nueva página del texto (beta)
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
1. Introduction
The grammaticalization of nouns and verbs into adpositions has been widely discussed in the grammaticalization literature. This paper is about a much less well known grammaticalization path, by which deictic spatial adverbs (i.e., words such as English here or there) are converted into adpositions. In this grammaticalization process, the source construction is the LOCATIVE APPOSITION CONSTRUCTION, defined as a construction consisting of the juxtaposition of two coreferential spatial expressions, a deictic spatial adverb, and a spatial expression whose nucleus is a noun (as in English here in the village or there on the table).
In section 2, I discuss evidence for this grammaticalization path in Tswana (Bantu). In Tswana, three words that were originally deictic spatial adverbs, and still have the ability to act as deictic spatial adverbs (fa [fá] ‘here’, ko [kó] ‘there’, and mo [mó] ‘in this/that’), are also found in constructions in which they can be analyzed as emerging prepositions.1 Moreover, mo [mó] also has non-spatial uses in which its behavior is that of a full-fledged preposition.
In section 3, I propose that this grammaticalization process is the consequence of the routinization of the LOCATIVE APPOSITION CONSTRUCTION (here in the village, there on the table), i.e., the tendency toward extending its use to contexts where the precision added by the deictic spatial adverb has no communicative relevance.
In section 4, I show that Jóola Fóoñi (Atlantic) also has spatial prepositions showing evidence of resulting from the grammaticalization of deictic spatial adverbs in the LOCATIVE APPOSITION CONSTRUCTION.
In section 5, I evoke the case of other languages showing an unusual frequency of constructions combining a deictic spatial adverb and a non-decitic locative expression referring to the same place, which constitutes the initial stage of the grammaticalization path DEICTIC LOCATIVES > SPATIAL ADPOSITIONS.
Section 6 summarizes the main conclusions.
2. Preposition-like uses of deictic spatial adverbs in Tswana2
2.1. General remarks on locative marking in Sotho-Tswana varieties
Tswana (aka Setswana) is a Bantu language belonging to the Sotho-Tswana dialect cluster. It is spoken by approximately 5 million speakers in Botswana and South Africa.
A remarkable characteristic of Sotho-Tswana, shared by the other Southern Bantu languages, is the loss of the distinction between several locative genders,3 and a system of locative marking that, contrary to the situation found in other groups of Bantu languages, does not interfere with gender agreement.4
Tswana, like the other Sotho-Tswana varieties, has a single locative gender including very few nouns (in Tswana, the only nouns belonging to the locative gender are golo [χʊ̀lɔ̀] and felo [fὶlɔ̀], both meaning ‘place’), and two locative markers that convert noun phrases into locatives, i.e., expressions fulfilling the role of ground in spatial relationships.
As illustrated in (1), the locative markers attach to the first word of the noun phrase they convert into a locative. One of them (-ng [-ŋ̀]) is a suffix, the other (go [χʊ́-]) is a prefix.5
(1) a. lokwalo lole lo lotona
lʊ̀kwálɔ̀ lólé ló lʊ̀tʊ́ná
book(11) that(11) lk(11) big(11)
‘that big book’
b. lokwalong lole lo lotona
lʊ̀kwáló-ŋ̀ lólé ló lʊ̀tʊ́ná
book(11)-loc that(11) lk(11) big(11)
‘on/in that big book’
c. go lole lo lotona
χʊ́-lólé ló lʊ̀tʊ́ná
loc-that(11) lk(11) big(11)
‘on/in that big one’
The choice between the two locative markers is entirely determined by the grammatical nature of the first word of the noun phrase they attach to, without any semantic implication. If the first word of the noun phrase is a common noun, with a few exceptions, the suffix -ng [-ŋ̀] is selected. In all the other cases (including pronouns and proper names), the conversion of a noun phrase into a locative is marked by the prefix go [χʊ́-].
Spatial adverbs, toponyms, and a few common nouns (including golo [χʊ̀lɔ̀] and felo [fὶlɔ̀] ‘place’) fulfill the role of ground in spatial relationships without taking locative marking - example (2).
(2) Ke ba bonye Gaborone maabane.
kὶbábóɲί χàbʊ́rónὶ mààbâːnὶ.
I.saw.them(2) Gaborone yesterday
‘I saw them in Gaborone yesterday.’
Locative marking does not occur in complex locatives either. Complex locatives are head-modifiers constructions in which the noun phrase in the role of ground modifies a word that can also be used by itself as a non-deictic spatial adverb, such as morago [mʊ̀ráχʊ́] ‘behind’ or godimo [χʊ̀dímʊ́] ‘up’. The adverbs in question are former locative gender nouns that have lost the ability to act as the nucleus of noun phrases, and this explains why, in complex locatives, the noun phrase in ground role is flagged by the same genitival linker as nouns in the role of adnominal possessor, and the genitive linker in complex locatives expresses locative gender agreement (or ‘class 17 agreement’ according to the terminology traditionally used in Bantu studies) - example (3).
(3) a. lobota lwa ntlo
lʊ̀bɔ̀tà lwá-ńtɬʊ́
wall(11) gen(11)-house(9)
‘the wall of the house’
b. morago ga ntlo
mʊ̀ráχʊ̀ χá-ńtɬʊ́
behind gen(17)-house(9)
‘behind the house’
As illustrated in (4), locative marking is not sensitive to the distinction between static location, source of motion, and destination of motion. With motion verbs, the assignment of the role of source of motion or destination of motion to locatives is a lexical property of verbs.
(4) a. O ile nokeng.
ʊ́ìlé nʊ̀kêː-ŋ̀
(s)he(1).has.gone river(9)-loc
‘(S)he has gone to the river.’
b. Ba tswa nokeng.
bátswá nʊ̀kêː-ŋ̀
they(2).come.from river(9)-loc
‘They come from the river.’
c. Re tlaa šapa nokeng.
rὶtɬààʃápá nʊ̀kêː-ŋ̀
we.shall.swim river(9)-loc
‘We shall swim in the river.’
However, applicative derivation can modify the role assigned to locatives by motion verbs. For example, a locative complement of huduga [húdúχá] ‘change residence’ (5a) can only be interpreted as referring to the source, whereas with the applicative derivate hudugela [húdúχέlà], destination is the only possible interpretation.
(5) a. Ba tlaa huduga Kanye
bátɬáàhúdúχá kàːɲέ
they(2).will.change.residence Kanye
‘They will move from Kanye.’
b. Ba tlaa hudugela Kanye.
bátɬáàhúdúχέlà kàːɲέ
they(2).will.change.residence.appl Kanye
‘They will move to Kanye.’
2.2. The locative prepositions of Tswana
One of the properties that distinguish Tswana from the other Sotho-Tswana varieties is that, in Tswana, in addition to the locative markers mentioned in §2.1, locatives are usually also flagged by one of the three prepositions fa [fá], ko [kó], or mo [mó]. As illustrated in (6), this equally concerns locatives marked by the suffix -ng [-ŋ̀], as in (6a), locatives marked by the prefix go [χʊ́-], as in (6b), unmarked locatives (i.e., noun phrases having the ability to fulfill the role of ground in spatial relationships without locative marking), as in (6c), and complex locatives, as in (6d).
(6) a. Ke ile (ko) motseng.
kὶìlé (kó) mʊ̀tsί-ŋ̀
I.went loc village(3)-loc
‘I went to the village.’
b. Ke ile (ko) go rre.
kὶìlé (kó) χʊ́-r̀ːrέ
I.went loc loc-my.father(1)
‘I went to my father’s.’
c. Ke ile (ko) Gaborone.
kὶìlé (kó) χàbʊ́rôːnὶ
I.went loc Gaborone
‘I went to Gaborone.’
d. Lenong le tshela (ko) godimo ga dithaba.
lὶnòŋ̀ lίtsʰίlà (kó) χʊ̀dímʊ̀ χá-dítʰâːbà
vulture(5) it(5).lives loc up gen(17)-mountains(10)
‘Vultures live on the top of mountains.’
As a rule, the locative prepositions are optional, but in spontaneous discourse, locatives preceded by prepositions are much more frequent than prepositionless locatives.6 To take just an example, of the 49 locative expressions found in the first two pages of the entry tshimo (‘field’) of the Tswana-English Glosbe Online Dictionary,7 42 are introduced by one of the three locative prepositions, whereas only 7 are prepositionless.
It is also symptomatic that, in elicitation sessions, consultants tend to use the locative prepositions systematically, but if asked about their obligatoriness / optionality, they almost always answer that the construction is still grammatical if the preposition is deleted.
Contrary to the locative markers, which are in complementary distribution and give no precision about the nature of the spatial relationships, the locative prepositions are sensitive to some semantic distinctions. Cole (1955: 341) describes their meaning as follows:
- ko [kó] “implies relative remoteness, and therefore indicates that the specified locality at, to or from which the action is effected or directed is relatively distant”;
- fa [fá] “implies relative proximity, and therefore indicates that the specified locality at, to or from which the action is effected or directed is relatively nearby”;
- mo [mó] “indicates that the action is effected in, on or round, or directed into or out of, the specified locality, without reference to the distance involved”.
2.3. The locative prepositions of Tswana as reflexes of the Proto-Bantu demonstratives of locative genders
Demuth & Mmusi (1997) erroneously analyzed the locative prepositions of Tswana as noun prefixes that, according to their analysis, would be the reflexes of the Proto-Bantu prefixes of nouns belonging to one of the reconstructed locative genders 16, 17, and 18. As discussed in detail in Creissels (2011), this analysis must be rejected, both from the point of view of the synchronic status of fa [fá], ko [kó] and mo [mó], and from the point of view of their etymology:
- synchronically, fa [fá], ko [kó] and mo [mó] do not have the tonal behavior of prefixes, but of words; for example the downstep in ko go bontšhe [kó !χʊ́bóɲ̀tʃʰé] ‘on the ostriches’ should not occur if ko [kó] were a prefix (Creissels 2011: 40);
- historically, as already acknowledged by Cole (1955: 341), fa [fá], ko [kó] and mo [mó] can only be the reflexes of the Proto-Bantu demonstratives of locative genders.
In present-day Tswana, fa [fá], ko [kó] and mo [mó] are still used as demonstrative determiners with the two nouns that constitute the sole locative gender of Tswana. In this use, they all behave synchronically as belonging to sole locative gender of Tswana, but carry the same semantic implications as when they act as prepositions: (7a) implies distant location, (7b) implies nearby location, and (7c) implies reference to interiority. In (7a-b-c), the subject phrase is invariably resumed in the verb form by the subject index of the sole locative gender of Tswana go [χʊ́-], regardless of the choice of the demonstrative.
(7) a. Golo ko ga go a siama.
χʊ̀lɔ̀ kó χàχʊ́àsíâːmà
place(17) dem(17) it(17).is.not.good
‘That place is not good.’
b. Golo fa ga go a siama.
χʊ̀lɔ̀ fá χàχʊ́àsíâːmà
place(17) dem(17) it(17).is.not.good
‘That place is not good.’
c. Golo mo ga go a siama.
χʊ̀lɔ̀ mó χàχʊ́àsíâːmà
place(17) dem(17) it(17).is.not.good
‘That place is not good.’
In Bantu languages, the demonstratives of locative genders are also used as deictic spatial adverbs, i.e., as the equivalents of English here and there. Consequently, it comes as no surprise that in Tswana, fa [fá], ko [kó] and mo [mó] are also found in contexts such as those illustrated in (8), in which they can only be analyzed as deictic spatial adverbs that can be glossed ‘here’, ‘there’ and ‘in this/that’, respectively.
(8) a. Ba tswa ko.
bá-tswàː kó
they(2).are.coming.from there
‘They are coming from there.’
b. Ba tla fa.
bá-tlàː fá
they(2).are.coming here
‘They are coming here.’
c. Peba e tsenye mo.
pὲbà ίtséɲὶː mó
mouse(9) it(9).entered in.this/that
‘The mouse entered there.’
The possibility of using fa [fá], ko [kó] and mo [mó] as locative deictic adverbs raises the question of whether the constructions in which they act as prepositions (as in example (6) above) should not be rather analyzed as constructions in which a deictic spatial adverb is in apposition to a non-deictic locative (‘there in the village’, ‘there at my father’s’, ‘there in Gaborone’, ‘there on the top of the mountains’). The reason for rejecting this analysis is that, as illustrated in (9), fa [fá], ko [kó] and mo [mó] as deictic spatial adverbs can be used in apposition to locatives, but then, they are postposed, and their presence has no incidence on the optional use of the corresponding preposition. Semantically, the difference between (9a) and (9b) is that (9b) insists on the communicative relevance of the deictic meaning carried by ko [kó] as an adverb.
(9) a. (ko) motseng
(kó) mʊ̀tsί-ŋ̀
loc village(3)-loc
‘in the village’
b. (ko) motseng ko
(kó) mʊ̀tsί-ŋ̀ kó
loc village(3)-loc there
‘in the village there’
Moreover, the loss of the original deictic semantics of the locative prepositions is evidenced both by bilingual texts and the reaction of consultants in elicitation. In bilingual texts such as the Tswana-English Glosbe Online Dictionary, in the majority of cases, the English equivalents of Tswana sentences including locative expressions introduced by prepositions are devoid of any indication of spatial deixis. Similarly, in elicitations sessions, consultants tend to systematically use the locative prepositions, even when asked to give the Tswana equivalents of English sentences including no indication of spatial deixis.
To conclude, historical and synchronic data converge in supporting the conclusion that Tswana has spatial prepositions resulting from the grammaticalization of words that originally were demonstratives of locative genders also used as deictic spatial adverbs.
2.4. The particular case of the preposition mo [mó]
Given the optionality of the spatial prepositions of Tswana as described in §2.2., it may be argued that they should be characterized as emerging prepositions, rather than full-fledged prepositions. In this perspective, it is interesting to mention that a property distinguishing the preposition mo [mó] from the other two spatial prepositions is that it also has non-spatial uses in which it is obligatory, and consequently behaves as a full-fledged preposition assigning locative marking to its complement. For example, in (10), mo legodung [mó lίχòdù-ŋ̀] does not answer the question kae [káί] ‘where?’, but rather mo go mang [mó !χʊ́-máŋ́] ‘from whom?’, and in this construction, the preposition is obligatory.
(10) Ga ke reke dikgomo mo legodung.
χàkίrékί díqʰòmʊ́ !mó lίχòdù-ŋ̀
I.don’t.buy cows(10) loc thief(5)-loc
‘I don’t buy cows from a thief.’
This is also the case when mo [mó] is used to flag a noun phrase in the role of standard of comparison, as in (11).
(11) Pitse e tona mo tonking.
pìtsί !ίtʊ́ná !mó !tóŋ́kîː-ŋ̀
horse(9) it(9).is.big loc donkey(9)-loc
‘Horses are bigger than donkeys.’
Interestingly, in this use, mo [mó] requires locative marking even for the nouns (for example, toponyms) that have the ability of fulfilling the role of ground in spatial relationships without overt locative marking. For example, the addition of a locative affix to the toponym Kanye in (12a) would result in ungrammaticality, whereas in (12b), the locative prefix is obligatory.
(12) a. Kitso o agile (ko) Kanye.
kítsɔ́ !ʊ́áχílè (kó) kàːɲέ.
Kitso he(1).has.built loc Kanye
‘Kitso has built a house in Kanye.’
b. Gaborone o motona mo go Kanye.
χàbʊ́rónὶ ʊ́mʊ̀tʊ́ná !mó χʊ́-kàːɲέ.
Gaborone it(1).is.big loc loc-Kanye
‘Gaborone is bigger than Kanye.’
This observation confirms that, in its non-spatial uses, mo [mó] has gone a step further in the grammaticalization process.
3. Analysis: the grammaticalization process
As observed by Hopper & Traugott (1993: 201), a crucial aspect of grammaticalization is the process of routinization of language, by which “a form or a combination of forms occurs in discourse with increasing frequency, and from being an “unusual” way of making or reinforcing a discourse point comes to be the “usual” and unremarkable way to do so”.
The literature on grammaticalization abounds in examples of grammaticalization processes whose starting point is an increasingly frequent use of a given construction, which implies an increasing tendency to extend the use of the construction in question to contexts in which it has little communicative relevance. This routinization of the use of a construction initially motivated by the desire to express a precise semantic or pragmatic value, results in a weakening of the semantic or pragmatic value carried by the construction, which paves the way for further evolutions.
The reasons why a process of routinization affects a particular construction at a particular point in the evolution of a given language are poorly understood. I have nothing to propose about the reasons why the grammaticalization process I hypothesize for Tswana did not occur in the other Sotho-Tswana languages. However, this does not alter the fact that the importance of routinization processes in language evolution is hardly debatable.
In the present case, the lack of historical documentation precludes an analysis whose details could be supported by empirical data. It is, however, reasonable to hypothesize that the acquisition of the status of preposition by words that were initially demonstratives of locative genders also used as deictic spatial adverbs results from the routinization of a source construction I propose to designate as LOCATIVE APPOSITION CONSTRUCTION, in which two locative expressions referring to the same place are juxtaposed to each other: a locative expression whose nucleus is a noun and a deictic spatial adverb (here in the house, there at the river).
The LOCATIVE APPOSITION CONSTRUCTION is certainly available in a great number of languages. The reason why it is rarely mentioned in descriptive grammars is that, in its literal interpretation, it cannot be expected to be very frequent in discourse, since it normally implies that, for reasons of emphasis, the speaker decides to express the description of a place and a deictic specification or the same place separately (here in the house), rather than by inserting a demonstrative modifier (in this house).
With this background, it is reasonable to assume that the initial stage in the grammaticalization process that resulted in the situation described in §2 was the routinization of the LOCATIVE APPOSITION CONSTRUCTION, i.e., the extension of its use to contexts in which there is no need to insist on the precision added by the deictic spatial adverb. The generalization of this use of deictic spatial adverbs results in that deictic spatial adverbs accompanying non-deictic locative expressions tend to lose their marked deictic value and to act as prepositions whose presence is required regardless of questions of communicative relevance. In other words, the LOCATIVE APPOSITION CONSTRUCTION tends to be reanalyzed as a prepositional phrase with the former deictic spatial adverb in the role of preposition. However, in Tswana, of the three deictic spatial adverbs engaged in this evolution, mo [mó] is the only one for which the process has been completed, at least in its non-spatial uses. For the time being, the other two spatial prepositions, as well as mo [mó] in its locative uses, can only be characterized as emergent prepositions.
4. Prepositions originating from deictic spatial adverbs in Jóola Fóoñi (Atlantic)
Jóola Fóoñi, an Atlantic language spoken in South Western Senegal by approximately half a million speakers, is another language showing compelling evidence that the routinization of the LOCATIVE APPOSITION CONSTRUCTION may result in the grammaticalization of deictic spatial adverbs into full-fledged adpositions. The details are slightly different, since in Jóola Fóoñi, the conversion of deictic spatial adverbs into prepositions was accompanied by a reduction of their form, and by the deletion of the preposition that flagged the non-deictic locative expression in the source construction.
Jóola Fóoñi has a multifunction preposition dι which, in addition to being used as a comitative preposition, is widely used as a spatial preposition. In its spatial use, illustrated in (13), dι implies no further specification of the relationship between the figure and the ground.
(13) a. kupupur dι esukey.
they.went.out.from loc the.village
‘They left the village.’
b. nιrεmbεnrεmbεn katɔjak dι kɐkubeerɐk.
I.put the.leaf loc the.cover
‘I put the leaf on the cover.’
c. najajaw dι ɐmoori.
(s)he.went loc soothsayer
‘(S)he went to a soothsayer.’
d. nιtɔɔkɔɔtɔɔk dι εlʊʊpεy.
I.found.him/her loc the.house
‘I found him/her at home.’
In the corpus of naturalistic texts analyzed by Alain Christian Bassène and myself for the purposes of our project of writing a reference grammar of Jóola Fóoñi, three other spatial prepositions occur with a very high frequency in contexts in which dι could equally be used: baa, taa, and dɐɐ.8 The first two specify the location as relatively vague (baa) or relatively precise (taa), whereas dɐɐ can be glossed as ‘in’.
(14) a. nιtɔɔkɔɔtɔɔk baa bιιtab.
I.found.him/her loc the.rice.field
‘I found him/her at the rice-field.’
b. εyenεy εñʊñʊb taa fulimetɐf.
the.dog it.crouched.down loc the.door
‘The dog crouched down at the door.’
c. kʊrεmbεnrεmbεn εmaanaay dɐɐ buntuŋɐb.
they.put the.rice loc the.granary
‘They put the rice in the granary.’
Dι and baa / taa / dɐɐ are largely interchangeable, with, however, an interesting difference in their distribution: toponyms can occur in the role of ground either unmarked or combined with baa / taa / dɐɐ, but not in combination with dι. Another interesting observation is that our corpus includes a few occurrences of baa / taa / dɐɐ followed by dι.
Given the high frequency of baa / taa / dɐɐ in our corpus, it is remarkable that none of the available descriptions of Jóola Fóoñi (Weiss 1939, Sapir 1965, Hopkins 1995) acknowledges their existence. They are not mentioned in Sapir et al. (1993) dictionary either. However, the fact that Sapir (1965) explicitly mentions the existence of the LOCATIVE APPOSITION CONSTRUCTION suggests that this construction, which is quite rare in our own corpus, prominently featured in the data he analyzed.
The deictic spatial adverbs of Jóola Fóoñi are given in Table 1. Their first syllable coincides with the three prepositions illustrated in (14), and they express the same distinction between vague location, precise location, and interiority.
Table 1 The deictic spatial adverbs of Jóola Fóoñi
| proximal | unspecified for distance | distal | |
| vague location | baabε | baabʊ | baaba |
| precise location | taatε | taatʊ | taata |
| interiority | dɐɐre | dɐɐru | dɐɐrɐ |
The examples of the LOCATIVE APPOSITION CONSTRUCTION given by Sapir (1965) and Sapir et al. (1993) are quoted in (15). Although Sapir does not comment on this point, it is interesting to observe that the preposition dι, which would have been obligatory in the absence of the deictic spatial adverb, becomes optional in the LOCATIVE APPOSITION CONSTRUCTION (15f), which paves the way to the reanalysis of the deictic spatial adverb as a preposition.
(15) a. baabε dι esukey
here loc the.village
‘here in the village’
b. taata dι fankaf
there loc the.courtyard
‘there on the courtyard’
c. baaba dι falaf
there loc the.river
‘there by the river’
d. dɐɐrɐ dι kalιmbιsak
there loc the.room
‘there in the room’
e. baabʊ Dakaar
there Dakar
‘in Dakar’
f. baabε εlʊʊpεy
here the.house
‘here at the house’
This suggests that the spatial prepositions occurring with a high frequency in our corpus result from the reduction and reanalysis of the LOCATIVE APPOSITION CONSTRUCTION. At some point in the evolution of the LOCATIVE APPOSITION CONSTRUCTION, the deictic spatial adverb lost its second syllable (and consequently, the indication it provided about deixis), and the preposition dι became optional, so that what was initially the reduced form of a deictic spatial adverb in the LOCATIVE APPOSITION CONSTRUCTION was eventually reanalyzed as a preposition:
5. Other languages showing evidence of the grammaticalization path DEICTIC SPATIAL ADVERB > ADPOSITION
5.1. ’Eeja ‘there’ as an incipient postposition in Hoocąk
In his account of grammaticalization in Hoocąk (Siouan), Helmbrecht (2020: 920-921) describes a construction in which the deictic spatial adverb ’eeja ‘there’ can be viewed as an “incipient postposition”. Contrary to the neighboring Siouan languages, Hoocąk does not have postpositions. As illustrated in (16), spatial expressions can be formed by combining the noun in the role of ground (here xee ‘hill’) with a relational noun specifying the kind of spatial relation that holds between the figure and the ground (here hihak ‘top’), plus ’eeja ‘there’, as in (16).
(16) xee hihak ’eeja
hill top there
‘on top of a hill’
The Hoocąk locative constructions with a relational noun “obligatorily require a general locative marker ’eeja ‘there’, which is on the way to become a postposition. This rule applies also for toponyms. ’Eeja ‘there’ is originally a local adverb which contrasts deictically with ’eegi ‘here’ and is still frequently used adverbially. The deictic contrast disappears in constructions like the one in (16)” (Helmbrecht 2020: 921).
5.2. Languages with an unusual frequency of the LOCATIVE APPOSITION CONSTRUCTION
An unusual frequency of the LOCATIVE APPOSITION CONSTRUCTION in discourse is mentioned in descriptions of the following languages: Baule (Kwa), Beng (Mande),9 and Classical Nahuatl.
Example (17), from Creissels & Kouadio’s (1977: 364-365) grammar, illustrates the tendency toward routinization of the LOCATIVE APPOSITION CONSTRUCTION in Baule.
(17) a. ɔ̀ fìn Ábìjân (lɔ̀).
(s)he comes Abidjan (there)
‘(S)he is coming from Abidjan.’
b. ɔ̀ bòlì í kòfí í nyrún (lὲ).
(s)he hit him/her Kofi his face (here)
‘(S)he hit him/her in the presence of Kofi.’
As regards Beng, Paperno (2014: 55) observes that “A locative phrase consists of either a locative noun, which may or may not have syntactic dependents on the left, or a noun phrase with a locative postposition. On top of that, locative phrases also usually have a deictic marker on the right edge” (emphasis mine). Example (18) is from Paperno (2014: 56).
(18) Ó dɔ́ lɔ́ɔ́ wó we̅.
(s)he.cpl stop market in there
‘(S)he stopped at the market.’
Similarly, Launey (1981: 55-56) emphasizes the frequency of the LOCATIVE APPOSITION CONSTRUCTION in Classical Nahuatl. In this construction, non-deictic locatives are accompanied by nica̅n ‘here’ or o̅mpa ‘there’.
The routinization of the LOCATIVE APPOSITION CONSTRUCTION observed in such languages is precisely the situation which, according to the analysis put forward in §3, creates the conditions for a possible reanalysis of deictic spatial adverbs as adpositions.
6. Conclusion
In this paper, I have analyzed data from two sub-Saharan languages (Tswana and Jóola Fóoñi) that support the hypothesis according to which, in addition to the grammaticalization scenarios already discussed in the general literature on grammaticalization, the creation of adpositions may also result from a scenario whose initial stage is the routinization of the LOCATIVE APPOSITION CONSTRUCTION, eventually leading to its reanalysis as an adpositional phrase in which the role of adposition is fulfilled by the former deictic spatial adverb. In §5, after mentioning Hoocąk (Siouan) as another language in which the possibility of this grammaticalization path has been recognized, I have shown that a situation that could constitute the initial stage in the grammaticalization process converting deictic spatial adverbs into postpositions can be observed in some other languages.
