Symposium B3: Etymology and the diachrony of valency change in Uralic

Organizers: Riho Grünthal and Johanna Nichols

Monday, August 22 – Tuesday, August 23, Department of Slavic Studies (Slawistik), room 1

(Original symposium call)

Monday 22 August


Johanna Nichols:

Introduction: Etymology and the diachrony of valence change in Uralic

This introduction for Workshop B.3 will summarize the issues from our call for papers, namely:

Abundant derivational morphemes and overlapping lexical and functional properties in verbal morphology are characteristic of very many Uralic languages. One of the keys to examine these phenomena more closely is to clarify the relationship of underived and derived in verb sets such as causative verbs and decausativizing mechanisms. Given that derivational morphology is widely used in the Uralic languages, we seek to bring new light to its importance for etymology and the diachrony of languages. In this respect, the evidence of the Uralic languages is also of more general interest.

We are interested in identifying the role of different morphological and structural units in the diachronic development of valence and verb.

  • Does the evidence of Uralic branches and individual languages show similar patterns of change, or are they mutually contradictory?
  • How is grammatical information transferred from morphology to lexemes?
  • To what extent does the pairing of verbs, i.e. the relationship between an underived and derived words, reveal the diachrony of grammatically encoded lexemes?
  • What are the origins of valence-related suffixes in Uralic languages?
  • What valence-related derivational suffixes can be reconstructed for Proto-Uralic (PU) or intermediate branches?
  • For what PU verbs can valence be reconstructed?
  • Can we reconstruct PU derived verbs consisting of a verb root and valence-changing suffix?
  • Was causativization as predominant in PU as in many contemporary Uralic languages?
  • Was transitivity generally an inherent lexical property in PU, or were most/many verbs ambitransitive (labile)?

The introduction will justify and flesh out those questions and, where relevant, mention workshop papers that will bear on them and some results from the literature. It will raise two further questions that come from the organizers' own work:

  • What information needs to go into an etymological dictionary in order to facilitate comparative morphosyntax?
  • Does comparison based entirely on verbs yield different phylogenies and typologies from established ones?

It will close with a brief overview of the workshop papers and the issues they raise.


Kaisla Kaheinen:

Means of valency change in Samoyed languages – historical, areal, and universal tendencies

The Samoyed languages are often described as particularly synthetic among the Uralic language family, especially with reference to verbal morphology. The northernmost languages have the richest and most complex verbal morphology, with three inflections (intransitive, transitive, and reflexive) and a large number of derivational suffixes. These are sometimes hypothesized to point towards connections with the so-called Paleo-Siberian languages, although concrete evidence of contacts is lacking (cf. Katzschmann 2008: 28). A more close-up look into the valency changing strategies of four Samoyed languages – Kamas, Selkup, Tundra Nenets, and Nganasan – reveals that the patferns of causativization and decausativization in Samoyed resemble quite closely those found in Uralic languages in general (see Grünthal et al. 2021), although single verbs display particularly interesting areal connections as well as historical developments.

Based on data of 21 verbs from 21 Uralic languages collected during the UrCaus project, I compare verb-specific as well as language-specific patterns of valence change in Samoyed to the rest of Uralic. The canonical valence-changing operation in Samoyed is suffixal causativization, achieved by the use of *-ptA, a stlll productive causativizing suffix going back to Proto-Samoyed and, ultimately, to Proto-Uralic. However, individual verbs show a great deal of variation, such as sound change giving rise to synchronically opaque relationship between transitive and intransitive verbs of common origin. Verbs such as ’open’, ’break’, and ’hang’ seem to favor decausativizing patterns. An interesting parallel between Selkup, Kamas, and Ob-Ugric languages is found in a metaphoric extension of the verb pair ’eat/feed’ to the meaning ’burn’ – literally ’to be eaten by fire’. The development cannot be of recent origin, and thus it has to be either an old East Uralic areal feature or a coincidence based on the metaphoric attractiveness of the concept of eating.

In general, verbs displaying divergent patterns seem to be either language-specific, or belong to the so-called core vocabulary, and thus less susceptible to regularizing analogy. Valency-changing operations in Samoyed do not diverge from common Uralic patterns, and at least this part of Samoyed morphosyntax does not seem to be affected greatly by contact-induced change.

Grünthal, Riho; Arjava, Heini; Lehtinen, Jyri & Nichols, Johanna 2021: Basic causative verb patterns in Uralic: Retention and renewal in grammar and lexicon. – Silvia Luraghi & Elina Roma (eds.): Valency over time. Diachronic perspectives on valency patterns and valency orientation. 209–234. London–New York: Mouton de Gruyter.
Katzschmann, Michael 2008: Chrestomathia Nganasanica. Texte – Übersetzung – Glossar – Grammatik. – Norderstedt: Books on demand GmbH.


Riho Grünthal:

Causative alternation in Uralic: valence in transition

Since a wordlist-based typology of basic valence orientation was proposed in 2004 (Nichols et al. 2004) there has been work on individual languages, some historical comparison using the typological idea but no exhaustive survey of a language family, and virtually no work of this type on Uralic. This paper remedies these gaps with an exhaustive survey of the Uralic languages and dialects for which adequate lexical coverage is available, using an expanded version of the Nichols et al. wordlist, and proposing some major typological and historical trends in Uralic.

Semantically, the list consists of universally basic meanings, characteristically expressed by verbs. For the Uralic languages two parameters are crucial. Firstly, is the non-causal member of the pair an intransitive or transitive verb and, secondly, is the verb semantically animate or inanimate, i.e. what kind of semantical correlation does it have with the subject (if intransitive) or object (transitive).

The verbs include basic human actions as ‘laugh’, ‘eat’, ‘fear’, universally attestable meanings as ‘die’ which also shows more semantic and language-specific variation in the nature of the process (Botne 2003) but also others which are conceptualized in more multiple ways, for instance ‘be open’, ‘dry’, ‘awake’ etc. It is characteristic of Uralic that the causal member of pairs such as ‘feed’ corresponding to ‘eat’ is formally derived from the non-causal, whereas other strategies such as reducing non-causals from causal verbs occur considerably less frequently in comparison with causativization. Certain verbs such as Finnish syödä, North Saami borrat and Mari kočkaš ‘eat’ can be ambitransitive. The basic pattern of North Saami borrat ‘eat’ : borahit ‘feed’ illustrates the derivation from an underived verb stem (1–2) (cf. Nickel & Sammallahti 2011: 580–591)

North Saami

(1)Dat mearkkaša ahte dat borrá eará elli-id...
it mean.3SG that it eat.3SG other animal-PL.GEN(-ACC)
‘It means that it eats other animals.’

(2) Ma-id sii galge-t bora-hi-t mána-i-de?
What-PL.GEN(-ACC) they must-3PLeat-CAUS-INF child-PL-ILL
‘What should they give the children to eat?’

The relationship between the basic verb and its causal counterpart may vary lacking a productive causative or decausative marker in either of them as in the corresponding Udmurt verb pair śiyny ‘eat’ : śudyny ‘feed’. In the given case the causativizing strategy is based on suppletion which, arguably, is frequent in Uralic. A fundamental difference between the intransitive and transitive verbs is that the causativization of transitive verbs does not always involve valence change. The core questions of our paper are:

(1) What kind of derivational suffixes are used for causativization, what is their distribution and do they represent different diachronic layers in Uralic?

(2) What kind of similarities and dissimilarities can be found in the causativization of intransitive and transitive verbs?


Botne, Robert 2003. To die across languages. Linguistic Typology 7. 233–278.

Dolovai, Dorottya. 2001. A többszörös műveltés az obi-ugor nyelvekben. Néprajz és nyelvtudomány 41. 77–93.

Nichols, Johanna, David A. Peterson, and Jonathan Barnes. 2004. Transitivizing and detransitivizing languages. Linguistic Typology 8:2. 149–211.

Nickel, Klaus Peter and Pekka Sammallahti. 2011. Nordsamisk grammatik. Karasjok: Davvi Girji.

Shibatani, Masayoshi and Prashant Pardeshi. 2002. The causative continuum. Masayoshi Shibatani, ed., The Grammar of Causation and Interpersonal Manipulation, 85–126. Amsterdam: Benjamins.


Jeremy Bradley, András Czentnár & Nikolett F. Gulyás:

Valence changing strategies in the Volga-Kama linguistic area

The paper at hand represents a survey of contact linguistically motivated convergence and divergence within the Volga-Kama Sprachbund. The Turkic and Uralic languages of this region (Tatar, Bashkir, Chuvash; Mari, Udmurt, peripherally Mordvin, Komi-Permyak, and Komi-Zyrian) all make extensive use of both valence-increasing (i.e. transitivizing, causative, etc.) and valence-decreasing (i.e. reflexive, passive, etc.) derivational suffixes. In all languages there is a strong preference for valence-increasing operations, but Russian with its strong preference for valence-decreasing operations might be strengthening the position of reflexive/passive suffixes in the region. The specific valence-changing suffixes used in these languages are not generally borrowings; their cognate forms are widely used throughout their respective language families. Our focus will thus lie on their function rather than on their form: we will survey the typological microvariation that can be attested in the region in the domain of valence-changing operations. The research questions we have investigated (by consulting reference materials, corpora, and native speakers) include:

i) Can more than one causative suffix attached to a verbal stem in a variety (1)?
ii) Does a variety allow for double accusative marking in a causative construction (2)?
iii) Does a variety allow for quasi-causatives (3)?
iv) Does a variety allow for impersonal (subjectless) passives that can have a direct object marked with the accusative case (4)?
v) Does the variety allow for a canonical passive construction with an overt Agent (5)?
vi) Does a variety allow for a valence-decreasing suffix to be used to indicate the non-volitional nature of an activity (6)?

Our talk will examine these questions within the languages of the region, focussing on convergence across genealogical boundaries and divergence within genealogical groupings. Special attention will be given to closely related languages or varieties that have been subject to different contact situations (Meadow Mari ↔ Hill Mari, Moksha Mordvin ↔ Erzya Mordvin, Komi-Zyrian ↔ Komi-Permyak, Tatar ↔ Bashkir), and indeed exhibit both structural and functional differences in this domain. We will examine how these differences correlate to their different historical contact situations.

(1) Tudo vaty-ž-lan vüd-ym yry-kt-ykt-en.
3sg wife-px3sg-dat water-acc heat-caus-caus-3sg.pst2
‘He made his wife heat water.’ (Meadow Mari)

(2) Vospitaťeľńica nylaš-ez iźy-t-e.
kindergarten_teacher girl-acc sleep-caus-prs.3sg
‘The kindergarten teacher makes the girl sleep.’ (Udmurt)

(3) Min-e qaltıra-t-a.
1sg-acc tremble-caus-prs.3sg
‘I’m trembling.’ (Tatar)

(4) Ala-kunam ožno suas jylmy-m=at tunem-alt-yn.
sometime earlier Tatar language-acc=and study-refl-pst2.3sg
‘Sometime in the past, Tatar was studied as well.’ (Meadow Mari)

(5) ?Tyn-eš pur-ty-maš [...] akušerka dene yšt-alt-yn.
faith-lat enter-caus-nmlz [...] midwife with do-refl-pst2.3sg
‘The baptism was done by a midwife.’ (Meadow Mari)

(6) Menam onmöśśi-ś-öma.
1sg.gen fall_asleep-refl-ptcp.pst
‘I fell asleep (unintentionally).’ (Komi-Permyak)





Juha Kuokkala:

The Uralic passive-automative *-w : Reflexes and reconstructions reconsidered 

In the established Proto-Uralic reconstruction, a detransitivizing verbal suffix *-w is reconstructed on the basis of the Saami -ōj ~ -uvv- (SaaN oidnot ‘to be seen’, čoalbmaduvvat ‘to get entangled’), Finnic -u/-ü (Fi. kuulua ‘to be heard’), Mordvin -v (Erza nejavoms ‘to be seen’) automative/passive suffixes and the Mansi passive marker -w (Middle Konda teewø ‘he gets eaten’). In Finnic languages, the suffix has a variant -pu/-pü, usually occurring after a long stressed root syllable as in Fi. juopua ‘to get drunk’, but in Livonian also after bisyllabic roots, e.g. rikābõ ‘to get broken, spoiled’. This variant has sometimes been addressed in the literature as a “strong grade” form of -u/-ü, though more often as prothesized or based on the -pa/-pä participle. As more substantive evidence in favor of the “strong grade variant” hypothesis has apparently not been presented before, a survey is conducted in the current presentation to find out whether there are grounds for reconstructing a detransitivizing Uralic *-p(V) suffix and what relationship this kind of a suffix may have to the traditional set of *-w suffixes. In particular, possible unnoticed cases of ancient consonant-stem derivatives are scrutinized, such as SaaS gåarpodh ‘to get ice crust (of water)’ ~ SaaL goarre ‘hoarfrost’ (< PU *kari ‘skin, bark’) and Liv. kuorbõ, Est. kõrbeda, Erza kurvams ‘to burn (intr.)’ ~ SaaN goardit ‘to heat, roast’. Finally, the implications of reconstructing *-p instead of or in parallel with *-w are discussed, considering the overall picture of Proto-Uralic morphology as well as the possible solutions offered to the problem of divergent reflexes of inherited labial vowel suffixes in Saami and Finnic.

Tuesday, 23 August


Juho Pystynen:

A Proto-Uralic causative derivation pattern redefended 

According to the standard model of root structure in Proto-Uralic (PU), nominal and verbal basic roots were bisyllabic and ended in a vowel. Only two types of root-final ("stem") vowels have been generally accepted: the *A-stems, ending in open *a/*ä, and the *ә-stems, ending in a more closed vowel (notated also *i, *ii, *e). As a part of the lexical root, stem vowels are usually not considered to carry any specific meaning.

Katz (1996), however, has proposed a causativizing "stem ablaut" pattern *-ә- → *-A-. Some of the Samoyedic data Katz takes as a starting point is by now reconstructed differently: as a primary transitive *A-stem verb and a derived reflexive verb in *-O- < PU *-A-w- (cf. Salminen 2012: 340). However, other parts of his data seem to remain viable. My talk will build further on evidence especially from Samic and Finnic, where the distinction between PU *-A-, *-ә- and *-V-w- remains clearly observable. Numerous new doublets can be added beside the previously known cases, most often between a nominal in *- and a verb in *-A-. Examples include:

*ipsә 'smell' ~ *ipsä- 'to smell' (> e.g. Northern Sami haksa ~ haksit)
*künčә 'nail' ~ *künčä- 'to plough' (> e.g. Finnish kynsi ~ kyntää)
*kajә 'dawn' > *kaja- 'to dawn' (> e.g. Finnish koi ~ kajastaa)
*je̮kә 'divide' ~ *je̮ka- 'to divide' (> e.g. Erzya jav ~ javoms)

Instead of the pre-Uralic vowel contraction scenario sketched by Katz, I however propose a link to the known process of consonant-stem alternants of PU *ә-stems, allowing a simple agglutinative analysis of *A-causatives, e.g. *ips- → *ips-ä-. This derivation pattern has particularly interesting implications for Mari. It has long been known that the two verb conjugations in the language correlate with transitivity: em-verbs tend to be transitive, am-verbs intransitive. Ravila (1939) has however proposed the contrast to originally derive from the PU stem vowel contrast: *A-stem verbs > em-verbs, *ә-stem verbs > am-verbs. The transitivity correlation Ravila attributes in part to the PU transitivity-increasing suffixes *-ta-, *-tta-, *-kta-, and the denominal intransitive verb suffix *--; he is more skeptical of the proposal by Budenz to treat am-verbs as continuing PU reflexive derivatives in *-w-. The current proposal now allows additionally treating seemingly unsuffixed em-verbs as continuing PU *A-causatives, and a few cases can be indeed identified where a transitive em-verb corresponds to cognates otherwise pointing to a PU *ә-stem, e.g. Mari kuškems 'to churn butter' ~ Finnish koskea 'to touch'.

Katz, Hartmut. 1996. Eine ururalische Kausativbildung. – Mészáros, Edit (ed.): Ünnepi könyv Mikola Tibor tiszteletére, 156–164. Szeged.
Ravila, Paavo. 1938. Über die entstehung des tscheremissischen Konjugationssystems. – FUF 25, 1–25.
Salminen, Tapani. 2012. Traces of Proto-Samoyed vowel contrasts in Nenets. – Per Urales ad Orientem. Iter polyphonicum multilingue (SUST 264), 339–358.


Heini Arjava & Riho Grünthal:

Valence-changing morpheme stacking in Finnic and other Uralic languages: The diachrony of fused stems 

Typologists have long been aware of the close diachronic relationship between causativizing and intransitivizing morphology, and causatives are typical diachronic sources of passives and reflexives (see Haspelmath 1990, Zúñiga & Kittilä 2019: 226 and passim). This connection can also be observed in the history of verb derivation in many Finnic languages, where verbs with valence-changing morphology have served as stems for new valence-changing derivations, frequently resulting in stacked morpheme chains, as in (1) below.

Our study is based on a database of 22 Uralic languages, consisting of 21 verb sets of intransitives and their corresponding semantic causatives (e.g. ‘die’ and ‘kill’) (cf. Nichols, Peterson & Barnes 2004). The pairs of the set are often derived from each other either by causativizing, decausativizing or both, and in Finnic, the typical morphemes used in this valence alternation are the inherited suffixes -tA (Tr), and -u- (Intr). A historical valence-changing chain of four suffixes is illustrated by the verb istuutua 'sit down' in Finnish (1a), and similar chains also feature in Veps (1b) and Votic (1c); note that the same original stem produces more stacking in Veps.

(1a) Finnish: (WU) *isa- ‘sit’ > TR *ista- > INTR istu- ‘sit’ (arch. ‘sit down’) > INTR istu-utu- ’sit down’

(1b) Veps: (F) *süte- ‘strike (flint)’ > TR *sütt- > INTR süttu- > TR sütuta- ‘ignite’

(1c) Votic: (F) *süte- ‘strike (flint)’ > TR süttä- ‘ignite’

In our paper, we will analyze questions and hypotheses including, i) are transitive or intransitive suffixes more prone to fuse into verb stems, and what is the distribution of alternating transitive-intransitive stacking (as in examples above) vs. the rarer Tr-Tr or Intr-Intr stacking; ii) how do verb semantics correlate with the loss of segmentable suffixes; iii) how much do individual verb stems vary in their degree of fusion and semantic fading, and iv) how much of the stacking can be explained by the reanalysis of valence-changing morphemes as telic aspectual markers?

The diachrony of stacking is diverse and transparent in Finnic, but we will also give examples of the other, less studied Uralic branches. More generally speaking, derivation and valence change have gained only little attention in Uralic studies, and hardly any from the viewpoint of contemporary typology.


Haspelmath, M. (1990) The grammaticization of passive morphology. Studies in Language. International Journal sponsored by the Foundation “Foundations of Language”, 14(1), 25-72.

Hakulinen, L. (2000) Suomen kielen rakenne ja kehitys.

Nichols, J., Peterson, D. A., & Barnes, J. (2004) Transitivizing and detransitivizing languages.Linguistic typology, 8(2), 149-211.

SSA = Suomen sanojen alkuperä: Etymologinen sanakirja. Eds. Itkonen, E., U.-M. Kulonen et al. (1992) Helsinki: Suomalaisen Kirjallisuuden Seura.

Zúñiga, F., & Kittilä, S. (2019). Grammatical Voice. Cambridge: Cambridge University Press.


Jack Rueter:

Transitivity morphology in literary corpora of three Uralic languages, tip of the iceberg

This paper makes a quick survey of verbs directly related to an 18-verb list (Nichols et al 2004) and their most frequent corresponding forms in corpora from the three Uralic languages, Erzya, Moksha and Komi-Zyrian. It notes there are numerous verbs that occur before the first representatives of the verb list and seeks to apply their morphological analysis to the understanding of verb list members.

Although some open-source corpora is available through the Universal Dependencies project and the Fennougrica collections at the National Library of Finland, additional corpora used for Erzya and Moksha are from the ERME-corpora, much still to be published, and the Komi-Zyrian corpora developed by FULab in Syktyvkar. The initial method is to find the most frequent representative from each of the verb types and then extract information on what kinds of morphology is encountered. This, of course, is only the first step in developing an understanding of inflectional variation to be found in a transitivity morphology of Uralic languages.

Of the verb types found higher in frequency, it will be noted that modal verbs expressing necessity or possibility appear to be in the present or non-past, as do the copulas. There is a tendency to be in a past tense when the meaning of the verb involves the notion of inchoativity, change of state or completion in Erzya and Moksha.In Komi-Zyrian, however, verbs such as ‘say’ and ‘it appears’, one used with parataxis, tend to be in the present. In Moksha, the verb ‘say’ invokes a subject-object conjugation in this very position, and it is only with the verb ‘ask (someone)’ that the subject-object conjugation occurs in both Erzya and Moksha.

Tense variation can be observed in literary corpora. While all three languages adhere to a third person singular agreement on the verb, there is a disparity with regard to tense. The Komi-Zyrian sample prefers first preterit in 18 verb types, whereas Erzya has two finite verb in non-past, and Moksha four.

Two of the three languages can indicate both the subject and the object in the conjugation, and this is used in four of the verb types. Erzya and Moksha both use a subject-object strategy for the verbs ‘see’ (non-derived transitive), ‘open’ (ambitransitive) and ‘hang’ (augmented transitive), Erzya applies this strategy to the verb ‘straighten’ (augmented transitive), as well.

Etymological, causative marking is not always what it looks like. It is observed in 5 of the Erzya verbs, whereas two are transitive, one is ambitransitive, one is intransitive, and the fifth with a causative + frequentative derivation set is intransitive. Moksha only has two verbs with causative marking, of which one is transitive and the other a (caus. + freq.) intransitive.

The verbs for laughter in all three languages prefers a reductive derivation expressing a sudden eruption of laughter .