ŋili ‘underwater’), parietsj xaruat ~ Pur Płyiqtyiq kałwt ~ TN Piryityiq xarəd ‘Surgut’, the first word being derived from pыrj ~ Pur płya ~ TN pirya ‘pike’. These examples show that Castrn’s notation of certain vowels varies in a way that makes the correct interpretation of metaphony more difficult. Tentatively it may be assumed that the dialects recorded by Castrn did not differ markedly from other attested varieties of Forest Nenets, in which metaphony is a regular feature.

Turning to morphology, it must be noted that Castrn recorded verbs mainly in their 1st person singular forms, and it is in the respective personal suffixes that the most striking parallelisms of the Toropkova material with Tundra Nenets can be found. Starting with the objective conjugation, there are examples such as ˜amau ~ Pur ŋmam ~ TN ŋəmaw ‘I ate it’, waaptangau ~ Pur waptaŋam ~ TN wabtaəw ‘I turned it over’ (the lack of ŋ in the Tundra Nenets word-form is due to a relatively recent sound change, and in Tundra Nenets folklore archaic forms such wabtaŋaw are marginally possible), laadangau ~ Baly raatngam ~ Pur łatŋam ~ TN ladəw ‘I hit it’. Common are also records with apparent variation in the suffix, e.g. njadangau -m [= m written on top of u] ~ Pur nyataŋam ~ TN nyadaəw ‘I helped him, I added it’, tangau ~ tangam ~ Pur taŋam ~ TN taəw ‘I brought it, I gave it’, teamdangau -m ~ Pur temtaŋam ~ TN temtaəw ‘I bought it’, haptangau ~ kaptangam ~ Pur kptaŋam ~ TN xəbtaəw ‘I extinguished it (the fire)’, while forms with only m are less common, e.g. jihelangam ~ Pur jexałaŋam ~ TN yexaraəw ‘I do not know it’, pilingam ~ Pur pyłyiŋam ~ TN pyiryeəw ‘I cooked it up’. Occasionally, the suffix variant u is stroked through and m written in its place by Castrn, e.g. haamlangau ~ kaamlangam ~ Pur kaqmłaŋam ~ TN xaqwraəw ‘I fell it’. As can be seen from the relevant examples, the placement of m besides or instead of u may occur concomitantly with the addition of initial k instead of h. Both of these corrections seem make the material more like the actual dialect spoken in Toropkova, in other words, those variants of 1st person singular forms in the objective conjugation with m can be regarded as representative records of the Toropkova dialect, which then accords with the other Forest Nenets dialects.

The 1st person singular possessive suffix of nouns is formally and historically identical with the 1st person singular personal suffix in the objective conjugation of verbs, but in modern Forest Nenets, the old suffix m has been largely replaced by an innovative suffix j. In Castrn’s material, which includes a list of 1st person singular possessive forms, perhaps collected rather mechanically, the predominant suffix variant is notably u, reminiscent of Tundra Nenets w attached to vowel stem nouns. Were it not for one instance of the current Forest Nenets suffix, namely lambu ~ lambai ~ Pur łmpaj ~ TN ləmpaw ‘my ski’, as well as one anomalous record, jilepšem -u ~ Pur jłyipsyj ~ TN yilyebcyəm ‘my wild reindeer’, with the variant u attached to a consonant stem noun, the records would indeed give an impression that the possessive forms derived basically from Tundra Nenets. Given the presence of the normal Forest Nenets form and the error in the application of the Tundra Nenets suffix, it is yet plausible to assume that the actual Toropkova dialect employed a system of possessive declension which differed little from mainstream Forest Nenets.

In the subjective conjugation of the verbal inflection, the modern suffix of the 1st person singular in Forest Nenets is t while in Tundra Nenets it is dm. In Forest Nenets, however, the suffix t derives through a recent sound change from tm, and judging from Castrn’s records, this archaic variant still existed both in Toropkova and Baly at the time. When we see examples such as aarmadm ~ Pur ŋałmat ~ TN ŋarmadm ‘I grew’, the similarity of dm to the Tundra Nenets suffix is basically coincidental, involving the voicing tendency discussed above and an older rather than a Tundra Nenets influenced variant of the suffix. There is a lot of variation in Toropkova, e.g. aarmadm ~ aarmat ‘I grew’, jungat (t stroked through) -dm ~ Pur jŋat ~ TN yuədm ‘I warmed up’, pišengat ~ Pur pysyŋat ~ TN pyisyŋadm ‘I am laughing’, and in Baly the archaic variant seems to predominate, e.g. jiringatm ~ Pur jłyiŋat ~ TN yilyeədm ‘I live’, or witjengat -tm ~ Pur wytyiŋat ‘I drink water’, although the number of examples is not large.

There are other examples of the retention of a final nasal after the schwa, e.g. laemang = Lmŋ ~ Pur Lm ‘Lyamin’, confirming that the loss of the consonant in this environment was an on-going process at Castrn’s time, as it still was in the early 20th century. In the verbal suffix, the final m was for a while supported by the respective preterite forms from which it has later disappeared through analogy, e.g. Baly maatm = matm ~ Pur mat ‘I said’ : preterite maatamš = matmsy ~ Pur matsy ‘I said (earlier)’.

There is an alternative way of forming the 1st person singular of the subjective conjugation that is only found in Forest Nenets. Uncharacteristically of the conjugation system, the personal suffix mq is attached directly to the verb stem, e.g. pyinmq ‘I am afraid’, cf. pyinŋat id. Exactly this form is also recorded by Castrn as piinam, but there are also instances of a suffix variant u, e.g. mueu ~ Pur memq ‘I am’, as well as variation, e.g. jakham -u ~ Pur jaqkmq ‘I itch’. Since there are no Tundra Nenets forms of the type *pynəwq, the only way of explaining the records with u is that they represent a simple, albeit inconsistently applied substitution process.

Among the subjective conjugation 1st person singular forms there are also genuinely anomalous formations. For instance, there are four records meaning ‘I am in a hurry’, namely teatam ~ tetau ~ tetangam -dm, of which the two first represent the shorter variant described in the preceding paragraphy (~ Pur teqtmq) and the added suffix variant dm creates a word-form that corresponds regularly to Pur teqtŋat, but the more original record with a suffix variant m is both unattested in later material and fully unexpected as well as rare, hence anomalous, even within Castrn’s corpus. Other examples would be wadjungam -dm ~ Lyamin watyuŋat ‘I grew up’, corrected in the same manner, jilingat -u -m ~ Pur yłyiŋat ‘I live’, with two mislead corrections from the original correct form as it happens, jušiliengam ~ jušiljengat -m ~ Pur jnsyiłyeŋat ‘I listen’, with one uncorrected and one misconstructed form, koonjungam ~ ? koonjungadam ~ Lyamin konyuŋat ‘I sleep’, exhibiting simple variation with the correct form preceded by a question mark, and jaadelŋau ~ Lyamin jatirŋat ‘I walk’, with nothing but the anomalous form. There are a few such cases in the Baly material as well. Their occurrence seems to signal that certain confusion existed between Castrn and his informants when 1st person forms were collected.

In the reflexive conjugation, there are many examples of a suffix variant u in the 1st person singular forms where only m is known from Forest Nenets otherwise, e.g. haamjeu ~ kaamjeu ~ Pur kaqmjm ~ TN xaqmiəwq ‘I fell’, but again, variation exists, e.g. ˜amdjeu ~ xmdjom (where is stroked through and e and ea are written in its place; x is an odd symbol for ŋ) ~ Pur ŋamtjm ~ TN ŋamtiəwq ‘I sat down’, and there are instances of only m, e.g. tjiijem ~ Pur tyijm ~ TN tyyəwq ‘I flew away’. It is not difficult to conclude that the attestations with m reflect the every-day language spoken in Toropkova, while those with u derive from a substitution of the native suffix with the one known from Tundra Nenets.

Besides the 1st person singular, there is another morphological feature that is recorded in large numbers in Castrn’s material, namely the future tense, the formation of which is partly different in the Nenets languages. In Tundra Nenets, there are two independent suffixes, attached to consonant stems, and ŋko attached to vowel stems, while in Forest Nenets, the suffix variant t parallels both formally and functionally with TN , but for vowel stems, a distinct variant n is used. In the Toropkova material, we notice that both ngu and na are attested as future markers in vowel stem verbs, but the Tundra Nenets type, as examplified in the form for ‘I will help him’ in the above table, is admittedly more common. Crucially, there are also plenty of examples of the normal Forest Nenets type, e.g. jillanangau ~ Pur jłnŋam ~ TN yilŋkuw ‘I will lift it’, jurnangam ~ jurnangau ~ Pur jłnŋam ~ TN yurŋkuw ‘I will forget it’, jurknajeu ~ Pur jłknjm ~ TN yurkŋkuwq ‘I will get up’. In the case of intransitive verbs, a surprising but fitting state of affairs is revealed: all of the recorded forms are anomalous in the above defined sense as they employ a suffix variant u (= *w) in the subjective conjugation, e.g. nunguu ~ Pur nunŋat ~ TN nŋkudm ‘I will stand’ (while nungadm -t ~ Pur nuŋat ‘I am standing’ shows the expected non-future form), or njohanangam ~ Pur nyoxanŋat ~ TN nyoxaŋkudm ‘I will sweat’, for which the correct form was noted at Baly by adding t on top of the m of the Toropkova record. As an example of another kind of anomaly in the future formation in the Toropkova material, the variant ngu may be attached to consonant stems, at least when the verb appears to end in a vowel in Castrn’s records because of the absence of a symbol for the glottal stop: ˜adanguu ~ Pur ŋtqtŋam ~ TN ŋədtəəw ‘I will tear it’. The only possible conclusion seems to be that Castrn recorded future forms too mechanically, without paying respect to the factual forms used by the informants, and that the native future formation in Toropkova followed the regular Forest Nenets pattern.

Finally, it is the lexicon found in the Toropkova material that provides the final evidence that it can only derive from a genuine Forest Nenets dialect with few actual influences from Tundra Nenets. The number of lexical differences between the Nenets languages is large, and at least the following exclusively Forest Nenets words are present in the Toropkova corpus: jajqk ‘hazel grouse’, jamp pi ‘snake’, jetuqku ‘dace (a species of fish)’, kpiqsya ‘spoon’ (where Castrn has a misprint with initial l), kpta- ‘far’ (cf. TN ŋa- id.), mani ‘sack’, nipta- ‘rest’ (cf. TN nila- id.), nyalu ‘flame’ (cf. TN leyo id.), ŋan ‘louse’ (cf. TN pəncyeq id.), paqłpta- (> Pur paqłta-) ‘deceive’ (cf. TN tempra- id.), pnsłi (~ psłi) ‘old’ (cf. TN nyewxi id.), py ‘aspen’ (cf. TN nyurka id.), syat ‘coal’ (cf. TN yataqma id.), syeju ‘soul, breath’ (cf. TN yntq ‘breath’), sy ‘tongue’ (cf. TN nyamyu id.), tłyał- ‘cry, shout’, tasy ‘merchandise’, tonł ‘grass’, tyanyu ‘broad’, xalaqku ‘animal, bird’ (cf. TN sarmyik id.), xlmyiqk ‘sable’ (cf. TN tos id.), waptq ‘cover, lid’. There are obvious archaisms among them, for instance Forest Nenets ‘tongue’ derives from Proto-Uralic, while ‘louse’ has a wide distribution among Samoyed languages, but from the point of view of the current topic, only the isoglosses matter. The Toropkova material also contains Forest Nenets neologisms such as wyt ŋłi ‘beaver’ (literally “underwater”; cf. TN lyidyaŋk ‘beaver’), jłknu (> Pur jłnu) ‘morning’ derived from jłk- ‘to get up’ (cf. TN xwi ‘morning’), or ŋta wyax ‘finger’ (literally “streak of the hand”; cf. TN ŋudah tarka ‘finger’), and of course Khanty loan-words not known in Tundra Nenets, notably jłnas ‘shirt’, jŋkł ‘mouse’, jelim ‘shame’, ŋasni ‘sheep’, wntł ‘otter’, wsya ‘hello’, xiłnyi ‘gold’.

Semantic differences, representing largely Forest Nenets innovations, also keep the Toropkova material firmly apart from Tundra Nenets: munuqŋat ‘I speak’ (cf. TN mnoqŋadm ‘I make noises’), myj ‘malitsa’ (cf. TN myuy ‘inner’), nyampa ‘forehead’ (cf. TN peya id., nyampa ‘crown of the head’), ŋmtt ‘a species of grass’ (cf. TN ŋəmteq ‘flower’), syn ‘steam, tobacco’ (cf. TN syun ‘steam’), wiqnyu ‘spring (season)’ (cf. TN weqnyuy ‘first fish in the spring’), xq- ‘to want, to be strong’ (cf. TN səq- ‘to be strong’, xərwa- ‘to want’), xt ‘oven (indoors)’ (cf. TN sd ‘clay’).

A wide range of irregular morphological and phonological correspondences can be detected between Forest Nenets and Tundra Nenets, and again, what Castrn recorded in Toropkova is in full accordance with other Forest Nenets data. Representative words include jłyu ‘friend’ (cf. TN yuryo id.), kontw ‘ptarmigan’ (cf. TN xontyeq id.), kjqku ‘birch’ (cf. TN xo id.), ksyeq ‘how’ (cf. TN xən-cyerq id.), nyłyi ‘grandfather’ (cf. TN yiryi id.), pyełyimytya ‘young woman’ (cf. TN pyryibtya id.), tym ‘tooth’ (cf. TN tyibya id.), wat (< watŋ) ‘corral’ (cf. TN waq id.), xoju ‘calf’ (cf. TN syu id.).

As opposed to the overwhelming presence of typically Forest Nenets features in its lexicon, there remain a handful of words in the Toropkova material that actually bear resemblance to Tundra Nenets unparalleled by any other attested Forest Nenets variety. Firstly, two words for ‘blood’ are recorded, namely the commonly known kem as well as weja, which is identical with Tundra Nenets weya. We may be dealing with an archaism of the Toropkova dialect, as weya also has cognates in Enets, and the common Forest Nenets wiji ‘broth’ can be understood as a derivation of the otherwise unattested weja. The second word in the Toropkova material with a cognate only in Tundra Nenets is juna ‘horse’ ~ TN yuna, an ancient Turkic loan-word in Samoyed, which must have been more widely used in Forest Nenets before the adoption of the Khanty loan-word law id. which appears in the Baly material. The third word that could be a common Nenets word only preserved in Toropkova is tiuko ‘lungs’, the root of which derives from Proto-Uralic, and a similar derivation tyiwak is the only Tundra Nenets word meaning the same, while the common Forest Nenets word for ‘lungs’ is kpuj, found in the Toropkova and Baly corpora as well. Further, the normal Forest Nenets word for ‘spruce’ is kq ~ kat, both variants being recorded from Baly, but the Toropkova material contains kaadы ~ kaadi, apparently close to and possibly borrowed from Tundra Nenets xadi id. The word maarat ‘town’ has a close parallel only in the European dialects of Tundra Nenets, far away from the Forest Nenets language area, where mərq is used in the same meaning, but given the unexpected ablative form maardahat, it cannot be the exact cognate of any other recorded word, so it remains a candidate for a Tundra Nenets loan. Lastly, and most conspicuously, we find the word šibeku ‘mummy’ with an unexpected internal consonant as well as vague meaning, as the usual Forest Nenets word is symyaqku ‘female parent bird’ ~ TN syibyako id. Since the Tundra Nenets by is the result of a recent sound change, this may indeed be a loan-word from Tundra Nenets in the Toropkova dialect.

Loans from Tundra Nenets are known to exist in Forest Nenets dialects, although they appear to be limited to a couple of words, notably jawi ‘polar bear’ < TN yawi id. and (Lyamin) paryiŋota ‘czar’ < TN (dial.) paryeŋoda id. Although many ‘Tundraisms’ were detected in the morphological data collected by Castrn, the Toropkova dialect possesses likewise very few Tundra Nenets loan-words. The logical conclusion of this discrepancy is that the morphological aberrance of the Toropkova material, itself subject to unwarranted variation, does not reflect the actual spoken dialect of the area, but the informants made a conscious effort to modify their performance following the model provided by the Tundra Nenets speaking Castrn himself, and their own, presumably rather meagre knowledge of Tundra Nenets; while few if any individuals in Toropkova were fluent in Tundra Nenets, they must have been acquainted with Tundra Nenets folklore, which continues to be popular far beyond the Tundra Nenets language area.

Castrn obviously noted the unusual presence of variants in his data, but given the short period of time he could spend with Forest Nenets speakers and the small number of them, even a scholar of his stature had no chance of retrieving the exact information. Crucially, with provisory Tundra Nenets elements now singled out from the Toropkova corpus, it is possible to reassess the significance of the actual Toropkova dialect for Forest Nenets studies.

References

Castrn 1846 = Anteckningar om Samojediskans frvandtskap med de Finska Sprken, af M. A. Castrn. Suomi, Tidskrift i fosterlndska mnen. 1845. Femte rgngen. Utgifven p Finska Litteratur-Sllskapets frlag. Helsingfors, J. Simelii Arfvingar, 1846. 177–186.

Castrn 1854 = M. Alexander Castrn’s Grammatik der samojedischen Sprachen. Im Auftrage der Kaiserlichen Akademie der Wissenschaften herausgegeben von Anton Schiefner. Nordische Reisen und Forschungen von Dr. M. Alexander Castrn [7]; St. Petersburg 1854.

Castrn 1855 = M. Alexander Castrn’s Wrterverzeichnisse aus den samojedischen Sprachen. Im Auftrage der Kaiserlichen Akademie der Wissenschaften bearbeitet von Anton Schiefner. Nordische Reisen und Forschungen von Dr. M. Alexander Castrn [8]; St. Petersburg 1855.

Castrn 1856 = M. Alexander Castrn’s Reiseberichte und Briefe aus den Jahren 1845–1849. Im Auftrage der Kaiserlichen Akademie der Wissenschaften herausgegeben von Anton Schiefner. Nordische Reisen und Forschungen von Dr. M. Alexander Castrn [2]; St. Petersburg 1856.

Castrn 1960 = Castrns Aufzeichnungen ber das Waldjurakische. — Samojedische Sprachmaterialien. Gesammelt von M. A. Castrn und T. Lehtisalo. Herausgegeben von T. Lehtisalo. Mmoires de la Socit Finno-Ougrienne 122; Helsinki 1960. 262–316.

Castrn mscr. = Anmrkningar fver den Kondinska dialecten af Samojediskan, gjorda i byn Toropkowa wid Obfloden 1845 om sommaren. Manuscripta Castreniana XI, Samoiedica 5, Jurak-Samoiedica 4: 219–294.