Conditioned Animacy Marking in Nivkh1)
KEYWORDS: Nivkh, animacy, causation, verb formation
It is well known that the language Nivkh has a special case marking for the syntactic causee, which is traditionally named as "dative-accusative case (дательно-винительный пажедь)" in Russian linguistics. In this paper, I will point out that the marking is a particular linguistic case of animacy markings, which appear in a variety of formal devices in many languages of the world. In this language, however, the animacy marking is highly syntactically restricted, in that only the syntactic causee is marked as animate with the suffix -ax if it is human or personified under special syntactic conditions in which it occurs. The principle this language makes use of for this purpose is simple and well motivated, because it is restricted to the case to represent the intentionality of the causee in a straight-forward way. But in this context, there are some linguistic issues worth thinking about, among others, the question, why it is necessary to use such a special marking under special syntactic conditions. The reason goes back to the fact that the suffix -gu is inherently a grammatical device for transitivization, and that implies that the causation is in this language somewhat a part of of transitivization, or at least the latter makes use of the same verb formation device, and that, therefore, these two syntactic processes are distinguished only by use of different case markings.
1. Some Preliminaries
1.1. Case Marking
To begin with, let us get a look at the case marking system of the North-Western dialect of this language2):
(1) case markings in North-West dialect in a overlook (Gruzdeva 1998)
a. absolutive case: φ (zero-marking)
b. dative/accusative case: -ax, -X
c. comparative case: -yk
c. locative case: -unine/-uin/iin/iun/-n
d. locative/absolutive: -uÄe/-Äe. -ux/-x
e. dative/additive: -toX/--roX---doX/-rX /-tX
f. limitative: -toÄo/--roÄo/-doÄo/; tH-yky/-rHyky; -tHXa/-rHXa/-rGa
g. instrumental: -kir/-Äir/-gir/-xir
The suffixes above are attached to a noun stem. In front of them can stand merely a plural suffix -gu/-ku/.... The allomorphs in each line above appear according to the phonological circumstances of the noun stems. Some forms come from simplification, and others result from consonant lenition in contact (about lenition cf. Shiraishi 2007).
In traditional grammars, cases are in general divided in two classes, the right (casus rectus) and the oblique ones (casus obliqus). In this sense, there is no doubt that the cases from (1c) to (1g) are to be classified as oblique. And only the case with no explicit marking, the absolutive case, belongs to the right case. We have to say that the right case of this language has no marking. But the case (1b), the dative/accusative case is peculiar among the cases in this language: it is neither right nor oblique. It appears only in a causative construction, so that it has to be called at best as a derived, or better, derivational case. Anyway, such a classification of cases into right and oblique says practically almost nothing relevant about the grammatical features of the case marking, especially it makes no use to explain the semantic and syntactic behavior of our "dative/accusative" case (1b) in question, namely the suffix -ax. If we regard it as derivational case, we have to show precisely in what syntactic structure it is used and how it is explained in semantic terms.
Now, the case marking is tightly related to the so-called valency features of the nouns they are attached to. Under valency we understand the syntactic potentials of a verb how many and what kind of nouns it requires to make a minimal sentence. The Nivkh verb vidJ (to go), for example, requires 2 nouns, to make a minimal sentence; the agent noun in absolutive (1a) and dative/additive (1e), or better to say, directive case, -toX. But the valence consists of two systems: the one determines the surface features, and the other reflects the so-called deep cases of the verb which dominates the nouns. But both are not necessarily in a simple relation of one-to-one correspondence. The Nivhk verb, e.g. xrodJ (< eÄrodJ = someone hangs something up to somewhere) requires three nouns all in absolutive case, despite that they refer to agent, object and place, respectively. Another good example is the verb kHimdJ (someone gives something to somebody). It requires three nouns, too, all in absolutive case, i.e. with no overt case marking. But it has semantic roles agent, object and patient (Krejnovich 1958). Therefore, the case markings cannot be necessarily predicted from the apparently corresponding deep cases. It is, in essence, particular-linguistically determined.
1.2. The so-called dative-accusative case
The traditional list of Nivkh case markings (1) contains, namely, a particular one to notice. It is the second one, the dative/accusative case, in Russian "дательно-винительный пажедь", which is commented by Panfilov 1965: "it marks only the noun designating an animate object which appears as object of a causative act". (p.131) And Grudzeva 1998 explains it with a more suitable remark "it marks the animate causee in a causative structure" with an example (p.19):
(2) N' -nanx φ n' ax pri-roX vi-gu-d.
my-elder-sister-NOM I-DAT/ACC forest- DAT/ADD go-CAUSE-Fin
"My elder sister let me go to the forest." (notation acc. to Grudzeva)
As far as I have seen, we have found no other remarks about this case marking in the literature. Crucial points about this case marking are, first, that it appears only in a causative construction. It marks the causee, i.e. the person who is commanded to pursue a required act. Second, the referent of the causee is a human being, or at most a personified animate. In other words, the suffix -ax designates that the noun is the agent of an embedded causative action pursued by a human being, or at least a personified animate agent. In short, it marks the intentional causee of a causative construction.
1.3. Verb derivation
The verb derivation in this language is pretty complicated. It seems to go far beyond the range of this paper even to get an overlook at main issues about them. So, we will take up only the five topics below (1.3.1 ‾ 1.3.5.) which are all directly related to the causative structure of this language.
1.3.1. Transitive-intransitive pairs
This language has a set of transitive-intransitive verb pairs with common stem features which, however, can have different stem top consonants. The opposition lies in that the stem top of the intransitive verb is a fortis consonant, while that of the corresponding transitive verb begins with a lenis of the same class, e.g. vi. tatadJ (to be important) : vt. ratadJ (to take important, to care). This makes an assumption possible that from a common source √Tata-dJ can be derived an intransitive verb tata-dJ with the top consonant fortis t- and a transitive rata-dJ with the corresponding lenis r-. This derivation type is pretty productive.
1.3.2. Valence Reduction with prefix i-/e-/j-
A set of transitive verbs, which are supposed to belong to the very old layer of the language, stand in opposition to the intransitive counterparts with the prefix i-/e-/j- (cf. Krejnovich 1958). The prefix denotes a generalized object which is conventionally fixed to something familiar in ordinary life. A transitive verb ardJ , e.g. is used in the context: agent gives prey to dogs. In this case, all of the arguments are marked with φ, like in the sentence: if ardJ ma qangu (he gave dogs dried sermon). But the corresponding intransitive verb jardJ with the prefixed j- reduces one argument. The inherent object changes here into the prefix. So, it lacks the inherent direct object designating prey, e.g. if j-ardJ qangu (to prey dogs). Note that the derived verb itself is yet transitive because the patient/beneficient object dogs remains in absolutive case, though the direct object prey disappeared. In case of transitive verbs with two arguments, e.g. nJivdJ (to eat something), the derived verb with the suffix inJdJ (to eat meal) is an intransitive, because it has only one absolutive argument, namely the subject., e.g. in the sentence like:
(3) t«tNan inJve! t«dJ als amra urdJra.
morning eat-Imp. this mushroom taste good-Fin-Aff mushroom eat-Imp
(Take breakfast!, This mushroom is good to eat. Eat mushroom.)
The prefix i-/e-/j- appears to intransitivize transitive verbs; it reduces one of the arguments they have inherently. It looks like a sort of incorporation of clitics representing the unspecified object of transitive verbs. Such an interpretation has to be discussed further. Anyway, it is sure that the prefixes have a syntactic function to reduce one valence from the inherent valence they have.
1.3.3. Transitivization with the derivative suffix -u
Nivkh has a morphological means to derive transitive verbs from intransitive by adding a derivative suffix -u to the intransitive stem, e.g. vt. l«rk-u-dJ (to float something ) from vi.l«rktJ (to swim). This derivative suffix is very productive and in many cases brings about with the fricativization (=lenition) of the initial consonant, if it is possible, i.e. viludJ (to make someone/something big) from pildJ (to be big) and Ä«rudJ ((to stop something) from kr«dJ (vi. to stop). With this transitivizing device the intransitive subject changes into a transitive object, where the both arguments stand in absolutive case, i.e. with φ.
1.3.4. Transitivization with the derivative suffix -gu/-ku
More productive is the suffix -gu /-ku.. With this suffix, too, an intransitive verb udJ (intr. to burn), e.g. changes into the transitive ugudJ as in the sentence: nJ i tHÄr ugudJ (I burned the tree) ＜tHÄr udJ ( the tree buns). The direct object here can be animate or inanimate like in the sentences below:
(4) a. oÄla-gu pan-gu-ny-te (let children grow up)
b. kapusta ban-gu-ny-te (let cabbage grow up)
child-Pl/cabbage grow-up-TR-Fut-Imp (TR:transitivization)
The suffix -gu/-ku is alo used to make verbs causative. The suffix is homomorphic in both usages, there is no morphological difference. This says that, in order to show the causative use unambiguously, a special marking is required to distinguish a causee from a simple direct object. In the sentence below, the argument of myr-(to go up), is the object of transitivized myr-gu (to let go up). Here, it is not marked explicitly whether it is a simple object or a voluntary agent.
(5) i-ranrH ittJ: nJ oGla pHuÄryr pal-dox myr-gu-ja (Panfilov 65, 48)
his-sister said: my child Refl-together mountain-Dir go-up-TR/CAUSE-Imp
(My sister said: bring my boy together up to the mountain.)
The object nJ oGla (my child) stands here in the absolutive case, so, it seems to be the object of both the verb pHuÄry (to bring together) and myr-gu (to let go up). But it cannot be interpreted as causee because it is not marked with -ax. See (12) and (12') bellow.
1.3.5. Causativization with the suffix -gu/-ku
The same suffix -gu/-ku is often attached to verbs with the object representing inherently human beings, or personified agents, like vaÄr-dJ (to be ashamed) in the sentence:
(6) «m«k pH-oGla-ax antX-ku-dox vaÄr-gu-dJ（Panfilov 1962,p.131）
[S1 mother[ S2 Relf-child-CAUSEE guest-Pl-Dat vi[be-ashamed]-CAUSE-Fin]]
(Mother makes the child ashamed to the guests.)
In the sentences like this, the agent in
the embedded phrase is marked with the -ax indicating the causee who is obliged to act by the compulsion of
the matrix agent. The embedded verbs can be transitive or intransitive; in case
that it is intransitive, the arguments are two at least like in the sentence
(7a), but in case of transitive verbs, at least three in the sentence (7b):
(7) a. «m«k pH-oGla-ax plicy-gu-dJ. (Mother let her child dance.)
[S mother [VP Refl-child-CAUSEE dance-CAUSE-Fin]]
b. «m«k pH-oGla-ax plaNrH oq xe-gu-dJ . (Mother let her child put on a long dress.)
[S mother [VP Refl-child-CAUSEE long dress put-on-CAUSE-Fin]]
2. Case Marking in Transitive and Causative Constructions
2.1. Case Making of Object Nouns
In Nivkh, object nouns are marked in absolutive case, namelyφ, regardless of its animacy. Compare the sentences in (8):
(8) a. Xevgun pH-ke-gu -φ se-gu-dJ-ra ＜Saveljeva/Taksami 1970 et.al)
[S Xevgun VP[Refl-net-Pl -Abs dry-TR-Fin-Affirm]]
(Xevgun dried his own net.
b. Xevgun pH«k«n -φ mu-gu-dJ-ra ＜Saveljeva/Taksami 1970 et.al)
[S Xevgun [VP Refl-brother-Abs die-TR-Fin-Affirm]]
(Xevgun let/made his brother die.)
c. nJi «Ärk« kHeq--umgu -φ pH-iÄ-gu-dJ-ra ( Panfilov1965,49-8)
[S I scarcely [VP fox-mother-Abs Refl-kill]- TR-Fin-Affim ]
(I scarcely let the fox-mother kill herself.)
In (8a), the embedded intransitive subject is a thing, which becomes dry by the act of the matrix subject Xevgun. The verb mu-dJ in (8b) is intransitive, it becomes transitive by means of the suffix -gu, in the meaning that the death has been brought about. But the matrix subject Xevgun is not necessarily responsible to his brother's death. His death is eventually nothing to do with Xevgun who only feels that he is responsible. Nivkh has another verb with the meaning to bring about death, namely iÄ-dJ/kHu-dJ/xu-dJ. This verb has an inherent meaning to take a prey by hunting. It represents an intentional act of the agent to kill something. The sentence (8c) says that the fox-mother scarcely fell in a trap, but the subject "I" succeeded in preventing her from suicide. The absolutive case marking of the fox-mother implies here that the possible suicide cannot be ascribed to herself, nor to the matrix subject.
From the sentences (8) we can see that the verbs with -gu make transitive sentences in any case, whose objects stand in absolutive case φ. It says that in this transitive construction, the object can be animate or inanimate with this case markingφ. The transitive object is in itself indifferent as to animacy.
2.2. Causee Marker
Nivkh has a special case marker -ax to indicate the causee who does the act forced to do by the matrix subject. Formally, the causee stands in the same syntactic position as the object noun, as illustrated in (9):
(9) a. causative sentence: [S CAUSER [VP CAUSEE-ax...V3) ]-gu-...]
b. transitive sentence: [S TR-subject [VP TR-object-φ...V3) ]-gu-...]]
The followings are the typical examples for causative sentences:
(10) a. nJi qan-ax mavr Nagi-gu-dJ-ra (Panfilov 65-48-4)
[S I [VP dog-ax goose hunt]-CAUSE-Fin-Affirm].
( I let the dog hunt the goose.)
b. nJi pHoGla-ax t«ftox eÄ-gu-dJ-ra (Panfilov 65-48-1)
[ S I [VP Refl-child-ax house-Dir return-]CAUSE-Fin-Affirm]
( I let my child return to the house.)
In both sentences the causee is marked by the marker -ax, though they are slightly different according to the animacy scale: the causee in (10a) is animate while that in (10b) human. But it is common in both cases in that both causees do the commanded work with their own will. The dog runs to the prey with his own will, so is the case, too, with the child who goes to the house obeying the command, but with his own will. Surely, their action is regarded not necessarily as voluntary in the normal sense, because it is a requested one. But it is carried out by the agent with their own will. Therefore, it is not yet precise enough to regard the causee only as animate/human. Essential is the intentionality of the causee to carry out the act commanded by the causer. The feature "animate/human" indicates only the extensional feature of the causee performing the action with his own will. Let us now call the feature instead as [+intention] in the sense "to do the action with his own will". We know the term "agent" implies the intentional actor inherently, but this meaning is not explicit somehow in ordinary grammatical usage. We pay attention by marking it with an intensional marking [+intention]. So, the causative sentence structure (9a) has to be supplemented by semantic information:
(11) [S N1 [VP N2-ax ... V]-gu-...]
where N1: Causer=the agent of Cause (=[V -gu])
N2: Causee=the agent [+intention] of V
V: the embedded verb V can be transitive or intransitive.
In the typical causative sentences in (10a), the transitive subject of the sentence is "I", human, and the causee is the transitive subject, a dog qan-ax [+intention], who was commanded to do the work to hunt Nag- (hunt) a goose mavr. Remark that in this case the embedded verb V is transitive.
In the second sentence (10b) the causer is the same as in (10a), i.e.human. The causee, who gets the command is the child, pHoGla, a human being, who goes home m«r- with [+intention].
Now, we find some sentences in which the transitive agent lacks the marker -ax, though the sentence structure seems to match the condition (11):
(12) . i-ranrH ittJ, nJoGla pH-iÄr«-r paldox m«r-gu-ja (Panfilov 1965-48-3)
[S his sister said [ VP 1 [VP 2 my child Refl-take-ing] mountain-Dir go-up]-TR/CAUSE-Imp]]
( My sister said, taking my child , go up to the mountain.)
The embedded verb phrase VP1 is complex. It contains an adverbial verb phrase VP2, which consists of a participle (in Russeian, deprihastie) V-stem+t/r. This participle phrase contains the object nJoGla (own child) in absolutive marking. Here we have two different way of grammatical interpretation: (a) the sentence (12) is transitive, because we can regard nJoGla (own child) as the object of VP2 as well as that of VP1. Or (b) the sentence (12) is interpreted as causative, in that the object remains in VP2 and the possible causee in V1 is omitted. In the latter interpretation we assume the following structure, so that the underlined nJoGla-ax is omitted because a reference identical NP occurs beforehand.
(12') i-ranrH ittJ, [VP1 [ VP2 nJoGla pH-iÄr«-r] nJoGla-ax paldox m«r-gu-ja ]]
In the second interpretation, we regard the sentence (12) as elliptical. A question arises here, whether it is possible to mark first occurrence of nJoGla with -ax tin order o show the causative meaning of the sentence. The answer may be negative, because then VP2 becomes ill-formed. The omission of the second occurrence of nJoGla-ax is surely a better option.
There is a more interesting example to support the elliptical interpretation. In the following sentence, the sentence meaning is focused to the intention to let the teacher explain the task.
(13) nJ i ucitJel-dox zadaca pH-«ks«m-gu -in« -t vi-dJ (Panfilov1965, 50-1)
[S I [VP1 teacher-Dir [VP2 task Refl-explain-CAUSE-want-Part] go-] VP1-Fin]
（I went to the teacher to let her explain me the task.)
The innermost verb phrase VP2 is a participle phrase with -t and it modifies the matrix verb vi-dJ (went). The embedded participle phrase is clearly causative, so that it should contain the causee, like in (13'):
(13') nJ i ucitJel-dox [ VP2ucitJel-ax zadaca pH-«ks«m-gu -in« -t ] vi-dJ .
But because of the precedence of the referent-identical noun ucitJel-dox, the causee noun is omitted, though this has a different case marking and belongs to the higher verb phrase VP1.
Now, compare the interpretation (12') and (13'). The causative marker -gu is attached in (12) to the matrix verb V1, so, the whole sentence is causativized. Therefore, the possible causee should appear here in the matrix verb phrase V1. On the other hand, in the sentence (13) the participle is causativized. The phrase "in order to make the teacher explain the task to me" contains the suffix -gu, so, the causee should occur in VP2 in (13'). In spite of this structural difference, however, the causee is omitted in both cases. So, we can assume that it makes no matter, where does occur the casusee in the syntactic constructions like above. It is anyway omitted, if a co-referential noun precedes it. This can imply that the causee marking is not so strong as the other case markings; in (12) it is weaker than the absolutive, and in (13) than an oblique one. But so far, it is not yet sufficient to establish a reasonable hierarchy of case markings in this language, the issue about hierarchy should be mentioned later.
3. Conditions for Causee Marking
3.1. Necessary Conditions
The case marking -ax appears, only when the suffix -gu is attached to the verb. But it is in general a marker to make intransitive verbs transitive. It has a function for transitivization inherently. In fact, we find no nouns with -ax-marking which would appear in the verb phrase without -gu. The suffix -ax has to co-occur always with -gu. Therefore, a causative construction in Nivkh has rigidly the form as (11) recited below. This formulation can be rewritten in a simpler way, but the remarks in (11) have to be supplemented. The simpler formulation is seen as (14):
(11) [S N1 [VP N2-ax ... V]-gu-...]
where N1: Causer=the agent of Cause (=[V -gu])
N2: Causee=the agent of V [+intention]
V: the inner verb V can be transitive or intransitive.
(14) [S CAUSER [VP CAUSEE[+intention] ... Vtr/intr]-CAUSE -...],
Now, let us ask about the semantic feature of the causee again. Question is, whether the traditional saying that it must be animate/human is sufficient in our case, too. Comparing the causative sentence (10a) with the transitive sentences (8b, c), we remark that an animate object is the grammatical object of a transitive as well as that of a causative predicate. This says that an animate/human object does not suffice to become a causee in a causative sentence. Crucial is that the derivative suffix -gu/ku co-ours with -ax to make a causative construction. The suffix -gu/ku only has the function to transitivize intransitive verbs in general. It makes, e.g. transitive sentences like in (8 a,b,c). In causative sentences the same suffix is applied. An interpretation is possible that is homomorphic suffix -gu/ku is used in transitive as well as in causative sentences. Or, it is possible to assume that the transitivizing suffix -gu/-ku is used in order to make causative sentences in which the causee has to be marked with -ax, at the same time. I think the latter interpretation may be correct. Anyway, in this language it is necessary for a causative sentence to be marked by the same suffix as in a transitive as to verb formation device. But it is not yet sufficient for a transitive sentence to become a causative construction as (14) above. The sufficient condition of (14) is that the agent of Vtr/intr is not merely animate/human, but the embedded agent does the requested act with his own will, namely an actor with intention. This situation is marked by -ax. In short, a Nivkh sentence has to be marked with the transitivizing suffix -gu (necessary condition) and at the same time the agent of the internal verb Vtr/intr is marked by the case marking -ax (sufficient condition).
Now, we rewrite the formulation (14) in a more general form:
(15) [VP 1 the agent of V-gu [VP2 [ [NP the agent [+intention] of V]-ax ] ...V-gu ...]
CAUSER CAUSEE CAUSE
Remark, a propos, that the suffix -gu is one of verb derivational suffixes in this language, parallel to the temporal -ny (future), the aspect -iv (durative), -Ä«t (perfective) and modal -in« (will) and so on. The relative order of the derivative suffixes is fixed: causative - aspect - tense - modal. These are attached to verb stems in an agglutinative fashion to make a long con-verb chain. But there are some more issues necessary to show precisely how they are arranged in the formation of verb complexes.
3.2. Restrictions for the Occurrence of -ax
We have seen above some cases in which causative nouns with -ax do not appear, although the conditions for causative construction are fulfilled. In the examples (12) and (13), we have seen that the causee with -ax do not occur if a co-referent noun precedes, where it is irrelevant, however, whether the referent-identical noun is a argument of the causativized verb or an element of the embedded participle phrase. In (12), the "omitted" noun with -ax is the subject of the verb m«rdJ (to go up)in VP1, and that in (13) is that of the transitive verb «ks«mdJ (to explain ) in VP2. In both cases, the possible causee is the agent of the causativized verbs Vtr/intr. They belong to syntactic valence of the verbs Vtr/intr with -gu. This means that the omitted nouns with -ax have to constitute the valency elemenents of the causativized verbs. Anyway, the omitted causee belongs to the terms which are easiest to remind of. It is not yet definitely clear, if this valency relation makes a condition for the ellipsis of the causee in general cases. But it is at least so far clear the syntactic relations the sentences (12) and (13) suggest. Provided that the valency relation makes a condition to omit the causee, we can formulate the conditions for ellipsis of the causee, as well as the well-formedness condtition for the omission as follows:
(16) a. The causee is omissible, if noun-ax is contained in the valency of V-gu
b. A causative construction is ill-formed, if noun-ax is not contained in the valency of V-gu
This assertion may be trivial, but it is worth noticing that the omissible causee nouns have to be the agent of the verbs with -gu, namely in a highest position in the valency hierarchy of the causativized verbs, ,even in case that the verbs are connected in a con-verbs-construction like in (12) and (13).
1) The outline of this paper was presented in the symposium "Cross-Linguistic Animacy" at the Spring Meeting of Japanese Linguistic Society (the 134th) at 16. June 2007 (at Reitaku University). In the symposium it became clear that we have to distinguish animacy-markings in general from animate (active) case systems. The latter is clearly related to the so-called split (animate versus non-animate) marking of the intransitive subject, as Tsunoda pointed out in his introductory comment of the symposium. But the former, the animate marking, appears in various forms in many particular languages: it is a grammatical phenomenon indifferent to the split marking of intransitive subject. The Nivkh animacy, I reported in the symposium, is a somewhat peculiar complex grammatical phenomenon, because (1) the language has a special case suffix -ax indicating an animate causee, (2) the suffix appears only in the embedded clause of a causative sentence if the causee is animate where the causativized verb itself can be transitive or intransitive, but (3) the Nivkh case system is, in essence, transitive, in the sense that the agentive, transitive subject and the object, too, are zero-marked: it belongs to a different case system from the so-called active case system like Goergian or Haida as presented by Kojima and Hori, respectively. So, Kazama is right, when he pointed out that there is something more to discuss about the animacy of Kikaijima dialect (presented by Matsumoto) and of Nivkh (Report on 134th Meeting of JLS, GENGO, Vol.36-No.9, p.100). I doubt rather if an active case system can be established as a typologically relevant type of case systems, but suspect that it is reduced to some specified grammatical phenomena in particular languages.
2) For the other dialect of this language, the South-East-Dialect in Sakhalin, we need a separate paper with new linguistic data.
3) The verb in the embedded phrase of a causative construction can be transitive or intransitive in this language. This implies that the grammatical phenomenon of animacy marking of causee in this language is nothing to do with the so-called split case marking of the subject of the intransitive verb in the so-called active case type mentioned in case typology.
Hattori, Takeshi 2000: Selected Writings, Hokkaido Publ. Center （服部健2000：『服部健著作集』、北海道出版企画センター）
Grudzeva, Ekaterina, Nivkh. Languages of the World/Material 111 Lincom Europa 1998
Kaneko, Tohru 2007: A Note on CAUSE in Nivkh. in Papers of Chiba University Eurasian Society(CES) No.9, pp.1-29
Krejnovich,A.E. 1958: Ob incorpopropriuvanii v Nivkhskogo Jazyke. Voprocy Jazyjaznania 1958-No/6.pp 12-33 Moskva-Leningrad
Panfilov,V.Z. 1962/1965: Grammatika Nivkhskogo Jazyka,Vol.1./Vol.2,IAN SSSR,
Saveljeva,V.H.& Ch.Taksami 1970: Nivxsko-Russkij Slovari, I Sov.Enz./Moskva
Shiraishi,Hidetoshi 2007:Topics in Nivkh Pholonogy., Groningen Dissertation in Linguistics 61, Groningen/Netherland
Takahashi Moritaka 1942, Sakhalin Gilyak Language,Asahi Shinbunsya.（高橋盛孝1942：『樺太ギリヤク語』、朝日新聞社）
cf. HP of Kaneko Tohru: www.ne.jp/asahi/kaneko-tohru/languages-nowar/
The language Nivkh has a special case marking for the causee. It occurs only in a syntactic construction like
(15) [VP 1 the agent of V-gu [VP2 [ [NP the agent [+intention] of V]-ax ] ...V-gu -...]
CAUSER CAUSEE CAUSE
N-ax co-occurs necessarily with V-gu, where V-gu itself the transitivized form of V which can be transitive or intransitive. N-ax depends, therefore, on the occurrence of V-gu. So, it may be regarded as syntactic option of -gu-transitivation of V.
The NP-ax can be omitted, but the condition for the omission is severe. First a reference-identical NP has to precede it in the same sentence. Second, the transitivized V-gu has to dominate the omitted NP-ax directly and the preceding NP at lest indirectly, like in the sentence
(13''') nJ i [NP1 ucitJel-dox [ VP2 [NP2 ucitJel-ax ] zadaca pH-«ks«m-gu -in« -t ] vi-dJ .
preceding NP omissible NP transitivized V
An ellipsis like this is permissible e.g. in Japanese, but surely somewhat ill-formed in many European tongues. The typological variety may be worth thinking about in more details.
The causee marker -ax has a semantic function to denote the animacy of the agent in the embedded VP. It belongs to the category of active case makings. This case marking denotes the animacy of an NP in a causative structure. It is one of the case markings of this language and denotes simply the animate causee. So, it is a derivational marker with a single grammatical function, so that it makes no net work nor any grammatical subsystem like an active case system. If such a grammatical device has to be understood as something beyond a simple case marking, the animacy marking in Nivkh is nothing other than a derivational case marking, Therefore, it is merely an isolated grammatical device something other than an active case system as in, say, Caucasus languages.