Navigation – Plan du site

AccueilNuméros8(2)On noncausal/causal alternations ...

On noncausal/causal alternations in Tima (Nuba Mountains, Sudan)

Gertrud Schneider-Blum


Le tima, une langue parlée dans les monts Nouba au Soudan, dispose de trois marqueurs de dérivation (avec divers allomorphes) pour augmenter ou diminuer la valence du verbe. La voix moyenne et le marquage antipassif sont utilisés pour réduire la valence. Le marquage causatif peut, d’une part, augmenter la valence des verbes monovalents et, d’autre part, fonctionner comme un marqueur anticausatif/passif pour les anciens verbes transitifs. La distribution de ces marqueurs de dérivation à travers les schémas d’alternance formelle augmentation, réduction et double marquage (tels que décrits par Nichols et al. 2004) est étudiée afin d’établir certains groupes de verbes. En raison du comportement particulier du marqueur causatif, à savoir qu’il peut être utilisé pour former un verbe causal ainsi qu’un verbe non causal, le lien entre les fonctions causatives et passives est examiné plus en détail en rapprochant les données du tima à un chemin de grammaticalisation, tel que suggéré par Haspelmath (1990), du causatif au passif en passant par le réflexif-causatif. Le tima s’écarte de ce chemin : une fonction causative-réflexive n’est pas attestée ; à la place, nous trouvons une fonction anticausative. L’hypothèse peu probable que la construction ergative du tima provient d’une grammaticalisation supplémentaire du morphème causatif-passif est également considérée.

Haut de page

Texte intégral

1. Introduction

  • 1 Examples for ambitransitive verbs are ɘ́màh ‘eat’, kɨ́mʌ́n ‘get satisfied, have eaten enough’ and (...)

1Tima, a language spoken in the Nuba Mountains of Sudan, has three verbal derivation marking options which play a role in noncausal/causal alternations, namely antipassive, middle and causative. These marking devices are briefly introduced in Section 2 and will then be discussed thoroughly in relation to noncausal/causal alternations in the main body of this contribution. At the same time, the question of whether the established noncausal/causal morphological patterns correlate with three of the different formal types described by Nichols et al. (2004: Section 2.4) (see also Haspelmath et al. 2014: Section 2, with differing terminology) will be studied. These are the strategies of augmentation (the noncausal verb is not marked, while the causal verb receives morphological marking), reduction (the causal verb is not marked, while the noncausal verb is overtly marked), and double derivation (in which both the noncausal and the causal verbs are marked). Thus, the present study focuses on the three derivation markers antipassive, middle and causative and their relationship to the formal types mentioned above. Other strategies, as established by Nichols et al. (2004), such as the ambitransitive strategy or suppletion, will not be considered here.1 See Table 1 for Tima examples of the three alternation patterns mentioned (‑ɨk and ‑ɘk are two realisations of the causative morpheme, ‑ɨl is one of the middle allomorphs, and ‑ak is one of the two antipassive allomorphs).

Table 1 — Formal noncausal-causal alternation patterns

Formal pattern Noncausal Causal Gloss
Augmentation tʌʌn tʌʌn‑ɨk boil
Reduction kudʌŋ‑ɨl kudʌŋ flip over
Double derivation kaar‑ak kaar‑ɘk grow

2The choice of verbs considered for the discussion was inspired by the list as compiled by Nichols et al. (2004: 156) as well as by the verb collection suggested by Dom et al. (2021), who in turn refer to Haspelmath’s (1993: 97) list of inchoative/causative verb pairs. Tima verbs for which the noncausal form occurs solely in a derived adjectival stem (see fn. 25), were excluded. Of course, verbs for which the relevant data was incomplete could not be considered either (e.g. ‘close’, ‘open’, and ‘freeze’). On the other hand, whenever the availability of data allows, other Tima verbs were added to the investigation (such as ‘free’ or ‘flip over’). Due to this adapted choice, not only does the question of whether the distribution of alternation patterns corresponds to a general tendency (see Haspelmath 1993: Section 6) remain to be answered in future research, but the presentation of solid numbers relating to “which correspondences are most and least frequent […], and what other typological properties are correlated with valence orientation” (Nichols et al. 2004: 155) also has to wait until the relevant data is available.

3This study on the noncausal/causal alternation ties in with a broader research topic, namely, to isolate the conditioning factors for choosing the ergative construction over other types of constructions in Tima, as well as trying to establish the origins of ergative marking in this language. Early during this research, the team working on Tima hypothesised that semantic causation is one of the crucial factors in that respect. Thus, it seemed appropriate not to take a purely form-based approach to causative marking but rather to look at noncausal/causal alternations of verbs. As will be shown below, groups of verbs which behave similarly (with respect to derivation marking and function) can be established. What remains to be analysed is the abstract logic behind the groupings. Hence, this part of the present publication is best characterised as a presentation of intermediary results from work in progress.

  • 2 Tima is not the only language in the wider area (covering Sudan, South Sudan, Ethiopia and beyond) (...)

4The established formal verb structures will also be related to possible word orders, thereby touching on the issue of subjecthood. By doing so, the discussion by Casaretto et al. (2020) on the emergence of Tima ergativity, in which causation plays a considerable role, will be enriched. The data which will be presented here supports the assumption formulated by Casaretto et al. (2020) that the passive construction is most probably not the origin of the ergative construction in Tima. In accordance with them, I assume — based upon both old and new data — that the rich derivation marking system, with interfering and overlapping meanings, might have been the source for this development (consider mainly Section 4 for a discussion of this).2

5The remainder of this contribution is structured as follows. After providing some background information in Section 2, which mainly serves the purpose of facilitating understanding of the examples in the main body of the text, I go into detail regarding noncausal/causal alternations in Tima in Section 3. This section will be subdivided into several sub-sections according to alternation patterns. Furthermore, the interplay between a causative and a passive/resultative reading of the causative marker with reference to middle marking will be considered in depth, and correlations to different realisation options of transitive verbs will be revealed. In Section 4, I will attempt to connect the data and findings with a grammaticalisation process, as described by Haspelmath (1990), and try to show how ergativity in Tima fits in. Here, I will also consider parallels between the situation as described for Tima with the findings known for Bantu languages. The last section will summarise the findings and formulate remaining thoughts and questions.

2. Background information

6Tima (tms) is an endangered language spoken in the Nuba Mountains of Sudan by roughly 7,000 speakers (Figure 1). Tima belongs to the Katloid language group, together with Katla and Julut (see e.g. Dimmendaal 2018: 383). Whether Katloid is a branch of Kordofanian, which in turn is part of the “Congo-Kordofanian” language phylum (as classified by Greenberg 1963), later relabelled as “Niger-Congo” (Williamson 1989; Schadeberg 1989), is nowadays dealt with rather cautiously (see e.g. Dimmendaal 2009; 2011; 2018; Schadeberg & Blench 2013; Blench 2013; Hammarström 2013; see Güldemann 2018 for a comprehensive review of the classification of African languages in general, and Kordofanian in particular).

Figure 1 — The Tima region in the Nuba Mountains (altitude: 600‑800 m)

Figure 1 — The Tima region in the Nuba Mountains (altitude: 600‑800 m)
  • 3 For most of the consonants, the relevant IPA symbol is used, with the following exceptions: retrofl (...)
  • 4 Tone in Tima has not yet been systematically investigated; therefore, the examples are phonetically (...)
  • 5 Dimmendaal (2009) was the first to sketch the ATR system of Tima; see also Bashir (2010; 2013) for (...)
  • 6 Tima orthographic symbols as opposed to IPA symbols:

7Tima has 21 consonants and 12 vowels.3 Like the related language Katla, Tima is a tone language and has vowel harmony with regard to the position of the tongue root.4 This means that it has two groups of vowels, categorised as [+ATR] vowels vs. [‑ATR] vowels.5 In a basic lexeme, only vowels of one or the other group may occur. For the sake of legibility, the graphemes, as agreed upon by the Tima community, represent the difference in ATR as a difference in vowel height (and not by means of the IPA diacritics).6 Most affixes harmonise with the ATR value of the root. Unlike Katla (Birgit Hellwig, p.c.), many of the affix and clitic vowels also harmonise with the adjacent root vowel with regard to frontness (see Bashir 2010).

  • 7 For the pronominal system in Tima, see Schneider-Blum (2013a).

8With respect to word order, Tima shows a tendency to prefer SV/AVO, as in (1), although OVA word order is also well attested, mainly occurring in longer chunks of discourse. With OVA word order, A is marked for ergative case by the homorganic nasal N=, as in (2). In ergative constructions, verbal indexing is with the agent, illustrated by (3) where cɛ̀ŋ ‘1sg:prf’ indexes the ergative agent, which is instantiated by the pronominal enclitic erg1sg’.7 The choice of transitive construction — AVO or OVA — depends on which constituent the speaker chooses to make the attentional centre, which occupies the sentence-initial position (for more details on the impact of the attentional centre on the choice of construction see Schneider-Blum forthcoming; see also Schneider-Blum & Hellwig 2018).

  • 8 The Tima data used for this paper was collected between 2007 and 2022. The source of a recorded exa (...)
(1) Áhmέt ánt̪ànà álɪ̀.
áhmέt án‑t̪ànà álɪ̀
Ahmed 3p:prf‑call Ali
‘Ahmed has called Ali.’ (20180130_1)8
(2) Álɪ̀ ánt̪ànà ŋ̀áhmɛ̀t.
álɪ̀ án‑t̪ànà ŋ̀=áhmɛ̀t
Ali 3p:prf‑call erg=Ahmed
‘Ahmed has called Ali.’ (20180130_1)
(3) Cíbʌ́ úkwʌ̀yʌ́wú cɛ̀ŋwúdʌ́yíkát̪áŋná.
c‑íbʌ́ ú=k‑wʌ̀yʌ́wú cɛ̀ŋ‑wúdʌ́‑y‑ík=á=t̪áŋ=ná
sg‑child dir=sg‑sausage_tree 1sg:prf‑burn‑ep‑caus=source=loc3p=erg1sg
‘I’ve burnt the fruit of the sausage tree.’ (20180130_17)

9Further possible word orders, such as AOV and OAV, are only allowed in connection with focus marking (for details see Schneider-Blum 2018). If the situational context allows for it, anaphoric zero reference for agent and patient is possible, with the following restrictions: A can only be zero if O is not the sentence-initial attentional centre. In other words: A cannot be zero if marked for ergative case.

10The verbal morphological structure is rather complex: TAM, person and derivation marking, as well as pronominal cross-referencing and several more or less productive derivation markers, are attested (see Table 2).

Table 2 — The verbal complex in Tima

  • 9 In the structure of the inflected verb, the first person marker is not obligatory.
Position Function Syntactic category
1 negation proclitic
2 aspect proclitic
3 (person (S, A))9 prefix
4 tense prefix
5 person (S, A) prefix
6 ROOT root
7/8/9/10 derivation 1/2/3/4 suffixes
11 completive/locative enclitic
12 applicative (DAT) enclitic
13 pronominal S enclitic
14 pronominal O enclitic
15 negation enclitic
  • 10 V of the middle and the causative suffix is the abstract representation of a vowel which harmonises (...)

11Tima uses two valency-decreasing devices, i.e. middle, marked by ‑Vl, and antipassive, marked by ‑Ak.10 Also attested is causative marking with ‑Vk, which may on the one hand increase the verb’s valency, but, in the circumstances described below, may also serve as a valency-decreasing device. Middle, antipassive and causative are the main derivational suffixes found in noncausal/causal pairs. They are mutually exclusive. Furthermore, the Tima verbal system is equipped with both ventive (valency neutral) and instrumental marking (valency increasing) devices, neither playing a major role in the coding of causation. All derivation markers mentioned so far are optional suffixes added only under relevant conditions (frozen forms are not considered here).

  • 11 “[…] high transitive marking in Tima, expressing a uniactional event, stands in paradigmatic opposi (...)

12Tima allows for high transitive and pluractional marking on some (but not all) transitive verbs. Verbs marked for high transitivity refer to a single action. The use of the pluractional verb stem indicates a repeated or long-lasting action. Additionally, a few verbs display a difference between low transitivity and pluractional marking through tone marking (see Dimmendaal & Schneider-Blum 2018: 133).11 The valency status of a transitive verb is not affected by (high or low) transitive vs. pluractional marking. In the context of the noncausal/causal alternation, transitive verbs, whether morphologically marked for transitivity/pluractionality or not, are treated as underived, as explained in more detail in Section 3.

13The marker for high transitivity can generally be isolated. The relevant morpheme is i/ɪ, in rare cases adapted to an adjacent vowel or merging with it. The marker for low transitivity is a. In contrast, pluractionality in Tima, applied to transitive as well as intransitive verbs, is not indicated by a single marking device and its marking generally cannot be separated from the root, unless we are dealing with reduplication.

Table 3 — Sequence of affixational derivation markers

Root (+ Slot 1) Slot 2 Slot 3 Slot 4
root (tr./plur) antip (‑Ak) mv (‑Vl) caus (‑Vk) vent (‑Vŋ) ins (‑aa)
  • 12 The middle voice morpheme has been labelled “reversive” or “separative” in earlier publications on (...)
  • 13 The combined source marker and locative marker, at̪aŋ, was formerly analysed and labelled as a comp (...)

14All marking options are sorted into four slots, depending on their vicinity to the verbal root (see Table 3). Slot 1 is reserved for transitivity or pluractionality marking, which are mutually exclusive. In Slot 2 we may find antipassive, middle and causative marking.12 There is no example attested that contains a combination of all three, or even of two of the three. Slot 3 is occupied by the ventive marker, and Slot 4 is reserved for the instrumental marker. In contrast to these derivational suffixes, the Tima valency increasing applicative morpheme ii/ɪɪ is considered a clitic, since the prepositional source marker a and the locative pronominal t̪aŋ, referring to an oblique referent, may intervene between the stem and the applicative marker (see Dimmendaal et al. in preparation).13

15Information on person and tense/aspect/mood (TAM) is generally coded before the root, if at all, and is dependent on word order and focus marking. If information on both TAM and person is present, person marking is interwoven with TAM marking. For 1st and 2nd persons, we additionally find pronominal enclitics, as in (3) with the ergative enclitic for 1sg. The ergative encliticised pronominal is only morphologically differentiated from the non-ergative form for 1sg, while the forms for 2sg/2pl and 1pl coalesce. Thus, the assignment of these forms to ergative case can only be made by analogy with the 1sg and third person forms. 3sg and 3pl have no encliticised forms, but their independent forms oppose ergative and non-ergative forms under the relevant conditions. They behave like ordinary nouns.

3. Noncausal/causal alternation options

16The valency for the majority of Tima verbal roots is inherent. That is, a certain verb is either monovalent, as is the case with áwʊ̀n ‘move (a little)’, or bivalent, as with táàn ‘beat’. Neither of these two verbs is provided with morphological marking indicating its transitivity status.

17The richness in derivation marking options allows easy altering of the valency of verbs. For instance, an intransitive verb may be suffixed by the causative marker to increase its valency. Consider the following two examples, (4) with the noncausal form of the verb and (5) with the causal one.

(4) Íídì àn‑tʌ́ʌ̀n.
íídì àn‑tʌ́ʌ̀n
pl:water 3p:prf‑boil
‘The water is boiling.’ (05.03.10_08_11)
(5) Tʌ́ʌ́nɨ̀k íídí.
tʌ́ʌ́n‑ɨ̀k íí
boil‑caus pl:water
‘Boil the water.’ (22.09.07‑122sg)

18Such alternations, in which the noncausal verb is not marked, while the causal verb is derived for the causative (augmentation strategy), are further discussed in Section 3.1.

19Section 3.2 considers verbs which at least synchronically have only derived forms. To be more exact, the verbs discussed have a noncausal antipassive form and a causal causative-marked form, such as pɨ̀ɽɨ́tʌ̀ʌ̀k ‘get free, escape’ vs. pɨ̀ɽítììk ‘let free’. In such cases, we are dealing with double derivation.

20The use of the causative marker as a valency decreasing (rather than increasing) device, with an anticausative or passive notion, is elucidated in Section 3.3. However, causative marking is just one device that is used in Tima in order to reduce a verb’s valency. The other one is the middle voice marker ‑Vl. Some transitive verbs have two causal forms that depend on event number, namely transitive and pluractional. Such verbs allow for two derived noncausal forms (middle or causative marking, respectively), again depending mainly on event number. Verbs which have only one transitive form invariably use the causative marker to decrease their valency in order to achieve a passive/resultative notion.

21Examples (6)‑(9) give some insight into what is to come. While (6) and (7) present the causal verb differentiated for pluractionality, the noncausal verb is derived by the middle voice in (8) to indicate a single event, and by the causative morpheme in (9) to indicate that several actions have been performed.

(6) ɽúmít̪áŋ súkwéérná ɪ̀ɪ̀sháhɪ̀.
jùɽúm‑í=t̪áŋ súkwéér=ná ɪ̀=ɪ̀sháhɪ̀
sink‑ht=loc3p sg.sugar=dem.prox dir=tea
‘Sink a piece of candy into the tea.’ (20211227_6)
(7) ɽùmàt̪áŋ yʌ́wùhnʌ̀ yànt̪ɪ̀ íídì.
júɽùm=à=t̪áŋ y‑ʌ́wùh=nʌ̀ y‑ànt̪ɪ̀ íídì
sink=source=loc3p pl‑stone=dem.prox loc‑inside pl:water
‘Sink the stones into the water.’ (20211227_6)
(8) Tɔ́lɔ́ ánjùɽùmùlàt̪áŋ.
tɔ́lɔ́ án‑jùɽùm‑ùl=à=t̪áŋ
sg.boat 3p:prf‑sink‑mv=source=loc3p
‘The boat sank.’ (20211227_6)
(9) Ìɽɔ̀lɔ́ ànjúɽùmùkàt̪áŋ.
ɪ̀ɽɔ̀lɔ́ àn‑júɽùm‑ùk=à=t̪áŋ
pl:boat 3p:prf‑sink‑caus=source=loc3p
‘The boats sank.’ (20211227_6)

22Of course, from a morphological point of view, verbs marked for high or low transitivity or pluractionality are derived verbs, but, as mentioned before, this derivation does not affect the valency of the verb. Also, as outlined in Section 2, transitive (high as well as low) and pluractional marking are assigned to one slot in the verb’s structure, while the three valency changing devices, middle, antipassive and causative, are assigned to another. That is why, with regard to the noncausal/causal alternation, both pluractional and the high and low transitive stems are treated as underived. Hence, the noncausal derivation here is following the reduction strategy.

23In the following sub-sections, I will go into detail regarding each pattern.

3.1 Non-derived noncausal meets derived causal

24The common feature of the noncausal verbs discussed in this section is that they are all intransitive and underived, while their causal counterparts are derived with the causative marker ‑Vk caus’ (see Table 4). The alternation between noncausal and causal events hence follows the augmentation pattern.

Table 4 — Augmentation strategy in Tima

Formal pattern Noncausal verb Causal verb
augmentation underived causative ‑Vk

25They are further subdivided into two groups (see Table 5). With the first group, the subject of the noncausal event is in the role of an agent, as in (10). The subject agent of the noncausal verb becomes an object agent of the causal verb; see (11) with an animate human agent and (12) with an inanimate agent.

  • 14 The k of certain roots, e.g. (k)at̪am ‘leave’, (k)aar ‘grow’, (k)amʊ ‘wash’, and (k)ʌbu ‘dig, plant (...)
(10) Kùhùnʌ̀ŋʌ̀ pɨ́nʌ́ kát̪ám.14
kùhùnʌ̀ŋ=ʌ́ pɨ́nʌ́ kát̪ám
now=foc pro3sg leave
‘(S)he has just gone (out).’ (19.02.07‑10)
(11) Át̪ámʊ̀k álɪ̀.
át̪ám‑ʊ̀k álɪ̀
leave‑caus Ali
‘Make Ali go out (e.g. by telling him).’ (20180130_12)
(12) Át̪ámʊ̀k íídí ʊ́twááɽɪ̀.
át̪ám‑ʊ̀k íídí ʊ́=twááɽɪ̀
leave‑caus pl:water dir=outside
‘Let/Make the water go out (e.g. by opening a barrier or by using a broom).’ (20180130_12)

26Animacy or volitionality seem not to be the decisive features in these constructions. As we saw when comparing (11) with (12), the object of the causal clause may be animate/volitional or inanimate/non-volitional, and this is also the case with the causer in such causal events; compare (13) with an animate and in principal volitional causer and (14) with an inanimate and non-volitional causer.

(13) Wáyέn àŋdɪ́yànɪ̀ɪ́k kɨ̀címbʌ́rí.
wáyέn àŋ‑dɪ́yànà‑ɪ́k kɨ̀címbʌ́rí
sg.father 3p:prf‑laugh‑caus sg.child
‘The father made the child laugh.’ (20.09.07‑38)
(14) T̪àmááná kɘ́málńa ándɪ́yànɪ̀ɪ̀kɨ́dʌ̀.
t̪‑àmáá= kɘ́‑mál=ńa án‑dɪ́yànà‑ɪ̀k‑ɨ́=dʌ̀
sg‑talk=dem.prox sg‑good=dem.prox 3p:prf‑laugh‑caus‑ep=1sg
‘The funny speech made me laugh.’ (20.09.07‑40)

27In the second subgroup, the subject of the intransitive/noncausal event is in the semantic role of an undergoer. In the causal event, the undergoer is coded as the object. The plain verb, which is an inchoative verb, is basic, as shown in (15); the transitive/causal verb is again derived for the causative, as demonstrated in (16) (see also (4) and (5)).

  • 15 This example has been elicited with material compiled by Skopeteas et al. (2006).
(15) Íɪ̀hwáà kwíyʌ́wááyáŋ úkúrtú tíín […] ɲ̀céŋwúdáát̪àŋ […].
ɪ́=ɪ̀hwáà kwíyʌ́‑w‑áá=yáŋ ú=kú‑rtú ꜜtíín […]
cond=people cop‑ep‑ins=loc3p dir=sg‑house inside
ɲ̀‑céŋ‑wúdʌ́=á=t̪àŋ […]
‘If people are inside the house, […], they will burn […].’ (Hamid_Skopeteas_15_1_2_D3b)15
(16) Cíbʌ́ ùkwʌ̀yʌ̀wú àŋwúdʌ́yíkát̪áŋ ŋ̀kày kàmɘ́lɘ̀l.
c‑íbʌ́ ù=k‑wʌ̀yʌ̀wú àŋ‑wúdʌ́‑y‑ík=á=t̪áŋ
sg‑child dir=sg‑sausage_tree 3p:prf‑burn‑epcaus=source=loc3p
ŋ̀=k‑ày kàmɘ́lɘ̀l
erg=sg‑the_one exorcising
‘The healer has burnt the fruit of the sausage tree.’ (20180130_16)

28Table 5 provides a non-exhaustive listing of examples of verbs with underived noncausals and causative-derived causals.

Table 5 — Non-derived noncausal/causative-derived causal verb forms

  • 16 In the case of wudʌ ‘burn’, the causal alternation also exhibits a differentiation between a singul (...)
Underived intr. Semantic role of subject Semantic value Derived tr. caus Semantic role of object
át̪àm agent leave át̪àmʊ̀k agent
dɪ̀yánà agent laugh dɪ̀yánɪ̀ɪ̀k agent
ʌ́t̪ì/ʌ́t̪ù agent lie down ʌ́t̪ììk agent
wudʌ undergoer burn wùdʌ́yìk/wúdèèk16 undergoer
hʊwan undergoer dry hʊ́wánt̪ɘ̀k undergoer
tʌʌn undergoer come to
tʌ́ʌ́nɨ̀k undergoer

29The semantic distinction seen in the intransitive event, with the subject being either in the role of an agent or of an undergoer, is also reflected in causal events. The former becomes an object agent, the latter an object undergoer (see Table 6).

Table 6 — Semantic roles in noncausal/causal constructions (augmentation)

Noncausal underived verb Causal derived verb
1st subgroup S(actor)V A(actor)VO(actor)
2nd subgroup S(undergoer)V A(actor)VO(undergoer)

30In terms of their argument structure constructions, causal events allow for both AVOcausee (e.g. (13)) and OcauseeVAcauser:erg (e.g. (16)) configurations. Thus, causative-derived causal verbs behave like any other transitive verb.

3.2 Derived noncausal meets derived causal

  • 17 Note that transitive verbs that are not causative-marked, i.e. high or low transitive and pluractio (...)

31In this alternation pattern we find an opposition between antipassive and causative forms, i.e. both the noncausal and the causal verbs are derived (see Table 7). The root can only be reconstructed by analogy with other verbs and their derivational potential, or by considering the form of the verbal noun, formed by a number marking prefix, the root and the middle voice suffix. Compare, for this purpose, the causative-marked verb forms opposed to the non-derived noncausals in Section 3.1. In its core function, the antipassive form allows for the omission of the direct object, which is implicitly still understood, while the causative marker demands that the direct object, which is still to follow, is mentioned (or recoverable from the preceding discourse). Antipassive marking is of interest for this paper on the causal/noncausal alternation because it may also have an anticausative function and may convey a reciprocal/reflexive/autobenefactive notion (see also Dimmendaal & Schneider-Blum 2018; Veit in preparation).17

Table 7 — Double derivation strategy in Tima

Formal pattern Noncausal verb Causal verb
double derivation antipassive ‑Ak
anticausative ‑Ak
causative ‑Vk

32The canonical function of the morpheme ‑Ak is illustrated with (17). With both derived verbs in the following two examples, the ‘sorghum’ is the affected undergoer, either implicitly, as in (17), or explicitly, as in (18). Both the antipassive-marked form and the causative-marked form have the agent in subject position. A reflexive/anticausative reading is then excluded.

(17) Ìwɔ̀rmáádɘ́hná cént̪úyùwʌ̀k.
ɪ̀wɔ̀rmáádɘ́h=ꜜná cén‑t̪úyù‑w‑ʌ̀k
pl:man=dem.prox 3p:ipfv‑thresh‑ep‑antip
‘The men are threshing.’ (12.04.09‑02‑04)
(18) Nʌ́t̪úyúk yέɛ̀h.
nʌ́‑t̪úyúk y‑έɛ̀h
2pl‑thresh:caus pl‑sorghum
‘Thresh (pl) the sorghum.’ (06.04.09_26‑13)

33In (19), which comes from a description about the use of plants, the complements of all three causative-marked verbs refer to an undergoer that, in this case, is not mentioned, since the referent, namely ‘the fruits of the baobab’, is deducible from the previous discourse.

(19) Ìt̪úyúkát̪áŋ ɪ̀tʊ́ɽʊ́ɽɪ́ɪ́kát̪áŋ íbʌ̀ ɪ̀bàhbàh ìkʌ́lùknɛ̀έy.
ì‑t̪úyúk=á=t̪áŋ ɪ̀‑tʊ́ɽʊ́ɽɪ́‑ɪ́k=á=t̪áŋ
pl‑thresh:caus=source=loc3p pl‑pull_out‑caus=source=loc3p
íbʌ̀ ɪ̀‑bàhbàh ì‑kʌ́lùk=nɛ̀έy
children‑astringent pl‑chew:caus=1pl.incl
‘We beat them, we take out the seeds, which have an astringent effect, we eat them.’ (31.01.08_33, use of plants (Adlaan): 48)

34Most occurrences of ‑Ak in our corpus entail the deletion of the object while the agent is preserved, thus, its label as antipassive marker is justified. However, since examples with the core function of antipassive marking are not relevant for the purposes of this paper, they are not further discussed here.

35While the detransitivised/antipassive forms of the verbs in (17) and (19) still pragmatically imply a direct object that is different from the agent, with other verbs, the antipassive form indicates that the action/spontaneous event is directed towards the agent itself. In the first case, it has an anticausative function, see the intransitive verb in (20) and compare it to the causal one in (21). In the second case, the verb conveys a reflexive/autobenefactive notion, see (25). Thus, antipassive marking, in addition to its canonical function of allowing to omit the object, may allow for the omission of an agent and preserves the patient, hereby gaining a reflexive/autobenefactive or anticausative function (Table 8).

36Again, the causee of the causative-marked verb is an undergoer, see (21)‑(22).

(20) Cɪ̀bɪ́ àŋkááràk. vs. (21) Áárɘ̀k cɪ̀bɪ́.
cɪ̀‑bɪ́ àŋ‑káár‑àk áár‑ɘ̀k cɪ̀‑bɪ́
sg‑tree 3p:prf‑grow‑antip grow‑caus sg‑tree
‘The tree has grown.’ ‘Grow a tree.’ (08.04.09, 2_04‑01)

37The following example pairs, (22)/(23) and (24)/(25), illustrate the causative and antipassive forms (the latter with an anticausative/reflexive function) of pɨɽitV ‘free’ and amV ‘wash’. The quality of the final root vowel cannot be determined as the verb does not occur in a simplex, non-derived form. When contrasting the causative form with the antipassive form, it becomes obvious that the final root vowel completely assimilates to the initial vowel of either suffix.

(22) Ádɪ̀yààmák ɨ́pɨ̀ɽìtíík kɘ̀mɘ̀rt̪áà.
ádɪ̀yààmák ɨ́‑pɨ̀ɽìtíík kɘ̀‑mɘ̀rt̪áà
hereafter p‑free:caus sg‑horse
‘Hereafter, they set the horse free.’ (01.10.07‑11, Adlaan Misiria, horsequarrel 57)
(23) Ímɪ̀ɪ̀ ámpɨ̀ɽɨ̀tʌ́ʌ́k.
ɪ́‑mɪ̀ɪ̀ ám‑pɨ̀ɽɨ̀tʌ́ʌ́k
pl‑goat 3p:prf‑free:antip
‘The goats escaped.’ (20180130_15)

38Similarly, (k)ámʊ̀ʊ̀k ‘wash:caus’ (with the meaning ‘wash sb.’), exemplified with (24), is opposed to (k)ámààk ‘wash:antip’ (with the meaning ‘wash yourself’), illustrated with (25). Example (24) also proves that causal verbs from this group can have their word order reversed to OcauseeVAcauser:erg (as is the case with the causative-derived verbs in Section 3.1).

(24) Kɨ̀címbʌ́rí cɛ̀ɛ̀kàmʊ́ʊ́knɛ̀έy.
kɨ̀címbʌ́ cɛ̀ɛ̀‑kàmʊ́ʊ́k=nɛ̀έy
sg.child 1pl:prf‑wash:caus=erg.1pl.incl
‘We washed the child (*We make the child wash itself/somebody/something).’ (Skype20180418)
(25) ɛ̀n àŋkʌ̀mùh kɨ̀címbʌ́rí ɘ̀kámààk
wέɛ̀n àŋ‑kʌ̀mùh kɨ̀címbʌ́rí ɘ̀‑kámààk
sg.mother 3p:prf‑let sg.child p‑wash:antip
‘The mother let the child wash (him-/herself).’ (example provided courtesy of Nataliya Veit)

39Some verbs which receive morphological marking in both the noncausal and the causal context are listed in Table 8. As already mentioned, though structurally comparable, the semantics of the derived noncausal verbs differ in that they either have a canonical antipassive reading or they indicate a spontaneous event / have a reflexive/autobenefactive reading. Table 9 illustrates the semantic roles of S/A and O referents in a schematised way.

Table 8 — Antipassive-derived noncausal/causative-derived causal verb forms

  • 18 It is not possible to form the imperative of aarak ‘grow (intr)’, since, as the speaker expressed h (...)
Derived intr.
Semantic role of
Derived tr.
Semantic role of
ɽʊ̀wàk agent clear ɽʊ̀k undergoer
t̪úyùwʌ̀k agent thresh t̪úyùk undergoer
t̪ɔ́ɔ̀l‑àk agent clean t̪ɔ́ɔ̀l undergoer
pɨ̀ɽɨ́tʌ̀ʌ̀k undergoer (refl) escape/free pɨ̀ɽítììk undergoer
ámààk undergoer (refl) wash ámʊ̀ʊ̀k undergoer
aarak18 undergoer (spont) grow áárɘ̀k undergoer

Table 9 — Semantic roles in noncausal/causal constructions (double derivation)

Noncausal derived verb Causal derived verb
1st subgroup S(actor)V A(actor)VO(undergoer)
2nd subgroup S(undergoer)V A(actor)VO(undergoer)

3.3 Derived noncausal meets non-derived causal

40The verbs listed in this section can again be divided into two subgroups; in both cases, the causal verb may appear unmarked or marked for transitivity or in its pluractional form (see Table 10).

  • 19 A grammatical constraint prevents the form marked for high transitivity to additionally take the mi (...)

41In the first group, the verb is either derived for middle voice or for causative marking in order to become intransitive, the conditions for which are explained below. In the second subgroup, only one noncausal verb form is attested, derived for the causative and having a passive/resultative notion. The division within the noncausal domain corresponds to the division in the causal domain. That is, we have two noncausal forms if there are two causal ones, and only one noncausal form if there is only one causal form. For the reasons outlined above, the causal forms are considered non-derived and hence basic, that is, the noncausal/causal alternation follows the reduction strategy.19 The subject of the noncausal event is the undergoer. We are thus dealing with a patient preserving valency decrease.

  • 20 See Kulikov (2011: 374; 2013: 267), who differentiates between a canonical passive (where the direc (...)

42As the examples will show, the detransitivising function of ‑Vk oscillates between an agentless passive and an anticausative meaning. Formally, the two readings cannot be differentiated, since, as will be discussed below, mentioning the agent in a passive construction is not possible in Tima.20

Table 10 — Reduction strategy in Tima (patient preserving)

Formal pattern Noncausal verb Causal verb
Reduction middle ‑Vl high transitive
anticausative/passive ‑Vk pluractional
no transitive marking

43Some verbs which differentiate single vs. plural events in the noncausal as well as the causal form are listed in Table 11.

Table 11 — Middle voice/causative noncausal/transitive causal verb forms

  • 21 When mɨ̀rɨ́nɨ̀l carries the meaning of ‘separate yourself (from others)’, the imperative is accepta (...)
intr. mv
intr. caus
role of
role of
mɨrɨnɨl21 mɨrɨnʌrɨnɨk undergoer divide mɨ̀rɨ́nɪ̀ mɨ̀rɨ́nʌ́rɨ̀n undergoer
cɪmɘl cɪmɘk undergoer gather cɪ̀mɪ́ cɪ̂m undergoer
kùdʌ́ŋɨ̀l kudʌŋɨk undergoer flip over kùdʌ́ŋì kùdʌ̀ŋ undergoer
ɽúmùl juɽumuk undergoer sink ɽúmì ɽùm undergoer

44Consider first the following two transitive/causal sentences. In (26), the causal verb is again marked for high transitivity, as shown with mɨ̀rɨ́nì ‘divide’. In (27), the verb occurs in its pluractional form, with pluractional marking being expressed here by partial reduplication.

(26) Mɨ̀rɨ́ní ɪ̀hwáánǎ.
mɨ̀rɨ́n‑í ɪ̀hwáá=nǎ
divide‑ht people=dem.prox
‘separate these people (context: they are fighting)’
(27) Wɔ́rt̪ɘ́máádɘ́h ámɨ̀rɨ̀nʌ̀rɨ̀nàt̪áŋ cɪ́bɪ́.
wɔ́rt̪ɘ́máádɘ́h á‑mɨ̀rɨ̀nʌ̀rɨ̀n=à=t̪áŋ cɪ́‑bɪ́ 3p:prf‑divide:plur=source=loc3p sg‑tree
‘The man divided the tree into pieces.’

45The following examples show that this verb also differentiates the intransitive/noncausal form for event number. Examples (28) and (29) show noncausal single events, while (30) and (31) illustrate plural events with the noncausal verb. When comparing these four examples, we can observe that the verb can be applied to inanimate items, as in (28) and (30), as well as animates, as in (29) and (31).

(28) Kʌ́wúhnʌ̌ àtɪ́ɪ́n dámák mɨ̀rɨ̀nɨ̀làt̪áŋ (ɲɪ̀hɪ̀ɪ́k).
k‑ʌ́wúh=nʌ̌ à‑tɪ́ɪ́n dámák
sg‑stone=dem.prox‑one then
mɨ̀rɨ̀n‑ɨ̀l=à=t̪áŋ (ɲ=ɪ̀‑hɪ̀ɪ́k)
divide‑mv=source=loc3p ins=pl‑two
‘This boulder split into two.’ (16.04.09‑16‑11x) (Figure 2)

Figure 2 — Stimulus for (28)

Figure 2 — Stimulus for (28)
(29) Làmàn ìjíyʌ́ŋ nʌ̌ŋ ŋ̀kɔ́ ɨ́mɨ̀rɨ̀nɨ̀làá.
làmàn ì‑jíyʌ́ŋ nʌ̌ŋ ŋ̀kɔ́ ɨ́‑mɨ̀rn‑ɨ̀l‑àá
when p‑came here cop p‑divide‑mv‑ins
‘When they came here, they (i.e. the Inguunyang people) divided into two.’ (031007_Daldum_ClanDividing 034)
(30) Cɪ̀bɪ́ ámɨ̀rɨ̀nʌ̀rɨ̀nɨ̀kàt̪áŋ.
cɪ̀‑bɪ́ á‑mɨ̀rɨ̀nʌ̀rɨ̀n‑ɨ̀k=à=t̪áŋ
sg‑tree 3p:prf‑divide:plurcaus=source=loc3p
‘The tree divided (branched) into many.’ (Figure 3)

Figure 3 — Stimulus for (30)

Figure 3 — Stimulus for (30)
(31) […] wʌ̀ kwɔ̀ɔ̀ lììní kʊ̀kɔ̀lɔ̀ŋɪ́ ɪ̀hwáá kárkámánέ; ɪ̀hwáá kárkámán ḿmɨ̀rɨ̀nʌ̀rɨ̀nɨ̀kàt̪áŋ, kwɔ̀ɔ́nɔ̀ lììní mák kɪ́tɪ́ɪ́nɪ́ kʌ́t̪ʌ̀ndʌ̀lí.
wʌ̀ kwɔ̀ɔ̀ lììn=í kʊ̀‑kɔ̀lɔ̀ŋ=ɪ́ ɪ̀hwáá
and family_father poss1pl.excl=sel sg‑big=sel people
kárkámán=έ; ɪ̀hwáá kárkámán ḿ‑mɨ̀rɨ̀nʌ̀rɨ̀n‑ɨ̀k=à=t̪áŋ, people Karkaman p‑divide:plurcaus=source=loc3p
kwɔ̀ɔ́=nɔ̀ lììn=í mák kɪ́‑tɪ́ɪ́n=ɪ́
family_father=dem.prox poss1pl.excl=sel then sg‑other=sel
‘[…] and our big family constitutes the people of Karkaman; the Karkaman people divided (into several subgroups), then our family is that of Kathanda’ (Videos_10.2007_15_AhmedKurmaadi_1‑6)

46Examples (26)‑(31) also show that the use of the pluractional form of the verb is not correlated to the number of the object in the causal event or, for that matter, the subject of the noncausal event, but rather to the number of actions/events happening.

47The derivation pattern of the translational equivalent of ‘gather’ is similar, in that we have two transitive forms (ht, as in (32) and (33), and plur, as in (34) and (35)) and two detransitivised forms, which are marked as middle (36) and causative (37), respectively.

(32) Cɪ̀máát̪áŋ yέɛ̀h.
cɪ̀m‑ɪ́=á=t̪áŋ y‑έɛ̀h
gather‑ht=source=loc3p pl‑sorghum
‘Gather the sorghum.’
(33) Mák cɪ̀mɪ́ɪ̀ŋ ɪ̀hwáà.
mák cɪ̀m‑ɪ́‑ɪ̀ŋ ɪ̀hwáà
then gather‑ht‑vent people
‘Then he gathered the people (around him).’ (010207_Jenge_LionHyena 033)
(34) Ìmúrɪ́k àɲcɪ́mát̪áŋ yέέh lɪ́yɔ̀ɔ̀.
ì‑múrɪ́k àɲ‑cɪ́m=á=t̪áŋ y‑έέh l‑ɪ́‑yɔ̀ɔ̀
pl‑Tima 3p:prf‑gather:plur=source=loc3p pl‑sorghum loc‑pl‑home
‘The Tima people have gathered the sorghum in their homes.’
(35) Ìhwáà àɲcɪ́m ŋ̀wɔ́rt̪ɘ́máádɘ̀h.
ɪ̀hwáà àɲ‑cɪ́m ŋ̀=wɔ́rt̪ɘ́máádɘ̀h
people 3p:prf‑gather:plur
‘The man gathered the people (at different places, in groups).’

48The verb cɪ̀mɪ́/cɪ̂m ‘gather’, in its high transitive and pluractional forms, entails a plural object; thus, the number of the object cannot have an influence on the type of verbal transitive marking that occurs. Which form is chosen rather depends on whether the items are gathered in one place (one action/event) or in groups/piles in different places (several actions/events). The correlations with single vs. plural events are again reflected in the noncausal forms in (36) and (37). Gathering in one place requires middle marking, as in (36), while gathering at different places demands causative marking, as in (37). The verb in both sentences is, but does not have to be, encliticised by =at̪aŋ. Its presence indicates that all the people who were expected are present.

(36) Ìhwáà ɪ̀‑cɪ́m‑ɘ́l=á=t̪áŋ.
ɪ̀hwáà ɪ̀‑cɪ́m‑ɘ́l=á=t̪áŋ
people p‑gather‑mv=source=loc3p
‘the people gathered (at one spot)’ (010207_Jenge_LionHyena 039)
(37) Ìhwáà àɲ‑cɪ́m‑ɘ̀k=à=t̪áŋ.
ɪ̀hwáà àɲ‑cɪ́m‑ɘ̀k=à=t̪áŋ
people 3p:prf‑gather: plurcaus=source=loc3p
‘The people have gathered (at different places).’

49As mentioned earlier, the subject of the intransitive, noncausal form discussed in this section has the role of an undergoer, as is also the case in (36) and (37). This is less clear, though, with the noncausal forms of ‘gather’ and ‘flip’ than with the noncausal forms of ‘divide’ or ‘sink’. With regard to ‘gather’ and ‘flip’, certain agentivity features (in the sense of Dowty 1991) are imperative as essential aspects of the subject. Thus, the semantics of ‘gather-mv’ does not allow yέɛ̀h ‘sorghum’ as the sole participant of the noncausal event due to the lack of any agentivity feature, as illustrated with (38). This also shows in the ungrammaticality of the English translation *‘the sorghum has gathered (in one place)’, which obviously mirrors Tima in this case.

(38) *Yέɛ̀h àɲcɪ́mɘ̀l/àɲcɪ́mɘ̀làt̪áŋ/àɲcɪ́mɘ̀k/àɲcɪ́mɘ̀kàt̪áŋ.
y‑έɛ̀h àɲ‑cɪ́m‑ɘ̀l(=à=t̪áŋ)
pl‑sorghum 3p:prf‑gather‑mv (=source=loc3p)
3p:prf‑gather‑caus (=source=loc3p)
‘The sorghum has been gathered.’
  • 22 “If verbs are marked by reduplication (partly or entirely), the meaning carries the notion of doing (...)

50What is possible in Tima, though, is the following construction with the reduplicated root cɪmcɪm.22 Suffixed by the causative marker, a passive reading of the causative, which will be discussed in detail below, is given. The encliticised =at̪aŋ in (39) is not obligatory. Its presence tells us that the gathering is complete; its absence indicates that it has only started.

(39) ɛ̀h àɲcɪ́mcɪ́mɘ̀kàt̪áŋ.
y‑έɛ̀h àɲ‑cɪ́mcɪ́m‑ɘ̀k=à=t̪áŋ
pl‑sorghum 3p:prf‑gather:reduplcaus=source=loc3p
‘The sorghum has been gathered.’
  • 23 Consider, in this context, Copley & Wolff (2014) on dispositionality requirements.

51By contrast, íídì ‘water’ is permitted as the subject with the middle-marked verb, as (40) shows. The feature that allows its occurrence is its potential to move autonomously.23 Note that íídì should be considered a plural/collective noun; it shows plural agreement on adjectival modifiers.

(40) Íídí àɲcɪ́mɘ́lát̪áŋ.
íídí àɲ‑cɪ́m‑ɘ́l=á=t̪áŋ
pl:water 3p:prf‑gather‑mv=source=loc3p
‘The water has gathered (in one place).’
  • 24 Middle marking is very productive, though, when it comes to adjectives (with a resultative notion) (...)

52The verbs of the second subgroup to be discussed here can only be detransitivised by using the causative marker to gain a passive/resultative interpretation, that is, they have only one noncausal form with the subject in the role of an undergoer. The causal verb may be differentiated for its degree of telicity, if the verbal semantics allow for that. For instance, the verb ‘hang’ has the telic high-transitive form tùhí ‘hang it up’ and the less telic pluractional – but still transitive – form túùh ‘hang them up (one by one)’, while the verb táàn ‘beat’ is inherently durative and thus has only one transitive form. When wanting to express ‘hitting once’, a different verb is used, i.e. hɔ́ɔ̀ ‘hit once’, which also has its own causative-marked passive form. Likewise, Tima has two verbs for ‘bite’, one expressing a single biting action, while the other one is used for repeated biting (see Table 12). For the semantically singular noncausal event, especially for the translational equivalent for ‘hang’, one might expect middle voice marking, but this is not the case.24

Table 12 — Causative-derived noncausal/transitive causal verb forms

  • 25 Again, it is because the feature control is missing that the imperative form cannot be formed, as t (...)
Derived intr. caus Semantic role of subject Semantic value Tr. causal form Semantic role of object
wʌrɨk undergoer lose wʌr25 undergoer
ŋalt̪ɘk undergoer miss target ŋàlɪ́ undergoer
tanɪɪk undergoer address, call tàná undergoer
hɔyʊk undergoer beat (uniactional) hɔ́ɔ̀ undergoer
taanɘk beat (pluractional) táàn
kalɘmɘk undergoer bite (uniactional) álɘ̀m undergoer
limiyiik bite (pluractional) límìyì
tuhuk undergoer hang (up) tùhí undergoer

53Mentioning an oblique with the noncausal verbs is possible, as demonstrated with (41), but mentioning the agent would yield an ungrammatical sentence, as illustrated with (42).

(41) Kúù cɛ̀ntáánɘ̀k./Kúù cɛ̀ntáánɘ̀k ɲ̀cʊ̀ràŋ.
k‑úù cɛ̀n‑táán‑ɘ̀k / k‑úù cɛ̀n‑táán‑ɘ̀k ɲ̀=c‑ʊ̀ràŋ
sg‑dog 3p:ipfv‑beat‑caus sg‑dog 3p:ipfv‑beat‑caus ins=sg‑stick
‘The dog is being hit./The dog is being hit with a stick.’ (20180130_1)
(42) *Kúù cɛ̀ntáánɘ̀k ŋ̀wɔ́rt̪ɘ́máádɘ̀h.
k‑úù cɛ̀n‑táán‑ɘ̀k ŋ̀=wɔ́rt̪ɘ́máádɘ̀h
sg‑dog 3p:ipfv‑beat‑caus ins/
‘The dog is being hit by the man.’

54The agent can only be overtly mentioned as subject of an underived transitive verb. This is true for any of the verbs listed in Table 11 and Table 12.

55In both subgroups discussed in this section, the semantic roles of S in the noncausal event with a derived verb and A in the causal event with an underived verb differ, while S in the former and O in the latter event share their roles (see Table 13), illustrating the patient preserving strategy.

Table 13 — Semantic roles in noncausal/causal constructions (reduction)

Noncausal derived verb Causal underived verb
S(undergoer)V A(actor)VO(undergoer)
  • 26 The only known exception of a middle-marked verb with passive reading is found with tòmó ‘kill’.
  • 27 Consider Kulikov (2013: 272f.): “Distinguishing passives without an Agent from nonpassive intransit (...)

56The discussion in this section shows that both middle marking and causative marking vary between passive and anticausative interpretation, with a strong tendency of the former to correlate with anticausative and the latter with a less obvious tendency to passive use.26 With some verbs, such as those listed in Table 12, the causative-derived noncausal can only be interpreted as passive. The question arises of why verbs follow different patterns and receive different interpretations. Considering the verbs in Table 12, it seems as if those verbs that allow for causative marking as the only detransitivising device designate actions that imply the existence of an agent.27 By contrast, the verbs listed in Table 11 do not presuppose that such an agent is responsible for the action, neither when being middle-marked nor when receiving causative marking. That is, an anticausative interpretation is more likely. Future research will hopefully shed more light on this issue.

4. Excursion: From causative to passive (and ergativity?)

  • 28 Tima is not unique in having a causative/passive polysemy. Kulikov (2001: 894), for example, lists (...)

57As has been demonstrated in the previous sections, the ‑Vk morpheme in Tima does in fact occur in clear causation contexts (see the verbs in Sections 3.1 and 3.2), but it covers additional functions, namely anticausative and passive (see Section 3.3).28 In this section, I would like to have a deeper look into the connection between the canonical use of the Tima causative marker and its use as a passive marker and will also consider a connection between Tima passive and the ergative construction.

58In order to see whether we are in fact dealing with a passive in Tima, I first cite the criteria for the passive in general, as stated by Haspelmath (1990), and compare them to the Tima data.

“A construction is called passive if:

  1. the active subject corresponds either to a non-obligatory oblique phrase or to nothing; and
  2. the active direct object (if any) corresponds to the subject of the passive; and
  3. the construction is somehow restricted vis-á-vis another unrestricted construction (the active), e.g. less frequent, functionally specialized, not fully productive.” (Haspelmath 1990: 27)

59All three criteria are met with the Tima data:

  1. A passive reading which mentions the agent is not attested. There are a few elicited examples where the cause (but not the causing agent) is optionally mentioned. With unmarked word order, the cause is postverbal and procliticised by a preposition.
  2. With a passive reading of ‑Vk-marked verbs, a change in word order is always involved, i.e. the object of the transitive sentence becomes the subject of the passive sentence and governs verbal agreement.
  3. Causative constructions in general, and specifically those with a passive interpretation, are less frequent than active sentences, as a count of causative marking in a number of recorded texts proves.29 Also, only certain transitive verbs authorise a passive reading.

4.1 Haspelmath’s grammaticalisation path

60The question arises as to how a causative morpheme can develop a passive function. Haspelmath (1990), relying on Keenan (1985) and Kormušin (1976), suggests several possible sources for passive morphology, among them a path which might be considered relevant for Tima, namely from causative (as source) via reflexive-causative to passive and possibly ergative (see Figure 4, based on Diagram 1 in Haspelmath (1990: 54), with the functions which are synchronically attested in Tima highlighted). After summarising Haspelmath’s argumentation, I will illustrate the attested different functions, causative, anticausative and passive, with Tima examples in Section 4.2.

Figure 4 — Passive grammaticalisation paths based on Haspelmath (1990: 54), red outlines added

Figure 4 — Passive grammaticalisation paths based on Haspelmath (1990: 54), red outlines added

61Haspelmath (1990: 46) illustrates the transition from causative to passive with English examples.

  1. I have the barber shave me. (causative)
  2. I have myself shaved by the barber. (reflexive-causative)
  3. I am shaved by the barber. (passive)

62According to Haspelmath (1990), the focus on the result of an action instead of on the activity itself is the turning point for causative forms being allowed to occur in constructions without mentioning the causer.

63The author compares passives arising from causatives with those from NP reflexives. Both involve “a loss of agency condition on causatives” (Haspelmath 1990: 47), whereas the “affectedness of the patient in causatives is another important similarity to the passive, where affectedness of the patient is often implied” (Haspelmath 1990: 47). Thus, “inactivization of the verbal situation, not participant backgrounding or foregrounding is generally the original (and in this sense basic) function of the passive” (Haspelmath 1990: 26). Furthermore, “[t]he two functions of agent backgrounding and patient foregrounding follow automatically from inactivization” (Haspelmath 1990: 60). Haspelmath also points out that one of the potential other functions of the passive is that of a resultative, i.e. it may refer to the “state resulting from a preceding action; with transitive verbs, the patient is the subject of the resultative, just as in the passive” (Haspelmath 1990: 33).

4.2 How does Tima fit into the scheme?

64While causative marking for the majority of Tima verbs serves to either increase or decrease the verb’s valency, there are a few verbs which have the ‑Vk extension, but a basic underived verb form with the inflected verb is not attested. We find the underived base only with the verbal noun (similar to the verbs we saw in Section 3.2 on double derivation). These verbs may occur, as expected, in a transitive context, as with the verb ‘drip’ in (43): dúkùk ìwʌ́ʌ̀ ‘make the ash water drip’.

65At least one of these derived verbs can also occur in a noncausal intransitive context, as in (61) ‘rain is dripping’. That is, synchronically, we find the same verb form with a canonical causative function and describing a natural process/spontaneous event.

(43) Dúkùk ìwʌ́ʌ̀. vs. (44) Kwààɽɘ́k céndúkùk.
dúk‑ùk ìwʌ́ʌ̀ kwààɽɘ́k cén‑dúk‑ùk
drip‑caus ash_water rain 3p:ipfv‑drip‑caus
‘Make the ash water drip.’ ‘Rain is dripping (i.e. it’s drizzling)’

66With other ‑Vk-marked verbs, we may encounter an anticausative notion as well as a passive notion. Consider (45), where we find an anticausative context with the verb ‘lose’, translated here as ‘it disappears’.

(45) Íìyʌ̀ʌ̀lúyé ílmʌ̀ŋʌ̀ŋt̪áŋ, pɨ́nʌ̀ wʌ́rɨ̀kàt̪áŋ.
í=ìyʌ̀ʌ̀lú=yé ílmʌ̀ŋʌ̀ŋ=t̪áŋ, pɨ́nʌ̀ wʌ́r‑ɨ̀k=à=t̪áŋ
cond=pus=foc neg=loc3p 3sg lose‑caus=source=loc3p
‘If it (i.e. the swelling) doesn’t contain pus, it disappears.’ (310108_33_Adlaan_UsePlants 012f.)

67In (46), this same verb has a passive reading (and speakers claim that there must have been somebody who lost the pen).

(46) Gálɘ́m àŋwʌ́rɨ̀kàt̪áŋ.
gálɘ́m àŋ‑wʌ́r‑ɨ̀k=à=t̪áŋ
sg.pen 3p:prf‑lose‑caus=source=loc3p
‘The pen got lost.’ (Skype20180418)

68Note that the causal verb àŋwʌ́rát̪áŋ ‘3p lost sth.’ is not causative-derived. Nor is the verbal noun kùwʌ̀rɨ́l ‘losing sth.’.

69Let us go back to the path described by Haspelmath (1990) and compare it to the situation found in Tima. The intermediate stage of reflexive-causative (cf. Haspelmath’s example I have the barber shave me) is not attested in Tima. Instead, we find a causative-marked verb with an anticausative function indicating a resultative state-change.

Haspelmath (1990) causative reflexive-causative passive
Tima causative anticausative passive

70Though in some way similar, there is a crucial difference between the two constructions: with the reflexive-causative, there is still an actor different from the undergoer performing the action, while, in the anticausative construction, the subject is the undergoer.

reflexive-causative I have the barber shave me S (actor) ≠ undergoer
anticausative he/she/it got lost S = undergoer

4.3 From passive to ergative?

71Haspelmath (1990: 62) further elaborates on the grammaticalisation path and concludes that “[p]assives may be further grammaticized into ergative constructions, thus turning into active constructions”. He claims that the “most important aspect of this change is the upgrading of the agent phrase to a central constituent of the clause” (Haspelmath 1990: 56), ending with its obligatoriness. Although, as outlined in Casaretto et al. (2020), we do not assume that the Tima ergative construction has its origin in the passive notion of causatives, the many possibilities for altering the verb’s valency with the help of the various derivation markers attested in Tima are regarded as having facilitated the emergence of the ergative construction.

72As we have seen, these valency changing processes allow for different constructions in the realm of transitivity. Thus, with the verb’s valency being reduced, the subject of the noncausal intransitive clause may occur in the role of an agent or an undergoer. Also, with derived or underived causatives, the object may either be agentive or patientive. The role of the subject in the noncausal form does not have to correlate with either the role of the subject or the object of the causal form.

  • 30 See McGregor (2017: 5, and references therein): “[…] instrumental case markers are among the most c (...)

73An ergative construction is permitted with any transitive clause. With regard to the marking of the ergative case, the Tima make use here of material which already existed. That is, the instrumental marker N=, a homorganic nasal, as exemplified in (41), serves at the same time as the ergative case marker, e.g. in (35).30 The common denominator is that an instrument and an agent may share the feature of causation.

74In the causative-marked passive construction, as exemplified above, the instrument as the means or cause of the action may but does not have to be expressed in an oblique phrase. Casaretto et al. (2020) outline in detail that one of the more unlikely hypotheses of the emergence of ergative constructions in Tima is that such an instrument may have gained subject properties, such as being indexed on the verb. Synchronically, there is no indication that it happened that way. Consider for example the passive sentence in (47), with an oblique phrase referring to the means/instrument by which the patient-subject was summoned.

(47) Cέntànɪ̀ɪ̀kɘ́dà ŋ̀kábʊ́ŋ.
cέn‑tànà‑ɪ̀k‑ɘ́=dà ŋ̀=k‑àbʊ́ŋ
1sg:ipfv‑call‑causep=1sg ins=sg‑horn
‘I was called/summoned by the horn (of the motorbike).’

75Although word order, as repeatedly demonstrated in this contribution, is relatively flexible in Tima, the instrument kàbʊ́ŋ ‘horn’ cannot be positioned sentence-initially. Thus, *kabʊŋ cɛntanɪɪkɘda ‘the horn called me’ would be ungrammatical. Indeed, kàbʊ́ŋ ‘the horn’ is not appropriate as an agent-subject at all, so the following (equivalent to (1)) is also not possible.

(48) *Kàbʊ́ŋ ánt̪ànà álɪ̀.
k‑àbʊ́ŋ án‑t̪ànà álɪ̀
sg‑horn 3p:prf‑call Ali
‘The horn has called/summoned Ali.’ (20180130_1)

76What would be possible would be an AVOObl construction like ‘Hamid called me/Ali by way of using a horn’.

77If causative marking occurs with a passive notion, ergative constructions are not an option. In other words, transitive constructions, whether with inherently transitive verbs or with verbs made transitive with the help of the causative marker, are the prerequisite for OVAerg word order, not causative marking per se. Note also that almost all causative-marked verbs are either valency increasing (canonical causative function) or valency decreasing (anticausative/passive function). Only exceptionally and in a frozen form can causative marking appear on both a causal verb and a noncausal verb. Whether it is possible to have the ergative construction then remains to be clarified.

78Although it might be understood from the data presented, I nevertheless want to explicitly point out that a passive and resultative reading of ‑Vk-marked verbs is only possible with certain inherently transitive verbs, marked for the causative and combined with agent demotion and patient promotion. An erstwhile intransitive verb occurring in an SV construction (with S being an undergoer) cannot be marked for the causative and thereby express a passive notion, as is demonstrated with tʌʌn ‘boil’. The causative here strictly implicates a further core argument, i.e. the causer, as exemplified with the OcauseeVAerg construction in (50). For convenience, example (4) with the non-derived intransitive verb is repeated here as (49).

(49) Íídì àn‑tʌ́ʌ̀n.
íídì àn‑tʌ́ʌ̀n
pl:water 3p:prf‑boil
‘The water is boiling.’ (05.03.10_08_11)
(50) Íídì cèntʌ́ʌ́nɨ̀knʌ́.
íídì cèn‑tʌ́ʌ́n‑ɨ̀k=nʌ́
pl:water 1sg:prf‑boil‑caus=erg1sg
‘I boiled the water.’ (Skype20180418)

79A passive reading of the causative-marked verb would be ungrammatical, as demonstrated with (51).

(51) *Íídì àn‑tʌ́ʌ́n‑ɨ̀k.
íídì àn‑tʌ́ʌ́n‑ɨ̀k
pl:water 3p:prf‑boil‑caus
‘The water has been brought to the boil.’

80To sum up: the morpheme ‑Vk in Tima has two functions, that of a causative marker and that of an anticausative/passive marker. With the former function, the verb is bivalent and causal and able to occur in ergative constructions, while with the latter function it is monovalent and hence noncausal and can only add the cause/means in an oblique phrase. There is generally a strict division between valency increasing and valency decreasing causative marking (unlike for certain Manchu verbs, as described by Nedjalkov 1993).

4.4 Double function as facilitating the emergence of the ergative construction?

81Similar to the verb dukuk ‘drip’, discussed in Section 4.2, there are other verbs with a derivation marker that cannot unambiguously be assigned to one function. However, the context usually disambiguates the attested functions.

82One such verb is the translational equivalent for ‘crack’, which has two causal forms, tɘ̀ráyɪ̀ ‘crack it’ and tɘ̀rárààk ‘crack them’, and two noncausal counterparts, tɘrayɪl and tɘraraak, both attested with an anticausative notion indicating a spontaneous event. That is, the pluractional causal verb is identical to the pluractional noncausal one, with the form of the derivation marker being reminiscent of the antipassive marker ‑Ak.

83The verb is first illustrated in transitive contexts: (52) contains the verb marked for high transitivity, tɘ̀ráyɪ̀, while in (53), we find the pluractional stem tɘ̀rárààk. In both sentences, mentioning the object is obligatory.

(52) Tɘ̀ráyɪ́ kíméŋ.
tɘ̀rá‑y‑ɪ́ kí‑méŋ
crack‑epht sg‑cucumis.species
‘Crack the cucumis fruit (to make two halves).’
(53) Wɔ́rt̪ɘ́máádɘ́h ántɘ̀ràràák/ántɘ̀ràrààkàt̪áŋ cɪ́bɪ́.
wɔ́rt̪ɘ́máádɘ́h án‑tɘ̀ràràák / án‑tɘ̀ràrààk=à=t̪áŋ 3p:prf‑crack:plur 3p:prf‑crack:plur=source=loc3p
‘The man split the tree into pieces (actions not completed/completed).’

84The detransitivised forms with their anticausative notions are exemplified in (54) and (55).

(54) Kwɛ̀έŋ àntɘ̀ràyɪ́l/ántɘ̀ràyɪ̀làt̪áŋ.
k‑wɛ̀έŋ àn‑tɘ̀rà‑y‑ɪ́l / án‑tɘ̀rà‑y‑ɪ̀l=à=t̪áŋ
sg‑bowl 3p:prf‑crack‑ep‑mv 3p:prf‑crack‑ep‑mv=source=loc3p
‘The bowl has cracked (once, by itself) (not completely/completely).’
(55) Cɪ̀bɪ́ ántɘ̀ràràák/ ántɘ̀ràrààkàt̪áŋ.
cɪ̀‑bɪ́ án‑tɘ̀ràràák / án‑tɘ̀ràrààk=à=t̪áŋ
sg‑tree 3p:prf‑crack:plur.caus 3p:prf‑crack:plur.caus=source=loc3p
‘The tree has cracked (often, by itself) (not completely/completely).’

85So far, this verb behaves, despite its different morphology, like those verbs presented in Section 3.3 which display a difference with regard to uniactionality vs. pluractionality, e.g. juɽum ‘sink’, whose detransitivised forms were exemplified in (8) and (9).

86A crucial difference from the verbs in Section 3.3 lies in the way a cause can be bound into the construction. For example, with juɽumul/juɽumuk ‘sink:mv/caus’, the verb must be further derived for instrumental marking when an additional participant is mentioned, as exemplified with (56) for the single event and (57) for the plural event. This is not the case with tɘraraak ‘crack:plur(caus)’, as a comparison between (58) and (59) (both underived for instrumental) reveals. Rather, the verb behaves like causative-marked verbs with passive usage, cf. (47) ‘I was called/summoned by the horn’, in that the causer may be added without further derivational marking on the verb.

(56) […] íjùɽùmùlààt̪áŋ ɲíídì.
í‑jùɽùm‑ùl‑àà=à=t̪áŋ ɲ=íídì
p‑sink‑mvins=source=loc3p ins=pl:water
‘[When the child jumped into the water] it sank in (lit.: with) the water.’
(57) […] íjùɽùmùkààt̪áŋ ɲíídì.
í‑jùɽùm‑ùk‑àà=à=t̪áŋ ɲ=íídì
p‑sink‑causins=source=loc3p ins=pl:water
‘[When the children jumped into the water] they sank in (lit.: with) the water.’
(58) Yááh kɪ́ŋέ ántɘ̀ràrààkáyɛ̀έn.
y‑ááh kɪ́‑ŋέ án‑tɘ̀ràrààk=á=yɛ̀έn
pl‑head sg‑mouth 3p:prf‑crack:plur.caus=source=loc1sg
‘My lips are cracked.’
(59) Yááh kɪ́ŋέ ántɘ̀ràrààkáyɛ̀έn ɲ̀cíìn.
y‑ááh kɪ́‑ŋέ án‑tɘ̀ràrààk=á=yɛ̀έn
pl‑head sg‑mouth 3p:prf‑crack:plur(caus?)=source=loc1sg
‘My lips are cracked from the cold.’ (07.04.09, 2_05‑15)

87What is more, cíín ‘cold’ can be the attentional centre and thus occupy the sentence-initial position, as demonstrated in (60), again without the verb being further derived.

(60) Cíín ántɘ̀ràrààkáyɛ̀έn yááh kɪ́ŋέ.
cí‑ín án‑tɘ̀ràrààk=á=yɛ̀έn y‑ááh kɪ́‑ŋέ
sg‑cold 3p:prf‑crack:plur=source=loc1sg pl‑head sg‑mouth
‘The cold cracked my lips.’

In (60), the instrument cíìn ‘cold’ is undoubtedly a core argument due to its non-markedness for case and to the verb not being derived for instrumental. The question arises whether, concerning ɲ̀cíìn, we are dealing with a core or an oblique participant in (59). It may still be considered an oblique by analogy with (47) ‘I was called/summoned by the horn’, with mention of the cause being optional. But it may be considered a core argument when drawing parallels with the cluster of sentences around ‘X split the tree’, i.e. (61)‑(63). The causer wɔ̀rt̪ɘ́máádɘ̀h ‘man’ in (61) is taken as a core argument by analogy with (62), where mentioning the agent wɔ̀rt̪ɘ́máádɘ̀h ‘man’ is obligatory and a transitive reading of the verb is mandatory, and with (63), proving that indexing on the verb is with the subject agent. Furthermore, an instrumental reading of N= for a human referent is excluded (the notion of accompaniment is transmitted by the more complex proclitic =na/=nʌ). Note that the verb in (61) and (63) is marked for pluractionality, while it receives high transitive marking in (62).

(61) Cɪ̀bɪ́ ántɘ̀ràrààkàt̪áŋ ŋ̀wɔ́rt̪ɘ́máádɘ̀h.
cɪ̀‑bɪ́ án‑tɘ̀ràrààk=à=t̪áŋ ŋ̀=wɔ́rt̪ɘ́máádɘ̀h
sg‑tree 3p:prf‑crack:plur=source=loc3p
‘The man split the tree into pieces.’
(62) Cɪ̀bɪ́ ántɘ̀ràyààt̪áŋ ŋ̀wɔ́rt̪ɘ́máádɘ̀h.
cɪ̀‑bɪ́ án‑tɘ̀rà‑y‑ɪ̀=à=t̪áŋ ŋ̀=wɔ́rt̪ɘ́máádɘ̀h
sg‑tree 3p:prf‑crack:htsource=loc3p
‘The man split the tree into two.’
(63) Cɪ̀bɪ́ cέntɘ̀ràrààkàt̪àŋná.
cɪ̀‑bɪ́ cέn‑tɘ̀ràrààk=à=t̪àŋ=ná
sg‑tree 1sg:ipfv‑crack:plur=source=loc3p=erg1sg
‘I split the tree into pieces.’

88It might be cases such as these, in patient preserving detransitivisation contexts with the derived verb forms not being differentiated with regard to their valency, which have contributed to the emergence of the ergative constructions. But, of course, one swallow does not make a summer, so we will have to look for verbs with similar patterns and other ambiguous verb structures.

4.5 Comparing Tima causative-passives with Bantu causatives

  • 31 Meeussen (1967: 92) lists, inter alia, the following derivational suffixes: ‑i‑ (and ‑i̧c‑?) causat (...)

89We close this excursion with some thoughts that have been addressed by Dimmendaal in his recent publications (e.g. 2011: 297; 2018: 396f.), pointing out that the situation in Tima as depicted above, is reminiscent of the situation described for Bantu languages. Dimmendaal (2018: 397) hypothesises that “[t]he causative marker ‑ɪk in Tima is cognate, presumably, with the Proto-Bantu impositive marker *‑ɪk‑“, and one might add with “neuter” *‑ɪk‑ (as labelled by Meeussen 1967).31 For the latter, the NP preceding the “neuter/passive” is always the subject (resulting from object raising). Meeussen explains:

“Neuter ‑ik‑ has special limitations. It is found as a kind of potential with ‑bón‑ « see » and ‑úngu̧‑ « hear », and with a small class of radicals which is at the same time a semantic class (« verba destruendi »): ‑át‑ « split », ‑bú̧n‑ « break », ‑bu̧nj‑ « tear », and doubtless some others.” (Meeussen 1967: 92)

  • 32 See also the two papers of Dom et al. (this issue) and Yoneda (this issue) that deal with the Proto (...)

90Though a potential reading is not attested in Tima, the undergoer may become the subject preceding the verb. According to Dimmendaal (2018: 397), the “Proto-Bantu causative marker *‑ici̜‑ (Meeussen 1967) would then be the result of increment of the former causative marker (*‑ik‑i̜ > *‑ici̜), a hypothesis already forwarded by Meinhof apparently, but rejected by a number of modern Bantuists (Larry Hyman, personal communication)”. The reconstructed Proto-Bantu causative suffix *‑i would then “most likely” be “cognate with the high transitive marker i/ɪ in Tima” (Dimmendaal 2014: 121).32

91In line with Haspelmath’s explanation that one of the other functions of the passive may be that of resultative and hence of referring to a state (see Haspelmath 1990: 33), the verbs (i.e. verba destruendi) Meeussen (1967) listed, for example, show the utmost affectedness in the neuter form. The Tima ‑Vk morpheme, provided with an anticausative/passive function, also reflects the result of the action rather than the action itself and has a stative meaning, or, as one of the speakers says, “the action is finished”.

5. Concluding remarks

92The main focus in this contribution is on the noncausal/causal alternation of Tima verbs. Tentative categorisations of a number of verbs allow them to be assigned to three groups (see also Table 14).

  1. These are non-derived noncausal verbs which can be derived by the causative marker to yield a transitive verb. A correlation between the role of the intransitive subject and the transitive object as agent or undergoer is given.
  2. In the second group, both the noncausal and the causal verb are marked, the former by the antipassive (possibly with a reflexive/anticausative notion), the latter by the causative. An underived verb form does not exist. With the verb having a reflexive/anticausative notion, the object of the causal construction shares its role with the subject of the noncausal construction.
  3. The third group of verbs consists of transitive causal verbs which are detransitivised by middle or causative marking. The object undergoer of the causal construction becomes the subject undergoer of the noncausal one. In more detail, this group can be divided into two. The first subgroup comprises those verbs which may be marked by middle or causative, depending on event number. Here, these noncausal forms generally have an anticausative function. The verbs of the second subgroup can only be suffixed by the causative marker. Here, the noncausal form has a passive function. In both subgroups, the result of the action is focused upon.

93The ultimate aim of this research on noncausal-causal alternation will be to see whether the majority of Tima verbs can be neatly classified. This aim is far from being reached, mainly due to a lack of data. However, a start has been made. That is, certain verb groups could be established on the basis of their alternation patterns combined with the type of derivation marking by considering the semantic roles of agent and patient. Future research with data from more verbs will show whether we are on the right track or whether we have to reconsider the whole matter from another angle. What I have shown, I hope, is that it hardly makes sense to look at just one derivation marker while ignoring others, due to the overlapping functions of the marking options considered.

94Transitive verbs — and this includes causative-derived transitives as discussed in Section 3.2 — can be detransitivised and preserve the patient by using either the antipassive (in its reflexive/anticausative function), or the middle or/and the causative marker. Antipassive marking with anticausative function is only attested for causative-derived causal verbs discussed in 3.2 (double derivation). Causative-marked causal verbs of the augmentation strategy (discussed in 3.1) are excluded from patient preserving detransitivisation. Other transitive verbs, i.e. high and low transitive as well as pluractional transitive verb, make use of middle and causative marking in order to achieve a resultative meaning, alternating between an anticausative and a passive notion (discussed in 3.3).

95As hinted at in the previous section, we find verbs which do not fit neatly into the established patterns. Without having all the relevant data for other verbs at one’s disposal, verbs like the previously discussed ‘crack’ or ‘drip’ remain notable exceptions. Nevertheless, the present discussion on the noncausal/causal relationship between two members of alternating verb pairs in Tima and the attempt to group the verbs discussed is a first step, and the research shows not only the language-internal variation but hopefully also adds to the cross-linguistic variability investigated in this volume.

Table 14 — Overview of noncausal/causal verb forms in Tima

Table 14 — Overview of noncausal/causal verb forms in Tima


96The extended research on Tima started in 2007 within a documentation project which was generously supported by the Volkswagen Foundation. Currently, research on this language is possible within the project ‘Split case marking and constituent order variation in East Africa’ under the umbrella of the CRC 1252 ‘Prominence in Language’ (Project-ID 281511265), funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation). I would like to express my sincere thanks to both foundations. I would also like to thank Birgit Hellwig and Gerrit Dimmendaal for their constant support, and not only with regard to linguistic aspects. Likewise, I thank all my colleagues from Khartoum University for their help over all these years. Furthermore, I am very grateful to the anonymous reviewers as well as Sebastian Dom for their critical and useful examination of this paper. Their criticism helped me to improve the paper considerably. I would also like to thank Mary Chambers for correcting the English at different stages of writing. And last but not least, my thanks go to Hamid Kafi Daldoum for his endless patience with me. Without him and the contribution of the whole Tima community, any research on Tima would have failed.


- / : / . / = morpheme boundaries acc. to Leipzig Glossing Rules
1/2/3 1st/2nd/3rd person
A agent
antip antipassive
appl applicative
atr advanced tongue root
caus causative
cond conditional
cop copula
dem demonstrative
dir direction
ep epenthetic element
erg ergative
excl exclusive
foc focus
fut future
ht high transitive
ideo ideophone
incl inclusive
ins instrumental
intr. intransitive
ipfv imperfective
loc locative
log logophoric
lt low transitive
mv middle voice
neg negation
O object
obl oblique
p person marker
pl plural
plur pluractional
prep preposition
prf perfect
pro pronoun
prox proximal
redupl reduplicated
refl reflexive
sel selective
S subject
sg singular
source source marker
sp species
spont spontaneous event
stat stative
tam tense/aspect/mood
tr. transitive
V 1. verb, 2. harmonising vowel
vent ventive
Haut de page


Alamin, Suzan. 2012. The nominal and verbal morphology of Tima, a Niger-Congo language spoken in the Nuba Mountains. Cologne: Rüdiger Köppe.

Bashir, Abeer. 2010. Phonetic and phonological study of the Tima language. Khartoum: University of Khartoum (PhD dissertation).

Bashir, Abeer. 2013. Conditions of feature specifications in Tima. In Thilo C. Schadeberg & Roger M. Blench (eds.), Nuba Mountain language studies, 269‑280. Cologne: Rüdiger Köppe.

Blench, Roger M. 2013. Splitting up Kordofanian. In Thilo C. Schadeberg & Roger M. Blench (eds.), Nuba Mountain language studies, 571‑586. Cologne: Rüdiger Köppe.

Casaretto, Antje, Gerrit J. Dimmendaal, Birgit Hellwig, Uta Reinöhl & Gertrud Schneider-Blum. 2020. Roots of ergativity in Africa (and beyond). Studies in African Linguistics 49(1). 111‑140.

Copley, Bridget & Phillip Wolff. 2014. Theories of causation should inform linguistic theory and vice versa. In Bridget Copley & Fabienne Martin (eds.), Causation in grammatical structures, 11‑57. Oxford: Oxford University Press.

Dimmendaal, Gerrit J. 2009. Tima. In Gerrit J. Dimmendaal (ed.), Coding participant marking: Construction types in twelve African languages, 331‑353. Amsterdam: John Benjamins.

Dimmendaal, Gerrit J. 2010. Ditransitive constructions in Tima. In Andrej Malchukov, Martin Haspelmath & Bernard Comrie (eds.), Studies in ditransitive constructions: A comparative handbook, 204‑220. Berlin: De Gruyter Mouton.

Dimmendaal, Gerrit J. 2011. Historical linguistics and the comparative study of African languages. Amsterdam: John Benjamins.

Dimmendaal, Gerrit J. 2014. Where have all the noun classes gone in Tima? In Carole De Féral, Maarten Kossmann & Mauro Tosco (eds.), In and out of Africa. Languages in question. In honour of Robert Nicolaï, 103‑125. Leuven: Peeters.

Dimmendaal, Gerrit J. 2018. Reconstructing Katloid and deconstructing Kordofanian. In Gertrud Schneider-Blum, Birgit Hellwig & Gerrit J. Dimmendaal (eds.), Nuba Mountain language studies: New insights, 383‑415. Cologne: Rüdiger Köppe.

Dimmendaal, Gerrit J. & Gertrud Schneider-Blum. 2018. Transitivity in Tima. In Rose-Juliet Anyanwu (ed.), Transitivity in African languages (Frankfurter Afrikanistische Blätter 26), 125‑146. Cologne: Rüdiger Köppe.

Dimmendaal, Gerrit J., Nataliya Veit & Gertrud Schneider-Blum. In preparation. Locatives as core arguments in Tima.

Dom, Sebastian, Ponsiano Sawaka Kanijo, Leora Bar-el & Malin Petzell. 2021. Questionnaire on the noncausal-causal alternation.

Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67(3). 547‑619.

Greenberg, Joseph H. 1963. The languages of Africa. Bloomington, IN: Indiana University (accessed 30 December 2022).

Güldemann, Tom. 2018. Historical linguistics and genealogical language classification in Africa. In Tom Güldemann (ed.), The languages and linguistics of Africa, 58‑444. Berlin: De Gruyter Mouton.

Hammarström, Harald. 2013. Noun class parallels in Kordofanian and Niger-Congo: Evidence of genealogical inheritance? In Thilo C. Schadeberg & Roger M. Blench (eds.), Nuba Mountain language studies, 549‑569. Cologne: Rüdiger Köppe.

Haspelmath, Martin. 1990. The grammaticization of passive morphology. Studies in Language 14(1). 25‑72.

Haspelmath, Martin. 1993. More on the typology of inchoative/causative verb alternations. In Bernard Comrie & Maria Polinsky (eds.), Causatives and transitivity, 87‑120. Amsterdam: John Benjamins.

Haspelmath, Martin, Andreea S. Calude, Michael Spagnol, Heiko Narrog & Elif Bamyacı. 2014. Coding causal-noncausal verb alternations: A form-frequency correspondence explanation. Journal of Linguistics 50(3). 587‑625.

Keenan, Edward L. 1985. Passive in the world’s languages. In Timothy Shopen (ed.), Language typology and syntactic description, 243‑281. Cambridge: Cambridge University Press.

Kormušin, Igor V. 1976. О пассивном значении каузативных глаголов [On the passive meaning of causative verbs]. Turcologica (1976). 89‑93.

Kulikov, Leonid I. 2001. Causatives. In Martin Haspelmath, Ekkehard König, Wulf Oesterreicher & Wolfgang Raible (eds.), Language typology and language universals, vol. 2(2), 886‑898. Berlin: Walter De Gruyter.

Kulikov, Leonid I. 2011. Voice typology. In Jae Jung Song (ed.), The Oxford handbook of linguistic typology, 368‑398. Oxford: Oxford University Press.

Kulikov, Leonid I. 2013. Middle and reflexive. In Silvia Luraghi & Claudia Parodi (eds.), The Bloomsbury companion to syntax, 261‑280. London: Bloomsbury.

McGregor, William B. 2017. Grammaticalization of ergative case marking. In Jessica Coon, Diane Massam & Lisa Demena Travis (eds.), The Oxford handbook of ergativity. Oxford: Oxford University Press.

Meeussen, A.E. 1967. Bantu grammatical reconstructions. Africana Linguistica 3. 79‑121.

Nedjalkov, Igor V. 1993. Causative-passive polysemy of the Manchu-Tungusic ‑bu/‑v(u). Linguistica Antverpiensa 27. 193‑202.

Nichols, Johanna, David A. Peterson & Jonathan Barnes. 2004. Transitivizing and detransitivizing languages. Linguistic Typology 8(2). 149‑211.

Schadeberg, Thilo C. 1989. Kordofanian. In John Bendor-Samuel & Rhonda L. Hartell (eds.), The Niger-Congo languages: A classification and description of Africa’s largest language family, 67‑80. Lanham, MD: University Press of America.

Schadeberg, Thilo C. & Roger M. Blench. 2013. The Nuba Mountain languages: An introduction. In Thilo C. Schadeberg & Roger M. Blench (eds.), Nuba Mountain language studies, 1‑24. Cologne: Rüdiger Köppe.

Schneider-Blum, Gertrud. 2013a. Personal pronouns in Tima. In Thilo C. Schadeberg & Roger M. Blench (eds.), Nuba Mountain language studies, 281‑296. Cologne: Rüdiger Köppe.

Schneider-Blum, Gertrud. 2013b. A Tima-English dictionary. Cologne: Rüdiger Köppe.

Schneider-Blum, Gertrud. 2017. Once or more often? On pluractionality marking in Tima. STUF - Language Typology and Universals 70(1). 163‑194.

Schneider-Blum, Gertrud. 2018. Constituent focus and selective marking in Tima. In Gertrud Schneider-Blum, Birgit Hellwig & Gerrit J. Dimmendaal (eds.), Nuba Mountain language studies: New insights, 255‑279. Cologne: Rüdiger Köppe.

Schneider-Blum, Gertrud. Forthcoming. The impact of attentional centering on ergative marking in Tima. STUF - Language Typology and Universals.

Schneider-Blum, Gertrud & Birgit Hellwig. 2018. Reference tracking in Tima and its interplay with ergative marking. Studies in Language 42(4). 970‑993.

Skopeteas, Stavros, Ines Fiedler, Sam Hellmuth, Anne Schwarz, Gisbert Fanselow, Caroline Féry & Manfred Krifka. 2006. Questionnaire on information structure (QUIS): Reference manual (Interdisciplinary Studies on Information Structure, Working Papers of the SFB 632 4). Potsdam: Universitätsverlag Potsdam.

Veit, Nataliya. In preparation. Semantic verb classes in Tima (Niger-Congo).

Williamson, Kay. 1989. Niger-Congo overview. In John Bendor-Samuel & Rhonda L. Hartell (eds.), The Niger-Congo languages: A classification and description of Africa’s largest language family, 3‑45. Lanham, MD: University Press of America.

Haut de page

Document annexe

Haut de page


1 Examples for ambitransitive verbs are ɘ́màh ‘eat’, kɨ́mʌ́n ‘get satisfied, have eaten enough’ and ŋʌ́ʌ̀l ‘smell’ (see Dimmendaal & Schneider-Blum 2018; Veit in preparation; the latter study also takes causative-derived ambitransitive verbs into account). An example for the noncausal/causal alternation by suppletion is, e.g. àmbúlúkát̪áŋ ‘they died’/ántòmààt̪áŋ ‘they killed it’.

2 Tima is not the only language in the wider area (covering Sudan, South Sudan, Ethiopia and beyond) which shows split argument marking. Casaretto et al. (2020: 116) state: “Tima […] shares some salient structural properties of split case marking with other languages of this region, albeit not the actual forms.” Presumably, “language-internal changes account for the development of split ergativity in Tima. It is nevertheless likely that ancient areal contact has played an additional, facilitating, role.” Currently, different types of split case marking systems are investigated in the project Split case marking and constituent order variation in East Africa within SFB 1252 at the University of Cologne.

3 For most of the consonants, the relevant IPA symbol is used, with the following exceptions: retroflex /ʈ/ is represented as <t>, /j/ as <y>, and /ɟ/ as <j>. This reflects the practical orthography, as developed in communication with the Tima Language Committee.

4 Tone in Tima has not yet been systematically investigated; therefore, the examples are phonetically marked. When words are mentioned in isolation, the tone of the citation form is marked. If a root/stem or other morpheme which is not also an independent word is mentioned in isolation, it is not marked for tone. The citation form for verbs is the imperative singular. If the semantics of a verb do not allow for a controlled action, an imperative is infelicitous. In such cases, the verbal root/stem is not marked for tone.

5 Dimmendaal (2009) was the first to sketch the ATR system of Tima; see also Bashir (2010; 2013) for a more detailed account.

6 Tima orthographic symbols as opposed to IPA symbols:

Grapheme + ATR IPA + ATR IPA - ATR Grapheme - ATR
i ɪ
e ɛ
ʌ a
o ɔ
u ʊ
ɨ ɨ̘ ɨ̙ ɘ

7 For the pronominal system in Tima, see Schneider-Blum (2013a).

8 The Tima data used for this paper was collected between 2007 and 2022. The source of a recorded example sentence is noted in brackets following the free translation and appears in different formats (reflecting the fact that data collection took place throughout various projects adhering to various conventions).

9 In the structure of the inflected verb, the first person marker is not obligatory.

10 V of the middle and the causative suffix is the abstract representation of a vowel which harmonises regarding its ATR value and frontness with the preceding vowel. It is, with a few morphophonologically conditioned exceptions, a high vowel. Abstract A only varies with regard to the ATR value, realised either as [a̘] <ʌ>] or [a̙] <a>.

11 “[…] high transitive marking in Tima, expressing a uniactional event, stands in paradigmatic opposition to pluractional marking, which generally indicates repeated action; low transitive marking (which is relatively rare, […]) tends to express atelic events, i.e. events involving duration” (Dimmendaal & Schneider-Blum 2018: 133f.). For more details regarding transitivity, see Dimmendaal & Schneider-Blum (2018). For detailed information on pluractional marking in Tima, see Schneider-Blum (2017).

12 The middle voice morpheme has been labelled “reversive” or “separative” in earlier publications on Tima, e.g. in Dimmendaal (2010), Alamin (2012) and Schneider-Blum (2013b).

13 The combined source marker and locative marker, at̪aŋ, was formerly analysed and labelled as a completive, since it often indicates the completion of an action. Thus, this combination can functionally be compared to culmination markers, as described by Copley & Wolff (2014), in that it signals entailed accomplishment – “culmination”, in their words – of the action.

14 The k of certain roots, e.g. (k)at̪am ‘leave’, (k)aar ‘grow’, (k)amʊ ‘wash’, and (k)ʌbu ‘dig, plant’, only shows up in certain environments and not in others. Generally, k is present if the verbal root is inflected for tense/aspect, and it is absent if the root occurs in the imperative. The alternation does not interfere with the noncausal/causal alternation.

15 This example has been elicited with material compiled by Skopeteas et al. (2006).

16 In the case of wudʌ ‘burn’, the causal alternation also exhibits a differentiation between a singular and a plural action, expressed by two different causative forms of the verb. For other causal verbs which have two forms, see Section 3.3.

17 Note that transitive verbs that are not causative-marked, i.e. high or low transitive and pluractional verbs, may also receive antipassive marking. In that case, the antipassive derivation is only attested with the agent preserving strategy.

18 It is not possible to form the imperative of aarak ‘grow (intr)’, since, as the speaker expressed himself, one does not have control over the action.

19 A grammatical constraint prevents the form marked for high transitivity to additionally take the middle (or causative) suffix, i.e. the middle (or causative) suffix replaces the high transitive marker.

20 See Kulikov (2011: 374; 2013: 267), who differentiates between a canonical passive (where the direct object is foregrounded, with the agent being backgrounded and marked as oblique) and an agentless passive where the agent is omitted. “Probably all languages that have a canonical passive can also freely omit the passive Agent […]. […] Next to the languages with canonical (“full”) passive, such as English, Latin, or Sanskrit, there are languages that cannot express the Agent in passive sentences.” (Kulikov 2013: 267). Kulikov (2013: 272) further states that “where the markers of the passive and anticausative overlap, passives without an overtly expressed Agent can be distinguished from anticausatives only by semantic criteria”. This is exactly what we find in Tima: a marker which comprises passive and anticausative functions.

21 When mɨ̀rɨ́nɨ̀l carries the meaning of ‘separate yourself (from others)’, the imperative is acceptable; it is not so in the context of ‘divide itself’.

22 “If verbs are marked by reduplication (partly or entirely), the meaning carries the notion of doing something by bits (i.e. many distributive events) […].” (Schneider-Blum 2017: 177) Thus, different options serving to indicate pluractionality allow for a further differentiation within pluractional marking.

23 Consider, in this context, Copley & Wolff (2014) on dispositionality requirements.

24 Middle marking is very productive, though, when it comes to adjectives (with a resultative notion) derived from verbs. Thus, for ‘hang’, the predicative adjectives àtùhúl ‘it is hanging’ and ìtùhúl ‘they are hanging’ are attested. Note that, here, causative marking is infelicitous for the singular as well as for the plural.

25 Again, it is because the feature control is missing that the imperative form cannot be formed, as the speaker suggested; instead we present the verbal root.

26 The only known exception of a middle-marked verb with passive reading is found with tòmó ‘kill’.

27 Consider Kulikov (2013: 272f.): “Distinguishing passives without an Agent from nonpassive intransitives (anticausatives) is one of the most complicated problems with which a linguist is confronted when undertaking a syntactic study of the verb. Alongside clear instances of passives, which raise no doubts by virtue of the inherent agentive semantics of the corresponding verb (cf. such predicates as build—is built), and doubtless anticausatives (cf. falls, grows), there is an area of uncertainty, that is, intransitive usages that allow for both passive and anticausative interpretations (cf. such meanings as “is born / arises”).” As the Tima speakers assert, all Tima verbs listed in Table 12 – with the possible exception of wʌr ‘lose’ – imply that an agent has performed the action.

28 Tima is not unique in having a causative/passive polysemy. Kulikov (2001: 894), for example, lists “Korean, some Altaic languages of Siberia (Tuvan, Yakut, Mongolian, Manchu and other Tungusic languages), some West African languages (Songhai, Dogon), [and] Bella Coola (Amerindian)” in which “causative markers can also function as passives”.

29 Consider Casaretto et al. (2020: 127) who, with regard to the occurrence of causative-marking in Tima, say that “a count of the ‑Vk suffix across 16 natural texts (763 intonation units) revealed that it most frequently occurs in its causative (47 times) and [anticausative] functions (8 times), and only once in its passive function”.

30 See McGregor (2017: 5, and references therein): “[…] instrumental case markers are among the most commonly suggested sources of ergative markers […].”

31 Meeussen (1967: 92) lists, inter alia, the following derivational suffixes: ‑i‑ (and ‑i̧c‑?) causative: ‑dim‑í̧‑ ‘make (someone) cultivate’; ‑ik1 impositive: ‑kú̧k‑ik‑ ‘put in a kneeling position’; ‑ik2 neuter: ‑bón‑ik‑ (bonek‑) ‘be in sight’.

32 See also the two papers of Dom et al. (this issue) and Yoneda (this issue) that deal with the Proto-Bantu causative suffix *‑i.

Haut de page

Table des illustrations

Titre Figure 1 — The Tima region in the Nuba Mountains (altitude: 600‑800 m)
Fichier image/png, 397k
Titre Figure 2 — Stimulus for (28)
Fichier image/png, 473k
Titre Figure 3 — Stimulus for (30)
Fichier image/png, 1,2M
Titre Figure 4 — Passive grammaticalisation paths based on Haspelmath (1990: 54), red outlines added
Fichier image/png, 53k
Titre Table 14 — Overview of noncausal/causal verb forms in Tima
Fichier image/png, 57k
Fichier image/jpeg, 46k
Fichier image/png, 756k
Haut de page

Pour citer cet article

Référence électronique

Gertrud Schneider-Blum, « On noncausal/causal alternations in Tima (Nuba Mountains, Sudan) »Linguistique et langues africaines [En ligne], 8(2) | 2022, mis en ligne le 31 décembre 2022, consulté le 29 mai 2024. URL : ; DOI :

Haut de page


Gertrud Schneider-Blum

University of Cologne

Haut de page

Droits d’auteur


Le texte seul est utilisable sous licence CC BY-SA 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search