2.2.1 An overview

Tables 2 and 3 contain a list of recurrent meaning correspondences between cases with noun phrases, and with clauses, as their scope. Table 2 features cases which combine core and non-core (oblique, or peripheral) functions (ergative-instrumental, ergative-ablative, etc.). Table 3 features cases that only have non-core functions (locative, allative, etc.). In each instance, I provide one or two illustrative examples of languages where such a phenomenon has been documented synchronically (the sources are listed in appendix 2).Footnote ¹⁰

Table 2

Meanings of cases having core uses, with noun phrases and with clauses

Table 3

Meanings of cases lacking core uses, with noun phrases and with clauses

2.2.2 Semantics of cases as clause linkers

Polysemous cases which combine core and oblique functions appear to be more likely to have a clause as their scope than purely core cases. The few examples of purely core cases marking clause linking involve Muskogean languages (e.g. Koasati, where the nominative case suffix -k is identical to the same-subject switch reference suffix, and the accusative -n to the different-subject marker: see Kimball Reference Kimball1991: 225, 391–395, 522–525; parallel phenomena in Choctaw are in Nicklas Reference Nicklas1974: 98, 211; see also Jelinek Reference Jelinek, Marácz and Muysken1989: 135–137, and Jacobsen Reference Jacobsen, Haiman and Munro1983: 176). In Yuman languages the erstwhile allative suffix -m came to be used as a same-subject marker in clause linking (Kendall Reference Kendall1975: 4, Langdon Reference Langdon, Campbell and Mithun1979: 630). In some Yuman languages it also developed into an object case and subsequently acquired the role of a clause complementizer (Jacobsen Reference Jacobsen, Haiman and Munro1983: 175). The correlation between nominal core objective case and a complementizer was pointed out for Diegueño by Gorbet (Reference Gorbet1973: 221); see also Gorbet (Reference Gorbet1976: 121–128; 1979: 261–263) for the putative development of subjective case into a same-subject clause-linking morpheme. Muskogean and Yuman (where the interpretation varies depending on the reconstruction and approach) have not been included in table 2, so as to keep it relatively simple.

In Lepcha and Bodic languages (Tibeto-Burman), and in Murinypata and Djambarrpuyngu (Australian), the ergative case has an additional, oblique-argument marking function, and is employed to link clauses. In Kusunda, the accusative-purposive case marks direct object, recipient, beneficiary and purpose (in addition to ‘dative subject’), and location; it is also employed as a clause-linking device. Tables 2 and 3 show that there are few if any semantic differences between core and non-core cases when they are used as clause-linking markers. The two sets of case-markers are presented in separate tables simply for ease of reference.

Cases and adpositions on noun phrases mark the functions of those noun phrases within the clause (see the overview in Blake Reference Blake2001). In contrast, the meanings of the same morphemes as clause linkers are consistent with the major semantic types of clause linking (see Dixon forthcoming b; and also the partial list in Thompson & Longacre Reference Longacre and Shopen1985: 177), which include temporal sequence, condition, cause, purpose, possible consequence, location, and manner. Case markers are also used as complementizers. There appear to be no examples of cases used for expressing conjunction or disjunction of clauses. Tables 2 and 3 show that a case may sometimes have a very similar meaning when it occurs with a noun phrase and with a clause. Other times, there are consistent differences.

Before proceeding to generalizations capturing these differences, we examine two examples, (a) from English, and (b) from Ket.

(a) English has a handful of prepositions which can also occur on a clause, marking its syntactic relationship with another clause. These are: after, before, since, until, till, and for. The meaning of most prepositions when used with a noun phrase and with a clause is the same: compare (6a) and (6b). Curly brackets indicate the boundaries of the noun phrase and of the clause within the scope of the preposition.

1. (6)

   1. (a) She had a hard time after {the death of her husband}.

   2. (b) She had a hard time after {her husband died}.

The preposition since is less straightforward: it has a temporal meaning when used with a noun phrase and either a temporal or a causal meaning when used with a clause. Both (7a) and (7b) are acceptable, and the preposition since has a temporal meaning in both examples; in (7a) its scope is a noun phrase, and in (7b) it is a clause.

1. (7)

   1. (a) I've been very lazy since {the end of summer school}.

   2. (b) I've been very lazy since {summer school ended}.

However, in its causal meaning since can only be used with a clause, as in (8a). This meaning with a noun phrase argument would not be grammatical, as shown in (8b).

1. (8)

   1. (a) Since {I disliked his manner}, I turned him down.

   2. (b) *Since my dislike for his manner, I turned him down.

Further discussion can be found in Long (Reference Long1965) and in Quirk et al. (Reference Quirk, Greenbaum, Leech and Svartvik1985: 659f.). The temporal and the causal meanings are semantically linked – if two events are mentioned together as following each other in time, it may be possible to infer that one is the cause of the other (see Thompson & Longacre Reference Longacre and Shopen1985: 181ff., Longacre Reference Longacre and Shopen1985). However, the fact that since expresses a causal relationship only when it links clauses alerts us to the fact that the context of use may entail different semantic overtones for what has traditionally been considered the same polysemous morpheme. Along similar lines, a locational case may have a somewhat different array of meanings with nouns and with clauses. This takes us to the next example.

(b) Ket (a Yenisseic language: Werner Reference Werner1997a: 105, 354) has nine clausal cases (not counting genitive), six of which can be used for clause linking.Footnote ¹¹ Adessive case means ‘at, towards’ when used with nouns, as shown in (9) (from Vajda Reference Vajda2004: 27). It can also mark the second argument of verbs denoting thinking (about something) or narrating (about something).

1. (9) bū láɣ-ìn-nà-ŋta òn sk du-o-il-daq3masc Selkup-pl-ap-ades many years 3masc.sj-d-pt-live‘He lived among the Selkups (lit. at the Selkups, that is, at their camp) for many years’

If used with clausal scope, the meaning of the adessive is causal, locative, or conditional. The causal meaning ‘because’ is illustrated in (10) (unglossed example from Werner Reference Werner1997a: 353; glosses from Anderson Reference Anderson and Vajda2004: 68):

1. (10) {at t-lɔver-a-vet-diŋta} at saʔj iñ-d-aqI 1-work-present-sf-ades I tea p/perf-1-give‘Because I work, give me tea’

The locational meaning of the adessive case is typical of nominal cases, while causal and conditional meanings are typical of clause-linking semantics (see Thompson & Longacre Reference Thompson, Longacre and Shopen1985: 177, and Dixon forthcoming b).

In a nutshell, the meanings of cases on noun phrases are consistent with the semantic functions of noun phrases, as recipients, beneficiaries, instruments, and locations. The meanings of cases as clause-linkers follow the major semantic types in linking clauses. These clause-linker meanings are intertwined: temporal sequence often has overtones of condition and cause; purposive clause linking may also indicate cause, or consequence, or ‘lest’ (Dixon forthcoming b). Based on the selection of language-specific correlations between the meanings of a case when occurring with a noun phrase and with a clause, as exemplified in tables 2 and 3, we can suggest a number of semantic correspondences between noun-phrase functions and clause-linking devices.

I. Noun-phrase markers with a dative, purposive, or benefactive meaning tend to have purposive meanings as clause linkers. But, despite its frequency, this is not a steadfast rule: in Ket, the dative case marker is used to mark locative clauses and even relative clauses (Werner Reference Werner1997a: 353), alongside ‘destination’ (Vajda Reference Vajda2004: 25).

II. Noun-phrase markers with locational meanings are likely to develop temporal connotations indicating relative time – ‘while, as soon as, after, upon’ and others – if they have a clause as their scope. This correlation is akin to the well-documented semantic extension from spatial to temporal notions in the domain of adverbs and of case markers (Haspelmath Reference Haspelmath1997; Heine & Kuteva Reference Heine and Tania2002: 40f., 179f., 183). This is supported by recent psychological experiments, confirming that ‘spatial representations are the source of temporal representations’ (Gentner et al. Reference Gentner, Imai and Boroditsky2002: 557).

The locational meaning of a noun-phrase marker can carry over to its meaning when occurring with a clause, but within a temporal rather than a spatial domain. For example, the perlative case means ‘along’ in Djambarrpuyngu noun phrases, and ‘concurrent with main clause’ when used on clauses containing motion predicates. An additional extension of ‘along’ when used with clausal scope is ‘the situation which is the channel or means for the main clause situation’ (Wilkinson Reference Wilkinson1991: 641f.). In Kham, lative ‘up to’ means ‘until’ when used with a clause (Watters Reference Watters2002: 317; see also examples in Blake Reference Blake1999: 307f., from Australian languages; and especially Genetti Reference Genetti1986, Reference Genetti, Traugott and Heine1991).

There can be additional extensions. A conditional meaning for a locative case marker when used as a clause-linker is documented for Ket; this can be viewed as an extension of its temporal meaning (Werner Reference Werner1997a: 354). Elative and ablative have conditional meanings in a number of languages (Rama, Qiang, Kham and Classical Tibetan). This development can be considered an extension of an erstwhile temporal meaning of a locative morpheme (see also Thompson & Longacre Reference Thompson, Longacre and Shopen1985, for links between condition and time in clause-linking). A locational case marker can express purpose. In some Tibeto-Burman languages, a locative marks purpose complement (as in Lepcha: Plaisier Reference Plaisier2006: 119f.; and Cogtse Gyarong: Nagano Reference Nagano, Thurgood and LaPolla2003: 487). Lichtenberk (Reference Lichtenberk, Traugott and Heine1991: 71–74) provides an explanation of how an ablative postposition came to mark positive purpose in Toqabaqita, and negative consequence ‘lest’ in Kwaio.

III. Noun-phrase markers with instrumental meaning have a causal or a temporal meaning when used as clause linkers. At least in one of the languages where this development has been attested the instrumental case on a noun can also express cause as a clausal linker, as in Tariana (Aikhenvald Reference Aikhenvald2003). We will see in section 3.4.2 that an instrumental case without a causal meaning is used for manner clause linking.

Other, more ‘exotic’ non-core cases are relatively straightforward. The similative case ‘like’ in Kwoma and in Limbu has the same meaning as a clause linker; the sociative marker on nouns expresses sequence of events in Yamphu and Toqabaqita. Approximative ‘about’ in Kham develops an appropriate clause-linking meaning of temporal ‘as long as’, or quantitative ‘as much as’ (Watters Reference Watters2002: 317).

In none of the instances mentioned in tables 2 and 3 using what looks like a nominal morpheme – a case or an adposition – on a verb does not imply an underlying nominalization.Footnote ¹² In the absence of an overt nominalizing morpheme, to say that a verb must be nominalized with a zero marker in order to be used with a case or an adposition is pure conjecture. There is no reason to believe that a verb like nu-nu-se ‘as soon as I come’ in (3) in Tariana is nominalized. If it were, it would be the only instance of a zero-marked nominalization in the language.

In addition, in Tibeto-Burman languages (such as Lepcha: Plaisier Reference Plaisier2006), the subject of a nominalization is marked differently from that of a subordinate clause whose predicate takes a case marker as a linking device. English also has a wide variety of nominalizing devices (see Dixon Reference Dixon2005: 322–352); but there are no language-internal reasons to consider dependent clauses like the ones in (6b) and (7b) and (8a) as ‘nominalized’. Along similar lines, Genetti (Reference Genetti, Traugott and Heine1991: 246) argues that in Classical Newari, ‘the first stage in the development of postpositions to subordinators’ was ‘the suffixation of nominal postpositions to fully inflected finite’ verb forms, ‘which lacked overt nominalizing suffixes’.

We conclude that, as expected, the semantics of ‘cases’ as clause-linkers fits in with the mould of semantic patterns of clause linking. A recurrent semantic correspondence is that holding between a spatial meaning of a nominal case and a temporal meaning of a clause linker, confirming that the domains of space and time share conceptual structure (see Boroditsky Reference Boroditsky2000: 25f. for experimental results to this effect, and the discussion of the underlying mental representations).

In terms of its historical development, the polysemy of nominal cases and clause-linkers has often been understood as a product of ‘grammaticalization’ of cases to clause linkers (in the conventional sense of the term, see Meillet Reference Meillet1912; Heine & Kuteva Reference Heine and Tania2002). The facts of Rama (Craig Reference Craig, Traugott and Heine1991: 471) corroborate this. Postpositions with noun phrase scope in Rama (see (4)) are free morphemes, while the corresponding markers with clausal scope are bound morphemes. This can be interpreted as pointing towards a unidirectional grammaticalization path:Footnote ¹³

1. (11) adposition (free morpheme) with noun-phrase scope>adposition with clausal scope (developed into a bound morpheme)

The direction of development in (11) is congruent with a general path of semantic change (especially in grammaticalization) whereby more specific meanings, or meanings based on a specific situation, become more general, or ‘based on the textual situation’ (Genetti Reference Genetti, Traugott and Heine1991: 249).Footnote ¹⁴

2.3.1 How clause-linking cases come to mark categories of a main clause

We can recall from table 1 that, when case markers occur on verb roots,Footnote ¹⁵ they are likely to express clausal categories such as aspect, modality, and mood.

Similar developments occur if case markers attach to deverbal nominalizations used as predicates of dependent clauses. If a dependent clause becomes reinterpreted as a main clause via ellipsis, the case morphemes are reinterpreted as aspect or modality markers. This path is summarized by Blake (Reference Blake1999, especially his table 3 on p. 304), for Australian languages.Footnote ¹⁶

For instance, a dependent apprehensive clause referring to possible negative consequence typically occurs with the main clause expressing a warning, or something to be avoided, as in (12), from Pitta-Pitta (Blake Reference Blake1999: 307, 310).

1. (12) wilakana-ya kiniyarri [nhan-(nh)a-ka piyawarli-lu patya-ka-la]hide-present girl she-acc-here dog-erg bite-[p]-lest‘The girl is hiding lest the dog bite her’

The case marker -la – which means ‘lest’ with clauses – has a causal meaning with noun phrases. It also marks the complement of verbs of fear (Blake Reference Blake, Dixon and Blake1979: 198), as in (13):Footnote ¹⁷

1. (13) kiniyari &ndsqrb;an-pa-ka alala kupakupa-lagirl she-nomin-here afraid old.man-causal‘The girl is frightened of the old man’

The main clause may be left out in Pitta-Pitta, and then the erstwhile dependent clause comes to be used on its own, with the same ‘apprehensive’ meaning, as in (14) (Blake Reference Blake1999: 310):

1. (14) nhan-(nh)a-ka piyawarli-lu patya-ka-lashe-acc-here dog-erg bite-[p]-lest‘The dog might bite her’

In Dixon's (Reference Dixon2002: 239) words, ‘types of subordinate clauses have been reanalyzed as main clauses, so that what were verbal suffixes marking subordination now take on TAM [tense, aspect, mood] values’. A typical development involves the aversive case ‘for fear of’ on a noun phrase to apprehensive modality on a verb in a clause (shown in (13) and (14)), and from dative case on noun phrases to purposive, intentional, or future on clauses (Dixon Reference Dixon2002: 237–239; Blake Reference Blake1999: 309f., 1993: 40).

Along similar lines, in Dyirbal -gu (Dixon Reference Dixon1972) is used both as dative on nouns and as purposive marker on verbs (of one conjugation). In Kalkatungu (Blake Reference Blake1999: 315), the dative case marker -ya forms a purposive in combination with past and proprietive markers (the resulting form is: -ny-tya-ya p-prop-dat). In this same language, a combination of -ya ‘dative’ (on nouns) and future -mi forms potential -mi-ya.Footnote ¹⁸

Reanalysis of dependent clauses as main clauses does not always proceed through the intermediary of a nominalization (as happens in Australian languages). The manifold meanings of -k and -m in Maricopa (Yuman) verbs in independent clauses appear to go back to their uses in dependent clauses (Gordon Reference Gordon and Munro1980: 140), but no nominalization is postulated.Footnote ¹⁹

2.3.2 ‘Aspectual’ cases

A clausal category can be expressed not only by a case marker on a predicate, but also by case on a noun phrase, as illustrated in this section. In numerous Balto-Finnic languages, especially Finnish, the choice of partitive vs. accusative case for the object of a number of verbs has aspectual implications.Footnote ²⁰ Examples (15) and (16) (from Kiparsky Reference Kiparsky, Butt and Geuder1998) illustrate this semantic effect. The ‘irresultative’, unbounded, and atelic event in (15) involves partitive case on the object; the bear was shot at, but not killed.

1. (15) Ammu-i-n karhu-ashoot-p-1sg bear-part‘I shot at the/a bear’

In contrast, the resultative and bounded version in (16) involves the object marked with the accusative. (16) ‘denotes an accomplishment, “to shoot dead”’, while (15) ‘denotes an activity’ (Kiparsky Reference Kiparsky, Butt and Geuder1998: 267):Footnote ²¹

1. (16) Ammu-i-n karhu-n shoot-p-1sg bear-acc‘I shot the/a bear (it is dead)’

Finnish also has intrinsically ‘unbounded’ verbs (such as ‘love’, ‘touch’) which require partitive objects, and intrinsically bounded verbs (such as ‘kill’, ‘find’) which require the accusative. The partitive case also denotes objects of indeterminate quantity (‘some’), and can be described as a marker of indefiniteness (see Sands Reference Sands2000; Sands & Campbell Reference Sands, Campbell, Aikhenvald, Dixon and Onishi2001).

In addition to this, in Finnish, ‘the use of local cases […] has aspectual implications. The inessive and adessive case imply a continuing activity and so have imperfective aspect, whereas the elative, illative, ablative and allative all imply an end-point of the activity and perfective aspect’ (Sands Reference Sands2000: 277; and a summary in Hakulinen Reference Hakulinen1961: 333). ‘The process as opposed to the result is indicated with the static-location cases (inessive and adessive)’ (Sands Reference Sands2000: 277). If the allative case is used, as in (17), the activity is considered to be a result; this station may be the train's final destination:

1. (17) Juna pysähtyi asema-lletrain stop.p.3sg station-all‘The train stopped at the station (lit. towards the station as its final destination)’

In (18), the adessive case implies that the station is ‘simply a stopping point on the train's journey, and the train continues on’. (Comparable phenomena in Estonian are addressed in Tuldava Reference Tuldava1994: 106f.).

1. (18) Juna pysähtyi asema-llatrain stop.p.3sg station-ades‘The train stopped at the station (lit. at the station as a passing point)’

In each of these instances, the way in which a noun phrase is marked affects the aspectual value of the clause. The semantic effect of case is reminiscent of the absolutive/dative case alternation in Warlpiri: marking the object (O) of a verb like ‘shoot’ as dative rather than as absolutive describes ‘the situation in which the effect normally resulting from the action denoted by the verb is, for one reason or another, aborted or else is subordinated in importance to the action itself’ (Hale Reference Hale and Swartz1982: 249). This phenomenon has further been characterized as a ‘conative’ case alternation which imparts a special ‘irresultative’ aspectual value to the clause as a whole (see Kiparsky Reference Kiparsky, Butt and Geuder1998: 266, 295f., and further references there). All this helps to confirm the status of aspect and modality as clausal categories which do not necessarily have to be marked specifically on the predicate.Footnote ²²

Historically speaking, the partitive case in the Balto-Finnic subgroup of Finno-Ugric comes from a locational case with a separative meaning ‘from’ (see Laanest Reference Laanest1975; Kiparsky Reference Kiparsky, Butt and Geuder1998). Correlations of case marking with aspect are generally considered a later development in Balto-Finnic. Larjavaara (Reference Larjavaara1991) hypothesizes that the locational ‘from’ case developed an aspectual overtone via a ‘quantificational’ meaning ‘some’ with a noun phrase. Further historical and comparative evidence indicates that, in Balto-Finnic, ‘the partitive's emergence as a structural case is a precondition for the rise of its aspectual function’ (Kiparsky Reference Kiparsky, Butt and Geuder1998: 305). That is, the noun-phrase-level function of a case is primary with respect to its other functions, such as marking aspect. This is comparable to the generalization under (11) above: a case or an adposition with noun phrase scope may develop into a case or an adposition with clausal scope, but not the other way round.

We now turn to similar phenomena in Manambu, a previously undescribed language from New Guinea.

3.3.1 The objective-locative case and completive aspect

The objective-locative case -Vm on noun phrases marks a second argument if the latter is completely involved in the action, or if it is completely affected, or if the action is telic. Consider the verb kwakə-. It means ‘find’ if the object is marked with the objective-locative in its function as ‘complete involvement case’, as in (19), and ‘search, look for’ if the object is unmarked, as in (20):

1. (19) {[a takwa:m] kwakə-ku} {wiya:r wula:l}dem.dist+fem.sg woman+lk+obj/loc look.for/find-compl.ss house+lk+all/instr go.inside+3fem.sgbas.p ‘After having found that woman, she went inside the house’

2. (20) [ñanugw amæy] kwakə-ya-banachildren mother look.for-come-1plsubj.nonpast+3fem.sgbas.nonpast‘We keep looking for children's mother’

The other major function of the objective-locative case is to mark a location which has been reached. An example is given in (21). The objective-locative case also conveys the idea of a completed action of reaching the cassowary's breast by climbing. (The cassowary, a flightless bird, is conceptualized here as a mythological woman. (21) comes from a myth about a man's head clinging to the cassowary's breast as if he was her baby.)

1. (21) [a məd lə-kə muñ-a:m] ata war-də-ldem.dist+fem.sg cassowary she-obl+fem.sg breast-lk+obj/loc then ascend-3masc.sgsubj.p-3fem.sgbas.p‘He (the man's head) went up to the cassowary's breast (reached it and stayed on it)’

If the destination has not yet been reached, the allative is appropriate, as in (22):

1. (22) pətəkaur ata war-də-lladder+all/instr then ascend-3masc.sgsubj.p-3fem.sgbas.p‘Then he went up a ladder (but did not reach the top)’

The objective-locative case thus conflates two functions: it marks a core or an oblique argument, and at the same time contributes an aspectual value of ‘completeness’ to the whole clause. This is reminiscent of the aspectual overtones of the partitive case in Finnish discussed in section 2.3.2, and even more so of the locational cases (see (17) and (18) above).

This same objective-locative case morpheme occurs on verbal roots. There, it indicates completion of an action, or total achievement of a state, e.g. (23):

1. (23) wun [də-kə-m] wukəmar-ə-mI he-lk-obj/loc forget-lk-obj/loc‘I completely forgot him’

The completive meaning of the objective-locative case on a verb is strongly reminiscent of the overtone of ‘complete’ involvement of an object or a location which is characteristic of the objective-locative as used with noun phrases. None of Manambu's relatives has any cognates of the objective-locative case. Consequently, we cannot tell which function of the -Vm form is historically prior. This is quite unlike the Balto-Finnic languages (see section 2.3.2), where – as we know – the primary function of cases with ‘aspectual’ overtones was to mark grammatical relations, while the aspectual overtones represent a later development.

3.3.2 The dative-purposive case and intentional modality

The dative-purposive case -Vk with noun phrases expresses intention and purpose, as in the first clause of (24), and the third argument of a ditransitive verb, as in the second clause:

1. (24) [sana:k ya-k-na-di] sa:n dayak kui-tukwamoney+lk+dat/purp come-fut-act.foc-3plbas.nonpast money they+dat give.to.third.person-proh‘They will come for money (to get money, with the intention to receive money), don't give them money’

The dative-purposive case is also used to mark reason (as in (26), on agwajapək ‘what for’), the object of verbs of emotions, such as ‘like’, ‘dislike’, ‘yearn for’, ‘refuse’, and the object of verbs of fear. It can also mark the object of an atelic verb with slightly frustrative overtones. We can recall, from (19), that the verb kwakə- means ‘find (telic action)’ if its object is marked with the objective-locative case, indicating complete involvement of the second argument; if the object is unmarked, the action is atelic (‘look for’), as in (20). If the object is instead marked with the dative-purposive case, the action is equally atelic, but with an additional frustrative overtone of ‘doing something in vain’:

1. (25) amæyik kwakə-danamother+lk+dat/purp look.for-3plsubj.nonpast+3fem.sgbas.nonpast‘They are looking for their mother in vain (and not finding her)’

The semantic effect of the objective-locative case in (19), as compared to that of the dative-purposive case in (25), is strongly reminiscent of the absolutive/dative case alternation in Warlpiri (see Hale Reference Hale and Swartz1982: 249; and section 2.3.2 above). The dative-purposive case imparts a modal frustrative value to the whole clause.

The dative-purposive case when occurring on a verbal root form marks intentional modality. In (26), it attaches to the verb root warya- ‘fight’:

1. (26) ñən wun-a-wa agwa-japək warya-kyou.fem I-lk-com what-thing+lk+dat fight-dat/purp‘I am going to fight, why (lit. what for) are you going to fight with me?’

The semantic link between the purposive meaning of the dative-purposive on nouns, as in (24), and on verbs, as in (26), is straightforward. The dative-purposive on verbs is also used as a clause-linking device – see the next section.

3.4.1 The dative-purposive case and same-subject purposive clauses

The intentional or purposive meaning of dative-purposive-marked predicates echoes the meaning of similarly marked noun phrases. A same-subject purpose complement to a verb of motion is shown in (27):

1. (27) {wun {mən-a:m karda-k} war-na-dəwun}I you:masc-lk+obj/loc take+down-purp.ss ascend-act.foc-1masc.sgbas.nonpast‘I have come up to take you downwards’

A noun marked with the dative-purposive and a verb marked with the same-subject verbal purposive are used in identical contexts. (28) illustrates a dative-purposive-marked verb warya- ‘to fight’ expressing purpose of speaking. In (29), a dative-purposive-marked noun di ‘shit’ is used.

1. (28) warya-k wa-nafight-purp.ss say-act.foc+3fem.sgbas.nonpast‘She intends to fight (lit. she said she was going to fight)’

2. (29) diya:k wa-nashit+lk+dat/purp say-act.foc+3fem.sgbas.nonpast‘She intends to have a shit (lit. she said for shit)’

There is no reason to consider the same-subject purposive form, or the intentional, a nominalization. These forms have no nominal properties whatsoever. Neither does the bare root of a verb. The intentional and the same-subject purposive are semantically similar. One may hypothesize that both go back to the dependent purposive, and postulate a subsequent reanalysis of the dependent purposive as a main clause, and of the verb form as intentional modality (along the lines of section 2.3.1 above). This analysis is plausible. However, it does not account for the lack of same-subject constraint when the morpheme occurs in intentional modality, and thus remains a mere hypothesis.

Whether an originally nominal case in fact got extended to a verbal environment in Manambu is an open issue. The cognates of the Manambu dative-purposive appear on both nouns and verbs in related languages. The Maprik dialect of Ambulas (Ndu family, Papua New Guinea: Wilson Reference Wilson1980: 68–75, 119f.) has a ‘referential’ case -ke which marks purpose, goal and theme (of a conversation). The same case-morpheme occurs as a marker of intentional modality in dependent clauses (cf. also Wendel Reference Wendel1993: 88, 102). This suggests that the dative-purposive in Proto-Ndu could have been just as polyfunctional as it is in the present-day languages, and that there is no reason to consider either the nominal or the verbal context to be diachronically prior.

3.4.2 The allative-instrumental case and same-subject manner clauses

Unlike the objective-locative and the dative-purposive cases which can mark core arguments, the allative-instrumental case always marks obliques. When used with noun phrases, the allative-instrumental case -Vr marks direction, as in (22), and instrument, as in (30):Footnote ²⁶

1. (30) [am-awa nəbi:r] ata vya-də-dibow-lk+com arrow+instr then hit-3masc.sgsubj.p-3plbas.p‘He then hit them with bow and arrow’

The allative-instrumental case has an additional meaning of ‘along’ (e.g. a road). It does not have any comitative overtones; neither is it used to express reason.

The allative-instrumental case appears on verbal roots to mark the predicate of same-subject manner clauses, as in (31).

1. (31) {ñam kuyar} {ñanugwa:k kamna:gw kui-la}chewed.food give.to.third.person+instr children+lk+dat food give.to.third.person-3fem.sgsubj.p+3fem.sgbas.p‘She gave food to children by giving (them) chewed food’

Tables 2 and 3 show that cross-linguistically an instrumental case occurring with clauses typically expresses causal or temporal linkage. The unusual meaning of manner linkage of the instrumental case in Manambu may be due to its lack of causal overtones; this is where it differs from the instrumental case in Tariana (see III in section 2.2.2 above). An instrumenal case typically acquires a manner extension (Blake Reference Blake2001). However, Manambu is unusual in that the allative-instrumental case marks manner only when used with verbs as a clause-linking device, and does not express any other semantic relationships (e.g. reason, as in other languages). A noun expressing manner is unmarked for case. Also note that, except in Manambu, case markers do not seem to be used for manner linkage in the languages for which descriptions were available (see table 3).

The allative-instrumental case has clear cognates throughout the Ndu family. However, its use with verb roots has not been documented for any language other than Manambu.

3.4.3 The substitutive case and substitutive clauses

The substitutive case when used with noun phrases means ‘instead of’. Just like the allative-instrumental case, it occurs on obliques only, as in (32).

1. (32) pilou-a-yæy kusu-wapwi a-takpillow-lk-subst wear-clothes impv-put‘Put clothes (on your bed) instead of a pillow’

The substitutive -yæy on a verbal root is a same-subject sequencing marker with the meaning ‘instead of, rather than’:

1. (33) {awarwa warya-yæy} {aka kəp lakati-dana}rec fight-subst:ss then just sort.out-3plsubj.nonpast+3fem.sgbas.nonpast‘Instead of fighting each other, they just sorted it out’

If the subjects are different, the substitutive case marker attaches to a partially inflected verb (which, as we can recall, implies cross-referencing just the subject). An example is (34).

1. (34) {ñən kiya-ñəna-yæy} {də-kə-m vya-təpul-kə-la-d}you.fem die-2fem.sgsubj.nonpast-subst he-obl-obj kill-‘hit’-fut-3fem.sgsubj.nonpast-3masc.sgbas.nonpast‘Instead of you dying, she will kill him’

The substitutive case has no cognates in other Ndu languages. That is, we have no historical information as to whether its adnominal or its verbal use is diachronically prior.