Printer Friendly

To be with X is to have X: comitatives, instrumentals, locative, and predicative possession(1).


Comitatives and instrumentals are characterized by markedly different behavior as to their combinability with additional functions. This is especially the case with possession and locative: normally, possession is compatible with comitative as an additional function, whereas the instrumental does not enter combinations with possession easily. Similarly, instrumentals display a propensity to associate with additional locative functions from which the comitative tends to be banned. However, if expressions of predicative possession involve a marker that also has the function of encoding the instrumental, such combinations of functions generally involve the comitative as well. Likewise, if markers fulfil both a comitative and a locative function, they typically express the instrumental as well. These combination patterns are attested in quite a variety of languages of different genetic, areal, and typological background. On the basis of these empirical facts -- mainly drawn from examples of predicative possession -- the concept of bridging function is introduced. Bridging functions serve the purpose of legitimizing/facilitating combinations of functions that are otherwise incompatible with each other. The recurrent patterns and the concept of bridging function are discussed with reference to recent hypotheses on the nature of possession and the relationship that holds between classes of functions.

1. Introduction

The present paper seeks to determine to what extent functions that display a certain tendency for being expressed by the same marker in a variety of languages also behave similarly when it comes to extending their domain over additional contexts. Comitatives(2) and instrumentals are renowned for being rather closely associated with each other in the bulk of (Indo-)European languages. Superficially, this close association would suggest that comitatives readily accept the same additional functions as the instrumental and vice versa. However, it can be demonstrated that this is not the case. For the present purpose, I check the combinability of comitative and instrumental with possessive(3) and locative, respectively. It is argued that comitative and instrumental have markedly different preferences. The combinability of comitative and instrumental with possessive and locative has a system of its own, which can be best understood by introducing the concept of bridging function. Comitatives need a kind of mediator in order to enter a pattern that involves the preferred partner of the instrumental. Likewise, the instrumental calls for a bridging function in order to combine with the favorite of the comitative.

In what follows, I take Croft's model (1991) of the causal chain and the grammaticalization model developed by Heine et al. (1991) as a starting point (section 2). In contradistinction to their approaches, I have opted for talking about functions (also: senses and contexts) instead of categories, in order to avoid suggesting any ontology of universal grammatical relations (Dryer 1997). For the same reason, the term combination (of functions, senses, or contexts) is preferred here, whereas "syncretism" and its derivations -- always in quotation marks -- is used when reference to the work of others is made or stylistic considerations call for some variation. Croft's and Heine's hypotheses on the link that might hold between comitatives/instrumentals and possession are scrutinized and checked against the results of my own research on the typological and universal aspects of comitatives and instrumentals (sections 3 and 4). The focus will be predominantly on those grammatical elements -- labeled relators in the remainder of the paper -- that primarily encode the relation that holds between the possessor and the possessum. If necessary, the scope of the discussion will be extended to the whole construction. Finally (section 5), a summary will highlight once again the major issues of the present paper. For expository reasons, I draw on two different samples: on the one hand, there is the world-wide sample, comprising 323 languages, used previously for the statistics in Stolz (1996); on the other hand, a purely European sample of 65 languages and regional varieties will be used extensively for the present purpose. Note that except for the occasional overlap of a handful of languages, the two samples are largely independent of each other.

2. Scheming something

In Croft (1991: 183-239), the foundations for a modern theory of (case) "syncretism" have been laid.(4) The model of the causal chain outlined in Croft's monograph distinguishes possible from impossible patterns of "syncretism" in terms of the distinction of antecedent vs. subsequent thematic roles, claiming that "no surface case marker will subsume both subsequent and antecedent thematic roles" (Croft 1991:187); see (1). In terms of the causal chain, possessive and locative are believed to be neutral outsiders (Croft 1991: 198). The theoretical justification of the causal chain and of the distinction of classes of relations is not at issue here. Rather, I am going to ascertain whether or not functions that correspond to members of the same class of relations are characterized by a similar "syncretistic" behavior. According to (1), comitative and instrumental are members of one and the same class, viz. antecedent. Given that combinations with members of the class of subsequent relations are actually blocked, the question arises of how the members of the so-called neutral class interact with those of the other classes; more precisely, how locative and possession combine with comitative and instrumental.
(1) Classes of thematic roles according to Croft (1991)

 Antecedent Subsequent Neutral

 cause result locative ("space")
 agent benefactive possession
 comitative recipient

The English preposition with is a case in point: like many other relators in quite a few languages, it has the functions of both comitative and instrumental, which is in line with their being classified as antecedent thematic roles. In addition, again like many of its translational equivalents, with can also be used to express attributive possession,(5) as, for example, in the English NP the man WITH the beard. However, possession -- like locative -- is a type of function that is said to be independent of but systematically related to those thematic roles that form part of the causal chain (Croft 1991: 198), although the systematic nature of this relationship does not exactly become clear from the description given (Croft 1991: 206-212). If one wants to find out how "syncretism" actually works, how it is limited, and how comitative and instrumental might possibly be related to possession and locative, if at all, it is necessary not only to evaluate the extant theoretical literature but also to have a closer look at the empirical data. The latter will be the primary task of my present study, with special focus on predicative possession.

Much in the same vein as Croft, Bernd Heine (1993, 1997a, 1997b) and his associates (Heine et al. 1991) suggest a set of so-called event schemas(6) that are said to be "stereotypical descriptions of [...] recurrent experiences" (Heine 1997b: 91). These schemas are propositional in form (Heine 1997b: 82) and are believed to belong to the cognitively rooted endowment that allows humans to develop strategies to cope with new experiences by giving new meanings/functions to old signs (Heine 1997a: 222). As to language, such schemas are depicted as highly important determinants of the ubiquitous process of grammaticalization (Heine et al. 1991: 36). Partly owing to the different characteristics of the language phenomena to be described and explained, the number of event schemas proposed and their spellout have been subject to some minor revisions during the past decade.(7) Irrespective of such changes, there is a common core for all of the inventories of event schemas (Heine 1993: 32). The chart in (2) surveys the most recent catalogue that is meant to exhaust the schematic options for the grammaticalization of expressions of predicative possession, which is one of the major linguistic areas for which the impact of event schemas has been scrutinized in some detail (Heine 1997a: 45-76, 1997b: 83-107).(8)
(2) Event schemas for predicative possession (Heine 1997a: 47)(9)

 Formula Label

 X takes Y action
 Y is located at X location
 X is with Y companion
 X's Y exists genitive
 Y exists for/to X goal
 Y exists from X source
 As for X, Y exists topic
 Y is X's (property) equation

In a manner of speaking, the event schemas hint at certain source--target relationships in terms of grammaticalization processes: the formulas and labels sketch the source that is employed to express the target concept (Heine et al. 1991: 39-44). In the case at hand, there are eight different sources for the expression of one target, viz. predicative possession. Furthermore, the event schemas not only allow for a strictly diachronic interpretation, with the source serving as the etymological ancestor of the target concept's expression, but also provide guidelines for synchronically attested patterns of "syncretism," that is, the source and target components of event schemas identify at least two functions that are likely candidates for being expressed by the very same linguistic means. Put differently, in a language where the LOCATION schema is used to express predicative possession (Heine 1997a: 50-53), for instance, Y BEING AT X and Y BELONGING TO X/X HAVING Y are formally indistinguishable; see (3).(10)
(3) Finnish

 a. [VIII, 32] existential/location


 Ei tahde-lla-ni ole tiiker-i-ta
 Neg.3Sg planet-Ade-Psr.1Sg be tiger-Pl-Part
 `There are no tigers on my planet.'

 b. [XV, 64] possession


 Minu-lla on kolme tulivuor-ta
 1Sg-Ade be:3Sg three volcano-Part
 `I have [lit. on/at me, there are] three volcanos.'

 c. Realization of the LOCATION schema in Finnish



In contemporary Finnish, existential/locational constructions are the basis for the expression of both spatial relations and predicative possession. Therefore, these cannot be told apart on purely structural grounds; see (3c). The only way to decide which of the two possible readings actually applies is the animacy test: if the participant inflected for adessive ranks high on the animacy hierarchy, as in (3b), then there is a fair chance that one is dealing with an instance of predicative possession -- though this is just a rule of thumb.(11) I will come back to the Finnish data in an instant (cf. below).

Among the above event schemas, the COMPANION schema is especially important. This is so not only because it accounts for almost 13% of the attested major construction types for predicative possession in the 100-language sample checked by Heine (1997a: 75).(12) More basically, it reflects the Lakoffian COMPANION metaphor, which, in a manner of speaking, is at the very heart of the grammaticalization model developed by the Cologne Africanists (Heine et al. 1991: 52). The original hypothesis put forward by Lakoff and Johnson (1980) claimed that, in universal perspective, grammaticalized expressions of accompaniment (= comitatives) and grammaticalized expressions of instrumentality (=instrumentals) are prone to be encoded "syncretistically." In a large-scale cross-linguistic check, however, I was able to demonstrate that this pattern is far from being a universal; rather, it represents a typically (Indo-)European minority solution (Stolz 1996: 114-144). The vast majority of the world's languages keeps comitatives and instrumentals formally distinct. "Syncretistic" relators such as, for example, the English preposition with, which combines the functions of comitative and instrumental, are relatively rare birds, so to speak.

In the formulaic assumedly metalinguistic labelling of the event schemas in (2), the COMPANION schema contains exactly this English preposition with. The label that has been suggested for the schema clearly indicates that its proponent had only the comitative function in mind when he coined it. Of course, this is in line with the idea that instrumentals display a higher degree of grammaticalization than the more basic/less abstract comitatives do (Heine et al. 1991: 159). One might ask whether or not this automatically implies that only combinations of comitative and predicative possession are to be expected, whereas the combinatory encoding of instrumental and predicative possession would be ruled out. As a matter of fact, English with is the paradigm case of a relator of comitative and instrumental that is also used in attributive possessive constructions -- as in, for example, a flower WITH three petals. To ask which of the two readings of with, if any, is the one that allows for a possessive interpretation is only a rhethorical question. Heine et al. (1991: 166) clearly argue in favor of the comitative:

Although the POSSESSIVE use can be immediately derived from the COMITATIVE sense, it cannot be located anywhere between COMITATIVE and INSTRUMENT and, hence, does not seem to be part of the COMITATIVE-INSTRUMENT-MANNER continuum. Rather, it appears to belong to a distinct branch [...].

This quote adds further weight to the assumption that instrumentals and possession are less closely associated with each other than are comitatives and possession. Of course, one should not oversimplify the facts by treating attributed possession and predicative possesion as the same thing (Heine 1997a: 186). Indeed, there is ample evidence that the two major types of possession differ considerably in their readiness to participate in combinations involving comitative/instrumental. In a test-based sample of 65 European languages, 38 make use of comitatives and/or instrumentals for the purpose of expressing at least one variety of attributive possession in one randomly picked sentence alone; see (4). If we go beyond this single sentence, the number of languages/regional varieties would inevitably increase further.(13)
(4) Cases of combinations involving comitative/instrumental and
 attributive possession in sentence [XVIII, 2]


 English: [[a flower].sub.possessor] [[three petals].sub.
 [[with].sub.relator] possessum]

 Albanian (Gheg) me, Albanian (Tosk) me, Alsatian mit, Asturian
 con, Bulgarian s, Croatian sa, Czech s, Danish med, Dutch met,
 Estonian -ga, Faroese vid, Frisian mei, Friulan cun, German mit,
 Greek me, Icelandic med, Kurdish bi, Ladinian (Gerdeina) cun,
 Latvian ar, Letzebuergesh mat, Limburgian (North) met,
 Limburgian (South) mit, Lithuanian su, Macedonian so, Malti bi,
 Moldavian cu, Norwegian med, Rumanian cu, Sardinian chin,
 Swedish med, Serbian s, Slovak s, Surselvian cun, Ukranian z,
 Vallader cun, Yiddish mit

Thus, it cannot be denied that English with and its many equivalents in the remaining 37 European languages encode instrumental (mostly along with comitative) and attributive possession, that is, the instrumental usage does not necessarily exclude the possessive usage of the same relator and vice versa. However, when it comes to predicative possession, the figures are much less impressive: there are no more than five languages in my European sample that allow for combinations of comitative and/or instrumental and predicative possession, viz. Finnish, Welsh, Irish, Icelandic, and Portuguese; see below. Four of these accidentally do not appear in (4).(14) Thus, it cannot be established with certainty whether or not identity of expressions of predicative possession with either comitative or instrumental tends to imply that there is also identity of expressions of attributive possession with either comitative or instrumental in the same language. It remains to be determined under which conditions a combination of instrumental and possessive is tolerated or even preferred and under which conditions instrumental and possessive do not go together well. With a view to coming closer to an answer to these questions, I will look into some aspects of predicative possession in a variety of languages and will discuss the relationship of these modes of expression with comitatives and instrumentals. In the remainder of the text, I count only those predicative possessive constructions as instances of the COMPANION schema that involve any copula and a morpheme that serves as translational equivalent of English with.(15)

3. Possessed companions vs. possessed instruments

According to Heine's statistics (Heine 1997a: 75), the COMPANION schema is especially frequent in constructions of predicative possession in languages of Africa and Oceania/Australia, whereas it is absent from Europe and Asia. This picture is largely corroborated by my own findings, which report a pronounced African and Oceanian/Australian preference for the COMPANION schema. As to Europe and Asia, the number of attested cases is really rather small, though not absolutely zero (Stolz 1996: 166); see below.

For a start, consider Swahili, a Niger-Kordofanian language of East Africa. In this language, there is indeed a pattern that combines the function of comitative and predicative possession: both make use of the relator na; see (5).(16)
(5) Swahili

 a. (Kwon 1995: 186)


 Hamisi a-na ki-tabu
 Hamisi 3Sg-[with.sub.comitative] C1.7-book
 `Hamisi has a book.'

 b. (Kwon 1995: 28)


 ma-ua na bustani
 C1.6-flower [with.sub.conjunction] garden
 y-ote i-li-pend-ez-a sana
 C1.9-all C1.9-Past-like-Caus-Ind much
 `The flowers and all the gardens were much liked.'

 c. (Brauner and Herms 1986: 61)


 ni-na-zungumz-a na mwalimu
 1Sg-Prog-chat-Ind [with.sub.preposition] C11.teacher
 `I am talking with/to the teacher.'

 d. Realization of the COMPANION schema in Swahili



On the other hand, the instrumental is regularly encoded by a distinct relator kwa; see (6). It is by no means possible to substitute kwa for na or vice versa.
(6) Swahili (Kwon 1995: 166)


mama a-na-kat-a nyama kwa ki-su
mother 3Sg-Pres-cut-Ind meat [with.sub.
 instrumental] C1.7-knife
`Mother is cutting the meat with a knife.'

This is a fairly common constellation not only in African languages but also outside this area in the group A of languages that make use of a translational equivalent of English with in constructions of predicative possession; see (7). The statistics are based on my larger sample, analyzed in Stolz (1996: 164-165).
(7) Combinations

 Language group Pattern Attestations Share (%)

 A Com = Pos [is not
 equal to] Ins 30 58.83
 B Com = Pos = Ins 11 21.57
 C Com [is not
 equal to] Pos = Ins 10 19.60
 Total 51 100

As is evident from the figures given in (7), comitative is more frequent in combinations with expressions of predicative possession, though, contrary to our expectations, instrumental-possessive combinations are not completely ruled out (cf. group C). Nevertheless, it is possible to derive three tendencies; see (8).
(8) Tendencies I

 a. If a language keeps comitative and instrumental formally
 distinct and if one of the two forms part of the mode of
 expression of predicative possession, then there is a 3:1
 probability that it is the comitative that belongs to this

 b. If in a language the marker for instrumental forms part of the
 mode of expression of predicative possession, then there is a
 50% probability that comitative and instrumental are not
 distinguished by formal means.

 c. If in a language the grammatical expression of predicative
 possession requires the use of the marker of either comitative
 or instrumental, then in four out of five cases the comitative
 is involved.

This asymmetry suggests that comitative-possessive combinations are the normal or even unmarked case, whereas instrumental-possession is something special, the marked case that calls for an explanation. Of the 21 cases of combinations of instrumental functions with those of predicative possession (groups B and C), slightly over 50% also imply comitative-instrumental as a pattern -- and thus a combination of comitative and possessive as well (group B). Of the remaining ten cases in group C, six are patterns that include the locative and two are patterns that include the conjunction AND -- one pattern involving both locative and AND (Stolz 1996: 170).(17) Only three languages -- all of which happen to be Amerindian languages (Stolz 1996: 191-194) -- display purely binary "syncretistic" patterns of instrumental and possessive; see (9).
(9) Combinations of instrumentals and possessive without participation
 of comitative (subdivisions of group C of [7])

 Group Pattern Attestations Share (%)

 C1 Ins = Pos 3 30
 C2 Ins = Loc = Pos 5 50
 C3 Ins = AND = Pos 1 10
 C4 Ins = AND = Loc = Pos 1 10
 Totals 10 100

The charts in (7) and (9) strongly suggest that the instrumental needs an additional function in order to enter into combinations with the possessive. This necessity is not shared by the comitative: of the 30 cases of comitative-possessive combinations, 18 do without any additional functions. Interestingly, even with those patterns that involve both comitative and instrumental as partners of predicative possession, eight out of eleven cases contain at least one more function, again mostly locative or AND conjunction (Stolz 1996: 170).(18) On the basis of this observation, I am in a position to add number (10) to the list of tendencies under (8) above.
(10) Tendencies II

 If a language allows for the marker of predicative possession to
 be part of the range of functions covered by the marker of
 instrumental, then, with only a very few exceptions, at least
 one additional function that is not equivalent to the comitative
 is involved in the combination.

This state of affairs supports the idea that the relationship that holds between instrumental and possessive is a mediated one, that is, there needs to be a kind of bridging function that provides the link between the two (basic) functions. Obviously, such a bridge is not necessarily called for if comitative--possessive combinations apply. With a view to a better understanding of this bridging function, some pertinent examples from my European sample will be discussed in some detail.

4. European bridge-building

4.1. Portuguese

As mentioned above, formal identity of markers of predicative possession and either comitative or instrumental are clearly dispreferred in Europe. This dispreference makes especially interesting exactly those cases that display such combinations against all odds on European linguistic territory. Consider for instance the six examples from Portuguese, all of which belong to a special facet of the wide range of possession types, viz. possession(19) of physical and mental states; see (11).
(11) Portuguese

 a. [II, 58]
 Entao ja sem grande paciencia porque
 then already without great patience because
 estava com pressa de comecar a
 be:Imperf:1Sg with haste of begin:Inf to
 desmontar o motor rabisquei este desenho
 take.apart:Inf Def engine Dem drawing
 `Since I was in a hurry to start taking my engine apart, I then
 tossed off this drawing already with not much patience left.'

 b. [X, 83]
 Estava com pena de nao poder ver o
 be:Imperf:3Sg with pain of Neg see:Inf Def
 seu por do Sol como dantes
 Psr.3Sg:Psm.Masc sunset as before
 `He was regretting that he could not watch his sunset as

 c. [XVII, 17]
 Ja estava com medo de se ter
 already be:Imperf:3Sg with fear of Refl3 have:Inf
 enganado de planeta quando reparou num
 mistake:Ptc of planet when catch.sight:Pret:3Sg in:Indef
 anel cor de lua a mexer areia.
 ring color of moon to move:Infin:Def sand
 `He was already afraid he had mixed up the planets, when he
 noticed a coil of gold, the color of moonlight, flashing across
 the sand.'

 d. [XXVI, 158]
 E sentou-se porque estava com medo
 and sit:Pret:3Sg-Refl3 because be:Imperf:3Sg with fear
 `And he sat down, because he was afraid.'

 e. [XXIV, 20]
 Via-as como em sonhos porque
 see:Imperf:1Sg-Obj.Fem:Pl as in dream:Pl because
 estava com uma ponta de febre por causa de
 be:Imperf:1Sg with Indef point of fever through reason of
 `I looked at them as if I were in a dream, because thirst had
 made me a little feverish.'

 f. [XXVI, 46]
 Estava com um olhar muito serio perdido la
 be:Imperf:3Sg with Indef look very serious lose:Ptc there
 muito ao longe
 very to:Def distance
 `He had a very serious look, as if lost somewhere far away.'

The construction under scrutiny is made up of an inflected form of the verb estar `to be' -- one of the two copulas of Portuguese, with estar being reserved for the encoding of temporary states(20) -- the relator com `with', which like its English equivalent with encodes comitative and instrumental (cf. [12]), and finally the possessum NP, which with its typical instantiations, such as fome `hunger', frio `cold', calor `heat', sono `tiredness', inveja `envy/jealousy', etc., normally belongs to the realm of feelings, emotions, moods, illnesses; see (13) (Endruschat 1997: 247-250). Less frequently, the possessum NP designates a concrete tangible object, a garment one is actually wearing. These cases are all instances of temporary or actual possession. The possessor is the grammatical subject of estar, which need not be lexically present in a pro-drop language like Portuguese.
(12) Portuguese

 a. [IV, 50] comitative
 O meu amigo ja se foi
 Def Psr.1Sg:Psm.Masc friend already Refl3 go:Pret.3Sg
 embora ha seis anos com a sua ovelha
 away Exi six year:Pl with Def.Fem Psr.3:Psm.Fem sheep
 `Already six years ago, my friend went away with his sheep.'

 b. [II, 3] instrumental
 preparei-me para tentar consertar o aviao
 prepare:Pret:1Sg-Refl1Sg for try:Inf repair:Inf Def plane
 com as minhas proprias maos.
 with Def:Fem:Pl Psr.1Sg:Psm.Fem:Pl own:Fem:Pl hand:Pl
 `I prepared myself for having a try at repairing the plane with
 my own hands.'

(13) Realization of the COMPANION schema in Portuguese(21)


 esta-v-a com medo
 be-Imperf-Subj.3Sg with fear
 `he was afraid.'

Except for the phenomenon exemplified in (11), Portuguese conforms very much to the SAE model familiar from English, German, and French. In the majority of cases, Portuguese makes use of a typical HABEO-verb ter `to have' for the purpose of expressing predicative possession; see (14).
(14) Portuguese [VII, 32]
 Tenho coisas mais serias com que me
 have:1Sg thing:Fem:Pl more serious:Fem:Pl with Rel 1Sg.Obj
 `I have more serious business to keep me occupied.'

Thus, in Portuguese, the use of the COMPANION schema is restricted to one facet of possession. This observation is only partially in line with Heine's (1997a: 92-93) statement that the COMPANION schema is "more likely to express physical and temporary or, more generally, alienable possession rather than inalienable possession".(22) Owing to the fact that the Portuguese construction estar COM Y is the usual way to express possession of physical and mental states that count among the prime candidates for inalienable possession (Heine 1997a: 10), one does not get very far with the more general notion of alienability/inalienability in this case. Probably, Portuguese exemplifies a peculiar combination of features of both alienable and inalienable possessa, namely temporariness and reduced objecthood.

4.2. Icelandic

A similar phenonemon can be observed for Icelandic. In my sample text, sentence (15) is a good example of an Icelandic predicative construction with vera MED Y `to be with Y'. The construction normally contains a pronominal or nominal possessor, an inflected form of the verb vera `to be', the preposition med `with' governing the inflectional accusative on the possessum, and the possessum NP. Again, med is a relator encoding both comitative and instrumental (cf. [16]), though outside the realm of possession there are some differences as to case government (Stolz 1996: 176-177).
(15) Icelandic [II, 53]

 Realization of the COMPANION schema in Icelandic



 Hann er med horn
 3Sg.Masc be:3Sg.Pres with horn (Acc.Pl)
 `It [= the lamb] has horns.'

(16) Icelandic

 a. [IV, 50] comitative
 pad eru pegar lidin sex ar sidan vinur min
 that be:3Pl when pass:Ptc six year.Pl since friend Psr.1Sg
 for med kind-in-a sina
 go:Pret:3Sg with sheep-Def-Acc Psr.3:Psm:Fem.Acc
 `Six years have already passed, since my friend went away with
 his sheep.'

 b. [XV, 58] instrumental
 Frasagnir landkonnuda eru fyrst skradar med
 report:Def:Pl explorer:Gen:Pl be:3Pl first note:Ptc with
 `The reports of the explorers are first written down with a

The regulations for the case government of Icelandic med being rather intricate, I deliberately skip a number of details in order to be able to generalize: dative government is possible with both the comitative and the instrumental reading; accusative government, however, only allows for a comitative or possessive reading. This fact obviously speaks in favor of (a) there being two distinct constructions, and (b) a closer association of comitative and possessive to the exclusion of the instrumental.

Besides the encoding of predicative possession by the vera-med-X construction, Icelandic has two fully-fledged HABEO verbs, viz. hafa `have' and eiga `to own', which mainly express de jure ownership, social relationships, and abstract possession (Stolz and Gorsemann forthcoming); see (17).
(17) Icelandic

 a. [II, 5]
 Eg atti taeplega vikuforda af
 1Sg own:Pret:1Sg hardly weekly.ration:Acc of
 `I had hardly a week's ration of potable water.'

 b. [IV, 54]
 Og eg get ordid eins og fullordna
 and 1Sg can.1Sg become:Ptc like grow-up:Ptc
 folkid sem hefur adeins ahuga a tolum
 people:Def.Ntr Rel have:3Sg only interest on number:Dat:Pl
 `And I could become like the grown-ups who only have an
 interest in numbers.'

Despite the overall similarities between the Icelandic and the Portuguese constructions, there is only one single sentence in the entire sample text in which both languages make use of the COMPANION schema for the same purpose, viz. possession of an illness (cf. [18b]), whereas in the remaining cases the two languages go their separate ways.
(18) Icelandic

 a. [X, 121]
 Hann var med miklum valdsmannsbrag
 3Sg be:Pret:3Sg with much authority
 `He had a lot of authority.'

 b. [XXIV, 20]
 Eg sa paer eins og i draumi pvi ad eg
 1Sg see:Pret:1Sg 3Pl:Acc like in dream:Dat because 1Sg
 var med ofurlitinn sotthita vegna porstans.
 be:Pret:1Sg with little:Acc fever:Acc because of thirst:Def:Gen
 `I looked at them as if I were in a dream, because thirst had
 made me a little feverish.'

 c. [XXV, 55]
 pu ert med fyriraetlanir sem mer er
 2Sg be:2Sg with plan:Acc.Pl Rel 1Sg.Dat be:3Sg
 okunnugt um [...]
 be.unfamiliar:Ptc about
 `You have plans that I do not know about.'

 d. [XXVI, 133]
 En hann var enn med ahyggjur
 but 3Sg.Masc be:Pret:3Sg still with worry:Acc.Pl
 `But he was still worrying.'

In fact, Icelandic med is used for another mix of features typically associated with inalienability and those that are bona fide instances of alienability. Simplifyingly, its functional domain comprises the expression of functions such as, for example, possession of body parts, certain part-whole relationships, a variety of mental and physical states, and temporarily possessed items of various kinds, clothes one is actually wearing, etc. (Stolz and Gorsemann forthcoming). Icelandic vera med covers a wider range of functions than Portuguese estar com.

Owing to the fact that there is no comparable differential morphological case government in Portuguese, it is not possible to find a formal criterion as in Icelandic to decide which of the two readings of com, comitative or instrumental, comes closer to the possessive function. However, in both Indo-European languages, there is a function of the constructions under scrutiny that is, strictly speaking, not at all an instance of predicative possession; see (19)-(20).
(19) Icelandic [I, 27]
 Eg heft mikid verid med fullordnu folki
 1Sg have:Pret:1Sg much be:Ptc with grow.up:Ptc:Dat people:Dat
 og haft af pvi nain kynni
 and have:Ptc therefore near knowledge:Acc
 `I have been together with grown-ups rather frequently and
 therefore have come to know them.'

(20) Portuguese (Busse 1994: 221)
 Estou com ele ha ja dois anos.
 be:1Sg with 3Sg.Masc Exi already two.Masc year:Masc:Pl
 `I have been together with him for two years now.'

As a matter of fact, both examples invite an associative reading based on a locative one, that is, the virtual copresence of two entities is stated in sentences (19)-(20). More precisely, these entities are maximaly animate, they refer to human beings who happen to be in the same -- abstract social or concrete local -- space at the same time. Admittedly, the locative function of med and com is not the most prominent or frequent among their various competing functions. In addition, it is restricted to combinations with animate NPs. Nevertheless, it is important to take note of the fact that besides the indiscriminate encoding of comitative, instrumental, and possession, there is one more member of the combination of functions for the two prepositions, viz. locative. The question still to be answered is whether or not it is this locative component that facilitates the association of instrumentals and possessives; see (9) above.

4.3. Finnish

The general importance of the locative component for the subject at hand becomes clearer as soon as the Finnish data are taken into consideration. In Finnish, there are quite a few competitors when it comes to encoding the comitative. Besides the now somewhat archaic inflectional comitative (cf. [21]), there is at least one very frequent postposition -- kanssa `with' -- that may pass as the principle marker of the comitative (cf. [22]). In contradistinction to the two Indo-European languages discussed above, Finnish keeps comitative and instrumental strictly distinct by formal means. The instrumental is encoded by the inflectional adessive, which in turn forms part of the subparadigm of local cases of modern Finnish (cf. [23]). Among a variety of other functions, the adessive usually encodes spatial relations corresponding to those expressed by the English prepositions at, on, close to (cf. [24]). The most frequent function of the adessive, however, is that of marking the possessor in constructions of predicative possession (cf. [25] and [3]). With 51 out of 135 occurrences of the adessive in my sample text, predicative possession outnumbers every other function of the Finnish adessive. For brevity's sake, I only list the two examples that directly correspond to parallel examples of Icelandic med and Portuguese com, respectively (cf. [25]).
(21) Finnish [VIII, 31] inflectional comitative
 Tiikeri-t saa-vat tulla kyns-ine-en
 tiger-Nom.Pl may-3Pl come-Inf claw-Pl:Com-Psr.3
 `May the tigers come with their claws!'

(22) Finnish [XXVI, 29] postpositional comitative
 Keskustele-t kaarme-id-en kanssa
 talk-2Sg snake-Pl-Gen with
 `You are talking with snakes!.'

(23) Finnish [VII, 25] instrumental
 Jos ei tuo tappi vielakaan helti-a iske-n si-ta
 if Neg.3Sg Dem bolt still.not loosen-Inf hit-1Sg Dem-Part
 `If this bolt still won't turn, I am going to knock it out with a

(24) Finnish [XI, 1] location (cf. [3a])


 Seuraava-lla tahde-lla asu-i Turhamainen.
 second-Ade planet-Ade live-Pret
 `On the second planet, there lived a conceited man.'

(25) Finnish possession (cf. [36])

 a. [II, 53] (cf. [16], Icelandic)


 Si-lla on sarve-t
 Dem-Ade be:3Sg horn-Nom.Pl
 `This one has horns.'

 b. [II, 58] (cf. [11a], Portuguese)
 Silloin karsivallisyyte-ni loppu-i ja kun
 then patience-Psr.1Sg end-Pret and as


 minu-lla ol-i kiire mootor-ia
 1Sg-Ade be-Pret haste engine-Part
 korjaa-maan tekais-i-n seuraa-va-n piirustuks-en
 repair-Inf3:Ill follow-Ptc-Gen drawing-Gen
 `Then my patience came to an end, and as I was in a hurry to
 repair the engine, I tossed off the following drawing.'

For certain types of attributive possession, the inflectional comitative and its postpositional competitor are not altogether uncommon as relators though they are often considered substandard; see (26). At least from the perspective of normative grammar, the adessive is not acceptable in constructions of attributive possession.(23) However, it is exactly the other way around with predicative possession: here, the various comitatives are banned, whereas the adessive is the only legal option for marking the possessor. This is paramount to saying that only the morpheme that also encodes the instrumental is involved in predicative possession.
(26) Finnish (Nau 1995: 134)(24)
 vuokra-tta-va-na huone piano-n kanssa
 let-Pass-Ptc-Ess room piano-Gen with
 `Room with piano to let.'

Looking beyond the peculiarities of the Finnish case, combinations of instrumental and locative are the second most frequent pattern in my worldwide sample (with number 1 being that of comitative and instrumental) of a selection of 17 patterns (Stolz 1996: 164-165). "Syncretistic" expressions of instrumental and locative account for about 20% of all attested cases of combinations of functions, whereas patterns involving comitative and locative have a share of slightly more than 2%! Interestingly, even the rate of ternary patterns made up of locative, instrumental, and comitative is higher (3.75%). Thus, there is evidence that the instrumental associates more easily with the locative than does the comitative. According to Heine (1997a: 75), with about 21%, the LOCATION schema is the most frequent among the schemas used for the expression of predicative possession in the languages of the world. On this basis, it is tempting to hypothesize that the cases of combinations of (comitative/)instrumental--possessive I have been discussing so far are in fact combinations of up to three different patterns, viz. locative--possessive, locative--instrumental, and comitative--instrumental, forming part of a conceptual network, (27), slightly different from the one sketched by Heine et al. (1991: 165-166).(25) It remains to be investigated whether or not possession types A and B are indeed distinct categories. Heine (1997a: 92-93) hypothesizes that the LOCATION schema and the COMPANION schema are both associated with temporary and physical possession. However, as mentioned above, the Portuguese and Icelandic regulations cut across the lexicon of potential possessa in very peculiar and very different ways. On such grounds, a language-independent distinction of alienable and inalienable possession is rendered almost impossible. More evidence is needed to decide whether or not the LOCATION schema and the COMPANION schema have affinities to different facets of temporary and physical possession and sundry types of possession.
(27) Conceptual network


If the combination of locative and possessive is assumed as the basic pattern, then we might state the following: the instrumental goes with the possessive (especially if possessive and locative are already part of a pattern), and the comitative goes with the locative more easily if instrumental and locative are in the combination as well. Put differently, the locative serves as the bridge between possession and instrumental. Similarly, the instrumental serves as the bridge between comitative and locative.

4.4. Celtic

As a kind of aside to the above discussion, some pertinent data from two members of the Celtic branch of the Indo-European macrophylum will be touched upon. A closer inspection would of necessity require a detailed recapitulation of the diachronic processes that eventually led to the present state of affairs in Irish and Welsh. For obvious reasons, such a flashback of Celtic language history has to be reserved for a follow-up study. In the present paper, I will make do with commenting tentatively on the diachronic implications of the synchronic facts. Actually, both Irish and Welsh have undergone a development that has brought about a constellation involving instrumental, comitative, possession, and locative that superficially looks quite different from those cases discussed above. However, this probably might turn out to be just a trompe-d'oeil.

4.4.1. Irish. In modern Irish, the preposition le `with' is used for a variety of purposes that should look rather familiar by now. First of all, le encodes the comitative, (28), and the instrumental, (29). There are also at times only idiomatic relics of a formerly more widespread locative usage, (30), corresponding to English against, along (Muller 1990). In addition, le is obligatory in constructions of predicative possession of the so-called belong type (Heine 1997a: 29-33); see (31).(26) The more regular construction of the HABEO type -- an instance of the LOCATION schema, by the way -- makes use of a different preposition, viz. ag `at' (Heine 1997a: 51). There is no fully fledged HABEO verb in Irish.
(28) Irish [XXI, 103] comitative
 Teann siad ag rince Deardaoin le cailin-i an bhaile
 go 3Pl on dance Thursday with girl-Pl Def village
 `On Thursday, they go to dance with the girls of the village.'

(29) Irish [XV, 58] instrumental
 Mar is le peann luaidhe a bhreactar sios tuairisci
 like Cop with pencil lead Rel write:Impers down description:Pl
 na dtaiscealaithe i dtus baire.
 Def:Pl investigator:Pl in beginning game
 `Because for a start, the descriptions by the explorers are
 down with pencil.'

(30) Irish (O Siadhail 1985: 105) locative
 Ta droim Chait leis an mballa
 be sit/stand Cait with:3Sg.Masc Def wall
 `Cait is sitting/standing with her back against the wall.'

(31) Irish possession

 a. [XIII, 1]
 An ceathru plainead ba le fear gno e
 Def fourth planet Cop.Pret with man business 3Sg.Masc
 `The fourth planet belonged to a businessman.'

 b. [XIII, 69]
 Ce leis iad?
 who with:3Sg.Masc 3Pl
 `To whom do they belong?'

 c. [XIII, 72]
 Le duine ar bith.
 with person at all
 `(They belong) to nobody.'

The few remains of an erstwhile predominant locative function of modern Irish le are paralleled by similarly relic-like locative readings -- perlative, prolative, or the like -- of other relators such as, for example, English with and Icelandic med (Kress 1982: 205). Their etymology suggests that the spatial semantics of these prepositions is the oldest historical layer, now on the retreat to the benefit of comitative, instrumental, and possession. On the basis of my above findings, I dare to speculate that le, with, and med -- presumably com, too, way back in early Latin -- all started out as spatial prepositions inviting successively a possessive and a comitative reading before the instrumental eventually came into play.

4.4.2. Welsh. Being a distant relation of Irish, Brythonic Welsh also lacks a proper HABEO verb. Furthermore, Welsh displays a rather complicated system of allomorphs, full synonyms, and partial synonyms surveyed in Stolz (1998). For brevity's sake, I will mention only one phenomenon, which was discussed by Heine (1997a: 100) in order to demonstrate how to reconstruct event schemas.

In modern Welsh, there are currently several competing prepositions that may be used in constructions of predicative possession. The high-style literary language prefers the preposition gan `with, by' marking the possessor. However, except for predicative possession, it is very hard to find any instance of gan that would pass as a fair translational equivalent of English with nowadays. The second most frequent function of gan is that of marking the agent in periphrastic passive constructions, which is an innovation in the Welsh grammatical system. Not surprisingly, what comes to mind first is a translation requiring English by. It is exactly this function of marking the agent of passive constructions that suggests that gan once had an instrumental function as well (Stolz forthcoming). Irrespective of the supposed locative, comitative, and instrumental prehistory of gan, none of these functions is synchronically attested for the preposition gan. However, the second preposition mentioned in passing by Heine (1997a: 100) is indeed a case in point: gyda `with' has a great deal, though not quite everything, in common with the typical cases presented above.

For stylistic considerations, gyda does not display the full range of its contemporary functions in every text. In my sample text, for instance, predicative possession is exclusively the job of gan. The only example in which gyda comes close to a possessive is (32).(27)
(32) Welsh [II, 3]
 A chan nad oedd mecanic na theithwyr gyda mi [...]
 and since Neg be:Pret:3Sg mechanic Neg passenger with 1Sg
 `And since I had neither a mechanic nor a passenger (with me) ...'

This is a typical constellation of a stative comitative involving animate participants in the slots for both the accompanee and the companion. This is indicative of the original function of gyda, which no doubt was to encode the comitative. As yet, there are only a very few and probably not fully acceptable instances of gyda being used with an instrumental reading, which is normally reserved for the preposition a or regional efo (Stolz 1998: 120-125). Those prepositions that encode the instrumental -- be it together with the comitative or without a comitative component -- are never part of a predicative possessive construction. In contradistinction to a and efo, predicative possession is probably the most frequent function of gyda; see (33).
(33) Welsh [Ifor 27]

 r-oedd ffagl gyda phob un o'-r ddau
 Dec-be:Pret:3Sg torch with all one of-Def two
 `Each of the two had a torch.'

With regard to the problems Welsh gyda poses, Heine (1997a: 100) states that "in modern Welsh [...] we are dealing with an instance of the Location Schema, rather than with Companion [...] in spite of the meaning of the preposition."

His argumentation is based exclusively on the marking strategy: simplifying, if the relator marks the possessum then this might indicate the presence of the COMPANION schema. However, if the relator marks the possessor then this is supposed to be an instance of the LOCATION schema (Heine 1997a: 102). Owing to the fact that the event schemas primarily have something to do with semantics, rather than with syntax, I have certain reservations about the validity of Heine's interpretation of this special case, as long as it cannot be proved that gyda either stems from an etymological source with a spatial meaning or has acquired spatial features during its history. In fact, gyda historically reflects the univerbation of a noun cyd `union' and the preposition a `with', a combination that invites a rather narrowly defined comitative interpretation (Stolz 1998: 124). Moreover, neither a nor gyda displays any properties of proper local prepositions in the extant sources. All this points to the possibility that the syntactic organization of the constructions used for predicative possession and the semantics of the relators used in the very same syntagms may be at odds -- from the point of view of the grammaticalization model. At present, it seems a little daring to ascribe any spatial component to gyda.

In short, this means that the regulations in Welsh instantiate a pure case of comitative-possessive "syncretism" without any participation of an instrumental component. In the light of the network in (27), this comes as no real surprise since there is no locative function of gyda that could serve as a bridge for the instrumental to enter the pattern with possessive. At the present stage, it cannot be decided if there is something tangible behind my ad-hoc distinction of possession type A and possession type B of (27). For the time being, I will make do with summarizing the empirical facts in (34), which is again suggestive of the dependence of the instrumental on the locative and of the comitative on the instrumental when it comes to participating in combinations with possessive and locative, respectively.
(34) Matrix of combinations(28)

Language Relator Comitative mental Possessive Locative

Portuguese com + + + *
Icelandic med + acc + - + -
 med + dat + + - +
Finnish adessive - + + +
 kanssa + - - -
Irish le + + + *
Welsh gyda + - + -

5. Conclusions

Bearing in mind that developing a general and viable theory of semantic compatibilities is presently one of the most important goals of at least some schools of linguistic thought, the intricacies of the behavior of comitatives and instrumentals might be considered a valuable testing ground for the extant hypotheses. First of all, it has turned out that two members of Croft's class of antecedent roles display divergent leanings as to their preferred associated functions (Stolz 1996: 172-177). Comitative and instrumental belong to the same subclass of so-called causal relations but their combinability with causally neutral categories such as, for example, possessive and locative differs considerably. As the evidence gathered by Heine (1997a) suggests, these two outsider relations (= locative and possessive) partake quite freely in combinations with each other. Furthermore, patterns involving one of the causally neutral outsiders and one of the antecedent relations are anything but rare. However, there are restrictions on the combinability of neutral relations and antecedent relations, or at least strong preferences. The locative primarily selects the instrumental, whereas possession prefers the comitative as member of the same pattern. In order to let one of the dispreferred functions participate in a combination, a bridging function is called for.

Let us therefore attribute some general importance to the concept of bridging function. Within the framework of Croft's approach, it might be useful to check whether or not the supposed blocking of "syncretism" between members of the antecedent and the subsequent class can be overcome by a conceptual detour requiring a bridging function of some kind. It is a fair guess that the prime candidate for the function of the conceptual bridge is the locative or the class of spatial relations. Therefore, one should investigate the possibility of the causally neutral categories serving as mediators between the two classes of causal relations.

As to possession, on the one hand, and comitatives and instrumentals, on the other, the European data once again support the idea that, in spite of their many common traits, the latter two functions do not behave like Siamese twins, as was formerly believed. Rather, they both display a lot of individual properties that allow for quite a variety of solutions as to the association with others. One might even go as far as concluding that there are two different types of possession: type A, which is comparatively easily accessible for instrumentals, and type B, which seems to have more in common with the comitative. Given the validity of the bridging-function concept, it is possible to hypothesize that possession type A differs from possession type B in so far as the latter lacks the static spatial features that the former probably possesses. Future studies will reveal to what extent the distinction of two possession types based on their association with either instrumental or comitative can actually stand the test.
University of Bremen

Received 15 June 2000
Revised version received
4 December 2000


(1.) The present paper has been made possible by skilfull support on the part of the members of my project crew at the University of Bremen. I am especially grateful to Sabine Gorsemann, Kai Herkstrotter, Sonja Kettler, Anna Sabater Fuentes, Ann-Lou Kleppa, Daniela Schuto, Oxana Schwarz, Christel Stolz, Cornelia Stroh, and Aina Urdze. The DFG (German Science Foundation) kindly financed my research within the special program Sprachtypologie. A word of thanks is also due to Andreas Ammann (Antwerp), Steven R. Fischer (Auckland), Bernd Heine (Cologne), Silvia Luraghi (Pavia), Enrique Palancar (Madrid), and Paolo Ramat (Pavia), who provided me with much needed pertinent literature and additional information. Last but not least, the two anonymous referees who commented on the first draft of the article deserve my gratitude. Of course, the sole responsibility for content and style of the present paper is exclusively mine. Correspondence address: FB10: Linguistik, Universitat Bremen, Postfach 330 440, 28334 Bremen, Germany. E-mail:

(2.) In the present paper, I use the terms comitative and instrumental as handy labels for a combination of functions/senses ascribed to a grammatical element, with comitative being the label for accompaniment + X and instrumental standing for means + Y. Accompaniment and means are core functions; X and Y represent potential additional senses or functions (including the highly improbable case of {0}). For the present purpose, X and Y exclude the functions of possessive/possession and locative; cf. below.

(3.) The label possessive/possession is used here much in the same was as it is employed in Heine (1997a). Thus, it serves as a cover term for quite a variety of functions reaching from the expression of legal ownership via that of abstract possession to the expression of part-whole relations. The question of whether or not possession is an especially fortunate terminological choice cannot be answered satisfactorily in the present paper. On the other hand, locative, as it is to be understood here, covers a variety of spatial senses ranging from inessive via ablative to prolative, etc. In both cases, i.e. for possessive as well as locative, it is assumed that the core functions, more often than not, occur in combinations with other functions, so that they -- similar to comitative and instrumental -- involve a component P or Q, respectively, which stand for a variety of readings. Again, these additional components are not to be equated with either comitative or instrumental: cf. above.

(4.) For a recent moderate criticism of Croft's approach, cf. Luraghi (forthcoming).

(5.) For the definition and catalogue of the various possession functions, I rely entirely on the inventory presented in Heine (1997a).

(6.) Formerly called "source propositions" (Heine et al. 1991: 36).

(7.) Heine et al. (1991: 36) started out with six distinctive schemas and Heine (1993: 31) lists nine, whereas Heine (1997a: 47, 1997b: 91) winds up with eight.

(8.) Of course, possession has also been the topic of much scholarly research work with at times a radically different scientific leaning (cf. inter alios Seiler 1983; Lehmann 1998). For brevity's sake, I refrain from surveying the state of the art of possession research in linguistics. For a brief history of ideas, cf. Heine (1997a: 1-44) and Stolz and Gorsemann (forthcoming: chapter 2).

(9.) Several of the formulas listed in (2) contain a verb that looks like an existential. This fact may give rise to the suspicion that there might be something more basic than the propositions themselves (Heine 1993: 32, 1997a: 57-58). With a view to keeping things as brief as possible, I skip discussing the issue of whether or not this something is identical with the "source concepts" presented in Heine et al. (1991: 32-36). Moreover, it might turn out to be rather difficult to distinguish instances of the LOCATION and COMPANION schemas from those that involve an existential.

(10.) Unless otherwise stated, the examples are taken from the various translations of the French original of Antoine de Saint-Exupery's Le petit prince. Roman numbers indicate the chapter, Arabic numbers the sentence in the chapter. Italics mark those elements of constructions that are relevant for the present argument. The English translations are all mine. Where they diverge from the English version of Le petit prince, my translations are closer to the structure of the sentence under scrutiny. The following abbreviations are used:
Acc = accusative
Ade = adessive
Caus = causative
Cl = class
Com = comitative
Cop = copula
Dat = dative
Dec = declarative
Def = definite
Dem = demonstrative
Ess = essive
Exi = existential
Fem = feminine
Gen = genitive
Ill = illative
Imperf = imperfect
Impers = impersonal
Inf = infinitive
Ins = instrumental
Ind = indicative
Indef = indefinite
Loc = locative
Masc = masculine
Neg = negation
Nom = nominative
Obj = object
Part = partitive
Pass = passive
Pl = plural
Pos = possessive
Pres = present
Pret = preterite
Psm = possessum
Psr = possessor
Ptc = participle
Refl = reflexive
Rel = relative
Sg = singular

The following sources are used for The Little Prince: Finnish: Packalen (1992) Icelandic: Bjornsson (1988) Irish: O Doibhlin (1997) Portuguese: Morais Varela (1987) Welsh: Dafis (n.d.)

"Ifor" is used to indicate Owen (1973).

(11.) Sulkala and Karjalainen (1992: 227-228) claim that the adessive is restricted to animate possessors; inanimate possessors are said to require the inflectional inessive.

(12.) The COMPANION schema is in fifth position on Heine's list. However, it's a fairly close run between ACTION schema, GENITIVE schema, and COMPANION schema, all of which oscillate around the 14% mark.

(13.) The other 27 members of my European sample are as follows: Aragones, Armenian, Azeri, Basque, Bielo-Russian, Breton, Catalan, Corsican, Finnish (cf. below), French, Galego, Gascognian, Georgian, Hungarian, Irish (cf. below), Italian, Ladinian (Badiota), Languedocian, Polish, Portuguese (cf. below), Provencal, Romani, Russian, Saame, Slovene, Spanish, Turkish, Welsh (cf. below).

(14.) In sentence [VII, 5], one gets Portuguese as flores COM espinhos and Welsh Blodau A phigau (arnyn nhw) for English (the) flowers WITH thorns (on them). Nau (1995: 134) reports how in modern colloquial Finnish certain constructions of attributive possession involving instrumentals or comitatives -- some of them being calques from Swedish -- are constantly gaining ground; for details cf. Stolz and Stroh (forthcoming).

(15.) Note, however, that these constructional properties are not restricted to expressions of predicative possession. A pertinent example of a nonpossessive use of the very same structure is the Slavic predicative instrumental such as, e.g.
(i) Russian [IX, 33]
 postarasja [byt'.sub.[copula]] [scastlivym.sub.[instrumental]]
 `try to be happy.'

(16.) Structurally speaking, na is a verboid equivalent to English to be with in (5a) carrying a subject prefix, whereas it is a conjunction and in (5b) and a preposition with, to in (5c) with no agreement morphology at all (Brauner and Herms 1986: 55-56).

(17.) The conjunctional and locative functions being among the most frequent additions to the comitative and instrumental functions, respectively (Stolz 1996), they are chosen here for an exemplification of the differential behavior of comitative and instrumental.

(18.) For the intricate distinction of AND and WITH cf. Stassen (2000).

(19.) As to the aptness of the term possession in such and sundry cases, cf. note 3.

(20.) It is also possible to use the preposition com in combination with the verb ser `to be', which normally indicates permanent states. Strictly speaking, the construction X ser COM Y does not fall outside the realm of possession. However, since it is an isolated idiomatic expression corresponding to English Y is X's business/task (Busse 1994: 375), it is too marginal and therefore will not be considered further in this study.

(21.) Odd as it may seem at first sight, (13) actually instantiates the COMPANION schema -- simply because it involves the preposition com, i.e. the translational equivalent of English with in Heine's inventory of schemas; cf. (2). There is no need for the NPs to meet certain animacy requirements or the like.

(22.) This statement is based at least in part on an earlier observation of mine, which evidently has to undergo a revision (Kilian-Hatz and Stolz 1993).

(23.) See note 11.

(24.) The example seems awkward not only because of the postpositional comitative but also because one would expect the first noun to be in the partitive, huoneita (Karlson 1984:211).

(25.) Conceptual networks and their likeness, called semantic maps, have been en vogue in linguistics for quite some time. A recent study that makes ample use of such maps is Haspelmath (1997).

(26.) According to Heine (1997a: 29), the two types -- belong vs. have -- can be distinguished inter alia on structural grounds. In the have type, the possessor is conceived of as the subject/agent, whereas in the belong type, this grammatical role is reserved for the possessum.

(27.) It needs to be investigated whether or not the fact that the sentence is a negative one determines the choice of gyda in (32).

(28.) * marks those cases where locative functions are only marginally attested but have been more frequent in earlier stages of the language.


Bjornsson, Porarinn (1988). Litli Prinsinn. Reykjavik: Bokautgafa Menningarjods.

Brauner, Siegmund; and Herms, Irmtraud (1986). Lehrbuch des modernen Swahili. Leipzig: Enzyklopadie.

Busse, Winfried (1994). Dicionario sintactico de verbos portugueses. Coimbra: Alemdina.

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

Dafis, Llinos Iorweth (n.d.). Y Tywysog Bach. Lerpwl: Gwasg y Brython.

Dryer, Matthew (1997). Are grammatical relations universal? In Essays on Language Function and Language Type Dedicated to Talmy Givon, Joan Bybee, John Haiman, and Sarah A. Thompson (eds.), 115-143. Amsterdam and Philadelphia: Benjamins.

Endruschat, Annette (1997). Die Praposition COM im Afroportugiesischen: Versuch einer neuen Interpretation ihrer spezifischen Verwendung. In Afrolusitanistik -- eine vergessene Disziplin in Deutschland?, Ruth Degenhardt, Thomas Stolz, and Hella Ulferts (eds.), 243-278. Bremen: University of Bremen.

Haspelmath, Martin (1997). Indefinite Pronouns. Oxford: Clarendon.

Heine, Bernd (1993). Auxiliaries. Cognitive Forces and Grammaticalization. New York and Oxford: Oxford University Press.

--(1997a). Possession. Cognitive Sources, Forces, and Grammaticalization. Cambridge: Cambridge University Press.

--(1997b). Cognitive Foundations of Grammar. New York and Oxford: Oxford University Press.

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

Karlsson, Fred (1984). Finnische Grammatik. Hamburg: Buske.

Kilian-Hatz, Christa; and Stolz, Thomas (1993). Grammatikalisierung und grammatische Kategorien. Ein Bericht aus der Pathologie. Essen: University of Essen.

Kress, Bruno (1982). Islandische Grammatik. Munich and Leipzig: Hueber.

Kwon, Joung-Mi (1995). Possession in Swahili. Cologne: Institut fur Afrikanistik.

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

Lehmann, Christian (1998). Possession in Yucatec Maya. Structures, Functions, Typology. Munich: LINCOM Europa.

Luraghi, Silvia (forthcoming). Syncretism and the classification of semantic roles. In Sprachtypologie und Universalienforschung 54(1).

Morais Varela, Joana (1987). O Principerzinho. Lisbon: Caravela.

Muller, Nicole (1990). Zur altirischen Praposition la. In Deutsche, Kelten und Iren, 150 Jahre deutsche Keltologie. Gearoid Mac Eoin zum 60. Geburtstag gewidmet, Hildegard L. C. Tristram (ed.), 115-123. Hamburg: Buske.

Nau, Nicole (1995). Moglichkeiten und Mechanismen kontaktbewegten Sprachwandels unter besonderer Berucksichtigung des Finnischen. Munich and Newcastle: LINCOM Europa.

O Doibhlin, Breandan (1997). An Prionsa Beag. Beal Feirste: Lagan.

O Siadhail, Micheal (1985). Lehrbuch der irischen Sprache. Hamburg: Buske.

Owen, Ivor (1973). Ifor bach. Dinbych: Gwasg Gee.

Packalen, Irma (1992). Pikku Prinssi. Porvoo, Helsinki, and Juva: Werner Soderstrom Osakeyhtio.

Ramat, Paolo (1999). Review of Bernd Heine, Possession. Archivo glottologico italiano 84, 115-128.

Seiler, Hansjakob (1983). Possession as an Operational Dimension of Language. Tubingen: Narr.

Stassen, Leon (2000). AND-languages and WITH-languages. Linguistic Typology 4(1), 1-54.

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.

--(1998). UND, MIT und/oder UND/MIT? Koordination, Instrumental und Komitativ -- kymrisch, typologisch und universell. Sprachtypologie und Universalienforschung 51(2), 107-130.

--(forthcoming). Comitatives vs. instrumentals vs. agents. In Aspects of Typology, Walter Bisang (ed.). Berlin: Akademie.

--(i.p.). On Circum-Baltic instrumentals and comitatives: to and fro coherence. In Circum-Baltic Languages, Maria Koptjevskaja-Tamm and Osten Dahl (eds.). Amsterdam and Philadelphia: Benjamins.

--; with Gorsemann, Sabine (forthcoming). Pronominal possession in Faroese and the parameters of alienability/inalienability. In Studies in Language.

--; and Stroh, Cornelia (forthcoming). Wenn Komitative Instrumentale sind und umgekehrt. In Kasusvariationen, Gerd Hentschel and Winfried Boeder (eds.). Oldenburg.

Sulkala, Helena: and Karjalainen, Merja (1992). Finnish. London and New York: Routledge.
COPYRIGHT 2001 Walter de Gruyter GmbH & Co. KG
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2001 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:Linguistics: an interdisciplinary journal of the language sciences
Geographic Code:1USA
Date:Mar 1, 2001
Previous Article:On the event structure of -ung nominals in German(*).
Next Article:The principle of pragmatic detachability in borrowing: English-origin discourse markers in Pennsylvania German(1).

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters