The origin and originality of passivization in Papiamentu

This paper is concerned with the diachrony of passivization in Papiamentu. While it is generally held in the literature that passivization in Papiamentu is a non-native 19th century borrowing from Dutch and Spanish, this paper argues that it is in fact an original, native feature of the Papiamentu grammar. With that purpose, the use of auxiliary-less passives in Early (19th / early 20th century) Papiamentu texts is illustrated and analyzed in detail. In addition, synchronic evidence will be adduced. This paper furthermore argues that Papiamentu inherited its passive morphology from proto-Upper Guinea Portuguese Creole. To bolster that claim, the auxiliary-less passives found in Early Papiamentu texts will be systematically compared with passivization patterns found in Upper Guinea Portuguese Creole.


Introduction
This paper is concerned with the diachrony of passivization in Papiamentu (PA) 1 .In the literature on passivization in PA, the view is held that PA, "in harmony with the majority of creole languages, originally did not have any passive" 2 (Eckkrammer 1993: 140).According to this view, passive morphology is thought not to have existed in PA prior to the introduction of the passive auxiliaries ser (< Spanish ser) and wòrdu (< Dutch worden) and to have been borrowed wholesale into PA from Dutch and Spanish.The central aim of this paper is to take issue with this view and to argue that, although the 1 Papiamentu is spoken on the Dutch Antilles (Bonaire and Curaçao) and Aruba (with status aparte since 1986) as well as by an important number of immigrants in the Netherlands.The dialectal differences between Aruba, Bonaire and Curaçao (jointly referred to as the ABC Islands) do not play a role in this paper. 2All quotes are my translations from the original.
above-mentioned passive auxiliaries doubtlessly constitute an innovation, passive morphology is in fact a native, original feature of the PA grammar.With this purpose, the central part of this paper provides examples of auxiliary-less, morphologically-marked passives from a set of Early PA (EPA) texts. 3 An added aim is to draw a parallel with the passive morphology found in the Upper Guinea branch of Portuguese Creole (UGPC) as spoken on the Cape Verde Islands (CV) 4 and in Guinea-Bissau and Casamance (GBC) in order to shed light on the origin of PA's passive morphology and, more generally, to strengthen the hypothesis that UGPC and PA share a common ancestor (Martinus 1996;Quint 2000b;Jacobs e.g. 2009a).
The structure is as follows: Section 1 introduces passivization in Modern PA (MPA) and describes the commonly held view, refuted in this paper, according to which passivization in PA is a non-native feature.Section 2 discusses the diachrony of PA's past participle morphology (on which passivization in PA crucially relies), as this conditions a proper understanding of the (comparative) data presented in the remaining sections.In Section 3, auxiliary-less passivization in UGPC is introduced to serve as a reference model against which the PA data presented in Sections 4 and 5 will be systematically mirrored.Section 4, then, provides examples of auxiliary-less passives found throughout a variety of EPA texts in order to support the central claim of this paper, the nativeness and originality of passive morphology in PA.This claim is lend additional synchronic support in Section 5, where it is shown that MPA still makes use of auxiliary-less passives in a small set of apparently fossilized expressions with the verb papia 'to speak' as well as by means of the high frequency modal verb ′meste~mesté~mester 'to need, be needed'.Following this, Section 6 briefly reflects on the integration of the auxiliaries PA ser and wòrdu before coming to a reassessment of the origin and originality of passive morphology in PA in Section 7. 3 Throughout the paper, when relevant, a distinction is made between Early PA (EPA) and Modern PA (MPA).The term EPA is used in relation to words and features no longer found in mainstream MPA, but present in 19th and/or early 20th century writings.I decided to take Lenz's (1928) landmark PA study as roughly marking the transition from EPA to MPA.Among the EPA texts studied for this paper are gospels, catechisms and short grammars written in a period stretching from 1775 to 1928.References follow throughout the paper. 4Within Cape Verdean Creole (CV), a distinction must be made between the varieties of Barlavento and Sotavento: the latter are generally considered to be the oldest and most basilectal varieties and are thus of most interest to this paper.The examples from CV in this paper concern the Santiago variety (SCV), as this is by far the best described and likely the oldest variety within the Sotavento branch.

Passivization in Modern Papiamentu 5
The PA passive consists of an auxiliary verb and a past participle (PP).A speaker of PA can choose its auxiliary from either ser6 , worde~wordo~wòrdu7 (henceforth wòrdu), or keda.The latter clearly is a recent innovation (Maurer 1988: 329)8 and will not be discussed further.PA ser and wòrdu appear to be in free variation (Kouwenberg & Ramos-Michel 2007: 319) (Maurer 1988: 330) (4) PA tur hende lo wòrdu pordoná all people FUT AUX forgiven 'everybody will be forgiven' (Heuvel, van Wel & van der Wal 1989: 233) As referred to briefly in the Introduction, the consensus has now been reached among PA specialists that the passive as exemplified above is a non--native feature, meaning that passive morphology is thought not to have existed in PA prior to the introduction of the auxiliaries ser and wòrdu in the 19th century; the Dutch and Spanish passive templates (with worden and ser respectively) are thought to have been borrowed wholesale into PA: [T]he Papiamentu passive (...) is in fact (...) a syntactic structure of chiefly Spanish derivation.(...) [T]he accretion of a passive, and of a past participle inflectionally distinguished from the hitherto uninflected, invariable Papiamentu verb, is one of the principle examples of the decreolizing Hispanization of literary Papiamentu.(Wood 1972a: 859 9 )   Similar affirmations are easily found, for instance, in Andersen (1974: 263), Clemesha (1981: 12  10 ), Busche (1993: 80  11 ), Eckkrammer (1993: 140), Kramer (2004: 110) and Sanchez (2005: 68), so that it is indeed legitimate to speak of a consensus.Interestingly, this consensus is very tangible also in the Antillean academic realm: not only is the use of passives "subject to controversy among native linguists" (Maurer 1988: 331), it is in fact "generally discouraged" (Birmingham 1970: 95) 12 and Eckkrammer (1993: 141) speaks of "efforts of numerous Antillean linguists to eradicate passives".The language purist Maduro, for instance, pleaded for avoiding passives as much as possible and condemned the "[w]rong use of passives: Papiamentu is a supporter of active forms" (Maduro 1971: 43-45;cf. 1991: 162).
The present paper takes issue with the consensus described above.

The diachrony of past participle morphology in Papiamentu 13
To facilitate a better understanding of the (comparative) data presented in the remainder of this paper, it is relevant to address the diachrony of past participle morphology in PA, which is what passivization in PA crucially relies on.Past participles in PA are productively derived from disyllabic verbs by shifting the stress from the first to the last syllable: PA ′kanta 'to sing' > kantá 'sung'.Contrary to disyllabic verbs, the infinitives of longer verbs are already stressed on the final syllable and are thus homophonous with their past participles.Thus, PA entregá can mean both '(to) submit' and 'submitted', according to the context. 14 Diachronically, PA's past participle morpheme is not a null morpheme, but rather a phonetically eroded reflex of the Iberian past participle suffix -/do/.An important clue in this respect is that the Early PA (EPA) texts and grammars analyzed for this paper testify to the use of an EPA past participle morpheme ′-/r/, either instead of, or in variation with ′-/Ø/.Examples of verb-participle pairs such as <jama> '(to) call' vs. <jamaar> 'called' abound in these EPA texts and can still be found as late as in Lenz (1928).Thus, for PA's past participle morpheme ′-/Ø/, we can reconstruct an erosion path that goes from etymological ′-/do/ via EPA ′-/r/ to MPA ′-/Ø/.Interestingly, this development had already been described for PA by Van Name (1869: 151): "Final d, with d which has become final by the loss of a following vowel, either passes into r or is dropped". 13This section provides a summary of findings presented in Jacobs (2008) regarding past participle formation in PA. 14 Interestingly, however, the orthography used in the EPA documents analyzed for this paper suggests that all verbs, regardless of their syllable length, were stressed penultimately well into the latter stages of the 19th century.Indicative of this is the Dutch-based orthography of these early documents, where double vowels are long vowels, providing an indication of stress.Such double vowels occur in penultimate syllables not only of disyllabic verbs, e.g.<kaanta> '(to) sing', but also of plurisyllabic verbs, e.g.<entreega> '(to) submit'.It is useful to add that PA, different from Dutch, does not have a length opposition in its vowel system.In other words, the orthographic practice of the texts is entirely based on the Dutch preference for long vowels in stressed positions (p.c.from anonymous reviewer).
Past participle formation in UGPC depends on the same mechanism as just described for PA: suffixation of the past participle morpheme to the verb stem.Unlike in PA, etymological -/do/ did not erode in UGPC, which has the suffix -/du/.Thus, from the infinitive kanta 'to sing', the past participle kan′tadu 'sung' is derived.
Tables 1 and 2 are meant to further underpin a phonetic reduction path from (a) Port.-/do/ via (b) proto-UGPC ′−/du/ to (c) EPA ′-/r/ and (d) MPA ′-/Ø/.The examples concern PA lexemes for which a UGPC origin can be hypothesized, either because the etymon is probably Portuguese, or because the UGPC and PA form share a salient phonetic alteration from the etymon, or both (for each example, this is indicated in a footnote).Table 1 shows the sound change at issue: mestedu '(to be) needed'20 <mesteer> mesté~mester Table 2 shows that the change from etymological -/dV/ via EPA ′−/r/ to MPA ′−/Ø/ also affected words other than past participles and serves to further strengthen the plausibility that MPA's past participle morpheme ′−/Ø/ derives from proto-UGPC -/du/.The hypothesis that the past participle morphology in PA and UGPC traces back to a common origin is further supported by the remarkable productivity and regularization of past participles in both creoles, which is uncommon among creoles (e.g.Grant 2001: 84).Regarding the formation of past participles in GBC, Kihm (1994: 243) points out: "Regularization has been thorough (…), so that there is no trace in Kriyol of the so-called 'irregular' forms such as escrito 'written' (Kriyol skirbidu 'be written'), feito 'done' (Kriyol fasidu 'be done'), and so forth".The same is true for PA: past participles like PA skirbí 'written', hasí 'made, done' or habrí 'opened' and the UGPC equivalents skirbidu, fasidu, abridu indicate that the past participle morphology has been regularized making redundant the irregular superstrate participles Sp. / Port.escrito, Sp. hecho / Port.feito and Sp.abierto / Port.aberto.
The regularization and productivity also becomes visible in participles whose corresponding Iberian infinitives end in −er: PA and UGPC again coincide, giving these participles a regular stress bearing [e] instead of the irregular [i] of the Iberian model 27 .Compare, for instance, the infinitives Sp. / Port.coser 'to sow' and their past participles cosido 'sowed' with the PA / SCV infinitive ′kose and the corresponding participles PA kosé -SCV kosedu, or, interestingly, Port.lamber 'to lick' and its participle lambido 'licked' with the PA / SCV stem form ′lembe and the corresponding participles PA lembé / SCV lembedu (Quint 2000b: 146). 28 To sum up, the mechanism of past participle formation is identical in PA and UGPC and the respective past participle suffixes, though synchronically distinct due to erosion, are historically cognate.

Auxiliary-less passives in Early Papiamentu
As noted, disregarding the synchronically different but diachronically cognate past participle suffixes, the only real difference between the MPA passive and the UGPC passive is that the latter lacks an auxiliary verb.Interestingly, however, passivization patterns found in EPA follow the same auxiliary-less passive template [TMA + PP] as just described for UGPC.Examples are provided below, accompanied by parallel examples from UGPC.
First, a methodological caveat is in order.Since PA ta functions not only as an imperfective aspect marker but also as an equative copula, participles preceded by ta are ambiguous between a resultative (copula + adjectival participle) and a passive (imperfective aspect marker + past participle) interpretation.Conveniently enough, the evangelical PA texts can be put side by side with the original Dutch Bible text, so as to assure that we are indeed dealing with passives, not with resultatives.Therefore, the respective passage from the Authorized Version of the Dutch Bible30 (AVDB) is provided whenever the example is drawn from an evangelical text.Note, furthermore, that in the examples that follow, an underscore [ _ ] marks the position where, according to MPA passivization rules, we would see the auxiliary verb wòrdu/ser.
(8) a. EPA mi sanger, (...) koe ta _ dramaar pa moetsjoe my blood that IMP _ (be) shed for many 'my blood, that is (being) shed for many' 31 (Kuiperi 1862: 30 b. SCV porta Ø abridu door PST/PRV (be) opened 'the door has been opened' (Quint 2000a: 266) 39   In addition, I found an interesting case of an auxiliary-less (non-stative) passive marked by zero in (14a), which is ungrammatical according to MPA rules but closely resembles UGPC participle patterns.( 14 of today can (be) extinguished 'the star of today can be extinguished' (Brito 1887: 393) Also in terms of word order, the correspondences between UGPC and EPA passivization patterns are striking.As noted by Baptista (2002: 113), "[i]n a number of cases, personal passives may involve subject-verb inversion (...) under the condition that the subject be a full NP".An example is (16b).The EPA passive proposition in (16a) conforms to this description: (16) a. EPA Ta pronunciá consonantenan mescos IMP (be) pronounced consonants same thing 'Consonants are (being) pronounced the same way' (Sintiago 1898: 8) b.SCV ta bendedu kuazi tudu kusa IMP (be) sold almost all thing 'almost everything is (being) sold' (Lang 2002: 159) In fact, passives of this type are still found in MPA and notably so with the verb papia 'to speak' (see examples in Section 5.1).According to Kouwenberg & Ramos-Michel (2007: 319), a construction such as (16a) is not a passive but rather "an active construction with an unexpressed subject which has arbitrary reference".This seems a possible interpretation considering that in MPA the infinitives of longer verbs are homophonous with their past participles, meaning that ta pronunciá could indeed be interpreted as an active predicate.However, as discussed in Jacobs (2008;cf. footnote 24), there is orthographic evidence that longer verbs in EPA behaved like disyllabic verbs in that they were stressed penultimately.This idea finds confirmation in the (Spanish-based) orthography used by Sintiago (1898) 40 , who placed an orthographic accent on the final syllable of forms that are clearly past participles (whether derived from disyllabic or from longer verbs, e.g.<pronunciá>, <formá>; cf.examples [18a, b]), while he does not use them on the final syllable of verbs (whether disyllabic or longer) that are clearly used actively (e.g.<pronuncia>, <forma>).This fact favors the interpretation of (16a) as an auxiliary-less passive involving subject-verb inversion, rather than as an active construction with an unexpressed subject in which case we'd expect <pronuncia> rather than <pronunciá>.
Interestingly, the passive interpretation of this type of clauses is also preferred by Abraham (2006: 24), who analyzed (17a).The example is taken from the Proclamasjon, the public announcement of the abolition of slavery in 1863.with boat 'planes and ships were bought' (Martinus 1996: 85) In keeping with Baptista's criteria (subject-verb inversion and a full NP as subject), Abraham (2006: 24) observes that (17a) "contains the passive verbs poeblika (publiká in modern spelling) (...) without an auxiliary or a BY--phrase.Strikingly, the passive subject e ley appears in postverbal position".The example is also given by Kouwenberg & Ramos-Michel (2007: 319), who, however, as mentioned, analyze it as "an active construction with an unexpressed subject which has arbitrary reference".The evidence from Sintiago (1898) suggests the opposite, i.e. that the analysis of (17a) as a passive is preferable.

Digression: On the reliability of evangelical EPA texts
As noted, the evangelical EPA texts from which most of the examples presented above were drawn offer the advantage that translations into other languages are available, facilitating interpretation while minimizing ambiguity (cf.remarks made in this respect by Bachmann [2005: 71], van den Berg [2007: 1], and Eckkrammer 1996).On the other hand, caution is called for, given that "[s]uch texts are produced in the majority of cases by non-native speakers, less concerned with accurately replicating native structures than the desire to amuse or to impart information" (Bickerton 1999: 98).Consequently, the evangelical sources cannot beforehand be assumed to carefully reflect the usage of passive morphology in spoken EPA; the authors Niewindt, Conradi, van Dissel and Kuiperi were certainly no native speakers of PA.
However, the evangelical EPA texts need not be dismissed as unreliable; in fact, clues are available suggesting that the authors received assistance in writing from native PA speakers, in the person of slaves, fellow clergymen, or church wardens.The 1825 PA catechism written by Niewindt provides one such clue: "Since this was one year after Niewindt's arrival, it is somewhat open to question how much of this work was his own" (Fouse 2002: 128).Coomans & Coomans-Eustatia (2002) express similar doubts: "It is surprising that at that time Niewindt was already publishing in Papiamentu.He had hardly been on Curaçao for one year and about his visit to Bonaire in November 1824 Niewindt tells that he couldn't do much because he didn't understand the language".Coomans & Coomans-Eustatia therefore assume that Niewindt received assistance in translating from either the church wardens or from (one of) the nine Venezuelan / Colombian priests present on Curaçao at that time. 42Needless to say, the aid of native speakers may well 42 At the same time, this anecdote shows Niewindt's awareness of the need to learn PA in order to successfully spread the word of faith among the locals.Fouse (2002: 127) confirms that as early as in 1769, "the Jesuit priest Rodier sent a letter to the Dutch Parliament in which he mentioned the need for priests to know Papiamentu".Reverend Kuiperi, for instance, already preached in PA 4 months after his arrival on Aruba in 1858 (Coomans 2001, in the postscript to Kuiperi (1862), no page numbers).As to Niewindt, for the present paper, one example, (15a), was taken from his 1837 catechism.This was 13 years after his arrival on Curaçao, and it seems likely that by that time he had mastered the language to an extent large enough to write a catechism without assistance.Such affirmations remain speculative, however.To be complete, it is noted that Conradi arrived on Curaçao in 1837; van Dissel in 1856.
have been common practice in the production of other religious PA texts of that time.It may also be noted that the authors, being clergymen, were concerned with the spread of faith amongst the (largely illiterate) working class: "their use of Papiamentu was nothing but instrumental to their social and religious power" (Broek 1988: 178).It would thus be of no use to them to integrate new features such as auxiliary-less passives, into their sermons, Catechisms and Bible translations if nobody were to understand these constructions.Note, furthermore, that the texts by Conradi (1844), Niewindt (1837) and van Dissel (1865), also exhibit basilectal characteristics such as relative time reference.In addition, the texts demonstrate a number of other words and features absent in MPA, but, significantly, still found in UGPC (see examples in Jacobs 2008: 67, 68;2009a: 331, 338, 342, 345-349;2009b).
But regardless of the exact authorship of the 19th century texts, Bickerton believes that "if a construction turns up at all in early texts, it surely must have existed at that time" (1999: 98, emphasis in original).If that is correct, we need not doubt the existence of auxiliary-less passive morphology in EPA.Also, the claim that auxiliary-less passives were common in EPA is supported by examples from non-evangelical writings, such as (16a), (17a), and is further strengthened by synchronic attestations of auxiliary-less passives, which is the topic of the next section.

The verb papia 'to speak'
Examples ( 18) and ( 19) show how the verb PA papia 'to speak' appears in auxiliary-less passive expressions in MPA: (18) a. MPA ta _ papiá ku IMP _ (be) said that 'it is (being) said that…' (Martinus 1996: 85) b.SCV ka ta papiadu na mesa NEG IMP spoken at table 'no talking at the table' (Bartens 1996: 43;cf. Veiga 1982: 175) (19) a. MPA tabata _ papiá ku PST IMP _ (be) said that 'it was (being) said that…' (Lucille Berry-Haseth, p.c.) b.SCV staba ta papiadu / fladu ma ANT IMP (be) said that 'it was being said that…' (Victor Barros, p.c.)More research is needed to determine whether these examples belong to a limited set of fossilized expressions or whether passives of this kind are still productively formed with verbs other than papia.

PA mesté~mester and SCV mestedu
Synchronic evidence of the use of auxiliary-less passives in PA is also provided by the passive use of PA mesté and its variant mester (henceforth mesté), the high frequency modal verb that appears both in active clauses meaning 'to need, must' and in passive clauses roughly meaning 'to be needed'.As noted in Section 2 (Table 1), PA mesté should diachronically be analyzed as the past participle of the verb PA ′meste.Note, though, that these two forms have now merged and, synchronically, are mere variants of one and the same modal verb meaning 'to need, be needed'.Below, the correspondences between PA ′meste~mesté and the SCV verb-participle pair ′meste -mestedu are highlighted.But before getting there, it is useful to explain why PA ′meste~mesté constitutes one of the most interesting items of the PA verbal system.
First, its etymon is the Old Portuguese noun mester43 'need' (≠ Sp. menester) which fell out of use in the course of the 16th century (Kihm 1994: 4).Since the Dutch settlement of Curaçao did not start prior to the mid-17th century, the presence of this 15th-16th century Portuguese item in PA requires an explanation.A plausible one is that this feature entered PA by way of Upper Guinea, where a creole based on Portuguese emerged as early as in the late 15th century.
Secondly, the syntactic behavior and semantic properties of the modal verb ′meste~mesté are in fact quite different from its etymon, as mentioned, the Old Portuguese noun mester, while fully coinciding with the SCV verb--participle pair ′meste -mestedu.Bernd Heine (p.c.) has noted that "grammaticalization of a noun into an auxiliary verb is in fact rare".It would thus be a weird stroke of luck if an Old Portuguese noun followed identical, and identically rare, paths of grammaticalization in two so distant places.If to these observations we add the striking correspondence of modal and other auxiliary verbs in PA and UGPC (Jacobs 2010), then there is little reason to look for the origin of PA ′meste~mesté in any place other than Upper Guinea.
The analysis provided above must still be supported with examples: the active use of PA / SCV ′meste is illustrated in (20a, b); the passive verbal use of the corresponding participles mesté -mestedu is demonstrated in (21a, b).(NB: The lack of a TMA marker in ( 20)-( 21) is due to the fact that these are stative predicates.)(20) a. PA bo meeste tien miëdoe 2sg need to have fear 'you must be afraid' (Conradi 1844: 10, Matthew  Recall, however, that, unlike SCV ′meste -mestedu, PA ′meste~mesté have merged and, synchronically, are mere variants of one and the same modal verb meaning either 'to need' (active) or 'to be needed' (passive) (Richard Hooi, p.c.; Igma van Putte-de Windt, p.c.).This implies that speakers of PA may in (20a) just as well take recourse to mesté and in (21a) prefer the variant ′meste (with mesté more frequently used). 44

On the integration of wòrdu and ser
The data provided in Section 4 and 5 suggest that PA started out with a passive of the type [TMA + PP].How this EPA passive template subsequently 44 PA mesté also occurs as a noun meaning 'need, desire ' (van Putte & van Putte-de Windt 2005: 290).As such, mesté occurs particularly in combination with the verb tin 'to have', as in mi tin mesté di bo 1s-have-need-of-2s 'I need you'.The noun mesté appears to have its equivalent in GBC, where from the verb misti~meste 'to want', the noun mistida~mesteda 'need, necessity, desire' is derived (Peck 1988: 146;Rougé 2004: 207).The form with /e/ is found in the conservative variety of Casamance, the form with /i/ is typical of Guinea-Bissau.Just as PA mesté, GBC mesteda can follow the verb 'to have': e tem mesteda pa fasi ... 3s-have-need-to--make 'he needs to make...' (Schuchardt 1889: 302).
The introduction and integration of the passive auxiliaries wòrdu and ser appears to have been triggered by a need to disambiguate.As noted 45 The texts studied for this paper suggest that PA started integrating and productively using variants of Dutch worden in the second half of the 19th century.In the early documents dating from 1775, 1776(e.g. in Maurer 1998: 203-206) : 203-206) and 1803 (e.g. in Martinus 1996: 33, 34) as well as in the longer texts by Conradi (1844), Niewindt (1833Niewindt ( , 1837) ) and van Dissel (1865), the use of passive auxiliaries is still entirely absent.I found the first mention of a passive with wòrdu in van Dissel (1857: 127), who, in a brief two-paged grammatical sketch of PA, gives <e boto ta wordt mirá> the-boat-IMP-AUX-watched 'the boat is being watched'.(This variant of wòrdu is not attested in any other PA source.)Kuiperi (1862, 1864[in Coomans & Coomans-Eustatia 2005]) provided the first texts with a strong tendency towards the use of passives with the auxiliary wòrdu (cf.Sanchez 2005: 161).The auxiliary ser integrated somewhat later, according to Sanchez (2008: 240), in the mid--twentieth century.However, examples of ser are already found in Sintiago (1898), van de Veen Zeppenfeldt (1928) and Lenz (1928: 131-133), sources not considered by Sanchez. Lenz (1928: 132) did however stress that the use of ser was still rare.
previously, due to the ambiguous status of PA ta (past tabata) as both an imperfective aspect marker and a resultative copula, without a proper context, EPA predicates of the type ta (/ tabata) + PP could acquire both a passive and a resultative interpretation. 46The ambiguity will have increased further still due to the erosion of the past participle morpheme discussed in Section 2. The erosion of this morpheme, with now only a stress shift distinguishing between the active verb and the participle, must have interfered with the perception of the active-passive distinction.
In addition to disambiguation, the integration of the auxiliaries can be explained in terms of "equivalence", a process by which "an existing category is restructured to be equivalent to a corresponding category of the model language" (Heine & Kuteva 2005: 133)  47 .In the case of the PA passive, the existing passive template was restructured (viz.complemented with auxiliaries) in order to increase equivalence with the passives of the model languages Dutch and Spanish. 48

Reassessing the origin and originality of Papiamentu's passive morphology
The auxiliaries wòrdu and ser understandably raise suspicions of artificiality; it is safe to assume that these have significantly nourished the view described in Section 1 according to which passivization in PA is a non-native feature borrowed wholesale from Dutch and Spanish.However, while these auxiliaries undoubtedly constitute a 19th century borrowing, the data presented in Sections 4 and 5 suggest that (auxiliary-less) passive morphology was part of PA's original morpho-syntactic make-up.
Other indications point in that same direction.Note, for instance, that passives in PA are by no means marginal to the language but in fact "very widespread and (...) not confined to urban or high-prestige use" (Wood 1972b: 645).In addition, passives are found "not only in written but also in spoken discourse" (Maurer 1988: 331;cf. Busche 1993: 79).The same observations apply to UGPC: a random scan through any text written in either SCV or GBC suffices to find a wealth of examples confirming that passivization is "impressively well developed" (Thiele 1991: 73) in UGPC.Unlike most creoles, then, the use of passives abounds both in PA and in UGPC and cuts across the variational continua, facts that can be taken to support the originality of passive morphology in PA.
It is furthermore useful to point out that in both creoles the distribution of passives is often at odds with (mainstream) Spanish and Portuguese: quite commonly, where the latter would take recourse to active reflexive constructions to convey a passive meaning, both PA and UGPC use a passive construction. 49In reference to SCV, for instance, Bartens (1996: 43) observed that an utterance such as SCV ka ta papiadu na mesa NEG-IMP-spoken-at--table 'no talking at the table' appears odd, "because, unlike Portuguese, there is no reflexivization but passivization".The same observation applies to (1), ( 7), ( 13) and ( 17)-( 23), all of which would commonly be translated into Spanish and Portuguese by means of pseudo-passive clauses with the reflexive pronoun se.
In short, unlike most creoles, both PA and UGPC are characterized by the heavy usage of morphologically marked passives in acrolectal as well as in basilectal varieties and in written as well as in spoken discourse, while the distribution of these passives differs from the Iberian lexifiers' passive.Not only are these strong arguments in favor of the native character of passive morphology in PA, they also underpin the hypothesized historical link between passivization in PA and UGPC.The correspondences illustrated in Sections 4 and 5 confirm Quint's (2000b: 206) observation that "a possible relation with the passive in SCV and GBC still shines through in PA's participle morphology" and ultimately also lend further support to the hypothesis that PA and UGPC share common ancestry as postulated by Martinus (1996), Quint (2000b) and Jacobs (2008Jacobs ( , 2009aJacobs ( , 2009bJacobs ( , 2010)).This hypothesis, though controversial, is now gradually finding more acceptance among creolists (e.g.Hagemeijer 2009;McWhorter 2010;Baptista, to appear). 49 Quint (2000a: 234) noticed the same for SCV: "SCV has largely extended the use of passives to verbs that do not allow passive constructions in Portuguese".

Final remarks
If the claim that passivization is a native PA feature gains a foothold in PA studies, this could have consequences for the normative debate referred to in Section 1.As indicated there, the view that PA's passive is a construction borrowed in its entirety from Dutch and Spanish flourishes not only among creolists but also among Antillian linguists; the didactic norm is to discourage the use of passives.Although the purpose of this paper is by no means language prescriptive or didactic, the available data make any crusade against the use of passives in PA seem rather baseless.Before closing, credit should still be given to Kouwenberg (in Sanchez 2005: 198) and Sanchez (2005: 198;cf. 2008: 231) for having drawn attention to the use of auxiliary-less passives in EPA texts, as discussed in the present paper: A fourth type of passive exists as well, one which may be the oldest of all: TMA marker + PAST PARTICIPLE (Kouwenberg, p.c.).Silvia Kouwenberg (p.c.) pointed out several examples of this kind of passive to me from the 1863 Proclamasjon (...).I did not examine these auxiliariless passives in this project, but in the future I will compare their use to that of the ser/wordu/keda passives.Here, I simply hypothesize that passives without a passivizing verb existed in Papiamentu before passivizing verbs were borrowed, and in fact served as a framework onto which the foreign passivizing verbs were later borrowed.
Note, however, that Sanchez's (2005) corpus included Conradi (1844) and Niewindt (1837); the abundant use of auxiliary-less passives in these texts apparently did not strike her as unusual.Moreover, the Proclamasjon in fact contains only one (rather than "several") predicate of the type [TMA + PP], namely (17a); however, as noted previously, (17a) is presented by Kouwenberg & Ramos-Michel (2007: 319) as an active predicate.Kouwenberg's position on the matter thus remains unclear.
Note, finally, that the passive morphology shared between PA and UGPC lends support to the view that the proto-UGPC-variety brought to Curaçao in the second half of the 17th century (as hypothesized in Jacobs 2009a) was a fully fledged creole (rather than a pidgin or jargon) introduced by native speakers.This can be inferred from the fact that pidgins and jargons generally lack complex inflectional morphology in general and morphologically marked passives in particular; the absence of passive morphology is presented by Parkvall (2000: 20) as "one of the most typical features of Pidgins" (cf.e.g.Markey & Fodale 1983: 69;Crowley 2008: 82).If this line of argumentation is correct, it is implied that we can discard hypotheses that were common in the 1970s and 80s according to which an ill-defined (Afro-) Portuguese proto--pidgin or trade jargon would be responsible for the (Afro-) Portuguese elements in PA.

Table 2 .
Examples of etym.′−/dV/ > proto-UGPC -/du/ > EPA ′−/r/ > MPA ′−/Ø/ Examples with the preverbal past imperfective marker tabata − which has a close relative in SCV staba ta and particularly in Barlavento CV tabata ~ tavata − are similarly rife in the EPA texts: 31 AVDB: "mijn bloed, hetwelk voor velen vergoten wordt". 32: "Wanneer het in de aarde gezaaid wordt." 33ce: http://www.mass.gov/Elwd/docs/dia/publications/ls_cape_verdean.pdfPassivescomposed of a simple past/perfective aspect marker (PA a, UGPC Ø) and a past participle in fact abound in UGPC (e.g.13b), while these are largely absent in the EPA documents analyzed for this paper.Pending further investigation, one can only speculate as to why this should be so 37 .Nonetheless, I found two examples, (13a) and (17a, to be discussed later), which might qualify as a passive of the type a + PP in EPA.Hij werd van gedaante veranderd."37Possiblyrelevant is that PA a derives from the 3rd person singular, ha, of Spanish haber 'to have', which, asMarkey & Fodale (1983: 70)note, "is notably excluded from formation of the periphrastic passive [in both Romance and Germanic]". 38PA hende as well as Sp.hender (both 'to cleft') are transitive verbs.The lack of a direct object thus justifies the passive rather than active reading.