Abstract
In the adnominal possessive construction of Mandinka (a West Mande language spoken in Gambia, Senegal, and Guinea Bissau), the possessor NP may be simply juxtaposed to the possessee or flagged by the postposition lá. This coding split shows unusual properties in comparison with most of the languages having formally similar coding splits. Crucially, any noun can be modified by an unflagged possessor, depending on the semantic nature of the possessor and of its relation to the possessee. The present-day distribution of the two constructions cannot be analyzed as reflecting an alienability contrast, nor can the alienability contrast account for the emergence of this coding split. The data analyzed in this study suggest that the lá-construction initially referred to situations where the possessor determines the location of the possessee, but is grammaticalized differently depending on the grammatical nature of the head noun and the semantic nature of the modifier. The Mandinka data also show the limits of explanations in terms of discourse frequency, and the need to admit that semantic features such as animacy and control may play a crucial role in the genesis of coding splits in the adnominal possession construction.
1 Introduction
Mandinka, spoken by approximately 1.5 million speakers in Gambia, Senegal, and Guinea Bissau, is the westernmost member of the Manding dialect cluster, included in the Western branch of the Mande language family:
Mandinka ⊂ Manding ⊂ West Mande ⊂ Mande
The Mande language family was included by Greenberg in the Niger-Congo phylum, but the evidence for a Niger-Congo affiliation for Mande is rather slim; Dimmendaal (2011), for example, argues that Mande is best treated as an independent language family.[1]
The literature on Mandinka is relatively small, but includes a comprehensive reference grammar (Creissels and Sambou 2013), with a chapter describing the coding split in the adnominal possessive construction. The present article discusses theoretical issues raised by the facts described in their chapter devoted to adnominal possession.
Two variants of the adnominal possessive construction must be distinguished in Mandinka. In the direct construction, the possessor NP immediately precedes the possessee noun, without any intervening morphological material, as in (1a), whereas in the indirect construction, the possessor NP is flagged by the postposition lá (glossed GEN when it fulfills the function of possessor flag in the adnominal possessive construction), as in (1b).[2]
| Jàatà-kúndáa | kòlóŋò |
| Jaata-neighborhood | well.D |
| ‘the well of the Jaatakundaa neighborhood’ | |
| Jàatà-kùndà-nkôo-lú | lá | kòlóŋò |
| Jaata-neighborhood-inhabitant.D-PL | GEN | well.D |
| ‘the well of the Jaata family’ | ||
This article addresses the question of the validity of the alienability contrast as a functional explanation for this coding split.[3] As developed by Rose and Van linden in the introductory chapter of this special issue, to which readers are referred for more references on alienability, inalienable possession is usually conceived as involving “either inextricable, essential or unchangeable relations between possessor and possessee” (Chappell and McGregor 1996: 3), in contrast to alienable possession conceived as involving less permanent or inherent types of associations (Chappell and McGregor 1989: 25). In the particular case of Mandinka, a superficial observation limited to the possessive relations commonly considered prototypical (ownership, kinship, and part/whole relationships) may suggest analyzing the coding split in the adnominal possessive construction of Mandinka as straightforwardly conditioned by alienability. However, a comprehensive description of the uses of the adnominal possessive construction leads to the opposite conclusion and suggests that the crucial notions in the historical development of the coding split observed in present-day Mandinka were location, control, and animacy, rather than alienability.
The article is organized as follows. Section 2 provides the necessary background information about Mandinka morphosyntax. Section 3 provides basic data about the coding split in the adnominal possessive construction. Section 4 describes the contrast between direct and indirect adnominal possessive constructions with ordinary nouns in the role of head, and discusses the historical scenario that may have resulted in the situation observed in present-day Mandinka. Section 5 describes the contrast between direct and indirect adnominal possessive constructions with verbal lexemes used as event nouns in the role of head. Section 6 discusses the contribution of Mandinka to the discussion about the possible explanations of coding splits in the adnominal possessive construction.
2 The basics of Mandinka morphosyntax
2.1 Nouns and verbs
Nominal lexemes are characterized by their ability to project NPs whose possible functions include those of subject and object. Verbal lexemes are typically found in the V slot of the verbal predication construction (see Section 2.2), but with the exception of sǎa ‘die’ (which nominalizes as sàayáa ‘death’), they can also be used as event-denoting nouns without any specific morphological marking, with modifiers representing their core arguments. However, as will be developed in Section 5, verbal lexemes used as event-denoting nouns do not behave exactly like ordinary nouns in the adnominal possession construction.
2.2 Verbal predication
A brief overview of verbal predication is in order for a proper understanding of the particularities of adnominal possessive constructions with verbal lexemes used as event-denoting nouns in the role of head (Section 5).
The most striking characteristic of clause structure in Mande languages is the extreme rigidity of the typologically unusual S-O-V-X constituent order in verbal clauses, and Mandinka is no exception. No operation, such as focalization or questioning, can trigger change in constituent order, and with the exception of some types of adjuncts, noun phrases or adpositional phrases cannot occur in topic position (on the left edge of the clause) without being resumed by a pronoun occupying the position they would occupy if they were not topicalized.
In transitive clauses, as illustrated in (2), both the subject and the object obligatorily precede the verb, and the subject obligatorily precedes the object. Assertive and interrogative transitive clauses always include an auxiliary-like element inserted between the subject and the object, called a predicative marker in the Mandeist tradition. Predicative markers are portmanteau morphemes encoding aspectual and modal distinctions and expressing polarity.[4] Obliques follow the verb and are standardly encoded as postpositional phrases (2c)–(2d). The subject and the object bear no mark of their syntactic role and are not indexed on the verb (or elsewhere in the clause).[5]
| Jàtôo | yè | dánóo | bàràmá. |
| lion.D | CPL.TR | hunter.D | hurt |
| ‘The lion hurt the hunter.’ | |||
| Dánòo | yè | jàtóo | bàràmá. |
| hunter.D | CPL.TR | lion.D | hurt |
| ‘The hunter hurt the lion.’ | |||
| Ì | yè | bànkôo-lú | táláa | kàbíilòo-lú | lè | têemá. |
| 3PL | CPL.TR | land.D-PL | divide | clan.D-PL | FOC | between |
| ‘They divided the lands between clans.’ | ||||||
| Kèê | kà | à | téerímáa | máakóyì | kódòo | tó. |
| man.D | ICPL | 3SG | friend | help | money.D | LOC |
| ‘The man helps his friend financially.’ | ||||||
In intransitive clauses, exemplified in (3), the subject precedes the verb. Like transitive subjects, intransitive subjects bear no mark of their syntactic role and are not indexed. Obliques behave in exactly the same way in transitive and intransitive clauses. With the exception of the completive positive (encoded by the predicative marker yè in transitive clauses, and by the verbal suffix -tá ∼ -tà in intransitive clauses), aspect, modality, and polarity are encoded by the same predicative markers as in transitive clauses.
| Yíróo | bòyí-tà | sílòo | kâŋ. |
| tree.D | fall-CPL.ITR | road.D | on |
| ‘The tree fell down on the road.’ | |||
| Nèê | kà | kómôŋ | jíyòo | kónò. |
| iron.D | ICPL | rust | water.D | in |
| ‘Iron rusts in water.’ | ||||
| Kèê | mâŋ | kúmá | mùsôo | yé. |
| man.D | CPL.NEG | talk | woman.D | BEN |
| ‘The man did not talk to the woman.’ | ||||
Pronouns occupy the same positions as NPs consisting of a nominal lexeme and its modifiers, and show no variation related to their syntactic roles.
2.3 Noun phrase structure
The structure of Mandinka noun phrases can be schematized as follows, with two possible positions for determiners. The position labeled ‘Det.₁’ is occupied by demonstratives, whereas all other types of determiners occupy the position labeled ‘Det.₂’:
(Adnom.poss.) (Det.₁) N (Attr.) (Num.) (Det.₂)
Mandinka has no agreement mechanism between nouns and their modifiers, and apart from a classifying suffix -maa optionally found with kinship terms and a few other nouns denoting interpersonal relationships, Mandinka has nothing similar to the phenomena described cross-linguistically as classifiers, noun classes, or grammatical genders.
Mandinka has a determiner -ò, sometimes labeled ‘definite marker’, which is better characterized as a default determiner. It originates from the grammaticalization of a demonstrative, and at some stage in the history of Manding languages, it probably acted as a definite article. However, synchronically, in most contexts, it carries no particular semantic specification, and must simply be present if the speaker does not consider it useful to select a determiner with a more specific meaning. The combination of nouns with the default determiner tends to behave as the default form of nouns, whereas the absence of the default determiner must be licensed by grammatical features of the noun phrase or of the clause in which it occurs. In particular, Mandinka speakers invariably use the ò-form of nouns for citation.
2.4 Personal pronouns
The emphatic versus non-emphatic contrast distinguishes personal pronouns from all other nominals (Table 1).
Emphatic and non-emphatic forms of personal pronouns.
| n.emph. | emph. | |
|---|---|---|
| 1SG | ŋ́ | ń-tè |
| 2SG | í | í-tè |
| 3SG | à | à-té |
| 1PL | ŋ̀ | ǹ-tè-lú ∼ ǹ-tò-lú |
| 2PL | álí ∼ álú | álí-tè-lú ∼ álú-tò-lú |
| 3PL | ì | ì-tè-lú ∼ ì-tò-lú |
As illustrated in (4), the third-person pronouns encode no gender or animacy distinction.
| Kèê | yè | kódòo | díi | mùsôo | lá. |
| man.D | CPL.TR | money.D | give | woman.D | POSTP |
| ‘The man gave the money to the woman.’ | |||||
| À | yè | à | díi | à | lá. |
| 3SG | CPL.TR | 3SG | give | 3SG | POSTP |
| ‘He/she gave it/him/her to him/her.’ | |||||
As also illustrated in (4), personal pronouns share with the other nominals the absence of any morphological encoding of their syntactic role, nor do they occupy positions distinct from those occupied by other nominals fulfilling the same functions. The only differences in the syntactic behavior of emphatic and non-emphatic forms are that non-emphatic forms (i) cannot be focalized, and (ii) are prosodically dependent on the following word, which prevents them from accessing positions in which they would necessarily be followed immediately by a pause (in particular, they cannot feature as left-dislocated topics, nor can they be used in the vocative function).
3 The adnominal possessive construction: first observations
3.1 Direct versus indirect adnominal possessors
NPs in adnominal possessor function invariably precede their head. Mandinka has a distinction between direct adnominal possessors, simply juxtaposed to their heads, and indirect adnominal possessors, flagged by the postposition lá. At first sight, one might think that this is a standard case of the alienability distinction at work, since with prototypical possessive constructions, the choice between the two variants of the adnominal possessive constructions is compatible with an explanation in terms of alienability (Nichols 1988; Rose and Van linden this issue). As illustrated in (5), the relationships between individuals and their body parts or blood relatives (prototypically inalienable) are among those requiring the direct construction (5a)–(5b), whereas the indirect construction is obligatory when the head noun refers to a concrete object that the referent of the possessor has at his/her disposal (5c).
| mùsóo | kùŋôo |
| woman.D | head.D |
| ‘the woman’s head’ | |
| díndíŋó | màmàmúsòo |
| child.D | grandmother.D |
| ‘the child’s grandmother’ | |
| kèê | lá | sàmàtôo-lú |
| man.D | GEN | shoe.D-PL |
| ‘the man’s shoes’ | ||
However, as will be discussed in detail in Sections 4 and 5, the notion of alienability does not provide a global explanation of the choice between the two variants of the adnominal possessive construction.
Before turning to this question, several general properties of the adnominal possessive construction are worth mentioning: the lack of pronoun versus noun split, the lack of choice between flagged and unflagged possessors in some conditions, the distinction between the adnominal possessive construction and N + N compounds, and the fact that no Mandinka noun can be characterized as either obligatorily possessed or non-possessible.
3.2 No pronoun versus noun split
There is no pronoun versus noun split in the adnominal possessive construction of Mandinka. As mentioned above, in predicative constructions, pronouns occupy the same positions as noun phrases and show the same form in subject, object and oblique roles. The same applies to the adnominal possessive construction, in which pronouns fulfill the role of modifier exactly like noun phrases. For example, as illustrated in (6), the second person singular pronoun occurs as í (n.emph.)/ítè (emph.), not only in subject, object and oblique function, but also in modifier function in the adnominal possessive construction, and combines with the postposition lá in the same conditions as a noun fulfilling the same function.
| í / íté | kùŋôo |
| 2SG (n.emph. / emph.) | head.D |
| ‘your head’ | |
| í / íté | màmàmúsòo |
| 2SG (n.emph. / emph.) | grandmother.D |
| ‘your grandmother’ | |
| í / ítè | lá | sàmàtôo-lú |
| 2SG (n.emph. / emph.) | GEN | shoe.D-PL |
| ‘your shoes’ | ||
3.3 Configurations with no possible choice between flagged and unflagged possessors
In the following three configurations, there is no possible choice between flagged and unflagged possessors.
When the role of head is fulfilled by the proprietive pronoun tâa ‘that of X’, ‘what belongs to X’ (exclusively used in the role of head in the adnominal possessive construction), the direct construction is invariably used, whatever the understood noun and its semantic relationship to the possessor – Example (7);
| í | lá | sàmàtôo-lú |
| 2SG | GEN | shoe.D-PL |
| ‘your shoes’ | ||
| í | díŋò-lú |
| 2SG | child.D-PL |
| ‘your children’ | |
| í | tàa-lú |
| 2SG | PROPR-PL |
| ‘yours (pl.)’, whatever the understood noun: shoes, children, etc. | |
When the noun in the role of head in the adnominal possessive construction combines with fánsúŋ- or súndíŋ- ‘own’, the direct construction is invariably used, whatever the head noun and its semantic relationship to the modifier – Example (8);
| í | là | búŋò |
| 2SG | GEN | room.D |
| ‘your room’ | ||
| í | (*là) | fánsúm-búŋò |
| 2SG | own-room.D | |
| ‘your own room’ | ||
When the noun in the role of head in the adnominal possessive construction is also modified by a demonstrative, the indirect construction is invariably used, whatever the head noun and its semantic relationship to the modifier – Example (9).
| í | díŋkèe |
| 2SG | son.D |
| ‘your son’ | |
| í | *(lá) | ñǐŋ | díŋkèe |
| 2SG | GEN | DEM | son.D |
| ‘this son of yours’ | |||
3.4 Adnominal possessive construction and N + N compounds
A brief mention of N + N compounds is in order here, since in some languages the distinction between N + N compounds and the adnominal possessive construction is problematic.
Generally speaking, compounding is a very productive type of word formation in Mandinka. In particular, nouns may be modified not only by noun phrases referring to specific entities in the role of modifier but also by bare nominal lexemes expressing generic reference to kinds, with which they form compounds. Moreover, contrary to the adnominal possessive construction, the formation of compounds involves specific tone rules that modify the tonal contour of the lexemes that combine into a compound. Example (10c) shows that the distinction is ensured by tone in cases where the segmental form is ambiguous. Moreover, N + N compounds show no coding split comparable to that found in the adnominal possessive construction: in general, the two terms of the compound are simply juxtaposed, and the same tone rules apply, whatever the semantic relationship underlying the formation of a compound.
| mùsôo | lá | sàmàtôo | vs. | mùsù-sámátòo |
| woman.D | GEN | shoe.D | woman-shoe.D | |
| ‘the/a shoe of the/a woman’ | ‘woman’s shoe’ | |||
| ñàŋkúmòo | féñòo | vs. | ñàŋkùmà-féñòo |
| cat.D | tail.D | cat-tail.D | |
| ‘the/a tail of the/a cat’ | ‘cat’s tail’ | ||
| bàtànsée | fàtôo | vs. | bàtànsèe-fátòo |
| eggplant.D | peel.D | eggplant-peel.D | |
| ‘the/a peel of the/an eggplant’ | ‘eggplant peel’ | ||
In the perspective of the hypothesis according to which the original function of the indirect construction of adnominal possession was to specify the locational nature of the relationship between Possessor and Possessee (as developed in Section 6 below), it is interesting to mention that Mandinka also has N + Postposition + N compounds in which the postposition specifies the locational nature of the relationship between the two nouns, such as wúlá-kónó-séewòo /bush-in-pig/ ‘warthog’.
3.5 Obligatorily possessed nouns, non-possessible nouns
Mandinka has neither obligatorily possessed nouns nor non-possessible nouns, in the sense defined and commented on in Rose and Van linden (this issue). On the one hand, relational nouns (i.e., nouns that are prototypically possessed) can be used without any modifier, and are then interpreted as expressing reference to unspecified possessors, as ‘foot’ in (11) and ‘mother’ in (12), or to specific possessors whose identity can be retrieved from the context.
| Sìŋôo | bé | dǔumà, | sàâ | bé | dǔumà, |
| foot.D | LCOP | on.the.ground | snake.D | LCOP | on.the.ground |
| ‘The foot is on the ground, the snake is on the ground, | |||||
| wǒo | ñóodáŋò | mâŋ | díyâa. | ||
| DEM | avoiding.each.other.D | CPL.NEG | be.easy | ||
| they can hardly avoid each other.’ (proverb) | |||||
| Álà | lá | kànôo | hádámàdíŋò | yé, | ||
| God | GEN | love.D | human.being.D | BEN | ||
| ‘God’s love for mankind | ||||||
| wǒo | lè | wàrá-tà | bâa | lá | díŋ-kánòo | tí. |
| DEM | FOC | be.great-CPL.ITR | mother.D | GEN | child-love.D | POSTP |
| is greater than a mother’s love for (her own) children.’ | ||||||
On the other hand, in Mandinka, the nouns that are not prototypically possessible are found quite commonly as heads of adnominal possessive constructions expressing relationships other than prototypical possession, for example, í là lúŋò ‘your day’ in the sense of ‘the day that has a special significance for you’, í lá sàâ ‘your snake’ in the sense of ‘the snake that frightened you’, í lá nìŋkínànkôo ‘your dragon’ in the sense of ‘the dragon you mention in the story you are telling’, etc.
4 Direct and indirect adnominal possessive constructions with ordinary nouns in the role of head
In the analysis of the contrast between flagged and unflagged possessors, adnominal possessive constructions with a verbal lexeme used as an event noun in the role of head must be treated apart, since they put into play a very straightforward rule which at the same time cannot be deduced from the general regularities about the choice between flagged and unflagged possessors. The special behavior of adnominal possessive constructions with verbal lexemes used as event nouns in the role of head will be analyzed in Section 5, while this section will focus on ordinary nouns in the role of head.
4.1 Factors determining the variation between direct and indirect adnominal possessive constructions
The following three types of relationships are commonly considered prototypical for possessive constructions (see Rose and Van linden, this issue, and references therein):
the relationships between human individuals and their body parts;
the relationship between human individuals and their blood relatives;
the relationship between human individuals and the objects they possess according to the rules that regulate individual property in a given society.
Within the limits of these three types of relationships, as shown by Example (5) above, the notion of alienability correctly predicts the distribution of direct and indirect adnominal possessive constructions. However, the adnominal possessive construction is far from being limited to the expression of prototypical possession. In fact, it encodes nothing more than the existence of a privileged relationship of any kind between the referents of the head noun and of the NP in the role of modifier. The point is that, for the relationships other than those commonly considered as constituting the semantic core of the linguistic notion of possession, it immediately turns out that alienability cannot explain the distribution of flagged and unflagged possessors in the adnominal possessive construction of Mandinka.
For example, it is difficult to imagine how the notion of alienability could explain the contrast in (13).
| Sènèkáalì | pérésídáŋò | vs. | Sènèkàalì-ŋkôo-lú | là | pérésídáŋò |
| Senegal | president.D | Senegal-inhabitant.D-PL | GEN | president.D | |
| ‘the president of Senegal’ | ‘the president of the Senegalese’ | ||||
Conversely, if the notion of alienability were crucial in the distribution of direct and indirect adnominal possessive constructions, it should not be possible to use the same construction for ‘the country where he was born’ (unbreakable relationship) and ‘the country over which he reigns’ (breakable relationship), as in (14).
| à | lá | bàŋkôo |
| 3SG | GEN | country.D |
| ‘her/his country’, with reference to any conceivable relationship between a human individual and a country: ‘the country where s/he was born’, ‘the country over which s/he reigns’, etc. | ||
Similarly, with léetáròo ‘letter’, à lá léetáròo (where à is the third person singular pronoun) can refer to any conceivable relationship between a person and a letter: ‘the letter s/he wrote’, ‘the letter s/he asked me to write for her/him’, ‘the letter s/he received’, ‘the letter s/he is holding in her/his hands’, ‘the letter s/he delivered’, ‘the letter s/he is waiting for’, etc. In fact, with léetáròo ‘letter’, the direct construction is only possible with an inanimate modifier: à léetáròo ‘the letter that goes with it’.
Such examples could be multiplied indefinitely. A crucial observation is that, regarding the productivity of the two variants of the adnominal possessive construction, there is a striking contrast between Mandinka and most of the languages that code, for example, ‘my bicycle’ differently from ‘my head’ and ‘my brother’. In most of the languages in which such a contrast can be observed, as a rule, the nouns that can fulfill the role of head in the construction used for ‘my head’ and ‘my brother’ constitute a small closed class. By contrast, in Mandinka, both variants of the adnominal possessive construction are productive. In fact, any noun can fulfill the role of head in the direct adnominal possessive construction, depending on the nature of the modifier.
In Mandinka, the best predictor of the choice between direct and indirect adnominal possessive constructions is animacy of the modifier. With some exceptions that will be examined in Sections 4.2 and 4.3, the default rule is that inanimate possessors require the direct construction, whereas animate possessors (including non-human animates) require the indirect construction. What creates the illusion of an alienability contrast is that, for reasons that will be discussed in Section 4.3, the default rule does not apply to constructions referring to body parts or kinship terms.
4.2 The adnominal possessive construction with inanimate possessors
When an ordinary noun in the role of head of the adnominal possessive construction combines with an inanimate modifier, the general rule is that irrespective of the head noun and the semantic nature of the relationship between the two terms of the construction, the direct construction is the only possible option. The only exception to this rule is that the presence of a demonstrative determiner invariably triggers the use of the indirect construction, as in (15b).
| à | kótòo |
| 3SG | meaning.D |
| ‘its meaning’ (speaking of a proverb) | |
| à | lá | ñǐŋ | kótòo |
| 3SG | GEN | DEM | meaning.D |
| ‘this meaning it has’ (speaking of a proverb) | |||
4.3 The adnominal possessive construction with animate possessors
When an ordinary noun in the role of head of the adnominal possessive construction combines with an animate modifier, the default choice is the indirect construction. There are, however, two important groups of exceptions to this rule. Possible explanations will be discussed in Section 6.
4.3.1 Nouns referring to body parts and related notions
The direct construction is used when the head noun refers to a body part of the referent of the modifier, giving rise to contrasts such as (16).
| sàajíyòo | kúlòo |
| sheep.D | skin.D |
| ‘the skin of the sheep’ | |
| à | lá | sàajìi-kúlòo |
| 3SG | GEN | sheep-skin.D |
| ‘his/her sheepskin’ | ||
The same behavior is observed for fùrée ‘corpse’.
| à | fùrêe |
| 3SG | corpse.D |
| ‘his/her corpse’ | |
| à | lá | fùrêe |
| 3SG | GEN | corpse.D |
| ‘the corpse of the person he/she killed’ | ||
The direct construction is also found with animate possessors of the following groups of nouns:
nouns denoting physical characteristics of individuals, such as sémbòo ‘force’, tímínòo ‘endurance’,
nouns denoting bodily emanations or excretions such as nìiníyòo ‘shadow’, sìnnôo ‘footprint’, táròo ‘sweat’, ñáajíyòo ‘tears’, dáajíyòo ‘saliva’, séeròo ‘smell’,
nouns denoting non-physical constituent parts of animate beings such as níyòo ‘soul’, hákílòo ‘intelligence’, sóndómòo ‘conscience’, jíkòo ‘personality’, fàlôo ‘destiny’, tànôo ‘taboo’, and nouns derived from verbs referring to individual behavior, such as táamá-ñàa ‘way of walking’ < táamà ‘walk’,
nouns denoting social characteristics of individuals such as tôo ‘individual name’, kòntóŋò ‘family name’, síyòo ‘ethnicity’,
yàâ ‘residence’,
jàmáanòo ‘time when s.o. lived’
However, the behavior of some of the nouns expressing such notions is difficult to explain. For example, dáajíyòo ‘saliva’ requires a direct construction, but kàarôo ‘sputum’ requires an indirect construction. Similarly, it is difficult to explain the use of the direct construction for nìnsôo kéekèe ‘the milk of the cow’, contrasting with the use of the indirect construction for sìisêe lá kílòo ‘the egg of the hen’.
4.3.2 Kinship terms and nouns referring to other types of interpersonal relationships
Kinship terms (including those referring to kinship based on marriage), are the other group of nouns with which animate noun phrases in the role of modifier occur in the direct adnominal possessive construction. In Mandinka, kinship terms are optionally marked by the classifying suffix -máa, whose presence excludes definiteness marking. With kèê, which has the non-relational meaning of ‘man’ and the relational meaning of ‘husband’, this gives rise to the contrast illustrated in (18).
| à | kèê | ∼ | à | kèe-máa |
| 3SG | man.D | 3SG | man-KIN | |
| ‘her husband’ | ||||
| à | lá | kèê |
| 3SG | GEN | man.D |
| ‘his/her man’, i.e., a man having with him/her a relationship of any kind, except marriage (for example, ‘the man he/she is speaking about’) | ||
Interestingly, mùsôo ‘woman, wife’ behaves differently from kèê ‘man, husband’, and constitutes in fact the only possible exception to the rule according to which kinship terms, including those referring to kinship based on marriage, require the direct construction. With mùsôo, the direct construction is obligatory if the classifying suffix for kinship terms -máa is used (à mùsù-máa ‘his wife’, exactly like à kèe-máa ‘her husband’). However, when the form mùsôo is used, the indirect construction is required (à lá mùsôo ‘his wife’ or ‘his woman’).
Grégoire (1984) showed that the explanation lies in the grammaticalization process that resulted in the creation of the postposition lá and the extension of its use to the adnominal possessive construction. In present-day Mandinka, lá is a multifunctional postposition, but historically, it is cognate with nouns attested throughout the Mande family with meanings such as ‘mouth’, ‘opening’, ‘edge’. Consequently, it can safely be concluded that lá grammaticalized first as a spatial postposition, and that the indirect adnominal possessive construction results from a process of semantic bleaching that has affected lá in a construction X lá Y, whose original meaning was ‘the Y located near X’. Starting from that, the crucial observation is that, in traditional Mandinka society (as is often the case in sub-Saharan Africa), when a man and a woman marry, the wife moves to the family residence of her husband. Consequently, it makes sense for a Mandinka man to designate his wife as ‘the woman who is at my place’, whereas the reverse would not make sense. Consequently, the fact that mùsôo ‘wife’ is the only kinship term requiring the indirect adnominal possessive construction can be explained as a retention of the original meaning of the postposition used in this construction.
In addition to kinship terms, the direct adnominal possessive construction is also required by nouns referring to symmetrical interpersonal relationships, such as téeròo ‘friend’, jáwòo ‘enemy’, fúláŋò ‘peer’, and co-participant nouns derived from verbs such as táa-ñòo ‘fellow traveler’ < táa ‘go’.
Finally, there are three pairs of nouns referring to interpersonal relationships that show the same behavior as ‘husband’ and ‘wife’, and for which the same explanation can be considered:
jìyàatíyòo ‘host’ requires the direct construction, whereas lúntáŋò ‘guest’ requires the indirect construction;
kàràmmôo ‘teacher, professor’ requires the direct construction, whereas kàràndíŋò ‘pupil, student’ requires the indirect construction;
màaríyòo ‘master’ requires the direct construction, whereas jòŋôo ‘slave’ requires the indirect construction.
In the case of ‘host’ and ‘guest’, it is obvious that a host can designate his/her guest as ‘the guest who is at my place’, but not the other way round. Similarly, at the time when slavery existed, slaves used to live at their master’s place. In the case of kàràmmôo ‘teacher, professor’ and kàràndíŋò ‘pupil, student’, the explanation is that in traditional Koranic teaching, pupils live in their teacher’s home. Interestingly, kàràmmôo ‘teacher, professor’ and kàràndíŋò ‘pupil, student’ behave in the same way when used with reference to European-style teaching, in spite of the fact that, in the context of European-style teaching, the original motivation of their behavior in the adnominal possessive construction is lost.
All the other nouns that refer to interpersonal relationships, including those referring to superiors in hierarchical relationships, follow the general rule for adnominal possessive constructions involving animate modifiers. For example, kìntáŋò ‘circumcisee’s mentor’ requires the indirect construction, contrary to kàràmmôo ‘teacher, professor’, jìyàatíyòo ‘host’, or màaríyòo ‘master’. The behavior of kìntáŋò may surprise, given the hierarchical relationship between a circumcisee and his mentor, but it is consistent with the fact that circumcisees do not live at their mentor’s place. This confirms that what matters for the asymmetrical pairs (‘husband’/‘wife’, ‘guest’/‘host’, ‘master’/‘slave’, and ‘teacher’/‘pupil’) is not hierarchical superiority, but the determination of residence.
5 Direct and indirect adnominal possessive constructions with verbal lexemes used as event nouns in the role of head
When verbal lexemes used as event nouns fulfill the function of head in the adnominal possessive construction, the choice between direct and indirect construction is determined by a very straightforward syntactic rule in which, contrary to the general case, animacy plays no role.
Contrary to ordinary nouns, which can fulfill the role of head in adnominal possessive constructions expressing any possible kind of semantic relationship between head and modifier, verbal lexemes used as event nouns can only combine with possessors corresponding to the subject and/or the object of the corresponding clause. The rule, which suffers no exception, is that subjects (irrespective of the transitive vs. intransitive distinction, and of their possible semantic roles) are transposed as flagged possessors, whereas objects are transposed as unflagged possessors. No difference is made between intransitive subjects whose referent is relatively agent-like or relatively patient-like. Example (19b) illustrates the use of a transitive verbal lexeme as an event noun.
| Ñǐŋ | kèê | yè | díndíŋò-lú | dóoyâa | kúnùŋ. |
| DEM | man.D | CPL.TR | child.D-PL | scold | yesterday |
| ‘This man scolded the children yesterday.’ | |||||
| Ñǐŋ | kèê | lá | díndíŋò-lú | dóoyàa | mâŋ | díyáa | ŋ́ | yè. |
| DEM | man.D | GEN | child.D-PL | scold.D | CPL.NEG | be.pleasant | 1SG | BEN |
| lit. ‘This man’s scold(ing) of children is not pleasant for me.’ | ||||||||
| → ‘I don’t like the way this man scolds children.’ | ||||||||
Example (20) illustrates the fact that even inanimate subjects are transposed as flagged possessors, in spite of the general rule according to which the indirect adnominal possessive construction is normally impossible with inanimate possessors.
| Sáñjí-fólòo | ké-tà | lè. |
| rain-first.D | occur-CPL.ITR | FOC |
| ‘The first rain has arrived.’ | ||
| sáñjí-fólòo | lá | kêe |
| rain-first.D | GEN | occur(ring).D |
| ‘the arrival of the first rain’ | ||
6 Discussion and conclusions
The aim of this article was to discuss the possibility of a functional explanation of the coding split in the adnominal possessive construction of Mandinka on the basis of a comprehensive description of the choice between the direct and the indirect construction.
As already discussed, the locative origin of the postposition lá and the particular behavior of the four pairs of nouns ‘husband’/‘wife’, ‘host’/‘guest’, ‘master’/‘slave’, and ‘teacher’/‘pupil’ in adnominal possession constructions suggest that the lá-construction was used initially with reference to situations where the referent of the modifier (the possessor) determines the residence of the head (the possessee). Crucially, determining the residence of the possessee can be viewed as a particular case of a broader notion of control exerted by the possessor in its relation to the possessee. In order to explain the situation observed in present-day Mandinka, one may imagine an intermediary stage in which the use of the lá-construction was extended to situations characterized by the more abstract feature of control exerted by the possessor on various aspects of its relationship to possessee, i.e., on aspects of the relationship between the possessor and the possessee not limited to location. Starting from that, apart from a few pairs of nouns referring to interpersonal relationships that have maintained the original situation, the distinction grammaticalized differently depending on the grammatical nature of the head noun and the semantic nature of the modifier.
With verbal lexemes used as event nouns in the role of head, the choice is conditioned now by the syntactic role of the modifier in the corresponding clause, and the fact that objects are transposed as unflagged possessors, whereas subjects are transposed as flagged possessors, is consistent with the lack of control that characterizes the semantic roles typically expressed by objects, contrasting with the control that characterizes the semantic role of agent, prototypically associated with the syntactic role of subject.
With ordinary nouns (i.e., nouns other than verbal lexemes used as event-denoting nouns without any morphological marking) in the role of head, inanimate modifiers (i.e., modifiers whose referent is conceived as inherently devoid of the ability to control anything) have continued to occur exclusively in the direct construction, except in the presence of a demonstrative modifier, which may perhaps be explained by the high degree of individuation implied by the demonstrative.
Finally, with ordinary nouns in the role of head and animate modifiers, the fact that animates are inherently characterized by their ability to exert control on other entities resulted in that the lá-construction has become the default construction, with, however, two groups of exceptions, for which two kinds of explanation can be considered. On the one hand, configurations in which the head noun is particularly prone to fulfill the role of head in the adnominal possessive construction by virtue of its lexical meaning, which is the case for body part terms and kinship terms, may tend to resist innovation. On the other hand, the situations with reference to which animate modifiers do not require the lá-construction typically involve lack of control on the part of the referent of the modifier. In fact, these two explanations are not exclusive of each other.
According to Nichols (1988):
There is no invariant semantic content to ‘alienability’. It is simply a formal split in the marking of adnominal constructions, with the more fused or archaic of the two marking types associated with exactly those nouns that are most often possessed. (1988: 31)
The analysis of the Mandinka data confirms that coding splits in the adnominal possessive construction are not always analyzable as a reflection of the alienability contrast, but at the same time shows the limits of explanations in terms of discourse frequency (as proposed by Haspelmath 2017), and the need to admit that semantic features such as animacy and control may play an important role in the genesis of coding splits in the adnominal possession construction.
As regards the possible role of discourse frequency in the development of the coding split observed in the adnominal possessive construction of Mandinka, it is true that discourse frequency may have favored the maintenance of the direct construction with head nouns referring to body parts or kinship terms. However, discourse frequency effects cannot be invoked to explain that, in Mandinka, the direct construction was also maintained with inanimate modifiers, irrespective of the head noun. Quite obviously, adnominal possessive constructions with inanimate modifiers are much less frequent in discourse than adnominal possessive constructions with animate modifiers. Consequently, the fact that they have maintained the direct construction is clearly an exception to the trend towards heavier morphological marking in configurations characterized by a relatively low discourse frequency.
Acknowledgments
I am very grateful to the editors of this special issue of Linguistics and to the anonymous reviewers for their insightful comments and suggestions, which greatly helped improve the final version of this article.
Abbreviations
- Adnom.Poss.
-
adnominal possessor
- Attr.
-
attributive adjective
- BEN
-
benefactive
- CPL
-
completive
- D
-
default determiner
- DEM
-
demonstrative
- Det.
-
determiner
- emph.
-
emphatic
- FOC
-
focalization
- GEN
-
genitive
- ICPL
-
incompletive
- ITR
-
intransitive
- KIN
-
classifying suffix optionally added to kinship terms
- LCOP
-
locational copula
- LOC
-
locative
- N
-
noun
- NEG
-
negative
- n.emph.
-
non-emphatic
- NP
-
noun phrase
- Num.
-
numeral
- O
-
object
- PL
-
plural
- POSTP
-
postposition[6]
- PROPR
-
proprietive pronoun
- S
-
subject
- SG
-
singular
- TR
-
transitive
- V
-
verb
- X
-
oblique
References
Chappell, Hilary & William McGregor. 1989. Alienability, inalienability and nominal classification. Proceedings of the Berkeley Linguistic Society 15. 24–36. https://doi.org/10.3765/bls.v15i0.1734.Search in Google Scholar
Chappell, Hilary & William McGregor. 1996. Prolegomena to a theory of inalienability. In Hilary Chappell & William McGregor (eds.), The grammar of inalienability, 3–30. Berlin & New York: Mouton de Gruyter.10.1515/9783110822137.3Search in Google Scholar
Creissels, Denis & Pierre Sambou. 2013. Le mandinka: Phonologie, grammaire, textes. Paris: Karthala.Search in Google Scholar
Dimmendaal, Gerrit. 2011. Historical linguistics and the comparative study of African languages. Amsterdam & Philadelphia: John Benjamins.10.1075/z.161Search in Google Scholar
Grégoire, Claire. 1984. Le syntagme déterminatif en mandé nord. Journal of African Languages and Linguistics 6(2). 173–193.Search in Google Scholar
Haspelmath, Martin. 2017. Explaining alienability contrasts in adpossessive constructions: Predicatiblity vs. iconicity. Zeitschrift für Sprachwissenschaft 36(2). 193–231. https://doi.org/10.1515/zfs-2017-0009.Search in Google Scholar
Nichols, Johanna. 1988. On alienable and inalienable possession. In William Shipley (ed.), In honor of Mary R. Haas, 557–609. Berlin & New York: Mouton de Gruyter.10.1515/9783110852387.557Search in Google Scholar
Vydrin, Valentin. 2009. On the problem of the Proto-Mande homeland. Вопросы языкового родства – Journal of Language Relationship 1. 107–142.Search in Google Scholar
© 2023 the author(s), published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 International License.
Articles in the same Issue
- Frontmatter
- Introduction: the limits of the explanatory potential of the alienability contrast
- Multi-variate coding for possession: methodology and preliminary results
- Categorizing possession in Zuanga-Yuanga and other Kanak languages (New Caledonia): a typological perspective
- Beyond alienability: factors determining possessive classes in Piaroa
- Questioning the relevance of alienability in Arawak linguistics: an innovative analysis of possession in Mojeño Trinitario
- When the alienability contrast fails to surface in adnominal possession: bound nouns in Harakmbut
- Non-canonical possessive constructions in Negidal and other Tungusic languages: a new analysis of the so-called “alienable possession” suffix
- Coding splits in the adnominal possessive construction and alienability: the case of Mandinka (West Mande)
- Explaining alienability splits in the use of overt and zero possessive marking: a source-oriented approach
Articles in the same Issue
- Frontmatter
- Introduction: the limits of the explanatory potential of the alienability contrast
- Multi-variate coding for possession: methodology and preliminary results
- Categorizing possession in Zuanga-Yuanga and other Kanak languages (New Caledonia): a typological perspective
- Beyond alienability: factors determining possessive classes in Piaroa
- Questioning the relevance of alienability in Arawak linguistics: an innovative analysis of possession in Mojeño Trinitario
- When the alienability contrast fails to surface in adnominal possession: bound nouns in Harakmbut
- Non-canonical possessive constructions in Negidal and other Tungusic languages: a new analysis of the so-called “alienable possession” suffix
- Coding splits in the adnominal possessive construction and alienability: the case of Mandinka (West Mande)
- Explaining alienability splits in the use of overt and zero possessive marking: a source-oriented approach