Atlas of Pidgin and Creole Language Structures Online

1. Introduction ⇫ ¶

The archipelago of Cape Verde is situated appro­xi­mate­ly 500 km to the west of Dakar (Senegal) in the At­lantic Ocean. Among its nine inhabited islands, Santiago in the Leeward group (Sotavento) is the most important by surface (991 sq km), population (some 250,000 people), and history (from the discovery around the year 1460 onwards). Both the old capital (Cidade Velha, previously known as Ribeira Grande) and the new capital (Praia) are located on Santiago.

     For historical reasons, the Portuguese-based San­tia­go Creole is closely related to the creole va­rie­ties on the archipelago’s other islands (see Baptista (2013) on Cape Verdean Creole of Brava, and Swolkien (2013) on Cape Verdean Creole of São Vicente), and also to the Por­tuguese-based creoles of Guinea-Bissau and Casa­man­ce (see Intumbo et al. (2013) and Biagui & Quint (2013)).

     Santiago Creole is the mother tongue of almost all of the island’s inhabitants. Another 250,000 speakers of Santiago Creole live in diaspora communities around the Atlantic Ocean, mainly in the USA (Boston), São Tomé and Príncipe, Portugal, Angola, Senegal, and the Netherlands.

2. Sociohistorical background ⇫ ¶

The discovery of the Cape Verdean archipelago between 1456 and 1460 provided the Portuguese with a va­luable port of call on the maritime routes along the African coast and, soon afterwards, towards the East Indies, the West Indies, and Brazil. In order to attract settlers and to guarantee Portuguese presence, royal privileges granted in 1466 and 1472 gave future colonists of Santiago the right to trade under favourable conditions on the west coast of Africa (cf. HGCV, Corpo documental I, 1988: 19–28). In Santiago and the nearby island of Fogo, cotton was grown and made into cloth and horses were raised. On the authority of the privileges, the colonists traded along the coast, exchanging the cloth and the horses for African goods and slaves. Only a minority of the slaves brought to Santiago and Fogo remained there, as the ma­jority were resold to Portuguese, Spanish, and Italian merchants who dispatched them to the European and American markets. The colonists had arrived without servants and without their spouses. A close cohabitation between the white master and several people of African origin (female partner(s) and servants of both sexes) must therefore have commenced immediately. Children born out of such unions were often freed. So from the second generation onwards, there were more “mulattos” than whites. The increase in the population of Santiago and in the wealth of its ruling class (the moradores) was only to last as long as the Portuguese were able to protect the monopoly along the West African coast, which the treaties of Alcáçovas (1479) and Tordesillas (1494) had granted them. From 1565 onwards, they became less and less successful at this. The African rulers increasingly dealt directly with the Dutch, the English, and the French. An economic crisis with deep repercussions for Santiago society ensued. Most of the landowner-merchants left the island and freed a large proportion of their slaves in order to relieve themselves of the responsibility of feeding them. Some of these emancipated slaves left for other islands (Brava, Santo Antão, São Nicolau) in search of land that could support them. A system of slavery was replaced with a system of tenancy with small tenants working the lands of absentee landowners. On the islands, the social boundaries between the freed and the enslaved and between whites, mixed, and blacks became blurred. Due to the isolation of the archipelago that took place at the end of this first but crucial period of the island’s history, the ethnic composition of the population of Santiago has changed little since that time.

In 1582, the report by sargento-mor Francisco de Andrade to the King provides a reliable estimate of the population of Santiago at the beginning of the crisis. Some 13,400 people were then living in Santiago. The two urban centers, Ribeira Grande and Praia, accommodated more than half the entire population of the island (cf. HGCV I 1991: 230–236), with 708 vizinhos (‘households’) and around 6,700 slaves living in them. In the inland areas there were some 600 whites and free mulattos and some 400 emancipated Africans who held about 5,000 slaves in their properties and houses. Unfortunately, there are no documents that provide direct information on the origin of the slaves. But there is indirect evidence that speakers of Wolof formed the majority amongst the first groups of slaves who disembarked at Santiago (Lang 2006). Later on, when the competition of the other European powers forced the Cape Verdean mer­chants to move the center of their activities more to the South, speakers of other Atlantic languages (Diola, Manjaku, Mankanya, Balanta, Papel, Bijago, Temne, etc.) and of Mande languages (especially Mandinka) must have become more and more numerous among the slaves who were brought to Santiago.

Santiago Creole certainly arose during this first period of the island’s history. More acrolectal varieties may have originated in the two urban centres and ports of the island, where Africans lived in close contact with their masters, and more basilectal ones in the countryside, where freed mulattos supervised the African slaves who grew cotton, raised horses, and wove cloth. At the time of the crisis at the end of the 16th century, the relative levelling in the remaining population of the differences between rich and poor and between town and country must have involved a similar levelling in linguistic terms and one that certainly favoured the more rural and basilectal varieties of Santiago Creole.

3. Sociolinguistic situation ⇫ ¶

In 1975, the former Portuguese colonies of Cape Verde and Guinea-Bissau became a single inde­pen­dent state. However, six years later, a coup d’état in Guinea-Bissau brought this alliance to an end. Since then, the Cape Verdean archipelago has formed the independent Republic of Cape Verde. In 1991, its one-party system was substituted by a multi-party system.

Although Portuguese is still the only official language of Cape Verde, the local inhabitants – nearly half a million people – almost all speak as their first language a Portuguese-based creole, which varies from island to island. So the term diglossia fits the overall linguistic situation very well. Those who have emigrated and their descendants (at least half of the Cape Verdeans) do not all have a complete command of the creole. In Santiago there is some geographic variation, the north-western region of Santa Caterina being considered the most conservative, and much social and register variation is found in Praia between basilectal and more Portuguese-influenced varieties of the educated people, who, due to professional reasons, speak much Portuguese. 

Serious efforts are being made to turn Cape Verdean Creole into the second official language of the country. In 1998, an official orthography (Alfabeto unificado para a escrita do Cabo-Verdiano, ALUPEC) was adopted, on experimental terms, for a period of five years. It has recently been confirmed as the only officially recognized writing system for Cape Verdean Creole. In this chapter, all examples are given in ALUPEC. Creole has also been declared a language of administration and education, but this de­ci­sion is still waiting for implementation. There are no television channels, radio stations, or newspapers functioning entirely in Creole, but Creole broadcasts and Creole articles can be found regularly in these media.

The first known grammar of Santiago Creole was written in 1885 by António de Paula Brito, an inhabitant of the island, and was published in 1888 by the Portuguese linguist Adolfo Coelho.  It is written in Santiago Creole (left column) and Portuguese (right column). Only in 1982 did a book appear, the Diskrison strutural di lingua Kabuverdianu by the Cape Verdean linguist Manuel Veiga, in which the description of Santiago Creole was kept strictly separate from the description of other varieties. The most recent published grammar of Santiago Creole is by Nicolas Quint (2000). Quint is also the author of the first dictionary dedicated exclusively to the Santiago variety (1996); he has repeatedly enlarged and recast it (see for instance Quint 1999). The most extensive dictionary of Santiago Creole up to now is Lang (ed.) 2002.

4. Phonology ⇫ ¶

Santiago Creole has eight oral vowel phonemes (see Table 1), which all have a nasal counterpart. The functional load carried by the /ɐ/:/a/ opposition is relatively small. In Table 1, the symbols used in the ALUPEC orthography are given in angle brackets.

For the sake of coherence and in order to help pronunciation, I ex­tend the use of the graphic accent to the open /a/, writing e.g. karapáti ‘tick’, as opposed to karapati ‘to cling to sb/sth’, with /ɐ/ in the stressed, penultimate syllable. The nasal vowels are represented in ALUPEC by the same symbols as their oral counterparts fol­lowed by an <n>.

     There are several reasons to consider the vowel system as having only three degrees of aperture rather than four: /e/ and /ɔ/ are extremely open so that even the mid vowels /e/ and /o/ are frequently mistaken for open vowels by outsiders; the three open vo­wels only occur in stressed syllables; all three open/mid oppositions are used to distinguish between nouns and verbs, e.g. karéka ‘bald head’ vs. kareka ‘to go bald’, karapáti ‘tick’ vs. karapati ‘to cling to sb/sth’, frónta ‘misfortune’ vs. fronta ‘to suffer a misfortune’ (see Quint 2001).

     In unstressed word-final position all oppositions in degree of aperture are neutralized. The three re­sul­ting archiphonemes are normally pronounced as [i], [ɐ], and [u]. The numerous diphthongs of Santiago Creole can all be interpreted as sequences of two simple vowels.

     The variety of Santiago Creole described here has 17 oral and three nasal consonantal phonemes. All oral consonants have pre­nasalized counterparts (see Lang 2007 for the role of nasality in Santiago Creole). In very basilectal varieties, there are no voiced fricatives as separate phonemes. Instead, there is a fourth nasal consonant /ŋ/ in words of African or onomatopoeic origin (meso- or acrolectal varieties have replaced it by a pre­na­sa­li­zed /g/, as in ŋánha ['ŋaɲɐ] ‘interior part of a corncob’, becoming ngánha ['ŋgaɲɐ]).

In the official Cape Verdean orthography ALUPEC, these consonants are represented by the same sym­bols with the exception of the post-alveolars and palatals – these are represented (from top to bottom) by <nh, x, j, tx, dj, lh>. Prenasalized consonants receive an <n> before the letter(s) which represent(s) the cor­res­ponding oral consonant.

Most syllables have a consonantal onset. In most cases there is a single con­so­nant in the onset, some­times there are two (kre ‘to want’, kláru ‘clear’, stángu ‘stomach’) or even three (spreme ‘to squeeze’, spli­ka ‘to explain’). Syllable codas never consist of more than one consonant, and only /r/, /l/, or /s/ are ad­mitted in this position. 

Word stress is not fixed, but placed on the penultimate syllable in most words. There is relatively little variation in pitch and syllable length within a (declarative) sen­ten­ce in Santiago Creole.

5. Noun phrase ⇫ ¶

There are no morphological case or gender distinctions in Santiago Creole nouns.

Female sex of animates may be in­dicated by specific lexemes (fidja ‘daughter’ as opposed to fidju ‘child’, badjadera ‘female dancer’ as op­posed to badjador ‘(male) dancer’, etc.) or by adding fémia ‘female’ (fi­dju fémia ‘daughter’ as opposed to fidju mátxu ‘son’).

Plural is generally indicated on the first element of the noun phrase where such an indication is pos­sible. When this element is a noun or an adjective, plural is marked by an -s in words ending in a vo­wel and by -is in words ending in a consonant (fidjus ‘children’, badjadoris ‘dancers’, etc.). However, like other grammatical markers, the plural marker is only used when the corresponding semantic value is not self-evident from the context.

Some adjectives such as bunitu ‘pretty’ may end in -a when used to characterize female persons (un ba­dja­dera bunita ‘a pretty (female) dancer’). Expressions such as un kása bunita ‘a pretty house’ (with adjective concord according to the gender of Portuguese casa ‘house’) or nhas fidjus ‘my children’ (with a plural marking on the possessive) on­ly occur in acrolectal varieties. Adjectives may precede or follow the noun. As in Romance lan­gua­ges, the preceding adjective modifies the meaning of the noun (as in un bon puéta ‘a poet writing good poetry’), the one which follows modifies the referent of the noun (as in un puéta bon ‘a poet who is good, for whatsoever reason’). 

In comparative constructions of equality, the standard is marked by sima:¹

(1)               Es       kása       li          e        áltu      sima    kel            la.

              DEM  house    here    be     high    as          DEM       there   

              ‘This house is as high as that one.’

Comparative constructions of inferiority are normally avoided in favor of constructions of superiority (‘A is higher than B’ instead of ‘B is less high than A’). In comparative constructions of superiority, the ad­jective is marked by más ‘more’ and the standard by ki or di ki:

(2)               Es          kása    li          e        más     áltu     (di) ki  kel            la.

              DEM     house  here    be     more      high    than      DEM       there.

              ‘This       house is higher than that one.’

Superlative is expressed by the same construction with a standard marked by di:

(3)               Es          kása         li          e        más     áltu        di      (es)        tudu.

              DEM     house       here    be     more      high       than    (DEM)  all

              ‘This house is the highest of all (these).’

There is a preposed indefinite article un (with a plural uns), but no definite article:

(4)       a.  Un      bes         un              ómi      di      lonji         bá        kása         di    un            mudjer.

              one      time       ART.INDF    man    of      far.away  go        house       of    ART.INDF  woman

              ‘One time, a man from far went to the house of a woman.’

       b. Mudjer  resebe    -l         ben      resebedu.

              woman    receive   -3SG   well     receive.PASS

              ‘The woman received him very well.’

The feminine form of the Portuguese indefinite article (unstressed uma) became a stressed indefinite ar­tic­le for augmentation. (The language also possesses two augmentative suffixes -on and -óna, which fre­quently co-occur with the augmentative indefinite article):  

(5)               Na    kel       baskudja,            N         diskubri     uma                    libron.

              in      DEM  rummage.about 1SG     discover     ART.INDF.AUGM  book.AUGM

              ‘In that rummaging through, I discovered a huge book.’

There are two demonstratives, both admitting pronominal and adnominal use, one with a plural form and neutral with respect to the distance from the speaker (kel, plural kes) and the other invariable and for pro­ximity to the speaker only (es). Both may be followed by a spatial adverb: kel (kása) li ‘this (house) here’, kel (kása) la ‘that (house) over there’ and es (kása) li ‘this (house) here’ (see examples 1–3).

     The independent pronouns carry stress, and the subject pronouns, object pronouns, and adnominal (pre­posed) possessives are unstressed (see Table 3). The a- in the independent personal pronouns is an optional topic mar­ker.

Algen ‘somebody’ may be used as an indefinite subject, but a passive construction is normally pre­ferred when the agent cannot or should not be specified:

In polite 2SG there are no unstressed adnominal possessives. In 2PL and polite 2SG, the stressed in­de­pen­dent pronouns must be used for object function. A possessive pronoun is illustrated in (7)

(7)               Kel         kása    e     di      kenha? –    E         di meu.

              DEM     house  be   of      whom        be        mine 

                ‘Whose house is this? – It is mine.’

As can be seen in Table 3, the possessive pronouns are made up of the preposition di ‘of’ followed by elements which are either identical to independent personal pronouns or which have no additional use. The sa­me forms serve as stressed adnominal possessives: kása di meu ‘my house’, kása di nho ‘your house’, etc.

Possessor constructions with two nouns vary in form but not in meaning: kása di pai, kása-l pai, kása pai ‘the house of the father’. The form -l, though probably of Wolof origin (cf. Lang 2009: 135–139), is felt to be a shorter variant of di.

The preposition ku ‘with’ is used for (pro)noun conjunction: mai ku pai ‘the mother and the father’, ami ku bo ‘I and you’. 

The numerals from 1 to 10 are un, dos, tres, kuátu, sinku, sax ~ sais, séti, oitu, nóvi, dés.

6. Verb phrase ⇫ ¶

There is no person, number, or gender concord in Santiago Creole verbs. Unless otherwise indicated, un­mar­ked forms of stative and dynamic verbs yield different temporal meanings:

(8)       a.  Es     ten      dos      fidju.         

              3PL  have    two      son            

              ‘They have two children.’      

         b.  Es     perde dos      fidju.

              3PL  loose   two      son

              ‘They lost two children.’

In some verbs, the unmarked form has present-time reference when it is used with a stative meaning and past reference when treated as a dynamic verb. Compare:

Santiago Creole has six verbal markers relating to aspect, mood, tense, and voice. Mood and aspect are expressed by pre­verbal particles (in this order), and relative tense and voice are expressed by verbal endings. The three preverbal mar­kers are ál (‘desire’), sa (‘progressivity’), and ta (‘imperfectivity’). The three verbal endings are -ba (‘anteriority’), -du (‘passivity’), and -da (‘anteriority + passivity’).

The modal marker ál has an alternative epistemic use which expresses a supposition. The inherent im­per­fectivity of a desire is never expressed by an additional ta. But sa ta may follow an epistemically used ál in order to express progressivity:

(10) a.    Dios  ál      dá      -u      sórti!              

              God  MOD  give   -you  luck               

              ‘May God make you lucky!’               

       b.    Ómi  ál      sa      ta      trabádja.

              man  MOD  PROG IPFV  work

              ‘The man will be working [now].’

Fitting the fact that progressivity implies imperfectivity, sa is always followed by ta, but, of course, ta is not always preceded by sa.

‘Future’ and ‘habituality’ are subsumed under ‘imperfectivity’. Es ta trabádja may thus mean ‘They work’, ‘They will work’, or ‘They use to work’. Stative verbs thus do not preclude the use of the im­per­fec­tivity marker ta. They simply make it superfluous whenever there is no need to express some more spe­cific variety of imperfectivity such as ‘future’, ‘habitual’, etc.: Es kre kunpanheru ‘They love each other’, but Es ta kre kunpanheru ‘They will love each other’.

The form -ba, often described as a past marker, more precisely indicates ‘anteriority’ (with respect to the moment of utterance in stative verbs and with respect to another past event in dynamic verbs):

(11)            Es     tenba                     tres           fidju. 

              3PL  have.ANT                    three        child

              ‘They had three children.’

(12)            E         lenbra        di      kusa      ki           sáibu          flába        -el.

              3SG    remember  of      thing     which    wise.man   tell.ANT    -3SG

              ‘She remembered what the wise man had told her.’

In hypothetical constructions which present the relevant condition as being unattainable, -ba, does not mark a temporal distance (‘anteriority’) but instead marks a modal distance (‘irreality’):

The form -du is mainly used in order to avoid mention of the agent and very often to give the clause an im­per­so­nal meaning:

(14)            Fládu        ma         Kabuberdiánus       gosta          di         grógu.

              say.PASS  COMP  Cape Verdeans       be.fond      of        rum

               ‘Cape Verdeans have been said to be fond of rum.’

-da is a contraction of -du + -ba, which in Brito’s grammar from 1885 (see §3 above) is still -duba (fláduba ‘it had been said’ instead of modern fláda).

Table 4 summarizes the use of the ver­bal markers with dynamic and stative verbs, and gives the Portuguese etymons of the Santiago Creole markers:

Except for the cases mentioned so far, all theoretically conceivable combinations of markers do occur (but remember that -da is itself a fusion of -du and -ba and thus may not combine with -du or -ba). The semantic results of such combinations are generally predictable.

Except in very rare cases, Santiago Creole always uses a copula in clauses which assign a characteristic to something (15), iden­tify something with something else (16), or locate something somewhere (17). As in Portuguese (and Spa­nish, and Catalan), Santiago Creole has two copulas, one for characterizations, identifications, and locations, the va­lidity of which is presented as being limited in time (sta, anterior stába), and another one for cha­rac­te­rizations, identifications, and localizations for which no such limitation is envisaged (e, anterior éra, in complementary distribution with ser, anterior sérba, used after auxiliary verbs, verbal particles, and in prepositionally introduced subordinate clauses). Each of the three cases is illustrated by the paired sentences under (15), (16), and (17):

(15)    a.  Si                    duénsa  e      grávi.   

              3SG.POSS     desease  be     serious

              ‘His disease is serious.’

         b.  Pamódi    ki           bu       sta     tristi?

              why         COMP  2SG    be        sad

              ‘Why are you sad?’

(17)    a.  Mudjer      éra        di      Práia.

              woman      be.ANT   of      Praia 

              ‘The woman was from Práia.’

         b.  Djánta     sta     na       mésa.

              dinner     be        in        table

              ‘Supper is on the table.’

Surprisingly, Santiago Creole makes a similar distinction between ten (without temporal limits) and tene (suggesting such limits) for ‘to have’:

(18)    a.  Djon   ten    dinheru.                  

              John   have    money                     

              ‘John has money (= John is rich).’       

         b.  Djon   tene   dinheru.

              John   have    money

              ‘John has money (= John has money with him).’

There are at least 26 verbal periphrases in Santiago Creole: three express modality, three express voice, two express re­la­tive tense or “taxis”, and eighteen express aspectuality. Only the first two of the three modal periphrasis pode fase ‘can do’, debe fase ‘must do’, and ten ki fase ‘have to do’ are regularly used for epistemic purposes:

(19)            E         pode/debe         sta    duenti.

              3SG    can/must   be        ill

              ‘He may/must be ill.’

Clausal negation is achieved by using ka 'not':

(20)            Na    mudjer     ka       ta           konfiádu!  

              in      woman    not        IPFV     trust.PASS

              ‘Never trust a woman!’

As can be seen from the preceding example, ka is normally placed in front of the verb and the verbal particles. There are two exceptions to this rule: (i) Negative imperatives are formed by placing ka in front of the subject pronoun (see example 32 below). (ii) Ka normally follows e, the only unstressed verb form of Santiago Creole:

(21)            Es       ómi        li       e       ka    Djon.

              DEM  man       here  be        not    John

              ‘This man here is not John.’

In negative clauses with indefinite pronouns ka co-occurs with the indefinite pronoun:

(22)            N      ka    odja       ningen.

              1SG  not    see         nobody

              ‘I didn't see anybody.’  

7. Simple sentences ⇫ ¶

The order of syntactic elements in the Santiago Creole clause is SVO. Objects in clauses with ditransitive verbs nor­­mal­ly occur in the order indirect object + direct object and are not marked by prepositions (double ob­ject construction):

There are no reflexive pronouns in the language. Reflexivity is expressed by kabésa ‘head’ (24), and re­ci­pro­city by kunpanheru ‘comrade’ (25):

(24)            Kel         trópa      nforka   kabésa pa     es      ka   obriga -l       trai      si       kolégas.

              DEM     soldier   hang     head      for    3PL  not  force   -3SG betray his    comrades

              ‘This soldier hung himself in order to avoid being obliged to betray his comrades.’ 

(25)            Vióla     ku          violon    ta           parse       ku        kunpanheru.

              vióla      with       violon    IPFV     resemble with    comrade

              ‘The “vióla” and the “violon” [two types of guitars] resemble each other.’

Verbs in imperative use do not take the 2SG subject pronoun (cf. 26a). Subject pronouns are however used with 2PL imperatives (cf. 26b) and in the prohibitive (cf. 32):

(26) a.    Plánta      midju!       

              cultivate  corn          

              ‘Cultivate corn!’      

       b.    Nhos     plánta         midju!

              2PL         cultivate     corn

              ‘Cultivate corn!’

Yes-no questions are distinguished from statements only by a rising intonation. The question words in content questions are normally fronted (27), but can also remain in place under certain conditions (28):

(28)            So     bu       fla   -m       e        módi!

              only  2SG    tell  -1SG   be     how

              ‘Just tell me how it is!’

The question words are kál ‘which’, kenha ‘who’, kusê ‘what’, undi ‘where’, ki ténpu ‘when’, módi ‘how’, kántu ‘how much’, and pamódi ‘why’.

     Focused elements are fronted and followed by the complementizer ki (29). When, in addition, the co­pu­la precedes the focused element, we clearly have a cleft construction. Focusing without the copula is how­ever more frequent (and even more so when the focused element is a question word): 

8. Complex sentences ⇫ ¶

Clauses are coordinated by the following conjunctions: y… ‘and…’, má(s)… ‘but…’, o… ‘or…’, o…o… ‘either… or…’, (nen…) nen… ‘(neither…) nor…’. Asyndetic coordination of clauses is much more fre­quent than coordination by y. There are three complementizers: neutral ki ‘that’ (30) and two com­ple­men­tizers which introduce object clauses after verbs of saying, thinking, and perception, i.e. assertive ma ‘that’ and interrogative si ‘whether’. In an enumeration of object clauses, ma and si may be replaced by ki from the second clause onwards (31):

(31)            E       fla     ma       e            mesteba        bá        Práia   

              3SG  say    COMP  3SG       need.ANT    go        Praia   

              y          ki          e        ta         saíba          sais   óra.

              and     COMP     3SG  IPFV  start.ANT six     hour 

              ‘He said that he had to go to Praia and that he would leave at six o’clock.’

Object clauses which express a wish, an expectation, or an anticipation of the main clause’s subject are in­troduced by pa ‘for’:

(32)            Ka       bu       kume   inda,   spéra     pa     buláxa   molse     primeru!

              not      you     eat       yet       wait       for       cookie    get.soft  first

              ‘Don't eat yet, wait until the cookie has got soft!’

Adverbial clauses are headed by subordinating conjunctions such as embóra ‘though’, (kel)óki ‘when’, in­kuántu ‘while, whereas’, kántu ‘when’, purki, pamodi ‘because’, komu ‘as’, sima ‘as soon as’, si ‘if’, sinon ‘otherwise’, timenti ‘so long as’, and a great number of complex subordinating conjunctions such as afin di ‘in order to’, alen di, pa len di ‘in addition to’, etc.  

Relative clauses, i.e. clauses which modify a noun, are introduced by the complementizer ki:

In cases where the referent of the antecedent functions as the subject or direct object of the relative clause, no resumptive pronoun occurs:

(34)            Kántu si                   mudjer     korda, e        txoma    Jáni    ki        ka    kudi.

              when  3SG.POSS   woman    awake 3PS   call         Jáni    COMP  not    answer

              ‘When his wife woke up, she called Jáni, who didn’t answer.’

This may even occur in other cases when there is no danger of misunderstanding:

In all other cases, a resumptive pronoun is used (no preposition stranding):

(36)            Bu       konxe  ómi      ki        N     papia ku     el.

              2SG    know  man    COMP  1SG  speak  with    3SG

              ‘You know the man I spoke with.’