Printer Friendly

On giving, receiving, affecting and benefitting in Jalonke *.

Abstract

The present article investigates the syntactic and semantic properties of verbs with optionally three participants in Jalonke. Jalonke, a variety of Yalunka, belongs to the Mande group of the Niger-Congo language stock and is spoken in and around Guinea (West Africa). In contrast to languages with ditransitive verbs, Jalonke has no verbs with three arguments. Rather, the third participant of the verbs in question is realized in a postpositional phrase, just like an adjunct, and is not distinguishable from adjuncts on semantic and syntactic grounds. The article gives an overview of the postpositions involved in the marking of the third participant of verbs whose second or third participants are Recipients, Beneficiaries, and to a lesser extent, Experiencers, since these are the roles attested across languages for verbs encoding three-participant events. Jalonke verbs with optionally three participants are explored according to two parameters of variation: the thematic roles and linking properties of their second and third participants and the alternation attested for verbs and classes of verbs. The article further systematizes which of the crosslinguistically attested linking patterns occur in Jalonke, and which semantic parameters govern differences between semantically related verbs and constructions in the language.

1. Introduction

This article explores the ways in which the participants of verbs with optionally three participants are marked in Jalonke, a Central Mande language of Guinea (West Africa). The article focuses on which participant is linked to object and which to postpositionally marked participant for individual verbs, and which alternation patterns in linking are attested for some verbs. Crosslinguistically, one of the participants of three-place predicates is typically marked as indirect object or as adjunct participant of a transitive verb. In some languages, syntactic criteria allow the identification of certain adjunct participants as oblique arguments. In view of the impossibility to classify participants marked with postpositions in Jalonke as arguments, and hence to identify three-place predicates for the language, the present article concentrates on verbs whose third participant, independent of its syntactic status, has one of the thematic roles generally attested for one of the participants of three-place predicates across languages. These crosslinguistically attested roles are Recipients, Beneficiaries and Experiencers. Given that Experiencers are not specific to three-participant verbs but also occur as participants of verbs with only two participants, their linking is not discussed in detail in the present article because of its focus on three-place verbs.

Jalonke does not have a single marking pattern for the syntactic expression of the relevant thematic roles. Rather, the language employs several postpositions to code the participant whose thematic role corresponds to a Recipient, Beneficiary, or Experiencer. In addition, Jalonke also allows several ways to link Recipient and Experiencer, but not Beneficiary, participants of the relevant events to argument and adjunct syntactic slots, reminiscent in some, but not all respects, to alternations widely known as the "Dative Alternation" and the "Locative Alternation" (Levin 1993). This article begins with an investigation of the syntactic encoding of participants in Jalonke. It further explores the semantics of the postpositions involved in the marking of the third participant of the verbs in question. In addition, the article examines the variations in linking for these verbs, employing the notions of frame semantics (Fillmore 1977, 1982; Fillmore and Atkins 1992; Fillmore and Kay 1993), lexical profiling (Langacker 1987), and syntactic profiling in order to account for the differences in linking between verbs, and for the alternations attested for a number of verbs.

The article is structured as follows: The remaining sections of the introduction provide essential information on Jalonke. Section 2 gives an overview of semantic and syntactic profiling of participants, crosslinguistically common marking and linking patterns for verbs with three participants and introduces the patterns attested in Jalonke. Section 3 has as its focus on the semantics of the relevant postpositions and introduces the three postpositions ma 'at' ra 'with' and be 'for' that mark animate Sources and Goals, Instruments, and Beneficiaries, and touches briefly upon the marking of Experiencers. Section 4 looks at variations in linking available for verbs with optionally three participants in Jalonke, ascribes the differences for verbs that appear exclusively in one linking pattern to differences in lexical or syntactic profiling, and discusses possible motivations for alternating verbs. A final section summarizes the findings.

Jalonke is the name used by the speakers for a variety of Yalunka spoken in the Futa Jalon, a mountainous region in the north of the West African country of Guinea. Yalunka belongs to the Mande branch of the Niger-Congo phylum. The different dialect areas of Yalunka are not contiguous to each other, but separated by large areas of predominantly Fula speaking communities. Not much is known about the exact number of speakers of the language in the different regions. Sources cite the numbers of 113,000 speakers of Yalunka for all areas (Platiel 1978), or of 87,000 for Guinea (Ethnologue 2004), but in the absence of reliable and recent census data, these numbers are not to be trusted too much. The numbers given in Platiel (1978) are based on old colonial census data given in Westermann and Bryan (1952) and de Lavergne de Tressan (1953), modified by adding an estimated population growth rate. The numbers appearing in the Ethnologue (2004) are almost identical to colonial census data from 1950, given in de Lavergne de Tressan (1953) as 87,875. As to the status of the language, it is at least known that it is a minor language and that the speakers in their majority are bilingual. This is even true for the enclaves still constituting homogeneous Yalunka areas. In the Northem areas--with the exception of Mali where Bambara is taking over (Denis Creissels, p.c.)--Fula is the dominating language. In the Southern areas two different Mande languages, Maninka and Koranko, are concurring with Yalunka.

Jalonke exhibits the typical Mande word order SOVX (Creissels 2000), X standing for all adjuncts. It is a configurational language with strong isolating tendencies. Consequently, grammatical relations are marked by word order, not by inflectional case. Participants other than subject and direct object are realized in postpositional phrases. There is no evidence for the grammatical relation of "indirect object" in the language; that is, there is no intermediate category, ranging after subjects and direct objects, but preceding oblique objects on the Accessibility Hierarchy (Keenan and Comrie 1977) with respect to relativization, the use of reflexives, and quantifiers. All NPs other than subject and direct object are realized as postpositional phrases, hence as syntactic adjuncts, in Jalonke.

2. Linking of the relevant participants crosslinguistically and in Jalonke

2.1. Lexical and syntactic profiling

As do many other languages, Jalonke allows several different syntactic realizations or linking patterns for verbs with optionally three participants. Since only two participants can be core arguments of the verb, the third one must be accorded adjunct status. The semantic participant thus downgraded is not always the same across verbs. More importantly, commonalities and differences between verbs can be ascribed to different levels of information. Identical conceptual structures, formalized in frame-semantic roles, properly introduced below, can be realized in different thematic roles; and identical thematic roles can be linked to different syntactic slots. To appreciate the first point, consider the following two examples, involving two distinct verbs of change of possession: (1)
(1) A xin -[??][??] fii -ma dii j[??]r -[??][??] ma.
 3SG milk -DEF Give -IPFV child young -DEF at
 'She Is giving milk to the young child.'
 (Chickens 004)

(2) Tumaani e samba kaseti -na 'a.
 Tumaani 3PL present cassette -DEF with
 'Tumaani presented them with a cassette.'
 (Labe 404)


In example (1), the Theme of 'give' is realized as the object, and the animate Goal (or Recipient) surfaces as an adjunct, marked by ma 'at'. In (2), what seems to be the animate Goal of 'present' is in object position and bears the Theme role. Additionally, what also seems to be a Theme, since it undergoes a change of location, the cassette, appears as an adjunct in (2). This participant is moreover marked like an Instrument by the postposition ra 'with'. Clearly, it is problematic for most grammatical frameworks, which do not admit multiple thematic role assignment, to assert that the Theme of (2) is realized as the Instrument of the change of possession event. Yet we would like to capture the intuition that the cassette in (2) has the same function in the change of possession as the milk in (1). The different solutions proposed for the thematic roles of English with-phrases further illustrate the problem of classifying their participants. Since the adjunct participant undergoes a change of location while serving as the Instrument, the with-phrase in English has been said to instantiate the "locatum argument" (Clark and Clark 1979), a "displaced Theme" (Rappaport and Levin 1988) or "with-Theme adjunct" (Jack-endoff 1990). (2) By keeping frame-semantic and thematic roles apart, one can account for the "theme-like" nature of the postpositionally marked participant more elegantly, and at the same time capture that it actually is an Instrument, despite its additional change of location features. As noted by Goldberg (2002), the with variant of the English Locative Alternation, whose linking patterns correspond to the Jalonke example in (2), is best characterized as a causative construction plus a with adjunct. This with adjunct encodes an entity that is manipulated by the Effector (3) and as such is an Intermediary or Instrument in the causal chain. That it is understood as an entity undergoing a change of location does not follow from its being a Theme.

The solution adopted here to account for the interpretation of with adjuncts and other cases of seemingly double thematic roles assignment goes back to Fillmore (1977), and has been operationalized by Cognitive Grammar (Langacker 1987) and Construction Grammar (Goldberg 1995). It consists of the assumption of a level of lexical information independent of thematic roles or syntactic realization, that of "semantic frames" and corresponding flame-semantic roles. A semantic flame conceptually "structures the word-meaning and [... ] the word 'evokes' the flame" (Fillmore 1982: 117). For the verb fii 'give', denoting a change of possession without exchange of money, such a conceptual or "semantic flame" crucially involves the roles of the initiator of the transfer or a "giver", as well as a "receiver", and the "gifts" that are going to be transferred (Fillmore and Kay 1993: 4.13). (4) By admitting identical frame-semantic roles we can account for the intuitive commonalities between the participants of the verbs fii 'give' and samba 'present' in examples (1) and (2), irrespective of the differences in thematic roles and consequent syntactic alignment present. The two verbs belong to a semantic domain that evokes the same frame-semantic roles: 'he' and 'Tumaani' instantiate the "giver"; 'the child' and 'they' the "receiver"; and 'the milk' and 'the cassette', the transferred "gift". But then how do the differences in thematic roles and syntactic realization between these two verbs come about? The answer is that frame-specific roles of a participant do not determine its thematic role (Fillmore and Kay 1993: 4.15). In Langacker's (1987) terras, differences in "lexical profiling" are responsible for this non-correspondence. The lexical semantics of a given verb treat or "profile" the frames differently: for Jalonke 'give', the "gift" is the Theme, and the "receiver" is the animate Goal. For Jalonke 'present', the "receiver" is treated as the Theme, whereas the "gift" serves as the Instrument to affect the Theme. (5) The argument structure then determines which thematic role will be realized in what syntactic position in default linking. The two different levels of linking are summarized in Table 1 for fii 'give', and in Table 2 for samba 'present with' for the example sentences (1) and (2), respectively.

Following Fillmore it is assumed here that the linking of identical frame-semantic roles to different thematic roles accounts for lexical differences like those between Jalonke fii 'give' and samba 'present'. Alternations or different syntactic realizations for verbs sharing a verb name are taken to present a different case. These alternations involve variation in linking from thematic roles to grammatical relations. Such ah alternation is exemplified for the verb doni 'lend' below:
(3) I a doni -m' an ma nde!
 2SG 3SG lend -IPFV 1SG at INACT
 'You will lend it to me!'
 (Pilgrim-Sall 011)

(4) N mani i doni ra n ma sii -na ma.
 1SG can 2SG lend with 1SG POSS goat -DEF at
 'I can lend you my goat.'
 (Nga 223)


With respect to the linear order of participants, this alternation corresponds to the English "Dative Alternation": in (3), the transferred Theme is syntactically profiled as the object, the animate Goal as the postpositionally marked participant. This sequencing is equivalent to the "to-variant" of the "Dative Alternation" (Birgit gave a bottle of whisky to her supervisor). In (4), an inverse linking order is present: the animate Goal is linked to object, the Theme to postpositionally marked participant. Note that in both examples, the "giver" corresponds to the Effector, linked to subject. Regarding the order of participants, but not their syntactic status, this ordering is consistent with the "double object variant" of the English "Dative Alternation" (Birgit gave me a fascinating book). In these cases, it is stipulated that the difference between the two alternants does not originate in a linking of identical frame-semantic roles to different thematic roles, but in linking differences one level lower, at the mapping of thematic roles to grammatical relations. These mapping differences are summarized in Table 3 and Table 4 for the verb doni 'lend' in examples (3) and (4). (6)

Simplified, one might classify the association of identical frame-semantic roles to different thematic roles as differences in lexical profiling, and the linking of identical thematic roles to different grammatical relations as differences in syntactic profiling. These two mechanisms will play a role in the detailed discussion of individual verbs and their linking variations in section 4.

2.2. Crosslinguistic variation and patterns attested in Jalonke for the linking of participants

Languages as well as constructions within languages are known to vary along crosslinguistically salient parameters with respect to the expression and linking of the participants of three-participant events. These parameters are:

i. The syntactic status of the third participant;

ii. Pivot properties of their participants;

iii. Linking of their participants in comparison to transitive verbs.

As already obvious from the examples given above, Jalonke, too, allows several patterns for the construal of three-participant events. These patterns are systematized in the following paragraphs in terms of the crosslinguistically relevant parameters.

2.2.1. The syntactic status of the third participant. As mentioned before, maximally two participants can be linked to direct argument positions in Jalonke. If a verb has more than two participants, the third participant is realized in a postpositional phrase. Adpositional phrases are under certain conditions analyzed as "oblique" arguments (Van Valin and LaPoUa 1997) or "complements" (Heine 1989). According to Van Valin and LaPolla (1997: 29) this is the case, for instance, if "the NPs in these PPs are represented in the semantic representation" of a verb (as the to-phrase with English give), and if these adpositional phrases can be promoted to argument positions (English John presented the award to Mary vs. John presented Mary with the award). In general, a combination of semantic and syntactic criteria is given for the classification of an NP as an argument or an adjunct:

i. Arguments are required by the verb in order to form a complete clause, while adjuncts are not;

ii. Arguments, but not adjuncts can undergo certain syntactic operations, for example, relativization or passivization;

iii. Arguments tend to be marked by case; adjuncts tend to be realized in adpositional phrases;

iv. Finally, arguments are said to express semantically necessary participants of an event, while adjuncts don't. Accordingly, arguments are more likely to refer to objects or persons, and adjuncts preferably refer to time, space, instruments, etc. (7)

These criteria are not only problematic in their applicability, but also fail to produce a clear-cut distinction between oblique arguments and adjuncts in Jalonke:

i. Optionality of "oblique arguments". Candidates for "oblique arguments" are encoded in postpositional phrases. These PPs can be freely ellipsed, as indicated through the brackets in the following two examples. (8)
(5) N bik -[??][??] [??]efu (Mariama ma).
 1AG pen -DEF lend (Mariama at)
 'I lent a pen (to Mariama).'

(6) N Mariama [??]efu (biku -na 'a ).
 1SG Mariama lend (pen) -DEF with
 'I lent Mariama a pen (lit. I lent Mariama with a pen).'


ii. A vailability of syntactic operations such as relativization and passivization. Syntactic tests reveal no differences between oblique arguments and adjuncts either: only direct objects can be passivized; but on the other hand, there are no constraints on relativization in terms of grammatical relations (see Section 2.2.2 for a detailed discussion).

iii. Differences in case marking between arguments and adjuncts. Adjuncts as well as participants that might be subcategorized by a verb can both occur in postpositional phrases, sometimes even in postpositional phrases headed by the same postposition, as in the following two examples. In (7), one would like to classify the PP as an adjunct, since soap is nota necessary participant in a washing event. This does not follow from the syntactic encoding alone, however, but also from real-world knowledge--while we know that is possible to be washed without soap, the adjunct character of the PP would be more questionable if it was 'with water', since most of us would regard water as indispensable for washing. In analogy the participant encoded in the PP in (8) is more likely to encode a participant semantically entailed by the event that the verb denotes, since an act of presenting necessarily contains a present:
(7) A dii -na ma- xaa (saafu -na 'a).
 3SG child -DEF DISTR- wash soap -DEF with
 'She washed the child (with soap).'

(8) Aissatu n samba -xi (bireeti -na 'a).
 Aissatou 1SG present -PF (bread -DEF with)
 'Aissatu has presented me (with bread).'
 (Ataya 180)


The PP in (8) cannot be promoted to direct object in the case of samba 'present', seemingly confirming the adjunct properties of the postpositionally marked participant. Nevertheless, for some verbs, such a promotion is possible, as shown in (9) and (10) below. The encoding in a PP alone therefore is a bad predictor for adjunct status. To finish, grammatical relations in Jalonke treat subjects and objects differently from all postpositional phrases, regardless of a possible argument-adjunct distinction within them.

iv. Semantic necessity of a participant. As already hinted at above, semantic factors such as whether a PP encodes a participant entailed by the event that the verb denotes, or a subcategorized adjunct in other terms, are crosslinguistically bad predictors of argumenthood. In Moseten (Sakel 2003) for instance, there are a transitive and an intransitive verb of eating. The verb 'look' in Arrernte (Wilkins in press) takes three arguments, just like the verb 'put', one of them corresponding to the location of the Theme. Across languages, many more verbs very that are close in meaning or seeming translation equivalents exhibit differences in the number and thematic roles of arguments they take. In view of this variation in linguistic encoding, it is clear that only syntactic evidence, not hypotheses on semantically necessary participants, can elucidate whether a participant is an argument or an adjunct of a given verb. Syntactic evidence does not result in a differentiation between oblique arguments and adjuncts for Jalonke.

This being said, there is nevertheless some evidence that some of the Jalonke verbs optionally taking a third participant take subcategorized adjuncts or "oblique arguments" (Van Valin and LaPolla 1997). As demonstrated above and spelled out in more detail below, the third participant of these verbs is syntactically an adjunct. However, for some verbs, in an alternative syntactic realization, this postpositionally marked participant can be linked to direct object, hence become a core argument. This possibility of alternation is illustrated through the two different realizations for the verb doni 'lend' in the following two examples:
(9) I a doni -m' an ma nde!
 2SG 3SG lend -IPFV 1SG at INACT
 'You will lend it to me!'
 (Pilgrim-Sall 011)

(10) N ma[??]i i doni ra n ma sii -na ma.
 1SG can 2SG lend with 1SG POSS goat -DEF at
 'I can lend you my goat.'
 (Nga 223)


This behavior favors an analysis of the alternating verbs in question as having three core participants--peripheral participants or adjuncts crosslinguistically cannot be advanced to a direct argument (Van Valin and LaPolla 1997). Thus, in English, the picture, which is an adjunct in (11), would qualify as an argument, because it can also be realized as the direct object of English present in (12). Xenos in (13), in contrast would not count as an argument, because it cannot be the direct object of English buy.

(11) I presented Loretta with a picture.

(12) I presented a picture to Loretta.

(13) I bought the picture frame at Xenos.

Not all verbs that might be said to involve three participants in Jalonke allow such alternating syntactic realizations, however -fii 'give' (cf. [1]) occurs only in one pattern, unlike its English translational equivalent give. Nevertheless, the non-alternating verbs in question do not all exhibit identical linking patterns. Rather, non-alternating three-place predicates realize one of the patterns attested for alternating verbs. The postpositionally marked participants of the non-alternating verbs are marked by those postpositions that also appear in third-participant marking of the alternating verbs, supporting the view that their participants involve identical thematic roles. Non-alternating verbs thus exhibit some similarity to alternating verbs in that the alternation in which the latter appear offers some evidence for the verbs having three arguments. Since it cannot be excluded that the verbs not attested in alternations so far can participate in them under specific circumstances, non-alternating and alternating optionally three-place verbs are treated on a par here. The scarcity of these verbs with all their participants syntactically expressed in the Jalonke corpus compiled so far combined with the difficulty of controlling and varying potentially relevant features for an alternation such as heaviness of the NP, animacy of the relevant participant, etc., in elicitations means that for the time being no commitment with respect to the argumenthood of the participant encoded in a PP can be made.

Overall, however, the syntactic adjunct character of the third participant distinguishes Jalonke from other languages or constructions within languages that employ the "double-object" or "direct" strategy (Margetts and Austin, this volume). In the "double object" or "direct strategy", the Effector, the Theme and the animate Goal (or Recipient) of a three-participant event are encoded like direct arguments (I gave Jurgen the paper). Other languages/constructions make use of the "indirect object" or "dative strategy" (Austin et al. in prep.; Blake 2001). In this strategy, the animate Goal is case-marked differently from direct objects and adjuncts (German Ich gab ihr einen Blumenstrauss 'I gave her a bunch of flowers' (lit. I-NOM gave her-DAT a bunch of flowers-ACC). The use of postpositions for the marking of the third participant that moreover have other functions than marking Goals reveals that Jalonke does not employ the indirect-object strategy. Jalonke adopts a third mechanism, termed the "oblique strategy" by Austin, Evans and Margetts, and marks the third participant with an adposition (I gave the paper to Jurgen). Most verbs in Jalonke may be said to instantiate the "locative" subtype of the "oblique strategy", selecting for a locative (or originally locative) postposition to mark the postpositionally marked participant coding the animate Source or Goal. Two groups of verbs in Jalonke use other variants of the "oblique strategy". Three verbs employ the "benefactive strategy" (Newman 1998; Margetts and "Austin, this issue) exclusively to encode transfer, marking the "target" participant of a transfer like a Beneficiary. For all other verbs of transfer, the Beneficiary is encoded differently from Source or Goal, the former being expressed by a PP headed by be 'for', the latter by ma 'at'. Other verbs follow exclusively or alternatively the "instrumental strategy", also crosslinguistically common according to Newman (1998) and Margetts and Austin (this issue). These verbs treat the "target" of a transfer as the Theme, affected by the "good" that is encoded as the Instrument of the transfer in a PP headed by ra 'with'.

Thus, in Jalonke, as is crosslinguistically the norm, the order of the second and third participant is the most salient parameter of variation. Since this order can differ regardless of the syntactic status of the third participant, in the following, I use labels that specify the thematic role of the second and third participants and their linear order. Thus, I use "Theme-Source/Goal strategy" to refer to those verbs of the "oblique strategy" that realize the Source/Goal as the postpositionally marked participant. Accordingly, I employ the labels "Theme-Beneficiary strategy" instead of "benefactive strategy" and "Theme-Instrument strategy" instead of "instrumental strategy". The correspondences between the labels proposed by Margetts and Austin (this issue) and my labels as well as brief characterizations of the strategies are given in Table 5.

2.2.2. Pivot properties of the participants. Passivization and relativization patterns are relevant to this second parameter that distinguishes languages and constructions according to the syntactic properties, or "pivot properties" (Van Valin and LaPolla 1997: 275ff.) of the arguments of three-participant verbs. Languages can, according to Bresnan and Moshi (1990) and Margetts and Austin (this issue), be symmetrical, that is, treat animate Goal and Theme in the same way with respect to control properties in the syntactic sense, as regards relativization, questioning, etc.. Languages can also differentiate in the morphosyntactic properties between animate Goal and Theme; then, they are asymmetrical. With respect to passivization, Jalonke three-place verbs behave like transitive verbs: whatever its thematic role, only the argument that is the direct object can be promoted to subject position. This promotion is barred for postpositionally marked participants, regardless of the thematic role they bear to the verb. Thus, Jalonke treats direct objects and postpositionally marked participants as formally different, regardless of their thematic roles. The following sets of active-passive pairs illustrate this asymmetry. They all feature the verb muga 'steal' for which either the Theme or the Source of the theft can be linked to direct object. In (14), the Theme is realized as the direct object, and the Source as the postpositionally marked participant. The only possible passive alternation promotes the direct object, in this case the Theme, to subject (15).
(14) E m[??]ntur -na mug a Maimuna ma.
 3PL watch -DEF steal Maimuna at
 S O PP
 'They stole a watch from Maimuna.'

(15) M[??]ntur -na muga Maimuna ma.
 watch -DEF steal Maimuna at
 S PP
 'A watch was stolen from Maimuna.'


In (16), muga 'steal' appears with the Source in direct object position, and the Theme realized as the postpositionally marked participant. Again, the participant occupying the direct object slot is promoted to subject in the passive counterpart in (17), although this time, its thematic role is that of the Source and not of the Theme of the transfer.
(16) E Maimuna muga m[??]ntur -na ma.
 3PL Maimuna steal watch -DEF at
 S O PP
 'They stole a watch from Maimuna (lit. They stole Maimuna at a
 watch).'

(17) Maimuna muga m[??]ntur -na ma.
 Maimuna steal watch -DEF at
 S PP
 'Maimuna was stolen a watch.'


Relativization in contrast, reveals no differences in pivot properties between arguments and postpositionally marked participants. In Jalonke, the relative pronoun naaxan (PL naaxee) appears in initial position of a corelative clause. The subordinate clause containing the relative marker is always the first clause, followed by the main clause. All arguments and postpositionally marked participants in a sentence can be relativized. (18) gives an example of the relativized NP being the subject of an intransitive relative clause; (19) illustrates the relativized NP as the subject of a transitive relative clause. (20) features an NP as the direct object of the relative clause, and (21) presents an postpositionally marked participant as relativized NP. Correlative clauses in Jalonke are often headless, as in (20) and (21).
(18) Mux -[??][??] naaxan fow n[??]n faa -ma dii -ra-
 person -DEF REL all DISC come -IPFV child CAUS
 -mini i
 exit at
 'The people who are all coming to a baptism,
 e taam -[??][??] s[??]t [??][??] nde.
 3PL taami -DEF find -IPFV INACT
 they will get taamina (a special bread).'
 (Diiram 054)

(19) N naaxan a fala -m' i b[??] j[??][??],
 1SG REL 3SG speak -IPFV 2SG for PART
 'I, who am talking to you now,
 n saa -xi saar -[??][??] ma.
 1SG lie -PF bed -DEF at
 I am lying on the bed.'
 (Jigijanna 057)

(20) Tumaani xucaa -na naaxan [??]in -xi, n
 Tuumani younger sibling -DEF REL cook -PF 1SG
 a don.
 3SG eat
 'Tumaani's younger sister, what she had cooked, I ate it.'
 (Labe 078)

(21) N[??][??]n[??] Siree on samba -xi naaxan na, i a
 mother Siree 1PL.I present -PF REL with 2SG 3SG
 xani.
 send
 'What mother Siree presented us with, you sent it.'
 (Labe 501)


2.2.3. Linking in comparison to transitive verbs. While all nominative-accusative languages map the Effector onto the subject slot, some languages treat the Theme of a three-participant verb parallel to the Theme of a transitive verb and single out the Source/Goal. These languages are called "direct object vs. indirect object" languages by Margetts and Austin (this issue). Others have the Source/Goal of a three-participant verb in the same syntactic slot as the Theme of a transitive verb, treating differently the Theme of a three-participant verb (or "notional direct object" in Dryer's [1986] terms). These languages distinguish between "primary" and "secondary" objects (Dryer 1986) rather than between "direct" and "indirect ones". (9) The situation is further complicated by the fact that languages can adopt more than one strategy. With respect to "direct-indirect" vs. "primary-secondary" alignment, this is also the case of Jalonke, as illustrated by the different alignment possibilities for muga 'steal' in (14) and (16) above. Since Jalonke has neither indirect nor secondary objects, but realizes both as postpositionally marked participants, for terminological clarity I use the terms "Theme-Source/Goal-strategy" instead of "direct-indirect strategy", and "Source/Goal-Theme-strategy" instead of "primary-secondary strategy", referring once more to the linear order of the second and third participants. Table 6 summarizes which of the crosslinguistically attested strategies are relevant for Jalonke, with the strategies occurring in the language given in italics.

The linking patterns of Jalonke and their variations are introduced in detail in Section 4.

3. Postpositions marking the postpositionally marked participant of verbs with optionally three participants in Jalonke

Three postpositions are involved in the marking the third participant of three-place verbs. Two of these postpositions--ma 'at' and ra 'with'--are, as the spatial postpositions of Jalonke and of Central Mande languages in general (see Lupke 2005; Trobs 1993, 1998), not specialized for the marking of Location, Path, Source, or Goal. Directional meaning components are lexicalized in verbs, not postpositions. B[??] 'for' is the only postposition that is not attested with spatial meanings--it exclusively marks Beneficiaries and occurs only with some verbs of transfer of information to mark the animate Goal/Beneficiary of a transfer of information. In the following, I will specify which specific Path relation, although determined by the directional semantics of the verb, not by the postposition, is marked, namely, either Source or Goal. Moreover, I will avoid the term Recipient, a convenient role label for languages that make grammatical distinctions sensitive to the animacy vs. inanimacy of the Goal AND sensitive to the coding of Sources vs. Goals (e.g. English to vs. from). Recipient is a misleading label for languages that make an animacy distinction but mark Sources and Goals indistinctly by the same adposition. This is the case of Jalonke, as examples (22) and (23) show, in which ma 'at' encodes the animate Source of sara 'buy', but the animate Goal of matii 'sell' respectively.
(22) N gatoo -na sara Hawa ma.
 1SG cookie -DEF buy Hawa at
 'I bought a cookie from Hawa.'

(23) A t[??]x -[??][??] matii Aissatu ma.
 3SG chicken -DEF sell Aissatu at
 'She solda chicken to Aissatu.'


With these facts in mind, let us now turn to the three postpositions of relevance here.

3.1. Ma marking location

Ma is a spatial postposition with highly general semantics. Its semantic schematicity probably goes back to the lexical noun it originated from, most likely a noun meaning 'place, location'. A cognate of Jalonke ma, ma 'place', is still attested in Bambara (Bailleul 1996). As a postposition (10) ma indicates relations that correspond best to English at, although it does not signal a specific topological relation of a Figure with respect to a Ground, unlike English at. Synchronically, Jalonke ma is most probably simply a generic Ground-denoting postposition, used as a default whenever no specific Figure-Ground relation is described.

The extensions of ma include the encoding of certain abstract types of predicative possession in nonverbal predications construed as located in/ at the possessor (24); location at a place with non-motion verbs (25); movement towards a Goal with certain motion verbs as in (26); and movement away from a Source for other motion verbs (27).
(24) Kaam[??] - na nxo ma!
 hunger -DEF 1PL.E at
 'We are hungry (lit. Hunger at us)!'
 (Alpha2-157)

(25) A wal -[??][??] tand -[??][??] maa.
 3SG work -IPFV courtyard -DEF at
 'He is working in the courtyard.'

(26) I sig -aa x[??][??] -[??][??] ma.
 2SG go -IPFV stranger -DEF at
 'You are going to the stranger.'
 (Mburee 097)

(27) On fan xa keli burun -na ma.
 IPI.I also SUBJ leave bush -DEF at
 'We, too, should leave (lit. from) the bush.'
 (Kiridina 234)



The uses of ma most relevant for the present study are not those where it marks Ground-denoting phrases in adverbial function as in (24)-(27), but those where it marks the third participant of an event. To anticipate the detailed discussion of the different relations that ma marks in these cases, let us look at events of change of possession and consider which participants are encoded by ma. In analogy to its function as a head of Ground-denoting adjuncts in descriptions of locative relations and motion events, the postposition here simply functions as the indicator of a generic relation. As for the motion verbs above, it is left to the verb to specify the nature and directionality of the change of possession event, for which I provisorily assume the same frame-semantic roles as for 'give'. Thus, in (28) ma serves to code the "giver", in (29) to code the "receiver" of a change of possession as the postpositionally marked participant.
(28) N wulu sul -la roni -xi n baaba
 1SG thousand five -DEF inherit -PFV ISG father
 ma.
 at
 'I have inherited five thousand (Guinean Francs) from my father.'

(29) E faa bande -na 'a, e a fii nxo ma.
 3PL come food -DEF P 3PL 3SG give 1PL.E at
 'They carne with food, they gave it to us.'
 (Alpha 018)


Ma also occurs to mark the postpositionally marked participant of some verbs of transfer of information. Thus in (30), it codes the "information", while the "asker" surfaces as the subject and the person "asked" as the object:
(30) N e maxorin e a telefon- nymeroo -na
 1SG 2PL ask 3PL POSS phone number -DEF
 ma.
 at
 'I asked them for their phone number.'
 (lettre 2-009)


For other verbs of transfer of information, either the person "asked" or the "information" can appear in postpositionally marked position, as in the case of makula 'ask, beg' in (31) and (32):
(31) O ji xax -aa makula n
 2P1 DEM.PROX favor -DEF ask 1SG
 xunjaa -na ma.
 younger.brother -DEF at
 'You asked this favor from my younger brother.'
 (Kiridina 111)

(32) Nxo t[??n o makula -xi naaxan ma ...
 1PL.E EMPH 2P1 ask -PFV REL at
 'What we have asked you for ...'
 (Xoro 045)


Given that ma probably goes back to a locative noun ma, the finding that it not only marks Grounds in events of location and change of location, but also participants of events of change of possession and of transfer of information is best explained in terms of metaphorical extension. Ma started out as a Ground-denoting marker. Since it was vague with respect to encoding Location, Source or Goal, the specification of the presence and direction of the change of location was contributed by the verb. Ina first metaphorical extension, change of possession was construed as similar to change of location, and ma marked "givers" and "receivers" of these events, although for a change of possession, no necessary change of location is entailed. Ina second step of metaphorical extension, ma encoded the "person asked" or the "information" of some communication events, treating them in analogy to change of possession. For all three types of events, it can be argued that they entail at least a metaphorical transfer of an entity from a Source to a Goal, applying a metaphor I will henceforth label "transfer-metaphor". In order to capture the common semantic frames of verbs following this metaphor, from now on I use the more general frame-semantic roles of "transferrer" for the entity that initiates the transfer. "Good" is used for the entity undergoing the transfer and "target" for the endpoint of the transfer, whenever no concrete frame-semantic roles for individual verbs or groups of verbs are referred to. A systematization showing the event types instantiating the "transfer metaphor" present with ma, their domain-specific frame-semantic roles and the generalization across these roles in bold face is presented in Table 7.

More difficult to assess is the use of ma to mark the postpositionally marked participants of certain verbs of psychological states. Here, it can code either the "affectee" (33) or the "affect" (34) of that state (borrowing terms used by Jackendoff (1990: 140) to describe the conceptual participants of psych-verbs):
(33) Xaran -na 'afan an ma.
 study -DEF be pleasant 1SG at
 'Studying is pleasant to me.'

(34) K[??]n[??] maa, n xa tin a ma ...
 but, DISC 1SG SUBJ agree 3SG at
 'But, well, I should agree to it ...'
 (Ataya 1186)


On the basis of the available evidence, it is difficult to judge whether these verbs are construed as the ultimate extension of the "transfer-metaphor", or whether the verbs denote a stative relation rather than having a deictic component. It is thus questionable which thematic roles the participants of these events are linked to. Moreover, verbs of psychological state and related semantic domains allow a wide variation with respect to the post-position heading the postpositionally marked participant phrase. For these reasons, verbs of psychological state are not treated further in this study.

3.2. Ra marking Instruments, Comitatives, and Manner

Ra 'with' is the second postposition involved in the marking of the third participant of verbs. As a postposition, (11) ra marks class inclusion (35) and equation (36) in nonverbal predications. In predications optionally containing a verbal predicate it expresses certain kinds of locative relations, covering mostly those that can also be construed as part-whole relations, like the tree and its leaves in (37). In verbal predications, the functions of ra cover Comitative (38), Instrument (39), and Manner (40).
(35) Gin[??] nan a ra.
 woman FOC 3SG with
 'She is a woman.'
 (Jiba 016)

(36) Manga -na a ra, [??] [??]!
 king -DEF 3SG with DISC
 'He is the king, yeh!'
 (Jigijanna 153)

(37) Bur[??]x[??] -nee mango -bil -la 'a.
 leaf -DEF.PL mango -tree -DEF with
 'The leaves are on the mango tree.'

(38) N faa ninge -nee ra.
 1SG come cow -DEF.PL with
 'I came with the cattle. (= I brought the cattle.)'
 (Ibrahima1-004)

(39) A lut -[??][??] i- bolon siizoo -nee ra.
 3SG rope -DEF IT cut scissor -DEF.PL with
 'He cut the rope with the scissors.'
 (CutandBreak-Alpha 024)

(40) A goro -ma a firifiri ra.
 3SG descend -IPFV 3SG spin with
 'He is descending spinning.'
 (Tomatoman-M 014)


The merger of Comitative, Instrument and Manner into one "case marker" makes Jalonke one further language that exhibits the widely attested, but not universal, syncretism of these roles (Croft 1991; Heine et al. 1991; Stolz 1996; among others). (12) Here, I am mainly concerned with the use of ra 'with' to mark the postpositionally marked participant of a three-place predicate as the Instrument of a transfer. Ra 'with' is marginally used alternatively to ma 'at' to mark the "target" of a transfer in the widest sense (41), but in most cases encodes its "good", treating it as its Instrument (42).
(41) Nxo banta fi fita Maimuna ra.
 1PL.E PF DEM.PROX show Maimuna with
 'We have (already) shown this to Maimuna.'
 (Diinaxu 005)

(42) Aissatu n samba -xi bireeti -na 'a.
 Aissatu 1SG present.with -PF bread -DEF with
 'Aissatu has presented me with a bread.'
 (Ataya 180)


3.3. B[??] marking Beneficiaries

B[??], the postposition encoding Beneficiaries, is more specialized than the other two postpositions examined so far. B[??] is used in verbless predications to mark predicative possession (43):
(43) Xuli m' aa be.
 tail NEG 3SG for
 'He (the chimpanzee) has no tail.'
 (Deemu 017)


In combination with verbal predicates, b[??] 'for' covers actions on somebody's behalf or for somebody's benefit or actions of intended transfer (to somebody's benefit). As such, its extension is roughly comparable to that of for in English (Barbara baked a cake for me):
(44) E band -[??][??] [??]in -ma nxo b[??] bui!
 3PL food -DEF cook -IPFV 1PL.E for DISC
 'They are cooking food for us, woah!'
 (Alpha 016)

(45) Nxo dubaa -xi i be.
 1PL.E say benedictions -PF 1SG for
 'We have said benedictions for you.'
 (Farewell1 001)


In combination with a verb encoding motion or direction, b[??] cannot be used to encode the Source or Goal (with the exception of three verbs of transfer of information, discussed in Section 4.1).

Table 8 gives an overview of the main extensions of the three postpositions in nonverbal and verbal predications.

To summarize, Jalonke employs three postpositions to mark the third participant of three-place verbs. Ma 'at' codes either the Source or the Goal of a change of location, depending on the directional semantics of the verb governing it. It is likely that the development of ma 'at' in marking the animate Goals and Sources of change-of-possession-events is due to a metaphorical extension. In this extension, changes of possession are treated like changes of location (even though, as in the case of roni 'inherit', they do not always entail change of location). A further metaphorical extension most plausibly underlies the use of ma 'at' with verbs of communication--again, these events seem to be construed metaphorically like caused changes of location of information or ideas. Ra 'with' marks the participants of certain intransitive verbs of psychological state and the "goods" of certain transitive verbs of transfer, most plausibly treating the former as Comitatives and the latter as Instruments of a transfer. B[??] 'for' is limited to select for Beneficiaries, with the exception of three verbs of transfer of information. The third participant of these verbs, which do not distinguish Goals and Beneficiaries, is marked by be 'for'. It is probable that the earlier extension of ma 'at' to cover situations entailing transfer has so far preempted the often attested development of a benefactive marker into a dative or marker of all "indirect objects" (Lehmann 1982b; Heine and Reh 1984; among others) in Jalonke. It has to be noted, however, that the verbs of transfer of information that encode their third participant in a PP headed by b[??] 'for', might constitute a bridging context in the grammaticalization of benefactives into datives. Such a development is crosslinguistically common (Heine and Kuteva 2002). It is possible that the occurrence of b[??] 'for' in these contexts reflects an ongoing grammaticalization process in the course of which it is taking over some of the functions of ma 'at'.

4. Linking patterns for verbs with optionally three participants in Jalonke

The following section looks at the ways in which the postpositions introduced in the previous paragraphs mark the third participant of verbs with optionally three participants in terms of variations in lexical and syntactic profiling. Sections 4.1, 4.2, and 4.3 group verbs according to their lexical profiling or the different association between frame-semantic and thematic roles. Section 4.4 explores the verbs that exhibit variation as to their syntactic profiling, or the linking variations from thematic roles to grammatical relations. Possible explanations for variations in lexical and syntactic profiling in terms of information structure and topicality are discussed.

4.1. Verbs employing the Theme-Source/ Goal strategy

The "Theme-Source/Goal strategy" treats the Theme of an optionally three-participant verb--(47) and (48)--in the same way as the Theme of a two-participant verb (46):
(46) E ning -[??][??] faxa.
 3PL cow -DEF kill
 Effector Theme
 S O
 'They killed a cow.'
 (Jiba 142)

(47) ... a xa ninge -nee fii n ma.
 3SG SUBJ cow -DEF.PL give 1SG at
 Effector Theme animate Goal
 S O PP
 'They should give me a cow.'
 (Mburee 187)

(48) Maa, e n xani Labe i, kasoo -n' i.
 DISC 3PL 1SG send Labe at prison -DEF at
 Effector Theme Goal
 S O PP
 'Well, they sent me to Labe, to prison.'
 (Alpha 2 166)


The Source- and Goal-oriented verbs of transfer sara 'buy (from)' (cf. [22] above) and matii 'sell (to)' (cf. [23] above) follow the "Theme-Source/ Goal strategy" exclusively. Among the verbs that adopt this strategy as the only one are, however, a number of verbs that use a different postposition to mark the Source or Goal of the transfer. Thus, verbs that denote the transfer of a message or information, jaabaa 'explain' and fala 'say, speak, tell' only take the postposition b[??] 'for' (otherwise used to mark Beneficiaries only) to express Beneficiary and/or animate Goal:
(49) O. a fala manga -nee b[??].
 2P1 3SG speak king -DEF.P1 for
 'You tell (it) to your kings
 o faa ji sunkutun -na x[??]n.
 3SG come DEM.PROX girl -DEF at
 that we came because of this girl.'
 (Kiridiina 069)

(50) A jaabaa n b[??]!
 3SG explain 1SG for
 'Explain it to/for me!'


Other verbs of transfer, such as s[??]b[??] 'write' or sara 'buy', differentiate between the marking of Beneficiaries and animate Goals or Sources. In (51) the participant encoded in the b[??]-phrase is a Beneficiary, and in (52) it is the animate Goal of writing a letter. In (53) the participant encoded in the b[??]-phrase can only be the Beneficiary, not the Source of buying, whereas in (54) it can only be the Source of the verb action.
(51) N l[??]t[??]r -na s[??]b[??] n faafa b[??].
 1SG letter -DEF write 1SG elder brother for
 'I wrote a letter for my elder brother (entails that I did it for
 his benefit or on his behalf, e.g., because he is illiterate).'

(52) N l[??]t[??]r -na s[??]b[??] n faafa ma.
 1SG letter -DEF write 1SG elder brother at
 'I wrote a letter to my elder brother.'

(53) Siga den, i ser -[??][??] sara ji
 go DISC 2SG medicament -DEF buy DEM.PROX
 b[??].
 for
 'Go, you buy medication for this one!'
 (Ibrahimal 147)

(54) N gatoo -na sara Hawa ma.
 1SG biscuit -DEF buy Hawa at
 'I bought biscuits from Hawa.'


Another verb of transfer of information, k[??]j[??]k[??]j[??] 'whisper', discussed further in Section 4.4 below, also appears in the "Theme-Source/Goal strategy" when its third participant appears in a PP headed by b[??] 'for'. Many verbs distinguish between animate Goal and Beneficiary through the use of different postpositions. It is probable that verbs of transfer of information behave differently, because in the case of information transfer, the two situation types never contrast in Jalonke--the animate Goal may be viewed as the Beneficiary and vice versa. In contrast to the other transfer verbs, they are not used to describe actions performed on behalf of or for the benefit of a third person that is not the animate Goal. (13)

The Jalonke verbs explored so far are specialized for transfer to or from human entities, roughly corresponding to situations of transfer to animate Goals expressed by the "double-object" construction of English (I sent Zara the quotes). In contrast to English, though, these verbs cannot appear in an equivalent to the English Dative Alternation (Loretta sent the application to London) used for inanimate Goals. (14) Rather, two different verbs, xani 'send' and x[??][??] 'send, commission' exclusively designate transfer to inanimate Goals. Thus, unlike in English, verbs and not constructions are specialized for the different kinds of transfer to animate vs. inanimate Goals. The locative postposition used with the two verbs x[??][??] 'send, commission' and xani 'send', varies according to the type of Goal referent--i 'at' in the case of toponyms (55), ta 'with' (56) in the case of body parts as Goals, and ma 'at' (57) with most other types of referents functioning as inanimate Goal.
(55) E e tongo, e e xani Kute i.
 3Pl 3Pl take 3Pl 3Pl send Kute at
 'They took him, they sent him to Kute.'
 (Jalonke 031)

(56) Alia i a wal -[??][??] xani jaare -na 'a!
 Allah 2SG POSS work -DEF send front -DEF with
 'May Allah make progress your work (lit. sent it to the front).'
 (lettre-1 043)

(57) A a tongo, a a xani wulaa -na ma
 3SG 3SG take 3SG 3SG send forest -DEF at
 'They took him, they sent him to the forest.'
 (Kweelenna 051)


Two verbs of transfer, sara 'buy' and matii 'sell' entail the exchange of money in their frame semantics. They can have the "price" of that exchange as a zero-marked third participant (58), but Goal/Source ANO price never occur in one clause.
(58) A band -[??][??] sara k[??]m[??] naani.
 3SG food -DEF buy hundred four
 'She bought food for four hundred (Guinean Francs).'
 (Labe 0279)


Thus, Jalonke follows the crosslinguistic generalization (Margetts and Austin, this issue) that a maximum of three overt arguments may be expressed by a simplex verb in a single clause, even though more participants might said to be present semantically.

Table 9 gives an inventory of all the Jalonke verbs exclusively attested in the Theme-Source/Goal strategy so far.

4.2. Verbs employing the Source/Goal-Theme strategy

In the "Source/Goal-Theme strategy", the grammatical relation of the Theme in the three-place predicate (60), is different from the one of the Theme of a transitive verb (59):
(59) E sii -na bura -a.
 3PL goat -DEF skin -IPFV
 Effector Theme
 S O
 'They are skinning a goat.'
 (Baptism 007)

(60) N e maxorin [??] [??] telefon- nymeroo
 1SG 3PL ask 3PL POSS phone -number
 Effector Goal Theme
 S O PP
 -na ma.
 -DEF at
 'I asked them for their phone number.'
 (lettre 2-009)


Transfer verbs employing only the "Source/Goal-Theme strategy" seem to lexically profile the animate Goal/Source as more affected or salient than the Theme, hence its linking to direct object. Note that an alternative approach to this linking strategy would be to postulate a difference in thematic roles that results in a syntactic alignment different from that of transitive verbs and from verbs following the Theme-Source/Goal strategy. Thus, one might want to consider the argument linked to object as the Theme, not as the animate Source/Goal (or Recipient). It is unclear, however, what the thematic role of the third participant would be on that account--ma 'at' does not mark Instruments in Jalonke, so unless one wants to stipulate idiosyncratic Instrument marking for a small group of verbs, this possibility is ruled out. Since a well accepted way to analyze this strategy is to admit that the participant structure of these verbs involves the thematic roles of Theme and animate Goal/Source just as for Theme-Source/Goal verbs, but that their linking properties differ (Dryer 1986), I follow this analysis.

Table 10 lists the verbs employing the "Source/Goal-Theme strategy" as the only one in the Jalonke corpus.

4.3. Verbs employing the Theme-Instrument strategy

The third strategy attested for Jalonke associates the "target" of a transfer with the Theme, and the "good" with the Instrument affecting the Theme. This strategy resembles the "Source/Goal-Theme strategy" with respect to the frame-semantic role that is linked to object: it is the "target". Nevertheless, the "Theme-Instrument strategy" and the "Source/ Goal-Theme strategy" differ in that the former exhibits consistency in syntactic profiling across two-place (61) and three-place (62) predicates.
(61) Nxo band -[??][??] cin.
 1PL.E food -DEF cook
 Effector Theme
 S O
 'We cooked food.'
 (Labe 042)

(62) Aissatu n samba -xi bireeti -na 'a.
 Aissatou ISG present -PFV bread -DEF with
 Effector Theme Instrument
 S O PP
 'Aissatu has presented me with a bread.'
 (Ataya 180)


The two transfer verbs of Jalonke that treat the "target" of a transfer as the entity affected by the "good" as the Instrument of that transfer, that is, as the Theme, have both the sense of 'present somebody with something'. The difference between the two verbs of presenting is that samba, but not kii, necessarily entails a previous displacement of the "giver"--he either came to visit or returned from a trip with a gift. Whether ratuu 'remind', the third verb instantiating this strategy exclusively, also follows the "transfer metaphor" must left be open for the moment. Table 11 lists the three verbs attested only in the Theme-Instrument strategy in Jalonke to date.

4.4. Alternations

If verbs alternate between strategies, they mainly fluctuate between the "Theme-Source/Goal" and the "Source/Goal-Theme" strategies. Examples for the alternations are given for k[??]j[??]k[??]j[??] 'whisper' below. K[??]j[??]k[??]j[??] does not distinguish the marking of Beneficiary/animate Goal in the "Theme-Source/Goal strategy".
(63) N gund -[??][??] k[??]j[??]k[??]j[??] Haamidu b[??].
 1SG secret -DEF whisper Haamidu for
 'I whispered a secret to Haamidu.'

(64) N Haamidu k[??]j[??]k[??]j[??] gund -[??][??] ma.
 1SG Haamidu whisper secret -DEF at
 'I whispered Haamidu a secret (lit. I whispered Haamidu at a
 secret).'


Table 12 gives an inventory of the verbs alternating between the Theme-Source/Goal and the Source/Goal-Theme strategy in Jalonke.

With respect to the linear order of participants, this alternation is equivalent to the English Dative Alternation that has received much linguistic attention. Although it is impossible to examine the wealth of statements on the semantic and pragmatic conditions of the alternation and to test whether they hold for Jalonke, too, some convergences and divergences can be pointed out. While many Jalonke verbs with animate Goals or Sources do not participate in the alternation, no verb with an inanimate Source/Goal does. Further, verbs specialized for transfer to or from inanimate Goals or Source (x[??][??] 'send, commission' and xani 'send') do not appear in the "Source/Goal-Theme strategy" at all. This finding puts Jalonke on a par with English with respect to the animacy restriction on the "double-object" variant of the Dative Alternation. Other observations made for English are not applicable to Jalonke: for instance, Levin and Rappaport Hovav (2003) notice in accordance with Bresnan and Nikitina (ms.) that, where no information structure and heaviness criteria overrule the choice, the alternants are distinguished by semantic properties of the Goal. If the Goal is animate and a kind of possessor, they state, it should enter the "double-object" alternant. If it is a purely spatial Goal, but also if it is not a type of possessor, it should be expressed in the "to-variant" of the alternation only. This finding is not paralleled by Jalonke: verbs such as kanta 'protect from' and kisi 'protect from' instantiate the "Source/Goal-Theme strategy". This strategy is reminiscent of the "double object", not of the prepositional variant of English, although the postpositionally marked participant clearly is not the resultant possessor of the transferred entity.

Only one verb, nefu 'lend', is attested to alternate between the "Theme-Source/Goal" and the "Theme-Instrumen" strategies in Jalonke. This is a noteworthy finding, since, according to Kone (1984: 110f.), the "Theme-Source/Goal" vs. "Theme-Instrument" alternation is the only alternation attested for three-place verbs in Bambara. Its different configurations are exemplified below:
(65) N bik -[??][??] [??]efu Mariama ma.
 1SG pen -DEF lend Mariama at
 'I lent a pen to Mariama.'

(66) N Mariama [??]efu biku -na a.
 1SG Mariama lend pen -DEF with
 'I lent Mariama a pen (lit. I lent Mariama with a pen).'


The "Theme-Instrument strategy" of Jalonke is reminiscent of the English "Locative Alternation" (Jurgen loaded boxes into the car vs. Jurgen loaded the car with boxes). For this alternation, it is most plausible, but remains to be tested, that it reflects a difference in affectedness, the "target" being more holistically affected in the with-variant than in the to-variant (Levin 1993). In English, the alternants have different entailments--the with-alternant entails the locative alternant, but not vice-versa, (Levin and Rappaport Hovav 2003; among others). Whether this difference in entailments also holds for Jalonke is impossible to judge on the basis of the evidence I collected in the field.

Verbs that oscillate between the "Source/Goal-Theme" and "Theme-Instrument" strategies are equally marginal in Jalonke--again, only one verb is known to have this behavior:
(67) A n t[??][??]nin -xi nde mug -aa ma.
 3SG 1SG accuse -PFV INACT theft -DEF at
 'He had accused me of theft.'

(68) A n t[??][??]nin -xi nde muga -na a.
 3SG 1SG accuse -PFV INACT theft -DEF with
 'He had accused me of theft (lit. with theft).'


Table 15 summarizes the main alignment patterns for the participants of three-participant events, limited to the most productive patterns.

While there are a number of verbs that instantiate one strategy exclusively, several verbs appear either in both the "Theme-Source/ Goal" and "Source/Goal-Theme" strategies or in both the "Source/ Goal-Theme" and "Theme-Instrument" strategies. No verb so far is attested with all three strategies. Very plausibly, this limitation in variation is motivated by conflicting semantic and pragmatic principles of information structure:

The DO/IO [direct object/indirect object] distinction follows semantic roles more closely: the DO of either a monotransitive or a ditransitive clause is prototypically a patient/theme, while the I0 is a recipient/beneficiary. The PO/SO [primary object/secondary object] distinction, in contrast, is linked more closely to discourse/pragmatic function. In ditransitive clauses, the IO tends to be more "topical" than the DO, since the IO is generally human and definite, and often 1st of 2nd person, the DO is generally non-human and indefinite, and almost invariably 3rd person. (Dryer 1986: 841)

Verbs that only follow a single strategy lexically either adopt a regular linking pattern OR respond to topicality features. Verbs adopting two strategies can choose between the two features, but the "Source/ Goal-Theme strategy" and the "Theme-Instrument strategy" are similar with respect to having the "notional indirect object" (in Dryer's terms) or "target" (in frame-semantic terms) in direct object position, even though they treat it once as an animate Goal and once as a Theme. Thus, they differ less crucially to each other than in contrast to the "Theme-Source/Goal strategy". The "Source/Goal-Theme" and the "Theme-Instrument" strategies can thus be viewed as two variations of a strategy that functions in terms of topicality. Both are opposed to the "Theme-Source/Goal strategy" that operates in terms of consistency of linking across two- and three-participant events.

It has to be stated, however, that based on my present knowledge of Jalonke, it is impossible to judge whether the alternations listed in this section constitute productive or lexeme-specific operations. This is due to the following reasons: even for well-studied alternations like the English dative alternation, principles of information structure such as "heaviness" of the constituent NPs are known to influence the admissibility of its variants. The influence of these and other criteria on the syntactic realization of three-place verbs is impossible to assess based on elicitation alone, and cannot be estimated from a small, field-based corpus, which is severely restricted in its informativeness through the fact that the verbs in question appear with only two participants in the majority of cases. Therefore, the exact levels of grammatical information relevant for the alternations cannot ultimately be stated here.

5. Conclusion

In this article was shown that the functions most typically associated with dative case crosslinguistically--the marking of "indirect objects" very often having the thematic roles of Source/Goal, and Beneficiary--are not expressed by a single "case marker" in Jalonke. The thematic role of Experiencer, also commonly marked by a dative, was excluded from the scope of the article because of its problematic syntactic and semantic status. A thematic role often not included in the treatment of "dative marking", the Instrument, was additionally considered, because it is an exclusive or alternate way for some verbs to mark the postpositionally marked participant of a three-participant event. By limiting the scope to the coding of the postpositionally marked participant of three-participant events, three postpositions could be identified as relevant. These postpositions differ among them in their degree of specialization: two of them, ma 'at' and ra 'with' cover a wide range of functions, while the third b[??] 'for' exclusively marks Beneficiaries. It was argued that ma 'at' is a generic Ground-denoting postposition with originally locative semantics that was extended to cover the marking of animate Sources and Goals of transfer in the widest sense because of the metaphorical resemblance of these events to caused change of location. In the discussion of ma, it was also shown that Jalonke postpositions do not distinguish between Location, Source and Goal, or locative, ablative and allative functions, because verbs and not postpositions lexicalize path relations. With respect to ra 'with', it was demonstrated that when it marks participants of a transfer-event, the verbs denoting these events lexically or syntactically profile the "target" as the affected entity or Theme of a caused change of state, and realize the "good" as the Instrument to affect the Theme. For b[??] 'for', it was shown that it is only used in those transfer-events where Beneficiary and animate Goal cannot be discriminated. It was further hypothesized that the existente of a separate postposition to mark animate Sources/ Goals has so far blocked the crosslinguistically-common development of a Beneficiary into a dative.

The article additionally explored the different options of Jalonke verbs for the lexical profiling of their semantic frames, identifying two different patterns of associating frame-semantic with thematic roles. These patterns diverge with respect to the thematic roles of the second and third participant, realizing either the "good" of the transfer as the Theme, and the "target" as the animate Goal, or the "target" as the Theme and the "good" as the Instrument. Verbs of communication realize the "good" as the Theme and the "target" asa Beneficiary. However, differences in alignment do not only come about because of differences in lexical profiling. It was demonstrated that verbs also vary with respect to the linking of thematic roles to grammatical relations, or syntactic profiling. The differences between the "Source/Goal-Theme" and "Theme-Source/Goal" strategies ate due to variations along this parameter. The inventory of strategies and possible alternations between them was given, and possible motivations in terms of the semantics of the Goal participant, of syntactic consistency for the marking of grammatical relations across two- and three-participant events, and of information structure considerations were given. Parallels to well-studied English lexicalization and alternation patterns were drawn where possible. It was pointed out that to some extent, predictions made for English regarding semantic constraints on syntactic profiling hold, as the animacy restriction on Goals realized as the direct object, but that not all predictions for English were borne out as applicable to Jalonke. It is not the case, for instance, that those Goals that are not possessors are barred from the direct object position. While there remain many points of interest to be addressed by future research, as for instance the extension and semantics of the Locative Alternation, this article has shown that regardless of grammatical differences, lexical and syntactic profiling of the marking of participants of transfer-events also exhibit some crosslinguistically comparable semantic criteria. A more detailed investigation might reveal more parallels and differences in the semantic motivation underlying lexical and syntactic profiling for the domain of transfer within and across languages.

Received 25 May 2005

Revised version received 14 December 2005

School of Oriental and Asian Studies, University of London

References

Bailleul, Charles (1996). Dictionnaire bambara-francais. Bamako: Editions Donniya.

Blake, Barry J. (2001 [1994]). Case. 2nd ed. Cambridge: Cambridge University Press.

Bresnan, Joan and Moshi, Lioba (1993). Object assymetries in comparative Bantu syntax. In Theoretical Aspects of Bantu Grammar, Sam A. Mchombo (ed.), 47-91. Stanford, CA: CSLI.

--and Nikitina, Tatiana (ms.). On the gradience of the dative alternation. Unpublished manuscript, Stanford University.

Clark, Eve V. and Clark, Herbert H. (1979). When nouns surface as verbs. Language 55(4), 767-811.

Creissels, Denis (2000). Typology. In African Languages: An Introduction, Bernd Heine and Derek Nurse (eds.), 231-258. Cambridge: Cambridge University Press.

Croft, William (1991). Syntactic Categories and Grammatical Relations: The Cognitive Organization of Information. Chicago and London: University of Chicago Press.

Dryer, Matthew S. (1986). Primary objects, secondary objects, and antidative. Language 62(4), 808-845.

Dumestre, Gerard (2003). Grammaire fondamentale du bambara. Paris: Karthala.

Ethnologue (2004). http://www.ethnologue.org.

Fillmore, Charles J. (1977). Topics in lexical semantics. In Current Issues in Linguistic

Theory, Roger W. Cole (ed.), 76-138. Bloomington: Indiana University Press.

--(1982). Frame semantics. In Linguistics in the Morning Calm, The Linguistic Society of Korea (ed.), 111-137. Seoul: Hanshin.

--and Atkins, Beryl T. (1992). Toward a frame-based lexicon: the semantics of RISK and its neighbors. In Frames, Fields, and Contrasts, Adrienne Lehrer and Eva Feder Kittay (eds.), 75-102. Hillsdale, NJ: Lawrence Erlbaum.

--and Kay, Paul (1993). Construction Grammar Coursebook: Course Reader for Linguistics X20. Berkeley: University of California, Berkeley.

Goldberg, Adele E. (1995). A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press.

--(2002). Surface generalizations: An alternative to alternations. Cognitive Linguistics 14(4), 327-356.

Haspelmath, Martin (2001). Argument marking in ditransitive alignment types. Paper presented at the 4th ALT meeting Santa Barbara, 19-22 July, 2001.

Heine, Bernd (1989). Adpositions in African languages. Linguistique Africaine 2, 77-127.

--and Reh, Mechthild (1984). Grammaticalization in African Languages. Hamburg: Buske.

--; Claudi, Ulrike; and Hunnemeyer, Friederike (1991). Grammaticalization: A Conceptual Framework. Chicago and London: University of Chicago Press.

--and Kuteva, Tania (2002). World Lexicon of Grammaticalization. Cambridge: Cambridge University Press.

Jackendoff, Ray (1990). Semantic Structures. Cambridge, MA: MIT Press.

Keenan, Edward L. and Comrie, Bernard (1977). Noun phrase accessibility and universal grammar. In Universal Grammar: 15 Essays, Edward L. Keenan (ed.), 3-45. London: Croom Helm.

Kone, Dramane (1984). Le verbe bambara: essai sur les proprietes syntaxiques et semantiques. Unpublished doctoral dissertation, University of Grenoble.

--(1984a). La combinatoire verbes-postpositions en bambara. Mandenkan 7, 1-14.

Lakoff, George and Johnson, Mark (1980). Metaphors We Live By. Chicago: University of Chicago Press.

Langacker, Ronald W. (1987). Foundations of Cognitive Grammar. Stanford, CA: Stanford University Press.

de Lavergne de Tressan, Michel (1953). Inventaire linguistique de l'Afrique Occidentale Francaise et du Togo. Dakar: IFAN.

Lehmann, Christian (1982). Thoughts on grammaticalization: a programmatic sketch. Arbeiten des Kdlner Universalienprojekts 48.

Levin, Beth (1993). English Verb Classes and Alternation. A Preliminary Investigation. Chicago and London: University of Chicago Press.

--and Rappaport Hovav, Malka (2003). The dative alternation revisited, paper presented at the Workshop on Verb Classes and Alternations, University of Stuttgart, January 2003.

Lupke, Friederike (2005). A grammar of Jalonke argument structure. Unpublished doctoral dissertation, Radboud University, Nijmegen. MPI Series in Psycholinguistics 30.

Masiuk, Nadine (1985). Differents types de circonstants de l'enonce a scheme verbal du Bambara. Mandenkan 10, 33-45.

Newman, John (1998). Recipients and 'give' constructions. In The Dative, Willy van Langendonck and William van Belle (eds.), 1-28. Amsterdam: John Benjamins.

Rappaport, Malka and Levin, Beth (1988). What to do with Theta-roles. In Thematic Relations, Wendy Wilkins (ed.), 7-36. San Diego, CA: Academic Press.

--and Levin, Beth (1998). Building verb meanings. In The Projection of Arguments, Miriam Butt and Wilhelm Geuder (eds.), 97-134. Stanford, CA: CSLI.

Sakel, Jeanette (2003). A grammar of Moseten. Unpublished doctoral dissertation, University of Nijmegen.

Song, Jae Jung (1990). On the rise of causative affixes. Lingua 82, 151-200.

Stolz, Thomas (1996). Some instruments are really good companions--some are not: on syncretism and the typology of instrumentals and comitatives. Theoretical Linguistics 23(1/2), 113-200.

Trobs, Holger (1993). Lokale Strukturen im Manding. Unpublished Master's thesis, University of Cologne.

--(1998). Funktionale Sprachbeschreibung des Jeli (West-Mande). Cologne: Rudiger Koppe.

Van Valin, Robert D. and LaPolla, Randy J. (1997). Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press.

Westermann, Diedrich and Bryan, Margaret A. (1952). The Languages of West Africa. London, New York and Toronto: Oxford University Press.

Wilkins, David (in press). Learning 'put' and 'look' in Arrernte. In Crosslinguistic Perspectives on Argument Structure: Implications on Learnability, Melissa Bowerman and Penelope Brown (eds.). Mahwah, NJ: Erlbaum.

Notes

* The research reported in this article was supported by the Max Planck Institute for Psycholinguistics. It is based on data collected during three field stays in Guinea between 1999 and 2002. I am grateful to all the inhabitants of Saare Kindia and in particular to my main consultants Mammadou Adama Bary, Alpha Bary, Mammadou Balde, Mammadou Aliou Bary and Souleymane 'Koubia' Bary, who have contributed to this study. I am indebted to many people for valuable comments and suggestions, among them Felix Ameka, Jurgen Bohnerneyer, Sonja EisenbeiSS, Birgit Hellwig, Anna Margetts, Loretta O'Connor and an anonymous referee. The members of the Language and Cognition research group at the Max Planck Institute for Psycholinguistics, the audience of the "Dative days" at the Department of African Studies at the University of Cologne in June 2002 where I presented earlier versions of this article, and the audience of the Workshop on the crosslinguistic encoding of three-participant events at the Max Planck Institute for Psycholinguistics in May 2003 equally contributed to shape the ideas developed here. I would like to thank all of them for their input. Obviously, the remaining shortcomings are mine. Correspondence address: Dept. of Linguistics, SOAS, University of London, Thornaugh Street, London WC1H 0XG, United Kingdom. E-mail: Friederike.luepke@soas.ac.uk.

(1.) The Jalonke examples are represented in terms of surface forms as they were uttered, not including for the sake of space the underlying forms in contexts where assimilations occurred. For a description of assimilatory processes, see (Lupke 2005). The following abbreviations are used for interlinearized glosses: 1--1st person; 2--2nd person; 3--3rd person; AUX--auxiliary; CAUS---causative derivational affix; DEF--definite determiner; DEM--demonstrative pronoun; DIM--diminutive derivation; DISC--discourse marker; DISTR---distributive derivational affix; DO--direct object; E--exclusive pronoun; EMPH--emphatic pronoun; FOC--focus marker; I--inclusive pronoun; INACT--marker of inactuality; IO--indirect object; IPFV--imperfective aspect; NEG--negation marker; N--noun; NP--noun phrase; O--object; P--postposition; PAST--past marker; PL--plural; PO--primary object; POSS--marker of alienable possession; PROX--proximal demonstrative; PV--perfect; S--subject; SG--singular; SO--secondary object; SUBJ--subjunctive; TAM--marker of tense, aspect or mood; V--verb; X--adjunct. In addition, the following conventions are used: "-" for affixes; "." for categories encoded by a portmanteau morpheme. Note that definiteness is a rather wide notion in Jalonke, ranging between definite determination and specificity. Where a source is given, the Jalonke examples are from texts; where not, they are elicited examples.

(2.) Note that for Rappaport and Levin (1988), the with variant ofthe Locative Alternation would have two Themes: the one that is linked to object is a Theme of a change of state; the one that is linked to adjunct is a Theme of change of location. For Jackendoff, in contrast, there is no dual assignment of the Theme relation, because he differentiates between the thematic roles of participants undergoing a change of state (Patients) and participants undergoing a change of location (Themes).

(3.) I follow Van Valin and Wilkins (1996) in distinguishing between Agents or participants that act volitionally and controlled in bringing about the eventuality denoted by the verb and Effectors, which merely bring about the eventuality denoted by the verb.

(4.) In order to keep the different levels of lexical semantic representation and grammatical functions separate, I use the following typographic conventions: frame-semantic roles appear between double quotes ("receiver"); thematic roles are given with initial capitals (Theme); and grammatical relations are written in normal font in lower case (subject).

(5.) Goldberg (1995), in her version of Construction Grammar, also uses the notions of 'semantic frames' and 'lexical profiling' in a similar way. The crucial difference to the approach taken here concerns assumptions about the levels at which semantic frames and thematic roles are situated. Goldberg assumes that verbs have participant roles, very reminiscent of semantic frames, and concrete instances of thematic roles, which fuse with thematic roles contributed only at the constructional and not the lexical, level. I follow a 'projectionist' approach (see Rappaport Hovav and Levin 1998 for a comparison of "constructionalist" and "projectionist" approaches to argument structure), which is based on the assumption that thematic roles are lexically specified, a view also shared by Fillmore and Kay's version of Construction Grammar (Fillmore and Kay 1993: 4.15). See Lupke (2005) for a detailed discussion.

(6.) So far, thematic roles have been presupposed but not properly defined. It is a general problem of theories of argument structure to come up with a set of thematic roles that are not merely postulated (see Dowty 1991; Van Valin and Wilkins 1996; inter alia, for a discussion of this problem). It is beyond the scope of this article to propose such an inventory of thematic roles. Rather, inspired by Van Valin and LaPolla (1997: Ch. 3), I postulate a set of thematic roles for Jalonke that is based on their syntactic reflection in default linking and do not commit myself to their crosslinguistic validity or language-internal exhaustiveness. See Section 3 for evidence for the thematic roles proposed arising from the postpositions marking the respective participants in default linking.

(7.) Kone (1984a) introduces a test to distinguish between arguments and adjuncts in Bambara, a language closely related to Jalonke. If it is possible to paraphrase a clause containing a candidate adpositional phrase with two clauses, featuring the verb 'do' (e.g. The women fought in the market vs. The women fought. They did it in the market), he takes this as evidence for the adjunct character of that adpositional phrase. Masiuk (1985) suggests 'extraposition', according to him one of the subtypes of topicalization in Bambara, as an additional test for argumenthood. As with Kone's test, three types of adpositional phrases can be differentiated through this diagnostic: oblique arguments, adjuncts and Goal or Source phrases of verbs of directed motion. Further research is required to determine the applicability of these tests to Jalonke and to compare the results.

(8.) Dumestre (2003: 180) mentions that in Bambara there are very rare cases in which the third participant of a verb, encoded in a PP, cannot be omitted. He gives two examples, containing singa (ma) 'lend (to)' and f[??] (k[??]) 'set up in place of'. For the Jalonke counterpart of 'lend', doni, it can be stated that the third participant can be freely ellipsed, so it seems, at least as far as the scarce data suggest, that the Bambara findings cannot be extended to Jalonke.

(9.) The distinction between direct vs. indirect and primary vs. secondary objects as first introduced by Dryer (1986) is now widely accepted (see Croft 1991). Dryer (1986) and Haspelmath (2001) reserve the terms for languages/constructions that treat the Recipient and Theme of a three-participant on a par with respect to case marking. This criterion corresponds to Margetts' and Austin's symmetry, and thus excludes languages that mark one Recipient or Theme as an oblique. Margetts and Austin use the term 'secondary strategy' in a different sense: they limit this notion to the grammatical relation the Theme of a three-participant bears in comparison to the Theme of a two-participant event. Independently of the case marking, they call it "secondary" if it ends up in a different syntactic slot. Since, regardless of the object properties of the arguments in question, this distinction is important, I follow Margetts and Austin in splitting up the direct-indirect and primary-secondary object strategies into two parameters. One of these parameters contrasts "symmetry" vs. "asymmetry"; the other "direct-indirect" vs. "primary-secondary strategy".

(10.) Ma is one of the most polyfunctional markers of Jalonke. Apart from its extensions as a postposition, it serves the following purposes: as a derivational verbal prefix it marks distributivity. As an inflectional verbal suffix, it expresses imperfective aspect. Finally, ma occurs as the marker signaling alienable possession in attributive possessive construction. It is plausible that ali theses instances of ma are not cases of homophony, bur diachronically related to the putative lexical noun 'place' grammaticalizing into the different target concepts.

(11.) Like ma, ra also functions as a preverb in Jalonke. As such, it serves as a causative derivational prefix probably going back to the instrumental, which is, according to Song (1990), a widely attested source for the rise of causative markers.

(12.) These extensions correspond to those widely attested for comitatives (Heine and Kuteva 2002). Although (going back to Lakoff and Johnson's [1980: 134f.] metaphor ["an instrument is a companion"]) a directional development of instruments into comitatives into manner adjuncts is often assumed, this directionality has not yet been confirmed unequivocally (Heine and Kuteva 2002). Hence, no directed grammaticalization chain is implied by the ordering of the meanings of ra in Jalonke.

(13.) Other verbs marginally occur with postpositions other than ma, ra and be. Matii 'sell' is attested with ma and with x[??]n 'at' to mark the animate Goal. jita 'show' is the only verb employing the postposition ra 'with' without following the "Theme-Instrument strategy"--it is the Theme that turns up as the object, the animate Goal as the adjunct, quite differently from the "Theme-Instrument" verbs introduced in Section 4.3. Since ra and x[??]n as Source/Goal markers for transfer verbs are only attested marginally, I treat these cases as idiosyncratic.

(14.) Although there are morphological and phonological restrictions on the distribution of the dative alternation in English (i.e. verbs of Latinate origin do not allow the double object construction), it is generally agreed (Levin 1993; Goldberg 1995; among others) that along with criteria of information structure, animacy features play a role for the dative alternation: verbs in the double object construction pose an animacy restriction on their Goal phrase.
Table 1. Association of semantic frames, thematic roles, and
grammatical relations for Jalonke 'give' in (1)

 'she' 'milk' 'child'

semantic frame "giver" "gift" "receiver"
thematic role Effector Theme animate Goal
grammatical relation subject object postpositionally
 marked participant

Table 2. Association of semantic frames, thematic roles, and
grammatical relations for Jalonke 'present with' in (2)

 'Tumaani' 'they' 'cassette'

semantic frame "giver" "receiver" "gift"
thematic role Effector Theme Instrument
grammatical relation subject object postpositionally
 marked participant

Table 3. Association of semantic frames, thematic roles, and
grammatical relations for Jalonke 'lend' in (3)

 'you' it, 'me'

semantic frame "giver" "gift" "receiver"
thematic role Effector Theme animate Goal
grammatical relation subject object postpositionally
 marked participant

Table 4. Association of semantic frames, thematic roles, and
grammatical relations for Jalonke 'lend' in (4)

 'I' 'you' 'goat'

semantic frame "giver" "receiver" "gift"
thematic role Effector animate Goal Theme
grammatical relation subject object postpositionally
 marked participant

Table 5. Correspondence between labels and characterization of
strategies

Label used by Labels used for Postposition
Margetts and Jalonke (linear that marks
Austin/Newman order of 2nd and the 3rd
 3rd participant) participant

"Locative strategy" "Theme-Source/ ma
 Goal strategy"

 "Source/Goal- ma
 Theme strategy"

"Benefactive "Theme-Beneficiary be
strategy" strategy"

"Instrumental "Theme-Instrument ra
strategy" strategy"

Label used by Linking of semantic frames
Margetts and to thematic roles and
Austin/Newman grammatical relations

"Locative strategy" "Good" as Theme [right arrow] object
 "Target" as Source/
 Goal [right arrow] 3rd participant

 "Target" as Source/
 Goal [right arrow] object
 "Good" as Theme [right arrow] 3rd
 participant

"Benefactive "Good" as Theme [right arrow] object
strategy" "Target" as Beneficiary [right arrow] 3rd
 participant

"Instrumental "Target" as Theme [right arrow] object
strategy" "Good" as Instrument [right arrow] 3rd
 participant

Table 6. Summary of crosslinguistically attested patterns for three
place verbs, with the patterns attested in Jalonke in italics

Parameters Strategies
Case marking direct dative oblique

 allative locative

Pivot properties symmetrical

Linking patterns parallel to transitive
 (Theme-Source/Goal strategy)

Parameters
Case marking oblique

 ablative benefactive instrumental genitive
 (Theme- (Theme-
 Beneficiary Instrument
 strategy) strategy)

Pivot properties asymmetrical

Linking patterns different to transitive
 (Source/Goal-Theme strategy)

Table 7. Semantic domain of events that instantiate the
"transfer-metaphor" (frame-semantic roles and generalization
over frame-semantic roles)

Caused change of location "mover" "moved" "endpoint"

Change of possession "giver" "gift" "receiver"

Communication "person asked" "information" "asker"
 "teller" "information" "told"

Generalization over the "transferer" "good" "target"
domain of transfer

Table 8. Extensions of the postpositions ma, ra, and b[??]

Post- Nonverbal
position predication Verbal predication

 Nondirectional verb Directional verb

ma Location Source/Goal

ra class inclusion Comitative
 equation some Instrument
 locative Manner
 relations

be possession Beneficiary
 (with some verbs of transfer
 of information: animate
 Goal/Beneficiary)

Table 9. All "Theme-Source/Goal strategy"-only verbs of transfer in
Jalonke

fala be 'tell, speak to, say to'
fii ma 'give to'
jaabaa be 'explain to/for'
jita ra 'show to'
luxun ma 'hide from'
ma-tii x[??]n/ma sell to' (lit. DISTR-stand [up])
sara ma 'buy from'
xee ma/ra/x[??]n/i 'send, commission'
xani ma/ra/x[??]n/i 'send'

Table 10. All Source/Goal-Theme strategy-only verbs of transfer in
Jalonke

kanta ma 'protect from, save from'
kisi ma 'protect from, save from'
niga ma 'learn to, teach to'
ra-sii ma 'advise, counsel (lit. CAUS-sow)'

Table 11. All Theme-Instrument strategy-only verbs of transfer
in Jalonke

kii ra 'present with, bestow with'
ra-tuu ra 'remind of (lit. CAUS-die)'
samba ra 'present with (after displacement of giver)'

Table 12. All verbs of transfer that alternate between the Theme-
Source/Goal and the Source/Goal-Theme strategies in Jalonke

doni ma 'lend, borrow (money)'
k[??]jek[??]je be[??]ma 'whisper'
ma-kula ma 'beg' (DISTR only)
ma-xorin ma 'ask' (DISTR only)
muga ma 'steal'
roni ma 'inherit'

Table 13. The one transfer verb that occurs in both the
Theme-Source/Goal and Theme-Instrument strategies in Jalonke

[??]efu ma/ra 'borrow, lend'

Table 14. The one transfer verb that occurs in both the
Source/Goal-Theme and Theme-Instrument strategies in Jalonke

t[??][??]nin ma/ra 'accuse'

Table 15. Participant marking and postpositions for the
different strategies

Strategy Role mapped Role mapped Role mapped to
 to Subject to direct object postpositionally
 marked
 participant

Theme-Source/Goal Effector Theme (Animate) Goal
strategy Effector Theme (Animate) Source

Source/Goal-Theme Effector (Animate) Goal Theme
strategy Effector (Animate) Source Theme

Theme-Instrument Effector Theme Instrument
strategy

Theme-Beneficiary Effector Theme Beneficiary
strategy

Strategy Postposition
 marking 3rd
 participant

Theme-Source/Goal ma
strategy

Source/Goal-Theme ma
strategy

Theme-Instrument ra
strategy

Theme-Beneficiary be
strategy
COPYRIGHT 2007 Walter de Gruyter GmbH & Co. KG
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Lupke, Friederike
Publication:Linguistics: an interdisciplinary journal of the language sciences
Geographic Code:60AFR
Date:May 1, 2007
Words:13782
Previous Article:Encoding three-participant events in the Lao clause (1).
Next Article:External possession and utterance interpretation: a crosslinguistic exploration *.
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters